Bleeding Preg

DRAFT FOR CONSULTATION
Pain and bleeding in early pregnancy: full guideline DRAFT (June 2012)
Pain and bleeding in early
pregnancy: assessment
and initial management of
ectopic pregnancy and
miscarriage in the first
trimester

National Collaborating Centre for Womens
and Childrens Health

Commissioned by the National Institute for
Health and Clinical Excellence

Draft for stakeholder consultation June 2012

Pain and bleeding in early pregnancy: full guideline DRAFT (June 2012) 1

Published by the RCOG Press at the Royal College of Obstetricians and Gynaecologists, 27 Sussex
Place, Regents Park, London NW1 4RG

www.rcog.org.uk

Registered charity no. 213280

First published year

Year National Collaborating Centre for Womens and Childrens Health

No part of this publication may be reproduced, stored or transmitted in any form or by any means,
without the prior written permission of the publisher or, in the case of reprographic reproduction, in
accordance with the terms of licences issued by the Copyright Licensing Agency in the UK
[www.cla.co.uk]. Enquiries concerning reproduction outside the terms stated here should be sent to
the publisher at the UK address printed on this page.

The use of registered names, trademarks, etc. in this publication does not imply, even in the absence
of a specific statement, that such names are exempt from the relevant laws and regulations and
therefore for general use.

While every effort has been made to ensure the accuracy of the information contained within this
publication, the publisher can give no guarantee for information about drug dosage and application
thereof contained in this book. In every individual case the respective user must check current
indications and accuracy by consulting other pharmaceutical literature and following the guidelines
laid down by the manufacturers of specific products and the relevant authorities in the country in
which they are practising.

This guideline has been fully funded by NICE. Healthcare professionals are expected to take it fully
into account when exercising their clinical judgement. However, the guidance does not override the
individual responsibility of healthcare professionals to make decisions appropriate to the
circumstances of the individual patient.

Implementation of this guidance is the responsibility of local commissioners and/or providers


Contents
Contents 1
1 Guideline summary 1
1.1 Guideline development group membership, NCC-WCH staff and acknowledgements 1
1.2 Foreword (or executive summary) 1
1.3 Care pathway 2
1.4 Key priorities for implementation 10
1.5 Recommendations 11
1.6 Key research recommendations 20
1.7 Research recommendations 23
1.8 Other versions of the guideline 23
1.9 Schedule for updating the guideline 24
2 Introduction 25
2.1 Pain and bleeding in early pregnancy 25
2.2 For whom is this guideline intended 25
2.3 Related NICE guidance 26
3 Guideline development methodology 27
3.1 Introduction 27
3.2 Developing review questions and protocols and identifying evidence 27
3.3 Reviewing and synthesising evidence 28
3.4 Emotional and psychological support 29
3.5 Incorporating health economics 30
3.6 Evidence to recommendations 31
3.7 Stakeholder involvement 32
4 Psychological support 33
4.1 Introduction 33
4.2 Psychological support 33
5 Early pregnancy assessment units 44
5.1 Introduction 44
5.2 Clinical and cost effectiveness of early pregnancy assessment units 44
5.3 Model for service organisation and delivery of EPAUs 48
6 Diagnosis of ectopic pregnancy and miscarriage 62
6.1 Signs and symptoms of ectopic pregnancy 62
6.2 Ultrasound for determining a viable intrauterine pregnancy 70
6.3 Accuracy of imaging techniques for diagnosis of an ectopic pregnancy 76
6.4 Diagnostic accuracy of two or more human chorionic gonadotrophin (hCG) measurements for
ectopic pregnancy 80
6.5 Diagnostic accuracy of two or more hCG measurements plus progesterone for ectopic
pregnancy 84
6.6 Diagnostic accuracy of two or more hCG measurements for viable intrauterine pregnancy 86
6.7 Diagnostic accuracy of two or more hCG measurements plus progesterone for viable
intrauterine pregnancy 89
7 Management of threatened miscarriage and miscarriage 93
7.1 Introduction 93
7.2 Progesterone for threatened miscarriage 93
7.3 Expectant management compared with active treatment of miscarriage 100
7.4 Surgical management compared with medical management of miscarriage 111
7.5 Misoprostol and mifepristone for managing miscarriage 123
Pain and bleeding in early pregnancy
2
7.6 Setting for surgical management of miscarriage 159
8 Management of ectopic pregnancy 164
8.1 Introduction 164
8.2 Surgical compared with medical management of ectopic pregnancy 164
8.3 Laparotomy compared with laparoscopy for ectopic pregnancy 175
8.4 Salpingectomy compared with salpingotomy for ectopic pregnancy 184
9 Anti-D rhesus prophylaxis 192
9.2 Anti-D rhesus prophylaxis for threatened miscarriage, miscarriage and ectopic pregnancy 192
9.3 Anti-D rhesus prophylaxis dose 196
10 Health Economics 200
10.3 Management of miscarriage 208
10.4 Management of ectopic pregnancy 213
11 References 238
12 Abbreviations and glossary 252
12.1 Abbreviations 252
12.2 Glossary 254

1 Guideline summary 1
1.1 Guideline development group membership, NCC- 2
WCH staff and acknowledgements 3
GDG members 4
Mary Ann Lumsden Professor of Gynaecology and Medical Education (Chair) 5
Fiona Blake Consultant psychiatrist 6
Nicola Davies General practitioner 7
Karen Easton Consultant nurse 8
Roy Farquharson Consultant obstetrician and gynaecologist 9
Joanne Fletcher Consultant nurse 10
Liz Jones Lay member (stood down July 2011) 11
Julie Orford Lay member (joined July 2011) 12
Caroline Overton Consultant obstetrician and gynaecologist 13
Shammi Ramlakhan Consultant in emergency medicine 14
Helen Wilkinson Lay member 15
National Collaborating Centre for Womens and Childrens Health 16
(NCC-WCH) 17
Lauren Bardisa-Ezcurra Research fellow (until April 2011) 18
Zosia Beckles Information scientist 19
Rupert Franklin Project manager 20
Maryam Gholitabar Research associate 21
Paul Jacklin Senior health economist 22
David James Clinical co-director (womens health) 23
Emma Newbatt Research associate 24
Roz Ullman Senior research fellow and clinical lead (midwifery) 25
External advisers 26
Tom Bourne Consultant gynaecologist 27
Janette Keit Consultant sonographer 28
David Roberts Professor of haematology 29
Acknowledgements 30
Additional support was received from: 31
Wahab Bello, Julie Hodge Allen, Juliet Kenny, Edmund Peston and Wendy Riches at the NCC-WCH 32
1.2 Foreword (or executive summary) 33
The final published guideline will include a foreword or executive summary. 34
35
2
1.3 Care pathway 1
A. Providing women with information and psychological support 2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
Treat all women with dignity and respect. Be aware that women will react to complications or the loss of a
pregnancy in different ways. Provide all women with information and support in a sensitive manner, taking
into account their individual circumstances and emotional response.
Healthcare professionals providing care for women with early pregnancy complications in any setting should
be aware that early pregnancy complications can cause significant distress for some women and their
partners. Healthcare professionals providing care for these women should be given training in sensitive
communication and breaking bad news.
Throughout a womans care, provide her with appropriate verbal and written information about:
what to expect during the time that she is waiting for an ultrasound scan
when and how to seek help if existing symptoms worsen or new symptoms develop, including a 24 hour
contact telephone number where appropriate
what to expect during the course of her management, including the length and extent of pain and/or
bleeding and potential side effects
what to expect during the recovery period, for example, when it is possible to resume sexual activity
and try to conceive again, and what to do if she becomes pregnant again
where to access support and counselling services, including leaflets with helpline numbers for support
organisations
how to find further information, such as return to work documents.
Ensure that sufficient time is available to provide this information. Arrange an additional appointment if
necessary.
After early pregnancy loss, offer women the option of a follow-up appointment with a healthcare
professional in either primary or secondary care according to the womans preference.

Guideline Summary
3
B. Initial clinical assessment 1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
! !
No abdominal tenderness or
signs of intra-abdominal bleeding

Consider vaginal examination
Pain and abdominal tenderness
on examination

Women with acute presentations of
shock or collapse are outside of the
scope of this guideline (exit pathway)

Women with pain and/or bleeding, or
atypical symptoms suggestive of early
pregnancy problems

Health care professional conducts
clinical history taking, physical
examination and urine pregnancy test

Positive pregnancy test Negative pregnancy test
(exit pathway)

Cervical excitation or pelvic
tenderness

Refer immediately for further
investigation at a dedicated early
pregnancy assessment service
(see section C)
No vaginal examination
performed or no cervical
excitation or pelvic tenderness

Refer to a dedicated early
pregnancy assessment service.
The urgency of the referral
depends on the clinical situation
(see section C).
Manage expectantly. Advise
women to take a urine pregnancy
test after a week and to return if
their symptoms worsen.

Women with:
pain or
a pregnancy of
uncertain gestation or
a pregnancy of 6
weeks gestation or
more, with more than
minimal bleeding
(spotting)
Women who are not in pain
with:
a pregnancy of less than
6 weeks gestation who
are bleeding or
a pregnancy of 6 weeks
gestation or more with
minimal bleeding
(spotting)
4
B. Initial ultrasound scan 1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
!
Ectopic pregnancy
(see section H)

Pregnancy of unknown
location (see section D)

Women referred to dedicated early
pregnancy assessment service
Transvaginal ultrasound scan
(unless transabdominal scan is indicated or is
the womans preference)

Women who have self-referred to
dedicated early pregnancy
assessment service due to a
previous ectopic pregnancy
Molar pregnancy
(outside scope of
guideline exit pathway)

Intrauterine pregnancy
(see section E)

!
Guideline Summary
5
D. Pregnancy of unknown location (PUL) 1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
!
! !
Ultrasound scan between 7
and 14 days

New or
worsening
symptoms

New or
worsening
symptoms

Women with pregnancy of unknown
location

Two serum hCG measurements (48 hours apart)

hCG change between less
than or equal to 50% decline
and less than or equal to
63% rise

hCG decrease of more than
50% (likely to be failing
pregnancy)

hCG increase of greater
than 63% (likely to be a
viable intrauterine
pregnancy)
Ask women to take a urine
pregnancy test in 14 days

!
!
Urgent clinical
review by
gynaecologist

No viable
intrauterine
pregnancy

Confirmed diagnosis of non-viable pregnancy:
Complete miscarriage (exit pathway)
Ectopic pregnancy (see section H)
Incomplete/missed miscarriage (see section F)
Molar pregnancy (outside scope of guideline - exit pathway)

Confirmed
intrauterine
pregnancy
(see section
E)

Negative pregnancy test and
resolution of symptoms;
therefore, no longer
pregnant though location not
determined
(exit pathway)

Positive pregnancy
test

Urgent clinical
review by
gynaecologist

Return for
clinical
review

Return for
clinical
review

Urgent clinical review by
gynaecologist and individualised
management including further
ultrasound scan

Viable
intrauterine
pregnancy

Routine
antenatal
care
(exit pathway)

6
E. Intrauterine pregnancy 1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
!

MSD 25.0 mm MSD > 25.0 mm

Second opinion
!
!
! ! !
CRL > 6.0 mm
(> 10 mm using
transabdominal
ultrasound)

CRL 6.0 mm
( 10 mm using
transabdominal
ultrasound)

Women with confirmed intrauterine
pregnancy

Visible fetal heartbeat

No visible fetal heartbeat
(uncertain viability)

In the presence of a fetal pole,
measure crown-rump length
(CRL)

In the fetal pole is not visible,
measure mean gestational
sac diameter (MSD)

Confirmed diagnosis:
Viable intrauterine pregnancy (refer for
routine antenatal care exit pathway)
Complete miscarriage (exit pathway)
Incomplete/missed miscarriage
(see section F)

Consider progesterone/
progestogen

Perform a second
scan a minimum of
14 days after the
first. Further scans
may be needed
before diagnosis.

Perform a second scan a minimum of 7 days after
the first scan before making a diagnosis
Viable intrauterine
pregnancy

Routine antenatal
care (exit pathway)
(
Return for further
assessment

No current
vaginal bleeding

Current vaginal
bleeding

Bleeding gets
worse or persists
beyond 14 days

Bleeding
stops

!
Guideline Summary
7
F. Management of miscarriage 1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
!

Medical management
(see section G)

Where clinically appropriate, offer
women a choice of:
manual vacuum aspiration in an
out-patient or clinic setting
evacuation in an operating theatre
under general anaesthetic

Surgical management
[Give 250 IU anti-D to Rh-
women]

Continued expectant
management
Review at a minimum of 14
days after the first follow-up
appointment
For women with retained
pregnancy tissue, discuss
and offer all treatment
options

Complete miscarriage
(exit pathway)

Advise women to return for
individualised management

Negative
pregnancy test

Positive
pregnancy test

Use expectant management
for 7-14 days as the first line
management strategy
Explore other management
options in women:
at increased risk of
haemorrhage (for example in
the late first trimester)
with prior adverse and/or
traumatic experience
associated with pregnancy,
such as previous miscarriage
or antepartum haemorrhage
at increased risk from the
effects of haemorrhage
Women with diagnosed
miscarriage

Offer ultrasound scan

Urine pregnancy test
after 3 weeks

!

Resolution of bleeding and pain
indicate that miscarriage has
completed

!
Bleeding and pain have not
started (suggesting process of
miscarriage has not begun) or
Bleeding or pain are persisting
and/or increasing (suggesting
incomplete miscarriage)

8
G. Medical management of miscarriage 1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
!
If bleeding has not commenced 24 hours
after treatment, there should be a clinical
review to determine individualised care

Follow-up to ensure there is no molar
pregnancy or ectopic pregnancy

No molar
pregnancy or
ectopic
pregnancy
(exit pathway)

Ectopic
pregnancy
(see
section H)

Molar
pregnancy
(exit
pathway)

Advise women to take a urine pregnancy test
3 weeks after medical management, unless
they experience worsening symptoms, in
which case advise them to return

Women undergoing medical
management of miscarriage

Offer anti-emetics and pain relief
as required

Women with incomplete
miscarriage

Use a single dose of 600
micrograms misoprostol vaginally
(unless oral is the womans
preference)

Women with missed
miscarriage

Use a single dose of 800
micrograms misoprostol vaginally
(unless oral is the womans
preference)

Negative pregnancy
test (exit pathway)

Positive pregnancy
test

!
!
Guideline Summary
9
G. Ectopic pregnancy 1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
Laparoscopy (in normal working hours) should be
performed whenever possible, taking into
account the condition of the woman and the
complexity of the surgical procedure
[Offer 250 IU anti-D to Rh- women]

Offer methotrexate as first
line treatment

Offer surgery where treatment
with methotrexate is not
acceptable to the woman

Women with any of the following:
ectopic pregnancy that is 3.5 cm or larger
ectopic pregnancy with fetal heartbeat
visible on ultrasound
ectopic pregnancy and an hCG level of
5000 IU/l or more
anticipated difficulties with follow-up

Women with diagnosed tubal ectopic pregnancy
Women who have all of the following:
no significant pain
an hCG level less than 5000 IU/l
an unruptured ectopic pregnancy smaller
than 3.5 cm with no visible heartbeat
availability for follow-up
Exercise additional consideration
(including need for urgent
admission) if hCG is greater than
1500 IU/l

In women with no risk
factors for infertility, perform
salpingectomy
In women with factors
prognostic of infertility, such
as contralateral tube
damage, consider
salpingotomy

Urine pregnancy test after 3
weeks (exit pathway)

Follow-up serum hCG
measurement at 7 days after
surgery, and then one per
week until a negative result is
obtained (exit pathway)

Two follow-up serum hCG
measurements in the first week (day 4
and 7) following treatment, and then one
per week following that until a negative
result is obtained (exit pathway)

Systemic methotrexate treatment

Salpingectomy Salpingotomy
10
This guideline recommends some drugs for indications for which they do not have a United Kingdom 1
(UK) marketing authorisation at the date of publication, if there is good evidence to support that use. 2
Many drugs do not have a license for use specifically in pregnant women, reflecting the fact that this 3
group is often excluded from studies. Unlicensed drugs are indicated with a footnote. 4
1.4 Key priorities for implementation 5
Number Recommendation See
section

Early pregnancy assessment units

4 A dedicated early pregnancy assessment service should be
available at least during Monday to Friday for women with pain
and/or bleeding in early pregnancy, where scanning can be carried
out and decisions about management made.
5.3

Signs and symptoms of ectopic pregnancy

8 All healthcare professionals involved in the care of women of
reproductive age should have access to pregnancy tests.
6.1

Ultrasound for determining a viable intrauterine
pregnancy

19 Offer women with pain and/or bleeding a transvaginal ultrasound
scan to identify the location of the pregnancy and whether there is a
fetal pole and heartbeat.
6.2

Diagnostic accuracy of hCG measurements with or
without progesterone for determining ectopic
pregnancy and viable intrauterine pregnancy

37 Assume women with a pregnancy of unknown location have an
ectopic pregnancy until the location is determined.
6.7

Management of miscarriage

50 Use expectant management for 7-14 days as the first-line
management strategy following confirmed diagnosis of a non-viable
pregnancy.
7.4

Setting for surgical management of miscarriage

67 Where clinically appropriate, offer women a choice of:
manual vacuum aspiration under local anaesthetic in an
evacuation in a theatre under general anaesthetic.
7.6

Laparotomy compared with laparoscopy for ectopic
pregnancy

77 When surgical treatment is indicated for women with an ectopic
pregnancy, laparoscopy should be performed whenever possible,
taking into account the condition of the woman and the complexity
of the surgical procedure.
8.3

Salpingectomy compared with salpingotomy for
ectopic pregnancy

81 Offer a salpingectomy to women undergoing surgery for an ectopic
pregnancy, unless they have other risk factors for infertility.
8.4
Guideline Summary
11
1
1.5 Recommendations 2
section

Psychological support

1 Treat all women with pain or bleeding in early pregnancy with
dignity and respect. Be aware that women will react to
complications or the loss of a pregnancy in different ways. Provide
all women with information and support in a sensitive manner,
taking into account their individual circumstances and emotional
response.
*

4.2
2 Healthcare professionals providing care for women with early
pregnancy complications in any setting should be aware that early
pregnancy complications can cause significant distress for some
women and their partners. Healthcare professionals providing care
for these women should be given training in sensitive
4.2
3 After an early pregnancy loss, offer the woman the option of a
follow-up appointment with a healthcare professional in either
primary or secondary care according to the womans preference.
4.2


4 A dedicated early pregnancy assessment service should be
available at least during Monday to Friday for women with pain
and/or bleeding in early pregnancy, where scanning can be carried
out and decisions about management made.
5.3
5 Dedicated early pregnancy assessment services should accept
self-referrals from women who have had a previous ectopic or
molar pregnancy. All other women with pain and/or bleeding should
be assessed by a healthcare professional before referral to an early
pregnancy assessment service.
5.3
6 Ensure that a system is in place to enable women referred to their
local dedicated early pregnancy assessment service to attend it
within 24 hours if the clinical situation warrants it. If the service is
not available at weekends, and the clinical symptoms warrant
further investigation, refer women to the nearest accessible facility
that offers specialist clinical assessment and ultrasound scanning.
5.3
5.3


7 When assessing women of reproductive age, be aware that they
may be pregnant, and think about offering a pregnancy test even
when symptoms are non-specific.
6.1
8 All healthcare professionals involved in the care of women of
reproductive age should have access to pregnancy tests.
6.1
9 Exclude the possibility of ectopic pregnancy, even in the absence of
risk factors (such as previous ectopic pregnancy), because about a
third of women with an ectopic pregnancy will have no known risk
6.1

* For further guidance about providing information, see NICE clinical guideline 138 Patient experience in adult NHS services
improving the experience of care for people using adult NHS services (2012)
12
section
factors.

Symptoms and signs

10 Be aware that atypical presentation for ectopic pregnancy is
common.
6.1
11 Be aware that ectopic pregnancy can present with a variety of
symptoms. Even if a symptom is less common, it may still be
significant. Symptoms of ectopic pregnancy include:
Common symptoms
o Abdominal or pelvic pain
o Amenorrhoea or missed period
o Vaginal bleeding without clots
Other reported symptoms
o Breast tenderness
o Gastrointestinal symptoms
o Dizziness, fainting or syncope
o Shoulder tip pain
o Urinary symptoms
o Passage of tissue
o Rectal pressure or defecation

12 Be aware that ectopic pregnancy can present with a variety of
signs. Signs of ectopic pregnancy include:
More common signs
o Pelvic tenderness
o Adnexal tenderness
o Abdominal tenderness
Other reported signs
o Cervical motion tenderness
o Rebound tenderness or peritoneal signs
o Pallor
o Abdominal distension
o Enlarged uterus
o Tachycardia (more than 100 beats per minute) or
hypotension (less than 100/60 mmHg)
o Shock or collapse
o Orthostatic hypotension.

1
2
section
13 If a woman has a positive pregnancy test the threshold for
considering an ectopic pregnancy should be low because the
symptoms and signs of ectopic pregnancy can resemble the
common symptoms and signs of other conditions - for example
gastrointestinal conditions or urinary tract infection.
6.1
14 Refer women with a positive pregnancy test and continuing or
worsening symptoms and signs that could suggest ectopic
pregnancy to a dedicated early pregnancy assessment service for
6.1
Guideline Summary
13
section
further evaluation.
15 If a woman needs to be referred to a dedicated early pregnancy
assessment service, explain the reasons for the referral and what
she can expect when she arrives there.
6.1

Signs

16 Refer immediately women with a positive pregnancy test and pain
and abdominal tenderness on examination to a dedicated early
pregnancy assessment service for further investigation.
6.1
17 If there is no abdominal tenderness or signs of intra-abdominal
bleeding, but there is still a suspicion of an ectopic pregnancy,
consider a vaginal examination to assess for cervical excitation or
pelvic tenderness. If either of these is present, together with a
positive pregnancy test, refer the woman immediately to a
dedicated early pregnancy assessment service.
6.1

pregnancy

18 Use expectant management for:
women with a pregnancy of less than 6 weeks gestation
who are bleeding but not in pain
women with a pregnancy of 6 weeks gestation or more
who have minimal blood loss (spotting) and who are not in
pain.
Advise these women to take a urine pregnancy test after a week
and to return if their symptoms worsen
Refer all other women to a dedicated early pregnancy assessment
service.
6.2
19 Offer women with pain and/or bleeding a transvaginal ultrasound
scan to identify the location of the pregnancy and whether there is a
fetal pole and heartbeat.
6.2
20 If a transvaginal ultrasound scan is unacceptable to the woman
offer a transabdominal ultrasound scan and explain the limitations
of this method of scanning.
6.2
21 Inform women that the diagnosis of miscarriage using one
ultrasound scan cannot be guaranteed to be 100% accurate and
there is a small chance that the diagnosis may be incorrect,
particularly at very early gestational ages.
6.2
22 If the crown rump length is 6.0 mm or less with a transvaginal
ultrasound scan (or 10.0 mm or less with a transabdominal
ultrasound scan) and there is no visible fetal heartbeat, perform a
second scan a minimum of 7 days after the first before making a
diagnosis.
6.2
23 If the crown rump length is more than 6.0 mm with a transvaginal
ultrasound scan (or 10.0 mm with a transabdominal ultrasound
scan), and there is no visible fetal heartbeat:
seek a second opinion on the viability of the pregnancy
and/or
perform a second scan a minimum of 7 days after the first
6.2
14
section
before making a diagnosis.
24 Do not measure the mean gestational sac diameter if there is a
fetal heartbeat.
6.2
25 Measure the mean gestational sac diameter if the fetal pole is not
visible.
6.2
26 If the mean gestational sac diameter is 25.0 mm or less with no
fetal pole, offer the woman a second scan, to be performed a
minimum of 14 days after the first. Further scans may be needed
before a diagnosis can be made.
6.2
27 If the mean gestational sac diameter is greater than 25.0 mm with
no fetal pole:
seek a second opinion on the viability of the pregnancy
and/or
perform a second scan a minimum of 7 days after the first
before making a diagnosis.
6.2
28 Do not use gestational age alone to determine whether a fetal
heartbeat should be visible.
6.2
29 Inform women that the date of their last menstrual period may not
give an accurate representation of gestational age because of
variability in the menstrual cycle.
6.2
30 Inform women what to expect while waiting for a repeat scan and
that waiting for a repeat scan has no detrimental effects on the
outcome of the pregnancy.
6.2
31 Give women specific verbal and written information on when and
how to seek help if symptoms worsen or new symptoms develop.
6.2
32 Give women a 24-hour contact telephone number so that they can
speak to someone who understands their needs and can advise on
appropriate care.
6.2
33 Provide women who have a confirmed miscarriage with written and
verbal information about where to access support and counselling
services (including links to useful websites).
6.2

Accuracy of imaging techniques for diagnosis of an
ectopic pregnancy

34 Offer a transvaginal ultrasound scan to diagnose or exclude an
ectopic pregnancy.
6.3
35 Consider a transabdominal ultrasound scan for women with an
enlarged uterus or other pelvic pathology, such as fibroids or an
ovarian cyst.
6.3
36 All ultrasound scans should be performed or reviewed by someone
with training in and experience of diagnosing ectopic pregnancies.
6.3

Diagnostic accuracy of hCG measurements with or
without progesterone for determining ectopic
pregnancy and viable intrauterine pregnancy

37 Assume women with a pregnancy of unknown location have an
ectopic pregnancy until the location is determined.
6.7
Guideline Summary
15
section
38 In women with a pregnancy of unknown location, place more
importance on clinical symptoms than on human chorionic
gonadotrophin (hCG) results, and review the womans condition if
any of her symptoms change, regardless of previous results and
assessments.
6.7
39 Use hCG measurements only for determining trophoblastic
proliferation, to help decide ongoing monitoring.
6.7
40 Do not use hCG measurements to determine the location of the
pregnancy.
6.7
41 Regardless of hCG levels, give women with a pregnancy of
unknown location written information about what to do if they
experience any new or worsening symptoms, including details
about how to access emergency care 24 hours a day. Advise
women to return if there are new symptoms or if existing symptoms
worsen.
6.7
42 Take two serum hCG measurements 48 hours apart to determine
subsequent management of a pregnancy of unknown location.
Take further measurement only after review by a senior healthcare
professional.
6.7
43 Inform a woman with an increase in serum hCG concentration
greater than 63% that she is likely to have a viable intrauterine
pregnancy. Offer her a transvaginal ultrasound scan between 7 and
14 days of the serum hCG test to confirm this:
if a viable intrauterine pregnancy is confirmed, offer her
routine antenatal care.
if a viable intrauterine pregnancy is not confirmed, refer her
for urgent clinical review by a gynaecologist.
6.7
44 For a woman with a decrease in serum hCG
concentration greater than 50%:
inform her that the pregnancy is unlikely to continue but
that this is not confirmed
provide her with written and verbal information about where
she can access support and counselling services
ask her to take a urine pregnancy test 14 days after the
serum hCG testand explain that:
o if the test is negative, no further action is
necessary.
o if the test is positive, she should return for urgent
clinical review by a gynaecologist.
6.7
45 For a woman with a change in serum hCG concentration between a
50% decline and 63% rise inclusive:
refer her for urgent clinical review by a gynaecologist and
individualised management
refer her for a repeat scan between 7 and 14 days
ask her to report worsening symptoms without delay.
6.7
46 For women with a pregnancy of unknown location, do not use
progesterone measurements together with hCG measurement as a
diagnostic test for either viable intrauterine pregnancy or ectopic
6.7
16
section
pregnancy.

Progesterone for threatened miscarriage

47 Consider progesterone/progestogen for women with a threatened
miscarriage (that is, vaginal bleeding and a confirmed intrauterine
pregnancy with a fetal heartbeat).

7.2
48 Inform women that although there is some evidence that
progesterone/progestogen can prevent a miscarriage, this evidence
is not strong.

7.2
49 Advise a woman with threatened miscarriage that:
if her bleeding gets worse, or persists beyond 14 days, she
should return for further assessment.
if the bleeding stops, she should book for, or return to,
routine antenatal care.
7.2


50 Use expectant management for 7-14 days as the first-line
management strategy following confirmed diagnosis of a non-viable
pregnancy.
7.4
51 Explain expectant management and that most women will need no
further treatment. Also provide written and verbal information about
further treatment options.
7.4
52 Give all women verbal and written information about:
what to expect throughout the process, including the likely
duration and severity of bleeding, advice on pain relief and
where and when to get help in an emergency
what to expect in the recovery period, such as how long
they should wait before resuming sexual activity and trying
to conceive.
Ensure that sufficient time is available to discuss these issues with
women. Arrange an additional appointment if necessary.
7.4
53 Inform women where to access support and counselling services.
Provide them with leaflets that include helpline numbers for support
organisations.
7.4
54 Explore management options other than expectant management if:
the woman is at increased risk of haemorrhage (for
example, she is in the late first trimester), or
she has previous adverse and/or traumatic experience
associated with pregnancy such as previous miscarriage or
antepartum haemorrhage, or
she is at increased risk from the effects of haemorrhage.
7.4
55 If the resolution of bleeding and pain indicate that the miscarriage
has completed during 714 days of expectant management, advise
the woman to take a urine pregnancy test after 3 weeks, and to
7.4

At the time of publication (June 2012), progesterone/progestogen did not have UK marketing authorisation for this indication.
Informed consent should be obtained and documented.
Guideline Summary
17
section
return if it is positive for individualised management.
56 Offer a repeat scan if after the period of expectant management the
bleeding and pain:
have not started (suggesting that the process of
miscarriage has not begun) or
are persisting and/or increasing (suggesting incomplete
miscarriage).
Discuss all treatment options with the woman to allow her to make
an informed choice.
7.4
57 Review the condition of a woman who opts for continued expectant
management at a minimum of 14 days after the first follow-up
appointment.
7.4

Misoprostol and mifepristone for managing
miscarriage

58 Offer vaginal misoprostol for the medical treatment of missed or
incomplete miscarriage. Oral administration is an acceptable
alternative if this is the womans preference.

7.5
59 For women with a missed miscarriage, use a single dose of 800
micrograms of misoprostol.
**

7.5
60 If bleeding has not started 24 hours after treatment, there should be
a clinical review to determine individualised care.
7.5
61 For women with an incomplete miscarriage, use a single dose of
600 micrograms of misoprostol. (800 micrograms can be used as
an alternative to allow alignment of treatment protocols for missed
and incomplete miscarriage.)

7.5
62 Do not offer mifepristone as a treatment for missed or incomplete
miscarriage.
7.5
63 Offer all women receiving medical management of miscarriage pain
relief and anti-emetics as required.
7.5
64 Inform women about what to expect throughout the process,
including the length and extent of bleeding and the potential side
effects of treatment including pain, diarrhoea and vomiting.
7.5
65 Advise women to take a urine pregnancy test 3 weeks after medical
management unless they experience worsening symptoms, in
which case advise them to return to the healthcare professional
responsible for providing their medical management.
7.5
66 For women with a positive pregnancy test, a follow-up process
should be in place to ensure that there is no molar or ectopic
pregnancy
7.5

Setting for surgical management of miscarriage

Although this use is common in UK clinical practice, at the time of publication (June 2012), misoprostol did not have UK
marketing authorisation for this indication. Informed consent should be obtained and documented.
** Although this use is common in UK clinical practice, at the time of publication (June 2012), misoprostol did not have UK
18
section
manual vacuum aspiration under local anaesthetic in an
7.6
68 Provide written and verbal information to all women undergoing
surgical management about the treatment options available and
what to expect during and after the procedure.
7.6

Surgical compared with medical management of
ectopic pregnancy

69 Give women relevant information in a variety of formats throughout
their care. This information should include:
information about post operative care (for women
undergoing surgery)
information about resuming normal activity
reference to the emotional impact of the loss of a baby
reference to potential future complications and what to do
next time the woman becomes pregnant
advice about how long to wait before trying to conceive
what follow-up support is available from within the hospital
and details of patient support groups provided
how the woman can contact a health care professional for
post-operative advice if needed, and who this will be
how to find further information for example, on returning
to work.
8.2
70 Inform women who have had an ectopic pregnancy that they can
self-refer to an early pregnancy assessment service in future
pregnancies if they have any early concerns.
8.2
what to expect throughout the course of their treatment and
recovery, especially the likely duration and severity of pain
and/or of bleeding
where and when to get help in an emergency.
8.2
72 Do not offer methotrexate to women who have been diagnosed with
an ectopic pregnancy unless they can return for follow-up.
8.2
73 Offer systemic methotrexate
as a first-line treatment to women

who have all of the following:
no significant pain
an unruptured ectopic pregnancy smaller than 3.5 cm with
no visible heartbeat,
an hCG level less than 5000 IU/l.
Offer surgery where treatment with methotrexate is not acceptable
to the woman.
8.2
74 Offer surgery as a first-line treatment to women with any of the
following:
8.2

Although this use is common in UK clinical practice, at the time of publication (June 2012), methotrexate did not have UK
Guideline Summary
19
section
an ectopic pregnancy that is 3.5 cm or larger,
an ectopic pregnancy with a fetal heartbeat visible on
ultrasound,
an ectopic pregnancy and an hCG level of 5000 IU/l or
more.
75 Exercise additional consideration with women who have an hCG
level greater than 1500 IU/l who choose medical treatment. Be
aware that:
their chance for needing further intervention is increased
they may need to be urgently admitted if their condition
deteriorates.
8.2
76 For women who have had methotrexate, take two serum hCG
measurements in the first week (day 4 and 7) following treatment
and then one hCG measurement per week until a negative result is
obtained.
8.2

Laparotomy compared with laparoscopy for ectopic
pregnancy

77 When surgical treatment is indicated for women with an ectopic
pregnancy, laparoscopy should be performed whenever possible,
taking into account the condition of the woman and the complexity
of the surgical procedure.
8.3
78 Surgeons providing care to women with ectopic pregnancy should
be competent to perform laparoscopic surgery.
8.3
79 Out-of-hours surgery should only be carried out as an emergency if
a woman is haemodynamically unstable or collapsed.
8.3
80 Commissioners and managers should ensure that equipment for
laparoscopic surgery is available.
8.3

Salpingectomy compared with salpingotomy for
ectopic pregnancy

81 Offer a salpingectomy to women undergoing surgery for an ectopic
pregnancy, unless they have other risk factors for infertility.
8.4
82 Consider salpingotomy as an alternative to salpingectomy for
women with factors prognostic of infertility such as contralateral
tube damage.
8.4
83 Inform women having a salpingotomy that up to one in five women
may need further treatment. This treatment may include
methotrexate administration and/or a salpingectomy.
8.4
84 For women who have had a salpingotomy, take one serum hCG
measurement at 7 days after surgery, then one hCG measurement
per week until a negative result is obtained.
8.4
85 Advise women who have had a salpingectomy that they should
take a urine pregnancy test after 3 weeks.
8.4

Anti-D rhesus prophylaxis

86 Offer anti-D rhesus prophylaxis at a dose of 250 IU (50
micrograms) to all rhesus negative women who have a surgical
9.3
20
section
procedure to manage an ectopic pregnancy or a miscarriage.
87 Do not offer anti-D rhesus prophylaxis to women who:
receive solely medical management for an ectopic
pregnancy or miscarriage
have a threatened miscarriage
have a complete miscarriage
have a pregnancy of unknown location.
9.3
88 Do not use a Kleihauer test for quantifying feto-maternal
haemorrhage.
9.3
1
1.6 Key research recommendations 2
Number Research recommendation See
section


RR 2 A national evaluation of early pregnancy assessment unit service
provision should be undertaken with statistical analysis to identify
factors affecting outcomes.
5.3

Why this is important

The first report of an early pregnancy assessment unit in England
was published over 20 years ago, and prompted the rapid
development of centres for the management of problems in early
pregnancy. Today there are an estimated 150 early pregnancy
assessment units in England and Wales (Association of early
pregnancy units, 2012). However, there is considerable variation
between centres in access to services and levels of care
provided. In addition, there has been very little good quality
research on the effectiveness of early pregnancy assessment
units in improving physical and emotional health.
A national audit of early pregnancy assessment units would help
to make up for this lack of information. Such an audit should be
along the lines of the National Caesarean Section Sentinel Audit,
a cross-sectional national survey of service configuration and
outcomes. Data recorded would include service location, opening
hours, and the healthcare professionals involved. Outcomes
would include time of attendance, length of stay, admission rates,
time to treatment and patient experience. Obtaining some of this
information would require units to undertake more formal follow-
up of patients than they may do currently, for the duration of the
audit. The evaluation should be structured to allow for
comparisons between different models of care.
Comparative outcome data collected would be used to conduct
an analysis of the cost effectiveness of early pregnancy
assessment units.

Guideline Summary
21
section

pregnancy

RR 4 How does the timing and frequency of ultrasound examination
affect diagnosis and outcomes of early pregnancy complications,
including patient experience and cost effectiveness?
6.2


The rationale behind the frequency of ultrasound to improve
diagnosis and outcomes of early pregnancy complications
addresses the problems associated with pregnancy of unknown
location and intrauterine pregnancy of uncertain viability. There is
no evidence base for timing of scans, and number of scans is
organised by individual units according to capacity and demand.
Some experts choose to wait 5 days between scans while others
will wait 10 to 14. These decisions are driven by resource
availability as well as clinical considerations. Discussions among
experts have failed to provide clear consensus, but all are agreed
that by 14 days a diagnosis will be clear. To establish the most
appropriate time for scans, the efficacy of scans taken after 14
days could be compared with scans taken after 7 days for
diagnosis of ectopic pregnancy or viability.


RR 5 Is progesterone/progestogen effective in treating threatened
miscarriage?
7.2

Approximately 20% of pregnancies miscarry in the first trimester
and many women will experience some bleeding and/or pain in
early pregnancy that does not cause miscarriage. In many
countries, women with bleeding and/or pain will be treated with
progesterone or progestogens to try and decrease the risk of
miscarriage. The evidence for the effectiveness of this treatment
has been inconclusive, but data from a meta-analysis of several
small studies suggest that progestogens are better than placebo.
However, there are theoretical risks to prescribing any treatment
in pregnancy and for many practitioners this will be a major
change in practice. The lack of strong evidence makes this a
priority area for research.
A multicentre randomised controlled trial of at least 800 women
treated with either progesterone/progestogen or placebo should
be conducted. The population would be women with pain and
bleeding and a spontaneous, confirmed, viable, singleton,
intrauterine pregnancy between 612 weeks gestation.
Progesterone/progestogen or placebo would be administered
from when bleeding starts until the end of the 13th week.
Pregnancy proceeding beyond the end of the first trimester might
be the primary outcome. Ideally, live birth would be measured,
but, this will make the trial much more expensive, and a majority
of pregnancies that are going to miscarry will do so in the first
trimester. However, pregnancy outcome should be a secondary

22
section
outcome, along with gestation at birth and presence of congenital
abnormalities.


RR 6 In women with confirmed miscarriage, does the type of
intervention (expectant, medical and surgical) impact on womens
experience, including psychological and emotional outcomes?
7.4


The management of miscarriage in the UK has changed in many
ways over the past two decades, particularly in the shift from
inpatient to outpatient or day case care and the introduction of
medical and expectant management as alternatives to surgery.
Despite these changes there is a lack of research into the effects
of these different approaches from the patients perspective, in
particular their psychological and emotional impact. Miscarriage
is distressing for most women, and the type of management itself
might affect womens need for counselling, with a resulting cost
to the NHS. Because of this it is an important area for research.
The deficiency in the literature could be addressed by a
comparative study of women having the different treatments
(expectant, medical or surgical) and in a variety of clinical
settings (for example, early pregnancy assessment unit,
gynaecological ward, or gynaecological emergency unit). The
data collected could be both quantitative (using validated
psychological health questionnaires) and qualitative (focusing
particularly on womens experience of the particular type and
setting of care).

ectopic pregnancy

RR 8 In women with ectopic pregnancy, does the type of intervention
(laparoscopy or medical management) impact on patient
8.2


Currently there is no evidence exploring the psychological impact
of the different treatments for ectopic pregnancy. However, the
emotional impact of the condition can be significant, in some
circumstances leading to post-traumatic stress disorder. A
qualitative comparative study should be undertaken to assess
how this impact can be reduced. This would help to maximise
womens emotional recovery in the short and long term, enable
women and clinicians to decide the optimum treatment method
and identify what support is needed for women during and after
the process. It could also reduce the cost to the NHS of providing
long-term counselling for affected women.

1
Guideline Summary
23
1.7 Research recommendations 1
section


RR 1 What is the psychological impact of the different types of early
pregnancy loss?
4.2


RR 2 A national evaluation of early pregnancy assessment unit service
provision should be undertaken with statistical analysis to identify
factors affecting outcomes.
5.3


RR 3 Research should be undertaken to design and validate a decision
tool for evaluating signs, symptoms and risk factors for correctly
identifying ectopic pregnancy
6.1

pregnancy

RR 4 How does the timing and frequency of ultrasound examination
improve diagnosis and outcomes of early pregnancy
complications, including patient experience and cost-
effectiveness?
6.2


RR 5 Is progesterone/progestogen effective in treating threatened
miscarriage?
7.2


RR 6 In women with confirmed miscarriage, does the type of
intervention (expectant, medical and surgical) impact on womens
7.4

Misoprostol and mifepristone for managing
miscarriage

RR 7 Is the combination of mifepristone and misoprostol more effective
than misoprostol alone in the medical management of
miscarriage?
7.5

ectopic pregnancy

RR 8 In women with ectopic pregnancy, does the type of intervention
(laparoscopy or medical management) impact on patient
8.2


RR 9 Does the administration of anti-D rhesus prophylaxis following pain
and bleeding in early pregnancy improve outcomes?
9.3
2
1.8 Other versions of the guideline 3
Links will be provided to the other versions of the guideline when they are published. 4
24
1.9 Schedule for updating the guideline 1
Clinical guidelines commissioned by NICE are published with a review date 3 years from the date of 2
publication. Reviewing may begin before 3 years have elapsed if significant evidence that affects 3
guideline recommendations is identified sooner. 4
5


2 Introduction 1
2.1 Pain and bleeding in early pregnancy 2
Pain and bleeding in early pregnancy has an adverse effect on the quality of life of many women. 3
Approximately 20% of pregnancies miscarry, and miscarriages can cause considerable distress. Early 4
pregnancy loss accounts for over 50,000 admissions in the UK annually (The Health and Social Care 5
Information Centre, 2012). The rate of ectopic pregnancy is 11 per 1000 pregnancies, with a maternal 6
mortality of 0.2 per 1000 estimated ectopic pregnancies (Cantwell et al., 2011). About two thirds of 7
these deaths are associated with substandard care. Women who do not access medical help readily 8
(such as women who are recent migrants, asylum seekers, refugees, or women who have difficulty 9
reading or speaking English) are particularly vulnerable. Improvement in the diagnosis and 10
management of early pregnancy loss is thus of vital importance, in order to reduce the incidence of 11
the associated psychological morbidity and avoid the unnecessary deaths of women with ectopic 12
pregnancies. 13
Women with pain and bleeding in early pregnancy will often seek guidance and help from their 14
general practitioners or from services within secondary care. These services can include accident and 15
emergency departments or dedicated early pregnancy assessment units (EPAUs). Accessibility of the 16
latter varies from region to region with wide variation in service provision in terms of staffing structure, 17
opening hours and acceptability of self-referral. Provision of an EPAU is a requirement of the National 18
Service Framework for children, young people and maternity care. However, there has been little 19
discussion on the cost effectiveness of EPAUs and it is not known whether they improve outcomes for 20
women with pain and bleeding in early pregnancy. 21
The guideline covers diagnosis of early pregnancy loss, including the use of ultrasound scanning and 22
biochemical testing. Investigations are not without cost and, in addition, the use of serial 23
measurements may delay decision making. The guideline includes guidance on when senior and/or 24
specialist advice should be sought in order to avoid errors and unnecessary delay. 25
Treatment for threatened miscarriage has been offered by many clinicians over the years, although it 26
is not freely available to all women. Even though progesterone/progestogen is not licensed for this 27
purpose in the UK, it is commonly prescribed in many countries around the world. The guideline will 28
examine the evidence for the risks and benefits of this treatment. 29
The clinical and cost effectiveness of expectant, surgical and medical management for miscarriage 30
and surgical and medical treatment of ectopic pregnancy will be considered, with reviews looking at 31
both the risks and benefits of each strategy in terms of clinical and psychological outcomes. Cost 32
effectiveness is an extremely important component of any guideline, in order to ensure that the limited 33
resources of the National Health Service are used to maximise health benefit for its users. The final 34
advice and selection of first line treatment takes this into account. 35
The guideline does not cover pregnancy after the first trimester (after 12 completed weeks of 36
pregnancy). It also does not deal with unusual conditions that present with pain and bleeding, such as 37
hydatidiform mole, which require a different form of treatment. Similarly, it does not consider recurrent 38
miscarriage, as this requires more specific investigation and management. 39
2.2 For whom is this guideline intended 40
This guidance is of relevance to those who work in or use the National Health Service (NHS) in 41
England and Wales, in particular: 42
GPs, primary care teams 43
26
professional groups who are routinely involved in the care of pregnant women, 1
especially early pregnancy care 2
health care professionals working in Accident and Emergency departments 3
GPs and primary care professionals who may encounter pregnant women in the course 4
of their professional duties, for example adult mental health professionals 5
those responsible for commissioning and planning healthcare services. 6
In addition, this guidance may be of relevance to professionals working in social 7
services and education/childcare settings. 8
2.3 Related NICE guidance 9
Published guidance 10
Antenatal and postnatal mental health. NICE clinical guideline 45 (2007). Available from 11
www.nice.org.uk/guidance/CG45 12
Antenatal care. NICE clinical guideline 62 (2008). Available from 13
Diabetes in pregnancy. NICE clinical guideline 63 (2008). Available from 15
Fertility. NICE clinical guideline 11 (2004). Available from 17
Hypertension in pregnancy. NICE clinical guideline 107 (2010). Available from 19
Pregnancy and complex social factors. NICE clinical guideline 110 (2010). Available 21
from www.nice.org.uk/guidance/CG110 22
Routine antenatal anti-D prophylaxis for women who are rhesus D negative. NICE 23
technology appraisal guidance 156 (2008). Available from 24
www.nice.org.uk/guidance/TA156 25
Surgical site infection. NICE clinical guideline 74 (2008). Available from 26
Venous thromboembolism reducing the risk. NICE clinical guideline 92 (2010). 28
Available from www.nice.org.uk/guidance/CG92 29
Patient experience in adult NHS services: improving the experience of care for people 30
using adult NHS services. NICE clinical guideline 138 (2012). Available from 31
Guidance under development 33
NICE is currently developing the following related guidance (details available from the NICE website): 34
Diabetes in pregnancy (update). NICE clinical guideline. Publication date to be 35
confirmed. 36
Fertility (update). NICE clinical guideline. Publication date to be confirmed. 37
38
39
40
41
42


3 Guideline development 1
methodology 2
3.1 Introduction 3
This guidance was commissioned by NICE and developed in accordance with the guideline 4
development process outlined in the 2009 edition of The Guidelines Manual 5
(www.nice.org.uk/guidelinesmanual) 6
In accordance with NICEs Equality Scheme, ethnic and cultural considerations and factors relating to 7
disabilities have been considered by the GDG throughout the development process and specifically 8
addressed in individual recommendations where relevant. Further information is available from: 9
www.nice.org.uk/aboutnice/howwework/NICEEqualityScheme.jsp. 10
3.2 Developing review questions and protocols and 11
identifying evidence 12
The GDG formulated review questions based on the scope (see Appendix A) and prepared a protocol 13
for each review question (see Appendix D). These formed the starting point for systematic reviews of 14
relevant evidence. Published evidence was identified by applying systematic search strategies (see 15
Appendix E) to the following databases: Medline (1950 onwards), Embase (1980 onwards), 16
Cumulative Index to Nursing and Allied Health Literature (CINAHL; 1982 onwards), and three 17
Cochrane databases (Cochrane Central Register of Controlled Trials, Cochrane Database of 18
Systematic Reviews, and the Database of Abstracts of Reviews of Effects). Searches to identify 19
economic studies were undertaken using the above databases, the NHS Economic Evaluation 20
Database (NHS EED), and the Health Technology Assessment (HTA) database. Four of the fourteen 21
searches were limited by date appropriate to the interventions being considered (questions relating to 22
biochemical and ultrasound diagnosis of miscarriage, effectiveness of early pregnancy assessment 23
units and treatment setting for management of miscarriage see protocols in Appendix D for details). 24
The searches were limited by language of publication (although publications in languages other than 25
English were not reviewed). Generic and specially developed search filters were used to identify 26
particular study designs, such as randomised controlled trials (RCTs). There was no systematic 27
attempt to search grey literature (conference abstracts, theses or unpublished trials), nor was hand 28
searching of journals not indexed on the databases undertaken. 29
For four of the review topics a joint search strategy was developed and run to cover more than one 30
question within that topic. The databases of identified titles and abstracts were then weeded and 31
papers allocated to their individual question before further weeding. This was carried out for the two 32
ultrasound reviews, the four reviews on human chorionic gonadotrophin and progesterone for 33
diagnosing early pregnancy loss, three reviews on the management of miscarriage (expectant 34
compared with active management, medical compared with surgical management, and dose of 35
mifepristone and misoprostol) and two reviews on anti-D rhesus prophylaxis. Following weeding 36
within each individual question full text versions of remaining studies were ordered. Each full text 37
version was then assessed for inclusion/exclusion against pre-defined criteria as detailed in the 38
protocol. Flow diagrams detailing these processes for each question can be found in Appendix F and 39
details of excluded studies in Appendix G. 40
Towards the end of the guideline development process, the searches were updated and re-executed 41
to include evidence published and indexed in the databases by 8
th
February 2012. 42
28
3.3 Reviewing and synthesising evidence 1
Evidence relating to clinical effectiveness was reviewed and synthesised according to the Grading of 2
Recommendations Assessment, Development and Evaluation (GRADE) approach (see 3
http://www.gradeworkinggroup.org/index.htm). In the GRADE approach, the quality of the evidence 4
identified for each outcome listed in the review protocol is assessed according to the factors listed 5
below, and an overall quality rating (high, moderate, low, or very low) is assigned by combining the 6
ratings for the individual factors. 7
Study design (as an indicator of intrinsic bias; this determines the initial quality rating) 8
Limitations in the design or execution of the study including concealment of allocation, 9
blinding, loss to follow up (these can reduce the quality rating) 10
Inconsistency of effects across studies: occurs when there is variability in the treatment 11
effect demonstrated across studies (heterogeneity) (this can reduce the quality rating) 12
Indirectness: the extent to which the available evidence fails to address the specific 13
review question (this can reduce the quality rating) 14
Imprecision: present when there is uncertainty around the estimate of effect, for 15
example when the confidence intervals are wide or the sample size or event rate is low 16
(this can reduce the quality rating) 17
Other considerations including large magnitude of effect, evidence of a dose-response 18
relationship, or confounding variables likely to have reduced the magnitude of an effect 19
(these can increase the quality rating in observational studies, provided no downgrading 20
for other features has occurred) 21
The type of review question determines the highest level of evidence that may be sought. For issues 22
of therapy or treatment, the highest possible evidence level is a well-conducted systematic review or 23
meta-analysis of randomised controlled trials (RCTs), or an individual RCT. In the GRADE approach, 24
a body of evidence based entirely on such studies has an initial quality rating of high, and this may be 25
downgraded to moderate, low, or very low if factors listed above are not addressed adequately. For 26
issues of prognosis, the highest possible level of evidence is a controlled observational study (a 27
cohort study or casecontrol study), and a body of evidence based on such studies would have an 28
initial quality rating of low, which might be downgraded to very low or upgraded to moderate or high, 29
depending on the factors listed above. 30
For each review question the highest available level of evidence was sought. Where appropriate, for 31
example, if a systematic review of RCTs or individual RCTs were identified to answer a question 32
directly, studies of a weaker design were not considered. Where systematic reviews of RCTs or RCTs 33
were not identified other appropriate experimental or observational studies were sought. For the 34
priority outcome of womens experience of care and psychological outcomes qualitative studies were 35
sought where appropriate. For diagnostic tests, test evaluation studies examining the performance of 36
the test were used if the accuracy of the test was required, but where an evaluation of the 37
effectiveness of the test in the clinical management of the condition was required, evidence from 38
RCTs or cohort studies was optimal. For studies evaluating the accuracy of a diagnostic test, 39
sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV) and likelihood 40
ratios for positive and negative test results (LR+ and LR, respectively), were calculated or quoted 41
where possible (see Table 3.1). 42
The GRADE system described above covers studies of treatment effectiveness. However, it is less 43
well established for studies reporting accuracy of diagnostic tests. For such studies, NICE 44
recommends using the Quality Assessment of Studies of Diagnostic Accuracy (QUADAS) 45
methodology checklist to assess study quality (see the NICE guidelines manual, 2009). 46
Some studies were excluded from the guideline reviews after obtaining copies of the corresponding 47
publications because they did not meet inclusion criteria specified by the GDG (see Appendix G). The 48
characteristics of each included study were summarised in evidence tables for each review question 49
(see Appendix H). Where possible, dichotomous outcomes were presented as risk ratios (RRs) or 50
Guideline development methodology
29
odds ratios (ORs) with 95% confidence intervals (CIs), and continuous outcomes were presented as 1
mean differences with 95% CIs or standard deviations (SDs). 2
The body of evidence identified for each review question (or part of a review question) was presented 3
in the form of a GRADE evidence profile summarising the quality of the evidence and the findings 4
(pooled relative and absolute effect sizes and associated CIs). Summary GRADE tables have been 5
reported in the main text, with the full GRADE evidence profiles reported in Appendix I. Where 6
possible, the body of evidence corresponding to each outcome specified in the review protocol was 7
subjected to quantitative meta-analysis. In such cases, pooled effect sizes were presented as pooled 8
risk ratios (RRs), pooled odds ratios (ORs), or weighted mean differences. By default, meta-analyses 9
were conducted by fitting fixed effects models, but where statistically significant heterogeneity was 10
identified random effects models were used. Where quantitative meta-analysis could not be 11
undertaken (for example, because of heterogeneity in the included studies or where studies were not 12
RCTs) the range of effect sizes reported in the included studies was presented. 13
Table 3.1 2 x 2 table for calculation of diagnostic accuracy parameters 14
Reference standard
positive
Reference standard
negative
Total
Index test result
positive
a (true positive) b (false positive) a+b
Index test result
negative
c (false negative) d (true negative) c+d
Total a+c b+d a+b+c+d = N (total
number of tests in study)
15
3.4 Emotional and psychological support 16
A key component of care for women with pain and bleeding in early pregnancy is the provision of 17
emotional and psychological support. In order to fully address this issue when considering the 18
evidence and drafting recommendations the GDG configuration included a consultant psychiatrist as 19
well as two lay members representing womens views and experiences of miscarriage and ectopic 20
pregnancy. 21
A specific systematic review was undertaken to investigate the most appropriate form of psychological 22
support to offer women with pain and/or bleeding in early pregnancy. In addition, and in recognition of 23
the fact that the provision of emotional and psychological support is important at each stage of the 24
care pathway, the GDG prioritised womens experience of care and psychological outcomes as key 25
outcomes for the reviews conducted on early pregnancy assessment units, management of 26
miscarriage and management of ectopic pregnancy. Evidence from qualitative studies was searched 27
for where appropriate in order to adequately address these outcomes. An additional qualitative review 28
was conducted to look at service organisation and delivery within early pregnancy assessment units 29
and how this might impact on quality of care including womens experience of care. 30
The scope of this guideline comprises diagnosis and initial management of women presenting with 31
pain and bleeding in the first trimester of pregnancy. The GDG felt that emotional and psychological 32
support for women was best addressed throughout the care pathway so as to underline that it is vital 33
at each stage. This decision was based on the fact that, in their experience, when care is provided in 34
a way that is supportive and sensitive to womens needs the psychological sequelae of pregnancy 35
loss can be lessened. Insensitive care during the acute phase of initial management can be 36
particularly hurtful and distressing and the GDG wished to address this. In addition, it was felt that the 37
recognition of which women would need interventions aimed at providing additional psychological 38
support after a first trimester pregnancy loss would come after this initial phase and it was appropriate 39
to consider providing this care later on, after the miscarriage or treatment of an ectopic pregnancy had 40
been completed. 41
30
The clinical evidence reviewed for management of miscarriage was based on RCT data. In addition 1
the review incorporated the qualitative data analysis that accompanied 3 of these trials. This data 2
describes womens experience of care and their emotional response to early pregnancy loss and its 3
management. These findings are reported in separate summary tables within the relevant chapter and 4
quotations used to illustrate the key themes identified (see tables 7.3 and 7.5). 5
In order to make explicit the GDGs consideration of womens psychological support when making 6
recommendations a separate sub-section has been created within the evidence to recommendations 7
section for each review. Summaries here will link the evidence from the systematic review, the GDGs 8
interpretation of the evidence and their own clinical experience with recommendations aimed at 9
improving psychological support for women. 10
Using these methods recommendations for emotional and psychological support will be integral to the 11
guideline, featuring in each chapter as appropriate in addition to the specific chapter where this is 12
addressed directly. In this way the GDG wanted to underline that this essential component of care 13
needs to be provided for all women alongside clinical care and treatment in a way that is appropriate 14
to each womans circumstances. 15
Recommendations aimed at improving womens emotional and psychological support during the initial 16
phases of care for pain and bleeding in early pregnancy include the provision of comprehensive 17
information, both verbal and written, so that women know what to expect during the process of their 18
care and what symptoms should prompt urgent return, written provision of a contact telephone 19
number and details of where a women can access additional ongoing support. Training in sensitive 20
communication and breaking bad news for staff working in a dedicated early pregnancy service is 21
recommended. There is also a recommendation to ensure that sufficient time is spent in consultations 22
so that issues can be discussed and provision made for an additional appointment to ensure this can 23
be done adequately. In addition, all women presenting with pain and bleeding in early pregnancy 24
should be offered the opportunity to return for a follow-up appointment in either primary or secondary 25
care, whichever the woman prefers. The guideline recommends that a dedicated early pregnancy 26
service should be available for women with pain and bleeding in early pregnancy as the GDG 27
believed this provision would improve both clinical and psychological care for women and would mean 28
women with early pregnancy complications would no longer need to be cared for on general 29
gynaecology wards or antenatal wards. The guideline also recommends that women who have had a 30
previous ectopic pregnancy should be able to self-refer directly into a dedicated early pregnancy 31
service in order to provide these women with the reassurance of quick, easy access to a specialist 32
service. Similarly, provision is made for women with previous adverse or traumatic experience 33
associated with pregnancy to be able to opt for active management of miscarriage if that is their 34
preference. The importance of womens choice is also highlighted throughout the guideline including a 35
choice of method of management of miscarriage when expectant management is not successful, a 36
choice of abdominal ultrasound where tranvaginal ultrasound is not acceptable and the choice of 37
surgical treatment for ectopic pregnancy where this is preferred over medical treatment. 38
A summary of the recommendations for information-giving and psychological support appears 39
towards the beginning of the care pathway in order to highlight its significance, simplify the pathway 40
and to avoid repetition. However, symbols are used throughout the care pathway to illustrate where 41
further specific information should be provided in order to highlight that this is essential throughout all 42
stages of care. 43
3.5 Incorporating health economics 44
The aims of the health economic input to the guideline were to inform the GDG of potential economic 45
issues relating to pain and bleeding in early pregnancy, and to ensure that recommendations 46
represented a cost effective use of healthcare resources. Health economic evaluations aim to 47
integrate data on benefits (ideally in terms of quality adjusted life years [QALYs]), harms and costs of 48
different care options. 49
The GDG prioritised a number of review questions where it was thought that economic considerations 50
would be particularly important in formulating recommendations. Systematic searches for published 51
economic evidence were undertaken for these questions. For economic evaluations, no standard 52
system of grading the quality of evidence exists and included papers were assessed using a quality 53
Guideline development methodology
31
assessment checklist based on good practice in economic evaluation. Reviews of the (very limited) 1
relevant published health economic literature are presented alongside the clinical effectiveness 2
reviews. 3
Health economic considerations were aided by original economic analysis undertaken as part of the 4
development process. For this guideline the areas prioritised for economic analysis were as follows: 5
Expectant compared with active management of miscarriage 6
Management of ectopic pregnancy 7
Progesterone for treatment of threatened miscarriage 8
Effectiveness of early pregnancy assessment units (EPAUs) 9
It should be noted that, due to a lack of relevant health economic literature and absence of clinical 10
effectiveness data, it was not possible to undertake economic analysis to determine the cost 11
effectiveness of EPAUs. 12
3.6 Evidence to recommendations 13
For each review question recommendations for clinical care were derived using, and linked explicitly 14
to, the evidence that supported them. In the first instance, the technical team drafted and the GDG 15
agreed short clinical and, where appropriate, cost effectiveness evidence statements which were 16
presented alongside the evidence profiles. Statements summarising the GDGs interpretation of the 17
evidence and any extrapolation from the evidence used to form recommendations were also prepared 18
to ensure transparency in the decision-making process. The criteria used in moving from evidence to 19
recommendations were as follows: 20
Relative value placed on the outcomes considered 21
Consideration of the clinical benefits and harms 22
Consideration of net health benefits and resource use 23
Quality of the evidence 24
Information giving and psychological support 25
Other considerations (including equalities issues) 26
In areas where no substantial clinical research evidence was identified, the GDG considered other 27
evidence-based guidelines and consensus statements or used their collective experience to identify 28
good practice. The health economics justification in areas of the guideline where the use of NHS 29
resources (interventions) was considered was based on GDG consensus in relation to the likely cost 30
effectiveness implications of the recommendations. The GDG also identified areas where evidence to 31
answer their review questions was lacking and used this information to formulate recommendations 32
for future research. 33
Towards the end of the guideline development process formal consensus methods were used to 34
consider all the clinical care recommendations and research recommendations that had been drafted 35
previously. The GDG identified 8 key priorities for implementation (key recommendations) and five 36
high-priority research recommendations. The key priorities for implementation were those 37
recommendations thought likely to have the biggest impact on the care of women with pain and 38
bleeding in early pregnancy and outcomes in the NHS as a whole; they were selected using two 39
rounds of anonymous voting amongst the GDG members. In the first round of voting each member 40
was asked to cast 10 votes and the five recommendations that received six or more votes were 41
promoted to become key priorities for implementation. A second round of voting was carried out for all 42
recommendations that received between three and five votes in the first round. Each GDG member 43
was asked to cast five votes. A further three recommendations received six or more votes in the 44
second round and were added to the list of key priority recommendations. The priority research 45
recommendations were selected in a similar way, with one round of voting leading to selection of five 46
key priority recommendations for research. 47
32
3.7 Stakeholder involvement 1
Registered stakeholder organisations were invited to comment on the draft scope and the draft 2
guideline. 3
[There will be more text to add here following stakeholder consultation] 4


4 Psychological support 1
4.1 Introduction 2
Becoming pregnant carries considerable psychological as well as physical and social significance. A 3
miscarriage is a physical loss but it also represents the loss of a potential relationship and, for some, 4
the loss of a potential new role. Many women will adjust without distress but some experience 5
considerable anxiety and sadness akin to grief following other important losses. In a minority the 6
miscarriage may precipitate psychological disorder such as anxiety or depression. Pain and bleeding 7
in early pregnancy may be also be distressing if it brings anxiety about the health and viability of the 8
pregnancy even if it does not end with miscarriage. In addition the experience of the early pregnancy 9
loss, especially if sudden or life-threatening, may generate symptoms associated with traumatic 10
stress. This is particularly relevant following ectopic pregnancy. 11
Pregnancy loss is not just about physical recovery and being ready to become pregnant again. The 12
GDG consider that good care includes sensitivity to the psychological impact of miscarriage. Even if 13
women have no extra psychological morbidity following the miscarriage they do express views about 14
what constitutes good care and preferences about how the condition should be managed to enhance 15
their recovery. 16
The GDG looked at the literature to determine if there were acceptable studies of interventions to 17
improve womens psychological adjustment after early pregnancy loss. The group also looked at 18
qualitative work which reported womens preferences about their care. 19
4.2 Psychological support 20
Review question 21
What interventions are the most effective for improving womens psychological and/or emotional 22
health following pain and/or bleeding in the first trimester of pregnancy? 23
Description of included studies 24
Eight studies are included in this review (Adolfsson et al., 2006; Lee et al., 1996; Neugebauer et al., 25
2006; Nikcevic et al., 1998; Nikcevic et al., 2007; Sjourn et al., 2010; Swanson, 1999; Swanson et 26
al., 2009). Six studies are randomised controlled trials (Adolfsson et al., 2006; Lee et al., 1996; 27
Neugebauer et al., 2006; Nikcevic et al., 2007; Swanson, 1999; Swanson et al., 2009), one study is a 28
quasi-randomised controlled trial (Sjourn et al., 2010) and one is an observational study (Nikcevic 29
et al., 1998). 30
Three studies were conducted in the USA (Neugebauer et al., 2006; Swanson, 1999; Swanson et al., 31
2009), three studies in the UK (Lee et al., 1996; Nikcevic et al., 1998; Nikcevic et al., 2007), one study 32
in France (Sjourn et al., 2010) and one in Sweden (Adolfsson et al., 2006). 33
The GDG prioritised anxiety and depression as key outcomes of interest. The interventions included 34
in the studies varied. A brief summary is given here (for further details see evidence table in Appendix 35
H). Three studies (Lee et al., 1996; Nikcevic et al., 2007; Sjourn et al., 2010) compared a single 36
psychological counselling session with no psychological counselling session and reported on the 37
psychological outcomes of anxiety and depression, and womens views/experiences of care. 38
Counselling sessions were 50 minutes with a psychologist 5 weeks post-miscarriage (Nikcevic et al., 39
2007), 60 minutes with a psychologist 2 weeks post-miscarriage (Lee et al., 1996) and a counselling 40
session (mean duration 37 minutes, SD = 14.38) on the day of surgical treatment for the 41
uncomplicated and unanticipated loss of pregnancy (Sjourn et al., 2010). The professional 42
providing the counselling session was not described in this study. 43
34
One study (Swanson 1999) compared three 1-hour counselling sessions at 1, 5 and 11 weeks after 1
enrolment in the study (mean time from miscarriage to study enrolment = 7.86 days, SD = 7.5) with no 2
counselling sessions and reported on the outcomes of anxiety and depression. Counselling sessions 3
were with the principal investigator or research associate (unclear whether midwives, nurses, 4
psychologists/psychiatrists, social workers). 5
One study (Neugebauer et al., 2006) compared a maximum of six counselling sessions with a 6
psychiatric social worker or psychiatrist with treatment as usual, which consisted of whatever lay 7
counselling or professional care women sought on their own initiative. Women receiving the 8
intervention could also seek other lay or professional care. The first counselling session was 9
scheduled for 60 minutes, the subsequent sessions were scheduled for 30 minutes. Women 10
concluded treatment whenever they wished; two women received six sessions and no others had 11
more than three sessions. 12
One study (Adolfsson et al., 2006) compared a single 60-minute structured follow-up visit with a 13
designated midwife with a single 30-minute standard follow-up with one of five midwives between 21 14
and 28 days post-miscarriage. The structured visit focussed specifically on emotional health issues. 15
One study (Swanson et al., 2009) compared nurse care (three 1-hour counselling sessions with a 16
nurse counsellor), self-care (three 18 minute videos featuring couples being coached in ways to 17
practice self and partner caring, plus two workbooks, one for the woman and one for her partner), 18
combined care (one 1-hour counselling session with a nurse counsellor followed by the self care 19
intervention) or control (no treatment). 20
One study (Nikcevic et al., 1998) surveyed women who had received a follow-up appointment with 21
their local hospital or general practitioner and women who were not offered a follow-up appointment. 22
Findings are presented below for each type of intervention. 23
One counselling session compared with no counselling session 24
Three studies (Lee et al., 1996; Nikcevic et al., 2007; Sjourn et al., 2010) compared a single 25
psychological counselling session with no psychological counselling session and reported on the 26
psychological outcomes of anxiety and depression, and womens views/experiences of care. 27
36% of women in the study by Nikcevic et al. 2007, 23% of women in Sjourn et al., 2010 and 0% of 28
women in Lee et al. 1996 had a history of miscarriage. It was unclear how many of the women with a 29
history of miscarriage had experienced recurrent miscarriage. The majority of women in all three 30
studies had children (51%, 56% and 68%). 31
Quantitative data for key outcomes are presented in a GRADE profile followed by descriptive findings 32
in a short summary table. 33
Table 4.1 GRADE findings for comparison of one psychological counselling session with no psychological 34
counselling session 35
Number of
studies
Number of women Effect Quality
One psychological
counselling
session
Mean (SD)
No psychological
counselling
Mean (SD)
Absolute
(95% CI)
Anxiety at 3 weeks (measured with: Hospital Anxiety and Depression Scale)
1 study
(Sjourn et al.,
2010)
Mean 7.21
(SD 3.02)
n = 50
Mean 9.06
(SD 3.95)
n = 52
MD 1.85 lower
(3.21 lower to 0.49
lower)
Moderate
1 study (Nikcevic Mean 7.2 Mean 6.7 MD 0.5 higher High
35
et al, 2007) (SD 5.2)
n = 33
(SD 4.1)
n = 33
(1.76 lower to 2.76
higher)
1 study
(Sjourn et al.,
2010)
Mean 6.22
(SD 3.52)
n = 45
Mean 7.16
(SD 4.25)
n = 37
MD 0.94 lower
(2.65 lower to 0.77
higher)
Moderate
Anxiety at 4 months (measured with: Hospital Anxiety and Depression Scale)
2 studies
(Lee et al., 1996,
Nikcevic et al.,
2007)
Mean 7.4
(SD 5.9)
n = 21

Mean 5.6
(SD 4.5)
n = 33
Mean 8.1
(SD 6.2)
n = 18

Mean 7
(SD 4.4)
n = 33
MD 1.23 lower
(3.1 lower to 0.64
higher)
High
Anxiety at 6 months (measured with: Hospital Anxiety and Depression Scale)
1 study
(Sjourn et al.,
2010)
Mean 5.33
(SD 3.42)
n = 33
Mean 6.5
(SD 3.49)
n = 34
MD 1.17 lower
(2.82 lower to 0.48
higher)
Moderate
Depression at 3 weeks (measured with: Hospital Anxiety and Depression Scale)
1 study
(Sjourn et al.,
2010)
Mean 3.93
(SD 3.38)
n = 50
Mean 5.08
(SD 3.6)
n = 52
MD 1.15 lower
(2.5 lower to 0.2
higher)
Moderate
1 study
(Nikcevic et al,
2007)
Mean 4.1
(SD 4.2)
n = 33
Mean 3.34
(SD 2.9)
n = 33
MD 0.7 higher
(1.04 lower to 2.44
higher)
High
1 study
(Sjourn et al.,
2010)
Mean 3.0
(SD 2.46)
n = 45
Mean 3.48
(SD 3.2)
n = 37
MD 0.48 lower
(1.74 lower to 0.78
higher)
Moderate
Depression at 4 months (measured with: Hospital Anxiety and Depression Scale)
2 studies
(Lee et al., 1996,
Nikcevic et al.,
2007)
Mean 3.2
(SD 4.2)
n = 21

Mean 2.8
(SD 4.1)
n = 33
Mean 4.8
(SD 7)
n = 18

Mean 3.7
(SD 3.7)
n= 33
MD 1.04 lower
(2.72 lower to 0.63
higher)
High
36
Depression at 6 months (measured with: Hospital Anxiety and Depression Scale)
1study
(Sjourn et al.,
2010)
Mean 2.24
(SD 2.79)
n = 33
Mean 2.44
(SD 2.5)
n = 34
MD 0.2 lower
(1.47 lower to 1.07
higher)
Moderate
Hospital Anxiety and Depression Scale: better = lower values [score of 11 was threshold for caseness in Nikcevic et al., 2007 1
and Lee et al., 1996] 2
3
Table 4.2 Womens views and experiences of care following one counselling session compared with no 4
Womens views/experiences of care
Nikcevic et al., 2007
[Moderate quality]
Helpfulness of 20-minute medical visit with obstetrician 5 weeks
post-miscarriage (received by women in both intervention and control
groups): 100% of women reported a moderate to strong agreement
regarding helpfulness of the medical consultation.
In a cohort of matched controls who received no medical consultation
and no psychological counselling ,30/47 (64%) women expressed that
some follow-up would have been helpful.
Helpfulness of 50-minute psychological counselling session 5
weeks post miscarriage: 94% of women agreed the psychological
counselling was at least moderately helpful
Sjourn et al., 2010
[Moderate quality]
Helpfulness of one psychological counselling session: at 3 weeks
post-miscarriage 43/50 (86%) women felt that the intervention was
helpful
Need for more support: at 10 weeks post-miscarriage 9/45 (20%)
women felt that the psychological counselling was insufficient and felt the
need for more support
Lee et al. 1996
[Moderate quality]
Helpfulness of 1 hour psychological counselling session: women
were asked to rate the helpfulness of the counselling session on a
100mm scale from extremely unhelpful (0) to extremely helpful (100):
mean = 74mm, SD = 21.1, n = 21
Positive comments on psychological counselling session: having
opportunity to express feelings and thoughts; having someone to talk to
who listened to them
Negative comments on psychological counselling session: having to
relive experience; limited medical knowledge of person carrying out the
debriefing.
Medical explanation: women commented that they would have liked a
more medical explanation for their miscarriage, as well as emotional
support
6
Evidence statements 7
Anxiety 8
One study found that anxiety at 3 weeks was lower in women who received one psychological 9
counselling session within 4 days of miscarriage, compared with women who did not receive 10
psychological counselling. This finding was statistically significant. The evidence for this finding was of 11
moderate quality. 12
37
Three studies measuring anxiety at different time points following miscarriage found no statistically 1
significant difference in anxiety in women who received one session of psychological counselling 2
compared with women who did not receive psychological counselling at 7 weeks (high quality), 10 3
weeks (moderate quality), 4 months (high quality) and 6 months (moderate quality). 4
Depression 5
Three studies measuring depression at different time points following miscarriage found no 6
statistically significant difference in depression in women who received one session of psychological 7
counselling compared with women who did not receive psychological counselling at 3 weeks 8
(moderate quality), 7 weeks (high quality), 10 weeks (moderate quality), 4 months (high quality) and 6 9
months (moderate quality). 10
Womens views/experiences of care 11
Helpfulness of psychological counselling 12
One study found that women moderately or strongly felt that a 20-minute counselling session with an 13
obstetrician at 5 weeks post-miscarriage was helpful. In the same study 94% of women who received 14
the additional 50-minute psychological counselling session felt that the psychological counselling was 15
at least moderately helpful. The evidence for this finding was of moderate quality. 16
One study found that, at 3 weeks post-miscarriage, 86% of women felt that a 1-hour psychological 17
counselling session on the day of surgical treatment of miscarriage was helpful. The evidence for this 18
finding was of moderate quality. 19
One study found that women rated the helpfulness of a 1-hour psychological counselling session at a 20
mean of 74mm on a visual analogue scale from extremely unhelpful (0mm) to extremely helpful 21
(100mm). Women found it helpful to have the opportunity to express feelings and thoughts, having 22
someone to talk to and listen to them. The evidence for this finding was of moderate quality. 23
Need for more support 24
One study found that 20% of women who received a 1-hour psychological counselling session felt 25
that the counselling was insufficient and felt the need for more support. The evidence for this finding 26
was of moderate quality. 27
Need for a medical explanation 28
One study reported that some women commented on the limited medical knowledge of the person 29
delivering the psychological counselling session and that women would have liked a more medical 30
explanation for their miscarriage, as well as emotional support. The evidence for this finding was of 31
Three counselling sessions compared with no counselling sessions 33
One study (Swanson 1999) compared three 1-hour counselling sessions at 1, 5 and 11 weeks after 34
enrolment in the study (mean time from miscarriage to study enrolment = 7.86 days, SD = 7.5) with no 35
counselling sessions and reported on the outcomes of anxiety and depression. 36
30% of women in this study (Swanson 1999) had a history of miscarriage. It was unclear how many of 37
the women with a history of miscarriage experienced recurrent miscarriage. The majority of women in 38
the study had children (54%). 39
This study used a Solomon four-group randomised experimental design: some women were 40
randomised to delayed outcome measurement to address the possibility that early outcome 41
measurement (survey completion) could in itself serve as a form of treatment. The data reported in 42
the below GRADE profile represents those women randomised to early outcome measurement as this 43
is the group for whom the most complete data set is available. 44
Table 4.3 GRADE findings for comparison of three psychological counselling sessions with no psychological 45
Number of
studies
Three counselling No psychological
Absolute
38
sessions counselling (95% CI)
Anxiety at 6 weeks (measured with: Profile of Mood States)
1 study
(Swanson, 1999)
Mean 10
(SD 5.4)
n = 43
Mean 11.5
(SD 7.3)
n = 40
MD 1.5 lower
(4.28 lower to 1.28
higher)
High
Anxiety at 4 months (measured with: Profile of Mood States)
1 study
(Swanson, 1999)
Mean 10.9
(SD 6.8)
n= 43
Mean 11.0
(SD 7.3)
n = 40
MD 0.1 lower
(3.14 lower to 2.94
higher)
High
Anxiety at 12 months (measured with: Profile of Mood States)
1 study
(Swanson, 1999)
Mean 8.7
(SD 5.6)
n = 43
Mean 9.3
(SD 7.3)
n = 40
MD 0.6 lower
(3.41 lower to 2.21
higher)
High
Depression at 6 weeks (measured with: Profile of Mood States)
1 study
(Swanson, 1999)
Mean 12.1
(SD 11)
n = 43
Mean 14.8
(SD 12.7)
n = 40
MD 2.7 lower
(7.83 lower to 2.43
higher)
High
Depression at 4 months (measured with: Profile of Mood States)
1 study
(Swanson, 1999)
Mean 9.8
(SD 8.7)
n = 43
Mean 12.6
(SD 13.7)
n = 40
MD 2.8 lower
(7.78 lower to 2.18
higher)
High
Depression at 12 months (measured with: Profile of Mood States)
1 study
(Swanson, 1999)
Mean 8.4
(SD 9.3)
n = 43
Mean 11.4
(SD 14.5)
n = 40
MD 3 lower
(8.28 lower to 2.28
higher)
High
Profile of Mood States scale: better = lower values 1
2
Anxiety 4
One study measuring anxiety at different time points following miscarriage found no statistically 5
significant difference in anxiety in women who received three sessions of psychological counselling 6
compared with women who did not receive psychological counselling at 6 weeks, 4 months and 12 7
months post-miscarriage. The evidence for these findings was high quality. 8
Depression 9
One study measuring depression at different time points following miscarriage found no statistically 10
significant difference in depression in women who received three sessions of psychological 11
counselling compared with women who did not receive psychological counselling at 6 weeks, 4 12
months and 12 months post-miscarriage. The evidence for these findings was high quality. 13
39
Up to six counselling sessions compared with treatment as usual 1
One study (Neugebauer et al., 2006) compared a maximum of six counselling sessions with a 2
psychiatric social worker or psychotherapist with treatment as usual, and reported on the outcome of 3
depression. 4
37% of women in this study (Neugebauer et al., 2006) had a history of miscarriage. It was unclear 5
how many of the women with a history of miscarriage experienced recurrent miscarriage. The number 6
of women with children was not reported. 7
Table 4.4 GRADE findings for comparison of a maximum of six psychological counselling sessions with 8
treatment as usual 9
Number of
studies
Maximum of six
counselling
sessions
Treatment as usual
Absolute
(95% CI)
Depression at 9 weeks (measured with: Hamilton Rating Scale for Depression-17 item)
1 study
(Neugebauer et
al., 2006)
Mean 11.6
(SD 8.2)
n = 10
Mean 12.9
(SD 8.3)
n = 9
MD 1.3 lower
(8.73 lower to 6.13
higher)
Very low
Hamilton Rating Scale for Depression (HAM-D-17): better = lower values 10
11
Depression 13
One study found no statistically significant difference in depression at 9 weeks post-miscarriage in 14
women who received up to a maximum of six sessions of psychological counselling compared with 15
women who received treatment as usual. The evidence for these findings was very low quality. 16
Structured midwifery follow-up compared with standard midwifery 17
follow-up 18
One study (Adolfsson et al., 2006) compared a single 60-minute structured follow-up visit with a 19
designated midwife with a single 30-minute standard follow-up with one of five midwives between 21 20
and 28 post-miscarriage and reported on the outcome of grief. Although not a priority outcome this 21
study is included as it is the only study found that investigated the effectiveness of an informal 22
intervention delivered by a nurse/midwife. 23
22% of women in this study (Adolfsson et al., 2006) had a history of miscarriage. It was unclear how 24
many of the women with a history of miscarriage experienced recurrent miscarriage. The number of 25
women with children was not reported. 26
Table 4.5 GRADE findings for comparison of structured midwifery follow up with standard midwifery follow up 27
Number of
studies
Structured follow
up reduction
change score (95%
CI)
Standard follow up
reduction change
score (95% CI)
Absolute
(95% CI)
Grief at 4 months (measured with: Perinatal Grief Scale Swedish short version)
1 study
(Adolfsson et al.,
23.5
(11.6 to 35.5)
17.5
(7.7 to 27.3)
No significant
difference
Low
40
2006) n = 43 n = 45 P = 0.43
Perinatal Grief Scale (total score): better = lower values 1
2
Grief 4
One study found no statistically significant difference in grief at 4 months post-miscarriage in women 5
who attended a structured follow-up visit with a midwife and women who attended a standard follow- 6
up visit with a midwife. The evidence for this finding was low quality. 7
Nurse caring, self-caring and combined caring interventions and no 8
treatment 9
One study (Swanson et al., 2009) compared nurse care (three 1-hour counselling sessions with a 10
nurse counsellor), self-care (three 18 minute videos featuring couples being coached in ways to 11
practice self and partner caring, plus two workbooks, one for the woman and one for her partner), 12
combined care (one 1-hour counselling session with a nurse counsellor followed by the self care 13
intervention) or control (no treatment). Three hundred and forty one couples were included in the 14
study (n= 682). 15
The number of women with a history of miscarriage and the number of women with children was not 16
reported. 17
Women in all three treatment groups exhibited a faster rate of recovery, measured with the Center for 18
Epidemiological Studies-Depression scale (CES-D), compared with women receiving no treatment. 19
However, only the nurse caring group met the authors criterion for substantial evidence (Bayesian 20
odds ratio > 3.2). The self-caring group had the highest proportion of women who did not return data 21
after baseline (25/172, 14.5%); the nurse caring group had the lowest proportion (1/168, 0.6%). 22
The study reported change in CES-D scores at 3, 5 and 13 months post-miscarriage (data was 23
extracted from a small graph and so numbers are not accurate). The changes in CES-D scores at 3 24
months were as follows: nurse caring 2.9, self-caring 2.3, combined caring 2.3 and no 25
treatment 2.2. The change in CES-D score at 5 months was: nurse caring 5.7, self-caring 26
4.9, combined caring 4.7 and no treatment 4.3. The changes in CES-D score at 13 months 27
were: nurse caring 8.2, self-caring 7.1, combined caring 6.9 and no treatment 6.2. The 28
evidence for this finding was of high quality. 29
One follow-up appointment compared with no follow-up appointment 30
One study (Nikcevic et al., 1998) surveyed women who had received a follow-up appointment with 31
their local hospital or general practitioner and women who were not offered a follow-up appointment. 32
The timing of the follow-up appointment and outcome measurement post-miscarriage was not 33
reported in the study. The healthcare professionals delivering the follow-up appointment were not 34
clearly reported. Findings from the study are summarised in the table below. 35
Table 4.6 Womens experience of care with one follow-up appointment compared with no follow-up appointment 36
Womens views/experiences of care
Nikcevic et al., 1998
[Low quality]
Desire for follow-up, offer of and attendance at follow-up: 187/204
(92%) women thought a follow-up appointment was desirable. Such an
appointment, with a local hospital or general practitioner, was offered to
61/204 (30%) women. 52/61 (85%) women attended the follow-up
appointment
Content of follow-up: 22/52 (42%) reported not being given the
opportunity to discuss feelings during the follow-up
Expectations from a follow-up clinic: 72% of women suggested clinic
should be conducted by a doctor, 28% would have preferred to see a
41
midwife or counsellor. 177/204 (87%) women reported it was 'very' or
'extremely' important to them to have an explanation as to why the
miscarriage happened.
Contact with the Miscarriage Association: prior to discharge from the
Harris Birthright Research Centre all women were given an information
leaflet that included the telephone number of the Miscarriage Association.
18/204 (9%) women had made contact, significantly more so in the group
that attended a follow-up clinic.
Emotional counselling: 73/204 (36%) women reported that they would
find emotional counselling helpful. The comparison between women who
expressed a wish for emotional counselling and those who did not
revealed that those who did not want counselling had significantly lower
levels of anxiety (t test = 2.44, d.f. = 200, P < 0.05), depression (t test =
2.51, d.f. = 200, P < 0.05) and grief (t test = 4.30, d.f. = 199, P <
0.001).
Women's opinions about ways to improve support from medical
professionals: many women wanted more information concerning the
reasons for their miscarriage and its implications, outlined the importance
of a sensitive and sympathetic attitude on the part of medical
professionals and emphasised the fact that the evacuation of the retained
products of conception after miscarriage is a trauma that is too often
dismissed as a routine surgical procedure by the medical staff involved.
1
Evidence to recommendations 2
The GDG had prioritised anxiety and depression as key outcomes for the review. However, when the 4
evidence was presented it was seen that for the majority of the studies the levels of anxiety and 5
depression being reported were very low and generally within a range considered normal. In addition, 6
the small differences in scores demonstrated between experimental and control groups were felt not 7
to be clinically significant. In contrast, findings from the qualitative studies were thought to be more 8
helpful and more credible. Thus womens reports of what they found helpful, their experience of care 9
and reports of what they would have liked were seen by the GDG as the most important findings from 10
this review. 11
Consideration of clinical benefits and harms 12
The provision of additional counselling sessions for women with pain and bleeding in early pregnancy 13
appears not to bestow any clinical benefits as measured by anxiety and depression scales. It seems 14
from the scores reported in the evidence reviewed that women with pain and bleeding in early 15
pregnancy are not suffering from high levels of anxiety or depression as represented by the 16
consistently low scores seen across studies, so perhaps it is not surprising that additional 17
psychological support has little impact on these parameters. The GDG recognised that the same may 18
well not be true for women with recurrent pregnancy loss or with pain and bleeding at a later stage of 19
pregnancy. The GDG also noted that the number and content of the counselling sessions did not 20
appear to make a difference to their effect on anxiety and depression, neither did the timing of such 21
sessions (although a general decrease in scores over time was noted) nor the professional training of 22
the person delivering the counselling sessions. 23
In contrast it was evident from the descriptive evidence that women who had received a follow-up 24
appointment valued this, and many women who had not had a follow-up appointment would have 25
liked one. Many women expressed a need for an explanation for the pregnancy loss and a preference 26
for a medical follow-up in order to discuss this. The importance of being treated in a sensitive and 27
sympathetic manner through all stages of care was also emphasised. The GDG cross-referred here to 28
NICE clinical guideline 138 Patient experiences in adult NHS services (2012) which makes a series of 29
recommendations on the importance of good communication and an understanding of patients 30
responses to events and the care provided. 31
42
The clinical benefits of additional psychological support for women with pain and bleeding in early 1
pregnancy were felt to outweigh any potential harms in that it was felt that women tend to self-select 2
in this respect. Evidence from the literature reviewed and from the GDGs own experience was that 3
women who would benefit from additional support would express a preference for it, whilst those who 4
preferred not to discuss the experience with a healthcare professional would not request such 5
sessions, or would choose not to attend. 6
Consideration of health benefits and resource uses 7
Findings from the descriptive evidence suggests that for some women a follow-up appointment is 8
perceived as valuable, particularly the provision of an opportunity to discuss possible reasons for the 9
pregnancy loss and helping plan for the future. The GDG believed that not all women experiencing 10
pain and bleeding in the first trimester of pregnancy would want or need this follow up, however they 11
believed that the offer of a follow-up appointment in itself would have a beneficial effect, even for 12
those women who chose not to take it up and thus felt it appropriate that this offer be made to all 13
women. The GDGs experience was that a minority of women would actually make and attend a 14
follow-up appointment, with most women finding support from family and friends and/or through third 15
sector agencies. The natural healing afforded by time passing would also reduce the number of 16
women needing this additional appointment. It was thus felt that the number of women attending for 17
follow-up would be fairly small and manageable within the existing resource allocated to providing 18
care to women with pain and bleeding in early pregnancy. 19
Quality of evidence 20
The evidence for the review ranged from high to very low. The quality of the descriptive findings 21
relating to womens experiences of care and preferences was moderate and low. Despite the 22
limitations of these studies the GDG felt their evidence was valuable, brought an important 23
perspective to the review findings and had high face validity in that it reflected their own experience. 24
Other considerations 25
The GDG noted that the location of the follow-up visit would be likely to have a bearing on how it is 26
experienced by women. Returning to the place where the diagnosis of pregnancy loss was first made 27
may be upsetting for some women, as would attendance at a location along with pregnant women 28
e.g. an antenatal clinic. Similarly some women might have a good relationship with their GP and 29
prefer to see them for follow-up care. For this reason the GDG believed it important that women 30
should be provided with a choice as to where to attend for a follow up appointment. 31
The GDG were disappointed to note that there was very little evidence on psychological support for 32
women following ectopic pregnancy, with the focus of the evidence base being on miscarriage. There 33
was concern that, whilst for many women the experience may be similar between the two types of 34
early pregnancy loss, any potential differences e.g. relating to perceptions of future fertility, have not 35
been explored in the literature. It was felt that an improved understanding of the differences between 36
womens experiences of miscarriage and ectopic pregnancy would help carers provide more 37
appropriate individualised support and a research recommendation was made to encourage work in 38
this area. 39
Recommendations 40
Number Recommendation
1 Treat all women with pain or bleeding in early pregnancy with dignity and respect.
Be aware that women will react to complications or the loss of a pregnancy in
different ways. Provide all women with information and support in a sensitive
manner, taking into account their individual circumstances and emotional
response.

2 Healthcare professionals providing care for women with early pregnancy
complications in any setting should be aware that early pregnancy complications
can cause significant distress for some women and their partners. Healthcare

For further guidance about providing information, see NICE clinical guideline 138 Patient experience in adult NHS services
improving the experience of care for people using adult NHS services (2012)
43
professionals providing care for these women should be given training in sensitive
3 After an early pregnancy loss, offer the woman the option of a follow-up
appointment with a healthcare professional in either primary or secondary care
according to the womans preference.
1
2
Number Research recommendations
RR 1 What is the psychological impact of the different types of early pregnancy loss?
3


5 Early pregnancy 1
assessment units 2
5.1 Introduction 3
Management of early pregnancy complications is often undertaken in Early Pregnancy Assessment 4
Units (EPAUs). EPAUs were established in the early 1990s and their location, opening times, and 5
accessibility vary throughout England and Wales. Where there is not an EPAU, women with early 6
pregnancy complications may be cared for within primary or secondary care. Comparisons of the 7
different models of care were sought to establish the clinical and cost effectiveness of EPAUs. In 8
addition, specific models of early pregnancy care were sought to determine if any particular model led 9
to better outcomes for women. 10
5.2 Clinical and cost effectiveness of early pregnancy 11
assessment units 12
Review question 13
What is the clinical and cost effectiveness of early pregnancy assessment units (EPAUs) in improving 14
womens clinical and psychological outcomes? 15
Four studies were included in this review (Bignardi et al., 2010; Bigrigg et al., 1991; Brownlea et al., 17
2055; Tunde-Byass & Cheung, 2009). 18
One UK retrospective study (Bigrigg et al., 1991) assessed the impact of an EPAU on duration of stay 19
in women with pain and bleeding in early pregnancy. Data were collected six months before the unit 20
opened and compared with data collected during the first year of the units operation. 21
One Australian retrospective study (Brownlea et al., 2005) examined the hypothesis that the 22
introduction of the Early Pregnancy Problem Service (EPPS) clinic reduced the length of stay in the 23
emergency department for women with early pregnancy pain and bleeding that did not require 24
hospital admission. The EPPS was established in June 1996. Data were collected from 2 months 25
before the establishment of the EPPS and compared with two months in 2003, seven years after the 26
EPPS was established. 27
One Australian prospective study (Bignardi et al., 2010) evaluated the impact of an ultrasound based 28
unit (Acute Gynaecology Unit [AGU]) in the management of women with acute gynaecology and early 29
pregnancy complications. The AGU provided ultrasound investigation to women at the initial visit by 30
the same person who took the history, and the ultrasound findings were interpreted in the context of 31
the clinical picture. 32
One Canadian retrospective study (Tunde-Byass & Cheung, 2009) assessed the impact of an early 33
pregnancy assessment clinic (EPAC) on the management of early pregnancy complications and its 34
effect on the number of emergency room visits. The EPAC was established in August 2005. Data 35
were collected from one year before the establishment of the EPAC and compared with two years 36
after the EPAC was established. 37
45
Evidence profile 1
Table 5.1 GRADE summary of findings for comparison of before and after the opening of an EPAU/AGU 2
Number of
studies
Before
EPAU/AGU
opened
After
EPAU/AGU
opened
Relative
(95% CI)
Absolute
(95% CI)
Length of stay in emergency department (minutes)
1 study
(Brownlea et
al. 2005)
Median 136 Median 107 Not calculable
(NC)
Median 29
higher (CI NC)
p < 0.001
Low
Re-presentation to emergency department
1 study
(Brownlea et
al. 2005)
14/87
(16%)
6/85
(7%)
2.28
(0.92 to 5.65)
90 more per
1000
(from 6 fewer to
328 more)*
Very low
1 study
(Tunde-Byass,
& Cheung,
2009)
431/1514
(28.5%)
384/1603
(24%)
1.19
(1.06 to 1.34)
46 more per
1000
(from 14 more
to 81 more)
Very low
Duration of stay for women requiring no treatment (hours)
1 study
(Bigrigg et al.,
1991)
Mean 36
(range 12 to 72
days)
Mean 2 NC MD 34 higher
(CI NC)
Low
Duration of hospital stay for women requiring evacuation of the uterus (hours)
1 study
(Bigrigg et al.,
1991)
Mean 72
(range 36 to 60)
Mean 24 NC MD 48 higher
(CI NC)
Low
Length of stay as inpatient (hours)
1 study
(Bignardi et al.,
2010)
Mean 13.9 Mean 4.6 NC MD 9.3 higher
(CI NC)
p = 0.011
Very low
Length of stay as outpatient (hours)
1 study
(Bignardi et al.,
2010)
Mean 4.1 Mean 0.75 NC MD 3.35 higher
(CI NC)
p = 0.0001
Very low
Proportion of women requiring hospital admission
1 study
(Brownlea et
al., 2005)
37/88
(42%)
29/81
(36%)
1.17
(0.80 to 1.72)*
61 more per
1000
(from 72 fewer
to 258 more)*
Very low
1 study
(Bignardi et al.,
48/133
(36.1%)
11/157
(7%)
5.15
(2.79 to 9.51)
291 more per
1000
Very low
46
2010) (from 125 more
to 596 more)*
Proportion of women who re-presented to emergency department with further pain and bleeding
1 study
(Brownlea et
al., 2005)
14/88
(16%)
6/81
(7%)
2.14
(0.90 to 5.21)*
84 more per
1000
(from 7 fewer to
312 more)*
p = 0.6
Very low
Proportion of women discharged within 3 hours from emergency department
1 study
(Brownlea et
al., 2005)
30/51
(60%)
44/52
(86%)
0.69
(0.25 to 0.88)*
262 fewer per
1000
(from102 fewer
to 635 fewer)*
p < 0.0001
Very low
* NCC calculation 1
Length of stay in emergency department 3
One study found a longer median length of stay in the emergency department before the 4
establishment of an EPAU compared with afterwards. This finding was statistically significant. The 5
evidence for this outcome was of low quality. 6
Re-presentation to emergency department 7
One study found no statistically significant difference in the number of women requiring representation 8
to emergency department before the establishment of an EPAU compared with afterwards. However 9
another study found higher number of women requiring representation to emergency department 10
before the establishment of an EPAU compared with afterwards. This finding was statistically 11
significant. The evidence for this outcome was of very low quality. 12
Duration of stay for women requiring no treatment 13
One study found a longer mean duration of hospital stay before the establishment of an EPAU 14
compared with afterwards for women presenting with pain and bleeding who required no treatment. 15
The statistical significance of this finding was not calculable. The evidence for this outcome was of 16
low quality. 17
Duration of hospital stay for women requiring evacuation of the uterus 18
One study found a longer mean duration of hospital stay before the establishment of an EPAU 19
compared with afterwards for women presenting with pain and bleeding who required evacuation of 20
the uterus. The statistical significance of this finding was not calculable. The evidence for this 21
outcome was of low quality. 22
Length of stay as inpatient 23
One study found a longer mean length of stay as inpatient before the establishment of an AGU 24
compared with afterwards for women presenting with pain and bleeding. This finding was statistically 25
Length of stay as outpatient 27
One study found a longer mean length of stay as outpatient before the establishment of an AGU 28
compared with afterwards for women presenting with pain and bleeding. This finding was statistically 29
Proportion of women requiring hospital admission 31
One study found no statistically significant difference in the number of women requiring hospital 32
admission for treatment of pain and bleeding in early pregnancy before the establishment of an EPAU 33
47
compared with afterwards. However another study found higher number of women requiring hospital 1
admission for treatment of pain and bleeding in early pregnancy before the establishment of an AGU 2
compared with afterwards. This finding was statistically significant. The evidence for this outcome was 3
of very low quality. 4
Proportion of women who re-presented to emergency department with further 5
pain and bleeding 6
One study found no statistically significant difference in the proportion of women who represented to 7
an emergency department with further pain and bleeding following initial treatment for pain and 8
bleeding in early pregnancy before the establishment of an EPAU compared with afterwards. The 9
evidence for this outcome was of very low quality. 10
Proportion of women discharged within 3 hours from emergency department 11
One study found a lower proportion of women who were discharged within 3 hours of being seen by a 12
doctor for pain and bleeding in early pregnancy before the establishment of an EPAU compared with 13
afterwards. This finding was statistically significant. The evidence for this outcome was of very low 14
quality. 15
The GDG had hoped to find studies that compared womens experiences of care in EPAUs with those 18
in other settings; unfortunately this evidence was not available. Similarly, there was no evidence on 19
whether women managed in an EPAU have better clinical or psychological outcomes than women 20
managed elsewhere. Therefore, the GDG based their decisions on the outcomes reported in studies 21
that are likely to impact on the cost of providing services in early pregnancy, namely the need for 22
hospital admission and the length of hospital stay. 23
The GDG noted that the introduction of an EPAU or AGU had resulted in a shorter length of stay in 25
both the emergency department and the outpatient clinic, a reduction in the proportion of women 26
requiring hospital admission, and a reduction in the number of women re-presenting to health 27
services. The GDG felt that these would be important considerations for women, in addition to having 28
potential cost implications as a result of a reduced inpatient stay. Despite the paucity of data, the 29
GDG were aware of considerable anecdotal evidence that women have better experiences of care in 30
EPAUs, in comparison to other models of care, such as care based on antenatal wards. They also felt 31
that it was likely that access to a dedicated early pregnancy service, as is provided in EPAUs, might 32
lead to better clinical outcomes for women. However, they recognised that research was needed to 33
confirm this hypothesis. 34
In light of the lack of clinical outcomes reported in the evidence, a cost-effectiveness analysis could 36
not be performed for this review question. A literature search was performed for health economic 37
evidence, which did not identify any relevant studies. Due to the potential cost savings from reduced 38
admissions and length of stay, the GDG felt that it was likely that an EPAU would be more cost- 39
effective than an inpatient service. However, they recognised that the cost of staffing a dedicated 40
service and the number of women attending the units would have implications for cost. Therefore, 41
they felt that it was important to recommend that health economics analysis be performed as soon as 42
clinical outcome data becomes available. 43
The quality of evidence was low, and there was a regrettable lack of evidence surrounding clinical 45
outcomes, womens experiences of care and cost-effectiveness. However, the GDG agreed that it 46
was important that all women should have access to some sort of dedicated early pregnancy 47
assessment service, which would provide ultrasound scanning and specialist clinical assessment for 48
women in early pregnancy. The GDG did not anticipate that this early pregnancy assessment service 49
would always be based in an EPAU, and did not feel able to recommend that this should be the case, 50
in light of the lack of evidence on their effectiveness. However, the currently operating EPAUs offer 51
this dedicated specialist service and it is generally thought that they improve womens experiences; 52
48
therefore, the GDG felt that it was vital to recommend that research be conducted to evaluate whether 1
EPAUs improve outcomes, both clinical and psychological, for women experiencing pain and bleeding 2
in early pregnancy. 3
The GDG believed that the availability of a dedicated service would minimise the risk of misdiagnosis 5
and help to reduce the anxiety that women with an early pregnancy loss might experience, especially 6
compared with attending services based in settings like antenatal units. In comparison with other 7
models of care, the group felt that staff in a dedicated service, who specialise in early pregnancy 8
issues, would be able to provide women with appropriate information and psychological support, 9
which might help to mitigate some of their anxieties and concerns. They agreed that staff providing 10
care in these settings should be appropriately trained in sensitive communication and breaking bad 11
news. They also felt that recommending that women could access this service promptly, where 12
clinically indicated, would help to ensure that women spent as little time as possible in a state of 13
uncertainty about their prognosis, and therefore would feel more informed and supported. 14
The GDG were aware that an EPAU may not be an appropriate model of care in every setting, for 16
example, very rural areas with low pregnancy rates. In light of this, and the lack of evidence on their 17
clinical and cost-effectiveness, the GDG did not feel able to recommend that EPAUs should be 18
established in areas currently lacking one. 19
5.3 Model for service organisation and delivery of 20
EPAUs 21
Review question 22
What is the appropriate model for service organisation and delivery of EPAUs? 23
Fourteen studies were included in this review (Akhter et al., 2007; Bignardi et al., 2010; Bigrigg & 25
Read, 1991; Brownlea et al., 2005; Davies & Geoghegan, 1994; Edey et al., 2007; Fox et al., 1999; 26
Harper 2003; Hill, 2009; Poddar et al., 2011; Sellapan et al., 2009; Shillito & Walker, 1997; Tunde- 27
Byass & Cheung, 2009; Twigg et al., 2002). 28
The majority of the included studies were conducted in the UK, with the exception of one from Ireland 29
(Akhter et al., 2007), one from Canada (Tunde-Byass & Cheung, 2009) and two from Australia 30
(Bignardi et al., 2010; Brownlea et al., 2005). 31
Of the included studies, four are observational studies that compared outcomes before and after 32
establishment of an early pregnancy assessment unit or clinic (EPAU or EPAC) or acute gynaecology 33
unit (AGU) (Bignardi et al., 2010; Bigrigg & Read, 1991; Brownlea et al., 2005; Tunde-Byass & 34
Cheung, 2009). Two of the included studies are cross-sectional studies, conducting using a postal 35
survey of EPAUs (Poddar et al., 2011; Twigg et al., 2002). The remaining included studies are 36
descriptive, non-comparative papers, detailing the experiences and model of care in a single unit. 37
Evidence profile 38
This review aimed to establish how different models of care within EPAUs might impact on service 39
outcomes, clinical outcomes and womens experiences of care. Due to the nature of the evidence that 40
was available for this review question, the findings are presented in a modified evidence profile, split 41
by study design. All of the evidence is of very low quality. 42
43
49
Table 5.2 Findings for service organisation and delivery of EPAUs
Study Characteristics of model of care Outcomes
Staffing Referral systems Availability of out-of-hours care Service outcomes Womens views and
experiences of care
Cross-sectional data from surveys of early pregnancy assessment units (EPAUs)
Poddar et
al., 2011
Practitioner performing
ultrasound (n/total (%))
Sonographers: 67/140
(47.9%)
EPAU nurse specialist:
12/140 (8.6%)
Trained midwife: 7/140
(5%)
Medical staff: 2/140 (1.4%)
Combination: 52/140
(37.1%)
Direct referral system for
women (n/total (%))
a. With previous EP:
125/140 (89%)
b. With recurrent
miscarriage: 113/140 (81%)
Availability of service in clinics (n/total (%))
Weekday:
3-5 hours each weekday: 47/135 (34.8%)
6-11 hours each weekday: 74/135 (54.8%)
3 days a week: 1/135 (0.7%)
2 hours a day: 1/135 (0.7%)
Mean opening time/hours: 7.3 3.6
Median (range) opening time/hours: 8 (2 -
24)

Weekend:
Full or partial weekend service: 42/140 (30%)
- Open Saturday and Sunday: 21/140 (15%)
- Open Saturday: 11/140 (7.9%)
- Open Sunday: 8/140 (5.7%)
Inconsistent weekend service: 2/140 (1.4%)

Availability of 24 hour contact telephone
None reported None reported
50
(n/total (%))
a. For women receiving conservative/medical
miscarriage management: 103/140 (74%)
b. For women receiving methotrexate for
ectopic pregnancy: 99/125 (79%)
Twigg et
al., 2002
Practitioner performing
ultrasound (%)
Ultrasonographer: (52.0)
Radiologist: (2.0)
Gynaecologist: (11.8)
Gynaecology nurse: (4.9)
Other: (2.9)
Midwife: (2.9)
Combination: (23.5)

54.9% of units said that
their scanning practitioners
had formal training in
breaking bad news.
51.5% of clinics said that
all patients were seen by a
gynaecologist.
95.8% said that they
received adequate
gynaecological back-up.

Proportion of clinics
accepting women by each
referral method (%)
Referral from other clinicians
and general practitioners
(GPs): 100
Direct from patients: 51
Other (e.g. midwives,
Accident and Emergency
[A&E], gynaecology): 21
Availability of service in clinics (%)
Weekday only: 77.4
Seven-day: 13.7
Once per week: 1
24-hour: 7
51
Descriptive data from individual EPAUs
Akhter et
al., 2007
Senior sonographer, junior
doctor and dedicated
counselling midwife.
(Consultant input is
available for complicated
cases)
502/603 (83%) of women
were self-referred.
The remainder were referred
by their GP or the A&E
department of other
hospitals
No.
The clinic is open Monday to Friday 7.30 am
to 10 am.
Number of patients seen
650 women attended during the
study period of approximately 6
months
Waiting time/hours (range):
1 3
Need for a repeat scan (n (%)):
121 (20%)
None reported
Bignardi et
al., 2010
A gynaecological
consultant is in charge of
the unit on a day-to-day
basis. History-taking,
clinical examination and
transvaginal ultrasound
are all undertaken by the
consultant.
Referrals must be made by
another practitioner (walk-
ins are not permitted). The
majority are referred by their
GP or the emergency
department.
No.
The clinic is open 9 am 1 pm Monday to
Friday.
Waiting time/minutes (mean)
a. To see trainee gynaecologist:
172
b. For ultrasound exam: 199
Length of stay/minutes (mean)
a. as an outpatient: 45
b. as an inpatient: 274
Admission rate (n (%))
a. Total: 11 (7)
b. For ultrasound: 4 (2.5)
None reported
Bigrigg &
Read,
1991
Women are seen by the
duty senior house office
(SHO), but a registrar and
consultant are available
on-site if needed.
GPs Unclear.
Open 7 days a week, with a limited on-call
system. Out-of-hours operating is avoided.
Number of women seen
In the first year, 771 women were
referred to the unit.
Length of stay/days
a. Maximum: 1.5
b. For women with viable IUP or not
pregnant: 0.08 (2 hours)
None reported
52
c. For women needing evacuation
of uterus: 1
Need for a repeat scan (%): 11
Brownlea
et al.,
2005
Women are reviewed by
an obstetrics and
gynaecology registrar who
performs transvaginal
ultrasound.
Referrals are received from
both the emergency
department (ED) and GPs.
Referrals from a non-ED
source (%)
Year 1: 26
Year 7: 48
Unclear.
The paper reports that referred patients are
reviewed on a weekday morning.
Number of patients seen in Jan-
Feb of years following
establishment of EPPS (n)
Year 1: 15
Year 7: 61
Proportion of patients
discharged from ED followed up
in EPPS (n/total (%))
Year 1: 11/54 (20%)
Year 7: 36/52 (69%)
Proportion of EPP patients re-
presenting to ED with further
pain and/or bleeding (n/total (%))
Year 1: 12/95 (13%)
Year 4: 12/82 (15%)
Year 7: 6/81 (7%)
Proportion of EPP patients
requiring hospital admission
(n/total (%))
Year 1: 41/95 (43%)
Year 4: 28/82 (34%)
Year 7: 29/81 (36%)
(p = 0.6 for trend)
None reported
53
Davies &
Geoghega
n, 1994
Nurse-led unit, with a team
of ward clerks, doctors,
scan stenographers and
phlebotomists.
Following the scan, a
registrar compares the
scan with the patients
history and makes a
diagnosis.
GPs No details given None reported None reported
Edey et
al., 2007
An audit of the unit found
that only 29% of the
women needed to be seen
by the junior doctor in the
clinic, with the rest being
managed by the
sonographer and nurse
practitioner.
Source of referrals (%)
GPs: 40
A&E: 2
(No further details given)
Unclear.
The clinic is open daily, but no further details
are given. It is unclear whether this includes
weekends or not.
Fox et al.,
1999
Out of 198 women, 120
(61%) were managed by a
nurse only. 78 (39%)
required medical
assessment.
Midwives or GPs No.
The clinic is open 5 days a week.
In 198 cases (100%), a nurse
practitioner made the correct initial
diagnosis in her assessment of the
womans condition.
None reported
Harper,
2003
Midwife provides care,
with later referral to
medical personnel if
needed.
Self-referrals are accepted,
as well as referrals from
miscarriage assessment
clinic, antenatal clinic, GPs,
A&E, team-based midwives.
Unclear.
Women are provided with a 24-hour telephone
advice number
Hill, 2009 Clinical nurse specialist
(runs service), two
ultrasonographers and an
on-call registrar when
Source of referrals (n/total
(%))
GP: 96/230 (42%)
No.
The clinic is open weekday mornings.
Patients seen on time (n/total
(%))
Yes: 217/237 (92%)
Women reported being
seen on time (n (%))
Yes: 79 (96%)
54
required EPAC: 44/230 (19%)
Consultant: 27/230 (12%)
A&E: 24/230 (10%)
Midwife: 14/230 (6%)
SHO: 9/230 (4%)
Antenatal: 8/230 (3%)
Registrar: 3/230 (1%)
Re-scan: 3/230 (1%)
Jas: 2/230 (0.9%)
No: 1/237 (0.4%)
Not stated: 12/237 (5%)
N/A: 2/237 (0.8%)
Did not attend: 5/237 (2%)
Acceptable wait for appointment
referral (n (%))
Yes: 161 (68%)
No: 1 (0%)
Probable rescans or further
treatment: 75 (32%)
Number of patients seen:
82 over a two-month period
No: 3 (4%)
Women felt wait for
appointment was
acceptable (n (%))
Yes: 76 (94%)
No: 5 (6%)
Women felt care in
scanning department was
given in a sensitive
manner (n (%))
Yes: 80 (99%)
No: 1 (1%)
Sonographer explained
results in a way that
women could understand
(n (%))
Yes: 81 (99%)
No: 1 (1%)
Women felt they were
given a thorough
explanation (n (%))
Yes: 81 (99%)
No: 1 (1%)
Women felt questions
were answered in a way
they could understand (n
(%))
Yes: 80 (98%)
No: 2 (2%)
Womens satisfaction with
interaction with different
55
staff (n/total (%))
a. Receptionist
Excellent: 27/63 (43%)
Good: 30/63 (48%)
Fair: 5/63 (8%)
Poor: 1/63 (2%)
b. EPAC Nurse Specialist
Good: 6/82 (7%)
Fair: 0/82
Poor: 0/82
c. Sonographers
Good: 14/81 (17%)
Fair: 1/81 (1%)
Poor: 0/81
d. Doctors
Good: 9/22 (41%)
Fair: 0/22
Poor: 0/22
Womens satisfaction with
privacy, dignity and care
(n/total (%))
a. Privacy
Good: 14/80 (18%)
Fair: 1/80 (1%)
56
Poor: 0/80 (0%)
b. Dignity
Good: 11/80 (14%)
Fair: 0/80 (0%)
Poor: 0/80 (0%)
c. Care
Good: 11/80 (14%)
Fair: 0/80 (0%)
Poor: 0/80 (0%)
Sellapan
et al.,
2009
Out of 198 women, 125
(66.5%) were managed by
midwives only. 45 (23.9%)
were managed by medical
staff.
(%))
GP: 90/188 (47.8%)
Emergency department:
17/188 (9%)
Self-referral: 31/188 (16.5%)
No details given Waiting time/minutes (n)
a. Up to 30 minutes: 95
b. Up to 60 minutes: 55
c. More than 60 minutes: 18
Average waiting time/minutes: 11
Need for a repeat scan: 25/188
(13.3%)
None reported
Shillito &
Walker,
1997
No details given Most referrals come through
GPs or A&E
No.
The clinic is open Monday to Friday from 8 am
to 12.30 pm; however staff deal with
telephone enquiries until 8 pm.
Workload per week (average): 30
Time of discharge (%)
Same day: 89
- Immediately: 80
- After same-day evacuation: 9
In a survey of 100 women,
over half wanted to see a
specialist nurse and less
than 10% expected to see a
doctor.
Tunde-
Byass &
Cheung,
Team of dedicated
gynaecologists and
experienced obstetrical
(%))
No.
The clinic is open three mornings per week
Number of women requiring
repeat ER assessment (n/total
None reported
57

2009 nurses. Gynaecologists
perform the ultrasound
scans.
Emergency room (ER):
557/1448 (38.5%)
Family physician: 445/1448
(30.7%)
Obstetrician-gynaecologist:
349/1448 (24.1%)
Midwife: 30/1448 (2.1%)
Other: 67/1448 (4.6%)
from 9 am to 12 noon. (%)): 738/3062 (24.1%)
58
Cross-sectional data from surveys of EPAUs 2
Staffing 3
Two surveys of EPAUs found that, in approximately half of units, scans were performed by 4
ultrasonographers. Less than a quarter of the clinics reported the scanning was done by either a 5
radiologist, gynaecologist, nurse, midwife or other practitioner. 6
Referral systems 7
One survey of EPAUs found that all units accepted referrals from other clinicians and general 8
practitioners (GPs), whereas only 51% of units accepted self-referrals. Another survey found that over 9
80% of units had direct referrals systems in place for women with a history of ectopic pregnancy or 10
recurrent miscarriage. 11
Availability of out-of-hours care 12
One survey of EPAUs (published 2002) found that 77% of clinics provided a weekday service only, 13
14% provided a seven-day service, 1% of clinics only provided a weekly service, and 7% provided 24 14
hour care. In a later survey (published 2011) 30% of units reported offering a full or partial weekend 15
service. The duration of opening hours during the week ranged from 2 to 24 hours. 16
Descriptive data from individual EPAUs 17
Staffing 18
Eleven studies reported details about the staffing structure of their unit, and described a variety of 19
models of care. Four studies reported that management, including ultrasound scanning, was 20
performed primarily by clinicians, whereas the remainder described a team approach, with varying 21
levels of input and on-site availability of clinicians. Three studies reported the proportion of women 22
seen by each type of practitioner, and found that 24% to 39% of women required assessment by a 23
doctor, with the remainder being successfully managed by nurses, midwives and ultrasonographers. 24
In one nurse-led unit, it was reported that 100% of women were correctly diagnosed by the nurse 25
practitioner in her initial assessment. 26
Referral systems 27
Twelve studies reported some details of their units referral systems, of which only three reported 28
accepting self-referrals. The majority of referrals were reported to be from GPs and emergency 29
departments. 30
Availability of out-of-hours care 31
Ten studies reported some detail about the units opening hours and availability of out-of-hours care. 32
In six of these, it was clear that out-of-hours care was not available, whereas in the remaining papers, 33
the details provided were not sufficient to establish availability. 34
Service outcomes 35
Five studies reported the number of women seen in the unit, which ranged from about 30 per month 36
to over 100. 37
Four studies reported waiting time, of which one reported that 92% of women were seen on time (no 38
criteria were reported). The average waiting time reported by the remaining units was between 11 39
minutes and 3 hours. 40
Two studies reported the length of stay, which ranged from 45 minutes for an outpatient, to a 41
maximum of over a day. The admission rate ranged from 7% in one study to over 30% in another. 42
One further study reported that 89% of women were discharged on the same day as presentation, 43
80% immediately and 9% after a same-day evacuation. 44
Three studies reported the number of women requiring a repeat ultrasound, which ranged from 11% 45
to 20%. Two studies reported the proportion of women re-presenting at the emergency department, 46
which ranged from 7% to 24%. 47
59
Womens views and experiences of care 1
One study found that, of women attending an EPAU, over half wanted to see a specialist nurse and 2
less than 10% expected to see a doctor. A second study assessed womens satisfaction in their 3
interaction with different staff, and found that women ranked their interaction with the nurse specialist 4
as excellent in 93% of cases, compared to 81% for the sonographer, 59% for the doctor and 43% for 5
the receptionist. 6
The same study found that over 80% of women rated their satisfaction with privacy, dignity and care 7
as excellent. The study reported that 99% of women felt that scanning was done in a sensitive 8
manner and the results were explained in a way that they could understand. Similarly, 99% of women 9
felt that they were given a thorough explanation and 98% felt that their questions were answered in a 10
way that they could understand. 11
The GDG wanted to evaluate whether the model of service organisation within an EPAU affected 14
womens clinical outcomes and experiences, and service outcomes. In particular, they felt that the 15
staffing structure, the ability of women to self-refer, and the accessibility of the service might affect 16
outcomes such as the length of hospital stay and need for admission, which are important 17
considerations for resource use. 18
There was a severe paucity of data linking aspects of the service organisation within the EPAU to 20
outcomes; therefore, the GDG felt that a research recommendation was warranted to audit models of 21
service delivery within EPAUs and link it to outcomes, both for women and for the service. 22
The evidence showed that EPAUs currently operate a number of different staffing models, ranging 23
from medical consultant-led to a team-based approach with varying levels of input by clinicians. 24
However, there was insufficient evidence to associate staffing structures with outcomes, and it was 25
recognised that any recommendations on personnel would have significant cost implications. 26
Therefore, the group did not feel able to recommend that units adopt a specific staffing model without 27
further evidence. 28
From their clinical experience working in EPAUs, the GDG decided that allowing all women to self- 29
refer to the units was not conducive to maximising quality of care. Whilst they understood that some 30
women may wish to attend for reassurance or for a scan early in a pregnancy that is progressing 31
normally, they felt that this was not an appropriate use of the resources of EPAUs, which are intended 32
for women experiencing a miscarriage or ectopic pregnancy. They felt that an excess of self-referrals 33
was likely to result in reduced quality of care for women with early pregnancy problems, as a result of 34
services becoming overstretched. Therefore, the group felt that it was appropriate to recommend that 35
women go through a triaging process prior to being referred to an EPAU. It was anticipated that 36
triaging might be performed by GPs, A&E, or NHS 24. However, the GDG felt that it was reasonable 37
to make an exception for women with a history of ectopic pregnancy, due to the risk of a repeat 38
ectopic pregnancy. Therefore, they recommended that these women be allowed to present at an 39
EPAU without being referred. 40
The GDG felt that the majority of women attending EPAUs could be managed successfully and safely 41
if the service was available during weekdays. Their experience was that the uptake of out-of-hours 42
service is generally low, and due to a lack of evidence on its added value, they did not feel that it was 43
appropriate to recommend that EPAUs should remain open 24 hours a day, seven days a week. 44
However, they did think that it was important that women who were referred to an early pregnancy 45
assessment service be seen in a timely manner. The group felt that, if a womans symptoms and 46
clinical situation suggested the need for further investigation or management, it was reasonable to 47
recommend that women be able to access a dedicated early pregnancy service (not necessarily an 48
EPAU) within 24 hours. They discussed the issue of accessibility of the service at the weekend, and 49
agreed that for a lot of women, waiting until Monday for a dedicated early pregnancy service would be 50
a safe and preferable option. However, they also noted that more immediate investigation would be 51
indicated in some cases, and therefore recommended that these women be referred to the nearest 52
60
service that could provide ultrasound scanning and specialist clinical assessment, likely to be a 1
gynaecology ward. 2
The GDG felt that limiting the number of women able to self-refer to early pregnancy assessment 4
units would reduce the cost of running such services, by minimising the number of appointments 5
given to women for whom the service was not intended. They hoped that this would allow resources 6
to be focused on women with the greatest need, i.e. a threatened or actual miscarriage or ectopic 7
pregnancy. Similarly, they agreed that, without evidence of any association with improved clinical 8
outcomes, it was not appropriate to recommend that units expend resources on providing a 24 hour 9
service or adopting a specific staffing model. 10
The quality of evidence for this review was universally very low; therefore, the GDG only felt able to 12
make limited recommendations about how early pregnancy assessment services should be 13
organised, based on their own experiences. The group thought that it was vital that research be 14
conducted to elucidate the most appropriate model of service organisation and delivery, in order to 15
maximise the benefit to women and cost effectiveness of the service. 16
As discussed above, the GDG felt that recommending that the majority of women be triaged before 18
referral to a dedicated early pregnancy service would improve care for women attending the service. 19
In particular, reducing the number of women attending the units unnecessarily should help to ensure 20
that healthcare professionals have more time and resources to counsel and support women 21
undergoing the potentially traumatic experience of early pregnancy loss. However, the group also 22
noted that for women with a previous ectopic pregnancy, the prospect of having another undiagnosed 23
ectopic pregnancy might cause significant fear and psychological problems. Therefore, they felt that it 24
was appropriate to enable these women to self-refer to early pregnancy services. 25
The group felt that women who have difficulty accessing health care, for example those with English 27
as a second language or travellers, merited extra consideration. However, after some discussion, the 28
GDG decided that allowing self-referrals would not be an effective way to remove the barriers to 29
access experienced by these groups due to the potential for over-burdening the EPAU with women 30
presenting without early pregnancy complications. It was decided that these women should be 31
assessed initially by either a GP or a health professional in an accident and emergency department. 32
Recommendations 33
4 A dedicated early pregnancy assessment service should be available at least during
Monday to Friday for women with pain and/or bleeding in early pregnancy, where
scanning can be carried out and decisions about management made.
5 Dedicated early pregnancy assessment services should accept self-referrals from
women who have had a previous ectopic or molar pregnancy. All other women with
pain and/or bleeding should be assessed by a healthcare professional before
referral to an early pregnancy assessment service.
6 Ensure that a system is in place to enable, women referred to their local dedicated
early pregnancy assessment service to attend it within 24 hours if the clinical
situation warrants it. If the service is not available at weekends, and the clinical
symptoms warrant further investigation, refer women to the nearest accessible
facility that offers specialist clinical assessment and ultrasound scanning.
34
61
Number Research recommendation
RR 2 A national evaluation of early pregnancy care service provision should be undertaken
with statistical analysis to identify factors affecting outcomes.

The first report of an early pregnancy assessment unit in England was published over
20 years ago, and prompted the rapid development of centres for the management of
problems in early pregnancy. Today there are an estimated 150 early pregnancy
assessment units in England and Wales (Association of early pregnancy units, 2012).
However, there is considerable variation between centres in access to services and
levels of care provided. In addition, there has been very little good quality research on
the effectiveness of early pregnancy assessment units in improving physical and
emotional health.
A national audit of early pregnancy assessment units would help to make up for this
lack of information. Such an audit should be along the lines of the National
Caesarean Section Sentinel Audit, a cross-sectional national survey of service
configuration and outcomes. Data recorded would include service location, opening
hours, and the healthcare professionals involved. Outcomes would include time of
attendance, length of stay, admission rates, time to treatment and patient experience.
Obtaining some of this information would require units to undertake more formal
follow-up of patients than they may do currently, for the duration of the audit. The
evaluation should be structured to allow for comparisons between different models of
care.
Comparative outcome data collected would be used to conduct an analysis of the
cost effectiveness of early pregnancy assessment units.


6 Diagnosis of ectopic 1
pregnancy and 2
miscarriage 3
6.1 Signs and symptoms of ectopic pregnancy 4
Review question 5
What are the signs and symptoms associated with ectopic pregnancy? 6
Introduction 7
Ectopic pregnancy is a relatively common and potentially life threatening complication of pregnancy. 8
Despite this, morbidity and mortality attributable to failure to consider the diagnosis, and therefore 9
missed or delayed diagnosis continues to be problematic. This is often due to misconceptions and 10
ignorance of symptoms and signs of an ectopic pregnancy. This review seeks to clarify the relative 11
importance of these individual factors in diagnosing an ectopic pregnancy. 12
Twenty-nine studies were included in this review (Aboud & Chaliha, 1998; Al-Suleiman & Khwaja, 14
1992; Banerjee et al., 1999; Barnhart et al., 2003; Barnhart et al., 2006; Bouyer et al., 2002; Buckley 15
et al., 1998; Choi et al., 2011; Clancy & Illingworth, 1989; Condous et al., 2007; Diamond et al., 1994; 16
Dimitry, 1989; Downey & Zun, 2011; Easley et al., 1987; Goksedef et al., 2011; Gonzalez & Waxman, 17
1981; Hutton & Narayan, 1986; Jabbar & Al-Wakeel, 1980; Jiao et al., 2008; Kazandi & Turan, 2011; 18
Larrain et al., 2011; Makinen et al., 1984; Menon et al., 2007; Michelas et al., 1980; Powers, 1980; 19
Raziel et al., 2004; Shaunik et al., 2011; Tsai et al., 1995; Wong & Suat, 2000). 20
The included studies consist of one case-control study (Barnhart et al., 2006) and four prospective 21
observational studies (Banerjee et al., 1999; Buckley et al., 1998; Condous et al., 2007; Shaunik et 22
al., 2011), whilst the remainder were retrospective case series. 23
The studies were conducted in the UK (Aboud & Chaliha, 1998; Banerjee et al., 1999; Clancy & 24
Illingworth, 1989; Condous et al., 2007; Dimitry, 1989), the USA (Barnhart et al., 2003; Barnhart et al., 25
2006; Buckley et al., 1998; Diamond et al., 1994; Downey & Zun, 2011; Easley et al., 1987; Gonzalez 26
& Waxman, 1981; Menon et al., 2007; Powers, 1980; Shaunik et al., 2011), France (Bouyer et al., 27
2002; Larrain et al., 2011;), Finland (Makinen et al., 1984), Greece (Michelas et al., 1980), Turkey 28
(Goksedef et al., 2011; Kazandi & Turan, 2011), New Zealand (Hutton & Narayan, 1986), China (Jiao 29
et al., 2008; Tsai et al., 1995), Singapore (Wong & Suat, 2000), South Korea (Choi et al., 2011), Israel 30
(Raziel et al., 2004), and Saudi Arabia (Al-Suleiman & Khwaja, 1992; Jabbar & Al-Wakeel, 1980). 31
All studies evaluated the presenting signs and symptoms of women diagnosed with an ectopic 32
pregnancy (EP), or reported the frequency of risk factors among the participants. Two studies 33
included only cases of ovarian pregnancy (Choi et al., 2011; Raziel et al., 2004), one study only 34
included cases of proximal ectopic pregnancy (Larrain et al., 2011) and one study included only cases 35
of caesarean scar pregnancy (Jiao et al., 2008). Three studies evaluated the signs and symptoms of 36
women initially classified as having pregnancies of unknown location (PUL) who were later diagnosed 37
with an ectopic pregnancy (Banerjee et al., 1999; Condous et al., 2007; Shaunik et al., 2011). 38
Diagnosis of ectopic pregnancy and miscarriage
63
Evidence profile 1
When more than one study reported a risk factor, symptom or sign, the reported % is the median 2
frequency derived from the reported frequency in each study. The range of frequencies given is 3
simply the minimum and maximum frequency as reported in the studies that included that particular 4
risk factor, symptom or sign. These have been presentd in order to provide some guidance as to 5
which risk factors, signs and symptoms are most frequently associated with ectopic pregnancy and to 6
illustrate the wide range of possible presenting symptoms and signs. 7
Evidence quality has been downgraded if the studies were retrospective, had a small sample size (N 8
50), or if five or fewer studies reported the finding and there was a serious issue of indirectness with 9
at least one of the study populations. Unusual study populations are detailed in the table. 10
Risk factors and symptoms are presented in order of decreasing frequency; signs are presented in 11
the order in which they might be elicited. 12
Table 6.1 GRADE summary of findings for the risk factors, symptoms, and signs of ectopic pregnancy 13
Risk factor, symptom,
or sign
Numbe
r of
studies
Number of
study
participants
Observed
frequency
[median %
(minimum
maximum)]
Other considerations
(e.g. study population)
Quality
Risk factors for ectopic pregnancy
Smoking 3 1990 48.1
(19 59.5)
Maximum reported in a
study of proximal ectopic
pregnancies
Low
No risk factors 6 691 37.2
(23.8 76.9)
study whose inclusion
criterion was presentation
with pain and/or bleeding
Moderate
Prior pelvic or
abdominal surgery
15 2963 22.5
(9.5 100)
study of 28 caesarean
scar pregnancies
(maximum is otherwise
48.8%)
Moderate
History of a sexually
transmitted infection
3 2498 19.7
(4.9 21.3)
Moderate
History of elective
abortion
7 1594 18.6
(14.9 47)
None Moderate
History of infertility 10 1367 16.9
(5.5 46.7)
study with 53% ruptured
EP (maximum is
otherwise 37.9%)
Moderate
History of miscarriage 7 3069 16.4
(11.3 33)
None Moderate
64
History of pelvic
inflammatory disease
18 3446 15.5
(4 44.4)
EP (maximum is
otherwise 33%)
Moderate
History of ectopic
pregnancy
16 4498 10.5
(2.2 27.9)
Minimum reported in a
EP (minimum is
otherwise 6%)
study of proximal ectopic
pregnancies (maximum is
otherwise 16.6%)
Moderate
History of IUCD use 18 5025 10.1
(2.2

68.4)
EP (minimum is
otherwise 4.1%)
study of 19 ovarian
otherwise 33.3%)
Moderate
Use of the oral
contraceptive pill
7 1302 6.7
(0.6 64.6)
population of 65 proximal
ectopic pregnancies
15.0%)
Moderate
Prior tubal surgery 11 3353 6
(3.9 18.7)
EP (minimum is
otherwise 4.8%)
Moderate
Endometriosis 5 671 4.7
(3.6 32.7)
study of 49 ovarian
otherwise 5%)
Low
Symptoms reported
Abdominal or pelvic
pain
21 3356 93.3
(42.9 100)
study of 49 ovarian
pregnancies (minimum is
otherwise 54.2%)
Moderate
Amenorrhea 11 2228 73
(8.2 98)
study of 49 ovarian
otherwise 64.3%)
EP (maximum is
otherwise 96.4%)
Moderate
Vaginal bleeding
with/without clots
25 3942 64.0
(14 82)
None Moderate
65
Breast tenderness 3 666 25.6
(8.9 31.7)
EP (minimum is
otherwise 25.6%)
Low
Gastro-intestinal
symptoms
10 1623 21.3
(2.0 48.3)
study of 49 ovarian
otherwise 15.5%)
Moderate
Dizziness/fainting/synco
pe
12 2334 21.1
(4.7 84.2)
study where the
denominator for this
finding was n=19
49%)
Moderate
Shoulder tip pain 7 818 19
(5 35.1)
None Moderate
Urinary symptoms 3 588 8.7
(5.9 23.3)
EP (minimum is
otherwise 8.7%)
Low
Asymptomatic 3 247 8.3
(2.9 18.4)
study of 49 ovarian
pregnancies
Low
Passage of tissue 2 143 7.6
(2.6 12.5)
Moderate
Rectal pressure or pain
on defecation
3 753 7
(6.7 9)
Median reported in a
EP
Low
Signs identified on examination
Shock or collapse 8 906 14.7
(2 23)
study of 49 ovarian
otherwise 10%)
Moderate
Pallor 1 97 44.3
(N/A)
None Moderate
Tachycardia or
hypotension
5 1077 21
(13.6 75.6)
EP (minimum is
otherwise 15%)
EP (maximum is
otherwise 23%)
Low
66
1
Orthostatic hypotension 3 580 4.6
(3.3 17.6)
None Moderate
Abdominal distension 2 259 36.8
(17.2 56.4)
EP
Low
Abdominal tenderness 11 1659 77.9
(7.4 91.2)
population of 27 EP
initially diagnosed as PUL
(minimum is otherwise
58.8%)
EP (maximum is
otherwise 91%)
Moderate
Rebound
tenderness/peritoneal
signs
9 1469 45.1
(23.1 86)
Moderate
Adnexal tenderness 7 821 82
(53.3 94.6)
EP (maximum is
otherwise 85.5%)
Moderate
Cervical motion
tenderness
8 1334 50.3
(33.3

86.5)
Moderate
Pelvic tenderness 1 177 91
(N/A)
None Moderate
Enlarged uterus 6 952 28.4
(15.4 36.7)
EP (minimum is
otherwise 17.6%)
Moderate
Adnexal mass 9 855 26.9
(5.1 55.7)
(minimum is otherwise
6.1%)
EP (maximum is
otherwise 49.5%)
Moderate
Palpable pelvic mass 2 242 15.7
(12.3 19)
None Moderate
67
Unless otherwise stated, all the evidence is of moderate quality. 2
Risk factors for ectopic pregnancy 3
Evidence from 6 studies showed that, on average, 37% of women with ectopic pregnancy had no risk 4
factors for ectopic pregnancy. 5
Evidence from 3 studies showed that, on average, 48% of women with ectopic pregnancy smoked 6
cigarettes (low quality). 7
Evidence from 15 studies showed that, on average, 23% of women with ectopic pregnancy had a 8
prior pelvic or abdominal surgery. 9
The evidence showed that 10 20% of women with ectopic pregnancy had a history of a sexually 10
transmitted infection (3 studies), a previous elective abortion (7 studies), a history of infertility (10 11
studies), a previous miscarriage (7 studies), a history of pelvic inflammatory disease (18 studies), a 12
previous ectopic pregnancy (16 studies), or a history of IUCD use (13 studies). 13
The evidence showed that less than 10% of women with ectopic pregnancy had a history of oral 14
contraceptive pill use (7 studies), prior tubal surgery (11 studies), or endometriosis (2 studies, low 15
quality). 16
Symptoms reported 17
The evidence showed that the majority of women with ectopic pregnancy presented with abdominal or 18
pelvic pain (93%, 21 studies), amenorrhea (73%, 11 studies) or vaginal bleeding (64%, 25 studies). 19
The evidence showed that 20 30% of women with ectopic pregnancy presented with breast 20
tenderness (3 studies, low quality), gastro-intestinal symptoms (10 studies), or dizziness, fainting or 21
syncope (12 studies). 22
The evidence showed that 10 20% of women with ectopic pregnancy presented with shoulder tip 23
pain (7 studies). 24
The evidence showed that less than 10% of women with ectopic pregnancy presented with urinary 25
symptoms (3 studies, low quality), passage of tissue (2 studies), rectal pressure or pain on defecation 26
(3 studies, low quality), or no symptoms (3 studies, low quality). 27
Signs identified on examination 28
The evidence showed that the majority of women with ectopic pregnancy had pelvic tenderness (91%, 29
1 study), adnexal tenderness (82%, 7 studies), or abdominal tenderness (78%, 11 studies). 30
The evidence showed that 40 75% of women with ectopic pregnancy had cervical motion 31
tenderness (8 studies), pallor (1 study), or rebound tenderness or peritoneal signs (9 studies). 32
The evidence showed that 20 40% of women with ectopic pregnancy had abdominal distension (2 33
studies, low quality), an enlarged uterus (6 studies), an adnexal mass (9 studies), or tachycardia or 34
hypotension (5 studies, low quality). 35
The evidence showed that less than 20% of women with ectopic pregnancy had a palpable pelvic 36
mass (2 studies), were collapsed or in shock (8 studies), or had orthostatic hypotension (3 studies). 37
In conducting this review, the group was keen to identify uncommon signs and symptoms associated 40
with ectopic pregnancy. 41
The group recognised that risk factors were not a helpful method for identifying women with an 43
ectopic pregnancy, as about a third of women with an ectopic pregnancy had no identifiable risk 44
factors. The group agreed that even in the absence of risk factors, it would still be necessary for a 45
healthcare professional to rule out the possibility of ectopic pregnancy, and thus agreed that they 46
should not be used as a diagnostic aid. 47
68
The group recognised that there was a wide range of symptoms associated with ectopic pregnancy, 1
including some non-specific symptoms such as gastro-intestinal symptoms. Given this, the group felt 2
that there is value in healthcare professionals always considering pregnancy in women of childbearing 3
age presenting with these symptoms, and thus should consider conducting a pregnancy test. If a 4
woman is found to be pregnant, the GDG agreed that ectopic pregnancy should always be 5
considered, as many of the symptoms of ectopic pregnancy are the same as those of pregnancy in 6
general. They felt strongly that all health care professionals that provide care to women of 7
reproductive age should have access to pregnancy tests, in order that women with a suspected 8
ectopic pregnancy can be identified and referred promptly and appropriately. 9
The evidence showed that pain or tenderness in the abdominal region was associated with ectopic 10
pregnancy. As a result, the group recommended that women with pain in these areas and a positive 11
pregnancy test should be immediately referred for further assessment. For women with bleeding but 12
no abdominal tenderness, the group agreed that a vaginal examination should be performed to elicit 13
the presence of cervical excitation or pelvic tenderness. 14
Consideration of health benefits and resource use 15
The group recognised that lowering the index of suspicion for ectopic pregnancy was likely to lead to 16
an increase in the number of women being offered a pregnancy test and thus an increase in the 17
number of women referred to an early pregnancy assessment service. However, the group felt that 18
this approach was very likely to be cost effective given the potential large QALY loss associated with 19
missing a diagnosis of ectopic pregnancy. 20
The majority of evidence was of moderate quality and as such, the group felt that it was sufficient to 22
make recommendations. 23
From their clinical experience, the GDG thought that uncertainty about what was happening would 25
increase womens anxiety. Therefore, they felt that it was important that women who required referral 26
were given information about why the referral was necessary, and what they might expect when they 27
arrived at the early pregnancy assessment service. 28
The GDG was keen to emphasise that ectopic pregnancy can present with a variety of symptoms and 30
agreed that it would be helpful for healthcare professionals to be able to refer to a list of potential 31
signs and symptoms (recommendations 11 and 12). Although reported in the evidence, the GDG felt 32
that it was not appropriate to include adnexal mass or palpable pelvic mass in the signs and 33
symptoms table. From their clinical experience, they noted that palpation can increase the risk of an 34
ectopic pregnancy rupturing. They also felt that, while palpation might once have been used in the 35
diagnosis of ectopic pregnancy, the development of new diagnostic modalities (such as transvaginal 36
ultrasound and biochemical tests), has meant that it is no longer the most appropriate tool. 37
The GDG felt that a decision tool, incorporating risk factors, signs and symptoms, could be very 38
valuable in decreasing the likelihood of women with ectopic pregnancies being misdiagnosed and 39
therefore mismanaged. They group noted that the evidence from this review was not sufficient to 40
develop and validate such a tool, and therefore decided that a research recommendation was 41
warranted. The group felt that such a tool would be extremely valuable in allowing health care 42
professionals, particularly non-specialists, to evaluate a womans likelihood of having an ectopic 43
pregnancy and to determine the level of urgency of any resulting referral. 44
Recommendations 45
7 When assessing women of reproductive age, be aware that they may be pregnant,
and think about offering a pregnancy test even when symptoms are non-specific.
8 All healthcare professionals involved in the care of women of reproductive age
69
should have access to pregnancy tests.
9 Exclude the possibility of ectopic pregnancy, even in the absence of risk factors
(such as previous ectopic pregnancy), because about a third of women with an
ectopic pregnancy will have no known risk factors.

Symptoms and signs
10 Be aware that atypical presentation for ectopic pregnancy is common.
11 Be aware that ectopic pregnancy can present with a variety of symptoms. Even if a
symptom is less common, it may still be significant. Symptoms of ectopic pregnancy
include:
Common symptoms
o Abdominal or pelvic pain
o Amenorrhoea or missed period
o Vaginal bleeding without clots
Other reported symptoms
o Breast tenderness
o Gastrointestinal symptoms
o Dizziness, fainting or syncope
o Shoulder tip pain
o Urinary symptoms
o Passage of tissue
o Rectal pressure or defecation
12 Be aware that ectopic pregnancy can present with a variety of signs. Signs of
ectopic pregnancy include:
More common signs
o Pelvic tenderness
o Adnexal tenderness
o Abdominal tenderness
Other reported signs
o Cervical motion tenderness
o Rebound tenderness or peritoneal signs
o Pallor
o Abdominal distension
o Enlarged uterus
o Tachycardia (more than 100 beats per minute) or hypotension (less
than 100/60 mmHg)
o Shock or collapse
o Orthostatic hypotension.
13 If a woman has a positive pregnancy test the threshold for considering an ectopic
pregnancy should be low because the symptoms and signs of ectopic pregnancy
can resemble the common symptoms and signs of other conditions - for example
gastrointestinal conditions or urinary tract infection.
14 Refer women with a positive pregnancy test and continuing or worsening symptoms
and signs that could suggest ectopic pregnancy to a dedicated early pregnancy
assessment service for further evaluation.
15 If a woman needs to be referred to a dedicated early pregnancy assessment
service, explain the reasons for the referral and what she can expect when she
arrives there.

Signs
16 Refer immediately women with a positive pregnancy test and pain and abdominal
tenderness on examination to a dedicated early pregnancy assessment service for
further investigation.
70
17 If there is no abdominal tenderness or signs of intra-abdominal bleeding, but there
is still a suspicion of an ectopic pregnancy, consider a vaginal examination to
assess for cervical excitation or pelvic tenderness. If either of these is present,
together with a positive pregnancy test, refer the woman immediately to a dedicated
early pregnancy assessment service.
1
2
RR 3 Research should be undertaken to design and validate a decision tool for evaluating
signs, symptoms and risk factors for correctly identifying ectopic pregnancy
3
6.2 Ultrasound for determining a viable intrauterine 4
pregnancy 5
Review question 6
What is the diagnostic value of ultrasound for determining a viable intrauterine pregnancy? 7
Introduction 8
The application of ultrasound is well established and important in the assessment and evaluation of 9
early pregnancy events and complications. Its use in early pregnancy assessment may be routine but 10
practices vary considerably. Although high resolution transvaginal ultrasound has been widely 11
adopted, the limitations of accuracy in defining ultra-small structures, such as a fetal heart at early 12
gestations, are acknowledged. The aim of this review was to identify the point at which viability of a 13
pregnancy can be definitively confirmed using ultrasound. This threshold also represents the point at 14
which miscarriage can be definitively diagnosed. 15
Fifteen studies were included in this review (Abaid et al., 2007; Abdallah et al., 2011; Bree et al., 17
1989; Brown et al., 1990; Cacciatore et al., 1990; de Crespigny, 1988; Ferrazzi et al., 1993; Goldstein, 18
1992; Hassan et al., 2009; Levi et al., 1988; Levi et al., 1990; Pennell et al., 1991; Rempen, 1990; 19
Rowling et al., 1999; Steinkampf et al., 1997). 20
The included studies consist of nine prospective observational studies (Abdallah et al., 2011; Bree et 21
al., 1989; Cacciatore et al., 1990; de Crespigny, 1988; Goldstein, 1992; Hassan et al., 2009; Pennell 22
et al., 1991; Rempen, 1990; Rowling et al., 1999), five retrospective observational studies (Abaid et 23
al., 2007; Ferrazzi et al., 1993; Levi et al., 1988; Levi et al., 1990; Steinkampf et al., 1997) and one 24
partially retrospective observational study (Brown et al., 1990). 25
The studies were conducted in the UK (Abdallah et al., 2011; Hassan et al., 2009), Germany 26
(Rempen, 1990), Italy (Ferrazzi et al., 1993), Finland (Cacciatore et al., 1990), the USA (Abaid et al., 27
2007; Bree et al., 1989; Brown et al., 1990; Goldstein, 1992; Pennell et al., 1991; Rowling et al., 1999; 28
Steinkampf et al., 1997), Canada (Levi et al., 1988; Levi et al., 1990), and Australia (de Crespigny, 29
1988). 30
All included studies evaluated the use of transvaginal ultrasound for visualising fetal cardiac activity in 31
intrauterine pregnancies, and stratified their findings by gestational age, crown-rump length or 32
gestation sac size. Two studies additionally compared the performance of transabdominal ultrasound 33
in visualising cardiac activity (Ferrazzi et al., 1993; Pennell et al., 1991). 34
71
Evidence profile 1
Table 6.2 GRADE summary of findings for evaluation of ultrasound for determining a viable intrauterine 2
pregnancy 3
Number of
studies
Type of ultrasound
scan (transvaginal
(TVU) or
transabdominal
(TAU))
Number of women
scanned for fetal
cardiac activity,
and stratified by
fetal size/ age
(Total study
participants)
Threshold at
which 100% of
fetuses that later
proved to be
viable can be
identified
Quality
Visualisation of cardiac activity by crown-rump length / mm
1 study
(Rempen, 1990)
5-MHz TVU 292
(363)
3 High
1 study
(Pennell et al.,
1991)
5-MHz/3.5-MHz
TAU
5-MHz/7.5-MHz
TVU
175
(175)
TAU: 9
TVU: 5
High
1 study
(Hassan et al.,
2009)
TVU 1174
(1174)
6.0 Moderate
1 study
(Abaid et al.,
2007)
8-MHz TVU 179
(179)
3.5 Moderate
1 study
(Brown et al.,
1990)
5-MHz TVU 375
(375)
5 Moderate
1 study
(Abdallah et al.,
2011)
6-12-MHz TVU 24
(1060)
5.3 Low
1 study
(Levi et al., 1990)
6.5-MHz TVU 71
(71)
4.0 Low
1 study
(Goldstein, 1992)
5-MHz/7.5-MHz
TVU
96
(96)
4 Low
Visualisation of cardiac activity by gestation sac diameter / mm
1 study
(Rempen, 1990)
5-MHz TVU 354
(363)
18.3 * High
1 study
(Abdallah et al.,
2011)
6-12-MHz TVU 183
#

(1060)
21 Moderate
266
#
(1060)
21
1 study
(Bree et al., 1989)
7-MHz TVU 53
(53)
> 9 Moderate
72
1 study
(Rowling et al.,
1999)
9-5-MHz TVU 39
(39)
13 Low
1 study
(Levi et al., 1988)
6.5-MHz TVU 35
(62)
16 Very low
1 study
(de Crespigny,
1988)
5-MHz TVU 353
(353)
> 12 Low
1 study
(Steinkampf et al.,
1997)
5-MHz TVU 82
(82)
19** Very low
1 study
(Cacciatore et al.,
1990)
5-MHz/6.5-MHz
TVU
20
(22)
> 18*** Very low
Visualisation of cardiac activity by gestational or menstrual age (days)
1 study
(Rempen, 1990)
5-MHz TVU 252
(363)
46
(menstrual age)
High
1 study
(Bree et al., 1989)
7-MHz TVU 53
(53)
> 40
(gestational age)
Moderate
1 study
(Steinkampf et al.,
1997)
5-MHz TVU 82
(82)
45.5**
(gestational age)
Very low
1 study
(Ferrazzi et al.,
1993)
5-MHz TAU/
5-MHz TVU
76
(598)
TAU: 37
TVU: 35
(menstrual age)
Very low
1 study
(Cacciatore et al.,
1990)
5-MHz/6.5-MHz
TVU
20
(22)
> 43***
(gestational age)
Very low
* Chorionic cavity diameter 1
** Point of 99% probability of visualisation 2
*** Point of reliable detection 3
# 183 scans showed a gestation sac without a visible embryo or yolk sac. 266 scans showed a gestation sac with a yolk sac, 4
but without a visible embryo. 5
The outcomes for this review were the thresholds of gestational age, CRL or gestation sac diameter 8
at which 100% of fetuses that later proved to be viable had cardiac activity visible on ultrasound. The 9
group noted that, even when women could be completely certain about the date of intercourse or last 10
menstrual period, variation in the menstrual cycle and rate of fetal growth might result in inaccurate 11
estimates of gestational age. Therefore, the GDG felt that the use of a gestational age threshold alone 12
was not appropriate for the determination of viability. From their own clinical experience, the group felt 13
that, where it was possible to measure, the crown-rump length would provide the most accurate 14
estimate of development; therefore, measurement of mean gestational sac diameter was only 15
recommended in cases where a fetal pole could not be identified. 16
73
The GDG felt that it was appropriate to set thresholds for the determination of a viable intrauterine 2
pregnancy; however, considering the consequences of misdiagnosing a viable intrauterine pregnancy 3
as a miscarriage, they felt that the thresholds should be based on the most conservative findings 4
reported in the studies. The GDG were aware of recent work by Pexsters et al. (2011) that 5
documented the potential for considerable intra- and inter-observer variation in measurements of CRL 6
and mean gestational sac diameter. They also noted that there was additional potential for variation in 7
measurements linked to the quality of scanning equipment and the skill level of the sonographer. In 8
light of these considerations and the findings from the included studies, the GDG determined that fetal 9
non-viability should not be diagnosed based on the absence of a heartbeat in fetuses with a CRL of 10
less than or equal to 6.0 mm or a mean gestational sac diameter of less than or equal to 25.0 mm, as 11
measured on a single transvaginal ultrasound. They felt that at such small sizes, it would not be 12
possible to determine whether a miscarriage had truly occurred or whether the embryo was simply too 13
small for there to be a visible heartbeat. Therefore, up to and including these thresholds, all women 14
should have a repeat scan to confirm the findings of the initial scan. The GDG discussed what would 15
be an appropriate interval between scans, balancing the fact that sufficient time would need to pass to 16
be able to confirm the diagnosis with the fact that women might understandably want an answer as 17
soon as possible. They felt that, for embryos with a measurable crown-rump length, an interval of 7 18
days would be sufficient to definitively demonstrate viability. However, they felt that a longer interval 19
would be required for women in whom only a gestation sac could be measured, and therefore 20
recommended that in such cases, the scan was performed after 14 days. 21
Due to the significant consequences of misclassifying a viable pregnancy as a miscarriage, the GDG 22
felt that it was reasonable to recommend that, in the absence of a visible heartbeat above these 23
thresholds (i.e. CRL > 6 mm, mean gestational sac diameter > 25 mm), all sonographers should seek 24
a second opinion before definitively diagnosing a non-viable pregnancy. However, they also realised 25
that some women might instead wish to wait and have a second confirmatory scan at a later date, and 26
felt that this was a reasonable alternative. 27
Based on their clinical experience, and evidence from another review comparing transvaginal and 28
transabdominal ultrasound for diagnosing ectopic pregnancy (see section 5.3 below), the GDG 29
decided that transvaginal ultrasound would generally be the optimal mode of scanning. However, they 30
recognised that, in some circumstances, a transvaginal ultrasound might not be appropriate or 31
acceptable to women, and therefore a transabdominal scan could be offered as an alternative. Based 32
on evidence from Pennell et al., (1991) the GDG decided that a higher CRL threshold of > 10 mm 33
should be used for diagnosing miscarriage on the basis of a single transabdominal ultrasound scan. 34
However, they felt it was safe to use the same mean gestational sac diameter threshold of > 25 mm, 35
because the larger size of the gestational sac results in almost equivalent accuracy with 36
transabdominal ultrasound. 37
The group did not feel that recommending a repeat ultrasound scan would add significantly to the 39
case load and resource use, because in practice this often happens anyway. However, they did feel 40
that cost-effectiveness should be a component of any research conducted in this area, and therefore 41
incorporated it into a research recommendation. Given the number of women requesting scans in 42
early pregnancy and the associated service and cost implications, the group felt that it was important 43
that research be done in this area, to determine the timing and frequency of ultrasound examinations 44
that would maximise improvements in diagnosis, outcomes and womens experience. 45
The quality of evidence ranged from high to very low quality and the GDG felt that, in conjunction with 47
their clinical experience, it was appropriate to make recommendations based on the findings. 48
As discussed above, although the GDG felt that transvaginal scanning was the optimal mode of 50
scanning, they understood that some women would choose to have a transabdominal scan instead. 51
The group fully supported this choice; however, they did feel that women should be given information 52
about the potential limitations of transabdominal scanning so that they could make a fully informed 53
choice. 54
74
From their own experience, and data about the risks of expectant management reported in another 1
review question (see section 6.3), the GDG felt that there would be minimal risk in recommending that 2
some women wait a week for a repeat scan, particularly when balanced against the consequences of 3
accidentally terminating a viable pregnancy after misdiagnosis. However, they felt that it was 4
important that women were informed about what to expect in the intervening week, and what 5
symptoms should prompt them to seek medical attention. Given this, they also recommended that 6
women should be provided with a 24 hour contact telephone number. The GDG were of the opinion 7
that, due to the fact that miscarriage is a potentially traumatic experience, it was important that 8
women be able to contact someone who would be able to give them accurate information and 9
appropriate support. Therefore, they recommended that this telephone number should allow women 10
to speak directly with someone with experience of dealing with early pregnancy complications, and 11
should not simply be a non-specific service like an emergency department. 12
The GDG were aware of a recent addendum to guidelines from the RCOG, which stated that all 14
women should receive a transvaginal ultrasound and that a diagnosis of miscarriage should only be 15
considered in women with an empty gestational sac of greater than or equal to 25 mm or a CRL 16
greater than or equal to 7 mm in the absence of a fetal heartbeat. 17
The GDG felt it important to note that not all women presenting to a health care professional with 18
bleeding in early pregnancy needed to be referred for a transvaginal ultrasound scan. For women who 19
were reporting to be less than 6 weeks pregnant with minor symptoms, or women of later gestation 20
with no pain and minimal blood loss or spotting, expectant management could be undertaken. The 21
GDG felt that this was justified, because at gestations of earlier than 6 weeks, the pregnancy is likely 22
to be too small to yield any information about viability. In addition, from their clinical experience, they 23
agreed that many women experience spotting in early pregnancy which resolves without need for 24
further intervention. However, the group also recognised that in all other cases, i.e. where women had 25
blood loss greater than spotting, were in pain, or where there was uncertainty about the pregnancys 26
gestation, referral to a dedicated early pregnancy service should be made in order that an ultrasound 27
scan could be carried out. 28
Recommendations 29
18 Use expectant management for:
women with a pregnancy of less than 6 weeks gestation who are bleeding
but not in pain
women with a pregnancy of 6 weeks gestation or more who have minimal
blood loss (spotting) and who are not in pain.
Advise these women to take a urine pregnancy test after a week and to return if
their symptoms worsen.
Refer all other women to a dedicated early pregnancy assessment service.
19 Offer women with pain and/or bleeding a transvaginal ultrasound scan to identify the
location of the pregnancy and whether there is a fetal pole and heartbeat.
20 If a transvaginal ultrasound scan is unacceptable to the woman offer a
transabdominal ultrasound scan and explain the limitations of this method of
scanning.
21 Inform women that the diagnosis of miscarriage using one ultrasound scan cannot
be guaranteed to be 100% accurate and there is a small chance that the diagnosis
may be incorrect, particularly at very early gestational ages.
22 If the crown rump length is 6.0 mm or less with a transvaginal ultrasound scan (or
10.0 mm or less with a transabdominal ultrasound scan) and there is no visible fetal
heartbeat, perform a second scan a minimum of 7 days after the first before making
75
a diagnosis.
23 If the crown rump length is more than 6.0 mm with a transvaginal ultrasound scan
(or 10.0 mm with a transabdominal ultrasound scan), and there is no visible fetal
heartbeat:
seek a second opinion on the viability of the pregnancy and/or
perform a second scan a minimum of 7 days after the first before making a
diagnosis.
24 Do not measure the mean gestational sac diameter if there is a fetal heartbeat.
25 Measure the mean gestational sac diameter if the fetal pole is not visible.
26 If the mean gestational sac diameter is 25.0 mm or less with no fetal pole, offer the
woman a second scan, to be performed a minimum of 14 days after the first.
Further scans may be needed before a diagnosis can be made.
27 If the mean gestational sac diameter is greater than 25.0 mm with no fetal pole:
seek a second opinion on the viability of the pregnancy and/or
perform a second scan a minimum of 7 days after the first before making a
diagnosis.
28 Do not use gestational age alone to determine whether a fetal heartbeat should be
visible.
29 Inform women that the date of their last menstrual period may not give an accurate
representation of gestational age because of variability in the menstrual cycle.
30 Inform women what to expect while waiting for a repeat scan and that waiting for a
repeat scan has no detrimental effects on the outcome of the pregnancy.
31 Give women specific verbal and written information on when and how to seek help if
symptoms worsen or new symptoms develop.
32 Give women a 24-hour contact telephone number so that they can speak to
someone who understands their needs and can advise on appropriate care.
33 Provide women who have a confirmed miscarriage with written and verbal
information about where to access support and counselling services (including links
to useful websites).
1
2
RR 4 How does the timing and frequency of ultrasound examination affect diagnosis and
outcomes of early pregnancy complications, including patient experience and cost
effectiveness?

The rationale behind the frequency of ultrasound to improve diagnosis and
outcomes of early pregnancy complications addresses the problems associated
with pregnancy of unknown location and intrauterine pregnancy of uncertain
viability. There is no evidence base for timing of scans, and number of scans is
organised by individual units according to capacity and demand. Some experts
choose to wait 5 days between scans while others will wait 10 to 14. These
decisions are driven by resource availability as well as clinical considerations.
Discussions among experts have failed to provide clear consensus, but all are
agreed that by 14 days a diagnosis will be clear. To establish the most appropriate
time for scans, the efficacy of scans taken after 14 days could be compared with
76
scans taken after 7 days for diagnosis of ectopic pregnancy or viability.
1
6.3 Accuracy of imaging techniques for diagnosis of an 2
ectopic pregnancy 3
Review question 4
What is the accuracy of transvaginal ultrasound compared with transabdominal ultrasound for 5
diagnosing ectopic pregnancy? 6
Introduction 7
The estimated prevalence of ectopic pregnancy is 1-2% worldwide with nearly 12,000 ectopic 8
pregnancies diagnosed each year in the UK. Associated costs are high due to repeated diagnostic 9
tests, delayed diagnosis and its treatment (Jurkovic & Wilkinson, 2011). Critical evaluation of patient 10
symptoms and signs remains important for the detection of ectopic pregnancy; however ultrasound 11
scanning remains the cornerstone of clinical diagnosis. Transabdominal ultrasound has, in recent 12
years been largely replaced by transvaginal ultrasound imaging bearing in mind that diagnostic 13
accuracy is experience-based and allied to constant vigilance for the potential presence of ectopic 14
pregnancy. The GDG therefore considered the evidence for application of the two different methods in 15
order to determine which should be used. 16
Five studies were included in this review. Three studies were conducted in the USA (Kivikoski et al., 18
1990; Shapiro et al., 1988; Thorsen et al., 1990), one in Austria (Schurz et al., 1990), and one in 19
Finland (Cacciatore, 1989). Four prospective studies (Cacciatore, 1989; Kivikoski et al., 1990; Shapiro 20
et al., 1988; Thorsen et al., 1990) compared the diagnostic accuracy of transvaginal and 21
transabdominal ultrasound in women with suspected ectopic pregnancy. One prospective study 22
(Schurz et al., 1990) evaluated reliability and advantages of transabdominal and transvaginal 23
ultrasound compared with clinical signs for detection of ectopic pregnancy. 24
In one prospective study (Kivikoski et al., 1990) women were first seen at 4-12 weeks amenorrhea at 25
the time of evaluation. Four prospective studies (Cacciatore, 1989; Schurz et al., 1990; Shapiro et al., 26
1988; Thorsen et al., 1990) did not report womens gestation at the time of ultrasound evaluation. 27
Evidence profile 28
The evidence is presented below in one profile. Diagnostic accuracy for ectopic pregnancy was 29
measured using transvaginal and transabdominal ultrasound. 30
Table 6.3 GRADE summary of findings for accuracy of diagnosing ectopic pregnancy using transvaginal or 31
transabdominal ultrasound 32
Number
of
studies
Number
of
women
Measure of diagnostic accuracy Quality
Sensitiv-
ity
Specif-
icity
Positive
predict-
tive
value
Negative
predict-
tive
value
Positive
likelihood
ratio
Negative
likelihood
ratio
Transvaginal ultrasound
1 study
(Shapiro
et al.,
1988)
25 90
(78 to
100)*
33
(20 to
86)*
90
(78 to
100)*
33
(20 to
86)*
1.36
(0.60 to
3.06)*
0.27
(0.05 to
2.17)*
Very
low
77
1 study
(Thorsen
et al.,
1990)
193 38
(26 to
50)*
100
(100 to
100)*
100
(100 to
100)*
78
(72 to
84)*
infinity* 0.6
(0.50 to
0.75)*
Moder-
ate
1 study
(Kivikosk
i et al.,
1990)
34 72
(56 to
90)*
100
(100 to
100)*
100
(100 to
100)*
53
(28 to
78)*
infinity* 0.26
(0.14 to
0.50)
Low
Transabdominal ultrasound
1 study
(Shapiro
et al.,
1988)
25 50
(29 to
70)*
Not
calculabl
e (NC)
NC NC NC NC Very
low
1 study
(Thorsen
et al.,
1990)
193 21
(11 to
32)*
100
(100 to
100)*
100
(100 to
100) *
73
(67 to
80)*
infinity* 0.7*
(0.67 to
0.89)
Moder-
ate
1 study
(Kivikosk
i et al.,
1990)
34 44
(25 to
63)*
100
(100 to
100)*
100
(100 to
100)*
31
(12 to
51)*
infinity* 0.55
(0.39 to
77)*
Low
* NCC calculation 1
Table 6.4 Additional diagnostic accuracy findings for transvaginal and transabdominal ultrasound from two 2
observational studies 3
Additional data
1 study
(Cacciatore,
1989)
Two very low quality studies also evaluated the diagnostic accuracy of transvaginal
and transabdominal ultrasound for ectopic pregnancy in women with clinical suspicion
of ectopic pregnancy. However, there was not adequate information reported in the
studies to calculate diagnostic accuracy measures. In one very low quality study
(Cacciatore, 1989) there was no statistically significant difference between the
transvaginal and transabdominal ultrasounds in detection of an adnexal mass (90%
vs. 80%), gestational sac (92% vs. 89%). More ectopic fetuses (21% vs. 0.0%),
ectopic sacs (69% vs. 44%), un-ruptured ectopic pregnancies (82% vs. 50%) and yolk
sacs or viable fetuses (49% vs. 0.0%) were detected by transvaginal ultrasound when
compared with transabdominal ultrasound.
1 study
(Schurz et al.,
1990)
In the other very low quality study (Schurz et al., 1990) reliability and advantages of
transabdominal and transvaginal ultrasound were compared with clinical signs for
detection of ectopic pregnancy. Clinical findings were more likely to lead to a correct
diagnosis of ectopic pregnancy than the findings obtained from transabdominal
ultrasound (58% vs. 25%; p < 0.05). However, clinical findings were less likely to lead
to a correct diagnosis of ectopic pregnancy than the findings obtained by transvaginal
ultrasound (26% vs. 95%; p < 0.05). This sample was examined at an earlier
gestational age than the sample used to compare abdominal ultrasound (hence the
difference in detection of ectopic pregnancy based on clinical findings).
4
78
In the evidence statements the following definitions have been used when summarising the levels of 2
sensitivity, specificity, positive predictive value (PPV) and negative predictive value (NPV): 3
High 90% and above 4
Moderate 75% to 89% 5
Low below 75% 6
The following terms have been used when summarising the positive and negative likelihood ratios: 7
Positive likelihood ratio: 8
Very useful > 10 9
Moderately useful > 5 to 10 10
Not useful < 5 11
Negative likelihood ratio: 12
Very useful 0 to 0.1 13
Moderately useful > 0.1 to 0.5 14
Not useful > 0.5 15
Evidence was identified for transvaginal and transabdominal ultrasounds to determine diagnostic 16
accuracy for ectopic pregnancy in women with clinical suspicion of ectopic pregnancy. The quality of 17
the evidence was moderate, low and very low for the included studies. 18
Transvaginal ultrasound 19
Three studies evaluated the diagnostic accuracy of transvaginal ultrasound for ectopic pregnancy in 20
women with suspected ectopic pregnancy. One very low quality study reported a high sensitivity, a 21
low specificity, a high PPV, a low NPV, a not useful positive likelihood ratio and a moderately useful 22
negative likelihood ratio. One moderate quality study reported low sensitivity, high specificity, a high 23
PPV, a moderate NPV, a very useful positive likelihood ratio and a not useful negative likelihood ratio. 24
One low quality study reported low sensitivity, high specificity, a high PPV, a low NPV, a very useful 25
positive likelihood ratio and a moderately useful negative likelihood ratio. 26
Transabdominal ultrasound 27
The same three studies evaluated the diagnostic accuracy of transabdominal ultrasound for ectopic 28
pregnancy in women with suspected ectopic pregnancy. One study of very low quality reported a low 29
sensitivity with no information provided about other diagnostic accuracy measures. One moderate 30
quality study reported low sensitivity, high specificity, a high PPV, a low NPV, a very useful positive 31
likelihood ratio and a not useful negative likelihood ratio. One low quality study reported a low 32
sensitivity, a high specificity, a high PPV, a low NPV, a very useful positive likelihood ratio and a not 33
useful negative likelihood ratio. 34
Diagnosis of ectopic pregnancy 35
Two very low quality studies also evaluated the diagnostic accuracy of transvaginal and 36
transabdominal ultrasound for ectopic pregnancy in women with clinical suspicion of ectopic 37
pregnancy. In one low quality study (Cacciatore, 1989) there was no statistically significant difference 38
between the transvaginal and transabdominal ultrasounds in detection of an adnexal mass and 39
gestational sac. More ectopic fetuses, ectopic sacs, un-ruptured ectopic pregnancies and yolk sacs or 40
viable fetuses were detected by transvaginal ultrasound when compared with transabdominal 41
ultrasound. 42
In the other very low quality study (Schurz et al., 1990) reliability and advantages of transabdominal 43
and transvaginal ultrasound were compared with clinical signs for detection of ectopic pregnancy in 44
two populations. Clinical findings were more likely to lead to a correct diagnosis of ectopic pregnancy 45
than the findings obtained from transabdominal ultrasound However; clinical findings were less likely 46
79
to lead to a correct diagnosis of ectopic pregnancy than the findings obtained by transvaginal 1
ultrasound. 2
Given the serious harm associated with failing to diagnose an ectopic pregnancy, the group agreed 5
that sensitivity was more important than specificity i.e. that it was more important to successfully 6
identify those women with an ectopic pregnancy and potentially create some false positives, than to 7
try to avoid false positives and thereby run the risk of failing to identify some women with an ectopic 8
pregnancy. 9
The evidence showed mixed results for both ultrasound methods. However, each study which looked 11
at transvaginal and transabdominal ultrasound showed that the transvaginal method had a higher 12
sensitivity. As a result, the group agreed that transvaginal ultrasound should be the preferred 13
approach but that womens views should be taken into account and accommodated where possible. 14
The group also identified from their clinical experience that there might be occasions when 15
transabdominal ultrasound would be the better option such as when women have an enlarged uterus 16
or other pelvic pathology. 17
The group did not feel that there was a difference in cost between the two different methods as the 19
time taken to perform them both would be the same. 20
The evidence varied in quality from very low to moderate. Nonetheless, taken in conjunction with the 22
groups clinical experience, it was felt sufficient to make recommendations 23
The group recognised that, whilst the method of performing an ultrasound scan can make a 25
difference, it is also extremely important that the scanning is undertaken by someone with specific 26
training and experience in identifying ectopic pregnancy. Not only can this can make a large 27
difference to the validity of the diagnosis, it can also impact the experience of the woman undergoing 28
the scan. From their clinical experience, the GDG felt that having a scan performed by a practitioner 29
without appropriate experience would be likely to make a woman feel anxious and uncertain about her 30
prognosis. In contrast, having a scan performed by a healthcare professional with experience in 31
diagnosing ectopic pregnancies would help to ensure that women felt informed and supported. 32
Recommendations 33
34 Offer a transvaginal ultrasound scan to diagnose or exclude an ectopic pregnancy.
35 Consider a transabdominal ultrasound scan for women with an enlarged uterus or
other pelvic pathology, such as fibroids or an ovarian cyst.
36 All ultrasound scans should be performed or reviewed by someone with training in
and experience of diagnosing ectopic pregnancies.
80
6.4 Diagnostic accuracy of two or more human 1
chorionic gonadotrophin (hCG) measurements for ectopic 2
pregnancy 3
Review question 4
What is the diagnostic accuracy of two or more hCG measurements for determining an ectopic 5
pregnancy in women with pain and bleeding and pregnancy of unknown location? 6
Introduction 7
The diagnosis of ectopic pregnancy remains problematic in a significant number of cases. Women 8
may present with a positive urine or blood pregnancy test, but no visible evidence of the location of 9
the pregnancy on an ultrasound scan (pregnancy of unknown location [PUL]). In these circumstances, 10
there is often a tension between not missing an ectopic pregnancy and not subjecting women with an 11
early intrauterine pregnancy to a battery of expensive and potentially unnecessary tests or 12
interventions. One possible alternative is to use serial measurements of human chorionic 13
gonadotrophin (hCG) as a diagnostic tool to try and identify women that might have an ectopic 14
pregnancy and those that are likely to have an early viable intrauterine pregnancy. The following 15
reviews evaluate the diagnostic accuracy of this approach, and consider whether there is added value 16
in the use of a single progesterone measurement. 17
Ten studies were included in this review (Barnhart et al., 2010; Condous et al., 2004; Condous et al., 19
2007; Dart et al., 1999; Daus et al., 1989; Hahlin et al., 1991; Mol et al., 1998; Morse et al., 2012; 20
Stewart et al., 1995; Thorburn et al., 1992). 21
The included papers consist of five prospective cohort studies (Condous et al., 2004; Condous et al., 22
2007; Hahlin et al., 1991; Mol et al., 1998; Thorburn et al., 1992) and five retrospective cohort studies 23
(Barnhart et al., 2010; Dart et al., 1999; Daus et al., 1989; Morse et al., 2012; Stewart et al., 1995). 24
The studies were conducted in the UK (Condous et al., 2004; Condous et al., 2007), the USA (Dart et 25
al., 1999; Daus et al., 1989; Morse et al., 2012; Stewart et al., 1995), the Netherlands (Mol et al., 26
1998) and Sweden (Hahlin et al., 1991; Thorburn et al., 1992). One study (Barnhart et al., 2010) was 27
conducted in both the UK and USA. 28
All the studies used two or more serum hCG measurements for diagnosis, and either reported 29
measures of diagnostic accuracy, or presented data that allowed calculations to be performed by the 30
technical team. Four studies evaluated the percentage change in hCG over 48 hours (Dart et al., 31
1999; Daus et al., 1989; Mol et al., 1998; Morse et al., 2012), one study evaluated the rate of change 32
of log hCG using two different thresholds (Stewart et al., 1995), and two studies evaluated the 33
diagnostic accuracy of an abnormal hCG score (Hahlin et al., 1991; Thorburn et al., 1992). A further 34
two papers evaluated the diagnostic accuracy of two different predictive models, M1 and M4 35
(Condous et al., 2004; Condous et al., 2007). Model M1 incorporated the hCG ratio (ratio of 36
concentration at 48 hours to concentration at 0 hours), and its performance was evaluated using three 37
different sets of parameters. Firstly, the model was evaluated using a probability threshold to 38
distinguish between ectopic pregnancies and non-ectopic pregnancies. Then, the authors 39
incorporated different statistical costs for misclassifying outcomes, using costs of one for 40
misclassifying failing pregnancies of unknown location and intrauterine pregnancies, and a cost of 41
either four or five for misclassifying an ectopic pregnancy. Model M4 incorporated the average hCG 42
concentration (from measurements at 0 and 48 hours), the ratio of the two hCG measurements, and 43
the quadratic effect of the hCG ratio. M4 was only evaluated using the optimal costs of one, one and 44
four for misclassifying a failing pregnancy of unknown location, an intrauterine pregnancy and an 45
ectopic pregnancy, respectively. The study participants were women with pain and bleeding in the first 46
trimester of pregnancy who had been classified as having a pregnancy of unknown location on 47
ultrasound. 48
81
Evidence profile 1
Table 6.5 GRADE summary of findings for the diagnosis of ectopic pregnancy using two or more hCG 2
measurements 3
Number
of
studies
Number
of
women
Sensitivity
% (95% CI)
Specificity
% (95%
CI)
Positive
predict-
tive
value %
(95% CI)
Negative
predict-
tive
value %
(95% CI)
Positive
likeli-
hood
ratio
(95% CI)
Negative
likeli-
hood
ratio
(95% CI)
% change in serum hCG in 48 hours: decline, or rise of < 50 %
1 study
(Mol et
al., 1998)
195 68.4
(53.6 to
83.2)*
11.5
(6.5 to
16.5)*
15.8
(10.2 to
21.3)*
60.0
(42.5 to
77.5)*
0.77
(0.62 to
0.97)*
2.75
(1.45 to
5.22)*
Low
1 study
(Daus et
al., 1989)
357 93.6
(86.6 to
100)*
18.4
(14.1 to
22.7)*
14.8
(10.8 to
18.9)*
95.0
(89.5 to
100)*
1.15
(1.05 to
1.26)*
0.35
(0.11 to
1.06)*
Very
low
1 study
(Dart et
al., 1999)
307 81.8
(68.7 to
95.0)*
16.8
(12.4 to
21.2)*
10.6
(6.8 to
14.4)*
88.5
(79.8 to
97.2)*
0.98
(0.83 to
1.16)*
1.08
(0.50 to
2.34)*
Low
% change in serum hCG in 48 hours: between a decline of 36-47% and a rise of 35%
1 study
(Morse
et al.,
2012)
1005 83.2
(77.7 to
88.8)
70.8
(67.7 to
73.9)
38.2
(33.4 to
43.0)
95.1
(93.4 to
96.8)
not
reported
(NR)
NR Very
low
1 study
(Morse
et al.,
2012)
1005 91.1
(86.8 to
95.3)
66.6
(63.4 to
69.8)
37.1
(32.6 to
41.7)
97.2
(95.8 to
98.5)
NR NR Very
low
1 study
(Morse
et al.,
2012)
1005 92.2
(88.2 to
96.2)
62.8
(59.5 to
66.1)
35.0
(30.6 to
39.3)
97.4
(96.0 to
98.7)
NR NR Very
low
Rate of change of log hCG: < 0.11
1 study
(Stewart
et al.,
1995)
36
#
89.7
(81.8 to
97.5)*
37.3
(25.0 to
49.6)*
58.4
(48.2 to
68.7)*
78.6
(63.4 to
93.8)*
1.43
(1.15 to
1.77)*
0.28
(0.12 to
0.63)*
Very
low
Rate of change of log hCG: < 0.14
1 study 36
#
98.3 22.0 55.3 92.9 1.26 0.08 Very
82
(Stewart
et al.,
1995)
(94.9 to
100)*
(11.5 to
32.6)*
(45.7 to
64.9)*
(79.4 to
100)*
(1.10 to
1.45)*
(0.01 to
0.58)*
Low
Abnormal hCG score
1 study
(Hahlin
et al.,
1991)
307 88.7
(83.8 to
93.6)*
47.3
(39.3 to
55.3)*
64.4
(58.0 to
70.7)*
79.6
(71.1 to
88.0)*
1.68
(1.43 to
1.98)*
0.24
(0.15 to
0.38)*
Low
1 study
(Thorbur
n et al.,
1992)
261 81.1
(73.0 to
89.2)*
43.9
(36.4 to
51.3)*
43.2
(35.7 to
50.7)*
81.5
(73.6 to
89.5)*
1.44
(1.22 to
1.71)*
0.43
(0.27 to
0.68)*
Low
Model M1: using probability thresholds
1 study
(Condou
s et al.,
2004)
196 83.3
(62.3 to
100)**
88.0
(83.4 to
92.7)**
31.3
(15.2 to
47.3)**
98.8
(97.1 to
100)**
6.97
(4.37 to
11.11)*
0.19
(0.05 to
0.67)*
Low
Model M1: using costs 1, 1 and 4
1 study
(Condou
s et al.,
2004)
196 83.3
(62.3 to
100)**
86.4
(81.5 to
91.4)**
28.6
(13.6 to
43.5)**
98.8
(97.1 to
100)**
6.13
(3.94 to
9.56)*
0.19
(0.05 to
0.68)*
Low
1 study
(Condou
s et al.,
2007)
173 73.3
(51.0 to
95.7)**
87.3
(82.2 to
92.5)**
35.5
(18.6 to
52.3)**
97.2
(94.5 to
99.9)**
5.79
(3.48 to
9.66)**
0.31
(0.13 to
0.71)**
Low
Model M1: using costs 1, 1 and 5
1 study
(Condou
s et al.,
2004)
196 91.7
(76.0 to
100)**
84.2
(79.0 to
89.5)**
27.5
(13.7 to
41.3)**
99.4
(98.1 to
100)**
5.82
(4.00 to
8.46)*
0.10
(0.02 to
0.65)*
Low
Model M4
1 study
(Condou
s et al.,
2007)
173 80.0
(59.8 to
100)

**
88.6
(83.7 to
93.6)

**
40.0
(22.5 to
57.5) **
97.9
(95.6 to
100) **
7.02
(4.25 to
11.61)

**
0.23
(0.08 to
0.62)

**
Low
1 study
(Barnhart
et al.,
2010)
(2
included
cohorts:
UK and
adjusted
USA)
431 80.8
(65.6 to
95.9)**
88.9
(85.8 to
92.0)**
31.8
(20.6 to
43.1)**
98.6
(97.4 to
99.8)**
7.27
(5.21 to
10.14)*
0.22
(0.10 to
0.48)*
Very
low
544 54.8
(45.2 to
64.4)**
87.7
(84.7 to
90.8)**
51.4
(42.1 to
60.7)**
89.2
(86.2 to
92.1)**
4.47
(3.29 to
6.06)*
0.52
(0.46 to
0.64)*
Very
low
* Calculated by NCC-WCH technical team 1
83
# There were 36 women who received multiple hCG measurements, leading to a total of 117 pairs of hCG measurement used 1
for the diagnostic accuracy calculations 2
**Confidence intervals calculated by NCC-WCH technical team 3
The following definitions have been used when summarising the levels of sensitivity, specificity, PPV 5
and NPV: 6
High - 90% and above 7
Moderate - 75% to 89.9% 8
Low - 75% and below 9
The following terms have been used when summarising the positive and negative likelihood ratios: 10
Positive likelihood ratio: 11
Very useful > 10 12
Moderately useful > 5 to 10 13
Not useful < 5 14
Negative likelihood ratio: 15
Very useful 0 to 0.1 16
Moderately useful > 0.1 to 0.5 17
Not useful > 0.5 18
Percentage change in serum hCG concentration in 48 hours 19
One study evaluated the use of a decline or a rise <50% in hCG over 48 hours for the diagnosis of 20
ectopic pregnancy. The study reported a low sensitivity, low specificity, low PPV, low NPV, not useful 21
positive likelihood ratio and not useful negative likelihood ratio. The evidence for this finding was of 22
low quality. 23
One study evaluated the diagnostic accuracy of a decline or a rise <63% in hCG over 48 hours for the 24
diagnosis of ectopic pregnancy. The study reported a high sensitivity, low specificity, low PPV, high 25
NPV, not useful positive likelihood ratio and moderately useful negative likelihood ratio. The evidence 26
for this finding was of very low quality. 27
One study evaluated the diagnostic accuracy of a decline or a rise <66% in hCG over 48 hours for the 28
diagnosis of ectopic pregnancy. The study reported a moderate sensitivity, low specificity, low PPV, 29
moderate NPV, not useful positive likelihood ratio and not useful negative likelihood ratio. The 30
evidence for this finding was of low quality. 31
One study evaluated the diagnostic accuracy of a change in hCG between a decline of 36-47% and a 32
rise of 35%. The study reported a moderate sensitivity, low specificity, low PPV and high NPV. The 33
study did not report likelihood ratios. The evidence for this finding was of very low quality. 34
rise of 53%. The study reported a high sensitivity, low specificity, low PPV and high NPV. The study 36
did not report likelihood ratios. The evidence for this finding was of very low quality. 37
rise of 71%. The study reported a high sensitivity, low specificity, low PPV and high NPV. The study 39
did not report likelihood ratios. The evidence for this finding was of very low quality.Rate of change of 40
log hCG concentration. 41
One study evaluated the diagnostic accuracy of a rate of change of log hCG <0.11 for the diagnosis of 42
ectopic pregnancy. The study reported a moderate sensitivity, low specificity, low PPV, moderate 43
NPV, not useful positive likelihood ratio and moderately useful negative likelihood ratio. The evidence 44
for this finding was of very low quality. 45
84
One study evaluated the diagnostic accuracy of a rate of change of log hCG <0.14 for the diagnosis of 1
ectopic pregnancy. The study reported a high sensitivity, low specificity, low PPV, high NPV, not 2
useful positive likelihood ratio and useful negative likelihood ratio. The evidence for this finding was of 3
very low quality. 4
hCG score 5
Two studies evaluated the diagnostic accuracy of an abnormal hCG score for the diagnosis of ectopic 6
pregnancy. They both reported a moderate sensitivity, low specificity, low PPV, moderate NPV, not 7
useful positive likelihood ratio and moderately useful negative likelihood ratio. The evidence for this 8
finding was of low quality. 9
Models 10
One study evaluated the diagnostic accuracy of model M1, using a probability threshold, for the 11
diagnosis of ectopic pregnancy. The study reported a moderate sensitivity, moderate specificity, low 12
PPV, high NPV, moderately useful positive likelihood ratio and moderately useful negative likelihood 13
ratio. The evidence for this finding was of low quality. 14
Two studies evaluated the diagnostic accuracy of model M1, incorporating costs of 1, 1 and 4, for the 15
diagnosis of ectopic pregnancy. One study reported a moderate sensitivity, moderate specificity, low 16
ratio. The evidence for this finding was of low quality. One study reported a low sensitivity, moderate 18
specificity, low PPV, high NPV, moderately useful positive likelihood ratio and moderately useful 19
negative likelihood ratio. The evidence for this finding was of low quality. 20
One study evaluated the diagnostic accuracy of model M1, incorporating costs of 1, 1 and 5, for the 21
diagnosis of ectopic pregnancy. The study reported a high sensitivity, moderate specificity, low PPV, 22
high NPV, moderately useful positive likelihood ratio and very useful negative likelihood ratio. The 23
Two studies evaluated the diagnostic accuracy of model M4 for the diagnosis of ectopic pregnancy. 25
One study reported a moderate sensitivity, moderate specificity, low PPV, high NPV, moderately 26
useful positive likelihood ratio and moderately useful negative likelihood ratio. The evidence for this 27
finding was of low quality. One further study incorporated two populations. In the UK population, the 28
study reported a moderate sensitivity, moderate specificity, low PPV, high NPV, moderately useful 29
positive likelihood ratio and moderately useful negative likelihood ratio. In the adjusted USA 30
population, the study reported a low sensitivity, moderate specificity, low PPV, moderate NPV, not 31
useful PPV and not useful NPV. The evidence for this finding was of very low quality 32
Please see section 6.7 where the evidence from all of the reviews which assess the use of hCG for 34
diagnosis has been considered. 35
6.5 Diagnostic accuracy of two or more hCG 36
measurements plus progesterone for ectopic pregnancy 37
Review question 38
What is the diagnostic accuracy of two or more hCG measurements plus progesterone for 39
determining an ectopic pregnancy in women with pain and bleeding and pregnancy of unknown 40
location? 41
Three studies were included in this review (Condous et al., 2004; Gevaert et al., 2006; Hahlin et al., 43
1991). 44
The included papers consist of two prospective cohort studies (Condous et al., 2004; Hahlin et al., 45
1991) and one retrospective study (Gevaert et al., 2006). The studies were conducted in the UK 46
(Condous et al., 2004; Gevaert et al., 2006) and Sweden (Hahlin et al., 1991). 47
85
All of the studies used the combination of two or more serum hCG measurements and progesterone 1
for diagnosis, and either reported measures of diagnostic accuracy, or presented data that allowed 2
calculations to be performed by the NCC-WCH technical team. One study evaluated the diagnostic 3
accuracy of an abnormal hCG score in conjunction with a progesterone concentration of less than 4
30nmol/l (Hahlin et al., 1991), and two studies evaluated the diagnostic accuracy of predictive models 5
(Condous et al., 2004; Gevaert et al., 2006). The Bayesian model (parameter prior model [PPM]) 6
incorporated the hCG ratio (ratio of concentration at 48 hours to concentration at 0 hours), 7
progesterone concentration at 48 hours, and the number of gestation days. The model M3 8
incorporated the hCG ratio, the average progesterone concentration (from measurements at 0 and 48 9
hours), and maternal age in years. The study participants were women with pain and bleeding in the 10
first trimester of pregnancy who had been classified as having a pregnancy of unknown location on 11
ultrasound. 12
measurements plus progesterone 14
Number
of
studies
Number
of
women
Sensitivity
% (95% CI)
Specificity
% (95%
CI)
Positive
predict-
ive value
% (95%
CI)
Negative
predict-
tive
value %
(95% CI)
Positive
likeli-
hood
ratio
(95% CI)
Negative
likeli-
hood
ratio
(95% CI)
Abnormal hCG score and progesterone concentration < 30nmol/l
1 study
(Hahlin,
et al.,
1991)
307 71.7
(64.7 to
78.7)*
58.8
(50.9 to
66.7)*
65.1
(58.1 to
72.2)*
65.9
(57.8 to
74.0)*
1.74
(1.40 to
2.16)*
0.48
(0.36 to
0.64)*
Low
Bayesian model (parameter prior model)
1 study
(Gevaert
et al.,
2006)
257 77
(Not
calculable
[NC])
83
(NC)
NC NC 4.5
(NC)
0.28
(NC)
Very
Low
measurements plus progesterone (Model M3) 17
Number of studies Number of women Area under the ROC
curve
(95% CI)
Quality
Model M3
1 study
(Condous et al., 2004)
195 Test set: 0.836 (0.693 to
0.979)
(Other diagnostic
accuracy measures not
reported and not
calculable)
Low
18
86
hCG score and progesterone concentration 2
One study evaluated the diagnostic accuracy of an abnormal hCG score in combination with a 3
progesterone <30nmol/l for the diagnosis of ectopic pregnancy. The study reported a low sensitivity, 4
low specificity, low PPV, low NPV, not useful positive likelihood ratio and moderately useful negative 5
likelihood ratio. The evidence for this finding was of low quality. 6
Models 7
One study evaluated the diagnostic accuracy of Bayesian model PPM for the diagnosis of ectopic 8
pregnancy. The study reported a moderate sensitivity, moderate specificity, not useful positive 9
likelihood ratio and moderate useful negative likelihood ratio; PPV and NPV were not reported. The 10
evidence for this finding was of very low quality. 11
One study evaluated the diagnostic accuracy of model M3 for the diagnosis of ectopic pregnancy. 12
Only the area under the ROC curve was reported, therefore other diagnostic accuracy measures 13
could not be assessed. The evidence for this finding was of low quality. 14
Please see section 6.7 for a combined evidence summary of all of the reviews which assess the use 16
of hCG for diagnosis. 17
measurements for viable intrauterine pregnancy 19
Review question 20
What is the diagnostic accuracy of two or more hCG measurements for determining a viable 21
intrauterine pregnancy in women with pain and bleeding and pregnancy of unknown location? 22
Six studies were included in this review (Dart et al., 1999; Daus et al., 1989; Hahlin et al., 1991; Mol et 24
al., 1998; Morse et al., 2012; Stewart et al., 1995). 25
The included papers consist of two prospective cohort studies (Hahlin et al., 1991; Mol et al., 1998) 26
and four retrospective cohort studies (Dart et al., 1999; Daus et al., 1989; Morse et al., 2012; Stewart 27
et al., 1995). The studies were conducted in the USA (Dart et al., 1999; Daus et al., 1989; Morse et 28
al., 2012; Stewart et al., 1995), the Netherlands (Mol et al., 1998) and Sweden (Hahlin et al., 1991). 29
All the studies used two or more serum hCG measurements for diagnosis, and either reported 30
measures of diagnostic accuracy, or presented data that allows calculations to be performed by the 31
NCC-WCH technical team. Four studies evaluated the percentage change in hCG over 48 hours (Dart 32
et al., 1999; Daus et al., 1989; Mol et al., 1998; Morse et al., 2012), one study evaluated the rate of 33
change of log hCG (Stewart et al., 1995), and one study evaluated the diagnostic accuracy of a 34
normal hCG score, calculated by plotting the initial hCG value against the rate of change of the serum 35
level of hCG (Hahlin et al., 1991). The study participants were women with pain and bleeding in the 36
first trimester of pregnancy who had been classified as having a pregnancy of unknown location on 37
ultrasound. 38
Evidence profile 39
Table 6.8 GRADE summary of findings for the diagnosis of viable intrauterine pregnancy using two or more hCG 40
measurements 41
Number Number Measure of diagnostic accuracy Quality
87
of
studies
of
women
Sensitivity
% (95% CI)
Specificity
% (95%
CI)
Positive
predict-
tive
value %
(95% CI)
Negative
predict-
tive
value %
(95% CI)
Positive
likeli-
hood
ratio
(95% CI)
Negative
likeli-
hood
ratio
(95% CI)
% change in serum hCG in 48 hours: rise > 35 %
1 study
(Morse
et al.,
2012)
1005 92.3
(89.0 to
95.6)
94.0
(92.3 to
95.7)
84.2
(79.9 to
88.4)
97.2
(96.0 to
98.4)
NC NC Very
low
% change in serum human chorionic gonadotrophin (hCG) in 48 hours: rise > 50 %
1 study
(Mol et
al., 1998)
195 93.3
(80.7 to
100)*
91.1
(87.0 to
95.3)*
46.7
(28.8 to
64.5)*
99.4
(98.2 to
100)*
10.50
(6.45 to
17.09)*
0.07
(0.01 to
0.49)*
Low
1 study
(Morse
et al.,
2012)
1005 82.6
(78.0 to
87.3)
97.2
(96.0 to
98.4)
91.1
(87.4 to
94.7)
94.2
(92.5 to
95.8)
NC NC Very
low
1 study
(Daus et
al., 1989)
357 87.1
(78.8 to
95.4)*
98.0
(96.4 to
99.6)*
90.0
(82.4 to
97.6)*
97.3
(95.5 to
99.2)*
42.82
(19.28 to
95.09)*
0.13
(0.07 to
0.25)*
Very
low
1 study
(Dart et
al., 1999)
307 75.5
(63.9 to
87.1)*
95.3
(92.7 to
97.9)*
76.9
(65.5 to
88.4)*
94.9
(92.2 to
97.6)*
15.97
(9.01 to
28.34)*
0.26
(0.16 to
0.41)*
Low
1 study
(Morse
et al.,
2012)
1005 72.6
(67.1 to
78.1)
98.1
(97.1 to
99.1)
93.1
(89.5 to
96.6)
91.2
(89.2 to
93.1)
NC NC Very
low
Rate of change of log hCG: > 0.11
1 study
(Stewart
et al.,
1995)
36
#
80.0
(62.5 to
97.5)**
87.6
(81.1 to
94.2)**
57.1
(38.8 to
75.5)*
95.5
(91.2 to
99.8)*
6.47
(3.65 to
11.47)*
0.23
(0.09 to
0.55)*
Very
low
Rate of change of log hCG: > 0.14
1 study
(Stewart
et al.,
1995)
36
#
65.0
(44.1 to
85.9)**
99.0
(97.0 to
100)**
92.9
(79.4 to
100)*
93.2
(88.3 to
98.1)*
63.05
(8.74 to
454.93)*
0.35
(0.19 to
0.64)*
Very
low
Normal hCG score
88
1 study
(Hahlin
et al.,
1991)
307 94.5
(89.3 to
99.7)*
91.9
(88.4 to
95.4)*
78.4
(69.8 to
87.0)*
98.2
(96.4 to
100)*
11.64
(7.54 to
17.98)*
0.06
(0.02 to
0.15)*
Low
# There were 36 women who received multiple hCG measurements, leading to a total of 117 pairs of hCG measurement used 2
for the diagnostic accuracy calculations 3
** Confidence intervals calculated by NCC-WCH technical team 4
Percentage change in serum hCG concentration in 48 hours 6
One study evaluated the diagnostic accuracy of a > 35% rise in hCG over 48 hours for the diagnosis 7
of viable intrauterine pregnancy. The study reported a high sensitivity, high specificity, moderate PPV 8
and high NPV. The study did not report likelihood ratios. The evidence for this finding was of very low 9
quality. 10
of viable intrauterine pregnancy. The study reported a high sensitivity, high specificity, low PPV, high 12
NPV, very useful positive likelihood ratio and very useful negative likelihood ratio. The evidence for 13
this finding was of low quality. 14
of viable intrauterine pregnancy. The study reported a moderate sensitivity, high specificity, high PPV 16
and high NPV. The study did not report likelihood ratios. The evidence for this finding was of very low 17
quality. 18
of viable intrauterine pregnancy. The study reported a moderate sensitivity, high specificity, high PPV, 20
low NPV, very useful positive likelihood ratio and moderately useful negative likelihood ratio. The 21
of viable intrauterine pregnancy. The study reported a moderate sensitivity, high specificity, moderate 24
PPV, high NPV, very useful positive likelihood ratio and moderately useful negative likelihood ratio. 25
The evidence for this finding was of low quality. 26
of viable intrauterine pregnancy. The study reported a low sensitivity, high specificity, high PPV and 28
high NPV. The study did not report likelihood ratios. The evidence for this finding was of very low 29
quality. 30
Rate of change of log hCG concentration 31
One study evaluated the diagnostic accuracy of a rate of change of log hCG > 0.11 for the diagnosis 32
of viable intrauterine pregnancy. The study reported a moderate sensitivity, moderate specificity, low 33
ratio. The evidence for this finding was of very low quality 35
One study evaluated the diagnostic accuracy of a rate of change of log hCG > 0.14 for the diagnosis 36
of viable intrauterine pregnancy. The study reported a low sensitivity, high specificity, high PPV, high 37
NPV, very useful positive likelihood ratio and moderately useful negative likelihood ratio. The 38
hCG score 40
One study evaluated the diagnostic accuracy of a normal hCG score for the diagnosis of viable 41
intrauterine pregnancy. The study reported a high sensitivity, high specificity, moderate PPV, high 42
NPV, very useful positive likelihood ratio and very useful negative likelihood ratio. The evidence for 43
this finding was of low quality. 44
89
Please see section 6.7 where the evidence from all of the reviews which assess the use of hCG for 2
diagnosis has been considered. 3
measurements plus progesterone for viable intrauterine 5
pregnancy 6
Review question 7
What is the diagnostic accuracy of two or more hCG measurements plus progesterone for 8
determining a viable intrauterine pregnancy in women with pain and bleeding and pregnancy of 9
unknown location? 10
One study was included in this review (Hahlin et al., 1991). The included paper is a prospective cohort 12
study, conducted in Sweden. The study used the combination of a normal hCG score and 13
progesterone > 30nmol/l for diagnosis, and included women with pain and bleeding in the first 14
trimester of pregnancy who had been classified as having a pregnancy of unknown location on 15
ultrasound. 16
Evidence profile 17
Table 6.9 GRADE summary of findings for the diagnosis of viable intrauterine pregnancy using two or more hCG 18
measurements plus progesterone 19
Number
of
studies
Number
of
women
Sensitivity
% (95% CI)
Specificity
% (95%
CI)
Positive
predict-
tive
value %
(95% CI)
Negative
predict-
tive
value %
(95% CI)
Positive
likeli-
hood
ratio
(95% CI)
Negative
likeli-
hood
ratio
(95% CI)
Normal hCG score and progesterone > 30nmol/l
1 study
(Hahlin
et al.,
1991)
307 93.2
(87.4 to
99.0)*
94.4
(91.5 to
97.4)*
84.0
(76.0 to
91.9)*
97.8
(95.9 to
99.7)*
16.77
(9.85 to
28.54)*
0.07
(0.03 to
0.17)*
Low
Evidence Statements 21
One study evaluated the diagnostic accuracy of a normal hCG score in combination with a 22
progesterone > 30nmol/l for diagnosing viable intrauterine pregnancy. The study reported a high 23
sensitivity, high specificity, moderate positive predictive value, high negative predictive value, very 24
useful positive likelihood ratio and very useful negative likelihood ratio. The evidence for this finding 25
was of low quality. 26
In these reviews, it had been hoped to identify whether the use of two or more hCG measurements 29
(with or without progesterone) was appropriate for diagnosing both an ectopic pregnancy, and a 30
viable intrauterine pregnancy. The group were therefore looking for a test which provided both high 31
sensitivity, and high specificity. However, whilst the evidence indicated a number of different ways of 32
90
documenting the change in hCG levels (measuring percentage change, measuring log change, or 1
applying various mathematical models), none of these methods showed the use of hCG to be useful 2
as a test for comprehensively and definitively diagnosing either an ectopic pregnancy or a viable 3
intrauterine pregnancy. 4
The group recognised that some of the papers considered mathematical models. Whilst they 6
understood the intrinsic potential value that such models could have, the models considered in the 7
review had not been widely validated nor did they lend themselves to ease of use. Thus it was not felt 8
appropriate to recommend their use. 9
The evidence showed that the use of the rate of log change hCG or a change in hCG levels of greater 10
than 63% both provided a high negative predictive value but a low positive predictive value. In other 11
words, these measures were effective as tests to rule out the presence of an ectopic pregnancy but 12
not effective as tests to specifically identify the presence of an ectopic pregnancy. Thus, the group felt 13
that the value of the tests would be as a risk stratification tool to determine the urgency and type of 14
care that each woman requires, with the most urgent care being focused on those women in whom an 15
ectopic pregnancy was more likely. 16
The group recognised that whilst measuring the rate of log change hCG appeared to be a slightly 17
more accurate diagnostic test than simply measuring the rate of change of hCG, the findings were 18
similar for both approaches. It was felt that measuring an increase of greater than 63% would be 19
easier to calculate and a more useful measure in clinical practice than calculating the rate of log 20
change. In addition, the findings regarding the greater than 63% change were likely to be more valid 21
as the sample size of the study was much larger. As a result, the group agreed to recommend the use 22
of an hCG increase of greater than 63%. The evidence suggested that approximately 17% women 23
with a PUL would fall into this category and would have a high chance of having a viable intrauterine 24
pregnancy. Daus et al., (1989) reported that 60/357 women with a PUL had an hCG rise of greater 25
than 63%. Of these 54/60 were ultimately diagnosed with a viable IUP. Similar findings were reported 26
by Dart et al., (1999) who found that 52/307 of women with a PUL had an hCG rise of greater than 27
66% and 40/52 of these were ultimately diagnosed with a viable IUP. 28
The group agreed that for women with an hCG increase of > 63%, it would be appropriate to offer a 29
routine ultrasound scan within 7-14 days in order to confirm the location of the pregnancy. 30
The GDG interrogated the evidence to determine an appropriate lower threshold to identify those 31
women likely to have a failing pregnancy and for whom a different management strategy could be 32
used. Of the four papers which evaluated a decline of hCG, one (Daus et al., 1999) used any 33
decline, one used a decline of 36-47% (Morse et al., 2012), whilst the other two (Mol et al., 1998; 34
Dart et al., 1999) evaluated a decline of greater than 50%. These two studies suggested that 32-36% 35
of women with a PUL would fall into the latter category (63/195 and 109/307 respectively), and were 36
therefore likely to have a failing pregnancy. In these studies none of these women had a viable 37
intrauterine pregnancy and the risk of an ectopic pregnancy was low (about 1% when both studies 38
were combined). In Morse et al (2012), which used a threshold for decline of 36-47%, a higher 39
proportion of women (over 3%) were misclassified as having a miscarriage when they were eventually 40
diagnosed with an ectopic pregnancy. From these data and their own clinical experience, the group 41
felt that a decline in hCG levels of greater than 50% was highly likely to indicate a failing pregnancy 42
and that this was an appropriately conservative threshold to use. 43
The group agreed that for women with a decline in hCG levels of greater than 50% it would not be 44
necessary to conduct a repeat ultrasound scan, but instead that they should be asked to do a urine 45
pregnancy test in two weeks time. If this was negative and the woman was asymptomatic no further 46
action would be necessary. However, if it were positive then the woman should return to the 47
dedicated early pregnancy service for a further clinical review and individualised management by a 48
gynaecologist. 49
For the remaining group of women (i.e. those with an hCG level change between a decline of less 50
than or equal to 50% and a rise of less than or equal to 63%) it was agreed that review by a 51
gynaecologist would be warranted. 52
91
The group wished to highlight that hCG should not be used for a confirmatory diagnosis and that final 1
confirmation could only be provided by either an ultrasound scan, or a negative pregnancy test (to 2
identify a failed pregnancy). 3
The group considered the evidence that was available for the use of progesterone levels in 4
conjunction with hCG. They noted that for both diagnosis of ectopic pregnancy and diagnosis of viable 5
intrauterine pregnancy, there was little or no improvement in the negative predictive value of the tests 6
as compared with using hCG alone. 7
The group felt it important to stress the importance of symptoms over hCG. The group agreed that all 8
women, regardless of their hCG level, should be given written information about what to do if they 9
experience any new or worsening symptoms, including details about how to access emergency care. 10
The GDG felt that the use of large numbers of hCG measurements was not an effective use of 12
resources, both in terms of the womans care and the cost of the tests. They wanted to avoid large 13
numbers of tests being performed without a diagnosis being made, and agreed that the use of more 14
than two hCG measurements should only be undertaken following review by a specialist health care 15
professional. Considering that there was no evidence of any added value of progesterone in making a 16
diagnosis and the performing the test has associated costs, the GDG agreed that progesterone 17
should not be used in the assessment of women with pain and bleeding and a PUL. 18
The quality of the evidence was either low or very low. The GDG was aware of the limitations of the 20
studies but ultimately agreed that there was sufficient evidence to make recommendations. 21
The GDG felt that it was important that women be given a realistic likely prognosis for their 23
pregnancy, based on their hCG levels, but also be informed that further confirmation would be 24
needed. For women whose pregnancies might be unlikely to continue, they thought that this would 25
avoid giving women false hope and ensure that they felt informed about the likely outcome. For these 26
women, the group also felt that it was very important that they be given information about where and 27
how to access support and counselling services. 28
Recommendations 29
37 Assume women with a pregnancy of unknown location have an ectopic pregnancy
until the location is determined.
38 In women with a pregnancy of unknown location, place more importance on clinical
symptoms than on human chorionic gonadotrophin (hCG) results, and review the
womans condition if any of her symptoms change, regardless of previous results
and assessments.
39 Use hCG measurements only for determining trophoblastic proliferation, to help
decide ongoing monitoring.
40 Do not use hCG measurements to determine the location of the pregnancy.
41 Regardless of hCG levels, give women with a pregnancy of unknown location
written information about what to do if they experience any new or worsening
symptoms, including details about how to access emergency care 24 hours a day.
Advise women to return if there are new symptoms or if existing symptoms worsen.
42 Take two serum hCG measurements 48 hours apart to determine subsequent
management of a pregnancy of unknown location. Take further measurement only
after review by a senior healthcare professional.
43 Inform a woman with an increase in serum hCG concentration greater than 63%
that she is likely to have a viable intrauaterine pregnancy. Offer her a tranvaginal
92
ultrasound scan between 7 and 14 days of the serum hCG test to confirm this:
If a viable intrauterine pregnancy is confirmed, offer her routine antenatal
care
If a viable intrauterine pregnancy is not confirmed, refer her for urgent
clinical review by a gynaecologist.
44 For a woman with a decrease in serum hCG concentration greater than 50%:
inform her that the pregnancy is unlikely to continue but that this is not
confirmed
provide her with written and verbal information about where she can access
support and counselling services
ask her to take a urine pregnancy test 14 days after the serum hCG test.
explain that:
o If the test is negative, no further action is necessary.
o If the test is positive, she should return for urgent clinical review by
a gynaecologist.
45 For a woman with a change in serum hCG concentration between a 50% decline
and 63% rise inclusive:
refer her for urgent clinical review by a gynaecologist and individualised
management
refer her for a repeat scan between 7 and 14 days
ask her to report worsening symptoms without delay.
46 For women with a pregnancy of unknown location, do not use progesterone
measurements together with hCG measurement as a diagnostic test for either
viable intrauterine pregnancy or ectopic pregnancy.



7 Management of 1
threatened miscarriage 2
and miscarriage 3
7.1 Introduction 4
Threatened miscarriage is the commonest complication of early pregnancy, occurring in 5
approximately 20% of pregnant women before 20 weeks gestation (Sotiriadis et al, 2004). Although 6
many women who have threatened miscarriage go on to have a successful pregnancy there is an 7
increase in risk of miscarriage of 2.6 times and 17% of women with threatened miscarriage go on to 8
have further pregnancy complications. In the UK, it is estimated that around a quarter of a million 9
pregnancies each year end in a miscarriage (The Miscarriage Association, 2011). This loss is 10
associated with a significant amount of physical and psychological morbidity. This chapter presents 11
evidence and guidance for clinically and cost effective care for women with miscarriage considering 12
both clinical and psychological outcomes. 13
Review question 15
What is the effectiveness of progesterone in improving outcomes in women with threatened 16
miscarriage? 17
Introduction 18
Progesterone is an essential hormone secreted by the corpus luteum that provides early pregnancy 19
support until placental production takes over at 10 to 12 weeks gestation. Historically, low levels of 20
circulating progesterone have been linked to impending miscarriage and the presence of associated 21
vaginal bleeding. It has been postulated, therefore, that a lack of progesterone is a cause of 22
miscarriage rather than a secondary signal of failing pregnancy. 23
This review analyses the evidence from published studies where progesterone/progestogen 24
supplementation has been introduced in pregnancies complicated by first trimester threatened 25
miscarriage (presence of vaginal bleeding before 12+6 weeks gestation). Various outcomes were 26
examined to determine any detrimental effect or proven efficacy. 27
Six studies were included in this review (Duan et al., 2010; El-Zibdeh et al., 2009; Pandian, 2009; 29
Gerhard et al., 1987; Omar et al., 2005; Palagiano, et al. 2004). 30
Four studies are randomised trials (El-Zibdeh et al., 2009; Pandian, 2009; Gerhard et al., 1987; 31
Palagiano, et al. 2004) and two are observational studies (Duan et al., 2010; Omar et al., 2005). One 32
study was conducted in Jordan (El-Zibdeh et al., 2009), two in Malaysia (Pandian, 2009; Omar et al., 33
2005), one in Germany (Gerhard et al., 1987), one in China (Duan et al., 2010), and one in Italy 34
(Palagiano et al., 2004). 35
Three included studies (Duan et al., 2010; Gerhard et al., 1987; Palagiano, et al. 2004) assessed the 36
efficiency of progesterone administration in women with bleeding in early pregnancy and three other 37
94
studies (El-Zibdeh et al., 2009; Pandian, 2009; Omar et al., 2005) evaluated the effect of progestogen 1
(dydrogesterone) on pregnancy outcomes for threatened miscarriages. Route of administration varied 2
and consisted of intramuscular administration in one study (Duan et al., 2010), oral in three studies 3
(El-Zibdeh et al., 2009; Pandian, 2009; Omar et al., 2005) and vaginal pessary in two studies 4
(Gerhard et al., 1987; Palagiano et al., 2004). 5
Evidence profile 6
Table 7.1 GRADE summary of findings for comparison of progesterone with no treatment or placebo 7
Number of
studies
Number of women or mean SD Effect Quality
Progesterone /
progestogen
No treatment /
placebo
Relative
(95% CI)
Absolute
(95% CI)
Term birth
1 meta-
analysis of 2
studies
(El-Zibdeh et
al., 2009;
Gerhard et al.,
1987)
88/112
(78.6%)
61/86
(70.9%)
RR 1.12
(0.95 to 1.32)
85 more per
1000
(from 35 fewer
to 227 more)
Low
Preterm birth
1 meta-
analysis of 2
studies
(El-Zibdeh et
al., 2009
Pandian,
2009)
12/182
(6.6%)
9/155
(5.8%)
RR 1.10
(0.48 to 2.52)
6 more per 1000
(from 30 fewer
to 88 more)
Low
1 study
(Duan et al.,
2010)
66/532
(12.4%)
2257/21054
(10.7%)
RR 1.16
(0.92 to 1.46)
17 more per
1000
(from 9 fewer to
49 more)
Very low
Miscarriage (any route)
1 meta-
analysis of 4
studies
(El-Zibdeh et
al., 2009;
Gerhard et al.,
1987;
Palagiano et
al., 2004;
Pandian,
2009)
31/224
(13.8%)
51/197
(25.9%)
RR 0.53
(0.35 to 0.79)
122 fewer per
1000
(from 54 fewer
to168 fewer)
Low
1 study
(Omar et al.,
2005)
3/74
(4.1%)
11/80
(13.8%)
RR 0.29 (0.09 to
1.02)
P = 0.05*
98 fewer per
1000
(from 125 fewer
to 3 more)
Very low
Management of threatened miscarriage and miscarriage
95
Miscarriage in women with vaginal bleeding (stratified analysis)
1 study
(Omar et al.,
2005)
2/29
(6.9%)
6/37
(16.2%)
RR 0.43
(0.09 to 1.95)
92 fewer per
1000
(from 148 fewer
to 154 more)
Very low
Miscarriage in women with vaginal spotting (stratified analysis)
1 study
(Omar et al.,
2005)
1/45
(2.2%)
5/43
(11.6%)
RR 0.19
(0.02 to 1.57)
94 fewer per
1000
(from 114 fewer
to 66 more)
Very low
Miscarriage in women with fetal heart activity (stratified analysis)
1 study
(Omar et al.,
2005)
1/31
(3.2%)
3/34
(8.8%)
RR 0.37
(0.04 to 3.33)
56 fewer per
1000
(from 85 fewer
to 206 more)
Very low
Miscarriage in women with presence of yolk sac (stratified analysis)
1 study
(Omar et al.,
2005)
0/23
(0%)
1/25
(4%)
RR 0.36
(0.02 to 8.45)
26 fewer per
1000
(from 39 fewer
to 298 more)
Very low
Miscarriage in women with regular intrauterine gestational sac (stratified analysis)
1 study
(Omar et al.,
2005)
2/7
(28.6%)
3/5
(60%)
RR 0.48
(0.12 to 1.88)
312 fewer per
1000
(from 528 fewer
to 528 more)
Very low
Miscarriage (oral progesterone)
1 meta-
analysis of 2
studies
(El-Zibdeh et
al., 2009;
Pandian,
2009)
27/182
(14.8%)
42/155
(27.1%)
RR 0.54
(0.35 to 0.84)
125 fewer per
1000
(from 43 fewer
to 176 fewer)
Low
Miscarriage (vaginal progesterone)
1 meta-
analysis of 2
studies
(Gerhard et al.,
1987;
Palagiano et
al., 2004)
4/42
(9.5%)
9/42
(21.4%)
RR 0.47
(0.17 to 1.30)
114 fewer per
1000
(from 178 fewer
to 64 more)
Low
Pregnancy rate at 20 weeks
1 study
(Pandian,
2009)
84/96
(87.5%)
68/95
(71.6%)
RR 1.22
(1.05 to 1.42)
157 more per
1000
(from 36 more
Low
96
to 301 more)
1 study
(Omar et al.,
2005)
71/74
(95.9%)
69/80
(86.3%)
RR 1.11
(1.01 to 1.23)
95 more per
1000
(from 9 more to
198 more)
Very low
Pregnancy rate at 20 weeks in women with vaginal bleeding (stratified analysis)
1 study
(Omar et al.,
2005)
27/29
(93.1%)
31/37
(83.8%)
RR 1.11
(0.93 to 1.32)
92 more per
1000
(from 59 fewer
to 268 more)
Very low
Pregnancy rate at 20 weeks in women with vaginal spotting (stratified analysis)
1 study
(Omar et al.,
2005)
44/45
(97.8%)
38/43
(88.4%)
RR 1.11
(0.98 to 1.24)
97 more per
1000
(from 18 fewer
to 212 more)
Very low
Pregnancy rate at 20 weeks in women with fetal heart activity (stratified analysis)
1 study
(Omar et al.,
2005)
30/31
(96.8%)
31/34
(91.2%)
RR 1.06
(0.94 to 1.2)
55 more per
1000
(from 55 fewer
to 182 more)
Very low
Pregnancy rate at 20 weeks in women with presence of yolk sac (stratified analysis)
1 study
(Omar et al.,
2005)
23/23
(100%)
24/25
(96%)
RR 1.04
(0.93 to 1.16)
38 more per
1000
(from 67 fewer
to 154 more)
Very low
Placental abruption
1 study
(Duan et al.,
2010)
5/532
(0.94%)
153/21054
(0.73%)
RR 1.29
(0.53 to 3.14)
2 more per 1000
(from 3 fewer to
16 more)
Very low
Hypertensive disorders in pregnancy
1 meta-
analysis of 2
studies
(El-Zibdeh et
al., 2009;
Pandian,
2009)
19/182
(10.4%)
17/155
(11%)
RR 1.00
(0.54 to 1.88)
0 fewer per
1000
(from 50 fewer
to 97 more)
Low
1 study
(Duan et al.,
2010)
16/532
(3%)
974/21054
(4.6%)
RR 0.66
(0.41 to 1.08)
16 fewer per
1000
(from 27 fewer
to 4 more)*
Very low
Gestational diabetes
1 study
(Duan et al.,
37/532
(7%)
1141/21054
(5.4%)
RR 1.28
(0.94 to 1.76)
15 more per
1000
Very low
97
2010) (from 3 fewer to
41 more)
Intrahepatic cholestasis of pregnancy
1 study
(Duan et al.,
2010)
51/532
(9.6%)
1712/21054
(8.1%)
RR 1.18
(0.9 to 1.54)
15 more per
1000
(from 8 fewer to
44 more)
Very low
Pain score at the end of 5 day treatment (mean SD)
1 study
(Palagiano et
al., 2004)
0.4 0.7
n = 25
2.4 0.8
n = 25
not calculable MD 2.0 lower
(2.42 lower to
1.58 lower)
p < 0.001
Moderate
* p = 0.05 calculated by RevMan and p = 0.03 reported in the paper 1
Term Birth 3
One meta-analysis of two studies found no statistically significant difference in term birth in women 4
who received progesterone/progestogen treatment compared with women who had no treatment. The 5
Preterm birth 7
One meta-analysis of two studies and one further study found no statistically significant difference in 8
preterm birth in women who received progesterone/progestogen treatment compared with women 9
who had no treatment. The evidence for this finding was of low and very low quality. 10
Miscarriage 11
One meta-analysis of four studies found a reduced incidence of miscarriage in women who received 12
progesterone/progestogen treatment compared with women who had no treatment. This finding was 13
statistically significant. The evidence for this finding was of low quality. One further study found that 14
incidence of miscarriage was lower in women who received progestogen treatment compared with 15
women who had no treatment. This finding was statistically significant. When stratified sub-group 16
analyses were conducted for women with vaginal bleeding, women with vaginal spotting, women with 17
fetal heart activity, women with presence of yolk sac, and women with regular intrauterine gestational 18
sac the findings favoured the progesterone group but the difference was no longer statistically 19
significant. The evidence for this finding was of very low quality. 20
Miscarriage (oral progesterone) 21
One meta-analysis of two studies found a reduced incidence of miscarriage in women who received 22
oral progestogen treatment compared with women who had no treatment. This finding was statistically 23
significance. The evidence for this finding was of low quality. 24
Miscarriage (vaginal progesterone) 25
One meta-analysis of two studies found no statistically significant difference in incidence of 26
miscarriage in women who received vaginal progesterone treatment compared with women who had 27
no treatment. The evidence for this finding was of low quality. 28
Pregnancy rate at 20 weeks 29
Two studies found that the rate of pregnancy at 20 weeks was higher in women who received 30
progestogen treatment compared with women who had no treatment. This finding was statistically 31
significant in both studies. The evidence for this finding was of low quality in one study and very low in 32
the other. When stratified sub-group analyses were consucted for women with vaginal bleeding, 33
women with vaginal spotting, women with fetal heart activity, and women with presence of yolk sac 34
the findings favoured progesterone treatment but the difference was no longer statistically significant. 35
The evidence for these findings was of very low quality. 36
98
Placental abruption 1
One study found no statistically significant difference in incidence of placental abruption in women 2
who received progesterone treatment compared with women who had no treatment. The evidence for 3
this finding was of very low quality. 4
Hypertensive disorders in pregnancy 5
One meta-analysis of two studies and one further study found no statistically significant difference in 6
incidence of hypertensive disorders of pregnancy in women who received progesterone/progestogen 7
treatment compared with women who had no treatment. The evidence for this finding was of very low 8
quality. 9
Gestational diabetes 10
One study found no statistically significant difference in incidence of gestational diabetes in women 11
who received progesterone treatment compared with women who had no treatment. The evidence for 12
this finding was of very low quality. 13
Intrahepatic cholestasis of pregnancy 14
One study found no statistically significant difference in intrahepatic cholestasis of pregnancy in 15
women who received progesterone treatment compared with women who had no treatment. The 16
Pain score at the end of 5 day treatment 18
One study found that the mean pain score at the end of the five days of treatment was lower in 19
women who received progesterone treatment compared with women who had placebo treatment. 20
This finding was statistically significant. The evidence for this finding was of moderate quality. 21
Health economics 22
The cost-effectiveness analysis undertaken for this guideline suggested that progesterone for 23
threatened miscarriage was cost-effective when compared with no treatment. In the base case 24
analysis progesterone dominated no treatment, producing an incremental QALY gain and a cost 25
saving of 49. Progesterone was cost saving because the savings from averted miscarriage more 26
than offset treatment costs. Probabilistic sensitivity analysis found progesterone to be cost-effective in 27
99.93% of Monte Carlo simulations. In sensitivity analysis when a much higher treatment cost of 200 28
was assumed, progesterone still had an 83% probability of being cost-effective at a willingness to pay 29
threshold of 20,000 per QALY. The model is described in more detail in section 9.2. 30
The group agreed that the most important outcomes were the rate of term pregnancy, miscarriage, 33
and pregnancy rate beyond 20 weeks gestation. They also recognised that the side effects 34
associated with progesterone treatment is an important outcome. 35
The RCT evidence showed that progesterone treatment and placebo or no treatment of miscarriage 37
had a similar effect on term and preterm birth and also on the clinical or medical side effects in 38
women with threatened miscarriages. Progesterone or dydrogesterone treatment was significantly 39
associated with fewer miscarriages, less severe pain, and higher rate of pregnancy at 20 weeks 40
gestation, in women with threatened miscarriages. A significant difference in favour of progesterone 41
for the pregnancy rate at 20 weeks was also shown in a small and underpowered observational study. 42
However, when stratified analysis was performed based on vaginal bleeding, vaginal spotting and 43
presence of yolk sac, the result was no longer statistically significant though this could be due to the 44
very small sample size. 45
The GDG acknowledged the importance of the chosen end point for each of the included studies. The 46
studies chose different end points (some at 20 weeks gestation, others at birth), but none adopted an 47
end point beyond the birth and none included neonatal congenital abnormalities as an outcome. The 48
GDG were concerned that routine administration of progesterone might interfere with natural 49
99
miscarriages that were associated with genetic abnormalities in the fetus. This in turn could result in 1
an increase in the rate of neonatal abnormalities or later miscarriages. 2
The evidence showed that progesterone treatment and placebo or no treatment of threatened 3
miscarriage had similar effects on complications of pregnancy (gestational diabetes, hypertensive in 4
pregnancy and Intrahepatic cholestasis of pregnancy). 5
Separately from the issue of whether or not progesterone should be offered, the GDG agreed that 6
there should be a follow-up procedure in place for women with a threatened miscarriage. They agreed 7
that if the bleeding gets worse, women should be advised to return in order to receive further 8
assessment. They also agreed that there should be a follow-up scan if the bleeding persists in order 9
to determine whether or not the pregnancy is still viable. The group wished to strike a balance 10
between offering women reassurance, but also ensuring that not all women would need to return for a 11
scan. Ultimately, it was agreed that 14 days would be an appropriate time to wait before offering a 12
follow-up scan. 13
The economic evaluation undertaken for this guideline demonstrated that progesterone treatment was 15
likely to be a cost-effective treatment for women with threatened miscarriage as it not only reduces 16
the threat of miscarriage but also produces net savings as a result of averted miscarriage. However, 17
the GDG had reservations about some of the studies which informed the treatment effect size in the 18
health economic model. Furthermore, in the UK, it is not usual practice to offer progesterone for 19
threatened miscarriage, and the group had concerns about recommending a significant change in 20
practice based on such poor quality evidence. 21
The evidence available for this question was generally of low or very low quality. The few studies that 23
were included generally had a low numbers of participants, and the GDG was concerned that even 24
the randomised trials that were included were potentially subject to bias. Two of the trials were funded 25
by manufacturers of progesterone, including one trial with a single author and which had a high loss 26
to follow up. 27
Overall, the GDG felt that the evidence was insufficient to recommend the routine use of 28
progesterone, particularly as there was no demonstrated significant difference in the rate of term birth. 29
However, given that there was some evidence of benefit, that the treatment was cost-effective and 30
there was no evidence of harm; they felt that a womans choice would be an important consideration 31
and that it was appropriate to provide this option. The group felt strongly that further, high quality 32
studies investigating the efficacy of progesterone were needed, and decided that this was a priority 33
area for research, particularly considering outcomes of term birth, late miscarriage and incidence of 34
congenital abnormalities. 35
Given their reservations about the quality of the studies, the GDG thought that it was important that 37
women were made aware that the evidence for the efficacy of progesterone is not strong. They felt 38
that this would avoid giving women false hope about the prognosis for their pregnancy and facilitate 39
them making an informed choice about whether or not to take progesterone/progestogen. As detailed 40
above the provision of follow-up support for women with worsening or persistant bleeding was also 41
considered important and a recommendation made to reflect this. 42
Recommendations 43
47 Consider progesterone/progestogen for women with a threatened miscarriage (that
is, vaginal bleeding and a confirmed intrauterine pregnancy with a fetal heartbeat).
***

48 Inform women that although there is some evidence that progesterone/progestogen

*** At the time of publication (June 2012), progeterone/progestogen did not have UK marketing authorisation for this indication.
100
can prevent a miscarriage, this evidence is not strong.

49 Advise a woman with threatened miscarriage that:
if her bleeding gets worse, or persists beyond 14 days, she should return
for further assessment.
if the bleeding stops, she should book for, or return to, routine antenatal
care.
1
2
RR 5 Is progesterone/progestogen effective in treating threatened miscarriage?

Approximately 20% of pregnancies miscarry in the first trimester and many women
will experience some bleeding and/or pain in early pregnancy that does not cause
miscarriage. In many countries, women with bleeding and/or pain will be treated
with progesterone or progestogens to try and decrease the risk of miscarriage. The
evidence for the effectiveness of this treatment has been inconclusive, but data
from a meta-analysis of several small studies suggest that progestogens are better
than placebo. However, there are theoretical risks to prescribing any treatment in
pregnancy and for many practitioners this will be a major change in practice. The
lack of strong evidence makes this a priority area for research.
A multicentre randomised controlled trial of at least 800 women treated with either
progesterone/progestogen or placebo should be conducted. The population would
be women with pain and bleeding and a spontaneous, confirmed, viable, singleton,
intrauterine pregnancy between 612 weeks gestation. Progesterone/progestogen
or placebo would be administered from when bleeding starts until the end of the
13th week. Pregnancy proceeding beyond the end of the first trimester might be the
primary outcome. Ideally, live birth would be measured, but, this will make the trial
much more expensive, and a majority of pregnancies that are going to miscarry will
do so in the first trimester. However, pregnancy outcome should be a secondary
outcome, along with gestation at birth and presence of congenital abnormalities.
3
7.3 Expectant management compared with active 4
treatment of miscarriage 5
Review question 6
How effective is expectant management of miscarriage compared with active treatment for improving 7
womens clinical and psychological outcomes? 8
Introduction 9
Although historically miscarriages were often treated with a surgical procedure, there are now other 10
less invasive options available, in the form of medical treatment or expectant management. However, 11
the ability of women to access each mode of management varies across England and Wales, a fact 12
which could be attributed to uncertainty about their relative efficacy and risk of complications. 13
Therefore, these reviews aimed to establish which treatment option is the most clinically and cost 14
effective, recognising the importance of womens psychological outcomes. This includes establishing 15

101
whether simply allowing the natural process of miscarriage to complete its course is just as good as 1
medical or surgical treatment. 2
Twelve studies were included in this review (Blohm et al., 1997; Chipchase & James, 1997; Ngai et 4
al., 2001; Nielsen & Hahlin, 1995; Nielsen et al., 1996; Nielsen et al., 1999; Shelley et al., 2005; Smith 5
et al., 2006; Smith et al., 2009; Trinder et al., 2006; Wieringa-de Waard et al., 2002a; Wieringa-de 6
Waard et al., 2002b). 7
One qualitative paper (Smith et al., 2006) and the following eleven reports of seven RCTs were 8
included (Blohm et al., 1997; Chipchase & James, 1997; Ngai et al., 2001; Nielsen & Hahlin, 1995; 9
Nielsen et al., 1996; Nielsen et al., 1999; Shelley et al., 2005; Smith et al., 2009; Trinder et al., 2006; 10
Wieringa-de Waard et al., 2002a; Wieringa-de Waard et al., 2002b). 11
The RCT papers report the outcomes and follow-up data of seven trials, conducted in the UK (2 trials: 12
Chipchase & James, 1997 and Smith et al., 2009 / Trinder et al., 2006), Australia (Shelley et al., 13
2005), Sweden (2 trials: Nielsen & Hahlin, 1995 / Nielsen et al., 1996 / Blohm et al., 1997 and Nielsen 14
et al., 1999), The Netherlands (Wieringa-de Waard et al., 2002a / Wieringa-de Waard et al., 2002b) 15
and Hong Kong (Ngai et al., 2001). The qualitative study is the follow-up to an RCT conducted in the 16
UK, including both participants and non-participants of the trial (Smith et al., 2006). 17
All studies compared expectant management with medical and/or surgical management of 18
miscarriage (both of which in isolation or in combination were defined as active by the GDG), and 19
reported at least one priority outcome. The trials were all conducted in developed countries, and their 20
populations include women with missed miscarriages and/or women with ongoing miscarriages. One 21
RCT (Wieringa-de Waard et al., 2002b) also included a group of women who did not accept 22
randomisation but instead chose to be managed according to their preference. They were followed up 23
in exactly the same way as the randomised group and their outcomes were analysed separately and 24
compared to the randomised group. 25
Evidence profile 26
Thirteen outcomes (grouped in seven broad categories) were chosen by the GDG as being of priority 27
to inform recommendations. 28
Heterogeneity was low (under 60%) for all outcomes except for bleeding duration (2 trials) where it 29
was 86%; therefore for this outcome, the NCC-WCH technical team used a random effects model in 30
the meta-analysis. The high heterogeneity could be the result of the fact that one of the trials 31
compared expectant with medical management whereas the other compared expectant with surgical 32
management. 33
Table 7.2 GRADE summary of findings for comparison of expectant management with active treatment 34
Number of
studies
Number of women or average Effect Quality
Expectant Active Relative
(95% CI)
Absolute
(95% CI)
Need for unplanned intervention*
1 meta-
analysis of 6
studies
(Ngai et al.,
2001; Nielsen
& Hahlin,
1995; Nielsen
et al., 1999;
Shelley et al.,
2005; Trinder
238/672
(35.4%)
181/1020
(17.7%)
RR 2.28
(1.93 to 2.7)
227 more per
1000
(from 165 more
to 302 more)
High
102
et al., 2006;
Wieringa-de
Waard et al.,
2002a)
Infection (incidence up to 15 days)
1 meta-
analysis of 7
studies
(Chipchase &
James, 1997;
Ngai et al.,
2001; Nielsen
& Hahlin,
1995; Nielsen
et al., 1999;
Shelley et al.,
2005; Trinder
et al., 2006;
Wieringa-de
Waard et al.,
2002a)
17/691
(2.5%)
30/1038
(2.9%)
RR 0.82
(0.46 to 1.44)
5 fewer per
1000
(from 16 fewer
to 13 more)
High
Gastrointestinal side effects (number of events)
1 study
(Ngai et al.,
2001
12/87
(13.8%)
25/90
(27.8%)
RR 0.5
(0.27 to 0.92)
139 fewer per
1000
(from 22 fewer
to 203 fewer)
High
Need for a blood transfusion
1 meta-
analysis of 4
studies
(Ngai et al.,
2001; Shelley
et al., 2005;
Trinder et al.,
2006;
Wieringa-de
Waard et al.,
2002a)
8/507
(1.6%)
4/911
(0.4%)
RR 3.39 (1.08 to
10.61)
10 more per
1000
(from 0 more to
42 more)
High
Duration of bleeding (days)
1 study
(Trinder et al.,
2006)
Median 12
(IQR 7 to 15)
n=398
Medical:
Median 11
(IQR 7 to 15)
n=398
Surgical:
Median 8
(IQR 4 to 14)
n=402
not calculable
(NC)
Expectant vs.
medical
Median 1 higher
(confidence
interval NC)
p (NC)
Expectant vs.
surgical
Median 4 higher
(confidence
High
103
interval NC)
p < 0.0001
1 meta-
analysis of 2
studies
(Nielsen &
Hahlin, 1995;
Nielsen et al.,
1999)
n=179 n=134 NC MD 0.28 higher
(1.64 lower to
2.20 higher)
Moderate
1 study
(Ngai et al.,
2001)
Mean 15
(SD not
reported (NR))
n=29
Mean 14.6
(SD NR)
n=30
NC MD 0.4 higher
(confidence
intervals NC)
p value NR
Moderate
1 study
(Wieringa-de
Waard et al.,
2002a)
Median 17
(IQR 10 to 26
n=64
Median 13
(IQR 9 to 17)
n=58
NC Median 4 higher
(confidence
interval NC)
p=0.04
Moderate
1 study
(Chipchase &
James, 1997)
Median 4
(range 0 to 7)
n=19
Median 2
(range 0 to 7)
n=16
NC Median 2 higher
(confidence
interval NC)
NS (p value NR)
Moderate
Pain: duration (days)
1 study
(Nielsen &
Hahlin, 1995)
Mean 1.92
(SD 1.47)
n=103
Mean 1.69
(SD 1.46)
n=52
NC MD 0.230
higher
(0.263 lower to
0.723 higher)
NS (p > 0.03)
Moderate
1 study
(Wieringa-de
Waard et al.,
2002a)
Median 14
(IQR 7 to 24)
n=64
Median 11
(IQR 6 to 26)
n=58
NC Median 3 higher
(confidence
interval NC)
NS (p value not
reported)
Moderate
1 study
(Chipchase &
James, 1997)
Median 0
(range 0 to 5)
n=19
Median 0
(range 0 to 2)
n=16
NC Median 0 higher
(confidence
interval NC)
NS (p value NR)
Moderate
1 study
(Shelley et al.,
2005)
Median 3.0
(range 0.0 to
11.0)
n=15
Medical:
Median 3.0
(range 0.2 to
16.0)
n=11
Surgical:
Median 2.0
NC Expectant vs.
medical
Median 0 higher
(confidence
interval NC)
Expectant vs.
surgical
Median 1 higher
Moderate
104
(range 0.2 to
12.0)
n=12
(confidence
interval NC)
p values NR
Pain severity
1 study
(Nielsen et al.,
1999)
Mean 62.0
(SD 30.1)
n=62
Mean 66.1
(SD 26.3)
n=60
NC MD 4.10 lower
(12.97 lower to
4.77 higher)
NS (p value NR)
Moderate
1 study
(Shelley et al.,
2005)
Median 3
(range 1 to 7)
n=15
Medical:
Median 3
(range 1 to 8)
n=11
Surgical:
Median 3
(range 1 to 10)
n=12
NC Median
difference NC
(confidence
interval NC)
NS between 3
groups but p
values NR
Moderate
Unplanned admissions**
1 study
(Trinder et al.,
2006)
196/398
(49.2%)
104/800
(13%)
RR 3.79
(3.09 to 4.65)
363 more per
1000
(from 272 more
to 475 more)
High
Womens satisfaction
1 study
(Nielsen et al.,
1999)
Mean 25.2
(SD 25.6)
n=62
Mean 28.6
(SD 24.8)
n=60
NC MD 3.40 lower
(11.32 lower to
4.52 higher)
P = 0.174
Moderate
1 study
(Chipchase &
James, 1997)
19/19
(100%)
14/16
(87.5%)
RR 1.14
(0.95 to 1.38)
125 more per
1000
(44 fewer to 388
more)
Low
Anxiety
1 study
(Nielsen et al.,
1996)
Mean 57.5
(SD 12.4)
n=58
Mean 57.5
(SD 14.0 )
n=28
NC MD 0.00 higher
(5.92 lower to
5.92 higher)
P > 0.30
Moderate
1 study
(Wieringa-de
Waard et al.,
2002b)
Mean/SD NR
n=46
Mean/SD NR
n=36
NC MD NC
(confidence
interval NC)
P = 0.09
Low
1 study
(Shelley et al.,
3/15
(20%)
5/22
(22.7%)
RR 0.88
(0.25 to 3.14)
27 fewer per
1000
(171 fewer to
Low
105
2005) 486 more)
Mental health
1 study
(Wieringa-de
Waard et al.,
2002b)
Mean/SD NR
n=46
Mean/SD NR
n=36
NC MD 7.4 in favour
of expectant
(confidence
interval NC)
P = 0.004
Low
1 study
(Shelley et al.,
2005)
Mean 37.1
(SD 13.0)
n=15
Mean 39.3
(SD 14.2)
n=22
NC MD 2.2 lower
(11.54 lower to
7.14 higher)
Low
Live birth rate
1 study
(Smith et al.,
2009)
177/224
(79.0%)
373/465
(80.2%)
RR 0.99
(0.91 to 1.07)
12 fewer per
1000
(from 74 fewer
to 55 more)
Moderate
Fertility
1 study
(Chipchase &
James, 1997)
9/12
(75%)
6/9
(66.7%)
RR 1.13
(0.64 to 1.98)
83 more per
1000
(from 241 fewer
to 654 more)
Low
1 study
(Blohm et al.,
1997)
Cumulative
conception rate:
0.7
Cumulative
conception rate:
0.6
NC NS
(P value NR)
Low
*Unplanned interventions include: surgery (including repeat surgery) due to first-line treatment failure, emergency surgery prior 1
to allocated treatment, surgical completion on maternal request, or treatment to deal with a complication of the initial treatment 2
**Unplanned admission is an admission to hospital during the trial that was not pre-specified in the methodology as part of the 3
management protocol 4
Two studies undertook additional data collection and analysis as a follow-up to an RCT. One of these 5
is a UK study that includes qualitative data collection and analysis (Smith et al, 2006); the other was 6
conducted in the Netherlands and compares quality of life and anxiety scores for both randomised 7
and not randomised women taking part in a trial (Wieringa-de Waard et al, 2002b). Findings from 8
these studies are summarised in table 6.3 with quotations used to illustrate key themes identified from 9
the qualitative research. 10
Table 7.3 Additional findings evaluating expectant and active miscarriage management strategies 11
Qualitative data on emotional/psychological outcomes and womens satisfaction with management
1 study
(Smith et al.,
2006)
(High*)
Areas with general consensus:
Fear: There was near uniform fear of intervention, especially anaesthetic and a perception of
hospitalisation and surgery as traumatic events.
"...yeah... I didn't really want to have anything done. I thought it was bad enough having lost it,
without having to have any more fiddling around."
Predictability: Women wanted a predictable end, so they could get on with their lives, and they
wanted their management and symptoms to have a predictable course.
And it was like: I wanted it done, I wanted it done now. I wanted to get
home for tea, sort of thing, that was how I was: cant we just do it.
106
Need for more information: Women felt they did not know what to expect in terms of bleeding
and pain, and wanted more details on the timing, duration and effects of interventions.
I wanted to. I didnt want to sort of just go home and wait for a
miscarriage, erm, ... because I, I didnt know what to expect at all.
Areas with wider variation in responses:
Appropriateness: Some women queried whether intervention was necessary and wished to be
allowed to miscarry naturally themselves, whereas others were in favour of something being
done to help expedite completion.
I didnt want a D & C, I didnt I know it sounds silly, cos the baby was
already dead, but I dont agree with abortion, and things like that, and to
me it felt the same; I wanted to do it on my own, and I got the D & C.
I remember thinking about the three options, and coming to the
conclusion that, at least a D & C was quick because at the time Id
been off work for 3 weeks already and I just thought: I dont want to
wait anymore, particularly as I dont know whats going to happen.
Awareness of the event: Some women felt benefit in experiencing the event, to allow them to
say goodbye, whereas others preferred surgery to avoid consciousness of the miscarriage.
its very clean, very quick, wonderful operation, but, in a way, I think
probably letting it miscarry helps to grieve in a funny way, because youre
going through your grief all of the time that you are waiting for it to go,
and then it goes, and you do a sort of mental realignment or whatever,
you know, you have time to sort of prepare yourself.
The baby: A few women wanted to see it and say goodbye, whereas others were scared
about what they might see and wanted to avoid it. Some women wanted to avoid intervention
because in the case of a misdiagnosis they felt that they would have been responsible for the
babys death.
but you know, I just sort of thought: whats that there? You know and,
then, sort of waited, and then when you pull the flush, its like a real
goodbye, you know.
Pain and bleeding: Pain and bleeding were mentioned mostly by medical and expectant
groups. Experiences of pain varied considerably, whereas bleeding was generally described as
being a lot.
They said it would be like a contraction, but I mean, it wasnt like a
contraction at all, really it was like very strong period pain I likened
it to when I first started my periods, when I was sort of 13.
I mean, looking back on it, I bled for about 40 hours, and had 40
hours of pain and bleeding; but I think that the actual psychological
support I had was so much better, that it didnt seem that bad.
Care received: A minority of women in the medical and surgical groups described a lack of
caring by staff. In contrast, several women in the expectant group commented that although the
experience was upsetting for them they found it reassuring to be at home.
... and I hated it! The whole thing was cold! It was so insensitive, it was
horrible! I will never forget how insensitive, and cold it felt.(woman who
had surgical treatment)
so, you know, I thought: no, Ill be at home, Ill be safe, and if theres
any real problems, Ive got a phone number to ring, or my GP, or well
just call... (woman who chose expectant management)
Emotional and psychological outcomes in non-randomised women
107
1 study
(Wieringa-de
Waard et al.,
2002b)
(Very low*)
In addition to the randomised women (n=82), this study reports the outcomes of women who
chose to be managed according to their own preference (n=147). Their outcomes are analysed
separately, and in comparison with the randomised group.
Within the preference group, there were no statistically significant differences between the
mental health scores and anxiety scores of women who chose expectant management (n=61)
and women who chose active treatment (n=86).
Comparing randomised and non-randomised women
When comparing women who were randomised to expectant management and women who
chose expectant management, there were no significant differences in mental health score and
anxiety.
Women who were randomised to active treatment had significantly worse mental health scores
than women who chose active treatment (p=0.03).
Within the randomised group, no differences were found between women who were randomly
allocated to the mode of management for which they had expressed a slight preference for, and
those who were randomised to the other mode of management.
Note: Other than expressing a strong preference for a specific management option it is unclear
whether randomised and non-randomised women were comparable. Mean values for all scores
at different assessment times were not reported as figures in the text but only in graphs from
which it is impossible to extract accurate values.
* Qualitative studies not ranked in GRADE but using NICE quality assessment for qualitative studies 1
Need for unplanned intervention 3
One study found that the need for unplanned intervention was higher in women who received 4
expectant management compared with women who received active treatment. This finding was 5
statistically significant. The evidence for this outcome was of high quality. 6
Infection 7
One meta-analysis of seven studies did not find a statistically significant difference in infection for 8
women who received expectant management compared with women who received active treatment. 9
The evidence for this outcome was of high quality. 10
Gastrointestinal side effects 11
One study found that gastrointestinal side effects were lower in women who received expectant 12
management compared with women who received active treatment. This finding was statistically 13
significant. The evidence for this outcome was of high quality. 14
Need for a blood transfusion 15
One study found that the need for a blood transfusion was higher in women who received expectant 16
Duration of bleeding 19
One study found that the duration of bleeding was longer in women who received expectant 20
significant. The evidence for this finding was of moderate quality. One study found that the duration of 22
bleeding was longer in women who received expectant management compared with women who 23
received surgical management. This finding was statistically significant. The evidence for this finding 24
was of high quality. However, the same study did not find a statistically significant difference in 25
duration of bleeding in women who received expectant management compared with women who 26
received medical management. The evidence for this finding was of moderate quality. One meta- 27
analysis of two studies and one further study did not find a statistically significant difference in 28
duration of bleeding in women who received expectant management compared with women who 29
108
received active treatment. The evidence for this finding was of moderate quality. One study reported 1
duration of bleeding in a manner that did not allow assessment of statistical significance. 2
Pain 3
Three studies did not find a statistically significant difference in the duration of pain for women who 4
received expectant management compared with women who received active treatment. One further 5
study reported this outcome in a manner that did not permit assessment of statistical significance. The 6
evidence for this outcome was of moderate quality. 7
Two studies did not find a statistically significant difference in the severity of pain for women who 8
received expectant management compared with women who received active treatment. The evidence 9
for this outcome was of moderate quality. 10
Unplanned admissions 11
One study found that unplanned admissions were higher in women who received expectant 12
Womens satisfaction 15
Two studies did not find a statistically significant difference in satisfaction for women who received 16
expectant management compared with women who received active treatment. The evidence for this 17
outcome was of moderate quality in one study and low in the other. 18
Anxiety 19
Three studies did not find a statistically significant difference in anxiety for women who received 20
outcome was of moderate quality in one study and low in the others. 22
Mental health 23
One study found that mental health scores were higher in women who received expectant 24
significant. One further study found no statistically significant difference in mental health scores 26
between the two groups. The evidence for this outcome was of low quality. 27
Live birth rate 28
One study did not find a statistically significant difference in the live birth rate for women who received 29
outcome was of moderate quality. 31
Fertility 32
Two studies did not find a statistically significant difference in fertility for women who received 33
Emotional and psychological outcomes 36
One qualitative study found that when comparing expectant management and active treatment there 37
was general consensus among women related to the following issues: fear of intervention, especially 38
anaesthetic, hospitalisation and surgery; desire to have a predictable course/end in terms of 39
management and symptoms, and need for more information in terms of bleeding and pain, details on 40
the timing, duration and effects of interventions. In contrast there were areas with great variation in 41
responses: appropriateness and need for intervention (to miscarry naturally vs. to have something 42
done to help expedite completion); awareness of the event (benefit in experiencing the event and 43
saying goodbye vs. avoiding consciousness of the miscarriage); the baby (seeing it and saying 44
goodbye vs. being scared about what they might see and feeling responsible for the babys death in 45
case of misdiagnosis); degree and experiences of pain and bleeding and care received (lack of caring 46
by staff vs. being reassuring to be at home). The evidence for these findings was of high quality. 47
One study found that mental health scores were worse in women who were randomised to active 48
treatment compared with women who had chosen active treatment. This difference was statistically 49
109
The same study did not find a statistically significant difference between women randomised to 1
expectant management and women who chose expectant management for the following outcomes: 2
mental health score; anxiety. In addition, the study did not find a statistically significant difference 3
between women who chose expectant management and women who chose active treatment for the 4
following outcomes: mental health score, anxiety. The evidence for these outcomes was of very low 5
quality. 6
Health economics 7
One good quality economic evaluation of the MIST trial (Petrou et al., 2006) undertaken in an English 8
setting found that the mean cost of surgical management was 200 more expensive than medical 9
management (95% CI 122 to 278) and that the mean cost of medical management was 273 (95% 10
CI 160 to 376) more expensive than expectant management. There was more than a 50% chance 11
that expectant management was the most cost-effective treatment if the decision maker was not 12
prepared to spend more than 70,000 to prevent a single gynaecological infection. 13
Five studies of lower quality (Hughes et al., 1996; Graziosi et al., 2005; Niinimaki et al., 2009; Rocconi 14
et al., 2005; You & Chung, 2005) were generally congruent with the economic evaluation of the MIST 15
trial in that they all found surgical management to be more expensive than medical management. 16
Where the studies also considered expectant management (Rocconi et al., 2005; You & Chung, 17
2005) this was found to be the least expensive alternative apart from a decision tree artefact where 18
expectant management cost exceeded those of medical management because the expected 19
management decision could be overridden by a patient preference for surgical treatment (You & 20
Chung, 2005). For a full description of the literature review, see section 9.3. 21
The group felt that the most important clinical outcomes were the need for an unplanned intervention, 24
the incidence of infection, gastrointestinal side effects and the need for a blood transfusion. Whilst 25
they recognised that live birth rate is an important outcome, they did not feel that it was likely the 26
evidence would be of a sufficient size to show a statistically significant difference. Womens emotional 27
and psychological outcomes were also prioritised. The group also recognised that womens 28
satisfaction with their care is an important outcome, but felt that this is a difficult outcome to capture in 29
research as the women are not able to express their satisfaction with the care they received in 30
comparison to their care that they didnt. Thus it is difficult to determine which management strategy 31
women are likely to prefer. 32
The evidence showed that expectant and active treatment of miscarriage were similar in terms of 34
severity of pain experienced, anxiety, womens satisfaction with care, number of days in pain and the 35
incidence of infection. Expectant treatment of miscarriage was associated with fewer gastro-intestinal 36
side effects, but women having expectant management of miscarriage had more days of bleeding and 37
a significantly greater chance of needing a blood transfusion. Unplanned intervention was significantly 38
greater in the expectantly managed group. 39
The review of health economic literature clearly demonstrated that expectant management is the most 41
cost-effective approach by a significant margin, even taking into account the additional care that some 42
women may require as the result of emergency procedures and unplanned interventions. As a result, 43
the group agreed that, for the majority of women, expectant management should be the first-line 44
treatment that is offered. They felt that, for most women, expectant management would be an 45
acceptable or even preferable alternative, as it negates the risk of intervening and accidentally 46
terminating a viable pregnancy. In addition, they noted the almost universal fear of intervention that 47
was reported in the qualitative study. 48
However, the group also recognised that expectant management may not be appropriate in some 49
circumstances. In particular, they noted the increased rate of blood transfusion following expectant 50
management. From their clinical experience, the GDG felt that this could be due to an increase in 51
severe blood loss at later gestation. Based on clinical judgement, they agreed that, for women at 52
110
increased risk of haemorrhage, such as those in the late first trimester, other management options 1
should be considered. They were aware that this might have cost implications in terms of treatment, 2
but felt that this should be weighed against the increased likelihood of a blood transfusion and need 3
for emergency care in this subset of women, and the risk of poor outcomes following haemorrhage. 4
The GDG noted that there was a failure rate associated with expectant management of miscarriage, 5
which is captured in a rate of further intervention of over 35%. They also recognised that for some 6
women, a lengthy period of expectant management would not be acceptable; therefore the group felt 7
that it was appropriate to designate an endpoint, after which women in whom expectant management 8
had thus far been unsuccessful could choose to undergo an intervention to expedite the process. 9
From their clinical experience, they felt that the natural course of a miscarriage might generally be 10
expected to occur over a period of 7 to 14 days from the start of bleeding, and therefore that women 11
should be reviewed after this time. Women should be rescanned after 7 to 14 days unless the 12
symptoms are improving or have resolved when a pregnancy test after 3 weeks would be sufficient. 13
The evidence available for this question was generally of high or moderate quality. In addition, it was 15
felt that the health economic literature was of sufficient quality that it would be used to inform the 16
review, rather than needing to develop a bespoke model for the question. Whilst the quality of 17
evidence was not high for all outcomes, it was for the majority, including those which the group placed 18
most emphasis on, and thus the group felt confident in using the evidence to develop their 19
recommendations. 20
One of the key themes from the qualitative data analysed in the review was the fact that women 22
wanted more information regarding what to expect during their treatment. As expectant management 23
may be an unfamiliar concept to some women, the GDG felt that it was vital that they be informed 24
about the process, including what to expect in terms of pain and bleeding, what pain relieving 25
measures they might use as well as what options would be available in the event that expectant 26
management does not result in the completion of the miscarriage. The GDG also felt that it was 27
important that women be given details of support organisations, so that they could access on-going 28
support and counselling services. The GDG recognised that there can be a lot of information to give 29
to women and agreed that healthcare professionals should ensure that sufficient time is made 30
available to discuss all of the relevant issues with women and that this be supported with written 31
information for the woman to take away. In some instances, this might also mean arranging an 32
additional appointment. 33
Although the qualitative studies for this review were looking at womens experiences of different 34
treatment options, the GDG agreed that the evidence highlighted the more general point that women 35
react to complications and the loss of pregnancy in different ways. The GDG therefore wished to 36
stress the importance of giving information in a sensitive way, taking into account each individual 37
womans emotional response. 38
From their clinical experience, they also recognised that, for women with prior traumatic experience of 39
miscarriage, the prospect of expectant management could be completely unacceptable. Therefore, in 40
light of the potential for increased psychological sequelae, the group felt that other management 41
options should be considered for these women. 42
The GDG recognised that there may be some groups of women for whom it would not be appropriate 44
to recommend expectant management due to their increased risk of haemorrhage or adverse effects 45
of haemorrhage; for example, women who are unable to receive blood transfusions, or women who 46
would have difficulty accessing follow-up care. 47
111
7.4 Surgical management compared with medical 1
management of miscarriage 2
Review question 3
How effective is surgical management of miscarriage compared with medical management for 4
improving womens clinical and psychological outcomes? 5
Introduction 6
Surgical management of miscarriage has been the standard management of miscarriage. Surgical 7
completion of miscarriage carries a risk of complications and often requires a general anaesthetic. 8
Medical management may provide an alternative treatment, but needs to be studied in terms of 9
maternal outcomes including emotional and psychological outcomes as well as pain, bleeding and 10
adverse clinical events including re-admission and unplanned intervention. 11
Twenty seven studies were included in this review (Chung et al., 1999; Dabash et al. 2010; Dao et al., 13
2007; Davis et al., 2007; de Jonge et al., 1995; Demetroulis et al., 2001; Egarter et al., 1995; Fang et 14
al., 2009; Graziosi et al., 2004; Graziosi et al., 2005a; Graziosi et al., 2005b; Harwood & Nansel, 15
2008; Hinshaw, 1997; Lee et al., 2001; Montesinos et al., 2011; Moodliar et al., 2005; Muffley et al., 16
2002; Niinimaki et al., 2006; Sahin et al., 2001; Shelley et al., 2005; Shwekerela et al., 2007; Smith et 17
al., 2006; Smith et al., 2009; Tam et al., 2005; Taylor et al., 2011; Trinder et al., 2006; Zhang et al., 18
2005). 19
One included study is a qualitative study which follows up both participants and non-participants of an 20
RCT conducted in the UK (Smith et al., 2006). One included study is a partially randomised trial, 21
conducted in the UK, which included both women who had chosen their method of management, and 22
those who had been randomised to medical or surgical management (Hinshaw, 1997). The remainder 23
of the included studies report the outcomes and follow-up data of sixteen randomised controlled trials, 24
conducted in the UK (2 trials: Demetroulis et al., 2001; Smith et al., 2009 / Trinder et al., 2006), 25
Australia (Shelley et al., 2005), Austria (Egarter et al., 1995), Burkina Faso (Dao et al., 2007), China 26
(Fang et al., 2009), Ecuador (Montesinos et al., 2011), Egypt (Dabash et al. 2010), Finland (Niinimaki 27
et al., 2006), Ghana (Taylor et al., 2011), Hong Kong (1 trial: Chung et al., 1999 / Lee et al., 2001 / 28
Tam et al., 2005), the Netherlands (1 trial: Graziosi et al., 2004 / Graziosi et al., 2005a / Graziosi et 29
al., 2005b), South Africa (2 trials: de Jonge et al., 1995; Moodliar et al., 2005), Tanzania (Shwekerela 30
et al., 2007), Turkey (Sahin et al., 2001), and the USA (2 trials: Davis et al., 2007 / Harwood & 31
Nansel, 2008 / Zhang et al., 2005; Muffley et al., 2002). 32
All studies compared medical and surgical management of miscarriage, and reported at least one 33
priority outcome. The trials were conducted in both developed and developing countries, and their 34
populations include women with missed miscarriages and/or women with ongoing miscarriages. 35
Evidence profile 36
Sixteen outcomes (grouped into eight broad categories) were chosen by the GDG as being of priority 37
to inform recommendations. In outcomes with high heterogeneity (over 60%), the NCC-WCH 38
technical team used a random effects model (the remaining outcomes used fixed effects models) and 39
explored the heterogeneity with sensitivity analyses. Findings from these sensitivity analyses are as 40
follows: 41
Gastro-intestinal side effects 42
Overall heterogeneity (89%) was not significantly reduced by considering only trials in developed 43
countries, considering trials where only drugs administered vaginally were used, considering only 44
trials where at least one oral drug was used or considering only trials where the surgical arm received 45
general anaesthetic. The technical team also explored the effect of dosage, with and without mode of 46
administration. None of the combinations tested reduced the heterogeneity to below 60%. The high 47
heterogeneity could be as a result of the variety of medical regimens used, the combination of three 48
112
gastro-intestinal side effects in to one overall outcome, or the fact that it was a patient-reported 1
outcome. 2
Unplanned visits to a medical facility 3
Considering only developed countries reduced the heterogeneity from 64% to 25%, and resulted in 4
there being no significant difference between the number of unplanned visits in the medical and 5
surgical arms. This could be as a result of different patterns of healthcare-seeking behaviour in 6
developed and developing countries. 7
Satisfaction 8
Overall heterogeneity (77%) was not reduced by considering only developed countries, or removal of 9
the trials in which the surgical arm received only local or verbal anaesthesia. The high heterogeneity 10
could be a result of factors not generally reported in the trials, for example variable waiting times for 11
surgery (reported in one trial as a contributing factor to dissatisfaction in the surgical patients). 12
Table 7.4 GRADE summary of findings for comparison of medical management with surgical management 13
Number of
studies
Number of women (%) or average Effect Quality
Medical Surgical Relative
(95% CI)
Absolute
(95% CI)
Need for unplanned intervention*
1 meta-
analysis of 18
studies

545/2553
(21.3%)
55/2186
(2.5%)
RR 8.13
(6.26 to 10.55)
179 more per
1000
(from 132 more
to 240 more)
High
Infection (incidence up to 15 days)
1 meta-
analysis of 7
studies

23/1455
(1.6%)
24/1113
(2.2%)
RR 0.9
(0.51 to 1.57)
2 fewer per
1000
(from 11 fewer
to 12 more)
High
Gastro-intestinal side effects (number of events)
1 meta-
analysis of 12
studies**
994/4358
(22.8%)
260/3346
(7.8%)
RR 2.36
(1.39 to 4.00)
106 more per
1000
(from 30 more
to 233 more)
Moderate
1 meta-
analysis of 8
studies

15/1353
(1.1%)
8/1063
(0.8%)
RR 1.6
(0.74 to 3.42)
5 more per 1000
(from 2 fewer to
18 more)
High
1 meta-
analysis of 5
studies
(Demetroulis
et al., 2001;
Egarter et al.,
1995; Graziosi
et al., 2004;
Moodliar et al.,
n=245 n=241 not calculable
(NC)
MD 1.31 higher
(0.73 to 1.89
higher)
p < 0.0001
High
113
2005; Sahin et
al., 2001)
1 study
(Trinder et al.,
2006)
Median 11
(IQR 7-15)
n=398
Median 8
(IQR 4-14)
n=402
NC Median 3 higher
(confidence
interval NC)
p = 0.0004
High
1 study
(Davis et al.,
2007)
Median 12
(IQR 9-14)
n=428
Median 10
(IQR 7-12)
n=135
NC Median 2 higher
(confidence
interval NC)
Moderate
1 study
(Dabash et al.
2010)
Mean 3.23
(SD NR)
n=327
Mean 2.73
(SD NR)
n=316
NC MD 0.5 higher
(confidence
interval NC)
p < 0.01
Moderate
1 study
(Taylor et al.,
2011)
Mean 2.86
(SD NR)
Mean 1.64
(SD NR)
NC MD 1.22 higher
(confidence
interval NC)
p = 0.001
Moderate
1 study
(Chung et al.,
1999)
Mean 9.1
(SD not
reported (NR))
n=321
Mean 9.3
(SD NR)
n=314
NC MD 0.2 lower
(confidence
interval NC)
p = 0.48
Low
1 study
(Montesinos et
al., 2011)
Mean 3.4
(SD NR)
Mean 3.0
(SD NR)
NC MD 0.4 higher
(confidence
interval NC)
p = 0.223
Low
1 study
(Dao et al.,
2007)
Mean 3.1
(SD NR)
n=223
Mean 2.9
(SD NR) n=224
NC MD 0.2 higher
(confidence
interval NC)
p = 0.09
Very low
Pain: duration (days)
1 study
(Demetroulis
et al., 2001)
Mean 4.7
(SD 2.4)
n=36
Mean 2.8
(SD 1.6)
n=35
NC MD 1.9 higher
(0.95 to 2.85
higher)
p < 0.0001
High
1 study
(Dabash et al.
2010)
Mean 2.63
(SD NR)
n=327
Mean 2.63
(SD NR)
n=316
NC MD 0 higher
(confidence
interval NC)
p = 0.98
Moderate
1 study
(Shelley et al.,
2005)
Median 3.0
(Range 0.2-
16.0)
Median 2.0
(Range 0.2-
12.0)
NC Median 1 higher
(confidence
interval NC)
Moderate
114
n=11 n=12
1 study
(Montesinos et
al., 2011)
Mean 2.5
(SD NR)
Mean 2.6
(SD NR)
NC MD 0.1 lower
(confidence
interval NC)
p = 0.739
Moderate
1 study
(Taylor et al.,
2011)
Mean 1.44
(SD NR)
Mean 1.34
(SD NR)
NC MD 0.1 higher
(confidence
interval NC)
p = 0.44
Low
1 study
(Chung et al.,
1999)
Mean 0.17
(SD NR)
n=321
Mean 0.25
(SD NR)
n=314
NC MD 0.08 lower
(confidence
interval NC)
p = 0.30
Low
1 study
(Dao et al.,
2007)
Mean 1.4
(SD NR)
n=223
Mean 1.3
(SD NR)
n=224
NC MD 0.1 higher
(confidence
interval NC)
p = 0.08
Very low
Pain: severity score/10
1 meta-
analysis of 3
studies
(Graziosi et al.,
2004; Moodliar
et al., 2005;
Zhang et al.,
2005)
n=602 n=263 NC MD 2.3 higher
(1.92 to 2.68
higher)
p < 0.00001
High
1 study
(Shelley et al.,
2005)
Median 3
(Range 1-8)
n=11
Median 3
(Range 1-10)
n=12
NC Median 0 higher
(confidence
interval NC)
Moderate
1 study
(Fang et al.,
2009)
Score <3: 17/45
(37.8%)
Score <3: 12/30
(40%)
RR 0.94 (0.53-
1.68)
24 fewer per
1000
(from 188 fewer
to 272 more)
Low
Pain: severity score/7
1 study
(Dao et al.,
2007)
Mean 2.32 (SD
NR)
n=223
Mean 2.73
(SD NR)
n=224
NC MD 0.41 lower
(confidence
interval NC)
p = 0.047
Moderate
1 study
(Shwekerela et
al., 2007)
Mean 3.0
(SD NR)
n=150
Mean 3.5
(SD NR)
n=150
NC MD 0.5 lower
(NC)
p < 0.001
Moderate
Unplanned visits to a medical facility
115
1 meta-
analysis of 5
studies
(Chung et al.,
1999; Dabash
et al. 2010;
Dao et al.,
2007;
Demetroulis et
al., 2001;
Zhang et al.,
2005)
148/1375
(10.8%)
42/1026
(4.1%)
RR 1.67
(0.74 to 3.79)
27 more per
1000
(from 11 fewer
to 114 more)
Low
Unplanned admissions

1 study
(Trinder et al.,
2006)
72/398
(18.1%)
32/402
(7.9%)
RR 2.27
(1.53 to 3.37)
101 more per
1000
(from 42 more
to 189 more)
High
Satisfaction: reported incidence
1 meta-
analysis of 9
studies
#

1032/1093
(94.4%)
1024/1076
(95.2%)
RR 0.99
(0.96 to 1.03)
10 fewer per
1000
(from 38 fewer
to 29 more)
Low
Social function at 2 weeks: SF-36 score/100 (better indicated by higher values)
1 meta-
analysis of 2
studies
(Graziosi et al.,
2005a;
Harwood &
Nansel, 2008)
n=525 n=205 NC MD 0.69 lower
(2.7 lower to
1.32 higher)
p = 0.50
High
Social function at 2 weeks: SPS score (better indicated by lower values)
1 study
(Lee et al.,
2001)
Mean 0.14
(SD 0.26)
n=104
Mean 0.16
(SD 0.29)
n=111
NC MD 0.02 lower
(0.094 lower to
0.054 higher)
p = 0.93
High
Mental health at 2 weeks: SF-36 score/100 (better indicated by higher values)
1 meta-
analysis of 2
studies
(Graziosi et al.,
2005a; Shelley
et al., 2005)
n=79 n=66 NC MD 3.43 lower
(8.53 lower to
1.68 higher)
p = 0.19
Low
Live birth rate
1 meta-
analysis of 2
studies
(Smith et al.,
290/361
(80.3%)
304/365
(83.3%)
RR 0.96
(0.9 to 1.03)
33 fewer per
1000
(from 83 fewer
to 25 more)
Moderate
116
2009; Tam et
al., 2005)
1 study
(Graziosi et al.,
2005b)
n=69 n=57 RR 0.98
(0.66 to 1.5)
NC Very low
* Unplanned interventions include: surgery (including repeat surgery) due to first-line treatment failure, emergency surgery prior 1
to allocated treatment, surgical completion on maternal request, or treatment to deal with a complication of the initial treatment 2
18 studies Chung et al. 1999; Dabash et al. 2010; Dao et al., 2007; de Jonge et al., 1995; Demetroulis et al., 2001; Egarter 3
et al., 1995; Fang et al., 2009; Graziosi et al., 2004; Montesinos et al., 2011; Moodliar et al., 2005; Muffley et al., 2002; 4
Niinimaki et al., 2006; Sahin et al., 2001; Shelley et al., 2005; Shwekerela et al., 2007; Taylor et al., 2011; Trinder et al., 2006; 5
Zhang et al., 2005 6
7 studies - Chung et al., 1999; Moodliar et al., 2005; Sahin et al., 2001; Shelley et al., 2005; Shwekerela et al., 2007; Trinder 7
et al., 2006; Zhang et al., 2005; 8
** 12 studies - Chung et al., 1999; Dabash et al., 2010; Dao et al., 2007; Demetroulis et al., 2001; Egarter et al., 1995; Graziosi 9
et al., 2004; Montesinos et al., 2011; Moodliar et al., 2005; Shelley et al., 2005; Shwekerela et al., 2007; Taylor et al., 2011; 10
Zhang et al., 2005 11

8 studies - Dabash et al., 2010; Davis et al., 2007; de Jonge et al., 1995; Demetroulis et al., 2001; Graziosi et al., 2004; 12
Muffley et al., 2002; Shelley et al., 2005; Trinder et al., 2006 13
Unplanned admission is an admission to hospital during the trial that was not pre-specified in the methodology as part of the 14
management protocol 15
# 9 studies - Dabash et al. 2010; Dao et al., 2007; Demetroulis et al., 2001; Fang et al., 2009; Montesinos et al., 2011; 16
Niinimaki et al., 2006; Sahin et al., 2001; Shwekerela et al., 2007; Taylor et al., 2011 17
Two UK studies undertook additional data collection and analysis as a follow-up to RCTs. One study 18
(Smith et al, 2006) used qualitiative methods of data collection and analysis, the other used a simple 19
descriptive survey with data reported as descriptive statistics (Hinshaw, 1997). Findings from these 20
studies are summarised in table 6.5 with quotations used to illustrate key themes identified from the 21
qualitative research. 22
Table 7.5 Additional findings evaluating medical and surgical miscarriage management strategies 23
Qualitative data on emotional/psychological outcomes and womens satisfaction with care
1 study
(Smith et
al., 2006)
(High*)
Areas with general consensus:
Fear: There was near uniform fear of intervention, especially anaesthetic.
I was more worried about the anaesthetic, that sort of worries me, just sort of
being knocked out, and I'm always afraid about not waking up again....
Predictability: Women wanted a predictable end, so they could get on with their lives, and they
wanted their management and symptoms to have a predictable course.
I would have preferred to have a D & C, although Im not sure what that would
be like, exactly what that is, but, at least there would be an end to that, like you
know: one minute youre pregnant, and the next minute, its finished and you
can get on with your life.
Need for more information: Women felt they did not know what to expect in terms of bleeding and
pain, and wanted more details on the timing, duration and effects of interventions, particularly in the
medical group.
and I just thought: I dont want to wait any more, particularly because I dont
know whats going to happen, and, oh, the first time Id read a book about
miscarriage, and it, the most awful stories always get in there, I mean I was,
you always get those sorts of stories and you think, oh my God, you know,
what on earth is going to happen?
Areas with wider variation in responses:
117
Appropriateness: Some women queried whether intervention was necessary and wished to be
allowed to miscarry naturally themselves, whereas others were in favour of something being done to
help expedite completion.
it happened the next morning [when] I came home and it was a sense of
relief really, ... its ended the medical treatment, its just speeding it up its
not actually anyone else going in my body its just a little magic tablet its
midpoint its a kind treatment its not your baby whipped out of you, which
is what a D & C feels like to me.
Awareness of the event: Some women felt benefit in experiencing the event, to allow them to say
goodbye, whereas others preferred surgery to avoid consciousness of the miscarriage.
its very clean, very quick, wonderful operation, but, in a way, I think
probably letting it miscarry helps to grieve in a funny way, because youre going
through your grief all of the time that you are waiting for it to go, and then it
goes, and you do a sort of mental realignment or whatever, you know, you have
time to sort of prepare yourself.
The baby: A few women wanted to see it and say goodbye, whereas others were scared about what
they might see and wanted to avoid it. Some women wanted to avoid intervention because in the case
of a misdiagnosis they felt that they would have been responsible for the babys death.
I was very relieved that it had miscarried naturally cos I could cope with it
dying naturally, that wasnt a problem, with the thought of having it killed on
purpose, thats how I would have seen it.
Pain and bleeding: Pain and bleeding were mentioned mostly by medical group. Experiences of pain
varied considerably, whereas bleeding was generally described as being a lot.
I suppose to all intents and purposes, I had gone through labour, although,
obviously a different version, but I did feel, my body did feel as though Id gone
through labour, and of course, I had nothing to show for it.
Care received: A minority of women described a lack of caring by staff.
you felt like you were ... sort of on a conveyor belt and they just whacked
this mask over my face, it was almost like, you know: get through, lie down,
shut up [laughs] and we can get on with it, because you are slowing down the
process (woman who had surgical treatment)
Descriptive data from non-randomised women
1 study
(Hinshaw,
1997)
(Very
low*)
In a partially randomised trial, 54.2% women had a preference for one method and chose not to be
randomised; therefore they were given their preferred method of management. The reasons that
women preferred medical management were the avoidance of general anaesthesia/surgery (57.1%),
and the feeling that it was more natural or they were in control (35.7%). The reasons that women
preferred surgical management were the timescale (72.1%), issues of awareness (42.9%), avoidance
of pain/bleeding (40.8%) and perceived effectiveness (12.9%). Generally, acceptability was lower in
the medical arm, but was unaffected by whether women were randomised or chose their method of
management.
* Qualitative studies not ranked in GRADE but using NICE quality assessment for qualitative studies 1
Additional analysis of surgical complications 2
The rate of surgical complications is not reported as an outcome in the GRADE table, because it is 3
intuitive that the rate will be higher in women randomised to the surgical arm (the vast majority of 4
whom undergo surgery) when compared to women randomised to the medical arm (a minority of 5
whom end up having surgery). However, the GDG were concerned that potential side effects of 6
surgery should not be overlooked, and therefore further analysis was done based on data on surgical 7
complications that was reported in the studies: 8
118
Five trials (Chung et al., 1999; Egarter et al., 1995; Graziosi et al., 2004; Muffley et al., 2002; 1
Trinder et al., 2006) reported the rates of surgical complications, which included uterine 2
perforation, cervical laceration, haemorrhage and intrauterine synechia. The incidence of 3
surgical complications was generally low, ranging from 2% to 8% among women randomised 4
to surgery. The incidence was under 5% in four out of the five studies that reported it. 5
The potential for reducing the risk of complications through cervical priming prior to surgery 6
(as hypothesised in Chung et al., 1999) could not be assessed, because none of the studies 7
that assessed surgical complication rates reported priming. 8
The association of complication rate with type of miscarriage (e.g. missed or incomplete) 9
could not be explored because three of the studies only included women with missed 10
miscarriage, and in the remaining two studies over 75% of women had a missed miscarriage. 11
Need for an unplanned intervention 13
One meta-analysis of eighteen studies found that the need for an unplanned intervention was higher 14
in women who received medical management compared with women who received surgical 15
management. This finding was statistically significant. The evidence for this outcome was of high 16
quality. 17
Infection 18
One meta-analysis of seven studies did not find a statistically significant difference in the incidence of 19
infection for women who received medical management compared with women who received surgical 20
management. The evidence for this outcome was of high quality. 21
Gastro-intestinal side effects 22
One meta-analysis of twelve studies found that gastro-intestinal side effects were higher in women 23
who received medical management compared with women who received surgical management. This 24
finding was statistically significant. The evidence for this outcome was of moderate quality. 25
One meta-analysis of eight studies did not find a statistically significant difference in the need for a 27
blood transfusion for women who received medical management compared with women who received 28
surgical management. The evidence for this outcome was of high quality. 29
One meta-analysis of five studies and three other studies found that duration of bleeding was longer 31
in women who received medical management compared with women who received surgical 32
management. This finding was statistically significant. The evidence for this finding was of high quality 33
in the meta-analysis and one single study, and moderate quality in the other two studies. Three 34
further studies did not find a statistically significant difference in duration of bleeding between the two 35
groups. The evidence for this finding was low quality in two studies and very low quality in the other. 36
One further study reported duration of bleeding in a manner that did not allow assessment of 37
statistical significance. 38
Pain 39
One study found that the duration of pain was longer in women who received medical management 40
compared with women who received surgical management. This finding was statistically significant. 41
The evidence for this finding was of high quality. Five further studies did not find a statistically 42
significant difference in duration of pain between the two groups. The evidence for this finding was of 43
moderate quality in two studies, low in two studies and very low in the other. One further study 44
reported duration of pain in a manner that did not allow assessment of statistical significance. 45
One meta-analysis of three studies found that the severity of pain on a ten-point scale was higher in 46
women who received medical management compared with women who received surgical 47
management. This finding was statistically significant. The evidence for this finding was of high 48
quality. One further study did not find a statistically significant difference in severity of pain on a ten- 49
point scale between the two groups. The evidence for this finding was of low quality. One further 50
study reported severity of pain on a ten-point scale in a manner that did not allow assessment of 51
statistical significance. 52
119
Two studies found that the severity of pain on a seven-point scale was lower in women who received 1
medical management compared with women who received surgical management. This finding was 2
statistically significant. The evidence for this outcome was of moderate quality. 3
One meta-analysis of five studies did not find a statistically significant difference in the unplanned 5
visits to a medical facility for women who received medical management compared with women who 6
received surgical management. The evidence for this outcome was of low quality. 7
Unplanned admissions 8
One study found that unplanned admissions were higher in women who received medical 9
management compared with women who received surgical management. This finding was statistically 10
Satisfaction 12
One meta-analysis of nine studies did not find a statistically significant difference in the reported 13
satisfaction for women who received medical management compared with women who received 14
surgical management. The evidence for this outcome was of low quality. 15
Social function at 2 weeks 16
Three studies did not find a statistically significant difference in the social function score at 2 weeks 17
for women who received medical management compared with women who received surgical 18
management. The evidence for this outcome was of high quality. 19
Mental health at 2 weeks 20
One meta-analysis of two studies did not find a statistically significant difference in the mental health 21
score at 2 weeks for women who received medical management compared with women who received 22
surgical management. The evidence for this outcome was of low quality. 23
Live birth rate 24
One meta-analysis of two studies, and one further study, did not find a statistically significant 25
difference in the live birth rate for women who received medical management compared with women 26
who received surgical management. The evidence for this outcome was of moderate and very low 27
quality. 28
Womens preferences, emotional and psychological outcomes 29
One qualitative study found that there was general consensus among women regarding fear of 30
intervention, a desire for their treatment to follow a predictable course, and the need for more 31
information. In contrast, there was wider variation in womens feelings about the appropriateness of 32
intervention, awareness of the event, feelings about the baby, the degree of pain and bleeding 33
experienced, and the care received. The evidence for these findings was of high quality. 34
One partially randomised trial provided very low quality evidence that the acceptability of medical 35
management was lower than surgical management, but was unaffected by whether women were 36
randomised to, or had chosen, their method of management. Women who chose surgical 37
management stated timescale, issues of awareness, avoidance of pain/bleeding and perceived 38
effectiveness as the reasons for their preference. Women who chose medical management stated 39
avoidance of general anaesthesia/surgery, and the feeling that it was more natural and that they were 40
in control as the main reasons for their preference. However, the evidence for these findings was of 41
very low quality. 42
For this review, the GDG felt that both quantitative and qualitative evidence were equally valuable. 45
Clinically, they felt that the need for further intervention, requirement for a blood transfusion, and side 46
effects were important, in addition to other outcomes with cost implications, such as need for 47
admission. From their clinical experience, and the qualitative evidence, the group noted that womens 48
responses to miscarriage and their preferences for mode of management remain highly variable, and 49
therefore the views and experiences of women reported in the qualitative study were vital in informing 50
120
their decision. In contrast, the group did not feel that satisfaction with treatment and psychological 1
scores were particularly useful outcomes. What evidence there was showed little difference between 2
the two treatment arms, but the group felt that these outcomes are often difficult to capture accurately 3
in randomised populations, and are less informative than qualitative data that explores womens 4
experiences of their mode of management. 5
Medical management of miscarriage avoids the need for surgery in over 70% of women. The GDG 7
felt that this would be an important consideration for women, as surgery often requires a general 8
anaesthetic and has an associated risk of complications. The risk of surgical complications among 9
women randomised to surgery ranged from 2% to 8%, and included uterine perforation, haemorrhage, 10
cervical lacerations and synechia. However, medical management is also associated with a 11
significantly higher rate of unplanned intervention and unplanned admission. Gastro-intestinal side 12
effects such as vomiting are higher with medical management of miscarriage but the risk of infection 13
and haemorrhage requiring transfusion is similar for both forms of active management. From their 14
experience the GDG felt that this small risk of blood transfusion may be an over-estimation as in 15
current clinical practice few women receive one. 16
The results of the qualitative studies supported the GDGs view that individual women often have very 17
different priorities and expectations of their treatment. They recognised that whilst women may 18
strongly wish to avoid undergoing surgery, for some women the predictability, promptness, and high 19
likelihood of success following surgical treatment would be an attractive option. The GDG felt that, as 20
the majority of women would have undergone a week of expectant management as a first line 21
treatment, it was important that they be given a choice about how to proceed at that point. Although 22
the health economics analysis showed that medical management was more cost-effective than 23
surgical management, the GDG noted that this was based on estimates of first-line treatment, and 24
therefore could not directly be applied to women in whom expectant management had already failed. 25
They felt strongly that, after a period of expectant management, women should have the choice of 26
how to proceed, and therefore recommended that a discussion of the options take place. 27
The evidence showed that outcomes such as duration of pain seemed comparable in the medical and 28
surgical arms, although the tendency was that pain lasted longer and was more severe after medical 29
management. Similarly, the duration of bleeding was very variable but generally less in the surgical 30
arm than the medical arm. From their clinical experience, the GDG felt that womens experiences of 31
pain and bleeding after miscarriage tend to be extremely variable..The GDG noted that medical 32
management of miscarriage seems to be more successful (in terms of avoiding surgical intervention) 33
in women with incomplete or inevitable miscarriage, when compared to those with a missed 34
miscarriage. They also recognised that successful treatment was higher in studies that allowed longer 35
follow-up before surgical intervention and where follow-up was clinical rather than ultrasound 36
orientated. However, due to the differences between the studies, they did not feel that the evidence 37
was strong enough to make a recommendation that might supersede womens choice. Overall, they 38
noted that there were both advantages, in terms of avoiding surgery, and disadvantages, in terms of 39
the potential for increased pain and bleeding, of medical management and therefore that the 40
individual womans preference and specific clinical situation should inform the choice of second-line 41
management strategy. 42
Whilst both unplanned admissions and need for an unplanned intervention are higher after medical 44
management of miscarriage compared to surgical treatment, the health economics analysis of first 45
line treatment options calculated that medical management was more cost-effective due to the 46
reduced cost of the initial treatment. However, the GDG felt that, having recommended expectant 47
management (the most cost-effective option) as a first line management strategy on the grounds of 48
cost, it was appropriate that women have options if this strategy failed, particularly as the health 49
economics was based on outcomes of first-line treatment. They noted that, for women in whom 50
expectant management had not been successful, the success of medical management was likely to 51
be reduced, and therefore the associated costs of unplanned interventions and admissions would be 52
increased. 53
121
Much of the evidence for this review was of high or moderate quality, and, in particular, the GDG 2
welcomed the inclusion of the MIST trial, a high quality randomised controlled trial conducted in the 3
UK, with an associated qualitative study investigating womens views of different modes of 4
miscarriage management. The fact that it explored womens experiences so comprehensively, and 5
was conducted in the UK, led the GDG to believe that it was likely to represent the spectrum of 6
different views that women might have regarding their preferred treatment options. However, as it was 7
only one study involving a small group of women, the GDG felt that it was important to recommend 8
that further research be done to evaluate whether different modes of management impact on patient 9
experience and longer term psychological and emotional outcomes. 10
The GDG noted that an overarching theme from the qualitative data was the fact that women wanted 12
more information about what to expect, what their course of treatment would entail (including potential 13
complications), and what support would be provided (both immediately and longer term). It was noted 14
that a lack of information often led to uncertainty which could heighten womens anxiety. Therefore, 15
the group felt that it was important that women were informed about the possible course of events 16
following their chosen management course, including what to expect in terms of the duration and 17
severity of bleeding and where and when to get help in an emergency. In addition, it was the 18
experience of some of the group that women are often uncertain about what to expect in the recovery 19
period, and that they therefore need to be given more information about this, including details of how 20
to access counselling and other support services. 21
The GDG also felt it important to support womens choice following a period of expectant 22
management where this had not been successful. They recognised the potential for increased 23
psychological sequelae if women were denied a choice after 7-14 days of expectant management, 24
during which time they may have been continuing to bleed and desiring a prompt completion of the 25
process. 26
Choice of treatment is important for women and satisfaction is higher where women have been 28
offered and exercised their choice. The GDG recognised that for women with greater difficulty in 29
accessing health care (for example, women with English as a second language, drug users, 30
travellers, or those living in a remote area), surgical management might be preferable due to the 31
reduced need for unplanned intervention and unplanned admission. However, they felt that, patient 32
choice was still the most important consideration. 33
Recommendations 34
50 Use expectant management for 7-14 days as the firstline management strategy
following confirmed diagnosis of a non-viable pregnancy.
51 Explain expectant management and that most women will need no further
treatment. Also provide written and verbal information about further treatment
options.
what to expect throughout the process, including the likely duration and
severity of bleeding, advice on pain relief and where and when to get help
in an emergency
what to expect in the recovery period, such as how long they should wait
before resuming sexual activity and trying to conceive.
Ensure that sufficient time is available to discuss these issues with women. Arrange
an additional appointment if necessary.
53 Inform women where to access support and counselling services. Provide them with
122
leaflets that include helpline numbers for support organisations.
54 Explore management options other than expectant management if:
the woman is at increased risk of haemorrhage (for example, she is in the
late first trimester), or
she has previous adverse and/or traumatic experience associated with
pregnancy such as previous miscarriage or antepartum haemorrhage, or
she is at increased risk from the effects of haemorrhage.
55 If the resolution of bleeding and pain indicate that the miscarriage has completed
during 714 days of expectant management, advise the woman to take a urine
pregnancy test after 3 weeks, and to return if it is positive for individualised
management.
56 Offer a repeat scan if after the period of expectant management the bleeding and
pain:
have not started (suggesting that the process of miscarriage has not begun)
or
are persisting and/or increasing (suggesting incomplete miscarriage).
Discuss all treatment options with the woman to allow her to make an informed
choice.
57 Review the condition of a woman who opts for continued expectant management at
a minimum of 14 days after the first follow-up appointment.
1
2
RR 6 In women with confirmed miscarriage, does the type of intervention (expectant,
medical and surgical) impact on womens experience, including psychological and
emotional outcomes?

The management of miscarriage in the UK has changed in many ways over the past
two decades, particularly in the shift from inpatient to outpatient or day case care
and the introduction of medical and expectant management as alternatives to
surgery.
Despite these changes there is a lack of research into the effects of these different
approaches from the patients perspective, in particular their psychological and
emotional impact. Miscarriage is distressing for most women, and the type of
management itself might affect womens need for counselling, with a resulting cost
to the NHS. Because of this it is an important area for research.
The deficiency in the literature could be addressed by a comparative study of
women having the different treatments (expectant, medical or surgical) and in a
variety of clinical settings (for example, early pregnancy assessment unit,
gynaecological ward, or gynaecological emergency unit). The data collected could
be both quantitative (using validated psychological health questionnaires) and
qualitative (focusing particularly on womens experience of the particular type and
setting of care).
123
7.5 Misoprostol and mifepristone for managing 1
miscarriage 2
Review question 3
What is the most appropriate dose of misoprostol and mifepristone to provide for managing 4
miscarriage? 5
Introduction 6
Medical management of miscarriage has been offered to women suffering miscarriage for a number 7
of years with varying doses, timing and routes of administration of drugs being used. The reviews 8
carried out for the purposes of this guideline aimed to ascertain the most appropriate and efficacious 9
dose. The GDG considered the data for two groups of women: those having a missed miscarriage 10
and those having an incomplete miscarriage, as some of the clinicians on the GDG reported a 11
difference in treatment outcomes for these groups of women. 12
Twenty one studies were included in this review (Ayudhaya et al., 2006; Bagratee et al., 2004; 14
Blanchard et al., 2004; Blohm et al., 2005; Creinin et al., 1997; Kovavisarach & Jamnansiri, 2005; 15
Kovavisarach & Sathapanachai, 2002; Kushwah & Singh, 2009; Lelaidier et al., 1993; Lister et al., 16
2005; Ngoc et al., 2004; Ngoc et al., 2005; Pang et al., 2001; Paritakul & Phupong, 2011; Rita et al., 17
2006; Shah et al., 2010; Stockheim et al., 2006; Tang et al., 2003; Tang et al., 2006; Tanha et al., 18
2010; Wood & Brain, 2002). 19
All of the included studies were randomised controlled trials, and were conducted in the UK (Bagratee 20
et al., 2004), USA (Creinin et al., 1997; Lister et al., 2005), Canada (Wood & Brain, 2002), France 21
(Lelaidier et al., 1993), Sweden (Blohm et al., 2005), Israel (Stockheim et al., 2006), Hong Kong 22
(Pang et al., 2001; Tang et al., 2003; Tang et al., 2006), Thailand (Ayudhaya et al., 2006; 23
Kovavisarach & Jamnansiri, 2005; Kovavisarach & Sathapanachai, 2002; Paritakul & Phupong, 24
2011), Vietnam (Blanchard et al., 2004; Ngoc et al., 2004; Ngoc et al., 2005), India (Kushwah & 25
Singh, 2009; Rita et al., 2006), Pakistan (Shah et al., 2010), and Iran (Tanha et al., 2010). 26
The GDG decided that, for this review question, studies should only be included if they treated women 27
with incomplete miscarriages and women with missed miscarriages as separate populations. Four 28
studies only included women with incomplete miscarriages (Blanchard et al., 2004; Ngoc et al., 2005; 29
Pang et al., 2001; Paritakul & Phupong, 2011), and two studies included women with both incomplete 30
miscarriages and missed miscarriage, but reported at least one outcome separately for the two 31
populations (Bagatree et al., 2004; Blohm et al., 2005). The remainder of the studies only included 32
women with missed miscarriage. 33
All of the included studies evaluated the use of misoprostol (Ms) and/or mifepristone (Mf) for the 34
management of first trimester miscarriage. One study compared the efficacy of misoprostol alone to a 35
combined regimen of mifepristone and misoprostol (Stockheim et al., 2006). Four studies compared 36
different dosages of misoprostol using the same route of administration, of which one evaluated 37
vaginal misoprostol (Kovavisarach & Jamnansiri, 2005), one evaluated sublingual misoprostol (Tang 38
et al., 2006), and two evaluated oral misoprostol (Blanchard et al., 2004; Ngoc et al., 2005). Nine 39
studies compared misoprostol administered via different routes, of which three compared oral and 40
sublingual administration (Ayudhaya et al., 2006; Kushwah & Singh, 2009; Paritakul & Phupong, 41
2011), three compared sublingual and vaginal administration (Shah et al., 2010; Tang et al., 2003; 42
Tanha et al., 2010), and four compared oral and vaginal administration (Creinin et al., 1997; Ngoc et 43
al., 2004; Pang et al., 2001; Rita et al., 2006). Six trials were placebo controlled, of which five 44
evaluated misoprostol (Bagratee et al., 2004; Blohm et al., 2005; Kovavisarach & Sathapanachai, 45
2002; Lister et al., 2005; Wood & Brain, 2002), and one evaluated mifepristone (Lelaidier et al., 1993). 46
Evidence profile 47
The treatment regimens described in the GRADE tables and evidence statements detail the maximum 48
number of doses that women could receive; some women did not receive repeat doses if expulsion 49
124
had started. If multiple doses of misoprostol were given during the same visit, the interval between 1
doses was between three and six hours, except in one trial whose participants also received 2
supplemental doses after 12 hours (Kushwah & Singh, 2009). In four studies, a repeat dose was 3
given if expulsion had not occurred after 24 hours (Bagratee et al., 2004; Creinin et al., 1997; Lister et 4
al., 2005; Wood & Brain, 2002), and in one further study, the treatment regimen consisted of two 5
doses administered 48 hours apart (Stockheim et al., 2006) (for further details of study regimens 6
please see evidence tables in Appendix H). 7
Table 7.6 GRADE summary of findings for comparison of vaginal misoprostol with placebo for the management 8
of missed miscarriage 9
Number of
studies
Details of
treatment
regimen
(dose in
micrograms
unless
stated)
Number of women or
average
Effect Quality
Misoprostol
(Ms)
Placebo Relative
(95% CI)
Absolute
(95% CI)
Success of medical treatment
1 meta-
analysis of 2
studies
(Blohm et al.,
2005;
Kovavisarach &
Sathapanachai,
2002 )
400 vaginal
Ms
69/91
(75.8%)
37/89
(41.6%)
RR 2.10
(0.97 to
4.53)
457 more per
1000
(from 12
fewer to
1000 more)
Very low
1 study
(Bagratee et
al., 2004)
600 vaginal
Ms
(repeat after
24 h)
39/45
(86.7%)
11/38
(28.9%)
RR 2.99
(1.8 to 4.99)
576 more per
1000
(from 232
more to 1000
more)
High
1 meta-
analysis of 2
studies
(Lister et al.,
2005; Wood &
Brain, 2002)
800 vaginal
Ms
(repeat after
24 h)
35/44
(79.5%)
6/42
(14.3%)
RR 5.59
(2.62 to
11.93)
656 more per
1000
(from 237
more to 1000
more)
High
Need for further intervention
1 study
(Blohm et al.,
2005)
400 vaginal
Ms
8/57
(14%)
23/51
(45.1%)
RR 0.31
(0.15 to
0.63)
311 fewer
per 1000
(from 167
fewer to 383
fewer)
High
1 study
(Bagratee et
al., 2004)
600 vaginal
Ms
(repeat after
24 h)
6/45
(13.3%)
27/38
(71.1%)
RR 0.19
(0.09 to
0.41)
576 fewer
per 1000
(from 419
fewer to 647
fewer)
High
1 meta-
analysis of 2
800 vaginal 10/43 34/41 RR 0.28 597 fewer High
125
studies
(Lister et al.,
2005; Wood &
Brain, 2002)
Ms
(repeat after
24 h)
(23.3%) (82.9%) (0.16 to
0.49)
per 1000
(from 423
fewer to 697
fewer)
Unplanned visits to a medical facility
1 study
(Lister et al.,
2005)
800 vaginal
Ms
(repeat after
24 h)
0/18
(0%)
3/16
(18.8%)
RR 0.13
(0.01 to 2.3)
163 fewer
per 1000
(from 186
fewer to 244
more)
Moderate
Adverse effects: incidence of nausea and/or vomiting
1 study
(Kovavisarach
&
Sathapanachai,
2002)
400 vaginal
Ms
2/27
(7.4%)
1/27
(3.7%)
RR 2
(0.19 to
20.77)
37 more per
1000
(from 30
fewer to 732
more)
Very low
Adverse effects: incidence of nausea
1 study
(Bagratee et
al., 2004)
600 vaginal
Ms
(repeat after
24 h)
18/52
(34.6%)
16/52
(30.8%)
RR 1.12
(0.65 to
1.96)
37 more per
1000
(from 108
fewer to 295
more)
Low
1 study
(Lister et al.,
2005)
800 vaginal
Ms
(repeat after
24 h)
4/18
(22.2%)
3/16
(18.8%)
RR 1.19
(0.31 to
4.51)
36 more per
1000
(from 129
fewer to 658
more)
Moderate
Adverse effects: severity of nausea
1 study
(Blohm et al.,
2005)
400 vaginal
Ms
Mean 17.4
(SD 24.7)
n=64
Mean 14.9
(SD 23.8)
n=62
not
calculable
(NC)
MD 2.5
higher
(5.97 lower
to 10.97
higher)
p = 0.57
Low
Adverse effects: incidence of vomiting
1 study
(Bagratee et
al., 2004)
600 vaginal
Ms
(repeat after
24 h)
8/52
(15.4%)
7/52
(13.5%)
RR 1.14
(0.45 to
2.92)
19 more per
1000
(from 74
fewer to 258
more)
Low
1 study
(Lister et al.,
2005)
800 vaginal
Ms (repeat
after 24 h)
1/18
(5.6%)
3/16
(18.8%)
RR 0.3
(0.03 to
2.57)
131 fewer
per 1000
(from 182
fewer to 294
more)
Moderate
Adverse effects: severity of vomiting
126
1 study
(Blohm et al.,
2005)
400 vaginal
Ms
Mean 8.1 (SD
20.2)
n=64
Mean 7.3
(SD 21.7)
n=62
NC MD 0.8
higher
(6.53 lower
to 8.13
higher)
p = 0.85
Low
Adverse effects: incidence of diarrhoea
1 study
(Kovavisarach
&
Sathapanachai,
2002)
400 vaginal
Ms
2/27
(7.4%)
0/27
(0%)
RR 5
(0.25 to
99.51)
NC Very low
1 study
(Bagratee et
al., 2004)
600 vaginal
Ms
(repeat after
24 h)
11/52
(21.2%)
11/52
(21.2%)
RR 1
(0.48 to 2.1)
0 fewer per
1000
(from 110
fewer to 233
more)
Low
1 study
(Lister et al.,
2005)
800 vaginal
Ms
(repeat after
24 h)
1/18
(5.6%)
1/16
(6.3%)
RR 0.89
(0.06 to
13.08)
7 fewer per
1000
(from 59
fewer to 755
more)
Moderate
Adverse effects: severity of diarrhoea (maximum potential score not reported)
1 study
(Blohm et al.,
2005)
400 vaginal
misoprostol
Mean 7.5 (SD
15.0)
n=64
Mean 8.9
(SD 20.4)
n=62
NC MD 1.4 lower
(7.67 lower
to 4.87
higher)
p = 0.69
Low
Adverse effects: incidence of any gastrointestinal side effects
1 study
(Wood & Brain,
2002)
800 vaginal
misoprostol
(repeat after
24 h)
1/25
(4%)
not reported
(NR)
NC NC Low
Adverse effects: incidence of fever
1 study
(Kovavisarach
&
Sathapanachai,
2002)
400 vaginal
Ms
4/27
(14.8%)
0/27
(0%)
RR 9
(0.51 to
159.43)
NC Very low
Adverse effects: incidence of infection
1 study
(Blohm et al.,
2005)
400 vaginal
Ms
3/64
(4.7%)
0/62
(0%)
RR 6.78
(0.36 to
128.7)
NC Low
Adverse effects: incidence of pelvic inflammatory disease
127
1 study
(Bagratee et
al., 2004)
600 vaginal
Ms
(repeat after
24 h)
1/52
(1.9%)
0/52
(0%)
RR 3
(0.13 to
71.99)
NC Low
1 study
(Bagratee et
al., 2004)
600 vaginal
Ms
(repeat after
24 h)
Mean 11.65
(SD 4.4)
n=52
Mean 10.88
(SD 4.78)
n=52
NC MD 0.77
higher
(1 lower to
2.54 higher)
Moderate
Pain: incidence of menstrual cramping
1 study
(Lister et al.,
2005)
800 vaginal
Ms
(repeat after
24 h)
11/18
(61.1%)
5/16
(31.3%)
RR 1.96
(0.87 to
4.42)
300 more per
1000
(from 41
fewer to
1000 more)
Moderate
Pain: incidence of lower abdominal pain
1 study
(Kovavisarach
&
Sathapanachai,
2002)
400 vaginal
Ms
20/27
(74.1%)
6/27
(22.2%)
RR 3.33
(1.59 to
6.99)
518 more per
1000
(from 131
more to 1000
more)
Low
Pain: severity
1 study
(Blohm et al.,
2005)
400 vaginal
Ms
Mean 60.4
(SD 31.0)
n=64
Mean 43.8
(SD 37.1)
n=62
NC MD 16.6
higher
(4.64 to
28.56 higher)
p < 0.007
Low
1 study
(Bagratee et
al., 2004)
600 vaginal
Ms
(repeat after
24 h)
Mean 6.0
(SD 2.7)
n=52
Mean 5.4
(SD 2.7)
n=52
NC MD 0.6
higher
(0.44 lower
to 1.64
higher)
Moderate
1 study
(Lister et al.,
2005)
800 vaginal
Ms
(repeat after
24 h)
Mean 5.6
(SD NR)
n=16
Mean 5.2
(SD NR)
n=16
NC MD 0.4
higher
(confidence
intervals NC)
p = 0.806
Moderate
1 study
(Lister et al.,
2005)
800 vaginal
Ms
(repeat after
24 h)
14/15
(93.3%)
12/15
(80%)
RR 1.17
(0.88 to
1.55)
136 more per
1000
(from 96
fewer to 440
more)
Moderate
Satisfaction: score/10
128
1 study
(Bagratee et
al., 2004)
600 vaginal
Ms
(repeat after
24 h)
Mean 8.9
(SD 1.3)
n=52
Mean 8.7
(SD 1.5)
n=52
NC MD 0.2
higher
(0.34 lower
to 0.74
higher)
Moderate
1
Table 7.7 GRADE summary of findings for comparison of mifepristone with placebo for the management of 2
missed miscarriage 3
Number of
studies
Details of
treatment
regimen
Number of women or average Effect Quality
Mifepristone Placebo Relative
(95% CI)
Absolute
(95% CI)
1 study
(Lelaidier et
al., 1993)
600mg oral
Mf
19/23
(82.6%)
2/23
(8.7%)
RR 9.5
(2.49 to
36.19)
739 more
per 1000
(from 130
more to
1000 more)
Low
1 study
(Lelaidier et
al., 1993)
600mg oral
Mf
6/23
(26.1%)
19/21
(90.5%)
RR 0.29
(0.14 to
0.58)
642 fewer
per 1000
(from 380
fewer to 778
fewer)
Low
Adverse effects: incidence of endometritis
1 study
(Lelaidier et
al., 1993)
600mg oral
Mf
1/23
(4.3%)
1/21
(4.8%)
RR 0.91
(0.06 to
13.69)
4 fewer per
1000
(from 45
fewer to 604
more)
Low
Pain: incidence
1 study
(Lelaidier et
al., 1993)
600mg oral
Mf
12/23
(52.2%)
5/21
(23.8%)
RR 2.19
(0.93 to
5.17)
283 more
per 1000
(from 17
fewer to 993
more)
Low
4
Table 7.8 GRADE summary of findings for comparison of mifepristone plus misoprostol with misoprostol only for 5
the management of missed miscarriage 6
Number of
studies
Details of
treatment
regimen
(dose in
micrograms
unless
stated)
Number of women or
average
Effect Quality
Mifepristone
(Mf) +
misoprostol
(Ms)
Ms only Relative
(95% CI)
Absolute
(95% CI)
129
1 study
(Stockheim et
al., 2006)
Mf + Ms:
600mg oral
Mf, followed
48 hours
later by 400
oral Ms x 2
Ms only: 400
oral Ms x 2,
with a repeat
48 hours
later
38/58
(65.5%)
42/57
(73.7%)
RR 0.89
(0.7 to 1.13)
81 fewer per
1000
(from 221
fewer to 96
more)
Moderate
1 study
(Stockheim et
al., 2006)
Mf + Ms:
600mg oral
Mf, followed
48 hours
later by 400
oral Ms x 2
Ms only: 400
oral Ms x 2,
with the
same 48
hours later
20/58
(34.5%)
15/57
(26.3%)
RR 1.31
(0.75 to 2.3)
82 more per
1000
(from 66
fewer to 342
more)
Moderate
1
Table 7.9 GRADE summary of findings for comparison of vaginal misoprostol in different dosages for the 2
management of missed miscarriage 3
Number of
studies
Details of
treatment
regimen
(dose in
micrograms
unless
stated)
Number of women or
average
Effect Quality
600 vaginal 800 vaginal Relative
(95% CI)
Absolute
(95% CI)
1 study
(Kovavisarach
& Jamnansiri,
2005)
Vaginal
misoprostol
(Ms) 600 vs.
800
26/57
(45.6%)
39/57
(68.4%)
RR 0.67
(0.48 to
0.93)
226 fewer
per 1000
(from 48
fewer to 356
fewer)
High
1 study
(Kovavisarach
& Jamnansiri,
2005)
Vaginal Ms
600 vs. 800
2/57
(3.5%)
7/57
(12.3%)
RR 0.29
(0.06 to
1.32)
87 fewer per
1000
(from 115
fewer to 39
more)
Moderate
1 study Vaginal Ms 0/57 0/57 not NC Moderate
130
(Kovavisarach
& Jamnansiri,
2005)
600 vs. 800 (0%) (0%) calculable
(NC)
1 study
(Kovavisarach
& Jamnansiri,
2005)
Vaginal Ms
600 vs. 800
0/57
(0%)
2/57
(3.5%)
RR 0.2
(0.01 to
4.08)
28 fewer per
1000
(from 35
fewer to 108
more)
Moderate
1 study
(Kovavisarach
& Jamnansiri,
2005)
Vaginal Ms
600 vs. 800
10/57
(17.5%)
16/57
(28.1%)
RR 0.62
(0.31 to
1.26)
107 fewer
per 1000
(from 194
fewer to 73
more)
Moderate
Pain: incidence
1 study
(Kovavisarach
& Jamnansiri,
2005)
Vaginal Ms
600 vs. 800
30/57
(52.6%)
42/57
(73.7%)
RR 0.71
(0.53 to
0.96)
214 fewer
per 1000
(from 29
fewer to 346
fewer)
High
1
Table 7.10 GRADE summary of findings for comparison of sublingual misoprostol in different dosages for the 2
Number of
studies
Details of
treatment
regimen
(dose in
micrograms
unless
stated)
Number of women or
average
Effect Quality
600
sublingual
600 +
extended
course
sublingual
Relative
(95% CI)
Absolute
(95% CI)
1 study
(Tang et al.,
2006)
Sublingual
misoprostol
(Ms) 600 x 3
vs. 600 x 3 +
400 daily for
7 days
83/90
(92.2%)
84/90
(93.3%)
RR 0.99
(0.91 to
1.07)
9 fewer per
1000
(from 84
fewer to 65
more)
Moderate
Adverse effects: incidence of nausea on day 1
1 study
(Tang et al.,
2006)
Sublingual
Ms 600 x 3
vs. 600 x 3 +
400 daily for
7 days
38/90
(42.2%)
45/90
(50%)
RR 0.84
(0.61 to
1.16)
80 fewer per
1000
(from 195
fewer to 80
more)
Low
Adverse effects: incidence of nausea on days 2-9
1 study Sublingual
Ms 600 x 3
13/86 18/86 RR 0.72 59 fewer per Low
131
(Tang et al.,
2006)
vs. 600 x 3 +
400 daily for
7 days
(15.1%) (20.9%) (0.38 to
1.38)
1000
(from 130
fewer to 80
more)
Adverse effects: incidence of vomiting on day 1
1 study
(Tang et al.,
2006)
Sublingual
Ms 600 x 3
vs. 600 x 3 +
400 daily for
7 days
13/90
(14.4%)
14/90
(15.6%)
RR 0.93
(0.46 to
1.86)
11 fewer per
1000
(from 84
fewer to 134
more)
Low
Adverse effects: incidence of vomiting on days 2-9
1 study
(Tang et al.,
2006)
Sublingual
Ms 600 x 3
vs. 600 x 3 +
400 daily for
7 days
1/86
(1.2%)
5/86
(5.8%)
RR 0.2
(0.02 to
1.68)
47 fewer per
1000
(from 57
fewer to 40
more)
Low
Adverse effects: incidence of diarrhoea on day 1
1 study
(Tang et al.,
2006)
Sublingual
Ms 600 x 3
vs. 600 x 3 +
400 daily for
7 days
61/90
(67.8%)
63/90
(70%)
RR 0.97
(0.8 to 1.18)
21 fewer per
1000
(from 140
fewer to 126
more)
Moderate
Adverse effects: incidence of diarrhoea on days 2-9
1 study
(Tang et al.,
2006)
Sublingual
Ms 600 x 3
vs. 600 x 3 +
400 daily for
7 days
19/86
(22.1%)
38/86
(44.2%)
RR 0.5
(0.31 to
0.79)
221 fewer
per 1000
(from 93
fewer to 305
fewer)
Moderate
Adverse effects: incidence of fever on day 1
1 study
(Tang et al.,
2006)
Sublingual
Ms 600 x 3
vs. 600 x 3 +
400 daily for
7 days
52/90
(57.8%)
55/90
(61.1%)
RR 0.95
(0.74 to 1.2)
31 fewer per
1000
(from 159
fewer to 122
more)
Low
Adverse effects: incidence of fever on days 2-9
1 study
(Tang et al.,
2006)
Sublingual
Ms 600 x 3
vs. 600 x 3 +
400 daily for
7 days
0/86
(0%)
0/86
(0%)
not
calculable
(NC)
NC Low
Adverse effects: incidence of chills and rigor on day 1
1 study
(Tang et al.,
2006)
Sublingual
Ms 600 x 3
vs. 600 x 3 +
400 daily for
7 days
10/90
(11.1%)
13/90
(14.4%)
RR 0.77
(0.36 to
1.66)
33 fewer per
1000
(from 92
fewer to 95
more)
Low
132
Adverse effects: incidence of chills and rigor on days 2-9
1 study
(Tang et al.,
2006)
Sublingual
Ms 600 x 3
vs. 600 x 3 +
400 daily for
7 days
0/86
(0%)
0/86
(0%)
NC NC Low
1 study
(Tang et al.,
2006)
Sublingual
Ms 600 x 3
vs. 600 x 3 +
400 daily for
7 days
Median 11.5
(Range 5 - 35)
Median 11.0
(Range 6 -
42)
NC Median 0.5
higher
(confidence
interval NC)
NS (p-value
not reported)
Moderate
Pain: incidence on day 1
1 study
(Tang et al.,
2006)
Sublingual
Ms 600 x 3
vs. 600 x 3 +
400 daily for
7 days
88/90
(97.8%)
88/90
(97.8%)
RR 1
(0.96 to
1.05)
0 fewer per
1000
(from 39
fewer to 49
more)
Moderate
Pain: incidence on days 2-9
1 study
(Tang et al.,
2006)
Sublingual
Ms 600 x 3
vs. 600 x 3 +
400 daily for
7 days
66/86
(76.7%)
74/86
(86%)
RR 0.89
(0.77 to
1.03)
95 fewer per
1000
(from 198
fewer to 26
more)
Moderate
1
Table 7.11 GRADE summary of findings for comparison of oral misoprostol with sublingual misoprostol for the 2
Number of
studies
Details of
treatment
regimen
(dose in
micrograms
unless
stated)
Number of women or
average
Effect Quality
Oral Sublingual Relative
(95% CI)
Absolute
(95% CI)
1 study
(Ayudhaya et
al., 2006)
Oral: 400
misoprostol
(Ms) x 6
Sublingual:
400 Ms x 6
17/66
(25.8%)
15/70
(21.4%)
RR 1.2
(0.65 to
2.21)
43 more per
1000
(from 75
fewer to 259
more)
Low
1 study
(Kushwah &
Singh, 2009)
200mg oral
mifepristone
(Mf), plus:
Oral: 600 Ms
Sublingual:
42/50
(84%)
46/50
(92%)
RR 0.91
(0.79 to
1.06)
83 fewer per
1000
(from 193
fewer to 55
more)
Moderate
133
600 Ms
(+ 400 Ms x
3 after 12
hours if
needed)
Adverse effects: incidence of nausea or vomiting
1 study
(Ayudhaya et
al., 2006)
Oral: 400 Ms
x 6
Sublingual:
400 Ms x 6
3/66
(4.5%)
2/70
(2.9%)
RR 1.59
(0.27 to
9.22)
17 more per
1000
(from 21
fewer to 235
more)
Moderate
1 study
(Kushwah &
Singh, 2009)
200mg oral
Mf, plus:
Oral: 600 Ms
Sublingual:
600 Ms
(+ 400 Ms x
3 after 12
hours if
needed)
26/50
(52%)
17/50
(34%)
RR 1.53
(0.96 to
2.44)
180 more per
1000
(from 14
fewer to 490
more)
Low
1 study
(Kushwah &
Singh, 2009)
200mg oral
Mf, plus:
Oral: 600 Ms
Sublingual:
600 Ms
(+ 400 Ms x
3 after 12
hours if
needed)
22/50
(44%)
11/50
(22%)
RR 2
(1.09 to
3.68)
220 more per
1000
(from 20
more to 590
more)
Moderate
1 study
(Ayudhaya et
al., 2006)
Oral: 400 Ms
x 6
Sublingual:
400 Ms x 6
7/66
(10.6%)
6/70
(8.6%)
RR 1.24
(0.44 to
3.49)
21 more per
1000
(from 48
fewer to 213
more)
Moderate
1 study
(Kushwah &
Singh, 2009)
200mg oral
Mf, plus:
Oral: 600 Ms
Sublingual:
600 Ms
(+ 400 Ms x
3 after 12
hours if
needed)
28/50
(56%)
24/50
(48%)
RR 1.17
(0.8 to 1.7)
82 more per
1000
(from 96
fewer to 336
more)
Low
134
1 study
(Ayudhaya et
al., 2006)
Oral: 400 Ms
x 6
Sublingual:
400 Ms x 6
2/66
(3%)
15/70
(21.4%)
RR 0.14
(0.03 to
0.59)
184 fewer
per 1000
(from 88
fewer to 208
fewer)
High
1 study
(Kushwah &
Singh, 2009)
200mg oral
Mf, plus:
Oral: 600 Ms
Sublingual:
600 Ms
(+ 400 Ms x
3 after 12
hours if
needed)
26/50
(52%)
10/50
(20%)
RR 2.6
(1.41 to
4.81)
320 more per
1000
(from 82
more to 762
more)
Moderate
Adverse effects: incidence of chills
1 study
(Ayudhaya et
al., 2006)
Oral: 400 Ms
x 6
Sublingual:
400 Ms x 6
0/66
(0%)
4/70
(5.7%)
RR 0.12
(0.01 to
2.15)
50 fewer per
1000
(from 57
fewer to 66
more)
Moderate
Pain: incidence
1 study
(Ayudhaya et
al., 2006)
Oral: 400 Ms
x 6
Sublingual:
400 Ms x 6
40/66
(60.6%)
47/70
(67.1%)
RR 0.9
(0.7 to 1.16)
67 fewer per
1000
(from 201
fewer to 107
more)
Moderate
1 study
(Kushwah &
Singh, 2009)
200mg oral
Mf, plus:
Oral: 600 Ms
Sublingual:
600 Ms
(+ 400 Ms x
3 after 12
hours if
needed)
44/50
(88%)
23/50
(46%)
RR 1.91
(1.39 to
2.63)
419 more per
1000
(from 179
more to 750
more)
Moderate
1 study
(Kushwah &
Singh, 2009)
200mg oral
Mf, plus:
Oral: 600 Ms
Sublingual:
600 Ms
(+ 400 Ms x
3 after 12
hours if
needed)
36/50
(72%)
46/50
(92%)
RR 0.78
(0.65 to
0.95)
202 fewer
per 1000
(from 46
fewer to 322
fewer)
Moderate
135
1
Table 7.12 GRADE summary of findings for comparison of sublingual misoprostol with vaginal misoprostol for the 2
Number of
studies
Details of
treatment
regimen
(dose in
micrograms
unless
stated)
Number of women or
average
Effect Quality
Sublingual Vaginal Relative
(95% CI)
Absolute
(95% CI)
1 meta-
analysis of 2
studies
(Shah et al.,
2010; Tanha et
al., 2010)
Sublingual:
400 x 5/not
reported
(NR)
Vaginal: 400
x 5/NR
104/132
(78.8%)
61/129
(47.3%)
RR 1.4
(0.75 to
2.62)
189 more per
1000
(from 118
fewer to 766
more)
Very low
1 study
(Tang et al.,
2003)
Sublingual:
600 Ms x 3
Vaginal: 600
Ms x 3
35/40
(87.5%)
35/40
(87.5%)
RR 1
(0.85 to
1.18)
0 fewer per
1000
(from 131
fewer to 157
more)
Moderate
1 meta-
analysis of 2
studies
(Shah et al.,
2010; Tanha et
al., 2010)
Sublingual:
400 x 5/NR
Vaginal: 400
x 5/NR
28/135
(20.7%)
72/135
(53.3%)
RR 0.49
(0.16 to
1.44)
272 fewer
per 1000
(from 448
fewer to 235
more)
Very low
1 study
(Tang et al.,
2003)
Sublingual:
600 Ms x 3
Vaginal: 600
Ms x 3
4/39
(10.3%)
4/39
(10.3%)
RR 1
(0.27 to
3.72)
0 fewer per
1000
(from 75
fewer to 279
more)
Low
1 study
(Shah et al.,
2010)
Sublingual:
400 Ms x 5
Vaginal: 400
Ms x 5
5/25
(20%)
1/25
(4%)
RR 5
(0.63 to
39.79)
160 more per
1000
(from 15
fewer to
1552 more)
Low
1 study
(Tang et al.,
2003)
Sublingual:
600 Ms x 3
Vaginal: 600
Ms x 3
24/40
(60%)
20/40
(50%)
RR 1.2
(0.8 to 1.79)
100 more per
1000
(from 100
fewer to 395
more)
Moderate
136
1 study
(Tanha et al.,
2010)
Sublingual:
400 Ms x
NR
Vaginal: 400
Ms x NR
22/110
(20%)
13/110
(11.8%)
RR 1.69
(0.9 to 3.19)
82 more per
1000
(from 12
fewer to 259
more)
Low
1 study
(Tang et al.,
2003)
Sublingual:
600 Ms x 3
Vaginal: 600
Ms x 3
7/40
(17.5%)
9/40
(22.5%)
RR 0.78
(0.32 to
1.88)
50 fewer per
1000
(from 153
fewer to 198
more)
Moderate
1 study
(Tanha et al.,
2010)
Sublingual:
400 Ms x
NR
Vaginal: 400
Ms x NR
76/110
(69.1%)
40/110
(36.4%)
RR 1.9
(1.44 to
2.51)
327 more per
1000
(from 160
more to 549
more)
Moderate
1 study
(Tang et al.,
2003)
Sublingual:
600 Ms x 3
Vaginal: 600
Ms x 3
28/40
(70%)
11/40
(27.5%)
RR 2.55
(1.48 to
4.38)
426 more per
1000
(from 132
more to 930
more)
High
1 study
(Tanha et al.,
2010)
Sublingual:
400 Ms x
NR
Vaginal: 400
Ms x NR
26/110
(23.6%)
4/110
(3.6%)
RR 6.5
(2.35 to
18.01)
200 more per
1000
(from 49
more to 619
more)
Moderate
1 study
(Tang et al.,
2003)
Sublingual:
600 Ms x 3
Vaginal: 600
Ms x 3
23/40
(57.5%)
19/40
(47.5%)
RR 1.21
(0.79 to
1.84)
100 more per
1000
(from 100
fewer to 399
more)
Moderate
Adverse effects: incidence of chills or shivering
1 study
(Shah et al.,
2010)
Sublingual:
400 Ms x 5
Vaginal: 400
Ms x 5
6/25
(24%)
4/25
(16%)
RR 1.5
(0.48 to
4.68)
80 more per
1000
(from 83
fewer to 589
more)
Low
1 study
(Tang et al.,
2003)
Sublingual:
600 Ms x 3
Vaginal: 600
Ms x 3
6/40
(15%)
3/40
(7.5%)
RR 2
(0.54 to
7.45)
75 more per
1000
(from 34
fewer to 484
more)
Moderate
1 study
(Tang et al.,
Sublingual:
600 Ms x 3
Vaginal: 600
Median 12.5
(Range 4
36)
Median 12.0
(Range 5
79)
not
calculable
(NC)
Median 0.5
higher
(confidence
High
137
2003) Ms x 3 n=40 n=40 interval NC)
NS (p-value
NR)
Pain: incidence of cramps
1 study
(Tanha et al.,
2010)
Sublingual:
400 Ms x
NR
Vaginal: 400
Ms x NR
94/110
(85.5%)
62/110
(56.4%)
RR 1.52
(1.26 to
1.82)
293 more per
1000
(from 147
more to 462
more)
Moderate
Pain: incidence of severe pain
1 study
(Tanha et al.,
2010)
Sublingual:
400 Ms x
NR
Vaginal: 400
Ms x NR
77/110
(70%)
42/110
(38.2%)
RR 1.83
(1.4 to 2.4)
317 more per
1000
(from 153
more to 535
more)
Moderate
Pain: incidence of lower abdominal pain
1 study
(Tang et al.,
2003)
Sublingual:
600 Ms x 3
Vaginal: 600
Ms x 3
40/40
(100%)
40/40
(100%)
NC NC High
1 meta-
analysis of 2
studies
(Shah et al.,
2010; Tanha et
al., 2010)
Sublingual:
400 x 5/NR
Vaginal: 400
x 5/NR
116/135
(85.9%)
71/135
(52.6%)
RR 1.48
(0.94 to
2.32)
252 more per
1000
(from 32
fewer to 694
more)
Very low
1
Table 7.13 GRADE summary of findings for comparison of oral misoprostol with vaginal misoprostol for the 2
Number of
studies
Details of
treatment
regimen
(dose in
micrograms
unless
stated)
Number of women or
average
Effect Quality
Oral Vaginal Relative
(95% CI)
Absolute
(95% CI)
1 study
(Rita et al.,
2006)
Oral: 400
misoprostol
(Ms) x 3
Vaginal: 600
Ms x 2
18/50
(36%)
40/50
(80%)
RR 0.45
(0.3 to 0.67)
440 fewer
per 1000
(from 264
fewer to 560
fewer)
Moderate
1 study
(Creinin et al.,
Oral: 400 Ms
Vaginal: 800
3/12
(25%)
7/8
(87.5%)
RR 0.29
(0.1 to 0.79)
621 fewer
per 1000
(from 184
Very low
138
1997) Ms
(repeat after
24 h)
fewer to 788
fewer)
1 study
(Ngoc et al.,
2004)
Oral: 800 Ms
Vaginal: 800
Ms
89/101
(88.1%)
91/99
(91.9%)
RR 0.96
(0.87 to
1.05)
37 fewer per
1000
(from 119
fewer to 46
more)
Moderate
1 study
(Rita et al.,
2006)
Oral: 400 Ms
x 3
Vaginal: 600
Ms x 2
32/50
(64%)
10/50
(20%)
RR 3.2
(1.77 to
5.78)
440 more per
1000
(from 154
more to 956
more)
Moderate
1 study
(Ngoc et al.,
2004)
Oral: 800 Ms
Vaginal: 800
Ms
11/100
(11%)
7/98
(7.1%)
RR 1.54
(0.62 to
3.81)
39 more per
1000
(from 27
fewer to 201
more)
Low
Admission to a medical facility
1 study
(Ngoc et al.,
2004)
Oral: 800 Ms
Vaginal: 800
Ms
0/100 (0%) 2/98 (2%) RR 0.2
(0.01 to
4.03)
16 fewer per
1000
(from 20
fewer to 62
more)
Very low
1 study
(Rita et al.,
2006)
Oral: 400 Ms
x 3
Vaginal: 600
Ms x 2
25/50
(50%)
20/50
(40%)
RR 1.25
(0.81 to
1.94)
100 more per
1000
(from 76
fewer to 376
more)
Low
1 study
(Creinin et al.,
1997)
Oral: 400 Ms
Vaginal: 800
Ms
(repeat after
24 h)
6/12
(50%)
5/8
(62.5%)
RR 0.8
(0.37 to
1.74)
125 fewer
per 1000
(from 394
fewer to 463
more)
Very low
1 study
(Rita et al.,
2006)
Oral: 400 Ms
x 3
Vaginal: 600
Ms x 2
6/50
(12%)
3/50
(6%)
RR 2
(0.53 to
7.56)
60 more per
1000
(from 28
fewer to 394
more)
Low
1 study
(Creinin et al.,
1997)
Oral: 400 Ms
Vaginal: 800
Ms
(repeat after
3/12
(25%)
1/8
(12.5%)
RR 2
(0.25 to
15.99)
125 more per
1000
(from 94
fewer to
Very low
139
24 h) 1874 more)
1 study
(Ngoc et al.,
2004)
Oral: 800 Ms
Vaginal: 800
Ms
4/95
(4.2%)
14/95
(14.7%)
RR 0.29
(0.1 to 0.84)
105 fewer
per 1000
(from 24
fewer to 133
fewer)
High
1 study
(Rita et al.,
2006)
Oral: 400 Ms
x 3
Vaginal: 600
Ms x 2
5/50
(10%)
5/50
(10%)
RR 1
(0.31 to
3.24)
0 fewer per
1000
(from 69
fewer to 224
more)
Low
1 study
(Creinin et al.,
1997)
Oral: 400 Ms
Vaginal: 800
Ms
(repeat after
24 h)
5/12
(41.7%)
3/8
(37.5%)
RR 1.11
(0.36 to 3.4)
41 more per
1000
(from 240
fewer to 900
more)
Very low
1 study
(Ngoc et al.,
2004)
Oral: 800 Ms
Vaginal: 800
Ms
24/95
(25.3%)
23/95
(24.2%)
RR 1.04
(0.64 to
1.71)
10 more per
1000
(from 87
fewer to 172
more)
Moderate
Adverse effects: incidence of hyperpyrexia
1 study
(Rita et al.,
2006)
Oral: 400 Ms
x 3
Vaginal: 600
Ms x 2
2/50
(4%)
2/50
(4%)
RR 1
(0.15 to
6.82)
0 fewer per
1000
(from 34
fewer to 233
more)
Low
Adverse effects: incidence of fever or chills
1 study
(Ngoc et al.,
2004)
Oral: 800 Ms
Vaginal: 800
Ms
7/95
(7.4%)
7/95
(7.4%)
RR 1
(0.36 to
2.74)
0 fewer per
1000
(from 47
fewer to 128
more)
Moderate
1 study
(Creinin et al.,
1997)
Oral: 400 Ms
Vaginal: 800
Ms
(repeat after
24 h)
not reported
(NR)
Mean 10.0
(SD 2.8)
n=7
not
calculable
(NC)
NC Very low
1 study
(Ngoc et al.,
2004)
Oral: 800 Ms
Vaginal: 800
Ms
Mean 2.87
(SD NR)
n=95
Mean 2.69
(SD NR)
n=95
NC MD 0.18
higher
(confidence
interval NC)
NS (p-value
Moderate
140
not reported)
Pain: incidence
1 study
(Rita et al.,
2006)
Oral: 400 Ms
x 3
Vaginal: 600
Ms x 2
8/50
(16%)
5/50
(10%)
RR 1.6
(0.56 to
4.56)
60 more per
1000
(from 44
fewer to 356
more)
Low
1 study
(Ngoc et al.,
2004)
Oral: 800 Ms
Vaginal: 800
Ms
84/95
(88.4%)
85/95
(89.5%)
RR 0.99
(0.89 to
1.09)
9 fewer per
1000
(from 98
fewer to 81
more)
High
Pain: severity/10
1 study
(Creinin et al.,
1997)
Oral: 400 Ms
Vaginal: 800
Ms
(repeat after
24 h)
Mean 4.0
(SD 3.6)
n=11
Mean 5.9
(SD 2.7)
n=7
NC MD 1.9 lower
(4.82 lower
to 1.02
higher)
p = 0.33
Very low
1 study
(Ngoc et al.,
2004)
Oral: 800 Ms
Vaginal: 800
Ms
86/100
(86%)
88/98
(89.8%)
RR 0.96
(0.86 to
1.06)
36 fewer per
1000
(from 126
fewer to 54
more)
High
1
Table 7.14 GRADE summary of findings for comparison of vaginal misoprostol with placebo for the management 2
of incomplete miscarriage 3
Number of
studies
Details of
treatment
regimen
(dose in
micrograms
unless
stated)
Number of women or
average
Effect Quality
Misoprostol Placebo Relative
(95% CI)
Absolute
(95% CI)
1 study
(Bagratee et
al., 2004)
600 vaginal
misoprostol
(Ms)
(repeat after
24h)
7/7
(100%)
12/14
(85.7%)
RR 1.12
(0.84 to 1.5)
103 more per
1000
(from 137
fewer to 429
more)
Low
1 study
(Blohm et al.,
2005)
400 vaginal
Ms
0/7
(0%)
2/11
(18.2%)
RR 0.3
(0.02 to
5.46)
127 fewer
per 1000
(from 178
fewer to 811
Low
141
more)
1 study
(Bagratee et
al., 2004)
600 vaginal
Ms
(repeat after
24h)
0/7
(0%)
2/14
(14.3%)
RR 0.38
(0.02 to 6.9)
89 fewer per
1000
(from 140
fewer to 843
more)
Low
1
Table 7.15 GRADE summary of findings for comparison of oral misoprostol in different dosages for the 2
management of incomplete miscarriage 3
Number of
studies
Details of
treatment
regimen
(dose in
micrograms
unless
stated)
Number of women or
average
Effect Quality
600 oral 2 x 600 oral Relative
(95% CI)
Absolute
(95% CI)
1 meta-
analysis of 2
studies
(Blanchard et
al., 2004; Ngoc
et al., 2005)
Oral
misoprostol
(Ms) 600 vs.
2 x 600
199/236
(84.3%)
195/233
(83.7%)
RR 1.01
(0.94 to
1.09)
8 more per
1000
(from 50
fewer to 75
more)
Moderate
1 meta-
analysis of 2
studies
(Blanchard et
al., 2004; Ngoc
et al., 2005)
Oral Ms 600
vs.
2 x 600
35/234
(15%)
32/227
(14.1%)
RR 1.05
(0.69 to
1.59)
7 more per
1000
(from 44
fewer to 83
more)
Low
1 meta-
analysis of 2
studies
(Blanchard et
al., 2004; Ngoc
et al., 2005)
Oral Ms 600
vs.
2 x 600
48/235
(20.4%)
37/228
(16.2%)
RR 1.19
(0.57 to
2.46)
31 more per
1000
(from 70
fewer to 237
more)
Very low
1 meta-
analysis of 2
studies
(Blanchard et
al., 2004; Ngoc
et al., 2005)
Oral Ms 600
vs.
2 x 600
25/235
(10.6%)
24/228
(10.5%)
RR 1.01
(0.6 to 1.72)
1 more per
1000
(from 42
fewer to 76
more)
Low
1 study
(Ngoc et al.,
Oral Ms 600
vs.
51/149 68/145 RR 0.73
(0.55 to
127 fewer
per 1000
Moderate
142
2005) 2 x 600 (34.2%) (46.9%) 0.97) (from 14
fewer to 211
fewer)
Adverse effects: incidence of fever or chills
1 meta-
analysis of 2
studies
(Blanchard et
al., 2004; Ngoc
et al., 2005)
Oral Ms 600
vs.
2 x 600
27/235
(11.5%)
22/228
(9.6%)
RR 1.19
(0.7 to 2.02)
18 more per
1000
(from 29
fewer to 98
more)
Low
Duration of heavy bleeding (days)
1 study
(Ngoc et al.,
2005)
Oral Ms 600
vs.
2 x 600
Mean 0.8
(SD 0.8)
n=149
Mean 0.8
(SD 0.7)
n=145
not
calculable
(NC)
MD 0 higher
(0.17 lower
to 0.17
higher)
NS (p-value
NR)
Moderate
1 study
(Blanchard et
al., 2004)
Oral Ms 600
vs.
2 x 600
Mean 1.31
(SD not
reported (NR))
n=86
Mean 1.63
(SD NR)
n=83
NC MD 0.32
lower
(confidence
interval NC)
p = 0.21
Low
Duration of normal bleeding (days)
1 study
(Ngoc et al.,
2005)
Oral Ms 600
vs.
2 x 600
Mean 1.2
(SD 0.9)
n=149
Mean 1.2
(SD 1.2)
n=145
NC MD 0 higher
(0.24 lower
to 0.24
higher)
NS (p-value
NR)
Moderate
1 study
(Blanchard et
al., 2004)
Oral Ms 600
vs.
2 x 600
Mean 2.86
(SD NR)
n=86
Mean 2.76
(SD NR)
n=83
NC MD 0.1
higher
(confidence
interval NC)
p = 0.79
Low
Duration of light bleeding or spotting (days)
1 study
(Ngoc et al.,
2005)
Oral Ms 600
vs.
2 x 600
Mean 2.1
(SD 2.1)
n=149
Mean 1.8
(SD 2.1)
n=145
NC MD 0.3
higher
(0.18 lower
to 0.78
higher)
NS (p-value
NR)
Moderate
1 study
(Blanchard et
al., 2004)
Oral Ms 600
vs.
2 x 600
Mean 2.94
(SD NR)
n=86
Mean 2.88
(SD NR)
n=83
NC MD 0.06
higher
(confidence
interval NC)
Low
143
p = 0.89
Pain: incidence
1 meta-
analysis of 2
studies
(Blanchard et
al., 2004; Ngoc
et al., 2005)
Oral Ms 600
vs.
2 x 600
182/235
(77.4%)
183/228
(80.3%)
RR 0.97
(0.88 to
1.06)
24 fewer per
1000
(from 96
fewer to 48
more)
Moderate
Pain: severity/7
1 study
(Ngoc et al.,
2005)
Oral Ms 600
vs.
2 x 600
Mean 3.7 (SD
NR)
n=149
Mean 3.6
(SD NR)
n=145
NC MD 0.1
higher
(confidence
interval NC)
NS
(p-value NR)
Low
1 study
(Blanchard et
al., 2004)
Oral Ms 600
vs.
2 x 600
Mean 3.65
(SD NR)
n=85
Mean 4.09
(SD NR)
n=81
NC MD 0.44
lower
(confidence
interval NC)
p = 0.20
Low
1 meta-
analysis of 2
studies
(Blanchard et
al., 2004; Ngoc
et al., 2005)
Oral Ms 600
vs.
2 x 600
211/234
(90.2%)
199/226
(88.1%)
RR 1.02
(0.96 to
1.09)
18 more per
1000
(from 35
fewer to 79
more)
Moderate
1
Table 7.16 GRADE summary of findings for comparison of oral misoprostol with vaginal misoprostol for the 2
Number of
studies
Details of
treatment
regimen
(dose in
micrograms
unless
stated)
Number of women or
average
Effect Quality
Oral Vaginal Relative
(95% CI)
Absolute
(95% CI)
1 study
(Pang et al.,
2001)
Oral: 800
misoprostol
(Ms) x 2
Vaginal: 800
Ms x 2
67/105
(63.8%)
58/96
(60.4%)
RR 1.06
(0.85 to
1.31)
36 more per
1000
(from 91
fewer to 197
more)
Very low
1 study Oral: 800 Ms 36/103 37/95 RR 0.9 39 fewer per Very low
144
(Pang et al.,
2001)
x 2
Vaginal: 800
Ms x 2
(35%) (38.9%) (0.62 to
1.29)
1000
(from 148
fewer to 113
more)
1 study
(Pang et al.,
2001)
Oral: 800Ms
x 2
Vaginal: 800
Ms x 2
12/103
(11.7%)
7/95
(7.4%)
RR 1.58
(0.65 to
3.85)
43 more per
1000
(from 26
fewer to 210
more)
Low
1 study
(Pang et al.,
2001)
Oral: 800 Ms
x 2
Vaginal: 800
Ms x 2
6/103
(5.8%)
2/95
(2.1%)
RR 2.77
(0.57 to
13.38)
37 more per
1000
(from 9 fewer
to 261 more)
Low
1 study
(Pang et al.,
2001)
Oral: 800 Ms
x 2
Vaginal: 800
Ms x 2
62/103
(60.2%)
12/95
(12.6%)
RR 4.77
(2.74 to
8.27)
476 more per
1000
(from 220
more to 918
more)
Moderate
1 study
(Pang et al.,
2001)
Oral: 800 Ms
x 2
Vaginal: 800
Ms x 2
6/103
(5.8%)
11/95
(11.6%)
RR 0.5
(0.19 to
1.31)
58 fewer per
1000
(from 94
fewer to 36
more)
Low
1 study
(Pang et al.,
2001)
Oral: 800 Ms
x 2
Vaginal: 800
Ms x 2
Median 8
(Range 0 - 14)
n=97
Median 8
(Range 0 -
14)
n=89
not
calculable
(NC)
Median 0
higher
(confidence
intervals NC)
NS
(p-value not
reported
(NR))
Moderate
Pain: duration of pelvic pain (days)
1 study
(Pang et al.,
2001)
Oral: 800 Ms
x 2
Vaginal: 800
Ms x 2
Median 1
(Range 0 - 14)
n=97
Median 2
(Range 0 -
11)
n=89
NC Median 1
lower
(confidence
interval NC)
p = 0.02
Moderate
1
145
Table 7.17 GRADE summary of findings for comparison of oral misoprostol with sublingual misoprostol for the 1
Number of
studies
Details of
treatment
regimen
(dose in
micrograms
unless
stated)
Number of women or
average
Effect Quality
Oral Sublingual Relative
(95% CI)
Absolute
(95% CI)
1 study
(Paritakul &
Phupong,
2011)
Oral: 600
misoprostol
(Ms)
Sublingual:
600 Ms
28/32
(87.5%)
27/32
(84.4%)
RR 1.04
(0.85 to
1.26)
34 more per
1000
(from 127
fewer to 219
more)
Low
1 study
(Paritakul &
Phupong,
2011)
Oral: 600 Ms
Sublingual:
600 Ms
2/32
(6.3%)
5/32
(15.6%)
RR 0.4
(0.08 to
1.91)
94 fewer per
1000
(from 144
fewer to 142
more)
Low
1 study
(Paritakul &
Phupong,
2011)
Oral: 600 Ms
Sublingual:
600 Ms
7/32
(21.9%)
8/32
(25%)
RR 0.88
(0.36 to
2.13)
30 fewer per
1000
(from 160
fewer to 283
more)
Moderate
1 study
(Paritakul &
Phupong,
2011)
Oral: 600 Ms
Sublingual:
600 Ms
0/32
(0%)
0/32
(0%)
not
calculable
(NC)
NC Moderate
1 study
(Paritakul &
Phupong,
2011)
Oral: 600 Ms
Sublingual:
600 Ms
5/32
(15.6%)
9/32
(28.1%)
RR 0.56
(0.21 to
1.48)
124 fewer
per 1000
(from 222
fewer to 135
more)
Moderate
Adverse effects: incidence of fever/chills
1 study
(Paritakul &
Phupong,
2011)
Oral: 600 Ms
Sublingual:
600 Ms
9/32
(28.1%)
14/32
(43.8%)
RR 0.64
(0.33 to
1.27)
157 fewer
per 1000
(from 293
fewer to 118
more)
Moderate
Incidence of heavy bleeding
1 study Oral: 600 Ms 0/32 0/32 NC NC Moderate
146
(Paritakul &
Phupong,
2011)
Sublingual:
600 Ms
(0%) (0%)
Pain: incidence of pain/cramps
1 study
(Paritakul &
Phupong,
2011)
Oral: 600 Ms
Sublingual:
600 Ms
8/32
(25%)
10/32
(31.3%)
RR 0.8
(0.36 to
1.76)
62 fewer per
1000
(from 200
fewer to 237
more)
Moderate
Pain: severity/100
1 study
(Paritakul &
Phupong,
2011)
Oral: 600 Ms
Sublingual:
600 Ms
Mean 22.2
(SD 15.0)
n = 32
Mean 29.1
(SD 21.2)
n = 32
NC MD 6.9 lower
(15.9 lower
to 2.1 higher)
(p = 0.139)
High
1 study
(Paritakul &
Phupong,
2011)
Oral: 600 Ms
Sublingual:
600 Ms
28/32
(87.5%)
27/32
(84.4%)
RR 1.04
(0.85 to
1.26)
34 more per
1000
(from 127
fewer to 219
more)
Moderate
1
Management of missed miscarriage: comparison of vaginal misoprostol and 3
placebo 4
Success of medical treatment 5
One study found that the success of medical treatment was higher in women who received vaginal 6
misoprostol (600 micrograms, repeat after 24 hours) compared with women who received a placebo. 7
A meta-analysis of two studies also found that the success of medical treatment was higher in women 8
who received vaginal misoprostol (800 micrograms, repeat after 24 hours) compared with women who 9
received a placebo. These findings were statistically significant. The evidence for these findings was 10
of high quality. However, another meta-analysis of two studies did not find a statistically significant 11
difference in this outcome between women who received vaginal misoprostol (400 micrograms, 1 12
dose) and women who received a placebo. The evidence for this finding was of very low quality. 13
Need for further intervention 14
One study found that the need for further intervention was lower in women who received vaginal 15
misoprostol (400 micrograms, 1 dose) compared with women who received a placebo. A second 16
study found that the need for further intervention was lower in women who received vaginal 17
misoprostol (600 micrograms, repeat after 24 hours) compared with women who received a placebo. 18
A meta-analysis of two studies also found that the need for further intervention was lower in women 19
received a placebo. These findings were statistically significant. The evidence for this outcome was of 21
high quality. 22
One study did not find a statistically significant difference in the unplanned visits to a medical facility 24
for women who received vaginal misoprostol (800 micrograms, repeat after 24 hours) compared with 25
women who received a placebo. The evidence for this outcome was of moderate quality. 26
147
Adverse effects 1
One study did not find a statistically significant difference in the incidence of nausea and/or vomiting 2
for women who received vaginal misoprostol (400 micrograms) compared with women who received a 3
placebo. The evidence for this outcome was of very low quality. 4
One study did not find a statistically significant difference in the incidence of nausea for women who 5
received vaginal misoprostol (600 micrograms, repeat after 24 hours) compared with women who 6
received a placebo. One further study did not find a statistically significant difference in the incidence 7
of nausea for women who received vaginal misoprostol (800 micrograms, repeat after 24 hours) 8
compared with women who received a placebo. The quality of evidence for this outcome was low in 9
one study and moderate in the other. 10
One study did not find a statistically significant difference in the severity of nausea for women who 11
received vaginal misoprostol (400 micrograms) compared with women who received a placebo. The 12
One study did not find a statistically significant difference in the incidence of vomiting for women who 14
compared with women who received a placebo. The quality of evidence for this outcome was low in 18
one study and moderate in the other. 19
One study did not find a statistically significant difference in the severity of vomiting for women who 20
One study did not find a statistically significant difference in the incidence of diarrhoea for women who 23
received vaginal misoprostol (400 micrograms) compared with women who received a placebo. 24
Another study did not find a statistically significant difference in the incidence of diarrhoea for women 25
compared with women who received a placebo. The quality of evidence for this outcome was very low 29
in one study, low in one study and moderate in the other. 30
One study did not find a statistically significant difference in the severity of diarrhoea for women who 31
One study reported the incidence of any gastrointestinal side effects in a manner that did not permit 34
statistical comparison between the two arms. The evidence for this outcome was of low quality. 35
One study did not find a statistically significant difference in the incidence of fever for women who 36
One study did not find a statistically significant difference in the incidence of infection for women who 39
One study did not find a statistically significant difference in the incidence of pelvic inflammatory 42
disease for women who received vaginal misoprostol (600 micrograms, repeat after 24 hours) 43
compared with women who received a placebo. The evidence for this outcome was of low quality. 44
One study did not find a statistically significant difference in the duration of bleeding for women who 46
received a placebo. The evidence for this outcome was of moderate quality. 48
148
Pain 1
One study did not find a statistically significant difference in the incidence of menstrual cramping for 2
women who received vaginal misoprostol (800 micrograms, repeat after 24 hours) compared with 3
women who received a placebo. The evidence for this outcome was of moderate quality. 4
One study found that the incidence of lower abdominal pain was higher in women who received 5
vaginal misoprostol (400 micrograms, 1 dose) compared with women who received a placebo. This 6
finding was statistically significant. The evidence for this outcome was of low quality. 7
One study found that the severity of pain was higher in women who received vaginal misoprostol (400 8
micrograms, 1 dose) compared with women who received a placebo. This finding was statistically 9
significant. The evidence for this finding was of low quality. However, another study did not find a 10
statistically significant difference in this outcome between women who received vaginal misoprostol 11
(600 micrograms, repeat after 24 hours) and women who received a placebo. A further study did not 12
find a statistically significant difference in this outcome between women who received vaginal 13
misoprostol (800 micrograms, repeat after 24 hours) and women who received a placebo. The 14
evidence for this finding was of moderate quality. 15
Satisfaction 16
One study did not find a statistically significant difference in the incidence of satisfaction for women 17
One study did not find a statistically significant difference in the satisfaction score for women who 20
received vaginal misoprostol (600 micrograms, repeat after 24 hours)) compared with women who 21
Management of missed miscarriage: comparison of mifepristone and placebo 23
One study found that the success of medical treatment was higher in women who received 25
mifepristone compared with women who received a placebo. This finding was statistically significant. 26
The evidence for this outcome was of low quality. 27
One study found that the need for further intervention was lower in women who received mifepristone 29
compared with women who received a placebo. This finding was statistically significant. The evidence 30
for this outcome was of low quality. 31
Adverse effects 32
One study did not find a statistically significant difference in the incidence of endometritis for women 33
who received mifepristone compared with women who received a placebo. The evidence for this 34
Pain 36
One study did not find a statistically significant difference in the incidence of pain for women who 37
received mifepristone compared with women who received a placebo. The evidence for this outcome 38
Management of missed miscarriage: comparison of a combined regimen of 40
mifepristone plus misoprostol and misoprostol only 41
One study did not find a statistically significant difference in the success of medical treatment for 43
women who received mifepristone plus misoprostol compared with women who received misoprostol 44
only. The evidence for this outcome was of moderate quality. 45
One study did not find a statistically significant difference in the need for further intervention for 47
women who received mifepristone plus misoprostol compared with women who received misoprostol 48
only. The evidence for this outcome was of moderate quality. 49
149
Management of missed miscarriage: comparison of vaginal misoprostol dosages 1
One study found that the success of medical treatment was higher in women who received 800 3
micrograms of vaginal misoprostol compared with women who received 600 micrograms of vaginal 4
misoprostol. This finding was statistically significant. The evidence for this outcome was of high 5
quality. 6
Adverse effects 7
received 800 micrograms of vaginal misoprostol compared with women who received 600 micrograms 9
of vaginal misoprostol. The evidence for this outcome was of moderate quality. 10
There were no events in either arm for the outcome of incidence of vomiting. The evidence for this 11
Pain 19
One study found that the incidence of pain was higher in women who received 800 micrograms of 20
vaginal misoprostol compared with women who received 600 micrograms of vaginal misoprostol. This 21
finding was statistically significant. The evidence for this outcome was of high quality. 22
Management of missed miscarriage: comparison of sublingual misoprostol 23
dosages 24
women who received a single dose of sublingual misoprostol (600 micrograms) compared with 27
women who received an extended course of sublingual misoprostol (600 micrograms on day 1 and 28
then 400 micrograms daily on days 2-9). The evidence for this outcome was of moderate quality. 29
Adverse effects 30
One study did not find a statistically significant difference in the incidence of nausea on day 1 or days 31
2-9 for women who received a single dose of sublingual misoprostol (600 micrograms) compared with 32
then 400 micrograms daily on days 2-9). The evidence for these outcomes was of low quality. 34
One study did not find a statistically significant difference in the incidence of vomiting on day 1 or days 35
2-9 for women who received a single dose of sublingual misoprostol (600 micrograms) compared with 36
then 400 micrograms daily on days 2-9). The evidence for these outcomes was of low quality. 38
One study did not find a statistically significant difference in the incidence of diarrhoea on day 1 for 39
women who received a single dose of sublingual misoprostol (600 micrograms) compared with 40
then 400 micrograms daily on days 2-9). The same study found that the incidence of diarrhoea on 42
days 2-9 after treatment was higher in women who received an extended course of sublingual 43
misoprostol (600 micrograms on day 1 and then 400 micrograms daily on days 2-9) compared with 44
women who received a single dose of sublingual misoprostol (600 micrograms). The evidence for 45
these outcomes was of moderate quality. 46
One study did not find a statistically significant difference in the incidence of fever on day 1 for women 47
who received a single dose of sublingual misoprostol (600 micrograms) compared with women who 48
received an extended course of sublingual misoprostol (600 micrograms on day 1 and then 400 49
micrograms daily on days 2-9). The same study reported no events in either arm for the incidence of 50
fever on days 2-9. The evidence for these outcomes was of low quality. 51
150
One study did not find a statistically significant difference in the incidence of chills and rigour on day 1 1
for women who received a single dose of sublingual misoprostol (600 micrograms) compared with 2
then 400 micrograms daily on days 2-9). The same study reported no events in either arm for the 4
incidence of chills and rigour on days 2-9. The evidence for these outcomes was of low quality. 5
received a single dose of sublingual misoprostol (600 micrograms) compared with women who 8
received an extended course of sublingual misoprostol (600 micrograms on day 1 and then 400 9
micrograms daily on days 2-9). The evidence for this outcome was of moderate quality. 10
Pain 11
One study did not find a statistically significant difference in the incidence of pain on day 1 or days 2-9 12
for women who received a single dose of sublingual misoprostol (600 micrograms) compared with 13
then 400 micrograms daily on days 2-9). The evidence for these outcomes was of moderate quality. 15
Management of missed miscarriage: comparison of oral and sublingual 16
misoprostol 17
women who received oral misoprostol compared with women who received the same dose of 20
sublingual misoprostol (400 micrograms, 6 doses). Another study did not find a statistically significant 21
difference in the success of medical treatment for women who received oral misoprostol compared 22
with women who receive the same dose of sublingual misoprostol (200mg of mifepristone plus 600 23
micrograms of misoprostol, with 3 supplemental doses of 400 micrograms of misoprostol after 12 24
hours). The quality of evidence for this outcome was low quality in one study and moderate in the 25
other. 26
Adverse effects 27
One study did not find a statistically significant difference in the incidence of nausea or vomiting for 28
sublingual misoprostol (400 micrograms, 6 doses). The evidence for this outcome was of moderate 30
quality. 31
received oral misoprostol compared with women who receive the same dose of sublingual 33
misoprostol (200mg of mifepristone plus 600 micrograms of misoprostol, with 3 supplemental doses 34
of 400 micrograms of misoprostol after 12 hours). The evidence for this outcome was of low quality. 35
One study found that the incidence of vomiting was higher in women who received oral misoprostol 36
compared with women who received the same dose of sublingual misoprostol (200mg of mifepristone 37
plus 600 micrograms of misoprostol, with 3 supplemental doses of 400 micrograms of misoprostol 38
after 12 hours). This finding was statistically significant. The evidence for this outcome was of 39
One study found that the incidence of fever was higher in women who received sublingual misoprostol 41
compared with women who received the same dose of oral misoprostol (400 micrograms, 6 doses). 42
This finding was statistically significant. The evidence for this finding was of high quality. However, 43
another study found that the incidence of fever was higher in women who received oral misoprostol 44
compared with women who received the same dose of sublingual misoprostol (200mg of mifepristone 45
plus 600 micrograms of misoprostol, with 3 supplemental doses of 400 micrograms of misoprostol 46
after 12 hours). This finding was statistically significant. The evidence for this finding was of moderate 47
quality. 48
One study did not find a statistically significant difference in the incidence of chills for women who 49
received oral misoprostol compared with women who received the same dose of sublingual 50
misoprostol (400 micrograms, 6 doses). The evidence for this outcome was of moderate quality. 51
151
Pain 1
One study found that the incidence of pain was higher in women who received oral misoprostol 2
compared with women who received the same dose of sublingual misoprostol (regimen for both study 3
groups: 200mg of mifepristone plus 600 micrograms of misoprostol, with 3 supplemental doses of 400 4
micrograms of misoprostol after 12 hours). This finding was statistically significant. The evidence for 5
this finding was of moderate quality. However, another study found that there was no statistically 6
significant difference in the incidence of pain in women who received oral misoprostol compared with 7
women who received the same dose of sublingual misoprostol (400 micrograms, 6 doses). The 8
Satisfaction 10
One study found that the reported incidence of satisfaction was higher in women who received 11
sublingual misoprostol compared with women who received the same dose of oral misoprostol 12
(regimen for both study groups: 200mg of mifepristone plus 600 micrograms of misoprostol, with 3 13
supplemental doses of 400 micrograms of misoprostol after 12 hours). This finding was statistically 14
significant. The evidence for this finding was of moderate quality. 15
Management of missed miscarriage: comparison of sublingual and vaginal 16
misoprostol 17
One meta-analysis of two studies did not find a statistically significant difference in the success of 19
medical treatment for women who received sublingual misoprostol compared with women who 20
received the same dose of vaginal misoprostol (400 micrograms, 5 doses in one study and maximum 21
not reported in the other). A second study did not find a statistically significant difference in the 22
success of medical treatment for women who received sublingual misoprostol compared with women 23
who received the same dose of vaginal misoprostol (600 micrograms, 3 doses). The quality of 24
evidence for this finding was very low in the meta-analysis and moderate in the other study. 25
One meta-analysis of two studies did not find a statistically significant difference in the need for further 27
intervention for women who received sublingual misoprostol compared with women who received the 28
same dose of vaginal misoprostol (400 micrograms, 5 doses in one study and maximum not reported 29
in the other). A second study did not find a statistically significant difference in the need for further 30
intervention for women who received sublingual misoprostol compared with women who received the 31
same dose of vaginal misoprostol (600 micrograms, 3 doses). The quality of evidence for this finding 32
was very low in the meta-analysis and low in the other study. 33
Adverse effects 34
received sublingual misoprostol compared with women who received the same dose of vaginal 36
misoprostol (400 micrograms, 5 doses). A second study did not find a statistically significant 37
difference in the incidence of nausea for women who received sublingual misoprostol compared with 38
women who received the same dose of vaginal misoprostol (600 micrograms, 3 doses). The quality of 39
evidence for this finding was low in one study and moderate in the other. 40
misoprostol (400 micrograms, maximum number of doses not reported). A second study did not find a 43
statistically significant difference in the incidence of vomiting for women who received sublingual 44
misoprostol compared with women who received the same dose of vaginal misoprostol (600 45
micrograms, 3 doses). The quality of evidence for this finding was low in one study and moderate in 46
the other. 47
One study found that the incidence of diarrhoea was higher in women who received sublingual 48
micrograms, maximum number of doses not reported). A second study found that the incidence of 50
diarrhoea was higher in women who received sublingual misoprostol compared with women who 51
received the same dose of vaginal misoprostol (600 micrograms, 3 doses). These findings were 52
152
statistically significant. The quality of evidence for this finding was moderate in one study and high in 1
the other. 2
One study found that the incidence of fever was higher in women who received sublingual misoprostol 3
compared with women who received the same dose of vaginal misoprostol (400 micrograms, 4
maximum number of doses not reported). This finding was statistically significant. The evidence for 5
this finding was of moderate quality. However, another study did not find a statistically significant 6
difference in this outcome between women who received sublingual misoprostol and women who 7
received the same dose of vaginal misoprostol (600 micrograms, 3 doses). The evidence for this 8
finding was of moderate quality. 9
One study did not find a statistically significant difference in the incidence of chills or shivering for 10
women who received sublingual misoprostol compared with women who received the same dose of 11
vaginal misoprostol (400 micrograms, 5 doses). A second study did not find a statistically significant 12
difference in the incidence of chills or shivering for women who received sublingual misoprostol 13
compared with women who received the same dose of vaginal misoprostol (600 micrograms, 3 14
doses). The quality of evidence for this finding was low in one study and moderate in the other. 15
misoprostol (600 micrograms, 3 doses). The evidence for this outcome was of high quality. 19
Pain 20
One study found that the incidence of cramps was higher in women who received sublingual 21
micrograms, maximum number of doses not reported). This finding was statistically significant. The 23
One study found that the incidence of severe pain was higher in women who received sublingual 25
micrograms, maximum number of doses not reported). This finding was statistically significant. The 27
One study reported 100% incidence of lower abdominal pain in women who received sublingual 29
misoprostol and women who received the same dose of vaginal misoprostol (600 micrograms, 3 30
doses). The evidence for this outcome was of high quality. 31
Satisfaction 32
One meta-analysis of two studies did not find a statistically significant difference in the reported 33
incidence of satisfaction for women who received sublingual misoprostol compared with women who 34
received the same dose of vaginal misoprostol (400 micrograms, 5 doses in one study and maximum 35
not reported in the other). The evidence for this outcome was of very low quality. 36
Management of missed miscarriage: comparison of oral and vaginal misoprostol 37
One study found that the success of medical treatment was higher in women who received vaginal 39
misoprostol (600 micrograms, 2 doses) compared with women who received oral misoprostol (400 40
micrograms, 3 doses). A second study found that the success of medical treatment was higher in 41
women who received oral misoprostol (400 micrograms, repeat after 24 hours). These findings were 43
statistically significant. The quality of evidence for this finding was moderate in one study and very low 44
in the other. One further study did not find a statistically significant difference in this outcome between 45
women who received the same dose of vaginal and oral misoprostol (800 micrograms, 1 dose). The 46
One study found that the need for a further intervention was higher in women who received oral 49
misoprostol (400 micrograms, 3 doses) compared with women who received vaginal misoprostol (600 50
micrograms, 2 doses). This finding was statistically significant. The evidence for this finding was of 51
moderate quality. One further study did not find a statistically significant difference in this outcome 52
153
between women who received the same dose of vaginal and oral misoprostol (800 micrograms, 1 1
dose). The evidence for this finding was of low quality. 2
Admission to a medical facility 3
One study did not find a statistically significant difference in the incidence of admission to a medical 4
facility for women who received oral misoprostol compared with women who received the same dose 5
of vaginal misoprostol (800 micrograms, 1 dose). The evidence for this outcome was of very low 6
quality. 7
Adverse effects 8
received oral misoprostol (400 micrograms, 3 doses) compared with women who received vaginal 10
difference in the incidence of nausea for women who received oral misoprostol (400 micrograms, 1 12
dose) compared with women who received vaginal misoprostol (800 micrograms, 1 dose). The quality 13
of evidence for this outcome was low in one study and very low in the other. 14
One study found that the incidence of vomiting was higher in women who received vaginal 15
misoprostol (800 micrograms, 1 dose) compared with women who received oral misoprostol (800 16
micrograms, 1 dose). This finding was statistically significant. The evidence for this finding was of high 17
quality. However, another study found that there was no statistically significant difference in this 18
outcome between women who received vaginal misoprostol (600 micrograms, 2 doses) and oral 19
misoprostol (400 micrograms, 3 doses). A further study found that there was no statistically significant 20
difference this outcome between women who received vaginal misoprostol (800 micrograms, repeat 21
after 24 hours) and women who received oral misoprostol (400 micrograms, repeat after 24 hours). 22
The quality of evidence for this finding was low in one study and very low in the other. 23
difference in the incidence of diarrhoea for women who received oral misoprostol (400 micrograms, 1 27
dose) compared with women who received vaginal misoprostol (800 micrograms, 1 dose). Another 28
study did not find a statistically significant difference in the incidence of diarrhoea for women who 29
received oral misoprostol compared with women who received the same dose of vaginal misoprostol 30
(800 micrograms, 1 dose). The quality of evidence for this outcome was low in one study, very low in 31
one study, and moderate in one study. 32
One study did not find a statistically significant difference in the incidence of hyperpyrexia for women 33
who received oral misoprostol (400 micrograms, 3 doses) compared with women who received 34
vaginal misoprostol (600 micrograms, 2 doses). The evidence for this outcome was of low quality. 35
One study did not find a statistically significant difference in the incidence of fever or chills for women 36
who received oral misoprostol compared with women who received the same dose of vaginal 37
misoprostol (800 micrograms, 1 dose). The evidence for this outcome was of moderate quality. 38
(800 micrograms, 1 dose). The evidence for this finding was of moderate quality. One further study 42
reported duration of bleeding in a manner that did not permit a comparison between the two arms. 43
The evidence for this finding was of very low quality. 44
Pain 45
One study did not find a statistically significant difference in the incidence of pain for women who 46
difference in the incidence of pain for women who received oral misoprostol compared with women 49
who received the same dose of vaginal misoprostol (800 micrograms, 1 dose). The quality of 50
evidence for this outcome was low in one study and high in the other. 51
154
One study did not find a statistically significant difference in the severity of pain for women who 1
received oral misoprostol (400 micrograms, 1 dose) compared with women who received vaginal 2
misoprostol (800 micrograms, 1 dose). The evidence for this outcome was of very low quality. 3
Satisfaction 4
One study did not find a statistically significant difference in the reported incidence of satisfaction for 5
women who received oral misoprostol compared with women who received the same dose of vaginal 6
misoprostol (800 micrograms, 1 dose). The evidence for this outcome was of high quality. 7
Management of incomplete miscarriage: comparison of vaginal misoprostol and 8
placebo 9
women who received a placebo. The evidence for this outcome was of low quality. 13
women who received vaginal misoprostol (400 micrograms, 1 dose) compared with women who 16
received a placebo. A second study did not find a statistically significant difference in the need for 17
further intervention for women who received vaginal misoprostol (600 micrograms, repeat after 24 18
hours) compared with women who received a placebo. The evidence for this outcome was of low 19
quality. 20
Management of incomplete miscarriage: comparison of oral misoprostol dosages 21
One meta-analysis of two studies did not find a statistically significant difference in the success of 23
medical treatment for women who received two 600 microgram doses of oral misoprostol compared 24
with women who received a single dose of 600 microgram oral misoprostol. The evidence for this 25
intervention for women who received two 600 microgram doses of oral misoprostol compared with 29
women who received a single dose of 600 microgram oral misoprostol. The evidence for this outcome 30
Adverse effects 32
One meta-analysis of two studies did not find a statistically significant difference in the incidence of 33
nausea for women who received two 600 microgram doses of oral misoprostol compared with women 34
who received a single dose of 600 microgram oral misoprostol. The evidence for this outcome was of 35
vomiting for women who received two 600 microgram doses of oral misoprostol compared with 38
One study found that the incidence of diarrhoea was higher in women who received two 600 41
microgram doses of oral misoprostol compared with women who received a single 600 microgram 42
dose of oral misoprostol. This difference was statistically significant. The evidence for this finding was 43
of moderate quality. 44
fever or chills for women who received two 600 microgram doses of oral misoprostol compared with 46
155
Two studies did not find a statistically significant difference in the duration of heavy bleeding for 2
women who received two 600 microgram doses of oral misoprostol compared with women who 3
received a single dose of 600 microgram oral misoprostol. The quality of evidence for this outcome 4
was moderate in one study and low in the other. 5
Two studies did not find a statistically significant difference in the duration of normal bleeding for 6
women who received two 600 microgram doses of oral misoprostol compared with women who 7
received a single dose of 600 microgram oral misoprostol. The quality of evidence for this outcome 8
was moderate in one study and low in the other. 9
Two studies did not find a statistically significant difference in the duration of light bleeding for women 10
who received two 600 microgram doses of oral misoprostol compared with women who received a 11
single dose of 600 microgram oral misoprostol. The quality of evidence for this outcome was 12
moderate in one study and low in the other. 13
Pain 14
pain for women who received two 600 microgram doses of oral misoprostol compared with women 16
who received a single dose of 600 microgram oral misoprostol. The evidence for this outcome was of 17
Two studies did not find a statistically significant difference in the severity of pain for women who 19
received two 600 microgram doses of oral misoprostol compared with women who received a single 20
dose of 600 microgram oral misoprostol. The evidence for this outcome was of low quality. 21
Satisfaction 22
One meta-analysis of two studies did not find a statistically significant difference in the reported 23
incidence of satisfaction for women who received two 600 microgram doses of oral misoprostol 24
compared with women who received a single dose of 600 microgram oral misoprostol. The evidence 25
for this outcome was of moderate quality. 26
Management of incomplete miscarriage: comparison of oral and vaginal 27
misoprostol 28
misoprostol (800 micrograms, 2 doses). The evidence for this outcome was of very low quality. 32
misoprostol (800 micrograms, 2 doses). The evidence for this outcome was of very low quality. 36
Adverse effects 37
(800 micrograms, 2 doses). The evidence for this outcome was of low quality. 40
One study found that the incidence of diarrhoea was higher in women who received oral misoprostol 44
compared with women who received the same dose of vaginal misoprostol (800 micrograms, 2 45
doses). This finding was statistically significant. The evidence for this finding was of moderate quality. 46
156
(800 micrograms, 2 doses). The evidence for this outcome was of moderate quality. 4
Pain 5
One study found that the duration of pelvic pain was longer in women who received vaginal 6
misoprostol compared with women who received the same dose of oral misoprostol (800 micrograms, 7
2 doses). This finding was statistically significant. The evidence for this finding was of moderate 8
quality. 9
Management of incomplete miscarriage: comparison of oral and sublingual 10
misoprostol 11
sublingual misoprostol (600 micrograms). The evidence for this outcome was of low quality. 15
sublingual misoprostol (600 micrograms). The evidence for this outcome was of low quality. 19
Adverse effects 20
misoprostol (600 micrograms). The evidence for this outcome was of moderate quality. 23
One study reported no incidences of vomiting in women who received oral misoprostol and women 24
who received the same dose of sublingual misoprostol (600 micrograms). The evidence for this 25
One study did not find a statistically significant difference in the incidence of fever/chills for women 30
who received oral misoprostol compared with women who received the same dose of sublingual 31
Incidence of heavy bleeding 33
One study reported no incidences of heavy bleeding in women who received oral misoprostol and 34
women who received the same dose of sublingual misoprostol (600 micrograms). The evidence for 35
this outcome was of moderate quality. 36
Pain 37
One study did not find a statistically significant difference in the incidence of pain/cramps for women 38
who received oral misoprostol compared with women who received the same dose of sublingual 39
One study did not find a statistically significant difference in the severity of pain for women who 41
misoprostol (600 micrograms). The evidence for this outcome was of high quality. 43
Satisfaction 44
One study did not find a statistically significant difference in the incidence of reported satisfaction for 45
sublingual misoprostol (600 micrograms). The evidence for this outcome was of moderate quality. 47
157
The primary outcome measures for this question were the success rate of the treatment in terms of 3
live birth and the need for further interventions. The secondary outcomes were the various side 4
effects associated with treatment. Of these, the group felt that pain was less informative than the 5
others, as all women would be likely to experience a degree of pain as a natural consequence of the 6
miscarriage. It was also evident from the placebo trials that gastro-intestinal symptoms may 7
accompany miscarriage. 8
When considering the appropriate dose and mode of administration, the group considered the main 10
priority to be the efficacy of the treatment. The evidence compared a number of different doses and 11
regimens. For women with missed miscarriage, a single dose of 800 micrograms (oral or vaginal) of 12
misoprostol was the most effective overall. For women with an incomplete miscarriage, the evidence 13
suggested that a single dose of 600 micrograms of misoprostol was effective, but the group 14
recognised that units might prefer to use 800 micrograms for alignment of protocols. 15
For women with missed miscarriage, the group noted that, used at the same dose, both vaginal and 16
oral routes of administration had similar effectiveness, and that both were more effective than sub- 17
lingual administration. The majority of side effects did not show a difference by route. There was 18
contradictory evidence regarding diarrhoea: one study showed more diarrhoea associated with a 19
vaginal route of administration than with an oral route, but two further studies showed no statistically 20
significant difference which matched the GDGs clinical experience. The evidence suggested that 21
there were more side effects from sub-lingual administration than with vaginal administration. In 22
addition, the group recognised from their own clinical experience that there can often be difficulties 23
with sub-lingual administration as women are expected to hold up to 4 relatively large tablets under 24
their tongue for a long period of time. 25
For women with an incomplete miscarriage, one study found there was a significantly lower incidence 26
of diarrhoea with vaginal administration as compared with oral administration. A second study showed 27
no significant differences between oral and sub-lingual administration for all outcomes. 28
The group felt that an additional benefit of vaginal administration is that if women vomit after receiving 29
medication orally, this can interfere with the absorption of the drug and it can be difficult to determine 30
if the dose should be repeated. 31
Overall, the GDG agreed that vaginal misoprostol is the preferred treatment. However, they 32
recognised that it is important to take into account womens preferences and thus agreed that the oral 33
route would be appropriate if vaginal administration was not acceptable to the woman. 34
The group recognised that there was very little evidence which compared the efficacy of misoprostol 36
in combination with mifepristone compared with misoprostol alone for the treatment of miscarriage. 37
However, what evidence there was suggested that there was no difference in effectiveness. Given the 38
large cost of mifepristone compared with misoprostol
, the group agreed that mifepristone should 39

not be used routinely in the management of miscarriage. It was recognised that mifepristone is 40
currently used in current UK practice for this indication and that it would be therefore helpful to 41
conduct further research. They agreed that a trial was needed to determine definitively whether the 42
addition of mifepristone improves the success rate of the medical management, the results of which 43
could then be used to evaluate whether it is a cost-effective treatment. 44
The evidence for this question was drawn from RCTs, however the quality was mixed and ranged 46
from high to very low. Generally, the group felt that the evidence was of sufficient quality to make 47
recommendations, although they recognised that it would have been helpful to have had further 48
evidence related to the efficacy of mifepristone and misoprostol in combination. 49

A pack of three 200 mg mifepristone tablets costs approximately 50, whilst a pack of 60 200 microgram tablets costs
approximately 10
158
The group stressed the importance of providing women with information about the treatment, what to 2
expect as the miscarriage progresses, the potential side effects of treatment, and also the next steps 3
if the treatment is ineffective. They felt that informing women about the likely length and extent of 4
bleeding was a particular consideration, as this could be particularly distressing or worrying. 5
The GDG discussed the fact that the majority of women receiving medical treatment would now be 7
receiving it as second line treatment, as a result of recommendations about first line expectant 8
management made in section 6.4 above. All of the studies evaluated medical management as first 9
line treatment. However, the group noted that the women participating in the studies were likely to 10
have presented at a variety of different points in the course of their miscarriage, depending on their 11
symptoms and ability to access health care. For example, women with no symptoms might have 12
sought care, and been diagnosed with a missed miscarriage, weeks after the miscarriage had actually 13
occurred, and therefore would be reasonably comparable to women presenting earlier who had then 14
been expectantly managed for a period of time. Despite this, the group did accept that the treatment 15
success rates reported in the trials might be higher than would be expected in women in whom 16
expectant management had already failed, and that this would be a consideration for women 17
choosing their preferred second line treatment. However, they also felt that the relative efficacy of 18
different routes of administration would be unlikely to vary according to whether the treatment was 19
first or second line, and therefore, that the results of the comparisons reported in the trials could 20
safely be extrapolated to women actively choosing medical management after a period of expectant 21
management. 22
The group recognised that there are a number of side-effects associated with miscarriage itself 23
andwith the drugs used in its management, and that clinicians should treat these side effects at the 24
same time as providing miscarriage management. 25
The group agreed that, for women with a missed miscarriage, if the treatment is ineffective (i.e. 26
bleeding has not started) after 24 hours, there should be clinical review. The group also recognised 27
the importance of having a follow-up process in place to ensure that there is no molar or ectopic 28
pregnancy. The group agreed that all women should be advised to check that a urine pregnancy test 29
is negative 3 weeks after receiving medical treatment to determine if it has been effective. 30
Whilst misoprostol is commonly used to treat miscarriage, it is not currently licensed for use for this 31
indication and so womens consent should be obtained before it is used. 32
Recommendations 33
58 Offer vaginal misoprostol for the medical treatment of missed or incomplete
miscarriage. Oral administration is an acceptable alternative if this is the womans
preference.

59 For women with a missed miscarriage, use a single dose of 800 micrograms of
misoprostol.
****

60 If bleeding has not started 24 hours after treatment, there should be a clinical review
to determine individualised care.
61 For women with an incomplete miscarriage, use a single dose of 600 micrograms of
misoprostol. (800 micrograms can be used as an alternative to allow alignment of
treatment protocols for missed and incomplete miscarriage.)

**** Although this use is common in UK clinical practice, at the time of publication (June 2012), misoprostol did not have UK
159
62 Do not offer mifepristone as a treatment for missed or incomplete miscarriage.
63 Offer all women receiving medical management of miscarriage pain relief and anti-
emetics as required.
64 Inform women about what to expect throughout the process, including the length
and extent of bleeding and the potential side effects of treatment including pain,
diarrhoea and vomiting.
65 Advise women to take a urine pregnancy test 3 weeks after medical management
unless they experience worsening symptoms, in which case advise them to return
to the healthcare professional responsible for providing their medical management.
66 For women with a positive pregnancy test, a follow-up process should be in place
to ensure that there is no molar or ectopic pregnancy.
1
2
RR 7 Is the combination of mifepristone and misoprostol more effective than misoprostol
alone in the medical management of miscarriage?
7.6 Setting for surgical management of miscarriage 3
Review question 4
What is the effectiveness of surgical management of miscarriage in an outpatient (office) setting 5
compared with any other setting for improving womens clinical and psychological outcomes? 6
Introduction 7
Historically women diagnosed with a miscarriage who opted for surgical management would undergo 8
the procedure in a traditional theatre setting under a general anaesthetic. Recent advances in surgical 9
techniques and equipment have seen the introduction of surgical procedures being performed without 10
a general anaesthetic (but with some other form of anaesthesia/ analgesia/ sedation) in an outpatient 11
setting. This review sought to identify which settings are associated with better outcomes. 12
Four studies were included in this review. Three studies were conducted in the USA (Blumenthal et 14
al., 1994; Dalton et al., 2006; Edwards et al., 2007) and one in South Africa (De Jonge et al., 1994). 15
One retrospective observational study (Edwards et al., 2007) assessed and compared the efficiency, 16
post procedure quality of life, and acceptability of manual vacuum aspiration (MVA) performed in an 17
outpatient setting with electric vacuum aspiration (EVA) performed in a hospital in-patient setting in 18
women experiencing a first-trimester miscarriage. EVA was performed under either general 19
anaesthesia, monitored anaesthesia care (MAC), or spinal anaesthesia. Analgesia for MVA was 20
provided with a paracervical block of lidocaine. 21
One randomised clinical trial (De Jonge et al., 1994) assessed evacuation under systemic analgesia 22
(fentanyl and midazolam) in a treatment room compared with evacuation under general anaesthesia 23
in an operating theatre. 24
One prospective observational study (Dalton et al., 2006) examined womens satisfaction with an 25
office-based surgical procedure for early pregnancy failure and compared the resource use and cost 26
between office and operating room management. Anaesthesia for the MVA performed in an office 27
setting consisted of oral lorazepam, ibuprofen, and/or propoxyphene napsylate, with paracervical 28
block. EVA was performed under anaesthesia that included intravenous sedation, regional 29
anaesthesia, or general anaesthesia. 30
160
One quasi experimental study (Blumenthal et al., 1994) also examined the cost effectiveness of 1
performing manual vacuum aspiration curettage (MVAC) either in an emergency room or in a labour 2
room as an alternative to the traditional suction curettage (SC) in the operating room. For the SC 3
procedure, sedation was achieved with a combination of short acting benzodiazepines and narcotics 4
and MVAC was performed under systemic analgesia (fentanyl and midazolam). None of the women 5
had general anaesthesia in this study. 6
Evidence profile 7
The evidence from all 4 studies is presented below in one profile. 8
Table 7.18 GRADE summary of findings for comparison of operating room with office setting for surgical 9
management of miscarriage 10
Number of
studies
Operating
room/theatre
Outpatient/office
setting or
similar
Relative
(95% CI)
Absolute
(95% CI)
Emergency hospital visit on the same day of treatment
1 study
(Edwards et
al., 2007)
4/88
(4.6%)
3/67
(4.5%)
RR 1.01
(0.26 to 3.95)*
0 per 1000
(from 33 fewer
to 132 more)*
Very low
Tissue passed (reported within 48 hours after treatment)
1 study
(Edwards et
al., 2007)
14/79
(17.7%)
16/59
(27.1%)
RR 0.65
(0.35 to 1.22)*
95 fewer per
1000
(from 176 fewer
to 60 more)*
p = 0.19
Very low
Pain severity score (reported within 48 hours after treatment)
1 study
(Edwards et
al., 2007)
Mean 2.8
(SD 2.4)
(n = 79)
Mean 3.7
(SD 2.3)
(n = 62)
not calculable
(NC)
p = 0.03 Very low
Success rate (within 30 days after treatment)
1 study
(Edwards et
al., 2007)
81/83
(97.6%)
59/62
(95.2)
RR 1.02
(0.95 to 1.12)*
19 more per
1000
(from 48 fewer
to 114 more)*
p = 0.43
Very low
Fever (>38C) following treatment
1 study
(Edwards et
al., 2007)
4/83
(4.8%)
1/63
(1.6%)
RR 3.03
(0.47 to 19.9)*
32 more per
1000
(from 8 fewer to
300 more)*
p = 0.29
Very low
Waiting time: from emergency room admission to procedure (hours)
1 study Mean 7.18 Mean 3.45 NC MD 3.73 higher Very low
161
(Blumenthal et
al., 1994)
(SD 4.9) (SD 2.0) (from 1 higher
to 6 higher)
p < 0.01
1 study
(De Jonge et
al., 1994)
Mean 12.63
(range 1.08
70.25)
Mean 7.25
(0.25 63)
NC Median 5.38
higher (CI NC)
p < 0.0003
Moderate
Blood transfusion (number of women)
1 study
(De Jonge et
al., 1994)
24/68
(32.2%)
13/73
(17.8%)
1.98 (1.11 to
3.57)*
175 more per
1000
(from 18 more
to 458 more)*
p < 0.03
Low
Maximum total satisfaction (defined as maximum score on both satisfaction-related items)
1 study
(Dalton et al.,
2006)
26/46
(56%)
51/110
(46%)
1.21
(0.86 to 1.66)*
97 more per
1000
(from 65 fewer
to 306 more)*
p = 0.15
Very low
Post procedure infection
1 study
(Dalton et al.,
2006)
1/50
(2%)
2/115
(2%)
1.15
(0.15 to 8.55)*
3 more per
1000
(15 fewer to 131
more)*
p = 0.99
Very low
Blood loss (millilitres)
1 study
(Dalton et al.,
2006)
Mean 311
(SD 344)
Mean 70
(SD 106)
NC 241 higher
(171 higher to
310 higher)
p < 0.001
Very low
Need for re-evacuation
1 study
(Dalton et al.,
2006)
1/50
(2%)
4/115
(3%)
0.57
(0.08 to 3.67)*
15 fewer per
1000
(from32 fewer to
93 more)*
p = 0.68
Very low
* NCC calculation 1
Evidence was identified from four studies that reported efficiency of manual vacuum aspiration (MVA) 3
performed as an outpatient compared with electric vacuum aspiration (EVA) performed in an 4
operating room in women experiencing a first-trimester miscarriage. 5
One study found lower pain severity within 48 hours of treatment in women following EVA in an 6
operating room compared to MVA in an office setting. This finding was statistically significant. The 7
162
Two studies study found longer waiting times from admission to procedure in women undergoing EVA 1
in an operating room compared with MVA in office setting. This finding was statistically significant. 2
The evidence for this outcome was of moderate and very low quality. 3
One study found a higher proportion of women receiving a blood transfusion following EVA in an 4
operating room compared with MVA in an office setting. This finding was statistically significant. The 5
One study found higher mean blood loss in women following EVA in an operating room compared 7
with MVA in an office setting. This finding was statistically significant. The evidence for this outcome 8
was of very low quality. 9
One study did not find a statistically significant difference in the proportion of women that presented to 10
an emergency department on the same day of treatment following EVA in an operating room 11
compared with MVA in an office setting. The evidence for this outcome was of low quality. 12
One study did not find a statistically significant difference in the proportion of women that reported 13
passing tissue within 48 hours of treatment following EVA in an operating room compared with MVA 14
in an office setting. The evidence for this outcome was of very low quality. 15
One study did not find a statistically significant difference in success rate 30 days after treatment in 16
women following EVA in an operating room compared with MVA in an office setting. The evidence for 17
this outcome was of very low quality. 18
One study did not find a statistically significant difference in the proportion of women that developed 19
fever after treatment following EVA in an operating room compared with MVA in an office setting. The 20
One study did not find a statistically significant difference in maximum total satisfaction score in 22
women following EVA in an operating room compared with MVA in an office setting. The evidence for 23
this outcome was of very low quality. 24
One study did not find a statistically significant difference in post procedure infection in women 25
following EVA in an operating room compared with MVA in an office setting. The evidence for this 26
outcome was of very low quality. 27
One study did not find a statistically significant difference of need for re-evacuation in women 28
following EVA in an operating room compared with MVA in an office setting. The evidence for this 29
outcome was of very low quality. 30
The GDG considered the success rate of the treatment to be the most important outcome for this 33
question. In addition, the group felt that need for an emergency hospital visit was also an important 34
measure as an indicator of the comparative safety of the procedure in the different settings. 35
The evidence showed that the only differences between care in the different settings were the median 37
waiting time, the number of women requiring a blood transfusion, and the mean blood loss (all of 38
which favoured an outpatient setting). The group noted that the rate of blood transfusions, in the study 39
which reported this outcome, was particularly high in both arms and that this evidence was unlikely to 40
be applicable to a UK setting. In addition, the group noted that the study which reported the mean 41
blood loss outcome had more than twice as many women in one arm than the other and that this was 42
likely to have affected the results. 43
It was recognised that there would be an initial cost for units to set up the facility for performing 45
manual vacuum aspiration as an outpatient procedure. However, once these costs have been met, it 46
is likely that an outpatient setting would be cost-effective, given the reduced time for conducting the 47
procedure. 48
163
The evidence was generally of very low quality and so the group did not feel able to make a strong 2
recommendation that all surgical management should routinely be conducted as an outpatient 3
procedure. However, the group did feel that the evidence about reduced waiting times justified a 4
recommendation that units should be able to offer surgical management as an outpatient procedure in 5
order to provide women with a choice. 6
The GDG felt that it was important that women were given appropriate information about the different 8
treatment options and what to expect during the procedure, in order that they could make an informed 9
choice about their treatment. In addition, they agreed that women should be provided with information 10
about what to expect during the recovery period. 11
The group recognised that some women will prefer to have the procedure conducted under general 13
rather than local anaesthetic (i.e. in a theatre setting). In addition, they recognised that at later 14
gestations, it may not always be clinically appropriate to offer the procedure without a general 15
anaesthetic. As a result, they did not feel it appropriate to recommend that all surgical procedures be 16
conducted as an outpatient procedure. 17
Recommendations 18
manual vacuum aspiration under local anaesthetic in an out-patient or clinic
setting
68 Provide written and verbal information to all women undergoing surgical
management about the treatment options available and what to expect during and
after the procedure.
19
20
21


8 Management of ectopic 1
pregnancy 2
8.1 Introduction 3
The early diagnosis of ectopic pregnancy has increased with the advent of ultrasound and serum hCG 4
level assessment. Treatment options include surgical, medical and expectant management. Surgery, 5
by means of salpingectomy or salpingotomy, is performed laparoscopically or by open surgery. 6
Methotrexate is the most commonly used drug for the medical treatment of ectopic pregnancies. It 7
can be administered systemically, locally or both by various routes and requires constant vigilance of 8
effect and evaluation by serial follow-up. 9
The broader range of treatments has allowed womens choice to be an integral part of the decision 10
making process. This chapter examines the evidence for the effectiveness and safety of surgical and 11
medical treatment of tubal ectopic pregnancy in view of primary treatment success, tubal preservation 12
and future fertility. 13
8.2 Surgical compared with medical management of 14
ectopic pregnancy 15
Review question 16
How effective is surgical management of tubal ectopic pregnancy compared with medical 17
management for improving woemns clinical and psychological outcomes? 18
Ten studies were included in this review (Colacurci et al., 1998; Dias Pereira et al., 1998; Fernandez 20
et al., 1995; Fernandez et al., 1998; Hajenius et al., 1997; Moeller et al., 2009; Nieuwkerk et al., 1998; 21
Saraj et al., 1998; Sowter et al., 2001b; Zilber et al., 1996). 22
All the included studies are randomised trials; three of the studies were conducted in the Netherlands 23
(Hajenius et al., 1997; Dias Pereira et al., 1998; Nieuwkerk et al., 1998), two in France (Fernandez et 24
al., 1995; Fernandez et al., 1998), one in Denmark (Moeller et al., 2009), one in the USA (Saraj et 25
al.,1998;), one in Israel (Zilber et al., 1996), one in New Zealand (Sowter et al., 2001b), and one in 26
Italy (Colacurci et al., 1998). 27
All included studies compared medical and surgical management of ectopic pregnancy, and reported 28
at least one priority outcome. Surgical management consisted of salpingotomy/salpingostomy in all of 29
the included trials except for one trial (Sowter et al., 2001b) where some women (n = 7/28) were 30
treated with salpingectomy for various reasons. Medical management in all included trials consisted of 31
systemic or local injection of methotrexate. Eight trials (Colacurci et al., 1998; Dias Pereira et al., 32
1998; Fernandez et al., 1998; Hajenius et al., 1997; Moeller et al., 2009; Nieuwkerk et al., 1998; Saraj 33
et al., 1998; Sowter et al., 2001b) compared systemic methotrexate to surgery. In three of these trials 34
(Nieuwkerk et al., 1998; Dias Pereira et al., 1998; Hajenius et al., 1997) women were treated with 35
multiple doses of methotrexate and in the remaining five trials, they were treated with a single dose of 36
intra-muscular methotrexate injection. Three trials (Fernandez et al., 1995; Fernandez et al., 1998; 37
Zilber et al., 1996) compared local injection of methotrexate to surgery for management of ectopic 38
pregnancy. Local injection in the three trials consisted of direct injection of methotrexate into the 39
pregnancy site under laparoscopic guidance. One trial (Fernandez et al., 1998) compared local or 40
systemic injection of methotrexate to surgery for management of ectopic pregnancy. 41
Management of ectopic pregnancy
165
The trials were conducted in economically developed countries, and their populations include women 1
with confirmed ectopic pregnancy. 2
Evidence profile 3
Outcomes are reported in three evidence profiles: 4
Surgery compared with systemic methotrexate 5
Surgery compared with local methotrexate 6
Surgery compared with systemic or local methotrexate 7
In the outcomes with high heterogeneity (I
2
>

60%), the technical team used a random effects model 8
(the remaining outcomes used fixed effects models) and explored the heterogeneity with sensitivity 9
analyses as follows: 10
Success rate (surgery compared with systemic methotrexate) 11
Overall heterogeneity (69%) was not reduced by excluding the trial with multi-dose methotrexate; 12
however the heterogeneity was reduced to some extent (56%) by excluding the trial that treated some 13
women (n = 6/28) with salpingectomy. 14
Hospital stay (surgery compared with local methotrexate) 15
Excluding the study with small sample size (n < 50) reduced the heterogeneity from 100% to 0%. The 16
change could be a result of different hospitals policies regarding the hospital stay following the 17
treatment. 18
Table 8.1 GRADE summary of findings for comparison of surgery with systemic methotrexate 19
Number of
studies
Surgery Systemic
methotrexate
Relative
(95% CI)
Absolute
(95% CI) and p-
value if
reported
Success rate
1 meta-
analysis of 5
studies
(Fernandez et
al., 1998;
Hajenius et al.,
1997; Moeller
et al., 2009;
Saraj et
al.,1998;
Sowter et al.,
2001b)
187/215
(87%)
157/198
(79.3%)
RR 1.08
(0.99 to 1.19)*
63 more per
1000
(from 8 fewer to
143 more)*
Moderate
Pregnancy rate
1 meta-
analysis of 2
studies
(Dias Pereira
et al., 1998;
Moeller et al.,
2009)
48/92
(52.2%)
50/86
(58.1%)
RR 0.92
(0.71 to 1.18)*
47 fewer per
1000
(from 192 fewer
to 87 more)*
Low
Recurrent ectopic pregnancy
166
1 meta-
analysis of 2
studies
(Dias Pereira
et al., 1998;
Moeller et al.,
2009)
6/92
(6.5%)
3/86
(3.5%)
RR 1.65
(0.48 to 5.7)*
23 more per
1000
(from 18 fewer
to 164 more)*
Moderate
Resolution time (days)
1 meta-
analysis of 3
studies
(Colacurci et
al., 1998;
Fernandez et
al., 1995;
Fernandez et
al., 1998)
n = 84 n = 57 not calculable
(NC)
MD 8.8 lower
(11.99 to 5.62
lower)*
p < 0.0001
Moderate
Hospital stay (hours)
1 study
(Fernandez et
al., 1998)
Mean 46
(SD 8.4)
n = 49
Mean 24
(SD 1.2)
n = 22
NC MD 22 longer
(19.6 longer to
24.4 longer)*
p < 0.0001
Moderate
1 meta-
analysis of 3
studies
(Fernandez et
al., 1998;
Hajenius et al.,
1997; Moeller
et al., 2009)
8/130
(6.2%)
29/109
(26.6%)
RR 0.26
(0.12 to 0.55)*
197 fewer per
1000
(120 fewer to
234 fewer)*
High
Tubal preservation
1 study
(Hajenius et
al., 1997)
45/49
(91.8%)
46/51
(90.2%)
RR 1.02
(0.9 to 1.15)*
18 more per
1000
(from 90 fewer
to 135 more)*
High
Homolateral tubal patency
1 study
(Hajenius et
al., 1997)
23/39
(59%)
23/42
(54.8%)
RR 1.08
(0.74 to 1.57)*
44 more per
1000
(from 142 fewer
to 312 more)*
Moderate
Pain score 2 days after confirmative laparoscopy
1 study
(Nieuwkerk et
al., 1998)
Mean 68
(SD 23)
n = 34
Mean 79
(SD 21)
n = 38
NC MD 11 lower
(21.22 to 0.78
lower)*
p = NS**
(as reported in
Moderate
167
paper)
Pain score 2 weeks after confirmative laparoscopy
1 study
(Nieuwkerk et
al., 1998)
Mean 38
(SD 26)
n = 35
Mean 51
(SD 33)
n = 37
NC MD 13 lower
(26.68 lower to
0.68 higher)*
(p = 0.06**)
Moderate
Pain score 16 weeks after confirmative laparoscopy
1 study
(Nieuwkerk et
al., 1998)
Mean 15
(SD 21)
n = 30
Mean 19
(SD 27)
n = 34
NC MD 4 lower
(15.78 lower to
7.78 higher)*
p = NS**
Moderate
Depression score 2 weeks after confirmative laparoscopy
1 study
(Nieuwkerk et
al., 1998)
Mean 44
(SD 11)
n = 35
Mean 49
(SD 12)
n = 37
NC MD 5 lower
(10.31 lower to
0.31 higher)*
p = NS
Moderate
Depression score 16 weeks after confirmative laparoscopy
1 study
(Nieuwkerk et
al., 1998)
Mean 33
(SD 12)
n = 30
Mean 38
(SD 11)
n = 34
NC MD 5 lower
(10.67 lower to
0.67 higher)*
p = NS
Moderate
Overall quality of life score 2 days after confirmative laparoscopy
1 study
(Nieuwkerk et
al., 1998)
Mean 52
(SD 28)
n = 34
Mean 67
(SD 20)
n = 38
NC MD 15 lower
(26.36 to 3.64
lower)*
p < 0.05
Moderate
Overall quality of life score 2 weeks after confirmative laparoscopy
1 study
(Nieuwkerk et
al., 1998)
Mean 44
(SD 11)
n = 35
Mean 49
(SD 12)
n = 37
NC MD 5 lower
(10.31 lower to
0.31 higher)*
p < 0.05***
Moderate
Overall quality of life score 16 weeks after confirmative laparoscopy
1 study
(Nieuwkerk et
al., 1998)
Mean 23
(SD 20)
n = 30
Mean 27
(SD 20)
n = 34
NC MD 4 lower
(5.82 lower to
13.82 higher)*
p = NS
Moderate
* NCC-WCH calculation 1
** Women treated with methotrexate had consistently more pain than women treated with surgery at each separate time point 2
following laparoscopy. Based on the NCC-WCH calculation the pain difference 2 days following laparoscopy was statistically 3
significant (p < 0.03). This was not reported in the paper, however a significant treatment effect was demonstrated between the 4
2 groups when examined across all time points (2 days, 2 weeks, 4 weeks and 16 weeks; multivariate analysis of variance, p = 5
0.02) with more pain reported by women in the medical group at all time points. 6
168
*** This was the P value reported in the paper. However, an NCC-WCH calculation gave a P value of 0.07. 1
Table 8.2 GRADE summary of findings for comparison of surgery with local methotrexate 2
Number of
studies
Surgery Local
methotrexate
Relative
(95% CI)
Absolute
(95% CI) and p-
value if
reported
Success rate
1 meta-
analysis of 2
studies
(Fernandez et
al., 1995;
Fernandez, et
al. 1998)
66/69
(95.7%)
46/49
(93.9%)
RR 1.02
(0.93 to 1.11)*
19 more per
1000
(from 66 fewer
to 103 more)*
High
Pregnancy rate
1 meta-
analysis of 2
studies
(Fernandez, et
al. 1998; Zilber
et al., 1996)
16/28
(57.1%)
19/26
(73.1%)
RR 0.77
(0.53 to 1.12)*
168 fewer per
1000
(from 343 fewer
to 88 more)*
Low
1 meta-
analysis of 2
studies
(Fernandez et
al., 1998;
Zilber et al.,
1996)
2/38
(5.3%)
0/36
(0%)
RR 2.84
(0.31 to 26.08)*
not calculable
(NC)
Moderate
Resolution time (days)
1 study
(Fernandez et
al., 1998)
Mean 13.6
(SD 6.1)
n = 49
Mean 28.6
(SD 18.6)
n = 51
NC MD 15 lower
(20.38 to 9.62
lower)*
p < 0.0001
Moderate
1 study
(Zilber et al.,
1996)
Mean 13.9
(SD not
reported[NR])
n = 24
Mean 13.7
(SD NR)
n = 24
NC MD 0.2 higher
(CI NC)
p = NS
Moderate
Hospital stay (hours)
1 meta-
analysis of 3
studies
(Fernandez et
al., 1995;
n = 69 n = 71 NC MD 22 higher
(19.3 to 24.7
higher)*
p < 0.0001
Moderate
169
Fernandez et
al., 1998;
Zilber et al.,
1996)
1 meta-
analysis of 2
studies
(Fernandez et
al., 1998;
Zilber et al.,
1996)
3/75
(4%)
8/75
(10.7%)
RR 0.38
(0.1 to 1.36)*
66 fewer per
1000
(from 96 fewer
to 38 more)*
Moderate
Table 8.3 GRADE summary of findings for comparison of surgery with systemic and local methotrexate 2
Number of
studies
Surgery Methotrexate
(systemic or
local)
Relative
(95% CI)
Absolute
(95% CI) and p-
value if
reported
Spontaneous ongoing or term pregnancy
1 study
(Fernandez et
al., 1998)
15/37
(40.5%)**
21/37
(56.8%)**
RR 0.71
(0.44 to 1.16)*
165 fewer per
1000
(from 318 fewer
to 91 more)*
Low
1 study
(Fernandez et
al., 1998)
5/49
(10.2%)
1/51
(2%)
RR 5.2
(0.63 to 42.96)*
82 more per
1000
(from 7 fewer to
823 more)*
Low
** Excludes those who did not desire a pregnancy. Lost to follow-up included 4
Surgery compared with systemic methotrexate 6
Success rate 7
One meta-analysis of five studies did not find a statistically significant difference in the success rate in 8
women who received surgical management compared with women who received medical 9
management. The evidence for this finding was of moderate quality. 10
Pregnancy rate 11
One meta-analysis of two studies did not find a statistically significant difference in pregnancy rate in 12
management with systemic methotrexate. The evidence for this finding was of low quality. 14
Recurrent ectopic pregnancy 15
management with systemic methotrexate. The evidence for this finding was of moderate quality. 18
170
Resolution time 1
One meta-analysis of three studies found that resolution time was shorter in women who received 2
surgical management compared with women who received medical management with systemic 3
methotrexate. This finding was statistically significant. The evidence for this finding was of moderate 4
quality. 5
Hospital stay 6
One study found that hospital stay was longer in women who received surgical management 7
compared with women who received medical management with systemic methotrexate. This finding 8
was statistically significant. The evidence for this finding was of moderate quality. 9
One meta-analysis of three studies found that the need for further intervention was lower in women 11
who received surgical management compared with women who received medical management with 12
systemic methotrexate. This finding was statistically significant. The evidence for this finding was of 13
high quality. 14
Tubal preservation 15
One study did not find a statistically significant difference in tubal preservation in women who received 16
salpingotomy compared with women who received medical management with systemic methotrexate. 17
The evidence for this finding was of high quality. 18
Homolateral tubal patency 19
One study did not find a statistically significant difference in homolateral tubal patency in women who 20
received surgical management compared with women who received medical management with 21
systemic methotrexate. The evidence for this finding was of moderate quality. 22
Pain score 2 days after confirmative laparoscopy 23
One study did not find a statistically significant difference in pain scores 2 days after confirmative 24
laparoscopy in women who received surgical management compared with women who received 25
medical management with systemic methotrexate. The evidence for this finding was of moderate 26
quality. 27
Pain score 2 weeks after confirmative laparoscopy 28
One study did not find a statistically significant difference in pain scores 2 weeks after confirmative 29
quality. 32
Pain score 16 weeks after confirmative laparoscopy 33
One study did not find a statistically significant difference in pain scores 16 weeks after confirmative 34
quality. 37
Depression score 2 weeks after confirmative laparoscopy 38
One study did not find a statistically significant difference in depression scores 2 weeks after 39
confirmative laparoscopy in women who received surgical management compared with women who 40
received medical management with systemic methotrexate. The evidence for this finding was of 41
Depression score 16 weeks after confirmative laparoscopy 43
One study did not find a statistically significant difference in depression scores 16 weeks after 44
confirmative laparoscopy in women who received surgical management compared with women who 45
received medical management with systemic methotrexate. The evidence for this finding was of 46
Overall quality of life score 2 days after confirmative laparoscopy 48
One study found that overall quality of life scores 2 days after confirmative laparoscopy were lower in 49
171
management with systemic methotrexate. This finding was statistically significant. The evidence for 1
this finding was of moderate quality. 2
Overall quality of life score 2 weeks after confirmative laparoscopy 3
One study found that overall quality of life scores 2 weeks after confirmative laparoscopy was lower in 4
management with systemic methotrexate. This finding was statistically significant. The evidence for 6
this finding was of moderate quality. 7
Overall quality of life score 16 weeks after confirmative laparoscopy 8
One study did not find a statistically significant difference in overall quality of life scores 16 weeks 9
after confirmative laparoscopy in women who received surgical management compared with women 10
who received medical management with systemic methotrexate. The evidence for this finding was of 11
Surgery compared with local methotrexate 13
Success rate 14
One meta-analysis of two studies did not find a statistically significant difference in the success rate in 15
management with local methotrexate. The evidence for this finding was of moderate quality. 17
Pregnancy rate 18
management with local methotrexate. The evidence for this finding was of low quality. 21
management with local methotrexate. The evidence for this finding was of moderate quality. 25
Resolution time 26
One study found that resolution time was shorter in women who received surgical management 27
compared with women who received medical management with local methotrexate. This finding was 28
statistically significant. The evidence for this finding was of moderate quality. 29
One study did not find a statistically significant difference in resolution time in women who received 30
surgical management compared with women who received medical management with local 31
methotrexate. The evidence for this finding was of moderate quality. 32
Hospital stay 33
One meta-analysis of three studies found that hospital stay was longer in women who received 34
surgical management compared with women who received medical management with local 35
methotrexate. This finding was statistically significant. The evidence for this finding was of moderate 36
quality. 37
intervention in women who received surgical management compared with women who received 40
medical management with local methotrexate. The evidence for this finding was of moderate quality. 41
Surgery compared with methotrexate (systemic & local) 42
Spontaneous ongoing or term pregnancy 43
One study did not find a statistically significant difference in spontaneous ongoing or term pregnancy 44
in women who received surgical management compared with women who received medical 45
management with methotrexate (systemic or local). The evidence for this finding was of low quality. 46
172
One study did not find a statistically significant difference in recurrent EP in women who received 2
surgical management compared with women who received medical management with methotrexate 3
(systemic or local). The evidence for this finding was of low quality. 4
Health economics 5
A de novo economic model was developed for this guideline to assess the cost-effectiveness of 6
different treatment strategies for ectopic pregnancy. The model is described in more detail in section 7
10.4. 8
The model compared the cost-effectiveness of the following three treatment strategies in a population 9
of women diagnosed with an ectopic pregnancy but not requiring urgent surgical intervention: 10
Laparoscopic salpingectomy 11
Laparoscopic salpingotomy 12
Methotrexate 13
In the base-case analysis the evaluation took the form of a cost minimisation analysis with the 14
assumption that all women recover and that any differences in morbidity only exist in the very short 15
term. Methotrexate was the cheapest option at 1,432 followed by laparoscopic salpingectomy at 16
1,608. Laparoscopic salpingotomy, because of its relatively high re-intervention rate, was the most 17
expensive strategy at 2,205. A probabilistic sensitivity analysis of one million Monte Carlo 18
simulations found that methotrexate was cheapest in 99.65% of the simulations with a 0.35% 19
probability that laparoscopic salpingectomy was cheapest. 20
A number of one-way sensitivity analyses suggested that the ordinal ranking of strategies in terms of 21
their cost was not affected by large changes in parameter values. A further sensitivity analysis relaxed 22
the assumption about equivalence in treatment outcomes and estimated an incremental QALY gain 23
for the surgical alternatives when compared to methotrexate. This found that laparoscopic 24
salpingotomy was dominated by laparoscopic salpingectomy (a higher cost than laparoscopic 25
salpingectomy with no QALY gain). The incremental cost-effectiveness ratio of laparoscopic 26
salpingectomy against methotrexate was calculated to be almost 84,000 per QALY which would not 27
normally be considered cost-effective using advisory willingness to pay thresholds suggested by NICE 28
(NICE, 2009). 29
The GDG identified future reproductive outcomes as very important from the womans perspective. 32
The group also felt that outcomes such as the chances of treatment success, the likelihood of needing 33
another intervention, and the length of time required for follow-up would be important considerations 34
for women, as well as having cost implications. 35
Womens experiences of care and psychological outcomes were identified as being important. 36
However, evidence was only found for psychological outcomes and was not available for womens 37
experience of care. The effects demonstrated were short term, with no significant differences shown 38
after 2 weeks in most cases, and 4 months in the remainder. 39
The evidence in this review showed that women undergoing medical treatment required a longer 41
period of recovery and follow-up than women who had laparoscopic surgery. In some women, the 42
time to resolution was up to 6-7 weeks. Medical treatment was also associated with a higher need for 43
further intervention, which could include surgery or an additional dose of methotrexate. The GDG also 44
recognised that some women may have contraindications to surgery, or have a personal desire to 45
avoid a surgical procedure; therefore medical treatment would be appropriate in such circumstances. 46
For women choosing to have methotrexate treatment, the GDG agreed that systemic administration 47
was preferable, as it is easier to administer, less invasive, and is in line with current practice. 48
173
The GDG noted that the studies considered in this review had specific inclusion criteria for women 1
with an ectopic pregnancy who were felt to be at low risk for tubal rupture. Thus they felt that for 2
women who fell outside of these criteria, methotrexate treatment was not an appropriate choice of 3
treatment. Evidence from Sowter et al. (2001b) showed that the chance of treatment failure was 4
increased in women with higher initial hCG levels. At an initial hCG below 1000 IU/l, women had a 5
12% chance of requiring further treatment, but at an hCG over 1500 IU/l, this rose to 70%. The 6
evidence from Sowter et al. (2001b) also showed at an hCG level below 5000 IU/L, women had 91% 7
chance of successful treatment with either first or second dose of methotrexate administration. The 8
GDG felt that methotrexate treatment should only be recommended for a woman with hCG of less 9
than 5000 IU/l, and with an ectopic pregnancy of smaller than 3.5 cm. The GDG were also aware of 10
studies reporting that the risk of rupture was increased in women whose ectopic pregnancy had a 11
visible fetal heartbeat, and therefore felt that these women should be offered surgery as a first line 12
treatment. 13
The GDG were aware of cases of women who had experienced a failure of medical treatment, and 14
then re-presented later with tubal rupture or other severe symptoms; therefore, it was felt that follow- 15
up should be mandatory after any medical treatment and that women needed access to 24 hour 16
emergency care. Using their clinical experience, the group agreed that women should have two 17
follow-up appointments in the first week following treatment, and then one appointment per week until 18
a negative pregnancy test is performed. This follow-up protocol would ensure that women in whom 19
medical treatment was not effective would be identified early, before their condition became serious. 20
The group decided that, if there were any anticipated difficulties in follow-up, women should be 21
advised to have surgery as a first line treatment. In particular the group felt that women choosing 22
methotrexate treatment with an hCG of 1500 to 5000 IU/l merited extra consideration, due to the 23
increased risk of further intervention being needed that was demonstrated in Sowter et al., 2001b. 24
The GDG felt that, for a woman in this group, it is vital that clinicians explore her ability to access 25
emergency services, for example where she lives, before proceeding with methotrexate treatment. 26
The GDG noted that, while surgery has the advantage of a shorter resolution time and a reduced 28
need for further intervention, medical management is associated with a shorter stay in hospital. 29
Therefore, the group decided that this area was appropriate for a health economics analysis. Using a 30
cost minimisation approach, the model determined that methotrexate was the preferred treatment 31
approach. Taking into account the health economics, but also the potential risks inherent in tubal 32
rupture, the GDG decided to recommend methotrexate as the first line treatment for women meeting 33
the criteria discussed above, namely those with an unruptured ectopic pregnancy which is smaller 34
than 3.5cm with no visible heartbeat, and who have an hCG level less than 5000 IU/l. 35
The evidence was derived from randomised controlled trials, and ranged in quality from high to low for 37
the outcomes considered. The GDG felt that it was disappointing that studies had not reported more 38
information about outcomes among women with different initial hCG levels and size of ectopic 39
pregnancy. Four studies reported only including ectopic pregnancies below a certain size (ranging 40
from 3.5 cm to 4 cm), and only two discussed the effect of hCG values on outcomes. 41
The GDG emphasised the importance of providing good quality information to women who are in a 43
clinically stable condition, including accurate information on local management options and 44
appropriate counselling regarding the proposed procedure. The GDG were aware that, for some 45
women, methotrexate might be a clinically appropriate treatment option, but would not be acceptable 46
to the woman personally. In these circumstances, the GDG felt that the womans choice should be 47
supported. For all women with an ectopic pregnancy, the GDG thought that it was important that 48
women were given sufficient information about what to expect during the course of their treatment and 49
recovery. In particular, they felt women should be made aware of the amount of pain and/or bleeding 50
that might be expected, so that women knew that it was an expected part of the treatment process. 51
It was recognised that ongoing psychological support should be offered from the diagnostic period, 52
through discussions regarding treatment options and the relevant outcomes and into the post 53
operative period including information regarding resuming usual activity and details of patient groups 54
174
to contact for support. The information needs to cover details regarding the physical and emotional 1
recovery. 2
The GDG noted that these studies looked specifically at women with unruptured ectopic pregnancies, 4
and were therefore women not in extreme pain. The GDG did not feel that women in significant pain 5
were suitable for medical management, and therefore should be offered surgery. 6
Recommendations 7
69 Give women relevant information in a variety of formats throughout their care. This
information should include:
information about post operative care (for women undergoing surgery)
information about resuming normal activity
reference to the emotional impact of the loss of a baby
reference to potential future complications and what to do next time the
woman becomes pregnant
advice about how long to wait before trying to conceive
what follow-up support is available from within the hospital and details of
patient support groups provided
how the woman can contact a health care professional for post-operative
advice if needed, and who this will be
how to find further information for example on returning to work.
Inform women who have had an ectopic pregnancy that they can self-refer to an
early pregnancy assessment service in future pregnancies if they have any early
concerns.
70 Inform women who have had an ectopic pregnancy that they can self-refer to an
early pregnancy assessment service in future pregnancies if they have any early
concerns.
what to expect throughout the course of their treatment and recovery,
especially the likely duration and severity of pain and/or of bleeding
where and when to get help in an emergency.
72 Do not offer methotrexate to women who have been diagnosed with an ectopic
pregnancy unless they can return for follow-up.
73 Offer systemic methotrexate
as a first-line treatment to women who have all of

the following:
no significant pain
an unruptured ectopic pregnancy smaller than 3.5 cm with no visible
heartbeat,
an hCG level less than 5000 IU/l.
Offer surgery where treatment with methotrexate is not acceptable to the woman.
74 Offer surgery as a first-line treatment to women with any of the following:
an ectopic pregnancy that is 3.5 cm or larger,
an ectopic pregnancy with a fetal heartbeat visible on ultrasound,
an ectopic pregnancy and an hCG level of 5000 IU/l or more.
75 Exercise additional consideration with women who have an hCG level greater than

Although this use is common in UK clinical practice, at the time of publication (June 2012), methotrexate did not have UK
175
1500 IU/l who choose medical treatment. Be aware that:
their chance for needing further intervention is increased
they may need to be urgently admitted if their condition deteriorates.
76 For women who have had methotrexate, take two serum hCG measurements in the
first week (day 4 and 7) following treatment and then one hCG measurement per
week until a negative result is obtained.
1
2
RR 8 In women with ectopic pregnancy, does the type of intervention (laparoscopy or
medical management) impact on patient experience, including psychological and
emotional outcomes?

Currently there is no evidence which explores the psychological impact on women
of the different methods used to treat ectopic pregnancy. However, it is well known
that the emotional impact of the condition can be significant, in some circumstances
leading to post traumatic stress disorder. Therefore the GDG believe that it is
important to assess how this impact can be reduced, to maximise womens
emotional recovery both in the short and long term. Research to explore how the
impact can be minimised is necessary to enable women and clinicians to decide the
optimum treatment method for the patient and how to provide appropriate support to
them during and after the process. This could also reduce cost for the NHS of
providing long term counselling for this population.
3
8.3 Laparotomy compared with laparoscopy for ectopic 4
pregnancy 5
Review question 6
What is the effectiveness of laparotomy compared with laparoscopic techniques for managing tubal 7
ectopic pregnancy? 8
Introduction 9
Many women with ectopic pregnancies are managed surgically by either laparoscopy and/or 10
laparotomy. There is however some variation in practice in the way women with an ectopic pregnancy 11
with similar clinical characteristics are managed. The reasons for this variability are multifactorial, but 12
include differences in surgical training, availability of equipment and surgeon preference. This review 13
aims to consider the evidence comparing laparoscopy and laparotomy to investigate which approach 14
is more effective. 15
Fourteen studies were included in this review (Baumann et al., 1991; Chatwani et al., 1992; El 17
Tabbakh & El Sayes, 2002; Federici et al., 1994; Lo et al., 1999; Lundorff et al., 1991; Lundorff et al., 18
1992; Lundorff, 1997; Mehra et al., 1998; Mol et al., 1997; Murphy et al., 1992; Rizzuto et al., 2008; 19
Vermesh et al., 1989; Vermesh & Presser, 1992). 20
Five of the included studies reported the initial outcomes and follow-up data of two randomised 21
controlled trials, conducted in Sweden (Lundorff et al., 1991; Lundorff et al., 1992; Lundorff, 1997) and 22
176
the USA (Vermesh et al., 1989; Vermesh & Presser, 1992). Nine of the included studies were 1
prospective comparative observational studies, conducted in the UK (Baumann et al., 1991; Rizzuto 2
et al., 2008), the USA (Chatwani et al., 1992; Murphy et al., 1992), Italy (Federici et al., 1994), the 3
Netherlands (Mol et al., 1997), Hong Kong (Lo et al., 1999), Kuwait (El Tabbakh & El Sayes, 2002), 4
and India (Mehra et al., 1998). 5
All studies compared laparotomy with laparoscopy for the management of tubal ectopic pregnancies, 6
and reported at least one outcome of interest. However, one observational study also included seven 7
women with non-tubal ectopic pregnancies (Baumann et al., 1991), and another had a specific study 8
population of women with a ruptured ectopic pregnancy and significant haemoperitoneum (Rizzuto et 9
al., 2008). The latter is reported separately in the evidence profile below. 10
Evidence profile 11
Table 8.4 GRADE summary of findings for comparison of laparotomy with laparoscopy for the management of 12
tubal ectopic pregnancy 13
Number of
studies
Number of women or mean (SD) Effect Quality
Laparotomy Laparoscopy Relative
(95% CI)
Absolute
(95% CI) and p-
value if
reported
Subsequent viable intrauterine pregnancy
1 meta-
analysis of 2
studies
(Lundorff et al.,
1992;
Vermesh &
Presser, 1992)
27/66
(40.9%)
26/61
(42.6%)
RR 0.96
(0.64 to 1.45)
17 fewer per
1000
(from 155 fewer
to 191 more)
Low
Subsequent intrauterine pregnancy
1 meta-
analysis of 2
studies
(Lundorff et al.,
1992;
Vermesh &
Presser, 1992)
35/66
(53%)
35/61
(57.4%)
RR 0.92
(0.68 to 1.26)
43 fewer per
1000
(from 186 fewer
to 152 more)
Low
1 study
(Chatwani et
al., 1992)
12/35
(34.3%)
9/33
(27.3%)
RR 1.26
(0.61 to 2.59)
70 more per
1000
(from 106 fewer
to 433 more)
Very low
1 study
(Mehra et al.,
1998)
11/25
(44%)
46/86
(53.5%)
RR 0.82
(0.51 to 1.33)
95 fewer per
1000
(from 264 fewer
to 179 more)
Very low
1 study
(Murphy et al.,
1992)
5/10
(50%)
7/8
(87.5%)
RR 0.57
(0.29 to 1.12)
375 fewer per
1000
(from 620 fewer
to 105 more)
Very low
177
1 meta-
analysis of 2
studies
(Lundorff et al.,
1992;
Vermesh &
Presser, 1992)
9/66
(13.6%)
5/61
(8.2%)
RR 1.66
(0.59 to 4.69)
54 more per
1000
(from 34 fewer
to 302 more)
Low
1 study
(Mehra et al.,
1998)
1/25
(4%)
4/86
(4.7%)
RR 0.86
(0.10 to 7.35)
7 fewer per
1000
(from 42 fewer
to 295 more)
Very low
1 study
(Murphy et al.,
1992)
2/10
(20%)
0/8
(0%)
RR 4.09
(0.22 to 74.78)
200 more per
1000
(from 155 fewer
to 510 more)
Very low
Length of hospital stay (days)
1 meta-
analysis of 2
studies
(Lundorff et al.,
1991;
Vermesh et al.,
1989)
Means 5.4 and
3.3
(SD 1.5 and 1.1)
n=87
Means 2.2 and
1.4
(SD 0.69 and
0.55)
n=78
not calculable
(NC)
MD 2.55 higher
(1.28 to 3.83
higher)
Moderate
1 study
(Baumann et
al., 1991)
Mean 5.2
(SD 1.4)
n=27
Mean 1.7
(SD 1.2)
n=65
NC MD 3.5 higher
(3.05 to 3.95
higher)
p < 0.001
Very low
1 study
(Chatwani et
al., 1992)
Mean 4.70
(SD not
reported (NR))
n=61
Mean 1.27
(SD NR)
n=56
NC MD 3.43 higher
(confidence
intervals NC)
p < 0.05
Very low
1 study
(El Tabbakh &
El Sayes,
2002)
Mean 5.25
(SD 3.16)
n=23
Mean 2.14 (SD
1.81)
n=184
NC MD 3.11 higher
(2.24 to 3.98
higher)
Very low
1 study
(Federici et al.,
1994)
Mean 7.3
(SD 0.9)
n=7
Mean 2.8
(SD 0.7)
n=23
NC MD 4.5 higher
(3.84 to 5.16
higher)
p<0.001
Very low
1 study
(Lo et al.,
1999)
Mean 5.3
(SD NR)
n=164
Mean 2.65
(SD NR)
n=371
NC MD 2.65 higher
(confidence
intervals NC)
p = 0.0001
Very low
1 study Mean 3.52 Mean 1.48 NC MD 2.04 higher Very low
178
(Mehra et al.,
1998)
(SD 0.51)*
n=25
(SD 0.59)*
n=86
(1.80 to 2.27
higher)
p < 0.05
1 study
(Mol et al.,
1997)
Mean 8.89
(SD 2.33)
n=140
Mean 2.93
(SD 1.08)
n=115
NC MD 5.96 higher
(5.49 to 6.43
higher)
Very low
1 study
(Murphy et al.,
1992)
Mean 26.42
(SD 0.71)*
n=37
Mean 1.08
(SD 0.79)*
n=26
NC MD 25.34
higher
(24.96 to 25.72
higher)
p < 0.005
Very low
Need for further surgery
1 meta-
analysis of 2
studies
(Lundorff et al.,
1991;
Vermesh et al.,
1989)
3/87
(3.4%)
8/78
(10.3%)
RR 0.34
(0.09 to 1.22)
68 fewer per
1000
(from 93 fewer
to 23 more)
Low
1 study
(Baumann et
al.,
1991)
0/27
(0%)
2/65
(3.1%)
RR 0.47
(0.02 to 9.51)
16 fewer per
1000
(from 30 fewer
to 262 more)
Very low
1 study
(Federici et al.,
1994)
0/7
(0%)
0/23
(0%)
NC NC Very low
1 study
(Lo et al.,
1999)
1/164
(0.61%)
3/371
(0.81%)
RR 0.75
(0.08 to 7.20)
2 fewer per
1000
(from 7 fewer to
50 more)
Very low
1 study
(Murphy et al.,
1992)
0/37
(0%)
2/26
(7.7%)
RR 0.14
(0.01 to 2.84)
66 fewer per
1000
(from 76 fewer
to 142 more)
Very low
Need for methotrexate
1 study
(Lundorff,
1997)
0/57
(0%)
2/48
(4.2%)
RR 0.17
(0.01 to 3.44)
35 fewer per
1000
(from 41 fewer
to 102 more)
Low
1 study
(Murphy et al.,
1992)
0/37
(0%)
1/26
(3.8%)
RR 0.24
(0.01 to 5.6)
29 fewer per
1000
(from 38 fewer
to 177 more)
Very low
Need for surgery, methotrexate or expectant management
179
1 study
(Mol et al.,
1997)
1/140
(0.71%)
18/115
(15.7%)
RR 0.05
(0.006 to 0.34)
149 fewer per
1000
(from 104 fewer
to 156 fewer)
Very low
Readmission to hospital
1 study
(Chatwani et
al., 1992)
1/61
(1.6%)
1/56
(1.8%)
RR 0.92
(0.06 to 14.33)
1 fewer per
1000
(from 17 fewer
to 238 more)
Very low
1 study
(Lo et al.,
1999)
2/164
(1.2%)
8/371
(2.2%)
RR 0.57
(0.12 to 2.63)
9 fewer per
1000
(from 19 fewer
to 35 more)
Very low
Abdominal pain
1 study
(Lundorff,
1997)
3/57
(5.3%)
1/48
(2.1%)
RR 2.53
(0.27 to 23.50)
32 more per
1000
(from 15 fewer
to 469 more)
Low
Thromboembolic disease
1 study
(Mol et al.,
1997)
1/140
(0.71%)
0/115
(0%)
RR 2.47
(0.1 to 60.02)
7 more per 1000
(from 26 fewer
to 39 more)
Very low
Respiratory morbidity
1 study
(Mol et al.,
1997)
2/140
(1.4%)
0/115
(0%)
RR 4.11
(0.2 to 84.83)
14 more per
1000
(from 20 fewer
to 51 more)
Very low
1 study
(Murphy et al.,
1992)
1/37
(2.7%)
0/26
(0%)
RR 2.13
(0.09 to 50.36)
27 more per
1000
(from 104 fewer
to 138 more)
Very low
1 study
(El Tabbakh &
El Sayes,
2002)
6/23
(26.1%)
13/184
(7.1%)
RR 3.69
(1.56 to 8.77)
190 more per
1000
(from 40 more
to 549 more)
Very low
1 study
(Mol et al.,
1997)
10/140
(7.1%)
1/115
(0.87%)
RR 8.21
(1.07 to 63.22)
63 more per
1000
(from 1 more to
541 more)
Very low
1 study
(Murphy et al.,
1992)
2/37
(5.4%)
1/26
(3.8%)
RR 1.41
(0.13 to 14.70)
16 more per
1000
(from 33 fewer
to 527 more)
Very low
180
Intraoperative blood loss (millilitres)
1 study
(Vermesh et
al., 1989)
Mean 195
(SD 131.45)
n=30
Mean 79
(SD 98.59)
n=30
NC MD 116 higher
(55.95 to 176.05
higher)
p < 0.001
Moderate
1 study
(Baumann et
al., 1991)
Mean 269.0
(SD 258.90)
n=27
Mean 206.1
(SD 235.0)
n=65
NC MD 63 higher
(47.53 lower to
173.33 higher)
NS (p-value NR)
Very low
1 study
(El Tabbakh &
El Sayes,
2002)
Mean 270.7 (SD
138.4)
n=23
Mean 79.6 (SD
96.7) n=184
NC MD 191.08
higher
(146.61 to
235.55 higher)
p < 0.0001
Very low
1 study
(Lo et al.,
1999)
Mean 110.4
(SD NR)
n=164
Mean 129.2
(SD NR)
n=371
NC MD 18.8 lower
(confidence
intervals NC)
NS (p-value NR)
Very low
1 study
(Mehra et al.,
1998)
Mean 150
(SD 44.9)
n=25
Mean 140
(SD 51.9)
n=86
NC MD 10 higher
(12.72 lower to
32.72 higher)
NS (p-value NR)
Very low
1 study
(Murphy et al.,
1992)
Mean 115
(SD 115)
n=36
Mean 62
(SD 61)
n=26
NC MD 53 higher
(3.45 to 102.55
higher)
p < 0.001
Very low

Length of hospital stay (days)**
1 study
(Rizzuto et al.,
2008)
All patients
discharged after
3-4 days
n=5
All patients
discharged after
1-2 days
n=32
NC NC Very low
Need for further surgery**
1 study
(Rizzuto et al.,
2008)
0/5
(0%)
0/32
(0%)
NC NC Very low
* Calculated by the NCC-WCH technical team from data reported in hours in the study 1
** The results of Rizzuto et al., 2008 are reported separately due to the specific nature of the study population (women with a 2
ruptured ectopic pregnancy and significant haemoperitoneum) 3
181
Subsequent viable intrauterine pregnancy 2
subsequent viable intrauterine pregnancy for women who received a laparotomy compared with 4
women who received a laparoscopy. The evidence for this outcome was of low quality. 5
Subsequent intrauterine pregnancy 6
One meta-analysis of two studies, and three other studies, did not find a statistically significant 7
difference in the incidence of subsequent intrauterine pregnancy for women who received a 8
laparotomy compared with women who received a laparoscopy. The evidence for this outcome was of 9
low and very low quality. 10
One meta-analysis of two studies, and two other studies, did not find a statistically significant 12
difference in the incidence of recurrent ectopic pregnancy for women who received a laparotomy 13
compared with women who received a laparoscopy. The evidence for this outcome was of low and 14
Length of hospital stay 16
One meta-analysis of two studies, and eight other studies, found that the length of hospital stay was 17
longer in women who received a laparotomy compared with women who received a laparoscopy. This 18
finding was statistically significant. The evidence for this finding was of moderate quality in the meta- 19
analysis of two studies, and very low in the other studies. 20
Need for further surgery 21
One meta-analysis of two studies, and four other studies, did not find a statistically significant 22
difference in the need for further surgery for women who received a laparotomy compared with 23
women who received a laparoscopy. The evidence for this outcome was of low and very low quality. 24
Need for methotrexate 25
Two studies did not find a statistically significant difference in the need for methotrexate for women 26
who received a laparotomy compared with women who received a laparoscopy. The evidence for this 27
outcome was of low and very low quality. 28
Need for surgery, methotrexate or expectant management 29
One study found that the need for surgery, methotrexate or expectant management was lower in 30
women who received a laparotomy compared with women who received a laparoscopy. This finding 31
was statistically significant. The evidence for this finding was of very low quality. 32
Readmission to hospital 33
Two studies did not find a statistically significant difference in the need for readmission to hospital for 34
women who received a laparotomy compared with women who received a laparoscopy. The evidence 35
for this outcome was of very low quality. 36
Abdominal pain 37
One study did not find a statistically significant difference in the incidence of abdominal pain for 38
for this outcome was of low quality. 40
Thromboembolic disease 41
One study did not find a statistically significant difference in the incidence of thromboembolic disease 42
for women who received a laparotomy compared with women who received a laparoscopy. The 43
Respiratory morbidity 45
Two studies did not find a statistically significant difference in the incidence respiratory morbidity for 46
for this outcome was of very low quality. 48
182
Two studies found that the need for a blood transfusion was higher in women who received a 2
laparotomy compared with women who received a laparoscopy. This finding was statistically 3
significant. The evidence for this finding was of very low quality in both studies. One further study did 4
not find a statistically significant difference in the need for a blood transfusion between the two 5
groups. The evidence for this finding was of very low quality. 6
Intraoperative blood loss 7
Three studies found that intraoperative blood loss was higher in women who received a laparotomy 8
compared with women who received a laparoscopy. This finding was statistically significant. The 9
evidence for this finding was moderate in one study and very low in the others. Three further studies 10
did not find a statistically significant difference in intraoperative blood loss between the two groups. 11
The evidence for this finding was of very low quality. 12
One study that only included women with tubal rupture and significant haemoperitoneum reported 13
length of hospital stay in a manner that did not allow assessment of statistical significance. In the 14
same study there were no events in either arm for the outcome of need for further surgery. The 15
evidence for these outcomes was of very low quality. 16
The GDG identified future reproductive outcomes as the most important outcomes from the womans 19
perspective. 20
Length of hospital stay was also considered very important, both from a womans point of view and in 21
terms of cost to the NHS. 22
Other outcomes relating to the womans experience of care were also seen as important, particularly 23
pain experienced and incidence of complications requiring further care e.g. need for blood 24
transfusion, need for further surgery and need for readmission to hospital. 25
Although reported by many studies, the GDG agreed that intraoperative blood loss is not a useful 26
outcome as it is very difficult to measure/assess accurately. More importantly, even in those studies 27
which reported statistically significant differences in blood loss, the GDG did not consider these 28
differences to be clinically significant 29
The evidence suggested there is little difference in clinically significant outcomes between laparotomy 31
and laparoscopy. Laparoscopy is superior in terms of length of hospital stay, with no evidence of any 32
harm compared to laparotomy, including no difference in need for further treatment (either surgical or 33
medical). Whilst it was noted that one study (Mol et al., 1997) showed fewer incidences of 34
methotrexate administration, expectant management or further surgery with laparotomy, the GDG 35
believed that this finding was due to the fact that most (84%) of the women in the laparotomy arm 36
received excision surgery to remove the tube and ectopic pregnancy (reported in the paper as 37
radical surgery), whilst the majority (66%) in the laparoscopy arm received conservative surgery. 38
The group did not feel able to make a strong recommendation due to the poor level of evidence 39
regarding clinical benefits. 40
The evidence suggested that there is no difference in terms of health benefits between laparoscopy 42
and laparotomy, including the key outcome of subsequent successful pregnancy. 43
The length of hospital stay is shorter following laparoscopy compared with laparotomy thus this 44
aspect of care, when combined with the reduced likelihood of a blood transfusion, would use less 45
resources. The GDG noted the fact that laparoscopy is associated with the use of expensive 46
equipment which is prone to malfunction if not well-maintained and/or renewed regularly. In addition, 47
hospital units that employ disposable equipment will consume limited resources. However, on 48
balance, the group felt that the capital cost of the laparoscopic equipment, when spread across the 49
number of procedures that would be undertaken, would be outweighed by the saving in the number of 50
183
hospital in-patient bed days. Therefore, they felt it appropriate to recommend laparoscopy as the first 1
line technique for most surgeries, although they were aware that the availability and presence of 2
competent trained surgical personnel on a 24/7 basis may limit the minimally invasive option of 3
treatment. 4
The quality of the evidence was moderate to very low. Some of the studies included few women so it 6
was difficult to draw any conclusions as the number of adverse events was very low in the study 7
groups. Thus the GDG were unable to draw conclusions about the incidence of thromboembolic 8
complications, need for blood transfusion, abdominal pain or respiratory complications. This was 9
unfortunate, as all of these were considered important outcomes which would impact on womens 10
experience of care as well as resource use. 11
It is regrettable that there was no evidence about emotional outcomes as these are also considered 12
important. 13
Whilst the GDG had prioritised womens experience of care as a key outcome for this review no data 15
was reported for this. The GDG recognised that the shorter hospital stay associated with laparoscopic 16
surgery was likely to be valued by most women and that being separated from their family and friends 17
for a shorter time was likely to be beneficial in terms of emotional support and reduce the negative 18
psychological impact of surgery. 19
The GDG was aware that there may be some practitioners who are not competent to perform 21
laparoscopy, particularly in more complex cases. In addition, some surgeons are not comfortable with 22
laparoscopic surgery where the woman is collapsed or haemodynamically unstable. 23
An additional problem can arise where the laparoscopic equipment is unavailable/not working. The 24
GDG recognised from their clinical experience that equipment failures may lead to unnecessarily 25
lengthy operations. The decision as to which intervention is most appropriate therefore depends on 26
the relative expertise of the doctor, the equipment available, the complexity of the surgery required 27
and the condition of the woman. 28
The GDG felt that each unit should have at least some practitioners who are competent to perform 29
laparoscopy during daytime hours (e.g. 8am to 8pm). In order to achieve this, further investment in 30
training and equipment may be necessary. 31
Recommendations 32
77 When surgical treatment is indicated for women with an ectopic pregnancy,
laparoscopy should be performed whenever possible, taking into account the
condition of the woman and the complexity of the surgical procedure.
78 Surgeons providing care to women with ectopic pregnancy should be competent to
perform laparoscopic surgery.
79 Out-of-hours surgery should only be carried out as an emergency if a woman is
haemodynamically unstable or collapsed.
80 Commissioners and managers should ensure that equipment for laparoscopic
surgery is available.
33
184
8.4 Salpingectomy compared with salpingotomy for 1
ectopic pregnancy 2
Review question 3
What is the effectiveness of salpingectomy compared with salpingotomy in improving outcomes in 4
women with tubal ectopic pregnancy? 5
Introduction 6
Surgical treatment of ectopic pregnancy can be conservative, preserving the Fallopian tube; or 7
radical, removing the Fallopian tube. Salpingotomy is surgical incision of a Fallopian tube to remove 8
the ectopic pregnancy. Salpingostomy is the surgical formation of an opening in the Fallopian tube 9
where the fimbrial end has been closed by infection or chronic inflammation. In the literature there 10
appears to be a blurring of this definition with authors apparently using the terms interchangeably and 11
sometimes referring to salpingotomy and salpingostomy within the same paper. The term 12
salpingotomy will be used in this review to cover both salpingotomy and salpingostomy. A 13
conservative approach, salpingotomy, preserves the tube but bears the risk of incomplete removal of 14
the pregnancy tissue (persistent trophoblast). A radical approach, salpingectomy, bears no risk of 15
persistent trophoblast but leaves only one tube for reproductive capacity. It is unknown which type of 16
operation is better, especially for future fertility. The risk of additional treatment in the case of 17
persistent trophoblast after salpingotomy may be acceptable if compensated by retention of the 18
Fallopian tube and a small rise in intrauterine pregnancy rate. 19
Twenty one studies were included in this review (Bangsgaard et al., 2003; Becker et al., 2011; Bouyer 21
et al., 2000; Colacurci et al., 1998; DeCherney & Kase, 1979; dela Cruz & Cumming, 1997; Giambelli 22
et al., 1996; Gruft et al., 1994; Kuroda et al., 2009; Langebrekke et al., 1993; Mecke et al., 1989; Mol 23
et al., 1997; Mol et al., 1998; Ory et al., 1993; Parker et al., 1994; Sherman et al., 1982; Silva et al., 24
1993; Tahseen & Wyldes, 2003; Tulandi & Guralnick, 1991; Turan, 2011; Tuomivaara & Kauppila, 25
1988). 26
There were no relevant randomised controlled trials identified for this review question. Four of the 27
included studies were prospective observational studies (Becker et al., 2011; Bouyer et al., 2000; Mol 28
et al., 1997; Silva et al., 1993) and the remainder were retrospective observational studies. The 29
studies were conducted in the UK (Tahseen & Wyldes, 2003), Denmark (Bangsgaard et al., 2003), 30
Finland (Tuomivaara & Kauppila, 1988), France (Bouyer et al., 2000), Germany (Becker et al., 2011; 31
Mecke et al., 1989), Italy (Colacurci et al., 1998; Giambelli et al., 1996; Gruft et al., 1994), the 32
Netherlands (Mol et al., 1997; Mol et al., 1998), Norway (Langebrekke et al., 1993), Australia (Parker 33
et al., 1994), the USA (DeCherney & Kase, 1979; Ory et al., 1993; Silva et al., 1993), Canada (dela 34
Cruz & Cumming, 1997; Tulandi & Guralnick, 1991), Turkey (Turan, 2011), Israel (Sherman et al., 35
1982), and Japan (Kuroda et al., 2009). 36
The majority of the included studies compared salpingectomy with salpingotomy for the management 37
of tubal ectopic pregnancy; however in four of the studies, some of the participants received a 38
different type of radical or conservative surgery, such as a salpingo-oophorectomy or milking of the 39
tube (DeCherney & Kase, 1979; Mol et al., 1997; Sherman et al., 1982; Tuomivaara & Kauppila, 40
1988). 41
Evidence profile 42
Table 8.5 GRADE summary of findings for comparison of salpingectomy with salpingotomy for the management 43
of tubal ectopic pregnancy 44
Number of
studies
Number of women or cumulative
probability
Effect Quality
Salpingectomy Salpingotomy Relative Absolute
185
(95% CI) (95% CI) and p-
value if
reported
Subsequent live birth or full-term birth
1 study
(Silva et al.,
1993)
10/26
(38.5%)
19/60
(31.7%)
RR 1.21
(0.66 to 2.24)
67 more per
1000
(from 108 fewer
to 393 more)
Very low
1 study
(dela Cruz &
Cumming,
1997)
21/56
(37.5%)
16/34
(47.1%)
RR 0.8
(0.49 to 1.3)
94 fewer per
1000
(from 240 fewer
to 141 more)
Very low
1 study
(Mol et al.,
1998)
18/79
(22.8%)
22/56
(39.3%)
RR 0.58
(0.34 to 0.98)
165 fewer per
1000
(from 8 fewer to
259 fewer)
Very low
1 study
(Langebrekke
et al., 1993)
18/40
(45%)
38/58
(65.5%)
RR 0.69
(0.46 to 1.01)
203 fewer per
1000
(from 354 fewer
to 7 more)
Very low
1 study
(Gruft et al.,
1994)
23/71
(32.4%)
12/44
(27.3%)
RR 1.19
(0.66 to 2.14)
52 more per
1000
(from 93 fewer
to 311 more)
Very low
1 study
(Bangsgaard
et al., 2003)
21/68
(30.9%)
88/208
(42.3%)
RR 0.73
(0.49 to 1.08)
114 fewer per
1000
(from 216 fewer
to 34 more)
Very low
1 study
(Ory et al.,
1993)
29/50
(58%)
17/33
(51.5%)
RR 1.13
(0.75 to 1.69)
67 more per
1000
(from 129 fewer
to 355 more)
Very low
1 study
(DeCherney &
Kase, 1979)
21/50
(42%)
19/48
(39.6%)
RR 1.06
(0.66 to 1.71)
24 more per
1000
(from 135 fewer
to 281 more)
Very low
Subsequent intrauterine pregnancy
1 study
(Bouyer et al.,
2000)
18-month
cumulative rate
(95% CI): 57%
(44 to 70)
n=100
18-month
cumulative rate
(95% CI): 73%
(65 to 80)
n=166
Hazard ratio
0.56
(0.39 to 0.81)*
160 fewer per
1000
(confidence
interval not
calculable [NC])
Very low
1 study
(Becker et al.,
2011)
25/51
(49%)
122/145
(84.1%)
RR 0.58
(0.44 to 0.78)*
353 fewer per
1000
(from 185 fewer
Very low
186
to 471 fewer)
1 study
(Silva et al.,
1993)
14/26
(53.8%)
36/60
(60%)
RR 0.9
(0.59 to 1.35)
60 fewer per
1000
(from 246 fewer
to 210 more)
Very low
1 study
(Langebrekke
et al., 1993)
19/40
(47.5%)
40/58
(69%)
RR 0.69
(0.48 to 1)
214 fewer per
1000
(from 359 fewer
to 0 more)
Very low
1 study
(dela Cruz &
Cumming,
1997)
27/56
(48.2%)
23/34
(67.6%)
RR 0.71
(0.5 to 1.02)
196 fewer per
1000
(from 338 fewer
to 14 more)
Very low
1 study
(Mol et al.,
1998)
24/79
(30.4%)
30/56
(53.6%)
RR 0.57
(0.38 to 0.86)*
230 fewer per
1000
(from 75 fewer
to 332 fewer)
Very low
1 study
(Bangsgaard
et al., 2003)
39/68
(57.4%)
161/208
(77.4%)
RR 0.74
(0.6 to 0.92)
201 fewer per
1000
(from 62 fewer
to 310 fewer)
Very low
1 study
(Turan, 2011)
33/55
(60%)
23/35
(65.7%)
RR 0.91
(0.66 to 1.26)
59 fewer per
1000
(from 223 fewer
to 171 more)
Very low
1 study
(Tuomivaara &
Kauppila,
1988)
170/237
(71.7%)
59/86
(68.6%)
RR 1.05
(0.89 to 1.23)
34 more per
1000
(from 75 fewer
to 158 more)
Very low
1 study
(Sherman et
al., 1982)
75/104
(72.1%)
39/47
(83%)
RR 0.87
(0.73 to 1.04)*
108 fewer per
1000
(from 224 fewer
to 33 more)
Very low
1 study
(Giambelli et
al., 1996)
6-month
cumulative rate:
62.5%
(denominator
NR)
6-month
cumulative rate:
53.8%
(denominator
NR)
NC 87 more per
1000
(confidence
interval NC)
NS (p-value NR)
Very low
1 study
(Tulandi &
Guralnick,
1991)
24-month
cumulative
probability: 26%
n=24
24-month
cumulative
probability: 47%
n=34
NC 210 fewer per
1000
(confidence
interval NC)
p<0.05
Very low
1 study
(Tahseen &
38/97
(39.2%)
12/25
(48%)
RR 0.82
(0.51 to 1.32)
86 fewer per
1000
Very low
187
Wyldes, 2003) (from 235 fewer
to 154 more)
1 study
(Kuroda et al.,
2009)
17/40
(42.5%)
24/43
(55.8%)
RR 0.76
(0.49 to 1.19)
134 fewer per
1000
(from 285 fewer
to 106 more)
Very low
1 study
(Colacurci et
al., 1998)
2/11
(18.2%)
10/26
(38.5%)
RR 0.47
(0.12 to 1.81)
204 fewer per
1000
(from 338 fewer
to 312 more)
Very low
1 study
(Bouyer et al.,
2000)
10/100
(10%)
17/166
(10.2%)
RR 0.98
(0.47 to 2.05)
2 fewer per
1000
(from 54 fewer
to 108 more)
Very low
1 study
(Becker et al.,
2011)
7/51
(13.7%)
11/145
(7.6%)
RR 1.81
(0.74 to 4.42)
61 more per
1000
(from 20 fewer
to 259 more)
Very low
1 study
(Silva et al.,
1993)
2/26
(7.7%)
11/60
(18.3%)
RR 0.42
(0.1 to 1.76)
106 fewer per
1000
(from 165 fewer
to 139 more)
Very low
1 study
(Langebrekke
et al., 1993)
4/40
(10%)
4/58
(6.9%)
RR 1.45
(0.39 to 5.46)
31 more per
1000
(from 42 fewer
to 308 more)
Very low
1 study
(dela Cruz &
Cumming,
1997)
10/56
(17.9%)
4/34
(11.8%)
RR 1.52
(0.52 to 4.46)
61 more per
1000
(from 56 fewer
to 407 more)
Very low
1 study
(Ory et al.,
1993)
3/50
(6%)
8/33
(24.2%)
RR 0.25
(0.07 to 0.87)*
182 fewer per
1000
(from 32 fewer
to 225 fewer)
Very low
1 study
(Mol et al.,
1998)
7/79
(8.9%)
5/56
(8.9%)
RR 0.99
(0.33 to 2.97)
1 fewer per
1000
(from 60 fewer
to 176 more)
Very low
1 study
(Bangsgaard
et al., 2003)
8/68
(11.8%)
28/208
(13.5%)
RR 0.87
(0.42 to 1.83)
18 fewer per
1000
(from 78 fewer
to 112 more)
Very low
1 study
(Turan, 2011)
2/55
(3.6%)
6/35
(17.1%)
RR 0.21
(0.05 to 0.99)
135 fewer per
1000
(from 2 fewer to
Very low
188
163 fewer)
1 study
(Sherman et
al., 1982)
6/104
(5.8%)
3/47
(6.4%)
RR 0.9
(0.24 to 3.46)
6 fewer per
1000
(from 49 fewer
to 157 more)
Very low
1 study
(Tuomivaara &
Kauppila,
1988)
25/237
(10.5%)
10/86
(11.6%)
RR 0.91
(0.45 to 1.81)
10 fewer per
1000
(from 64 fewer
to 94 more)
Very low
1 study
(Giambelli et
al., 1996)
6-month
cumulative rate:
5.1%
(denominator
NR)
6-month
cumulative rate:
7.8%
(denominator
NR)
NC 27 fewer per
1000
(confidence
interval NC)
NS (p-value NR)
Very low
1 study
(DeCherney &
Kase, 1979)
6/50
(12%)
9/48
(18.8%)
RR 0.64
(0.25 to 1.66)
68 fewer per
1000
(from 141 fewer
to 124 more)
Very low
1 study
(Tulandi &
Guralnick,
1991)
24-month
cumulative
probability: 13%
n=24
24-month
cumulative
probability: 31%
n=34
NC 180 fewer per
1000
(confidence
interval NC)
p<0.05
Very low
1 study
(Kuroda et al.,
2009)
7/40
(17.5%)
4/43
(9.3%)
RR 1.88
(0.6 to 5.94)
82 more per
1000
(from 37 fewer
to 460 more)
Very low
1 study
(Colacurci et
al., 1998)
1/11
(9.1%)
1/26
(3.8%)
RR 2.36
(0.16 to 34.5)
52 more per
1000
(from 32 fewer
to 1000 more)
Very low
1 study
(Bouyer et al.,
2000)
1/178
(0.56%)
14/262
(5.3%)
RR 0.11
(0.01 to 0.79)
48 fewer per
1000
(from 11 fewer
to 53 fewer)
Very low
1 study
(Mol et al.,
1997)
1/157
(0.64%)
18/98
(18.4%)
RR 0.03
(0 to 0.26)
178 fewer per
1000
(from 136 fewer
to 184 fewer)
Very low
1 study
(Parker et al.,
1994)
1/103
(0.97%)
6/50
(12%)
RR 0.08
0.01 to 0.65)
110 fewer per
1000
(from 42 fewer
to 119 fewer)
Very low
1 study 0/25 14/153 RR 0.2 73 fewer per Very low
189
(Mecke et al.,
1989)
(0%) (9.2%) (0.01 to 3.32) 1000
(from 91 fewer
to 212 more)
1 study
(Giambelli et
al., 1996)
0/59
(0%)
4/55
(7.3%)
RR 0.1
(0.01 to 1.88)
65 fewer per
1000
(from 72 fewer
to 64 more)
Very low
1 study
(Mol et al.,
1997)
10/157
(6.4%)
1/98
(1%)
RR 6.24
(0.81 to 48.01)
53 more per
1000
(from 2 fewer to
480 more)
Very low
1 study
(Colacurci et
al., 1998)
0/13
(0%)
1/32
(3.1%)
RR 0.79
(0.03 to 18.13)
7 fewer per
1000
(from 30 fewer
to 535 more)
Very low
Surgical complications
1 study
(Mol et al.,
1997)
2/157
(1.3%)
3/98
(3.1%)
RR 0.42
(0.07 to 2.45)
18 fewer per
1000
(from 28 fewer
to 44 more)
Very low
1 study
(Mecke et al.,
1989)
0/25
(0%)
6/153
(3.9%)
RR 0.46
(0.03 to 7.85)
21 fewer per
1000
(from 38 fewer
to 269 more)
Very low
* Significance is altered when other factors influencing fertility are controlled for (using multivariate analysis or stratification) 1
The studies identified for this review question were generally of poor quality. Using GRADE criteria, 3
the evidence was of very low quality for every outcome. 4
Subsequent live birth or full-term birth 5
One study found that the proportion of women with a subsequent live birth or full-term birth was lower 6
in women who received a salpingectomy compared with women who received a salpingotomy. This 7
finding was statistically significant. A further seven studies did not find a statistically significant 8
difference in subsequent live birth or full-term birth between the two groups. 9
Subsequent intrauterine pregnancy 10
Five studies found that the proportion of women with a subsequent intrauterine pregnancy was lower 11
in women who received a salpingectomy compared with women who received a salpingotomy. This 12
finding was statistically significant. A further ten studies did not find a statistically significant difference 13
in subsequent intrauterine pregnancy between the two groups. 14
Three studies found that the proportion of women with a recurrent ectopic pregnancy was lower in 16
women who received a salpingectomy compared with women who received a salpingotomy. This 17
finding was statistically significant. A further twelve studies did not find a statistically significant 18
difference in recurrent ectopic pregnancy between the two groups. 19
190
Three studies found that the need for further intervention was lower in women who received a 2
salpingectomy compared with women who received a salpingotomy. This finding was statistically 3
significant. Two further studies did not find a statistically significant difference in the need for further 4
intervention between the two groups. 5
Two studies did not find a statistically significant difference in the need for a blood transfusion for 7
women who received a salpingectomy compared with women who received a salpingotomy. 8
Surgical complications 9
Two studies did not find a statistically significant difference in the incidence of surgical complications 10
for women who received a salpingectomy compared with women who received a salpingotomy. 11
Health economics 12
A de novo economic model was developed for this guideline to assess the cost-effectiveness of 13
different treatment strategies for ectopic pregnancy. Please see section 7.2 for a summary of the 14
models findings and section 9.4 for full details of the model. 15
The primary outcomes for this question were the reproductive outcomes and the need for further 18
intervention. The likelihood of a future viable intrauterine pregnancy and the possibility of a repeat 19
ectopic pregnancy are thought to be outcomes that are very important to most women. The chance of 20
a further intervention is important for informing women about the likely course of their recovery, as 21
well as having health economic implications. Secondary outcomes for this review included the need 22
for a blood transfusion, incidence of surgical complications, and ongoing pain, although evidence was 23
not available for the latter. 24
When considering which mode of surgery to recommend, the GDG felt that maintaining the womans 26
reproductive potential was a priority. Six of the studies showed that the proportion of women with an 27
intrauterine pregnancy or a subsequent live/full-term birth was lower in women who received a 28
salpingectomy, while the remainder showed no significant difference. However, the group noted that 29
the evidence was derived from observational studies, and in many of the papers it was reported that a 30
womans future reproductive desires and other fertility factors influenced the mode of surgery 31
performed, therefore biasing the results. When multivariate or stratified analysis was performed 32
(adjusting for other factors contributing to fertility, such as tubal pathology and history of infertility), the 33
effect became non-significant in two studies. In two other studies, the difference in the proportion of 34
women with an intrauterine pregnancy or a subsequent live/full-term birth was significant for women 35
with other factors prognostic of infertility, but not for the remainder of the women. 36
Three out of five studies showed that the need for a further intervention was significantly higher 37
following a salpingotomy. The GDG recognised that this would be an important consideration for 38
women, particularly as the second intervention might include a more radical procedure i.e. 39
salpingectomy. 40
For the outcome of recurrent ectopic pregnancy, there was a general trend that the incidence was 41
lower among women who received a salpingectomy; however due to the poor quality of the evidence, 42
the GDG did not feel that the effect was certain enough to base a recommendation on. 43
Overall, the GDG felt that for women without any coexistent fertility factors, future reproductive 44
potential was unlikely to be strongly affected by which mode of surgery was performed. However, for 45
women with factors prognostic of infertility, the evidence suggested that salpingotomy was associated 46
with a higher change of a subsequent intrauterine pregnancy (please refer to evidence tables in 47
appendix H, where stratified or adjusted analyses are reported for Becker et al., 2011; Bouyer et al., 48
2000; Sherman et al., 1982). . 49
191
The GDG noted that the evidence showed a higher incidence of further intervention following a 2
salpingotomy, and therefore that the comparison of salpingectomy and salpingotomy should be 3
incorporated in the health economic analysis for management of ectopic pregnancy. The cost 4
minimisation model (see section 9.4) showed that salpingectomy was preferable to salpingotomy in 5
terms of resource use. Given that the evidence around reproductive outcomes following 6
salpingectomy and salpingotomy was inconclusive, the group felt that for women without any 7
coexistent fertility factors, the recommended surgical treatment should be a salpingectomy. 8
The evidence for this review was of very low quality because it was drawn from observational studies, 10
the majority of which were retrospective. There was a high likelihood of bias in many of the studies, 11
because they had a select population and a womans future fertility desires and reproductive history 12
contributed to the choice of treatment. However, the GDG felt that the studies which performed 13
stratified or multivariate analyses controlled for some of these factors and therefore presented a more 14
accurate estimate of reproductive outcomes. 15
The likelihood of a future viable intrauterine pregnancy and the possibility of a repeat ectopic 17
pregnancy were prioritised as key outcomes as these were felt to be the most important to women. 18
The chance of a further intervention is important for informing women about the likely course of their 19
recovery and a recommendation was made to reflect this. 20
The GDG were aware that the current RCOG guidelines on this subject recommend salpingectomy 22
for all women. However, in light of the evidence, the GDG wanted to ensure that clinicians had the 23
option to perform a salpingotomy in women who had other factors that might reduce their fertility. 24
The GDG also discussed the ongoing ESEP study: a large multi-centre randomised controlled trial, 25
which compares salpingectomy and salpingotomy. Unfortunately, the trial will not be published in time 26
for inclusion in this guideline; however, in light of the fact that this study is being conducted, the GDG 27
did not feel that this area was a priority for a research recommendation. 28
Recommendations 29
81 Offer a salpingectomy to women undergoing surgery for an ectopic pregnancy,
unless they have other risk factors for infertility.
82 Consider salpingotomy as an alternative to salpingectomy for women with factors
prognostic of infertility such as contralateral tube damage.
83 Inform women having a salpingotomy that up to one in five women may need further
treatment. This treatment may include methotrexate administration and/or a
salpingectomy.
84 For women who have had a salpingotomy, take one serum hCG measurement at 7
days after surgery, then one hCG measurement per week until a negative result is
obtained.
85 Advise women who have had a salpingectomy that they should take a urine
pregnancy test after 3 weeks.
30
31


9 Anti-D rhesus 1
prophylaxis 2
9.1 Introduction 3
Haemolytic disease of the newborn is caused by the destruction of fetal red blood cells by maternal 4
antibodies against red cell antigens acquired from the father. Fifteen percent of all women are rhesus 5
(D) protein negative and therefore historically the vast majority of cases of haemolytic disease of the 6
newborn have been caused by the production of antibodies against the rhesus D antigen. Currently 7
there is confusion about whether anti-D prophylaxis is required to prevent sensitisation during 8
bleeding in the first trimester of pregnancy. The group therefore considered the risk of sensitisation in 9
the first 13 weeks of a pregnancy complicated by bleeding. In women in whom anti-D is required, the 10
group then looked at the most appropriate dose of anti-D that should be recommended to prevent 11
sensitisation. 12
9.2 Anti-D rhesus prophylaxis for threatened 13
miscarriage, miscarriage and ectopic pregnancy 14
Review question 15
Should anti-D rhesus prophylaxis be given to women with a threatened miscarriage, miscarriage or 16
ectopic pregnancy in the first trimester? 17
Eight studies were included in this review (Gavin, 1972; Katz & Marcus, 1973; Murray & Barron, 1971; 19
Murray et al., 1970; Simonovits et al., 1974; Simonovits et al., 1980; Visscher & Visscher, 1972; 20
Walsh & Lewis, 1970). 21
Of the included papers, five were non-comparative, descriptive studies, reporting the incidence of 22
sensitisation in women receiving no anti-D rhesus prophylaxis following first trimester obstetric events 23
(Katz & Marcus, 1973; Murray & Barron, 1971; Murray et al., 1970; Simonovits et al., 1980; Walsh & 24
Lewis, 1970). The remaining three papers were comparative studies, examining the effect of anti-D 25
rhesus prophylaxis on outcomes. One randomised controlled trial compared 300 micrograms of anti-D 26
rhesus prophylaxis with placebo, with the authors also reporting an additional prospective case series 27
of 9 women who did not receive any intervention (Visscher & Visscher, 1972). One study was a non- 28
randomised trial, comparing anti-D rhesus prophylaxis (dose not stated) with placebo (Gavin, 1972). 29
One prospective observational study compared outcomes in women who received anti-D rhesus 30
prophylaxis following a previous therapeutic abortion, and those who did not receive any prophylaxis 31
(Simonovits et al., 1974). 32
No studies were found that evaluated the use of anti-D rhesus prophylaxis after a threatened 33
miscarriage or ectopic pregnancy. Two studies evaluated outcomes in women who were diagnosed 34
with a miscarriage (Katz & Marcus, 1973; Visscher & Visscher, 1972). Three studies had a mixed 35
population, comprising both women receiving surgery for a miscarriage and women undergoing a 36
therapeutic abortion (Gavin, 1972; Murray & Barron, 1971; Murray et al., 1970). Three studies only 37
included women having a therapeutic abortion (Simonovits et al., 1974; Simonovits et al., 1980; 38
Walsh & Lewis, 1970). The GDG felt that, for this review, it was appropriate to include studies of 39
women having a therapeutic abortion. The reasons driving this decision were: the lack of evidence in 40
193
the populations of interest, the comparability of the surgical procedures, and the fact that the 1
outcomes of interest are biochemical rather than psychological. 2
Findings for the outcomes of interest reported are presented below in two evidence profiles. The first 3
profile includes the five non-comparative studies and the second includes the three comparative 4
studies. 5
Evidence profile 6
Table 9.1 GRADE summary of findings for series of women receiving no anti-D rhesus prophylaxis (non- 7
comparative data) 8
Number of studies Test used for antibody
detection*
Number of patients Quality
Incidence of sensitisation at 5-9 months following miscarriage/abortion
1 study
(Katz & Marcus, 1973)
Use of multiple tests
reported
1/36

(2.8%)
Very low
1 study
(Visscher & Visscher,
1972)
Enzyme-Coombs
screening procedure
0/9
(0%)
Very low
1 study
(Murray & Barron, 1971)
Indirect Coombs test 2/96
(2.1%)
Very low
Enzyme-treated cells 9/96
(9.4%)
1 study
(Murray et al., 1970)
(4.3%)
Very low
Lows papain 2/23
(8.7%)
Papain-treated cells 3/23
(13.0%)
1 study
(Walsh & Lewis, 1970)
(5.6%)
Very low
1 study
Use of multiple tests
reported
5/25
(20%)
Very low
Evidence of sensitisation in subsequent pregnancy
1 study
(Simonovits et al., 1980)
(0.8%)**
Very low
Papain-treated cells 6/386
(1.6%)**
1 study
(Visscher & Visscher,
1972)
Enzyme-Coombs
screening procedure
0/2
(0%)
Very low
Neonatal outcomes in sensitised women: delivery of hydropic infant or baby with hyperbilirubinemia
194
1 study
N/A 3/4
(75%)
Very low
Neonatal outcomes in sensitised women: positive direct Coombs test in baby born following
subsequent pregnancy
1 study
Direct Coombs test 2/3
(66.7%)
Very low
*The indirect Coombs test (also known as the indirect antiglobulin test) is currently the standard test for detecting whether a Rh- 1
woman has antibodies against the Rh D antigen present in her blood, and therefore whether she has been sensitised. 2
Historically, enzyme-treated red blood cells (such as those treated with papain) were used to improve the sensitivity of antibody 3
screening tests and were part of the screen for anti-D. However, tests using treated red blood cells detect a lot of non-specific 4
antibodies in addition to anti-D, and therefore in current practice they are only used in confirmatory tests and reference labs. 5
The direct Coombs test (also known as the direct antiglobulin test) is used to test a babys blood, and determine whether 6
maternal antibodies have bound to the babys red blood cells. This can be used to establish whether the baby is suffering from, 7
or is at risk of, haemolytic disease of the newborn. 8
The woman had weak antibody titre on admission, and then a titre of 1:4 at 5 months 9
** These are test results from the 2
nd
to 3
rd
month of second pregnancy. Test results from month 8-9 have also been reported in 10
the study, but are not reported here. 11
Table 9.2 GRADE summary of findings for anti-D rhesus prophylaxis compared with no intervention or placebo 12
(comparative data) 13
Number of
studies
Test used
for antibody
detection*
Anti-D
prophylaxis
Placebo /
no
intervention
Relative
(95% CI)
Absolute
(95% CI)
Incidence of sensitisation at 4-6 months following miscarriage/abortion
1 study
(Visscher &
Visscher, 1972)
Enzyme-
Coombs
screening
procedure
0/19
(0%)
0/29
(0%)
Not
calculable
(NC)
NC Very low
1 study
(Gavin, 1972)
Indirect
Coombs test
0/21
(0%)
2/36
(5.6%)
RR 0.34
(0.02 to
6.69)
37 fewer per
1000
(from 54
fewer to 316
more)
Very low
Evidence of sensitisation in subsequent pregnancy
1 study
(Visscher &
Visscher, 1972)
Enzyme-
Coombs
screening
procedure
0/3
(0%)
0/6
(0%)
NC NC Very low
1 study
(Simonovits et
al., 1974)
Anti-D:
papain-
treated cells
No
intervention:
indirect
Coombs test
and papain-
treated cells
for one
1/96**
(1.0%)
2/145
(1.4%)
RR 0.76
(0.07 to
8.21)
3 fewer per
1000
(from 13
fewer to 99
more)
Very low
195
woman; not
reported for
other
*The indirect Coombs test (also known as the indirect antiglobulin test) is currently the standard test for detecting whether a Rh- 1
woman has antibodies against the Rh D antigen present in her blood, and therefore whether she has been sensitised. 2
Historically, enzyme-treated red blood cells (such as those treated with papain) were used to improve the sensitivity of antibody 3
screening tests and were part of the screen for anti-D. However, tests using treated red blood cells detect a lot of non-specific 4
antibodies in addition to anti-D, and therefore in current practice they are only used in confirmatory tests and reference labs. 5
The direct Coombs test (also known as the direct antiglobulin test) is used to test a babys blood, and determine whether 6
maternal antibodies have bound to the babys red blood cells. This can be used to establish whether the baby is suffering from, 7
or is at risk of, haemolytic disease of the newborn. 8
**This woman delivered a Rh+ baby at the end of her second pregnancy and tested negative 6 months before birth; therefore 9
she is likely to have been sensitised in her second, full-term pregnancy 10
The evidence for each of the reported studies and outcomes is of very low quality. 12
Non-comparative data (see table 9.1) 13
Incidence of sensitisation at 5-9 months following miscarriage/abortion 14
One study found that the incidence of sensitisation was 2.8% among women who did not receive any 15
prophylaxis following a miscarriage; however the only case of sensitisation occurred in a woman who 16
had a weak antibody titre on admission. 17
One study did not report any incidences of sensitisation (using the enzyme-Coombs screening 18
procedure) among women who did not receive any prophylaxis following a miscarriage. 19
One study reported that the incidence of sensitisation was 2.1% using the indirect Coombs test, and 20
9.4% using enzyme-treated cells, among women who did not receive any prophylaxis following a 21
miscarriage or therapeutic abortion. 22
One study reported that the incidence of sensitisation was 4.3% using the indirect Coombs test, 8.7% 23
using Lows papain, and 13.0% using papainised cells, among women who did not receive any 24
prophylaxis following a miscarriage or therapeutic abortion. 25
One study reported that the incidence of sensitisation was 5.6% using the indirect Coombs test 26
among women who did not receive any prophylaxis following a therapeutic abortion. 27
One study reported that the incidence of sensitisation was 20% among women who did not receive 28
any prophylaxis following a miscarriage or therapeutic abortion; however definitive proof that the 29
miscarriage or therapeutic abortion was the cause of sensitisation was only available in 2/5 cases. 30
Evidence of sensitisation in subsequent pregnancy 31
One study reported that the incidence of sensitisation during a subsequent pregnancy was 0.8% 32
using the indirect Coombs test, and 1.6% using papain-treated red blood cells, in women who did not 33
receive any prophylaxis following a therapeutic abortion. 34
One study reported no incidences of sensitisation during a subsequent pregnancy (using the enzyme- 35
Coombs screening procedure) in 2 women who did not receive any prophylaxis following a 36
miscarriage. 37
Neonatal outcomes in sensitised women 38
One study found that 3 out of 4 women that were sensitised following a miscarriage delivered a 39
hydropic infant or baby with hyperbilirubinemia in a subsequent pregnancy. The same study found 40
that 2 out of 3 babies born to women sensitised after a miscarriage had a positive direct Coombs test. 41
Comparative data (see table 9.2) 42
Incidence of sensitisation at 4-6 months following miscarriage/abortion 43
One study did not find a statistically significant difference in the incidence of sensitisation for women 44
receiving anti-D rhesus prophylaxis following a miscarriage or therapeutic abortion compared with 45
196
women receiving a placebo. One further study did not report any incidences of sensitisation in women 1
receiving anti-D rhesus prophylaxis following a miscarriage and women receiving a placebo. 2
Evidence of sensitisation in subsequent pregnancy 3
One study did not find a statistically significant difference in the incidence of sensitisation during 4
subsequent pregnancies for women receiving anti-D rhesus prophylaxis following a therapeutic 5
abortion compared with women receiving a placebo. One further study did not report any incidences 6
of sensitisation during subsequent pregnancies in women receiving anti-D rhesus prophylaxis 7
following a miscarriage and women receiving a placebo. 8
Please see recommendations in section 8.3, where the evidence from all of the anti-D rhesus 10
prophylaxis reviews has been considered. 11
9.3 Anti-D rhesus prophylaxis dose 12
Review question 13
What is the appropriate dose of anti-D that should be administered to women with a threatened 14
miscarriage, miscarriage or ectopic pregnancy in the first trimester? 15
Three studies were included in this review (Hensleigh et al., 1977; Keith & Bozorgi, 1977; Stewart et 17
al., 1978). All of the studies were randomised controlled trials conducted in the USA, and evaluated 18
the administration of different doses of anti-D Rhesus prophylaxis to women following first trimester 19
therapeutic abortion. Two studies compared doses of 50 micrograms and 300 micrograms (Keith & 20
Bozorgi, 1977; Stewart et al., 1978). The third study compared three different doses of 73, 155 and 21
499 micrograms (Hensleigh et al., 1977). No studies were identified that compared different doses 22
following an ectopic pregnancy or miscarriage. 23
Evidence profile 24
Table 9.3 GRADE summary of findings for comparison of 50 micrograms and 300 micrograms of anti-D 25
prophylaxis 26
Number of
studies
50 micrograms 300
micrograms
Relative
(95% CI)
Absolute
(95% CI)
Detection of Rhesus antibodies at 6 months follow-up
1 meta-
analysis of 2
studies
(Keith &
Bozorgi, 1977;
Stewart et al.,
1978)
0/989
(0%)
0/81
(0%)
NC NC Very low
Adverse drug reaction
1 meta-
analysis of 2
studies
(Keith &
Bozorgi, 1977;
1/1218
(0.08%)
0/111
(0%)
RR 0.31
(0.01 to 7.61)
1 more
(from 33 fewer
to 5 more)*
Very low
197
Stewart et al.,
1978)
* NCC calculation 1
Table 9.4 GRADE summary of findings for comparison of 73, 155 and 499 micrograms of anti-D prophylaxis 2
Number of
studies
73
micrograms
155
micrograms
499
micrograms
Relative
(95% CI)
Absolute
(95% CI)
Incidence of sensitisation
1 study
(Hensleigh
et al.,
1977)
0/8 0/83 0/25 NC NC Very low
Adverse drug reaction
1 study
(Hensleigh
et al.,
1977)
0/8 0/83 0/25 NC NC Very low
3
The evidence for each of the reported studies and outcomes is of very low quality. 5
Comparison of 50 and 300 micrograms 6
Detection of rhesus antibodies at 6 months 7
One meta-analysis of two studies did not find any incidences of rhesus antibody detection in women 8
who received 50 micrograms or 300 micrograms of anti-D. The evidence for this outcome was of very 9
low quality. 10
Adverse drug reaction 11
adverse drug reaction for women who received 50 micrograms of anti-D compared with women who 13
received 300 micrograms of anti-D. The evidence for this outcome was of very low quality. 14
Comparison of 73, 155 and 499 micrograms 15
Incidence of sensitisation 16
One study did not find any incidences of sensitisation in women who received 73 micrograms, 155 17
micrograms or 499 micrograms of anti-D. The evidence for this outcome was of very low quality. 18
Adverse drug reaction 19
One study did not find any incidences of adverse drug reaction in women who received 73 20
micrograms, 155 micrograms or 499 micrograms of anti-D. The evidence for this outcome was of very 21
low quality. 22
For the review looking at the provision of anti-D rhesus prophylaxis, the key outcome of interest for 25
the GDG was the incidence of sensitisation following a miscarriage or therapeutic abortion, as it is this 26
which anti-D rhesus prophylaxis is supposed to prevent. This included both sensitisation in the current 27
pregnancy, and sensitisation in subsequent pregnancies. 28
198
For the review looking at the appropriate dose, the key outcome was the effectiveness of the 1
prophylaxis at different doses. 2
Trade off between clinical benefits and harms 3
The group recognised that there is little harm associated with the provision of anti-D rhesus 4
prophylaxis. Whilst there is always a potential risk of transferring blood-borne disease when 5
administering blood products, this risk is very low given the screening that is conducted before their 6
use. The group was not aware of any other adverse outcomes associated with the use of anti-D 7
rhesus prophylaxis, and there was only one such outcome reported in the available evidence. 8
By contrast, the group recognised that there is a clear health benefit in avoiding sensitisation if 9
possible. Sensitisation increases the chance of miscarriage in a later pregnancy and also increases 10
the chance that subsequent babies will develop a range of conditions including fetal heart failure, 11
hydrops (fluid retention), oedema, anaemia and rhesus disease. Rhesus disease, in turn, increases 12
the chance of the baby developing kernicterus which can cause brain damage or even death. 13
The group recognised that all of the evidence available for this topic was of very low quality. There 15
were only 3 studies looking at the use of anti-D rhesus prophylaxis which reported comparative data. 16
None of the studies was very large and it is unlikely that any were sufficiently powered to detect a 17
statistically significant difference. The group had hoped to see evidence both in women with a 18
threatened miscarriage and in women with ectopic pregnancy. However, none was available which 19
met the inclusion criteria. All of the studies reported data in women with either a miscarriage (the vast 20
majority of which were managed with surgery) or a therapeutic abortion. There was no evidence for 21
women undergoing medical management of miscarriage, and a very small proportion of women had a 22
complete miscarriage without intervention. 23
For the question of the appropriate dose, again, there were only three comparative studies. Whilst 24
one was relatively large, all were of very low quality. 25
Given the paucity of available evidence, the GDGs recommendations were mainly developed through 26
their own clinical experience and that of the clinical adviser for this topic. 27
Trade off between net health benefits and resource use 28
Although the quality of the evidence was very low for the descriptive studies, the group felt that, taken 29
as a whole, there was evidence of a risk of sensitisation for women if they did not receive anti-D 30
rhesus prophylaxis following a first trimester miscarriage or therapeutic abortion. The group 31
recognised that the comparative studies did not show a statistically significant difference in the rate of 32
sensitisation between women who did and did not receive anti-D rhesus prophylaxis. However, the 33
group believed that the small size of the studies meant that this was unlikely to be a true finding of no 34
effect. 35
The group was informed that the chance of sensitisation increases when there is a greater likelihood 36
of mixing between the maternal and fetal blood. As a result, there is an increased risk of sensitisation 37
when treating a miscarriage or ectopic pregnancy surgically. 38
Given the lack of evidence, the group did not feel it appropriate to recommend that women with a 39
miscarriage or ectopic pregnancy that resolves spontaneously, without intervention, routinely receive 40
anti-D rhesus prophylaxis. However, recognising the increased risk of sensitisation to women 41
undergoing a surgical intervention, they felt it appropriate to recommend that these women should 42
receive prophylaxis. 43
The group considered the population of women who will receive medical management for their 44
miscarriage or ectopic pregnancy. From their clinical experience and understanding, the effect of 45
misoprostol is to cause the body to mimic the physiological changes that occur during a 46
spontaneously completing miscarriage. Similarly, they felt that the risk of significant maternal and fetal 47
blood mixing during methotrexate treatment for an ectopic pregnancy was likely to be low. Given this, 48
they did not believe that it would lead to an increased risk of sensitisation, and thus agreed that 49
women receiving medical management for either miscarriage or ectopic pregnancy should not be 50
offered prophylaxis. 51
199
The evidence available for the review of the appropriate dose suggested that a 50 microgram dose 1
(250 international units) of prophylaxis was as effective as a larger dose. Given this, and the fact that 2
the 50 microgram dose is cheaper, the GDG agreed that this is the dose which should be provided. 3
The group noted from the evidence that some of the studies reported the use of a Kleihauer test in 5
which the maternal blood is stained to detect the presence of cells containing fetal haemoglobin and 6
the number of these cells is then manually counted. The group received expert advice that the 7
accuracy of the test decreases at low levels of fetal haemoglobin (less than 1 cell in every 10,000). 8
The group noted that at the time of gestation covered in the guideline, the levels of fetal haemoglobin 9
were likely to be very low, and thus the Kleihauer test was unlikely to give an accurate result. This 10
was supported by the data reported in the included studies, which showed low correlation between 11
the Kleihauer test results and risk of sensitisation. Given this, the group agreed that this specific 12
diagnostic test should not be used for this group of women. 13
Recommendations 14
86 Offer anti-D rhesus prophylaxis at a dose of 250 IU (50 micrograms) to all rhesus
negative women who have a surgical procedure to manage an ectopic pregnancy or
a miscarriage.
87 Do not offer anti-D rhesus prophylaxis to women who:
receive solely medical management for an ectopic pregnancy or
miscarriage
have a threatened miscarriage
have a complete miscarriage
have a pregnancy of unknown location.
88 Do not use a Kleihauer test for quantifying feto-maternal haemorrhage.
15
16
Number Research recommendations
RR 9 Does the administration of anti-D rhesus prophylaxis following pain and bleeding in
early pregnancy improve outcomes?
17
18
19
20
21
22
23
24
25
26

10 Health Economics 1
10.1 Introduction 2
The aims of the health economic input to the guideline were to inform the guideline development 3
group (GDG) of potential economic issues relating to pain and bleeding in early pregnancy and to 4
ensure that its recommendations represented a cost-effective use of healthcare resources. Health 5
economic evaluations aim to integrate data on benefits or harms (ideally in terms of quality adjusted 6
life years [QALYs]) and costs of different care options. 7
The GDG prioritised the clinical questions where it was thought that economic considerations would 8
be particularly important in formulating recommendations. For this guideline the areas prioritised for 9
economic analysis were: 10
progesterone for threatened miscarriage (see section 7.2 for summary and section 10.2 11
for full details) 12
management of miscarriage (see section 7.3 for summary and section 10.3 for full 13
details) 14
management of ectopic pregnancy (see section 8.2 for summary and section 10.4 for 15
full details). 16
Introduction 18
It has been suggested that a lack of progesterone may be the cause of miscarriage. If it was found 19
that progesterone/progestogen supplementation was effective in preventing miscarriage then given 20
the low cost of this intervention it is also likely to be cost-effective. 21
Review of the published health economic evidence 22
A health economics search of the literature for progesterone for threatened miscarriage identified five 23
articles. The abstracts of these articles were reviewed but none of them were analyses of women with 24
threatened miscarriage. Therefore a de novo health economic model was developed for the purposes 25
of this guideline. This model is described below. 26
Method 27
A simple decision analytic model was developed in Microsoft Excel to evaluate the cost- 28
effectiveness of progesterone/progestogen supplementation in pregnancies with threatened first 29
trimester miscarriage (presence of vaginal bleeding before 12+6 weeks gestation). A schematic of the 30
overall model structure is illustrated in Figure 10.1. 31
32
33
34
35
36
Health economics
201
Figure 10.1 Schematic of model decision tree 1
2
3
4
5
6
7
8
9
A number of sensitivity analyses were undertaken to assess the importance of parameter uncertainty 10
within the model. 11
Model probabilities 12
Clinical data was taken from the clinical review undertaken for this guideline. That review considered 13
a number of outcomes: 14
Term birth 15
Pre-term birth 16
Miscarriage 17
Pregnancy at 20 weeks 18
Placental abruption 19
Hypertensive disorders in pregnancy 20
Gestational diabetes 21
Intrahepatic cholestasis of pregnancy 22
Our model confined itself to miscarriage, the primary focus of the intervention. The model data on 23
miscarriage outcomes is shown in Table 10.1 and was taken from the meta analysis of four studies 24
(El-Zibdeh et al., 2009; Gerhard et al., 1987; Palagiano et al., 2004; Pandian., 2009) undertaken as 25
part of the clinical review undertaken from this guideline. 26
Table 10.1 Model Probabilities 27
28
29
30
31
32
Item Value Source Notes
Miscarriage rate with
progesterone
13.8% El-Zibdeh et al., 2009; Gerhard et
al., 1987; Palagiano et al., 2004;
Pandian., 2009
Guideline meta-analysis
Miscarriage rate no treatment 25.9% El-Zibdeh et al., 2009; Gerhard et
al., 1987; Palagiano et al., 2004;
Pandian., 2009
202
Costs and resource use 1
This analysis was undertaken from the perspective of the NHS and personal social services which is 2
in accordance with NICE guidelines methodology (NICE, 2009). Treatment alternatives were 3
compared using standard methods of incremental analysis and costs are based on 2010/11 prices. 4
Discounting was not needed as all model costs fall within a year of the commencement of the 5
intervention. 6
Table 10.2 shows the cost inputs used in this model. 7
Table 10.2 Model costs 8
Miscarriage 424 NHS Reference Costs
2010-11
Currency Code: MB08Z. Threatened or
spontaneous miscarriage
Non-elective (short-stay)
Progesterone/progestogen 2.10 Estimate Based on a tablet cost of 0.10 and one
tablet/day for 21 days
Health economic inputs 9
There is considerable uncertainty about the QALY loss from miscarriage, and none of the studies 10
identified by the search were informative for this parameter. Therefore, the model allows what-if 11
sensitivity analysis to assess the extent to which assumptions with respect to this parameter may 12
affect model conclusions. When generating an incremental cost per QALY it is necessary to have a 13
decision rule regarding the willingness to pay for a QALY to determine whether the benefits are being 14
obtained at an acceptable opportunity cost (the other NHS services that might have to be foregone if 15
the intervention is cost increasing). The base case values for QALY loss associated with miscarriage 16
and willingness to pay for a QALY are shown in Table 10.3. 17
Table 10.3 Health economic inputs 18
QALY loss from miscarriage 0.1 Illustrative Can be varied as part of a what-if
sensitivity analysis
Willingness to pay for a QALY 20,000 NICE (2009) Advisory cost-effectiveness
willingness to pay threshold
Probabilistic Sensitivity Analysis 19
Probabilistic sensitivity analysis was undertaken using Monte Carlo simulation. The model was run or 20
simulated a total of 10000 times. In each simulation the value of probabilistic model parameters, 21
shown in Table 10.4 and Table 10.5 below, are sampled from a probability distribution which reflects 22
sampling uncertainty in the data. 23
Table 10.4 Parameters for probabilistic sensitivity analysis (model probabilities) 24
Item Alpha Beta Distribution
Miscarriage rate with progesterone 31 193 Beta
Miscarriage rate no treatment 51 146 Beta
25
Health economics
203
Table 10.5 Parameters for probabilistic sensitivity analysis (model costs) 1
Item Mean SE Distribution
Miscarriage 424 15.42 Normal
Results 2
The base-case results are shown in Table 10.6. This showed progesterone to dominate no treatment, 3
producing cost savings and QALY gains as a result of averted miscarriages. 4
Table 10.6 Results for base-case analysis 5
6
Sensitivity Analysis 7
Probabilistic sensitivity analysis 8
The results of a probabilistic sensitivity analysis based on 10000 simulations are plotted in Figure 9
10.2. Progesterone treatment was cost-effective in 9993 of these simulations (99.93%). 10
Figure 10.2 Probabilistic sensitivity analysis result of incremental costs and QALYs of progesterone treatment 11
(n = 10,000) 12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
Two-way sensitivity analysis varying the QALY loss associated with miscarriage and the cost 30
of progesterone treatment 31
In this sensitivity analysis the cost of treatment is varied between 0-250 (the latter being much 32
higher than that used in the base case analysis). The QALY loss from miscarriage is varied between 33
0 - 0.1. In Figure 10.3. the results are plotted on a grid which indicates the cost-effectiveness 34
threshold across different values of these inputs. 35
Treatment Incremental cost Incremental QALY
Progesterone -49 0.012
204
Figure 10.3 Cost-effectiveness threshold for various QALY and treatment combinations holding other model 1
inputs constant at their base case value 2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
Two-way sensitivity analysis varying the QALY loss associated with miscarriage and the 17
willingness to pay for a QALY 18
With other variables held at their base case values the models conclusions are unaffected by 19
variation in QALY loss associated with miscarriage and the willingness to pay for a QALY. This is 20
because the cost saving from averted miscarriage more than offsets the treatment costs meaning that 21
the intervention is always cheaper. Therefore, for illustrative purposes in this example, a treatment 22
cost of 100 is assumed perhaps because medical therapy requires increased patient monitoring. The 23
results of this sensitivity analysis are shown in Figure 10.4 24
Figure 10.4 Cost-effectiveness threshold for various QALY and willingness to pay combinations for a treatment 25
cost of 100 and holding other model inputs constant at their base case value 26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
Health economics
205
Probabilistic sensitivity analysis but assuming a treatment cost of 200 1
In this analysis the probabilistic sensitivity analysis is re-run to capture parameter uncertainty relating 2
to treatment effectiveness and miscarriage costs. However, it tests the implications of assuming a 3
treatment cost of 200, almost one hundred fold more than in the base case analysis. The plot of the 4
10000 simulations is shown in Figure 10.5. At a willingness to pay of 20,000 per QALY, 5
progesterone was cost-effective in 94.5% of the simulations. A cost-effectiveness acceptability curve 6
(CEAC) is displayed in Figure 10.6. This shows the probability of either alternative being cost-effective 7
at different willingness to pay values. In this example, with a treatment cost of 200 and a QALY loss 8
of 0.1 from miscarriage, the CEAC suggests that progesterone is likely to be the most cost- 9
effectiveness treatment providing the willingness to pay for a QALY exceeds 13,000. At a willingness 10
to pay of 20,000 per QALY progesterone has an 83% chance of being cost-effective. 11
Figure 10.5 Probabilistic sensitivity analysis result of incremental costs and QALYs of progesterone treatment 12
(n = 10,000) with a treatment cost of 200 13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
206
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
100%
P
r
o
b
a
b
i
l
i
t
y

c
o
s
t
-
e
f
f
e
c
t
i
v
e
Willingness to pay
Progesterone
No treatment
Figure 10.6 Cost-effectiveness acceptability curve showing the probability that progesterone is cost-effective 1
treatment for threatened miscarriage at different willingness to pay for a QALY and with a treatment cost of 200 2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
Discussion 18
This analysis strongly suggests that progesterone is a cost-effective treatment for threatened 19
miscarriage. In the base case analysis progesterone dominates no treatment, being cheaper and 20
having a lower miscarriage loss which generates a QALY gain. This result is driven by the very cheap 21
cost of the intervention and the large 12.1 percentage point reduction in the absolute risk of 22
miscarriage with progesterone. Given the cost of miscarriage this gives an expected saving of 51.30 23
per woman which is well in excess of the cost of treatment. Providing progesterone treatment costs 24
less than 51.30 it will be cost saving with the point estimates of treatment effect size. Progesterone 25
is very cheap and unless medical therapy involves consumption of other NHS resources other than 26
the drug itself then the cost of treatment is unlikely to exceed the saving from averted miscarriage. 27
The fact that there is such a large saving associated with miscarriage explains the inverse relationship 28
between incremental cost and incremental QALY seen in Figures 10.2 and 10.5. High QALY gains 29
are associated with a large reduction in miscarriage which produces a concomitantly large reduction 30
in net costs (costs of treatment less downstream savings from reductions in miscarriage). In both 31
Figures 10.2 and 10.5 an overwhelming number of the simulations lie to the right of the vertical axis 32
and their relative frequency gives an estimate of the probability that treatment is effective given the 33
results reported in the meta-analysis. Where the cost of treatment is small relative to the cost saving 34
from averted miscarriage, as in the base-case analysis, then an overwhelming majority of the 35
simulations lie in the south-east quadrant of the cost-effectiveness plane, signifying dominance 36
(Figure 10.2). A very small number of simulations do lie in the north-east quadrant where cost- 37
effectiveness depends on the willingness to pay for a QALY. In the base case that occurs when the 38
sampled reduction in miscarriage is so small that downstream savings do not offset the small 39
treatment cost. In Figure 10.5 the simulations have been shifted vertically upwards by the 200 40
increase in treatment costs. In none of the 10,000 simulations does progesterone now produce cost 41
savings with most simulations concentrated in the north-east quadrant of the cost-effectiveness plane. 42
Whether progesterone would hypothetically still be considered cost-effective depends on the 43
willingness to pay for a QALY. As the CEAC in Figure 10.6 shows, 83% of these simulations would 44
still be considered to represent a cost-effective trade off of benefits for increased costs even at a 45
treatment cost way in excess of what would be realistic. This does of course also depend on the 46
QALY gain attributable to an averted QALY. 47
Health economics
207
Considerable uncertainty surrounds the QALY loss attributable to miscarriage. Some short-term 1
effects on the woman seem inevitable but longer term loss in health related quality of life may depend 2
on whether the woman goes on to have subsequent successful pregnancy outcomes. However, 3
where treatment is very cheap and there are large savings from successful treatment, as in the base 4
case analysis, then uncertainties about the QALY loss from miscarriage are unimportant because the 5
cost-effectiveness of treatment is derived from the intervention being cost saving even in the absence 6
of any QALY benefit. The sensitivity analysis results shown in Figure 10.3 suggest that quantifying the 7
QALY loss from miscarriage is only important for treatment costs of 50 and above. Similarly, as 8
shown in Figure 9.4 the willingness to pay for a QALY only becomes important when the intervention 9
is cost increasing overall. 10
The model has a number of limitations. The treatment effect size is clearly an important driver of the 11
cost-effectiveness conclusions and therefore this data in the model reflects the quality and limitations 12
of the studies on which it was based. Nevertheless, unless there are known biases in these studies in 13
favour of progesterone this represents the best available evidence to assess cost-effectiveness. 14
As noted earlier, the clinical review for this guideline considered several outcomes but this model 15
focused solely on miscarriage. Modelling always involves some simplification of the real world and the 16
art of modelling often involves knowing when simplification is reasonable. Some of the other 17
outcomes assessed in the clinical review term birth, pre-term birth, pregnancy rate at 20 weeks 18
can almost be seen as the other side of the miscarriage coin. If miscarriage occurs then there is a 19
corresponding reduction in those outcomes. As a result of the strong interdependence of these 20
outcomes with miscarriage and the added complexity of modelling these outcomes were not included 21
in the analysis. Whilst not always statistically significant at the 5% level the point estimates for these 22
outcomes were generally in favour of progesterone, as would be expected if progesterone reduced 23
miscarriage, and therefore their omission if anything biases the model against progesterone. The 24
evidence for hypertensive disorders gave a point estimate for the relative risk of 1. Clearly there is 25
uncertainty around this relative risk and it was thought that the inclusion of hypertensive disorders 26
would introduce considerable noise which might mask out the effects of miscarriage especially given 27
its relative high prevalence. Furthermore, the point estimate of risk for hypertensive disorders ever so 28
slightly favoured progesterone (10.4% v 11.0%) and therefore its omission does not introduce bias in 29
favour of progesterone into the model. 30
Placental abruption is a relatively rare event and the evidence indicated large confidence intervals 31
around the point estimate of relative risk. Although, the point estimate of relative risk favoured no 32
treatment this was a long way from achieving statistical significance (RR 95% CI: 0.53 to 3.54). 33
Furthermore, the risk for both treatment and no treatment was lower than the 1% risk cited for 34
pregnancy as a whole (http://emedicine.medscape.com/article/252810-overview#a0199 [accessed 35
January 2012]) and therefore it was decided that there was not good evidence that treatment 36
influenced placental abruption. Pain was not included as it contributes directly to health related quality 37
of life where the preferred measure is the QALY. The review found pain to be significantly lower in 38
women treated with progesterone which means that progesterone may produce QALY gains over and 39
above those relating to actual miscarriage. 40
The omission of gestational diabetes and intrahepatic cholestasis of pregnancy is more contentious. 41
Neither of the point estimates of relative risk for these outcomes, which both favoured no treatment, 42
achieved statistical significance at the 5% level. However, the relative risk for gestational diabetes 43
does approach statistical significance (RR 95% CI: 0.94 to 1.76), so the observed difference may not 44
be down to chance. The point estimates suggested that progesterone carried a number needed to 45
harm of 63 for gestational diabetes. 46
It has been estimated in a paper on the cost-utility of screening for gestational diabetes that the QALY 47
loss from a serious perinatal complication, an adverse outcome of gestational diabetes, is 2.1 QALYs 48
(Round et al., 2011). That same paper estimated that the risk of a serious perinatal outcome in 49
women with treated gestational diabetes was 0.017. If we accept that progesterone leads to a 1.6 50
percentage point increase in risk of gestational diabetes that translates to a 0.000576 average QALY 51
loss of women treated with progesterone for threatened miscarriage due to gestational diabetes
. 52
Using point estimates of risk, the QALY loss due to gestational diabetes would only outweigh the 53

Expected QALY loss due to gestational diabetes = 2.1 x 0.017 x 0.016 = 0.000576
208
QALY gain from averted miscarriage if the loss due to miscarriage was less than 0.0048 QALYs
*****
. 1
That same study assumes a treatment cost of gestational diabetes of 162. In addition there are also 2
costs of serious perinatal complications estimated at 1,184 to consider. However, in women with 3
treated gestational diabetes the risk of these complications is 0.017 and therefore the expected 4
serious perinatal complication cost per woman with gestational diabetes is 20, making a total cost of 5
gestational diabetes of 182 per woman. Accepting a 1.6 percentage increase in gestational diabetes 6
arising from progesterone for threatened miscarriage would mean that there would be an expected 7
downstream cost of approximately 3 due to gestational diabetes. This is relatively insignificant 8
compared to the expected 50 saving due to averted miscarriage and therefore because of the small 9
expected QALY loss and cost arising from gestational diabetes it is unlikely its omission from this 10
model would change the conclusion. 11
Conclusion 12
The model suggests that, not only is progesterone likely to be cost-effective in the treatment of 13
threatened miscarriage, but that it is also likely to be cost saving because the reduction in miscarriage 14
more than offsets the costs of treatment. Probabilistic sensitivity analysis suggested that this 15
conclusion was robust with respect to parameter uncertainty pertaining to treatment effect size and 16
miscarriage cost. Other sensitivity analysis suggested much higher treatment costs than assumed in 17
the base case analysis would be needed in order for the cost-effectiveness of progesterone to 18
become more equivocal. 19
10.3 Management of miscarriage 20
Introduction 21
As a result of high rates of gynaecological infection and resulting mortality, surgical evacuation 22
developed as the routine treatment for first trimester miscarriage (Ankum et al., 2001). However, there 23
has been an increasing willingness to consider alternatives such as expectant and medical 24
management in recent years (Ankum et al., 2001). Evidence based medicine requires that 25
consideration be given to the effectiveness of the different treatment alternatives but in any resource 26
constrained health care system, maximisation of health gain requires that consideration be given 27
additionally to the cost-effectiveness of the alternatives. Womens choice is an important determinant 28
of treatment within the NHS, but recommendations in this guideline are also informed by cost- 29
effectiveness as is required by the NICE guidelines manual (NICE, 2009). 30
Overview of the economic evidence 31
A total of 139 articles were identified by the search strategy. Based on the abstracts a total of six full 32
papers were obtained, of which five were included in this review (Graziosi et al., 2005c; Hughes et al., 33
1996; Niinimaki et al., 2009; Rocconi et al., 2005; Petrou et al., 2006). In addition, a further paper 34
(You & Chung, 2005) was identified from the references from one of the included studies (Niinimaki et 35
al., 2009) and was also included within the review. 36
One UK study compared the costs to the NHS of medical and surgical management for miscarriage 37
(Hughes et al., 1996). The analysis was conducted alongside a partially randomised trial where 38
enrolled women expressing a preference were given their treatment of choice (n=237) with the 39
remainder randomised to either surgical or medical evaluation (n=200). The precise form of medical 40
treatment varied according to the type of miscarriage. Mifepristone (200mg) followed by three 41
sequential oral doses of misoprostol 36-48 hours later was used for cases of missed miscarriage and 42
anembryonic pregnancy. Two sequential oral doses of misoprostol were used for women with an 43
incomplete miscarriage confirmed by clinical examination and ultrasound assessment. The study 44
found that the mean cost of medical management was 346 (95% CI 333 to 361) and the mean 45
cost of surgical management was 397 (95% CI 383 to 411). The finding that medical management 46
was cheaper was statistically significant (MD -51, p < 0.001). 47

***** Expected QALY loss due to miscarriage = 0.121 x QALY loss from miscarriage
0.121 x 0.0048 0.000576
Health economics
209
A US paper reported a decision analytic model used to compare (1) observation, (2) medical 1
management, (3) manual vacuum aspiration, (4) dilation and curettage (Rocconi et al., 2005). The 2
models end-point was treatment failure or cure and therefore combined management strategies were 3
not considered. Model inputs for clinical data were derived from a review of the published literature. 4
When literature derived estimates were not available local sources were consulted. Medical 5
management was vaginal misoprostol. In the observation strategy patients were seen weekly and 6
followed for a maximum period of 28 days. Costs were based on the perspective of a third-party payer 7
and were expressed in 2003 USD. Sources for cost data were the University of Alabama at 8
Birmingham reimbursement schedules, average wholesale drug costs and using 60% of the charge if 9
a cost wasnt available for a procedure. The results were as shown in Table 10.7.
The authors 10
conclude that, for their baseline analysis, MVA is the most cost-effective based on its ICER. However, 11
it doesnt automatically follow that the strategy with the lowest ICER is cost-effective, but rather the 12
willingness to pay threshold of the decision maker which determines whether the more efficacious 13
strategy is preferred given the increased cost (Drummond et al., 1997). 14
Table 10.7 Results from Roccini et al. (2005) 15
Strategy Cost per
patient
Cure rate Incremental cost-
effective ratio
MVA $754 95% $793 per cure
Observation $804 88% Dominated
Medical management $926 81% Dominated
Dilation & curettage $2,310 99% $38,900 per cure
16
A Dutch study undertook a cost analysis alongside a randomised trial of misoprostol versus curettage 17
following expectant management (Graziosi et al., 2005c). The authors stated that, a full cost- 18
effectiveness analysis was not undertaken because the trial found that both strategies were equally 19
effective. The analysis was undertaken on an intention-to-treat basis. The study enrolled 154 women 20
with early pregnancy failure who had been managed expectantly for a minimum of one week. 21
Resource use was prospectively collected for a period of up to six weeks. Curettage had to be 22
performed within a week of randomisation. Medical treatment was 4 tablets of 200 micrograms of 23
vaginal misoprostol given in an outpatient setting and repeated 24 hours later if necessary. In the 24
event of failed medical treatment, curettage was planned. Costs were based on a societal 25
perspective. Individual case record data was used to measure resource utilisation. Productivity losses 26
were estimated by self-completed patient questionnaires. Costs relating to trial protocol were 27
excluded from the analysis. The paper reported that the mean direct cost (defined as costs of medical 28
intervention) per patient was 433 for medical management compared to 683 for surgical 29
management (MD 250, 95% CI 184 to 316, p < 0.001). The difference in indirect costs (defined as 30
the costs of productivity losses) was not statistically significant at the 5% level (MD 58 95% CI -61 31
to 179, p = 0.51). A sensitivity analysis suggested that as long as misoprostol achieved a complete 32
evacuation rate of 31% or more (53% in this study) then medical management was the less costly 33
strategy. 34
A paper reported a decision analytic cost analysis comparing expectant, medical (misoprostol) and 35
surgical management for uncomplicated miscarriage in the first trimester (You & Chung, 2005). The 36
decision model used clinical inputs sourced from the literature and costs were assessed from the 37
perspective of a public health care provider in Hong Kong. In the surgical strategy women received 38
dilation and curettage within one day of inpatient care and one follow-up appointment in an outpatient 39
clinic. The model allowed for the possibility of surgical complications including the need for further 40
surgical evacuation, although very minor events with little or no resource consequences were not 41
incorporated. In the expectant management and surgical strategies, patient choice was allowed, in 42
that surgery could be chosen as an alternative if the other strategies were deemed unacceptable. In 43

This presentation of the results is slightly different to that which appears in the paper
210
the patients who accepted medical or expectant management it was assumed that patients would be 1
managed within an outpatient setting with two follow up visits within 2 weeks. The outcomes of these 2
strategies included complications necessitating surgical evacuation. The probabilities for the different 3
branches of the tree emanating from chance nodes were estimated from the literature. A cost per 4
patient was estimated which included primary treatment in the event of a spontaneous miscarriage, 5
the cost of surgical intervention when the primary treatment failed and the cost of other complications. 6
The base case analysis found that misoprostol was the cheapest strategy with a cost of USD $1,000. 7
The expectant and surgical strategy cost USD $1172 and USD $2007 per patient respectively. 8
A UK economic evaluation, taking a societal perspective and conducted alongside a randomised 9
controlled trial (MIST trial), compared the cost-effectiveness of expectant, medical and surgical in an 10
early pregnancy assessment unit setting (Petrou et al., 2006). Women randomised to expectant 11
management were allowed to go home with no intervention. Women allocated to medical 12
management with incomplete miscarriages were admitted to hospital and given a single dose of 800 13
micrograms of misoprostol. Medical management for women with incomplete miscarriages were pre- 14
treated with a single oral dose of 200 mg of mifepristone prior to admission 24-48 hours later for a 15
single vaginal dose of 800 micrograms of misoprostol. Surgical management involved admission of 16
women for the surgical evacuation of retained products. Gynaecological infection was the primary 17
outcome of the trial, covering the first 14 days and the first 8 weeks after entry into the trial. Trial data 18
collection forms were used to collect details of resource use by each study participant. A postal 19
questionnaire was used to collect data on community and social care contacts made by women as 20
well as childcare support and travel distance to healthcare providers that were attributable to their 21
miscarriage treatment. Unit costs, obtained from a variety of sources, were then used to cost the 22
resource inputs using 2001-02 prices. In total, resource use was collected from 1,200 women. The 23
mean cost of hospital care was 733 (SD = 845) in the expectant management group, 1,008 (SD = 24
644) in the medical management group and 1,210 (SD = 428) in the group allocated to surgical 25
treatment. The authors used non parametric bootstrapping to determine confidence intervals for the 26
mean difference in costs as the cost data was skewed. They reported that the mean cost of surgical 27
treatment was 200 (95% CI 122 to 278) more expensive than medical treatment and that the 28
mean cost of medical management was 273 (95% 160 to 376) more expensive than expectant 29
management. This finding wasnt affected when broader societal costs were included in the analysis. 30
Both expectant and medical management led to a non significant reduction in gynaecological 31
infections when compared with surgical management, a result suggesting that surgical management 32
is dominated (more expensive and less effective) than the alternatives. Medical management had the 33
least number of gynaecological infections and the point estimate of the incremental cost-effectiveness 34
of medical management relative to expectant management was approximately 63,000 per 35
gynaecological infection avoided. Non-parametric bootstrapping simulation was used to generate 36
1000 replications of the ICER and derive a concomitant cost-effectiveness acceptability curve. With a 37
willingness to pay of 10,000 to avoid a gynaecological infection there was a 97.8% probability that 38
expectant management was the most cost-effective strategy. Expectant management was found to 39
have the highest probability of being cost-effective up to a willingness to pay threshold of 70,000 per 40
gynaecological infection avoided. The authors note that the cost-effectiveness of expectant and 41
medical management may have been over-estimated if the higher rate of emergency consultations 42
and admissions observed in those groups led to an increased rate of health care utilisation beyond 43
the study follow-up period (8 weeks post randomisation). The authors also state that a preference 44
based measure of health outcome (e.g. QALY) would have been better for comparative purposes. 45
They acknowledge that it might have been possible to have mapped the trial outcomes onto a multi- 46
attribute utility measure, such as the SF-6D and then extrapolate to the QALYs attributable to each 47
management alternative. They suggest that doing so would have added little to the results of the 48
evaluation as none of the eight subscales of the UK Short Form-36 health-related quality of life 49
measure showed any significant difference within the trial with respect to the type of management. 50
A Finnish cost-effectiveness analysis (Niinimaki et al., 2009) evaluated the cost-effectiveness of 51
medical and surgical management of miscarriage using data from a previously published randomised 52
study (Niinimaki et al., 2006). In that study 49 patients were allocated to medical treatment and 49 53
patients were allocated to surgical treatment. In this study the cost of both treatment alternatives were 54
compared, both in terms of the initial allocated procedure but also costs arising from treatment 55
complications. The analysis was restricted to the 46 patients in each arm who received their allocated 56
treatment in the trial. Resource inputs (staffing, drugs, tests etc.) were estimated for surgical and 57
Health economics
211
medical treatment with an experienced health care professional estimating the staff time involved. 1
Taking the perspective of the health care provider, costs were calculated at 2007 prices using 2
institutional prices and charges to the county for outpatient and inpatient visits and procedures. For 3
each study participant the hospital files were used to determine the number, type and length of the 4
hospital visits which were then costed using these institutional prices. The authors report that the cost 5
for patients allocated to medical treatment was 455 against 489 for patients allocated to surgical 6
treatment. No confidence intervals or standard errors were reported for these point estimates or for 7
the 34 difference they report
. An incremental cost-effectiveness ratio (ICER) was then calculated 8

using patient satisfaction, pain and successful treatment (defined as no requirement for subsequent 9
intervention) as alternative measures of effect. The results presented by the study authors are shown 10
in Table 10.8 below: 11
Table 10.8 Incremental cost effectiveness ratios (Niinimaki et al., 2009) 12
Measure of effect Incremental cost of
surgery (n=46)
Incremental gain with
surgery (patients)
Incremental cost-effectiveness
ratio
Pain 1,688 12 141 per patient avoiding pain
Satisfaction 1,688 7 241 per satisfied patient
Successful treatment 1,688 5 338 per patient successfully
treated
13
However, for successful treatment the denominator is based on the patients allocated to each 14
treatment arm (n=49) and clearly an incremental cost based on treating 46 patients will be an 15
underestimate of the total incremental cost. Furthermore, for the satisfaction measure the 16
denominator was not identical between the comparators as well as being different from the patients 17
on which the total cost was based. In Table 10.9 below the ICERs are recalculated using data from 18
the paper but rather calculating the incremental cost per patient and the incremental gain as the 19
difference in the event rate. 20
Table 10.9 Recalculated incremental cost effectiveness ratios (Niinimaki et al., 2009) 21
Measure of effect Incremental cost per
patient
Incremental gain with
surgery (event rate)
Incremental cost-effectiveness
ratio
Pain 37 0.26 142 per patient avoiding pain
Satisfaction 37 0.12 308 per satisfied patient
Successful treatment 37 0.10 370 per patient successfully
treated
22
It should also be noted that, whilst there is considerable overlap, the patients on which the costs are 23
based are not completely the same as the patients for which pain, satisfaction and successful 24
treatment outcomes are based. The authors do not consider the willingness to pay for the benefits 25
attributable to surgery and therefore do not answer the question as to whether the additional costs of 26
surgical treatment represent a cost-effective use of resources. The authors conclude that it would not 27
be ethical to only offer medical treatment based on its lower cost. 28

For medical treatment they report a total cost of 22,282 which for 46 patients would be a cost per patient of 484. For
surgical patients they report a total cost of 23,970 which for 46 patients would be a cost per patient of 521 (difference 37)
212
Discussion 1
The main focus of this guideline was to compare the cost-effectiveness of expectant, medical and 2
surgical management for first trimester miscarriage. Our quality assessment suggested that study 3
quality varied quite widely across our six included studies. The evidence from these studies and some 4
of its limitations are discussed below. 5
All the studies included in this review found that surgical management was the most expensive option. 6
In the three studies that considered all three treatment alternatives (Petrou et al., 2006; Rocconi et al., 7
2005; You & Chung, 2005), two found expectant management to be the cheapest treatment option. 8
One study (You & Chung, 2005) reported that medical management was cheaper than expectant 9
management. However, the decision tree developed for this study allows for patient choice which is 10
not entirely consistent with a comparison of competing alternatives, a key tenet of economic 11
evaluation. In both medical and expectant strategies, patients have the option to reject that 12
management strategy of favour of surgery. The costs of the expectant and medical management 13
strategies therefore become a weighted average of two treatment alternatives with the weights 14
determined by the acceptance rate for the suggested management strategy. The acceptance rates 15
were 64% and 86% for expectant management and medical management respectively. Therefore, 16
their finding that medical management was the lowest cost treatment was an artefact of more of the 17
expectant management patients choosing surgery rather than medical management having 18
intrinsically lower costs. There is a sense that the alternatives are not competing if patients assigned 19
to one management strategy have the option to undergo one of the alternatives being evaluated. 20
Three of the studies were cost analyses or cost minimisation analyses. This study design is only 21
deemed appropriate in economic evaluation where the alternatives being evaluated are deemed to be 22
equivalent in terms of their benefits and harms. One of the cost studies (Graziosi et al., 2005c) 23
justified this approach on the basis that a recently performed RCT had found the effectiveness of 24
misoprostol and curettage to be equal (Graziosi et al., 2004). However, the justification for this 25
equivalence is based on a non-statistically significant difference at the 5% level, which is more a case 26
of no strong evidence of a difference rather than evidence of no difference. However, it is argued that 27
the non statistical significance rationale for cost minimisation analysis is inappropriate and that cost- 28
effectiveness analysis with probabilistic sensitivity analysis, utilising the probability distribution of 29
treatment effect, is the preferred approach (Briggs & OBrien, 2001). 30
A better rationale for a cost minimisation approach in this case is that a miscarriage is a time limited 31
event and therefore all women are ultimately cured. Furthermore, the morbidity associated with 32
miscarriage is generally only a short term phenomenon which means that the quality of life gains from 33
earlier resolution are relatively small. Where first-line treatment doesnt achieve a cure further 34
treatment options are available. One study uses cost per cure as its measure of cost-effectiveness, 35
with cure being defined as a negative pregnancy test at 28 days after commencement of treatment 36
(Roccini et al, 2005). As a result this study finds different ICERs for the different strategies, although 37
these are not presented in the paper. However, even correctly calculated, the meaningfulness of 38
these ICERs is questionable. The decision model does not reflect actual clinical practice which may 39
involve the offer of alternative treatment in the event of treatment failure, which of course has 40
additional resource implications. The value of the measure of effect of cure at 28 days may also be 41
questioned given the time limited nature of the condition. Even if we accept this as a proxy for earlier 42
resolution of symptoms/morbidity the authors offer no guide to the decision makers as to how this 43
benefit is to be valued. 44
Three of the studies were economic evaluations conducted alongside randomised controlled trials 45
(Hughes et al., 1996; Graziosi et al., 2005c, Petrou et al., 2006). However, one of the studies (Hughes 46
et al., 1996) used a partial randomisation approach which allowed women with a strong preference for 47
a particular treatment alternative to be included, although that can clearly dilute the benefits of 48
randomisation if this introduces systematic differences between the treatments in terms of patient 49
characteristics. It may also not reflect treatment options if decisions on treatment are to be determined 50
according to their cost-effectiveness. 51
Another evaluation was based on a previously published trial (Niinimaki et al., 2009). The trial on 52
which patient outcomes were derived had a small number of participants (n = 98). The trial found that 53
surgical management had fewer patients with pain, more satisfied patients and more patients with 54
successful treatment, defined as no requirement for further treatment. Thus it was possible from the 55
Health economics
213
data in the paper to calculate ICERs for all these different measures of effect for surgical 1
management and medical management. However, this information is of little use to decision makers 2
without some decision-rule as to what incremental benefit would be sufficient to justify the incremental 3
costs. Also, it can be argued that their definition of success is not the most appropriate as further 4
treatment, whilst having a resource implication, is likely to erode or eliminate any differences in health 5
related quality of life. Another major limitation with this study, especially given the small numbers of 6
patients on which it is based, is that no allowance is made for uncertainty. The results are only 7
provided as point estimates and no sensitivity analysis is undertaken. 8
Most of the included studies have important limitations in terms of their economic methods. However, 9
there was one very good quality economic evaluation (Petrou et al., 2006). This was based on the 10
collection of resource use data alongside a large randomised controlled trial (n = 1,200). Costs were 11
presented from a number of perspectives but they included the health service perspective and the 12
price year is clearly stated (2001-02). The costs of the alternative management strategies were 13
presented as mean values with standard deviations. Comparisons of the costs of different strategies 14
were reported as mean differences with confidence intervals. In addition to a cost comparison the 15
incremental cost-effectiveness was assessed using the cost per gynaecological infection prevented. 16
Uncertainty was taken into account using probabilistic sensitivity analysis by using nonparametric 17
bootstrapping to generate 1000 replications of each of the ICERs. This was made relevant to decision 18
making by the use of cost-effectiveness acceptability curves which found that expectant management 19
was likely to be the most cost-effective strategy up to a willingness to pay threshold of 70,000 per 20
gynaecological infection prevented. Using an advisory willingness to pay threshold of 20,000 per 21
QALY which is more consistent with NICE methods, expectant management would be most likely to 22
be cost-effective providing the gain from an averted gynaecological infection did not exceed 3.5 23
QALYs, as seems likely (3 QALYs is equivalent to an additional 3 years lived in perfect health). The 24
authors acknowledge that the fact that effectiveness has not been measured in a preference based 25
measure of health outcome (e.g. QALY) is a limitation but they do note that the MIST trial revealed no 26
significant differences in any of the eight sub-scales of the UK Short Form-36 health related quality of 27
life measure. This would suggest that the QALY differences between the alternative management 28
strategies are likely to be small and therefore unlikely to change a conclusion that expectant 29
management is likely to be the most cost-effective strategy. 30
Conclusion 31
The data from this review suggests that expectant management is the most cost-effective first line 32
treatment for the management of first trimester miscarriage. The GDG accepted that this evidence 33
was supported by one high quality published economic evaluation which was relevant to an NHS 34
context and which considered the relevant management alternatives (Petrou et al., 2006). Therefore, 35
they were prepared to make recommendations based on this evidence without the requirement for a 36
de Novo guideline model. 37
10.4 Management of ectopic pregnancy 38
Introduction 39
The incidence of ectopic pregnancy in the UK is approximately 1.1% (Lewis, 2007). It is important 40
because it can be fatal if left untreated. Technological advances mean that earlier diagnosis is now 41
possible which has helped extend the range of available treatments. However, uncertainty remains as 42
to whether treatment should be medical or surgical and, if the latter, whether surgery should be 43
conservative (salpingotomy) or radical (salpingectomy) and whether the surgical technique should be 44
laparotomic or laparoscopic. Clearly, within a context of trying to maximise health gain from finite 45
resources, the cost-effectiveness of these treatment alternatives also has to be considered. 46
Review of the published health economic evidence 47
A review of the literature was undertaken based on the three related clinical questions reviewed for 48
this guideline: 49
214
Medical compared with surgical management of ectopic pregnancy 1
A total of 33 articles were found using this search when applying a health economics filter. From 2
reading the abstracts of these papers, 12 full papers were obtained, of which eight were included in 3
this review. 4
A Canadian study (Yao et al., 1996) undertook a retrospective cost analysis of patients treated with 5
methotrexate and patients treated with laparoscopy. Treatment decisions were made by doctors with 6
no random allocation of patients. Therefore, it cannot be inferred that the patient groups are 7
comparable. Using direct medical costs they reported that the methotrexate group had a mean cost of 8
880 CAD (SE 160) compared to a mean cost of 1,840 CAD (SE 150) in the laparoscopic group, 9
with the difference statistically significant at the 5% level (p < 0.001). 10
A New Zealand paper (Sowter et al., 2001a) carried out a cost minimisation analysis alongside a 11
small randomised trial (n = 62). In the trial patients were randomised to either a single dose of 12
methotrexate or laparoscopic surgery. The study considered both direct and indirect costs 13
(productivity losses etc) but only the former is relevant using NICE methods (NICE, 2009). The 14
authors reported that the direct costs in the methotrexate group were 1,613 NZD (95% CI 1,116 NZD 15
to 2,061 NZD) lower than the direct costs in the laparoscopy group. 16
A decision analytic model (Morlock et al., 2000) compared the cost-effectiveness of intramuscular 17
methotrexate with laparoscopic salpingotomy for small unruptured ectopic pregnancy. Clinical model 18
inputs were based on a meta-analysis with costs based on local charge data at the authors US 19
hospital. With base case model values, laparoscopic treatment was more than 3,000 USD more 20
expensive than methotrexate per resolved ectopic pregnancy. The authors reported a number of 21
sensitivity analyses which continued to find methotrexate to be the cheaper treatment. 22
A French study (Lecuru et al., 2000) compared the direct costs of single dose methotrexate and 23
laparoscopy in the treatment of unruptured ectopic pregnancy. The study was prospective but women 24
having laparoscopic salpingectomy did so because they either refused methotrexate or because 25
methotrexate was contraindicated. Such studies are vulnerable to selection bias. Methotrexate was 26
found to be significantly cheaper than laparoscopic salpingectomy (1,145 Euros compared with 2,442 27
Euros, p = 0.006). The authors also noted that the saving with methotrexate was greatest for small 28
unruptured ectopic pregnancy because such cases require less hospitalisation and follow-up. 29
An earlier but similar French study (Robin et al., 1998) found that mean treatment costs with 30
methotrexate were 1,469 USD cheaper than with laparoscopic salpingotomy. 31
A US study (Alexander et al., 1996) compared the costs of methotrexate and laparoscopic 32
salpingotomy for women with a small unruptured ectopic pregnancy. The authors assumed that both 33
treatments would lead to a resolution of the ectopic pregnancy without maternal morbidity or long- 34
term morbidity, forming their justification for a cost minimisation approach. Clinical outcomes were 35
estimated from a review of the literature. The direct costs of each treatment alternative were 36
estimated using actual reimbursement rates of a third-party payer. The authors reported that 37
methotrexate was 2,536 USD less expensive in its best-case scenario and 1,124 less expensive in 38
its worst-case scenario. The authors additionally reported sensitivity analyses showing that 39
methotrexate remained less expensive across a wide range of probability and cost inputs. 40
A Dutch study (Mol et al., 1999), undertook an economic evaluation alongside a randomised control 41
study to compare systemic methotrexate with laparoscopic salpingotomy in haemodynamically stable 42
women with a confirmed unruptured tubal pregnancy. The authors justified a cost minimisation 43
approach by stating that the clinical outcomes of the trial were the same for the two treatment 44
alternatives. Using direct medical costs methotrexate was 769 USD (95% CI, 28 USD to 2384 USD) 45
more expensive than laparoscopic salpingotomy. 46
A US analysis (Creinin & Washington, 1993) compared the costs of surgery versus methotrexate for 47
the management of ectopic pregnancy. The direct medical costs of surgery were derived from billing 48
statements from all patients treated with ectopic pregnancy in 1991. They then estimated what 49
proportion of these patients could have been eligible for methotrexate using the following criteria: 50
haemodynamically stable, size of ectopic pregnancy smaller than 3.5 cm, the ectopic pregnancy was 51
located in the fallopian tube and not ruptured, no cardiac activity was present if an ultrasound was 52
performed before the procedure, no evidence of hepatic dysfunction or renal disease, a complete 53
Health economics
215
blood count did not reveal dyscrasia and no evidence of poor compliance. A cost for methotrexate 1
was then estimated using a published protocol for single dose methotrexate (Stovall et al., 1991). The 2
mean surgical cost of the 50 patients included in the analysis was 10,509 USD and it was estimated 3
that the cost of methotrexate in 1992 prices would be 1,495 USD. Using a methotrexate success rate 4
of 93% in the 30% of eligible patients the authors extrapolated that the use of methotrexate in these 5
patients could reduce the mean cost of ectopic pregnancy treatment to 7,951 USD per patient. 6
Laparotomy compared with laparoscopic techniques 7
The search strategy identified a total of 56 papers of which 10 were obtained in full. Five articles in 8
total are included in this review. 9
An Australian study (Lowe at al., 1998) retrospectively analysed 45 consecutive cases of surgically 10
treated ectopic pregnancy in order to compare laparoscopic and laparatomic techniques. Such study 11
designs are prone to selection bias. Laparoscopic treatment had a mean cost of 2,930 USD (95% CI, 12
2,458 USD to 3,402 USD) compared to 4,259 USD (95% CI, 3,666 USD to 4,852 USD). The 1,329 13
USD saving per patient with laparoscopy was statistically significant at the 5% level. 14
A Swedish cost-effectiveness analysis (Gray et al., 1995) compared laparoscopy versus laparotomy 15
for ectopic pregnancy alongside a randomised controlled trial. The authors reported that laparoscopy 16
dominated laparotomy, being equally successful and having lower costs. It was noted that the post- 17
operative inpatient stay for laparotomy was 2.9 days longer and this was an important driver of the 18
increased costs associated with that technique. 19
A Dutch study (Mol et al., 1997) compared the costs of laparoscopy and open surgery in women with 20
a tubal pregnancy. The analysis was undertaken on a group of consecutive patients having either 21
type of surgery. The authors note that treatment was successful in all 255 patients. This study found 22
that laparoscopic salpingectomy was cheaper than laparotomic salpingectomy (1,872 USD vs. 3,490 23
USD) and similarly that laparoscopic salpingotomy was also cheaper than the open surgery 24
alternative (2,125 USD vs. 3,420 USD). 25
A US paper (Foulk & Steiger, 1996) reported a retrospective cost analysis of the operative 26
management of ectopic pregnancy within the authors own hospital. The costs were: laparoscopy 27
alone 4,344 USD, laparoscopy converting to laparotomy 6,979 USD, laparotomy in haemodynamically 28
stable patients 5,333 USD and laparotomy in haemodynamically unstable patients 7,556 USD. The 29
authors note that the cost saving obtained with laparoscopy is reduced when including the 21% of 30
patients who convert to laparotomy from an intended laparoscopy. A threshold sensitivity analysis 31
suggested that the costs would be equivalent if the conversion rate reached 37%. In a similar vein it is 32
suggested that costs would be similar if the postoperative length of stay for laparotomy patients was 33
no more than two days. 34
A non-randomised prospective cohort study in the UK (Baumann et al., 1991) compared the costs of 35
laparotomy and laparoscopy management of ectopic pregnancy in haemodynamically stable women. 36
Costing was based on 20 randomly selected women from each treatment group and found that 37
laparoscopy produced an overall saving of 701.47 (p < 0.001) primarily as a consequence of a 38
shorter length of stay. 39
Salpingectomy compared with salpingotomy 40
The search strategy identified a total of 60 papers. However, the abstracts were reviewed and none 41
were relevant for this question. 42
Notwithstanding the findings of this review of the literature, much of which is quite dated and prone to 43
selection bias, the GDG felt it would be useful to additionally develop a new model for this guideline 44
reflecting the best available clinical evidence, based on the reviews undertaken for this guideline, and 45
relevant for an NHS context. This model is described below. The GDG thought that it was not useful 46
to consider laparotomic techniques, as this is rarely thought appropriate as a first-line treatment and 47
because the health economic evidence, limited as it is, suggests that laparoscopic techniques are 48
likely to be more cost-effective as well as preferable to the patient. 49
Method 50
A decision analytic model was developed in Microsoft Excel to compare the cost effectiveness of the 51
following treatment alternatives in women diagnosed with an ectopic pregnancy but not requiring 52
216
urgent surgical intervention as, by definition, medical intervention couldnt be considered as a 1
treatment alternative in such women. 2
Laparoscopic salpingotomy 3
Here patients are assigned to laparoscopic salpingotomy although it is assumed that a proportion of 4
these will convert to an open procedure. Such patients having an open salpingotomy will incur a 5
different procedure cost but it assumed that there are no costs incurred from conversion per se. The 6
surgery is assumed to have a complication and blood transfusion rate which differs according to 7
whether laparoscopic or open surgery is undertaken. It is assumed that the surgical complication and 8
blood transfusion rates are independent of success or failure (when further intervention is required 9
to treat the ectopic pregnancy). There is a risk of emergency admission in patients in whom the initial 10
procedure fails. Some women have salpingectomy as the second line treatment subsequent to failure 11
of the initial surgery. The remaining women in whom first line treatment fails receive methotrexate 12
second line. In patients failing with second line methotrexate, salpingectomy will be undertaken as a 13
third line treatment. Salpingectomy, whether offered as a second line or third line treatment, will 14
usually be performed laparoscopically but a proportion will convert to open salpingectomy. As for the 15
initial surgery, complication and blood transfusion rates will differ according to whether surgery is 16
laparoscopic or open. Second line or third line surgery is assumed to be 100% successful. There is 17
not assumed to be any further risk of emergency admission following second line treatment. 18
As with salpingotomy it is assumed that a proportion of laparoscopic salpingectomy patients will 20
convert to an open salpingectomy with a different procedure cost but no costs of conversion. Where 21
salpingectomy fails some patients will have a second line salpingectomy, which would normally be 22
laparoscopic but with some conversion to open surgery. Other patients failing with initial 23
salpingectomy receive methotrexate as the second line treatment. The assumptions, with respect to 24
surgical complications and blood transfusion, are the same as for salpingotomy. It is assumed that 25
there are no emergency admissions with this strategy. 26
Methotrexate 27
Patients are initially treated with methotrexate.In a proportion of cases, the treatment will fail in which 28
instance there is a risk of rupture which is assumed to lead to an emergency admission. A proportion 29
of patients who rupture will require a blood transfusion. There will also be some patients in whom first 30
line methotrexate fails who require an emergency admission without a rupture. It is assumed that a 31
proportion of those in whom methotrexate fails will have further methotrexate as a second line 32
treatment. The remaining patients in whom methotrexate fails first line will have laparoscopic 33
salpingectomy, with a proportion converting to open salpingectomy. Those in whom methotrexate fails 34
as second line will then have laparoscopic salpingectomy with some converting to open 35
salpingectomy as third line treatment. As with first line methotrexate fail, patients in whom 36
methotrexate fails second line have a risk of rupture, blood transfusion and emergency admission. Of 37
these a proportion will have an emergency admission as a result of rupture. Surgery carries the same 38
risk of complication and blood transfusion as in the previous strategies. 39
A schematic of the overall model structure is illustrated in Figure 10.7 with greater detail of the 40
individual treatment tree being shown in Figures 10.8 to 10.10 41
42
Health economics
217
Figure 10.7 Schematic of model decision tree

218
Figure 10.8 Salpingotomy decision tree 1
2
Figure 10.9 Salpingectomy decision tree 3
4
Health economics
219
Figure 10.10 Methotrexate decision tree 1
2
3
4
5
The structure of the decision tree was agreed by the GDG as being appropriate in an NHS context to 6
the population of interest, namely women with a diagnostically confirmed first ectopic pregnancy and 7
without emergency presentation. 8
The base case assessment of cost-effectiveness is undertaken using a cost minimisation approach, 9
where the optimal strategy is that which is the cheapest. Such an approach may be considered valid 10
when the effectiveness of all treatment alternatives is considered identical. Historically, this approach 11
has sometimes been used when the clinical data fails to reject a null hypothesis of no difference at the 12
5% level. However, the 5% cut-off for statistical significance is an arbitrary one and a failure to reject 13
the null hypothesis doesnt imply that treatment effectiveness is identical and the cost-minimisation 14
approach in such cases has been criticised (Briggs & OBrien, 2001). Nevertheless, the GDG 15
considered that a cost minimisation approach was reasonable in this context. The rationale was that 16
all women are ultimately cured and that this occurs within a fairly short time frame so that any QALY 17
differences between treatment alternatives would be small. 18
Probabilistic sensitivity analyses and other sensitivity analyses were undertaken to assess the 19
importance of parameter uncertainty within the model. A QALY loss variable for methotrexate 20
treatment was incorporated into the model so that its impact could be assessed as part of a what-if 21
sensitivity analysis. Methotrexate typically has a longer duration to resolution of symptoms than 22
surgery (see tables 7.1 and 7.2) as well as a risk of rupture which, although rare, can lead to death 23
(http://www.nlm.nih.gov/medlineplus/ency/article/000895.htm [accessed January 2012]). 24
Model probabilities 25
Where possible the probabilities were taken from the relevant clinical reviews undertaken for the 26
guideline. Data in the reviews related to treatment offered as first line and therefore, where additional 27
treatment was provided subsequent to first line treatment, it was assumed that efficacy would not be 28
different from that achieved first line. Where data wasnt available from the reviews to populate model 29
parameters, estimates taken from the literature or the GDG were used. 30
The model probabilities are shown in Table 10.10. 31
Table 10.10 Model Probabilities 32
220
Methotrexate first line success 79.3% Hajenius (1997), Moeller (2009),
Fernandez (1998), Saraj (1998),
Sowter (2001b)
Systemic methotrexate.
Methotrexate second line
success
79.3% Hajenius (1997), Moeller (2009),
Fernandez (1998), Saraj (1998),
Sowter (2001b
Assumed to be the same as for
first line but this assumption can
be relaxed as part of a sensitivity
analysis
Laparoscopic salpingectomy
success
97.4% Mol (1997)
Open salpingectomy success 100% Mol (1997)
Laparoscopic salpingotomy
success
77.6% Mol (1997)
Open salpingotomy success 95.5% Mol (1997)
Laparoscopic blood
transfusion risk
0.4% Mol (1997) Assumed that this would not
differ between salpingectomy
and salpingotomy and combined
data from both procedures (1/39
+ 0/76)
Open surgery blood
transfusion risk
7.1% Mol (1997) Assumed that this would not
differ between salpingectomy
and salpingotomy and combined
data from both procedures
(9/118 + 1/22)
Methotrexate blood
transfusion risk|Rupture
50.0% GDG estimate
Emergency admission|MTX
fail (no rupture)
10.0% GDG estimate
Emergency
admission|salpingotomy fail
10.0% GDG estimate
Laparoscopic surgical
complications
0.4% Mol (1997) There were 0/115 complications
in laparoscopic group. It is not
possible to calculate the
standard error for a proportion if
there are zero events. Without a
SE it isnt possible to sample
from a distribution. Therefore, a
common fix for this problem is to
add 0.5 to the events and non-
events (continuity correction)
Open surgery complications 3.6% Mol (1997) 5/140 in the study had surgical
complications (2/118 with open
salpingectomy open and 3/22
with salpingotomy open)
MTX|Salpingectomy fail 50.0% GDG estimate
MTX|Salpingotomy fail 35.0% GDG estimate
Rupture|MTX fail 1
st
line 15.0% GDG estimate
Health economics
221
1
Model costs and resource use 2
This analysis was undertaken from the perspective of the NHS and personal social services which is 3
in accordance with NICE guidelines methodology (NICE, 2009). Treatment alternatives were 4
compared using standard methods of incremental analysis and costs are based on 2010/11 prices. 5
Discounting was not needed as all model costs fall within a few weeks of the ectopic pregnancy 6
diagnosis. 7
Table 10.11 shows the unit cost for resource use inputs in the model whilst Table 10.12 shows the 8
assumptions, based on GDG estimates, about resource use associated with monitoring and follow-up 9
for each treatment alternative. 10
Table 10.11 Model costs 11
Methotrexate 425 NHS Reference Costs
2010-11
Currency Code: MA18C
Medical termination of pregnancy less than 14 weeks
gestation
Day case
Salpingectomy
(laparoscopy)
1,392 NHS Reference Costs
2010-11
Currency Code: MA09Z
Upper Genital Tract Laparoscopic / Endoscopic
Intermediate Procedures, Day case
Salpingectomy
(laparotomy)
2010-11
Currency Code: MA11Z
Upper Genital Tract Intermediate Procedures, Elective
Salpingotomy
(laparoscopy)
2010-11
See Salpingectomy (laparoscopy)
Salpingotomy
(laparotomy)
2010-11
See Salpingectomy (laparoscopy)
Emergency
admission
789 NHS Reference Costs
2010-11
Calculated as the difference in cost between an elective
and non-elective (long stay) Upper Genital Tract
Intermediate Procedure (Currency Code: MA11Z)
Blood
transfusion
297 http://www.hta.ac.uk/f
ullmono/mon1044.pd
Unit of blood 111.16 x 2
Matching 23.24
Updated to 2010 prices using HCHS index
Rupture|MTX fail 2
nd
30.0% GDG estimate
Salpingectomy|MTX fail 50.0% GDG estimate
Conversion to open
salpingectomy
10.0% GDG estimate
Conversion to open
salpingotomy
10.0% GDG estimate
222
Surgical
complications
1,333 Mol (1997), NHS
Reference Costs
2010-11, Plowman
(1999), Park (2011)
From Mol (1997) surgical complications were as follows:
UTI 40%, Pneumonia 40%, Thromboembolism 20%.
The following units costs were used: Thromboembolism
786, Pneumonia 941 UTI 1,997
These costs were then weighted using the proportions from
Mol (1997). The cost of Thromboembolism was estimated
from the NHS Reference Cost for DVT. UTI was estimated
from a study on hospital acquired infection in the UK
(Plowman et al., 1999) and updated to 2010 prices using
the HCHS index. Pneumonia was estimated from a US
study (Park et al., 2011). The study value in USD was
converted to GBP
(http://www.xe.com/ucc/convert/?Amount=1&From=USD&T
o=GBP [accessed January 2012]) and updated to 2010
prices using the HCHS index.
GP visit 36 Unit Costs of Health
and Social Care,
Curtis (2010)
Based on surgery consultation lasting 11.7 minutes and
including qualification and direct care staff costs
Clinic visit 141 NHS Reference Costs
2010-11
Consultant led first attendance: Face to face non-admitted
Full Blood Count 3.94 Liverpool Womens
Hospital (2011)
Obtained by GDG member
hCG 3.94 Liverpool Womens
Hospital (2011)
Kidney function 3.60 Liverpool Womens
Hospital (2011)
Liver function 4.72 Liverpool Womens
Hospital (2011)
Urine pregnancy
test
1 Estimate
Table 10.12 Quantity of Resource use by treatment 1
Item 1st
line
MTX Salpingectomy Salpingotomy Salpingotomy Salpingotomy Salpingotomy Salpingectomy MTX
2nd
line
Salpingectomy MTX MTX MTX Salpingectomy
3rd
line
Salpingectomy
Full Blood
Count
1 0 .0 0 1 1 1 1
hCG 5 0 3 2 3 5 5 4
Kidney
function
1 0 0 0 1 1 1 1
Health economics
223
Item 1st
line
MTX Salpingectomy Salpingotomy Salpingotomy Salpingotomy Salpingotomy Salpingectomy MTX
2nd
line
Salpingectomy MTX MTX MTX Salpingectomy
3rd
line
Salpingectomy
Liver
function
2 0 0 0 2 2 2 2
Urine
pregnancy
test
1 1 1 1 1 1 1 1
GP visit 0 0.5 0 0.5 0.5 0 0 0.5
Clinic visit 5 0.5 1 2.5 3.5 5 5 4
1
Probabilistic Sensitivity Analysis 2
Probabilistic sensitivity analysis takes into account sampling uncertainty in the data inputs in an 3
analogous way that sampling uncertainty is taken into account when generating confidence intervals 4
for point estimates of an effect size. A probability distribution for the parameter point estimate can be 5
established based on its point estimate, sampling variability and sample size. Probability sensitivity 6
analysis then involves Monte Carlo simulations where model parameters are repeatedly sampled from 7
their distribution to generate simulation results. The results of the simulation then provide quantitative 8
information on the extent to which the cost-effectiveness conclusion is sensitive to parameter 9
uncertainty. 10
The probabilistic parameters for the model inputs based on data obtained from the clinical review are 11
shown in Table 10.13. The probabilistic sensitivity analysis assumed that the methotrexate would 12
always have the same effectiveness when used subsequent to first line treatment as it would when 13
used as a first-line treatment. The probabilistic parameters for the treatment costs are shown in Table 14
10.14. 15
Table 10.13 Parameters for probabilistic sensitivity analysis (model probabilities) 16
Item Alpha Beta Distribution
Methotrexate success 157 41 Beta
Laparoscopic salpingectomy success 38 1 Beta
Open salpingectomy success 21 1 Beta
Laparoscopic salpingotomy success 59 17 Beta
Blood transfusion|laparoscopy 1 114 Beta
Blood transfusion|open surgery 10 130 Beta
Surgical complications|laparoscopy 0.5 114.5 Beta
Surgical complications|open surgery 5 135 Beta
17
224
Table 10.14 Parameters for probabilistic sensitivity analysis (model costs) 1
Item Mean SE Distribution
Methotrexate 425 22.66 Normal
Laparoscopic salpingectomy/salpingotomy 1,392 40.61 Normal
Open salpingectomy/salpingotomy 2,218 58.86 Normal
Surgical complications|open surgery 789 91.38 Normal
Results 2
The base-case results are shown in Table 10.15 and Figure 10.11. Using a cost minimisation 3
approach this suggests that methotrexate is the preferred strategy with a cost of 176 less than the 4
next cheapest alternative. 5
Of course, it is likely that there may be some small QALY differences between the different treatments 6
and the final two columns are used to illustrate the QALY gain that would be needed for the more 7
expensive option to be considered cost-effective relative to the next cheapest alternative. Where no 8
strategy is dominated, i.e. QALY gain from salpingectomy is not less than the QALY gain from 9
methotrexate, then an incremental QALY gain of 0.0088 or greater would be needed for 10
salpingectomy to be considered cost-effective relative to methotrexate. This is equivalent to health 11
state utility gain of 1.0 (the gain that is achieved in going from death to a life lived in perfect health) 12
sustained over approximately 77 hours. Alternatively, it would be equivalent to a health state utility 13
gain of 0.40 sustained over approximately 8 days. 14
Table 10.15 Results for base-case analysis 15
Treatment Cost Incremental cost Incremental QALY
needed
(no dominance)
Methotrexate 1,432 - -
Salpingectomy 1,608 176 0.0088
Salpingotomy 2,205 597 0.0298
16
Figure 10.11 Results for base-case analysis 17
18
19
20
21
22
23
24
Health economics
225
Calculations 1
The calculations below show how the results for each treatment alternative are derived
. 2
Salpingotomy 3
90% Laparoscopic salpingotomy 1
st
line 1,392 x 0.9 = 1,252.80 4
Of which: 5
Surgical complications 1,333 x 0.9 x 0.004 = 4.80 6
Blood transfusion 297 x 0.9 x 0.009 = 2.41 7
10% Open salpingotomy 1
st
line 2,218 x 0.1 = 221.80 8
Of which: 9
20.61% fail salpingotomy 1
st
line 12
Of which: 13
Emergency admission 789 x 0.2061 x 0.1 = 16.26 14
15
7.21% Methotrexate 2
nd
line 425 x 0.0721 = 30.64 16
12.06% Laparoscopic salpingectomy 2
nd
line 1,392 x 0.1206 = 167.88 17
Of which: 18
1.34% Open salpingectomy 2
nd
line 2,218 x 0.0134 = 29.72 21
Of which: 22
25
1.49% fail methotrexate 2nd line 26
1.34% Laparoscopic salpingectomy 3rd line 1,392 x 0.0134 = 18.65 27
Of which: 28
0.15% Open salpingectomy 3rd line 2,218 x 0.0015 = 3.33 31
Of which: 32
79.4% Salpingotomy follow up 35
Full Blood Count 3.94 x 0 x 0.794 = 0 36
hCG 3.94 x 3 x 0.794 = 9.39 37
Kidney function 3.60 x 0 x 0.794 = 0 38

Calculated answers may not exactly match those reported in the results above due to rounding
226
Liver Function 4.72 x 0 x 0.794 = 0 1
Urine Pregnancy test 1 x 1 x 0.794 = 0.79 2
GP visit 36 x 0 x 0.794 = 0 3
Clinic visit 141 x 3 x 0.794 = 335.86 4
5
5.7% Salpingotomy/methotrexate follow up 6
Full Blood Count 3.94 x 1 x 0.057 = 0.22 7
hCG 3.94 x 5 x 0.057 = 1.12 8
Kidney function 3.60 x 1 x 0.057 = 0.21 9
Liver Function 4.72 x 2 x 0.057 = 0.54 10
GP visit 36 x 0 x 0.057 = 0 12
Clinic visit 141 x 5 x 0.057 = 40.19 13
14
13.4% Salpingotomy/salpingectomy follow up 15
hCG 3.94 x 2 x 0.134 = 1.06 17
GP visit 36 x 0.5 x 0.134 = 2.41 21
Clinic visit 141 x 2.5 x 0.134 = 47.24 22
23
1.5% Salpingotomy/methotrexate/salpingectomy follow up 24
hCG 3.94 x 3 x 0.015 = 0.18 26
GP visit 36 x 0.5 x 0.015 = 0.27 30
Clinic visit 141 x 3.5 x 0.015 = 7.40 31
32
Average weighted cost = 2,205 33
34
Salpingectomy 35
90% Laparoscopic salpingectomy 1
st
line 1,392 x 0.9 = 1,252.80 36
Of which: 37
Health economics
227
10% Open salpingectomy 1
st
line 2,218 x 0.1 = 221.80 1
Of which: 2
1.04% Laparoscopic salpingectomy 2
nd
line 1,392 x 0.0104 = 14.48 5
Of which: 6
0.115% Open salpingectomy 2
nd
line 2,218 x 0.00115 = 2.55 9
Of which: 10
13
1.15% Methotrexate 2
nd
line 425 x 0.015 = 6.38 14
15
98.5% Salpingectomy follow up 16
hCG 3.94 x 0 x 0.985 = 0 18
GP visit 36 x 0.5 x 0.985 = 17.73 22
Clinic visit 141 x 0.5 x 0.985 = 69.44 23
24
1.15% Salpingectomy/methotrexate follow up 25
hCG 3.94 x 5 x 0.0115 = 0.23 27
GP visit 36 x 0 x 0.0115 = 0 31
Clinic visit 141 x 5 x 0.0115 = 8.11 32
33
35
Methotrexate 36
100% Methotrexate 1
st
line 425 x 1 = 425 37
20.7% fail Methotrexate 1
st
line 38
Of which: 39
228
15% rupture 789 x 0.207 x 0.15 = 24.50 1
Of which: 2
Blood transfusion 297 x 0.207 x 0.15 x 0.5 = 4.61 3
85% no rupture 4
Of which: 5
10% emergency admission 789 x 0.207 x 0.85 x 0.1 = 13.88 6
Methotrexate 2
nd
line 425 x 0.207 x 0.5 = 43.99 7
nd
line 1,392 x 0.207 x 0.5 x 0.9 = 129.66 8
Of which: 9
Surgical complications 1,333 x 0.207 x 0.5 x 0.9 x 0.004 = 0.50 10
Blood transfusion 297 x 0.207 x 0.5 x 0.9 x 0.009 = 0.25 11
Open salpingectomy 2
nd
line 2,218 x 0.207 x 0.5 x 0.1 = 22.96 12
Of which: 13
Surgical complications 1,333 x 0.207 x 0.5 x 0.1 x 0.036 = 0.50 14
Blood transfusion 297 x 0.207 x 0.5 x 0.1 x 0.071 = 0.22 15
16
2.14% fail Methotrexate 2
nd
line 17
Of which: 18
30% rupture 789 x 0.0214 x 0.30 = 5.07 19
Of which: 20
70% no rupture 22
Of which: 23
10% emergency admission 789 x 0.0214 x 0.70 x 0.1 = 1.18 24
rd
line 1,392 x 0.0214 x 0.9 = 26.81 25
Of which: 26
Surgical complications 1,333 x 0.0214 x 0.9 x 0.004 = 0.10 27
Open salpingectomy 3
rd
line 2,218 x 0.0214 x 0.1 = 4.75 29
Of which: 30
Surgical complications 1,333 X 0.0214 x 0.1 x 0.036 = 0.10 31
33
87.5% Methotrexate follow up 34
hCG 3.94 x 5 x 0.875 = 17.24 36
Health economics
229
0
500
1,000
1,500
2,000
2,500
0% 11% 22% 33% 44% 56% 67% 78% 89% 100%
Blood transfusion rate with MTX rupture
Salpingostomy
Salpingectomy
MTX
GP visit 36 x 0 x 0.875 = 0 1
Clinic visit 141 x 5 x 0.875 = 616.88 2
3
12.5% Methotrexate/salpingectomy follow up 4
hCG 3.94 x 4 x 0.125 = 1.97 6
GP visit 36 x 0.5 x 0.125 = 2.25 10
Clinic visit 141 x 4 x 0.125 = 70.50 11
12
14
Sensitivity Analysis 15
Probabilistic sensitivity analysis 16
A probabilistic sensitivity analysis based on one million simulations found that methotrexate had a 17
99.65% probability of being the cheapest treatment. Salpingectomy had a 0.35% probability of being 18
the cheapest treatment but salpingotomy was never the lowest cost option in any of the simulations. 19
Varying the blood transfusion rate with methotrexate rupture 20
In this analysis the proportion of patients who need blood transfusion after rupture is varied between 21
0-100%. The results are shown in Figure 10.12. 22
Figure 10.12 Sensitivity analysis varying the blood transfusion rate with methotrexate rupture 23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
230
0
500
1,000
1,500
2,000
2,500
0% 11% 22% 33% 44% 56% 67% 78% 89% 100%
Conversion to open surgery from laparoscopic salpingectomy
Salpingostomy
Salpingectomy
MTX
0
500
1,000
1,500
2,000
2,500
3,000
0% 11% 22% 33% 44% 56% 67% 78% 89% 100%
Conversion to open surgery from laparoscopic salpingotomy
Salpingostomy
Salpingectomy
MTX
Varying the conversion from laparoscopic salpingectomy to open surgery 1
In this analysis the proportion of patients having laparoscopic salpingectomy converting to open 2
salpingectomy is varied between 0-100%, see Figure 10.13 below. 3
Figure 10.13 Sensitivity analysis varying the proportion converting from laparoscopic to open salpingectomy 4
5
6
7
8
9
10
11
12
13
14
15
16
17
Varying the conversion from laparoscopic salpingotomy to open surgery 18
Here the proportion of patients converting to open salpingotomy after laparoscopic salpingotomy is 19
varied between 0-100% as is shown in Figure 10.14. 20
Figure 10.14 Sensitivity analysis varying the proportion converting from laparoscopic to open salpingotomy 21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
Health economics
231
0
500
1,000
1,500
2,000
2,500
0% 11% 22% 33% 44% 56% 67% 78% 89% 100%
Methotrexate rate following salpingectomy fail
Salpingostomy
Salpingectomy
MTX
0
500
1,000
1,500
2,000
2,500
0% 11% 22% 33% 44% 56% 67% 78% 89% 100%
Methotrexate rate following salpingtomy fail
Salpingostomy
Salpingectomy
MTX
Varying methotrexate rate if salpingectomy fails first line 1
In this sensitivity analysis the proportion of the patients having methotrexate after failed 2
salpingectomy is varied between 0-100% and is shown in Figure 9.15. 3
Figure 9.15 Sensitivity analysis varying the proportion having methotrexate after salpingectomy fails 4
5
6
7
8
9
10
11
12
13
14
15
16
17
Varying methotrexate rate if salpingotomy fails first line 18
Here the proportion of failed salpingotomy patients having methotrexate as second line treatment is 19
varied between 0-100%. The results are displayed in Figure 10.16. 20
Figure 10.16 Sensitivity analysis varying the proportion having methotrexate after salpingotomy fails 21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
232
0
500
1,000
1,500
2,000
2,500
80% 82% 84% 87% 89% 91% 93% 96% 98% 100%
Open Salpingectomy sucess
Salpingostomy
Salpingectomy
MTX
0
500
1,000
1,500
2,000
2,500
0% 6% 11% 17% 22% 28% 33% 39% 44% 50%
Rupture rate if MTX fails
Salpingostomy
Salpingectomy
MTX
Varying success rate of open salpingectomy 1
In the base case analysis it is assumed that open salpingectomy has no fail rate. Here that 2
assumption is relaxed and the effects are shown in Figure 10.17 below 3
Figure 10.17 Sensitivity analysis varying the success rate of open salpingectomy 4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
Varying rupture rate if methotrexate fails first line 19
Rupture is an important outcome in the event of treatment failure and Figure 10.18 shows the 20
consequences of varying this between 0-50%. 21
Figure 10.18 Varying the proportion with rupture after first line methotrexate treatment failure 22
23
24
25
26
27
28
29
30
31
32
33
34
35
Health economics
233
0
500
1,000
1,500
2,000
2,500
100 311 522 733 944 1,156 1,367 1,578 1,789 2,000
Cost of an emergency admission
Salpingostomy
Salpingectomy
MTX
0
500
1,000
1,500
2,000
2,500
100 644 1,189 1,733 2,278 2,822 3,367 3,911 4,456 5,000
Cost of a surgical complication
Salpingostomy
Salpingectomy
MTX
Varying cost of an emergency admission 1
In this analysis, graphed in Figure 10.19, the cost of an emergency admission is varied between 100 2
and 2,000 3
Figure 10.19 Varying the cost of an emergency admission 4
5
6
7
8
9
10
11
12
13
14
15
16
Varying the cost of surgical complications 17
Here the cost of a surgical complication is varied between 100 and 5,000 with the results illustrated 18
in Figure 10.20. 19
Figure 10.20 Varying the cost of surgical complications 20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
234
0
500
1,000
1,500
2,000
2,500
0% 11% 22% 33% 44% 56% 67% 78% 89% 100%
Salpingectomy following Methotrexate fail
Salpingostomy
Salpingectomy
MTX
0
500
1,000
1,500
2,000
2,500
50 267 483 700 917 1,133 1,350 1,567 1,783 2,000
Cost of a blood transfusion
Salpingostomy
Salpingectomy
MTX
Varying the cost of blood transfusion 1
The graph in Figure 10.21 shows the impact of varying the cost of a blood transfusion between 50 2
and 2,000. 3
Figure 10.21 Varying the cost of blood transfusion 4
5
6
7
8
9
10
11
12
13
14
15
16
Varying the proportion of patients having second line salpingectomy following first line 17
methotrexate treatment failure 18
Here the proportion of patients having failed methotrexate first line having surgery second line is 19
varied between none and all, as shown in Figure 10.22. 20
Figure 10.22 Varying the proportion having salpingectomy after first line methotrexate failure 21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
Health economics
235
0
500
1,000
1,500
2,000
2,500
0 11% 22% 33% 44% 56% 67% 78% 89% 100%
Rupture rate given MTX fail 2nd line
Salpingostomy
Salpingectomy
MTX
0
500
1,000
1,500
2,000
2,500
0% 11% 22% 33% 44% 56% 67% 78% 89% 100%
Emergency admission given MTX fail
Salpingostomy
Salpingectomy
MTX
Varying the proportion of ruptures in patients failing methotrexate second line 1
The effects of varying the proportion of women who rupture after second line methotrexate treatment 2
failure between 0-100% is shown in Figure 10.23. 3
Figure 10.23 Varying the proportion with rupture following second line methotrexate treatment failure 4
5
6
7
8
9
10
11
12
13
14
15
Varying the emergency admission rate after methotrexate treatment failure 16
In this analysis the proportion of patients who will require emergency admission after methotrexate 17
treatment failure is varied between 0-100% with the results displayed in Figure 10.24. 18
Figure 10.24 Varying the rate of emergency admission in women having methotrexate treatment failure 19
20
21
22
23
24
25
26
27
28
29
30
31
32
Introducing a QALY loss for methotrexate 33
The base case analysis took a cost minimisation approach on the assumption that all interventions 34
lead to cure. However, all of the interventions have associated and different complications, some of 35
which may carry a small risk of long term morbidity or even mortality. Also, the clinical review 36
undertaken for this guideline suggested that the resolution time was almost 9 days longer with 37
medical treatment than with surgery, which could result in a short term reduction in health related 38
quality of life. The base case analysis found that methotrexate was the cheapest option and by 39
assumption cost-effective. In sensitivity analysis it is often useful to bias the model against the 40
apparently preferred option to assess how robust the conclusion is to different assumptions. 41
236
One of the disadvantages of methotrexate compared to surgical alternatives relates to the risk of 1
rupture. Apart from the resource implications of rupture (which is addressed in the base case analysis 2
and other sensitivity analyses) the event is traumatic for the women and carries a small mortality risk. 3
It is estimated that in the UK there are 0.26 deaths per 100,000 pregnancies due to a rupture in 4
ectopic pregnancy (Cantwell et al., 2011). Using a QALY loss of 24.80 per death (National 5
Collaborating Centre for Womens and Childrens Health, 2011) that would represent a QALY loss of 6
0.000064 per woman with medically treated ectopic pregnancy due to the risk of death from 7
rupture
******
. Let us also assume that methotrexate carries a QALY loss due to a 9 day longer 8
resolution of the ectopic pregnancy. The maximum QALY loss if an ectopic pregnancy was equivalent 9
to death in terms of health related quality of life is 9 365 = 0.025. However, this doesnt seem 10
realistic for our model population who are women who are not initially presenting as an emergency. 11
One study (Sonneberg et al; 2004) estimated a short term disutility due to ectopic pregnancy of 12
0.08333 which for a duration of 9 days would amount to a QALY loss of 0.0021. If we add that to the 13
QALY loss due to the risk of death we have a total QALY loss of 0.0021 attributable to methotrexate 14
treatment
. The results of this are shown in Table 10.16. It is assumed that the surgical 15
alternatives give identical QALYs. 16
Table 10.16 Results assuming that Methotrexate has a QALY loss of 0.0021 17
18
19
20
21
22
23
In this analysis methotrexate would still be considered the cost-effective option because whilst 24
salpingectomy is more efficacious, the increased benefits are not deemed to be worth the additional 25
cost using a 20-30,000 willingness to pay for a QALY threshold. 26
If the loss in health state utility due to ectopic pregnancy was assumed to be 0.5 for the additional 9 27
days duration then the total QALY loss due to methotrexate would be 0.0110. Using this would give 28
the results shown in Table 10.17. In this scenario salpingectomy would be considered cost-effective, 29
falling within an incremental cost-effectiveness ratio of 20,000 per QALY. 30
Table 10.17 Results assuming that Methotrexate has a QALY loss of 0.0110 31
32
33
34
35
36
Discussion 37
The results of this model suggest that methotrexate was the cheapest of the three treatment 38
alternatives for women with an ectopic pregnancy but who are not presenting at the point of decision 39
with a need for urgent surgical intervention. As shown in Table 9.11 the lower cost of methotrexate is 40
driven by its low cost relative to surgical alternatives. This means that, although it has higher rates of 41
re-intervention than salpingectomy and greater follow-up costs, this is not enough to offset the initial 42
treatment saving. Salpingotomy was considerably more expensive than salpingectomy despite 43

****** 0.26 100,000 x 24.8 = 0.000064
For this sensitivity analysis we assume surgery has no long term QALY loss due adverse events
Treatment Cost Incremental cost Incremental QALY ICER
Methotrexate 1,432 - n/a n/a
Salpingectomy 1,608 176 0.0021 83,805
Salpingotomy 2,205 597 0 Dominated
Treatment Cost Incremental cost Incremental QALY ICER
Methotrexate 1,432 - n/a n/a
Salpingectomy 1,608 176 0.0110 15,999
Salpingotomy 2,205 597 0 Dominated
Health economics
237
identical initial procedural cost. This is because salpingotomy has higher rates of failure and more 1
follow up costs. 2
A probabilistic sensitivity analysis suggested that this finding was not sensitive to uncertainty in 3
parameters with well defined distributions derived by sampling methods, such as NHS treatment costs 4
and treatment success rates. However, a number of parameters were estimated using the clinical 5
opinion of the GDG and important uncertainty may remain with respect to these values. Nevertheless, 6
many of these estimates were subjected to one-way sensitivity analysis (Figures 10.12 to Figures 7
10.24) and shown to have a negligible impact on total costs. The upper and lower values used in the 8
sensitivity analysis were not intended to represent plausible extremes for these parameters but to 9
highlight the variation, or lack of it, across the entire range of theoretical values. Often the small 10
impact of these one way sensitivity analyses was a consequence of the very small number of patients 11
affected, which itself was a function of the clinical estimates of treatment success. So, for example, 12
varying the cost of blood transfusions between 100 and 2,000 will only have a minimal impact on 13
total costs because this cost is only relevant to a small minority of patients. The ordinal ranking of the 14
treatment costs only changed in one of the sensitivity analyses (see Figure 10.13) which were 15
predicated on nearly every patient having laparoscopic salpingectomy converting to an open 16
procedure, which does not reflect the clinical reality. 17
If it was deemed that to all intents and purposes these treatments were equivalent in terms of their 18
effectiveness and impact on health related quality of life then this model strongly suggests that 19
methotrexate is the preferred cost-effective option. However, if this is not thought to be the case then 20
the issue is whether the more expensive alternatives produce enough sufficient additional benefit to 21
be commensurate with their additional cost. 22
Given the point estimates of the costs of the different treatment strategies it was suggested that 23
salpingectomy would have to produce an average QALY gain of at least 0.0088 QALYs compared to 24
methotrexate to be considered cost-effective. Whilst this may not seem very much it needs to be seen 25
in the context of the time-limited nature of ectopic pregnancy and the limited difference between 26
treatments in terms of long term adverse outcomes. Indeed the mortality risk discussed in the final 27
sensitivity analysis relates to the entire population of ectopic pregnancies as a whole and it is likely to 28
be even lower in the model population where there is not initially a need for urgent surgical 29
intervention. Using an estimate of health state utility from one published study (Sonneberg et al.; 30
2004) it was suggested that methotrexate would still be considered cost-effective. Of course, 31
uncertainty surrounds that estimate, but the loss in health state utility required to make salpingectomy 32
cost-effective would have to be comparable to the loss reported in stable advanced non-small cell 33
lung cancer with no additional symptoms (Doyle at al., 2008) for example. 34
In the clinical review undertaken for this guideline comparing salpingectomy with salpingotomy the 35
evidence did not generally indicate better outcomes for salpingotomy. Therefore, in the light of the 36
proceeding discussion it is most unlikely that salpingotomy can be considered a cost-effective 37
treatment given its considerably higher costs. 38
The studies on which the clinical inputs in this model are based did not all restrict their population on 39
the basis of their hCG levels. However, from the reporting of the median values it is likely that most of 40
the women in the studies would have had an hCG of less than 5000 IU/L. The GDG are strongly of 41
the view that such patients are an important sub-group with a much higher risk of rupture and 42
therefore it is not thought that this analysis should inform recommendations for such patients. 43
Conclusion 44
This model suggests that methotrexate is almost 200 cheaper than salpingectomy and almost 800 45
cheaper than salpingotomy. Under a cost minimisation approach this would make methotrexate the 46
preferred alternative. Even if it is accepted that there are QALY differences between the treatment 47
alternatives it is far from clear that surgery offers sufficient additional benefit to make the additional 48
cost worthwhile. In particular, the higher costs of salpingotomy relative to salpingectomy in the 49
absence of any clear clinical benefit makes it reasonable to reject this as a treatment option on cost- 50
effectiveness grounds. 51
52
53

11 References 1
Abaid,L.N., As-Sanie,S., Wolfe,H.M. 2007. Relationship between crown-rump length and early 2
detection of cardiac activity. Journal of Reproductive Medicine; 5:375-378 3
Abdallah,Y., Daemen,A., Kirk,E., Pexsters,A., Naji,O., Stalder,C., Gould,D., Ahmed,S., Guha,S., 4
Syed,S., Bottomley,C., Timmerman,D., Bourne,T. 2011. Limitations of current definitions of 5
miscarriage using mean gestational sac diameter and crownrump length measurements: a 6
multicenter observational study. Ultrasound in Obstetrics & Gynecology; 5:497-502 7
Aboud,E., Chaliha,C. 1998. Nine year survey of 138 ectopic pregnancies. Archives of Gynecology 8
and Obstetrics; 2:83-87 9
Adolfsson,A., Bertero,C., Larsson,P.G. 2006. Effect of a structured follow-up visit to a midwife on 10
women with early miscarriage: a randomized study. Acta Obstetricia et Gynecologica Scandinavica; 11
3: 330-335 12
AEPU. 2012. Personal communication 13
Akhter,P., Padmanabhan,A., Babiker,W., Sayed,A., Molelekwa,V., Geary,M. 2007. Introduction of an 14
early pregnancy assessment unit: audit on the first 6 months of service. Irish Journal of Medical 15
Science; 1:23-26 16
Al-Suleiman,S.A., Khwaja,S.S. 1992. Ectopic pregnancy. Journal of Obstetrics and Gynaecology; 17
4:254-257 18
Alexander,J.M., Rouse,D.J., Varner,E., Austin,J.M.,Jr. 1996. Treatment of the small unruptured 19
ectopic pregnancy: a cost analysis of methotrexate versus laparoscopy. Obstetrics and Gynecology; 20
1:123-127 21
Ankum,W.M., Wieringa-de,Waard M., Bindels,P.J. 2001. Management of spontaneous miscarriage in 22
the first trimester: an example of putting informed shared decision making into practice. BMJ; 23
7298:1343-1346 24
Ayudhaya,O.P.N., Herabutya,Y., Chanrachakul,B., Ayuthaya,N.I.N., Prasertsawat,P. 2006. A 25
comparison of the efficacy of sublingual and oral misoprostol 400 microgram in the management of 26
early pregnancy failure: A randomized controlled trial. Journal of the Medical Association of Thailand; 27
SUPPL. 4:S5-S10 28
Bagratee,J.S., Khullar,V., Regan,L., Moodley,J., Kagoro,H. 2004. A randomized controlled trial 29
comparing medical and expectant management of first trimester miscarriage. Human Reproduction; 30
2:266-271 31
Banerjee,S., Aslam,N., Zosmer,N., Woelfer,B., Jurkovic,D. 1999. The expectant management of 32
women with early pregnancy of unknown location. Ultrasound in Obstetrics and Gynecology; 4:231- 33
236 34
Bangsgaard,N., Lund,C.O., Ottesen,B., Nilas,L. 2003. Improved fertility following conservative 35
surgical treatment of ectopic pregnancy. BJOG: An International Journal of Obstetrics and 36
Gynaecology; 8:765-770 37
Barnhart,K.T., Rinaudo,P., Hummel,A., Pena,J., Sammel,M.D., Chittams,J. 2003. Acute and chronic 38
presentation of ectopic pregnancy may be two clinical entities. Fertility and Sterility; 6:1345-1351 39
Barnhart,K.T., Sammel,M.D., Appleby,D., Rausch,M., Molinaro,T., Van Calster,B., Kirk,E., 40
Condous,G., Van Huffel,S., Timmerman,D., Bourne,T. 2010. Does a prediction model for pregnancy 41
of unknown location developed in the UK validate on a US population? Human Reproduction; 42
10:2434-2440 43
References

239
Barnhart,K.T., Sammel,M.D., Gracia,C.R., Chittams,J., Hummel,A.C., Shaunik,A. 2006. Risk factors 1
for ectopic pregnancy in women with symptomatic first-trimester pregnancies. Fertility and Sterility; 2
1:36-43 3
Baumann,R., Magos,A.L., Turnbull,A. 1991. Prospective comparison of videopelviscopy with 4
laparotomy for ectopic pregnancy. British Journal of Obstetrics and Gynaecology; 8:765-771 5
Becker,S., Solomayer,E., Hornung,R., Kurek,R., Banys,M., Aydeniz,B., Franz,H., Wallwiener,D., 6
Fehm,T. 2011. Optimal treatment for patients with ectopic pregnancies and a history of fertility- 7
reducing factors. Archives of Gynecology and Obstetrics; 1:41-45 8
Bignardi,T., Burnet,S., Alhamdan,D., Lu,C., Pardey,J., Benzie,R., Condous,G. 2010. Management of 9
women referred to an acute gynecology unit: impact of an ultrasound-based model of care. 10
Ultrasound in Obstetrics and Gynecology; 3:344-348 11
Bignardi,T., Burnet,S., Alhamdan,D., Lu,C., Pardey,J., Benzie,R., Condous,G. 2010. Management of 12
women referred to an acute gynecology unit: impact of an ultrasound-based model of care. 13
Ultrasound in Obstetrics and Gynecology; 3:344-348 14
Bigrigg,M.A., Read,M.D. 1991. Management of women referred to early pregnancy assessment unit: 15
care and cost effectiveness. BMJ; 6776:577-579 16
Bigrigg,M.A., Read,M.D. 1991. Management of women referred to early pregnancy assessment unit: 17
care and cost effectiveness. BMJ; 6776:577-579 18
Blanchard,K., Taneepanichskul,S., Kiriwat,O., Sirimai,K., Svirirojana,N., Mavimbela,N., Winikoff,B. 19
2004. Two regimens of misoprostol for treatment of incomplete abortion. Obstetrics and Gynecology; 20
5(1):860-865 21
Blohm,F., Friden,B.E., Milsom,I., Platz-Christensen,J.J., Nielsen,S. 2005. A randomised double blind 22
trial comparing misoprostol or placebo in the management of early miscarriage. BJOG: An 23
International Journal of Obstetrics and Gynaecology; 8:1090-1095 24
Blohm,F., Hahlin,M., Nielsen,S., Milsom,I. 1997. Fertility after a randomised trial of spontaneous 25
abortion managed by surgical evacuation or expectant treatment. Lancet; 349:995 26
Blumenthal,P.D., Remsburg,R.E. 1994. A time and cost analysis of the management of incomplete 27
abortion with manual vacuum aspiration. International Journal of Gynecology and Obstetrics; 3:261- 28
267 29
Bouyer,J., Coste,J., Fernandez,H., Pouly,J.L., Job-Spira,N. 2002. Sites of ectopic pregnancy: A 10 30
year population-based study of 1800 cases. Human Reproduction; 12:3224-3230 31
Bouyer,J., Job-Spira,N., Pouly,J.L., Coste,J., Germain,E., Fernandez,H. 2000. Fertility following 32
radical, conservative-surgical or medical treatment for tubal pregnancy: A population-based study. 33
British Journal of Obstetrics and Gynaecology; 6:714-721 34
Bree,R.L., Edwards,M., Bohm-Velez,M., Beyler,S., Roberts,J., Mendelson,E.B. 1989. Transvaginal 35
sonography in the evaluation of normal early pregnancy: Correlation with HCG level. American 36
Journal of Roentgenology; 1:75-79 37
Briggs,A.H., O'Brien,B.J. 2001. The death of cost-minimization analysis? Health Economics; 2:179- 38
184 39
Brown,D.L., Emerson,D.S., Felker,R.E., Cartier,M.S., Smith,W.C. 1990. Diagnosis of early embryonic 40
demise by endovaginal sonography. Journal of Ultrasound in Medicine; 11:631-636 41
Brownlea,S., Holdgate,A., Thou,S.T., Davis,G.K. 2005. Impact of an early pregnancy problem service 42
on patient care and Emergency Department presentations. Australian and New Zealand Journal of 43
Obstetrics and Gynaecology; 2:108-111 44
Brownlea,S., Holdgate,A., Thou,S.T., Davis,G.K. 2005. Impact of an early pregnancy problem service 45
on patient care and Emergency Department presentations. Australian and New Zealand Journal of 46
240
Buckley,R.G., King,K.J., Disney,J.D., Ambroz,P.K., Gorman,J.D., Klausen,J.H. 1998. Derivation of a 1
clinical prediction model for the emergency department diagnosis of ectopic pregnancy. Academic 2
Emergency Medicine; 10:951-960 3
Cacciatore,B., Stenman,U.H., Ylostalo,P. 1989. Comparison of abdominal and vaginal sonography in 4
suspected ectopic pregnancy. Obstetrics and Gynecology; 5(1):770-774 5
Cacciatore,B., Tiitinen,A., Stenman,U.H., Ylostalo,P. 1990. Normal early pregnancy: serum hCG 6
levels and vaginal ultrasonography findings. British Journal of Obstetrics and Gynaecology; 10:899- 7
903 8
Cantwell,R., Clutton-Brock,T., Cooper,G., Dawson,A., Drife,J., Garrod,D., Harper,A., Hulbert,D., 9
Lucas,S., McClure,J., Millward-Sadler,H., Neilson,J., Nelson-Piercy,C., Norman,J., O'Herlihy,C., 10
Oates,M., Shakespeare,J., de,Swiet M., Williamson,C., Beale,V., Knight,M., Lennox,C., Miller,A., 11
Parmar,D., Rogers,J., Springett,A. 2011. Saving Mothers' Lives: Reviewing maternal deaths to make 12
motherhood safer: 2006-2008. The Eighth Report of the Confidential Enquiries into Maternal Deaths 13
in the United Kingdom. BJOG: An International Journal of Obstetrics and Gynaecology; 111: 1-203 14
Chatwani,A., Yazigi,R., min-Hanjani,S. 1992. Operative laparoscopy in the management of tubal 15
ectopic pregnancy. Journal of Laparoendoscopic Surgery; 6:319-324 16
Chipchase,J., James,D. 1997. Randomised trial of expectant versus surgical management of 17
spontaneous miscarriage. British Journal of Obstetrics and Gynaecology; 7:840-841 18
Choi,H.J., Im,K.S., Jung,H.J., Lim,K.T., Mok,J.E., Kwon,Y.S. 2011. Clinical analysis of ovarian 19
pregnancy: a report of 49 cases. European Journal of Obstetrics, Gynecology, and Reproductive 20
Biology; 1:87-89 21
Chung,T.K., Lee,D.T., Cheung,L.P., Haines,C.J., Chang,A.M. 1999. Spontaneous abortion: a 22
randomized, controlled trial comparing surgical evacuation with conservative management using 23
misoprostol. Fertility and Sterility; 6:1054-1059 24
Clancy,M.J., Illingworth,R.N. 1989. The diagnosis of ectopic pregnancy in an accident and emergency 25
department. Archives of Emergency Medicine; 3:205-210 26
Colacurci,N., De,Franciscis P., Zarcone,R., Fortunato,N., Passaro,M., Mollo,A., Russo,G. 1998. Time 27
length of negativization of hCG serum values after either surgical or medical treatment of ectopic 28
pregnancy. Panminerva Medica; 3:223-225 29
Colacurci,N., Zarcone,R., De Franciscis, Mele,D., Mollo,A., de Placido, G. 1998. Tubal patency after 30
laparoscopic treatment of ectopic pregnancy. Panminerva Medica; 1:45-47 31
Condous,G., Okaro,E., Khalid,A., Timmerman,D., Lu,C., Zhou,Y., Van Huffel,S., Bourne,T. 2004. The 32
use of a new logistic regression model for predicting the outcome of pregnancies of unknown location. 33
Human Reproduction; 8:-1910 34
Condous,G., Van Calster,B., Kirk,E., Haider,Z., Timmerman,D., Van Huffel,S., Bourne,T. 2007. 35
Prediction of ectopic pregnancy in women with a pregnancy of unknown location. Ultrasound in 36
Obstetrics and Gynecology; 6:680-687 37
Condous,G., Van,Calster B., Kirk,E., Haider,Z., Timmerman,D., Van,Huffel S., Bourne,T. 2007. 38
Clinical information does not improve the performance of mathematical models in predicting the 39
outcome of pregnancies of unknown location. Fertility and Sterility; 3:572-580 40
Creinin,M.D., Moyer,R., Guido,R. 1997. Misoprostol for medical evacuation of early pregnancy failure. 41
Obstetrics and Gynecology; 5(1):768-772 42
Creinin,M.D., Washington,A.E. 1993. Cost of ectopic pregnancy management: Surgery versus 43
methotrexate. Fertility and Sterility; 6:963-969 44
Dabash,R., Ramadan,M.C., Darwish,E., Hassanein,N., Blum,J., Winikoff,B. 2010. A randomized 45
controlled trial of 400-mug sublingual misoprostol versus manual vacuum aspiration for the treatment 46
of incomplete abortion in two Egyptian hospitals. International Journal of Gynaecology and Obstetrics; 47
2:131-135 48
References

241
Dalton,V.K., Harris,L., Weisman,C.S., Guire,K., Castleman,L., Lebovic,D. 2006. Patient preferences, 1
satisfaction, and resource use in office evacuation of early pregnancy failure. Obstetrics and 2
Gynecology; 1:103-110 3
Dao,B., Blum,J., Thieba,B., Raghavan,S., Ouedraego,M., Lankoande,J., Winikoff,B. 2007. Is 4
misoprostol a safe, effective and acceptable alternative to manual vacuum aspiration for postabortion 5
care? Results from a randomised trial in Burkina Faso, West Africa. BJOG: An International Journal of 6
Dart,R.G., Mitterando,J., Dart,L.M. 1999. Rate of change of serial beta-human chorionic gonadotropin 8
values as a predictor of ectopic pregnancy in patients with indeterminate transvaginal ultrasound 9
findings. Annals of Emergency Medicine; 6:703-710 10
Daus,K., Mundy,D., Graves,W., Slade,B.A. 1989. Ectopic pregnancy. What to do during the 20-day 11
window. Journal of Reproductive Medicine; 2:162-166 12
Davies,M., Geoghegan,J. 1994. Developing an early pregnancy assessment unit. Nursing Times; 13
44:36-37 14
Davis,A.R., Hendlish,S.K., Westhoff,C., Frederick,M.M., Zhang,J., Gilles,J.M., Barnhart,K., 15
Creinin,M.D., National Institute of Child Health and Human Development Management of Early 16
Pregnancy Failure Trial. 2007. Bleeding patterns after misoprostol vs surgical treatment of early 17
pregnancy failure: results from a randomized trial. American Journal of Obstetrics and Gynecology; 18
1:31-37 19
De Cherney,A., Kase,N. 1979. The conservative surgical management of unruptured ectopic 20
pregnancy. Obstetrics and Gynecology; 4:451-455 21
de Crespigny,L.C. 1988. Early diagnosis of pregnancy failure with transvaginal ultrasound. American 22
Journal of Obstetrics and Gynecology; 2:408-409 23
De Jonge,E.T., Makin,J.D., Manefeldt,E., De Wet,G.H., Pattinson,R.C. 1995. Randomised clinical trial 24
of medical evacuation and surgical curettage for incomplete miscarriage. BMJ; 7006:662- 25
De Jonge,E.T., Pattinson,R.C., Makin,J.D., Venter,C.P. 1994. Is ward evacuation for uncomplicated 26
incomplete abortion under systemic analgesia safe and effective? A randomised clinical trial. South 27
African medical journal (Suid-Afrikaanse tydskrif vir geneeskunde); 8(1):481-483 28
Dela Cruz, A., Cumming,D.C. 1997. Factors determining fertility after conservative or radical surgical 29
treatment for ectopic pregnancy. Fertility and Sterility; 5:871-874 30
Demetroulis,C., Saridogan,E., Kunde,D., Naftalin,A.A. 2001. A prospective randomized control trial 31
comparing medical and surgical treatment for early pregnancy failure. Human Reproduction; 2:365- 32
369 33
Diamond,M.P., Wiser-Estin,M., Jones,E.E., DeCherney,A.H. 1994. Failure of standard criteria to 34
diagnose nonemergency ectopic pregnancies in a noninfertility patient population. Journal of the 35
American Association of Gynecologic Laparoscopists; 2:131-134 36
Dias,Pereira G., Hajenius,P.J., Mol,B.W., Ankum,W.M., van,der,V.1998. Fertility outcome after 37
systemic methotrexate and laparoscopic salpingostomy for tubal pregnancy.Fertility and Sterility; 38
3:S411 39
Dimitry,E.S. 1989. A ten year survey of 193 ectopic pregnancies. Journal of Obstetrics and 40
Downey,L.V., Zun,L.S. 2011. Indicators of potential for rupture for ectopics seen in the emergency 42
department. Journal of Emergencies Trauma and Shock; 3:374-377 43
Doyle,S., Lloyd,A., Walker,M. 2008. Health state utility scores in advanced non-small cell lung cancer. 44
Lung Cancer; 3:374-380 45
Drummond,M.F., Sculpher,M.J., Torrance,G.W., O'Brien,B.J., Stoddart,G.L. 2005. Methods for the 46
economic evaluation of health care programmes 47
242
Duan,L., Yan,D., Zeng,W., Yang,X., Wei,Q. 2010. Effect of progesterone treatment due to threatened 1
abortion in early pregnancy for obstetric and perinatal outcomes. Early Human Development; 1:41-43 2
Easley,H.A., Olive,D.L., Holman,J.F. 1987. Contemporary evaluation of suspected ectopic pregnancy. 3
Journal of Reproductive Medicine; 12:901-906 4
Edey,K., Draycott,T., Akande,V. 2007. Early pregnancy assessment units. Clinical Obstetrics and 5
Edwards,S., Tureck,R., Fredrick,M., Huang,X., Zhang,J., Barnhart,K. 2007. Patient acceptability of 7
manual versus electric vacuum aspiration for early pregnancy loss. Journal of Women's Health; 8
10:1429-1436 9
Egarter,C., Lederhilger,J., Kurz,C., Karas,H., Reisenberger,K. 1995. Gemeprost for first trimester 10
missed abortion. Archives of Gynecology and Obstetrics; 1:29-32 11
El Tabbakh,M.N., El Sayes,M.S. 2002. Tubal Ectopic Pregnancy: Laparoscopy vs. Laparotomy. Kasr 12
El Aini Medical Journal; 5(Supplement):367-382 13
El-Zibdeh,M.Y., Yousef,L.T. 2009. Dydrogesterone support in threatened miscarriage. Maturitas; 14
p:S43-S46 15
Fang,A., Chen,Q., Zheng,W., Li,Y., Chen,R. 2009. Termination of Missed Abortion in A Combined 16
Procedure: A Randomized Controlled Trial. Journal of Reproduction and Contraception; 1:45-49 17
Federici,D., Conti,E., Muggiasca,M.L., Ferrari,S., Arcaini,L., Brambilla,T., Meroni,M., Agarossi,A. 18
1994. Laparoscopic conservative surgery in tubal pregnancy. Minimally Invasive Therapy; 4:-201 19
Fernandez,H., Pauthier,S., Doumerc,S., Lelaidier,C., Olivennes,F., Ville,Y., Frydman,R.1995. 20
Ultrasound-guided injection of methotrexate versus laparoscopic salpingotomy in ectopic pregnancy. 21
Fertility and Sterility; 1:25-29 22
Fernandez,H., Yves Vincent,S.C., Pauthier,S., Audibert,F., Frydman,R. 1998. Randomized trial of 23
conservative laparoscopic treatment and methotrexate administration in ectopic pregnancy and 24
subsequent fertility. Human Reproduction; 11:3239-3243 25
Ferrazzi,E., Garbo,S., Sulpizio,P., Ghisoni,L., Levi,Setti:, Buscaglia,M. 1993. Miscarriage diagnosis 26
and gestational age estimation in the early first trimester of pregnancy: transabdominal versus 27
transvaginal sonography. Ultrasound in Obstetrics and Gynecology; 1:36-41 28
Foulk,R.A., Steiger,R.M. 1996. Operative management of ectopic pregnancy: a cost analysis. 29
American Journal of Obstetrics and Gynecology; 1:90-96 30
Fox,R., Savage,R., Evans,T., Moore,L. 1999. Early pregnancy assessment; a role for the 31
gynaecology nurse-practitioner. Journal of Obstetrics and Gynaecology; 6:615-616 32
Gavin,P.S. 1972. Rhesus sensitization in abortion. Obstetrics and Gynecology; 1:37-40 33
Gerhard,I., Gwinner,B., Eggert-Kruse,W., Runnebaum,B. 1987. Double-blind controlled trial of 34
progesterone substitution in threatened abortion. Biological Research in Pregnancy and Perinatology; 35
1(1ST Half):26-34 36
Gevaert,O., De Smet,F., Kirk,E., Van Calster,B., Bourne,T., Van Huffel,S., Moreau,Y., Timmerman,D., 37
De Moor,B., Condous,G. 2006. Predicting the outcome of pregnancies of unknown location: Bayesian 38
networks with expert prior information compared to logistic regression. Human Reproduction; 7:1824- 39
1831 40
Giambelli,E., Candiani,M., Natale,A., Gruft,L., De,MarinisS, Sambruni,I., Colombo,P., Busacca,M. 41
1996. Laparoscopic treatment of ectopic pregnancy: Analysis of 114 consecutive cases. Italian 42
Journal of Gynaecology and Obstetrics; 1:5-9 43
Goksedef,B.P., Kef,S., Akca,A., Bayik,R.N., Cetin,A. 2011. Risk factors for rupture in tubal ectopic 44
pregnancy: definition of the clinical findings. European Journal of Obstetrics, Gynecology, and 45
Reproductive Biology; 1:96-99 46
Goldstein,S.R. 1992. Significance of cardiac activity on endovaginal ultrasound in very early embryos. 47
References

243
Gonzalez,F.A., Waxman,M. 1981. Ectopic pregnancy. A retrospective study of 501 consecutive 1
patients. Diagnostic Gynecology and Obstetrics; 3:181-186 2
Gray,D.T., Thorburn,J., Lundorff,P., Strandell,A., Lindblom,B. 1995. A cost-effectiveness study of a 3
randomised trial of laparoscopy versus laparotomy for ectopic pregnancy. Lancet; 8958:1139-1143 4
Graziosi,G., Bruinse,H.W., Reuwer,P.J.H., Teteringen,O., Mol,B.J.W. 2005b. Fertility outcome after a 5
randomized trial comparing curettage with misoprostol for treatment of early pregnancy failure. 6
Human Reproduction; 20:1749-1750 7
Graziosi,G.C., Bruinse,H.W., Reuwer,P.J., van Kessel,P.H., Westerweel,P.E., Mol,B.W. 2005a. 8
Misoprostol versus curettage in women with early pregnancy failure: impact on women's health- 9
related quality of life. A randomized controlled trial. Human Reproduction; 8:2340-2347 10
Graziosi,G.C.M., Mol,B.W.J., Reuwer,P.J.H., Drogtrop,A., Bruinse,H.W. 2004. Misoprostol versus 11
curettage in women with early pregnancy failure after initial expectant management: A randomized 12
trial. Human Reproduction; 8:1894-1899 13
Graziosi,G.C., van der Steeg,J.W., Reuwer,P.H., Drogtrop,A.P., Bruinse,H.W., Mol,B.W. 2005c. 14
Economic evaluation of misoprostol in the treatment of early pregnancy failure compared to curettage 15
after an expectant management. Human reproduction; 4:1067-1071 16
Gruft,L., Bertola,E., Luchini,L., Azzilonna,C., Bigatti,G., Parazzini,F. 1994. Determinants of 17
reproductive prognosis after ectopic pregnancy. Human Reproduction; 7:1333-1336 18
Hahlin,M., Sjoblom,P., Lindblom,B. 1991. Combined use of progesterone and human chorionic 19
gonadotropin determinations for differential diagnosis of very early pregnancy. Fertility and Sterility; 20
3:492-496 21
Hajenius,P.J., Engelsbel,S., Mol,B.W., van,der,V, Ankum,W.M., Bossuyt,P.M., Hemrika,D.J., 22
Lammes,F.B. 1997. Randomised trial of systemic methotrexate versus laparoscopic salpingostomy in 23
tubal pregnancy. Lancet; 9080:774-779 24
Harper,J. 2003. Midwives and miscarriage: the development of an early pregnancy unit. MIDIRS 25
Midwifery Digest; 2:183-185 26
Harwood,B., Nansel,T., National Institute of Child Health and Human Development Management of 27
Early Pregnancy Failure Trial. 2008. Quality of life and acceptability of medical versus surgical 28
management of early pregnancy failure. BJOG: An International Journal of Obstetrics and 29
Hassan,R., Sandu,AanaL, Rich,K., Lal,S. 2009. Is transvaginal ultrasound a reliable test in the 31
diagnosis of early embryonic demise? Outcomes of embryos less than 6mm in crown-rump length 32
without cardiac activity. International Journal of Gynecology and Obstetrics; S536 33
Health and Social Care Information Centre. 2012. NHS Maternity Statistics, 2010-11. Available from 34
http://www.hesonline.nhs.uk/Ease/servlet/ContentServer?siteID=1937&categoryID=1815 (accessed 35
February 2012) 36
Hensleigh,P.A., Leslie,W., Dixon,E., Hall,E., Kitay,D.Z., Jackson,J.E. 1977. Reduced dose of Rho(D) 37
immune globulin following induced first-trimester abortion. American Journal of Obstetrics and 38
Hill,K. 2009. Improving services provided in an early pregnancy assessment clinic. Nursing Times; 40
6:18-19 41
Hinshaw,H.K.S. 1997. Medical management of miscarriage. Problems in early pregnancy - advances 42
in diagnosis and management; 284-295 43
Hua i Hsueh Tsa Chih - Chinese Medical Journal; 6:457-462 44
Hughes,J., Ryan,M., Hinshaw,K., Henshaw,R., Rispin,R., Templeton,A. 1996. The costs of treating 45
miscarriage: a comparison of medical and surgical management. British Journal of Obstetrics and 46
Gynaecology; 12:1217-1221 47
244
Hutton,J.D., Narayan,R. 1986. Is ectopic pregnancy too often diagnosed too late? New Zealand 1
Medical Journal; 794:3-5 2
Jabbar,F.A., Al-Wakeel,M. 1980. A study of 45 cases of ectopic pregnancy. International Journal of 3
Gynaecology and Obstetrics; 3:214-217 4
Jiao,L.Z., Zhao,J., Wan,X.R., Liu,X.Y., Feng,F.Z., Ren,T., Xiang,Y. 2008. Diagnosis and treatment of 5
cesarean scar pregnancy. Chinese Medical Sciences Journal; 1:10-15 6
Jurkovic,D., Wilkinson,H. 2011. Diagnosis and management of ectopic pregnancy. BMJ; 7811:1353- 7
1357 8
Katz,J., Marcus,R.G. 1973. Incidence of Rh immunization following abortion: possible detection of 9
lymphocyte priming to Rh antigen. American Journal of Obstetrics and Gynecology; 2:261-267 10
Kazandi,M., Turan,V. 2011. Ectopic pregnancy; risk factors and comparison of intervention success 11
rates in tubal ectopic pregnancy. Clinical and Experimental Obstetrics and Gynecology; 1:67-70 12
Keith,L., Bozorgi,N. 1977. Small dose anti-Rh therapy after first trimester abortion. International 13
Journal of Gynaecology and Obstetrics; 3:235-237 14
Kivikoski,A.I., Martin,C.M., Smeltzer,J.S. 1990. Transabdominal and transvaginal ultrasonography in 15
the diagnosis of ectopic pregnancy: a comparative study. American Journal of Obstetrics and 16
Gynecology; 1(1):123-128 [Erratum appears in Am J Obstet Gynecol 1990 Dec;163(6 Pt 1):2030] 17
Kovavisarach,E., Jamnansiri,C. 2005. Intravaginal misoprostol 600 microg and 800 microg for the 18
treatment of early pregnancy failure. International Journal of Gynaecology and Obstetrics; 3:208-212 19
Kovavisarach,E., Sathapanachai,U. 2002. Intravaginal 400 microg misoprostol for pregnancy 20
termination in cases of blighted ovum: a randomised controlled trial. Australian and New Zealand 21
Journal of Obstetrics and Gynaecology; 2:161-163 22
Krag Moeller,L.B., Moeller,C., Thomsen,S.G., Andersen,L.F., Lundvall,L., Lidegaard,O., Kjer,J.J., 23
Ingemanssen,J.L., Zobbe,V., Floridon,C., Petersen,J., Ottesen,B. 2009. Success and spontaneous 24
pregnancy rates following systemic methotrexate versus laparoscopic surgery for tubal pregnancies: a 25
randomized trial. Acta Obstetricia et Gynecologica Scandinavica; 12:1331-1337 26
Kuroda,K., Takeuchi,H., Kitade,M., Kikuchi,I., Shimanuki,H., Kumakiri,J., Kobayashi,Y., Kuroda,M., 27
Takeda,S. 2009. Assessment of tubal disorder as a risk factor for repeat ectopic pregnancy after 28
laparoscopic surgery for tubal pregnancy. Journal of Obstetrics and Gynaecology Research; 3:520- 29
524 30
Kushwah,B., Singh,A. 2009. Sublingual versus oral misoprostol for uterine evacuation following early 31
pregnancy failure. International Journal of Gynaecology and Obstetrics; 1:43-45 32
Langebrekke,A., Sornes,T., Urnes,A. 1993. Fertility outcome after treatment of tubal pregnancy by 33
laparoscopic laser surgery. Acta Obstetricia et Gynecologica Scandinavica; 7:547-549 34
Larrain,D., Marengo,F., Bourdel,N., Jaffeux,P., ublet-Cuvelier,B., Pouly,J.L., Mage,G., Rabischong,B. 35
2011. Proximal ectopic pregnancy: a descriptive general population-based study and results of 36
different management options in 86 cases. Fertility and Sterility; 3:867-871 37
Lecuru,F., Robin,F., Chasset,S., Leonard,F., Guitti,S., Taurelle,R. 2000. Direct cost of single dose 38
methotrexate for unruptured ectopic pregnancy. Prospective comparison with laparoscopy. European 39
Journal of Obstetrics, Gynecology, and Reproductive Biology; 1:1-6 40
Lee,C., Slade,P., Lygo,V. 1996. The influence of psychological debriefing on emotional adaptation in 41
women following early miscarriage: a preliminary study. British Journal of Medical Psychology; Pt 1: 42
47-58 43
Lee,D.T., Cheung,L.P., Haines,C.J., Chan,K.P., Chung,T.K. 2001. A comparison of the psychologic 44
impact and client satisfaction of surgical treatment with medical treatment of spontaneous abortion: a 45
randomized controlled trial. American Journal of Obstetrics and Gynecology; 4:953-958 46
Lelaidier,C., Baton-Saint-Mleux,C., Fernandez,H., Bourget,P., Frydman,R. 1993. Mifepristone (RU 47
486) induces embryo expulsion in first trimester non-developing pregnancies: a prospective 48
randomized trial. Human Reproduction; 3:492-495 49
References

245
Levi,C.S., Lyons,E.A., Lindsay,D.J. 1988. Early diagnosis of nonviable pregnancy with endovaginal 1
US. Radiology; 2:383-385 2
Levi,C.S., Lyons,E.A., Zheng,X.H., Lindsay,D.J., Holt,S.C. 1990. Endovaginal US: demonstration of 3
cardiac activity in embryos of less than 5.0 mm in crown-rump length. Radiology; 1:71-74 4
Lewis,G. 2007. Saving Mothers' Lives: reviewing maternal deaths to make motherhood safer 2003- 5
2005. The Seventh Report on Confidential Enquiries into Maternal Deaths in the United Kingdom. 6
Lister,M.S., Shaffer,L.E., Bell,J.G., Lutter,K.Q., Moorma,K.H. 2005. Randomized, double-blind, 7
placebo-controlled trial of vaginal misoprostol for management of early pregnancy failures. American 8
Journal of Obstetrics and Gynecology; 4:1338-1343 9
Lo,L., Pun,T.C., Chan,S. 1999. Tubal ectopic pregnancy: an evaluation of laparoscopic surgery 10
versus laparotomy in 614 patients. Australian and New Zealand Journal of Obstetrics and 11
Lowe,P.J. 1998. A casemix cost comparison of 2 treatments for ectopic pregnancy (Brief record). 13
Australian and New Zealand Journal of Obstetrics and Gynaecology; 3:333-335 14
Lundorff,P. 1997. Laparoscopic surgery in ectopic pregnancy. Acta Obstetricia et Gynecologica 15
Scandinavica Supplement; 81-84 16
Lundorff,P., Thorburn,J., Hahlin,M., Kallfelt,B., Lindblom,B. 1991. Laparoscopic surgery in ectopic 17
pregnancy. A randomized trial versus laparotomy. Acta Obstetricia et Gynecologica Scandinavica; 4- 18
5:343-348 19
Lundorff,P., Thorburn,J., Lindblom,B. 1992. Fertility outcome after conservative surgical treatment of 20
ectopic pregnancy evaluated in a randomized trial. Fertility and Sterility; 5:998-1002 21
Makinen,J., Nikkanen,V., Kivikoski,A. 1984. Problems and benefits in early diagnosis of ectopic 22
pregnancy. European Journal of Obstetrics, Gynecology, and Reproductive Biology; 6:381-391 23
Mecke,H., Semm,K., Lehmann-Willenbrock,E. 1997. Results of operative pelviscopy in 202 cases of 24
ectopic pregnancy. International Journal of Fertility; 2:93-94 25
Mehra,S., Gujral,A., Mehra,G. 1998. Endoscopic vs. conventional surgery for tubal gestation. 26
International Journal of Gynecology and Obstetrics; 3:297-298 27
Menon,S., Sammel,M.D., Vichnin,M., Barnhart,K.T. 2007. Risk factors for ectopic pregnancy: a 28
comparison between adults and adolescent women. Journal of Pediatric and Adolescent Gynecology; 29
3:181-185 30
Michelas,S., Creatsas,G., Fakas,G., Kaskarelis,D. 1980. Ectopic pregnancy: outcome of 152 cases; 31
International Surgery; 4:355-358 32
Miscarriage Association. 2011. Miscarriage and the workplace: a guide for employers. (available from: 33
http://www.miscarriageassociation.org.uk/wp/wp-content/uploads/2011/04/Miscarriage-the- 34
Workplace-Feb-20111.pdf) 35
Mol,B.W.J., Hajenius,P.J., Engelsbel,S., Ankum,W.M., Hemrika,D.J., van der Veen,F., 36
Bossuyt,P.M.M. 1999. Treatment of tubal pregnancy in the Netherlands: An economic comparison of 37
systemic methotrexate administration and laparoscopic salpingostomy. American Journal of 38
Mol,B.W., Hajenius,P.J., Engelsbel,S., Ankum,W.M., van der Veen,F., Hemrika,D.J., Bossuyt,P.M. 40
1997. An economic evaluation of laparoscopy and open surgery in the treatment of tubal pregnancy. 41
Acta Obstetricia et Gynecologica Scandinavica; 6:596-600 42
Mol,B.W.J., Hajenius,P.J., Engelsbel,S., Ankum,W.M., van der Veen,F., Hemrika,D.J., 43
Bossuyt,P.M.M. 1998. Serum human chorionic gonadotropin measurement in the diagnosis of ectopic 44
pregnancy when transvaginal sonography is inconclusive. Fertility and Sterility; 5:972-981 45
246
Mol,B.W.J., Hajenius,P.J., Engelsbel,S., Ankum,W.M., Van der Veen,F., Hemrika,D.J., 1
Bossuyt,P.M.M. 1997. An economic evaluation of laparoscopy and open surgery in the treatment of 2
tubal pregnancy. Acta Obstetricia et Gynecologica Scandinavica; 6:596-600 3
Mol,B.W.J., Matthijsse,H.C., Tinga,D.J., Huynh,T., Hajenius,P.J., Ankum,W.M., Bossuyt,P.M.M., van 4
der Veen,F. 1998. Fertility after conservative and radical surgery for tubal pregnancy. Human 5
Reproduction; 7:1804-1809 6
Montesinos,R., Durocher,J., Leon,W., Arellano,M., Pena,M., Pinto,E., Winikoff,B. 2011. Oral 7
misoprostol for the management of incomplete abortion in Ecuador. International Journal of 8
Gynaecology and Obstetrics; 2:135-139 9
Moodliar,S., Bagratee,J.S., Moodley,J. 2005. Medical vs. surgical evacuation of first-trimester 10
spontaneous abortion. International Journal of Gynaecology and Obstetrics; 1:21-26 11
Morlock,R.J., Lafata,J.E., Eisenstein,D. 2000. Cost-effectiveness of single-dose methotrexate 12
compared with laparoscopic treatment of ectopic pregnancy. Obstetrics and Gynecology; 3:407-412 13
Muffley,P.E., Stitely,M.L., Gherman,R.B. 2002. Early intrauterine pregnancy failure: a randomized trial 14
of medical versus surgical treatment. American Journal of Obstetrics and Gynecology; 2:321-325 15
Murphy,A.A., Nager,C.W., Wujek,J.J., Kettel,L.M., Torp,V.A., Chin,H.G. 1992. Operative laparoscopy 16
versus laparotomy for the management of ectopic pregnancy: a prospective trial. Fertility and Sterility; 17
6:1180-1185 18
Murray,S., Barron,S.L. 1971. Rhesus isoimmunization after abortion. British Medical Journal; 19
5766:90-92 20
Murray,S., Barron,S.L., McNay,R.A. 1970. Transplacental haemorrhage after abortion. Lancet; 21
7648:631-634 22
National Collaborating Centre for Womens and Childrens Health. 2011. Caesarean Section 23
National Institute for Health and Clinical Excellence. 2009. The guidelines manual 24
Neugebauer,R., Kline,J., Markowitz,J.C., Bleiberg,K.L., Baxi,L., Rosing,M.A., Levin,B., Keith,J. 2006. 25
Pilot randomized controlled trial of interpersonal counseling for subsyndromal depression following 26
miscarriage. Journal of Clinical Psychiatry; 8: 1299-1304 27
Ngai,S.W., Chan,Y.M., Tang,O.S., Ho,P.C. 2001. Vaginal misoprostol as medical treatment for first 28
trimester spontaneous miscarriage. Human Reproduction; 7:1493-1496 29
Ngoc,N.T., Blum,J., Westheimer,E., Quan,T.T., Winikoff,B. 2004. Medical treatment of missed 30
abortion using misoprostol. International Journal of Gynaecology and Obstetrics; 2:138-142 31
Ngoc,N.T.N., Blum,J., Durocher,J., Quan,T.T., Winikoff,B. 2005. A randomized controlled study 32
comparing 600 versus 1,200 microg oral misoprostol for medical management of incomplete abortion; 33
Contraception; 6:438-442 34
Nikcevic,A.V., Kuczmierczyk,A.R., Nicolaides,K.H. 2007. The influence of medical and psychological 35
interventions on women's distress after miscarriage. Journal of Psychosomatic Research; 3: 283-290 36
Nikcevic,A.V., Tunkel,S.A., Nicolaides,K.H. 1998. Psychological outcomes following missed abortions 37
and provision of follow-up care. Ultrasound in Obstetrics and Gynecology; 2: 123-128 38
Nielsen,S., Hahlin,M. 1995. Expectant management of first-trimester spontaneous abortion. Lancet; 39
8942:84-86 40
Nielsen,S., Hahlin,M., Moller,A., Granberg,S. 1996. Bereavement, grieving and psychological 41
morbidity after first trimester spontaneous abortion: comparing expectant management with surgical 42
evacuation. Human Reproduction; 8:1767-1770 43
Nielsen,S., Hahlin,M., Platz-Christensen,J. 1999. Randomised trial comparing expectant with medical 44
management for first trimester miscarriages. British Journal of Obstetrics and Gynaecology; 8:804- 45
807 46
Nieuwkerk,Pythia T., Hajenius,Petra J., Ankum,Willem M., Van der Veen,Fulco, Wijker,Wouter, 47
Bossuyt,Patrick M.M. 1998. Systemic methotrexate therapy versus laparoscopic salpingostomy in 48
References

247
patients with tubal pregnancy. Part I. Impact on patients' health-related quality of life. Fertility and 1
Sterility; 3:511-517 2
Niinimaki,M., Jouppila,P., Martikainen,H., Talvensaari-Mattila,A. 2006. A randomized study comparing 3
efficacy and patient satisfaction in medical or surgical treatment of miscarriage. Fertility and Sterility; 4
2:367-372 5
Niinimaki,M., Karinen,P., Hartikainen,A.L., Pouta,A. 2009. Treating miscarriages: a randomised study 6
of cost-effectiveness in medical or surgical choice. BJOG: An International Journal of Obstetrics and 7
Omar,M.H., Mashita,M.K., Lim,P.S., Jamil,M.A. 2005. Dydrogesterone in threatened abortion: 9
pregnancy outcome. Journal of Steroid Biochemistry and Molecular Biology; 5:421-425 10
Ory,S.J., Nnadi,E., Herrmann,R., O'Brien,P.S., Melton,L.J.,III. 1993. Fertility after ectopic pregnancy. 11
Fertility and Sterility; 2:231-235 12
Palagiano,A., Bulletti,C., Pace,M.C., DE,Ziegler D., Cicinelli,E., Izzo,A. 2004. Effects of vaginal 13
progesterone on pain and uterine contractility in patients with threatened abortion before twelve 14
weeks of pregnancy.Annals of the New York Academy of Sciences; 200-210 15
Pandian,R.U. 2009. Dydrogesterone in threatened miscarriage: A Malaysian experience. Maturitas; 16
1:S47-S50 17
Pang,M.W., Lee,T.S., Chung,T.K.H. 2001. Incomplete miscarriage: A randomized controlled trial 18
comparing oral with vaginal misoprostol for medical evacuation. Human Reproduction; 11:2283-2287 19
Paritakul,P., Phupong,V. 2010.Comparative study between oral and sublingual 600 microg 20
misoprostol for the treatment of incomplete abortion. Journal of Obstetrics and Gynaecology 21
Research; 5:978-983 22
Park,H., Rascati,K.L. 2011. Direct costs of pneumonia in the United States: An analysis of 2008 23
medical expenditure panel survey (MEPS) data. Value in Health; 7:A492- 24
Parker,J., Permezel,M., Thompson,D. 1994. Review of the management of ectopic pregnancy in a 25
major teaching hospital: Laparoscopic surgical treatment and persistent ectopic pregnancy. Australian 26
and New Zealand Journal of Obstetrics and Gynaecology; 5:575-579 27
Pennell,R.G., Needleman,L., Pajak,T., Baltarowich,O., Vilaro,M., Goldberg,B.B., Kurtz,A.B. 1991. 28
Prospective comparison of vaginal and abdominal sonography in normal early pregnancy. Journal of 29
Ultrasound in Medicine; 2:63-67 30
Petrou,S., Trinder,J., Brocklehurst,P., Smith,L. 2006. Economic evaluation of alternative management 31
methods of first-trimester miscarriage based on results from the MIST trial. BJOG: An International 32
Journal of Obstetrics and Gynaecology; 8:879-889 33
Pexsters,A., Luts,J., Van Schoubroeck, D., Bottomley,C., Van Calster, B., Van Huffel, S., Abdallah,Y., 34
D'Hooghe,T., Lees,C., Timmerman,D., Bourne,T. 2011. Clinical implications of intra- and 35
interobserver reproducibility of transvaginal sonographic measurement of gestational sac and crown- 36
rump length at 6-9 weeks' gestation. Ultrasound in Obstetrics and Gynecology; 5: 510-515 37
Plowman,R., Graves,N., Griffin,M., Roberts,J.A., Swan,A.V., Cookson,B., Taylor,L. 1999 . The 38
socioecomic burden of hospital acquired infection 39
Poddar,A., Tyagi,J., Hawkins,E., Opemuyi,I. 2011. Standards of care provided by Early Pregnancy 40
Assessment Units (EPAU): A UK-wide survey. Journal of Obstetrics & Gynaecology,J Obstet 41
Gynaecol; 7:640-644 42
Powers,D.N. 1980. Ectopic pregnancy: a five-year experience. Southern Medical Journal; 8:1012- 43
1015 44
Raziel,A., Schachter,M., Mordechai,E., Friedler,S., Panski,M., Ron-El,R. 2004. Ovarian pregnancy-a 45
12-year experience of 19 cases in one institution. European Journal of Obstetrics, Gynecology, and 46
Reproductive Biology; 1:92-96 47
248
Rempen,A. 1990. Diagnosis of viability in early pregnancy with vaginal sonography. Journal of 1
Ultrasound in Medicine; 12:711-716 2
Rita, Gupta,S., Kumar,S. 2006. A randomised comparison of oral and vaginal misoprostol for medical 3
management of first trimester missed abortion. JK Science; 1:35-38 4
Rizzuto,M.I., Oliver,R., Odejinmi,F. 2008. Laparoscopic management of ectopic pregnancy in the 5
presence of a significant haemoperitoneum. Archives of Gynecology and Obstetrics; 5:433-436 6
Robin,F., Lecuru,F., Bernard,J.P., Mac-Cordick,C., Boucaya,V., Taurelle,R. 1998. Methotrexate 7
provides significant cost savings for the treatment of unruptured ectopic pregnancy. Clinical Drug 8
Investigation; 5:405-411 9
Rocconi,R.P., Chiang,S., Richter,H.E., Straughn,J.M.,Jr. 2005. Management strategies for abnormal 10
early pregnancy: a cost-effectiveness analysis. Journal of Reproductive Medicine; 7:486-490 11
Round,J.A., Jacklin,P., Fraser,R.B., Hughes,R.G., Mugglestone,M.A., Holt,R.I. 2011. Screening for 12
gestational diabetes mellitus: cost-utility of different screening strategies based on a woman's 13
individual risk of disease. Diabetologia; 2:256-263 14
Rowling,S.E., Langer,J.E., Coleman,B.G., Nisenbaum,H.L., Horii,S.C., Arger,P.H. 1999. Sonography 15
during early pregnancy: Dependence of threshold and discriminatory values on transvaginal 16
transducer frequency. American Journal of Roentgenology; 4:983-988 17
Sahin,H.G., Sahin,H.A., Kocer,M. 2001.Randomized outpatient clinical trial of medical evacuation and 18
surgical curettage in incomplete miscarriage.[Erratum appears in Eur J Contracept Reprod Health 19
Care 2002 Mar;7(1):iv]. European Journal of Contraception and Reproductive Health Care; 3:141-144 20
Saraj,A.J., Wilcox,J.G., Najmabadi,S., Stein,S.M., Johnson,M.B., Paulson,R.J.1998. Resolution of 21
hormonal markers of ectopic gestation: a randomized trial comparing single-dose intramuscular 22
methotrexate with salpingostomy. Obstetrics and Gynecology; 6:989-994 23
Schurz,B., Wenzl,R., Eppel,W., Reinold,E. 1990. Early detection of ectopic pregnancy by transvaginal 24
ultrasound. Archives of Gynecology and Obstetrics; 1:25-29 25
Sjourn,N., Callahan,S., Chabrol,H. 2010. The utility of a psychological intervention for coping with 26
spontaneous abortion. Journal of Reproductive and Infant Psychology; 3: 287-296 27
Sellappan,K., Mcgeown,A., Archer,A. 2009. A survey to assess the efficiency of an early pregnancy 28
unit. International Journal of Gynecology and Obstetrics; S542- 29
Shah,N., Azam,S.I., Khan,N.H. 2010. Sublingual versus vaginal misoprostol in the management of 30
missed miscarriage. Journal of the Pakistan Medical Association; 2:113-116 31
Shapiro,B.S., Cullen,M., Taylor,K.J., DeCherney,A.H. 1988. Transvaginal ultrasonography for the 32
diagnosis of ectopic pregnancy Fertility and Sterility; 3:425-429 33
Shaunik,A., Kulp,J., Appleby,D.H., Sammel,M.D., Barnhart,K.T. 2011. Utility of dilation and curettage 34
in the diagnosis of pregnancy of unknown location. American Journal of Obstetrics and Gynecology; 35
2:130-130 36
Shelley,J.M., Healy,D., Grover,S. 2005. A randomised trial of surgical, medical and expectant 37
management of first trimester spontaneous miscarriage. Australian and New Zealand Journal of 38
Shelley,J.M., Healy,D., Grover,S. 2005. A randomised trial of surgical, medical and expectant 40
management of first trimester spontaneous miscarriage. Australian and New Zealand Journal of 41
Sherman,D., Langer,R., Sadovsky,G., Bukovsky,I., Caspi,E. 1982. Improved fertility following ectopic 43
pregnancy. Fertility and Sterility; 4:497-502 44
Shillito,J., Walker,J.J. 1997. Early pregnancy assessment units. British Journal of Hospital Medicine; 45
10:505-509 46
References

249
Shwekerela,B., Kalumuna,R., Kipingili,R., Mashaka,N., Westheimer,E., Clark,W., Winikoff,B. 2007. 1
Misoprostol for treatment of incomplete abortion at the regional hospital level: results from Tanzania. 2
BJOG: An International Journal of Obstetrics and Gynaecology; 11:1363-1367 3
Silva,P.D., Schaper,A.M., Rooney,B. 1993. Reproductive outcome after 143 laparoscopic procedures 4
for ectopic pregnancy. Obstetrics and Gynecology; 5(1):710-715 5
Simonovits,I., Bajtai,G., Kellner,R., Kerenyl,M., Rucz,L., Szilvas,R., Takacs,S. 1974. Immunization of 6
RhO(D)-negative secundigravidae whose first pregnancy was terminated by induced abortion. 7
Haematologia; 1-4:291-298 8
Simonovits,I., Timar,I., Bajtai,G. 1980. Rate of Rh immunization after induced abortion. Vox 9
Sanguinis; 3:161-164 10
Smith,L.F., Frost,J., Levitas,R., Bradley,H., Garcia,J. 2006. Women's experiences of three early 11
miscarriage management options: a qualitative study. British Journal of General Practice; 524:198- 12
205 13
Smith,L.F., Frost,J., Levitas,R., Bradley,H., Garcia,J. 2006. Women's experiences of three early 14
miscarriage management options: a qualitative study. British Journal of General Practice; 524:198- 15
205 16
Smith,L.F.P., Ewings,P.D., Quinlan,C. 2009. Incidence of pregnancy after expectant, medical, or 17
surgical management of spontaneous first trimester miscarriage: Long term follow-up of miscarriage 18
treatment (MIST) randomised controlled trial. BMJ; 7726:910- 19
Smith,L.F.P., Ewings,P.D., Quinlan,C. 2009. Incidence of pregnancy after expectant, medical, or 20
surgical management of spontaneous first trimester miscarriage: Long term follow-up of miscarriage 21
treatment (MIST) randomised controlled trial. BMJ; 7726:910- 22
Sonnenberg,F.A., Burkman,R.T., Hagerty,C.G., Speroff,L., Speroff,T. 2004. Costs and net health 23
effects of contraceptive methods. Contraception; 6:447-459 24
Sotiriadis, A., Papatheodorou, S., Makrydimas, G. 2004. Threatened miscarriage: evaluation and 25
treatment. BMJ; 329(7458): 152-155 26
Sowter,M.C., Farquhar,C.M., Gudex,G. 2001a. An economic evaluation of single dose systemic 27
methotrexate and laparoscopic surgery for the treatment of unruptured ectopic pregnancy. BJOG: An 28
Sowter,M.C., Farquhar,C.M., Petrie,K.J., Gudex,G. 2001b. A randomised trial comparing single dose 30
systemic methotrexate with laparoscopy for the treatment of unruptured tubal pregnancy. BJOG: An 31
Steinkampf,M.P., Guzick,D.S., Hammond,K.R., Blackwell,R.E. 1997. Identification of early pregnancy 33
landmarks by transvaginal sonography: analysis by logistic regression. Fertility and Sterility; 1:168- 34
170 35
Stewart,B.K., Nazar-Stewart,V., Toivola,B. 1995. Biochemical discrimination of pathologic pregnancy 36
from early, normal intrauterine gestation in symptomatic patients. American Journal of Clinical 37
Pathology; 4:386-390 38
Stewart,F.H., Burnhill,M.S., Bozorgi,N. 1978. Reduced dose of Rh immunoglobulin following first 39
trimester pregnancy termination. Obstetrics and Gynecology; 3:318-322 40
Stockheim,D., Machtinger,R., Wiser,A., Dulitzky,M., Soriano,D., Goldenberg,M., Schiff,E., 41
Seidman,D.S. 2006. A randomized prospective study of misoprostol or mifepristone followed by 42
misoprostol when needed for the treatment of women with early pregnancy failure. Fertility and 43
Sterility; 4:956-960 44
Stovall,T.G., Ling,F.W., Gray,L.A. 1991. Single-dose methotrexate for treatment of ectopic pregnancy. 45
Swanson,K.M. 1999. Effects of caring, measurement, and time on miscarriage impact and women's 47
well-being. Nursing Research; 6: 288-298 48
250
Swanson,K.M., Chen,H.T., Graham,J.C., Wojnar,D.M., Petras,A. 2009. Resolution of depression and 1
grief during the first year after miscarriage: a randomized controlled clinical trial of couples-focused 2
interventions. Journal of Women's Health; 8: 1245-1257 3
Tahseen,S., Wyldes,M. 2003. A comparative case-controlled study of laparoscopic vs laparotomy 4
management of ectopic pregnancy: an evaluation of reproductive performance after radical vs 5
conservative treatment of tubal ectopic pregnancy. Journal of Obstetrics and Gynaecology; 2:189-190 6
Tam,W.H., Tsui,M.H., Lok,I.H., Yip,S.K., Yuen,P.M., Chung,T.K. 2005. Long-term reproductive 7
outcome subsequent to medical versus surgical treatment for miscarriage. Human Reproduction; 8
12:3355-3359 9
Tang,O.S., Lau,W.N., Ng,E.H., Lee,S.W., Ho,P.C. 2003. A prospective randomized study to compare 10
the use of repeated doses of vaginal with sublingual misoprostol in the management of first trimester 11
silent miscarriages. Human Reproduction; 1:176-181 12
Tang,O.S., Ong,C.Y., Tse,K.Y., Ng,E.H., Lee,S.W., Ho,P.C. 2006. A randomized trial to compare the 13
use of sublingual misoprostol with or without an additional 1 week course for the management of first 14
trimester silent miscarriage. Human Reproduction; 1:189-192 15
Tanha,F.D., Feizi,M., Shariat,M. 2010. Sublingual versus vaginal misoprostol for the management of 16
missed abortion. Journal of Obstetrics and Gynaecology Research; 3:525-532 17
Taylor,J., Diop,A., Blum,J., Dolo,O., Winikoff,B. 2011. Oral misoprostol as an alternative to surgical 18
management for incomplete abortion in Ghana. International Journal of Gynaecology and Obstetrics; 19
1:40-44 20
Thorburn,J., Bryman,I., Hahlin,M., Lindblom,B. 1992. Differential diagnosis of early human 21
pregnancies: impact of different diagnostic measures. Gynecologic and Obstetric Investigation; 4:216- 22
220 23
Thorsen,M.K., Lawson,T.L., Aiman,E.J., Miller,D.P., McAsey,M.E., Erickson,S.J., Quiroz,F., 24
Perret,R.S.1990. Diagnosis of ectopic pregnancy: Endovaginal vs transabdominal sonography. 25
American Journal of Roentgenology; 2:307-310 26
Trinder,J., Brocklehurst,P., Porter,R., Read,M., Vyas,S., Smith,L. 2006. Management of miscarriage: 27
expectant, medical, or surgical? Results of randomised controlled trial (miscarriage treatment (MIST) 28
trial). BMJ; 7552:1235-1240 29
Trinder,J., Brocklehurst,P., Porter,R., Read,M., Vyas,S., Smith,L. 2006. Management of miscarriage: 30
expectant, medical, or surgical? Results of randomised controlled trial (miscarriage treatment (MIST) 31
trial). BMJ; 7552:1235-1240 32
Tsai,H.D., Chen,H.Y., Yeh,L.S. 1995. A 12-year survey of 681 ectopic pregnancies. Chung 33
Tulandi,T., Guralnick,M. 1991. Treatment of tubal ectopic pregnancy by salpingotomy with or without 34
tubal suturing and salpingectomy [Erratum appears in Fertil Steril 1991 Jun;55(6):1213-4]. Fertility 35
and Sterility; 1:53-55 36
Tunde-Byass,M., Cheung,V.Y. 2009. The value of the early pregnancy assessment clinic in the 37
management of early pregnancy complications. Journal of Obstetrics and Gynaecology Canada: 38
JOGC; 9:841-844 39
Tunde-Byass,M., Cheung,V.Y. 2009. The value of the early pregnancy assessment clinic in the 40
management of early pregnancy complications. Journal of Obstetrics and Gynaecology Canada: 41
JOGC; 9:841-844 42
Tuomivaara,L., Kauppila,A. 1988. Radical or conservative surgery for ectopic pregnancy? A follow-up 43
study of fertility of 323 patients. Fertility and Sterility; 4:580-583 44
Turan,V. 2011. Fertility outcomes subsequent to treatment of tubal ectopic pregnancy in younger 45
Turkish women. Journal of Pediatric and Adolescent Gynecology; 5:251-255 46
Twigg,J., Moshy,R., Walker,J.J., Evans,J. 2003. Early pregnancy assessment units in the United 47
Kingdom: An audit of current clinical practice. Journal of Clinical Excellence; 4:391-402 48
References

251
Vermesh,M., Presser,S.C. 1992. Reproductive outcome after linear salpingostomy for ectopic 1
gestation: a prospective 3-year follow-up. Fertility and Sterility; 3:682-684 2
Vermesh,M., Silva,P.D., Rosen,G.F., Stein,A.L., Fossum,G.T., Sauer,M.V. 1989. Management of 3
unruptured ectopic gestation by linear salpingostomy: a prospective, randomized clinical trial of 4
laparoscopy versus laparotomy. Obstetrics and Gynecology; 3(1):400-40 5
Visscher,R.D., Visscher,H.C. 1972. Do Rh-negative women with an early spontaneous abortion need 6
Rh immune prophylaxis? American Journal of Obstetrics and Gynecology; 2:158-165 7
Walsh,J.J., Lewis,B.V. 1970. Transplacental haemorrhage due to termination of pregnancy. Journal of 8
Obstetrics and Gynaecology of the British Commonwealth; 2:133-136 9
Wieringa-de Waard, M., Hartman,E.E., Ankum,W.M., Reitsma,J.B., Bindels,P.J., Bonsel,G.J. 2002b. 10
Expectant management versus surgical evacuation in first trimester miscarriage: health-related quality 11
of life in randomized and non-randomized patients. Human Reproduction; 6:1638-1642 12
Wieringa-de Waard, M., Vos,J., Bonsel,G.J., Bindels,P.J., Ankum,W.M. 2002a. Management of 13
miscarriage: a randomized controlled trial of expectant management versus surgical evacuation. 14
Human Reproduction; 9:2445-2450 15
Wong,E., Suat,S.O. 2000. Ectopic pregnancy--a diagnostic challenge in the emergency department. 16
European Journal of Emergency Medicine; 3:189-194 17
Wood,S.L., Brain,P.H. 2002. Medical management of missed abortion: a randomized clinical trial. 18
[Erratum appears in Obstet Gynecol 2002 Jul;100(1):175 Note: Dosage error in published abstract; 19
MEDLINE/PubMed abstract corrected]. Obstetrics and Gynecology; 4:563-566 20
Yao,M., Tulandi,T., Kaplow,M., Smith,A.P. 1996. A comparison of methotrexate versus laparoscopic 21
surgery for the treatment of ectopic pregnancy: a cost analysis. Human Reproduction; 12:2762-2766 22
You,J.H., Chung,T.K. 2005. Expectant, medical or surgical treatment for spontaneous abortion in first 23
trimester of pregnancy: a cost analysis. Human Reproduction; 10:2873-2878 24
Zhang,J., Gilles,J.M., Barnhart,K., Creinin,M.D., Westhoff,C., Frederick,M.M., National Institute of 25
Child Health Human Development (NICHD) Management of Early Pregnancy Failure Trial. 2005. A 26
comparison of medical management with misoprostol and surgical management for early pregnancy 27
failure. New England Journal of Medicine; 8:761-769 28
Zilber,U., Pansky,M., Bukovsky,I., Golan,A. 1996. Laparoscopic salpingostomy versus laparoscopic 29
local methotrexate injection in the management of unruptured ectopic gestation. American Journal of 30
Obstetrics and Gynecology; 3 (1):600-602 31

12 Abbreviations and 1
glossary 2
12.1 Abbreviations 3
A&E accident and emergency unit
AEPU Association of Early Pregnancy Units
AGU acute gynaecology unit
CI confidence interval
CRL crown-rump length
D&C dilation and curettage
DASS Depression Anxiety Stress Scales
ED emergency department
EDIS emergency department information system
EP ectopic pregnancy
EPAC early pregnancy assessment clinic
EP(A)U early pregnancy (assessment) unit
EPPS early pregnancy problem service
ER emergency room
ERPC evacuation of retained products of conception
EUP extra-uterine pregnancy
EVA electric vacuum aspiration
FMH fetomaternal haemorrhage
GP general practitioner
GRADE Grading of Recommendations Assessment, Development and Evaluation
HADS Hospital Anxiety and Depression Scale
hCG -human chorionic gonadotrophin
HCP healthcare professional
HSG hysterosalpingography
ICER incremental cost effectiveness ratio
ICSI intracytoplasmic sperm injection
ICT Indirect Coombs test
IM intramuscular
IMI intramuscular injection
IPUV intrauterine pregnancy of uncertain viability
Abbreviations and glossary
253
IU international unit(s)
IUCD intrauterine contraceptive device
IUP intrauterine pregnancy
IV intravenous
IVF in-vitro fertilisation
MAC monitored anaesthesia care
Mf mifepristone
Ms misoprostol
MSD mean gestational sac diameter
MTX methotrexate
MVA(C) manual vacuum aspiration (curettage)
NC not calculable
NR not reported
NS not significant
PFC plaque-forming cells
PID pelvic inflammatory disease
PUL pregnancy of unknown location
RBC red blood cells
RCOG Royal College of Obstetricians and Gynaecologists
Rh rhesus
RPOC retained products of conception
SC suction curettage
SD standard deviation
SEM standard error of the mean
SF-36 Short Form-36
SHO senior house officer
STAI State-Trait Anxiety Inventory
STI sexually transmitted infection
TAS transabdominal scan
TAU transabdominal ultrasound
TB tuberculosis
TVS transvaginal scan
TVU transvaginal ultrasound
US ultrasound
VAS visual analogue scale
WHO World Health Organization
1
254
12.2 Glossary 1
Active
treatment/management
Intervention to manage a condition, usually in the form of surgery or
management with drugs.
Anti-D Immunoglobulin that binds to, and causes the removal of, any Rhesus D
positive red blood cells that have passed from the fetus into the maternal
circulation
Ectopic pregnancy A pregnancy located outside of the uterine cavity, usually in the fallopian tube
Electric vacuum aspiration The use of suction, generated using an electric pump, to remove the contents
of the uterus through the cervix
Expectant management Management approach in which a patient is closely monitored (for example
with observations, scans or blood tests) but treatment is not administered,
waiting to assess whether the condition will resolve naturally.
hCG ratio A measure of the pattern of increase or decrease in hCG levels, calculated by
dividing one hCG measurement by the previous measurement
Hospital Anxiety and
Depression scale
A 10 point scale used to determine levels of anxiety and depression
experienced by patients. A score of 1 indicates a potential need for psychiatric
treatment, whereas a score of 10 represents clinical stability.
Human chorionic
gonadotrophin
Glycoprotein hormone produced during pregnancy by the embryo and later the
placenta. It can be detected using blood tests or urine pregnancy tests.
Incomplete miscarriage A diagnosed non-viable pregnancy in which bleeding has begun, but
pregnancy tissue remains in the uterus.
Indirect Coombs test A test to detect antibodies against red blood cell antigens, e.g. it can be used
to detect antibodies against Rhesus D
Inevitable miscarriage A diagnosed non-viable pregnancy in which bleeding has begun and the
cervical os is open, but pregnancy tissue remains in the uterus
Intrauterine pregnancy of
uncertain viability
A pregnancy that is located within the uterus, without a visible heartbeat and
which cannot be diagnosed as a confirmed miscarriage following an ultrasound
scan.
Kleihauer test A method of quantifying the amount of fetal haemoglobin that has passed into
the maternal bloodstream.
Laparoscopy Surgical operation performed within the abdomen or pelvis using cameras and
instruments inserted through small incisions.
Laparotomy Open surgical procedure, involving a surgical incision into the abdomen.
Manual vacuum aspiration The use of suction, generated using a manual pump, to remove the contents of
the uterus through the cervix
Methotrexate Anti-metabolite drug which can stop the development of a pregnancy and
therefore has been used in the medical management of ectopic pregnancy.
Methotrexate can be administered systemically or locally at the site of the
pregnancy (either ultrasound guided or using laparoscopic injection).
Missed miscarriage A non-viable pregnancy identified on ultrasound scan, without associated pain
and bleeding (also known as early fetal demise, delayed miscarriage or silent
miscarriage)
Os The opening of the cervix
Pregnancy of unknown
location
A descriptive term used to classify a pregnancy when a woman has a positive
pregnancy test but no pregnancy can be visualised on ultrasound scan
Recurrent miscarriage The loss of three or more pregnancies before 23
+6
weeks of gestation
Abbreviations and glossary
255
RhoGAM Brand of rhesus D immune globulin.
Salpingectomy Surgical removal of the Fallopian tube.
Salpingo-oophorectomy Surgical removal of the Fallopian tube and ovary.
Salpingostomy Surgical formation of an opening of Fallopian tube where the fimbrial end (the
end closest to the ovaries) has been closed by infection or chronic
inflammation.
Salpingotomy Surgical incision of a Fallopian tube to remove an ectopic pregnancy.
Sensitisation Development of antibodies against antigens found on fetal red blood cells.
Short Form-36 scale A survey of patient health, consisting of eight scales: vitality, physical
functioning, bodily pain, general health perceptions, physical role functioning,
emotional role functioning, social role functioning and mental health. The
scores are summed and transformed into a scale of 0-100. The scale can also
be divided into two aggregate measures, the Physical Component Summary
and the Mental Component Summary.
State-Trait Anxiety Inventory A questionnaire containing two 20-item scales covering both current (state)
and background (trait) anxiety. Items are rated on a 4-point scale with total
scores ranging from 20 to 80 where higher scores represent higher levels of
anxiety.
Surgical completion of
miscarriage
Treatment of a miscarriage using a surgical procedure (e.g. dilation and
curettage, vacuum aspiration); also known as evacuation of retained products
of conception.
Threatened miscarriage Vaginal bleeding in the presence of a viable pregnancy in the first 23
+6
weeks
of pregnancy
1

Bleeding Preg

Uploaded by

Copyright:

Available Formats

Bleeding Preg

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Bleeding Preg

Uploaded by

Copyright:

Available Formats

What is the guideline about?

What methods are discussed for diagnosing ectopic pregnancy and miscarriage?

DRAFT FOR CONSULTATION

as a first-line treatment to women

, the group agreed that mifepristone should 39

as a first-line treatment to women who have all of

. An incremental cost-effectiveness ratio (ICER) was then calculated 8

You might also like