The n e w e ng l a n d j o u r na l of m e dic i n e

Original Article

A Randomized Trial of Progesterone

in Women with Bleeding in Early Pregnancy
A. Coomarasamy, A.J. Devall, V. Cheed, H. Harb, L.J. Middleton, I.D. Gallos,
H. Williams, A.K. Eapen, T. Roberts, C.C. Ogwulu, I. Goranitis, J.P. Daniels,
A. Ahmed, R. Bender‑Atik, K. Bhatia, C. Bottomley, J. Brewin, M. Choudhary,
F. Crosfill, S. Deb, W.C. Duncan, A. Ewer, K. Hinshaw, T. Holland, F. Izzat,
J. Johns, K. Kriedt, M.-A. Lumsden, P. Manda, J.E. Norman, N. Nunes,
C.E. Overton, S. Quenby, S. Rao, J. Ross, A. Shahid, M. Underwood,
N. Vaithilingam, L. Watkins, C. Wykes, A. Horne, and D. Jurkovic​​

A BS T R AC T

BACKGROUND
Bleeding in early pregnancy is strongly associated with pregnancy loss. Progester- The authors’ full names, academic de‑
one is essential for the maintenance of pregnancy. Several small trials have sug- grees, and affiliations are listed in the
Appendix. Address reprint requests to
gested that progesterone therapy may improve pregnancy outcomes in women who Dr. Coomarasamy at the Institute of Me‑
have bleeding in early pregnancy. tabolism and Systems Research, College
of Medical and Dental Sciences, Univer‑
METHODS sity of Birmingham Edgbaston, Birming‑
ham B15 2TT, United Kingdom, or at
We conducted a multicenter, randomized, double-blind, placebo-controlled trial to a​.­coomarasamy@​­bham​.­ac​.­uk.
evaluate progesterone, as compared with placebo, in women with vaginal bleeding
N Engl J Med 2019;380:1815-24.
in early pregnancy. Women were randomly assigned to receive vaginal supposito- DOI: 10.1056/NEJMoa1813730
ries containing either 400 mg of progesterone or matching placebo twice daily, Copyright © 2019 Massachusetts Medical Society.
from the time at which they presented with bleeding through 16 weeks of gesta-
tion. The primary outcome was the birth of a live-born baby after at least 34 weeks
of gestation. The primary analysis was performed in all participants for whom
data on the primary outcome were available. A sensitivity analysis of the primary
outcome that included all the participants was performed with the use of multiple
imputation to account for missing data.
RESULTS
A total of 4153 women, recruited at 48 hospitals in the United Kingdom, were
randomly assigned to receive progesterone (2079 women) or placebo (2074 women).
The percentage of women with available data for the primary outcome was 97%
(4038 of 4153 women). The incidence of live births after at least 34 weeks of gesta-
tion was 75% (1513 of 2025 women) in the progesterone group and 72% (1459 of
2013 women) in the placebo group (relative rate, 1.03; 95% confidence interval
[CI], 1.00 to 1.07; P = 0.08). The sensitivity analysis, in which missing primary
outcome data were imputed, resulted in a similar finding (relative rate, 1.03; 95%
CI, 1.00 to 1.07; P = 0.08). The incidence of adverse events did not differ signifi-
cantly between the groups.
CONCLUSIONS
Among women with bleeding in early pregnancy, progesterone therapy adminis-
tered during the first trimester did not result in a significantly higher incidence of
live births than placebo. (Funded by the United Kingdom National Institute for
Health Research Health Technology Assessment program; PRISM Current Con-
trolled Trials number, ISRCTN14163439.)

n engl j med 380;19 nejm.org  May 9, 2019 1815

The New England Journal of Medicine
Downloaded from nejm.org by ampri yuwana loyra on May 23, 2019. For personal use only. No other uses without permission.
Copyright © 2019 Massachusetts Medical Society. All rights reserved.
 The n e w e ng l a n d j o u r na l of m e dic i n e

M
iscarriage affects one in five ing. Women who have experienced at least three
pregnancies.1 Miscarriage can cause ex- prior pregnancy losses, however, may benefit from
cessive bleeding, infection, and com- progesterone therapy in the first trimester.”9 We
plications associated with surgical treatment,2 conducted the multicenter, randomized, parallel-
as well as substantial psychological harm, in- group, double-blind, placebo-controlled PRISM
cluding anxiety, depression, and post-traumatic (Progesterone in Spontaneous Miscarriage) trial
stress disorder.3,4 to investigate whether treatment with progester-
Progesterone, which is produced by the corpus one would result in a higher incidence of live
luteum in the ovary, is necessary to prepare the births among women with bleeding in early
endometrium for implantation of the embryo pregnancy than placebo.
and thus is an essential hormone for a success-
ful pregnancy. Additional progesterone is pro- Me thods
duced when an embryo implants in the endome-
trium and during early placental development. Trial Oversight
Subsequently, beginning at approximately 12 weeks The PRISM trial was approved by the United
of pregnancy, the placenta becomes the dominant Kingdom Medicines and Healthcare Products
source of progesterone.5 Regulatory Agency, the United Kingdom National
The physiological importance of progesterone Research Ethics Service Committee (South Cen-
has prompted researchers, physicians, and pa- tral Oxford), and the National Health Service
tients to consider progesterone supplementation research and development department at each
during early pregnancy to prevent miscarriages. participating hospital. The trial was conducted
Progesterone supplementation in early pregnancy at clinics that are part of the trial research net-
has been attempted in two contexts: the first is work of the Tommy’s National Centre for Mis-
to prevent miscarriages in asymptomatic women carriage Research, which is funded by Tommy’s
who have a history of recurrent miscarriages, Charity. Progesterone and placebo were pur-
and the second is to rescue a pregnancy in chased from Besins Healthcare. This company
women who have started to bleed during early had no role in the design of the trial; in the
pregnancy.6 We addressed the first scenario in a collection, analysis, or interpretation of the data;
previous issue of the Journal7 and found no bene­ or in the preparation of the manuscript. Trial
ficial effect of progesterone in women with a oversight and monitoring were provided by a
history of unexplained recurrent miscarriages. trial steering committee and by an independent
The current trial focuses on women with vaginal data and safety monitoring committee. The first,
bleeding in early pregnancy. second, and last authors vouch for the accuracy
A Cochrane review (originally published in and completeness of the data and analyses and
2007 and last updated in 2018)6 of 7 randomized for the fidelity of the trial to the protocol, avail-
trials of progestational agents that involved able with the full text of this article at NEJM.org.
women with bleeding in early pregnancy showed
a significantly lower risk of miscarriages among Trial Participants
women who received progesterone than among The participants in the PRISM trial were re-
those who received placebo or no treatment cruited at 48 hospitals in the United Kingdom.
(odds ratio, 0.64; 95% confidence interval [CI], Women were eligible for enrollment in the trial
0.47 to 0.87) but noted that the trials were small if they were 16 to 39 years of age, if they had
(the largest trial had a sample size of 191) and completed less than 12 weeks of pregnancy, if
had methodologic weaknesses. Another Cochrane they presented with vaginal bleeding, and if they
review of 13 trials of progestational agents that had an intrauterine gestational sac that was vis-
involved women with recurrent miscarriages was ible on ultrasonography. The upper threshold of
originally published in 2003 and was last up- 39 years for maternal age was chosen because
dated in 2018.8 The American College of Obste- the probability of miscarriages due to chromo-
tricians and Gynecologists reviewed the evidence somal abnormalities increases with advancing
and concluded, “For threatened early pregnancy age,10 and progesterone treatment could not be
loss, the use of progestins is controversial, and expected to prevent such miscarriages. Partici-
conclusive evidence supporting their use is lack- pants were excluded if at the time of presenta-

1816 n engl j med 380;19 nejm.org  May 9, 2019

The New England Journal of Medicine

Downloaded from nejm.org by ampri yuwana loyra on May 23, 2019. For personal use only. No other uses without permission.
Copyright © 2019 Massachusetts Medical Society. All rights reserved.
 Progesterone in Women with Bleeding in Early Pregnancy

tion the fetal crown–rump length was 7 mm or ongoing pregnancy at 12 weeks of gestation,
longer with no visible heartbeat; if the gesta- miscarriage (defined as loss of pregnancy before
tional sac was a mean of 25 mm or greater in 24 weeks of gestation), live birth before 34 weeks
diameter with no visible fetal pole on ultraso- of gestation, ectopic pregnancy, stillbirth (defined
nography; if they had evidence of ectopic preg- as intrauterine death after at least 24 weeks of
nancy; if they had life-threatening bleeding; if gestation), termination of pregnancy, the week of
they had current or recent use of progesterone gestation at delivery, birth weight, size (small or
supplementation; if they had contraindications large) for gestational age, preeclampsia, Apgar
to progesterone therapy (i.e., a history of liver scores, survival at 28 days of neonatal life, and
tumors; current genital or breast cancer, severe congenital abnormalities, as well as other ante-
arterial disease, or acute porphyria; or a history natal, intrapartum, postpartum, and neonatal
during pregnancy of idiopathic jaundice, severe outcomes. A detailed list of all secondary out-
pruritus, or pemphigoid gestationis); or if they comes is provided in the Supplementary Appen-
were participating in any other blinded, placebo- dix, available at NEJM.org. We attempted to
controlled trials of medicinal products in preg- collect outcome data for all participants who
nancy. All the participants provided written in- underwent randomization, regardless of adher-
formed consent. ence to the trial-group assignment.

Trial Assignments Statistical Analysis

Participants were randomly assigned, in a 1:1 We calculated that 1972 women would need to
ratio, to administer to themselves vaginal supposi- be included in each trial group to provide 90%
tories containing either 400 mg of micronized power to detect a minimally important absolute
progesterone (Utrogestan, Besins Healthcare) difference of 5 percentage points between the
or matching placebo twice daily, from the time progesterone group and the placebo group in the
of randomization through 16 completed weeks of incidence of live births after at least 34 weeks of
gestation (or earlier if pregnancy ended before gestation (65% vs. 60%), at a two-sided alpha
16 weeks). If vaginal administration was not level of 0.05. This minimally important differ-
preferred, participants could administer the sup- ence was chosen on the basis of a national survey
positories rectally. Randomization was performed of clinical practitioners in the United Kingdom.
through a secure, centralized Internet facility We planned to include 4150 women in the trial
with the use of minimization to balance the to account for an expected 5% loss to follow-up.
trial-group assignments according to maternal The analysis of the primary outcome was
age (<35 years vs. ≥35 years), body-mass index performed according to the intention-to-treat
(BMI [the weight in kilograms divided by the principle; a main analysis of all available data
square of the height in meters], <30 vs. ≥30), was supplemented by a sensitivity analysis of the
fetal heart activity (present vs. absent), estimated primary outcome that included all participants
gestation at presentation (<42 days vs. ≥42 days), and took into account any missing data with the
and amount of vaginal bleeding (pictorial blood- use of multiple imputation.12 A Poisson regres-
loss assessment chart [PBAC] score of ≤2 vs. sion model with robust standard errors was used
score of ≥3; scores range from 1 to 4, with to estimate the relative rates and corresponding
higher scores indicating greater vaginal blood 95% two-sided confidence intervals, with adjust-
loss).11 The appearance, route, and timing of ad- ment for the minimization variables. This meth-
ministration of progesterone and placebo were od has been shown to be appropriate and to be
identical. Participants, physicians, and trial nurses less prone to convergence issues than other
were unaware of the trial-group assignments similar methods.13
throughout the course of the trial. For the primary outcome, a P value was gener-
ated with the use of a two-sided chi-square test.
Outcome Measures The statistical analysis plan did not include a
The primary outcome was the birth of a live- provision for correction for multiplicity when
born baby after at least 34 completed weeks of the analyses of the secondary outcomes were
gestation. Secondary outcomes included the time performed. Therefore, the results are reported as
from conception to the end date of pregnancy, point estimates and 95% confidence intervals,

n engl j med 380;19 nejm.org  May 9, 2019 1817

The New England Journal of Medicine
Downloaded from nejm.org by ampri yuwana loyra on May 23, 2019. For personal use only. No other uses without permission.
Copyright © 2019 Massachusetts Medical Society. All rights reserved.
 The n e w e ng l a n d j o u r na l of m e dic i n e

without P values. For continuous outcomes, a the suppositories vaginally and 1% (51 of 3662
linear regression model was used to estimate women) administered them rectally.
mean differences, with the same adjustment that
was used in the analysis of the primary out- Outcomes
come. The widths of the confidence intervals The incidence of live births after at least 34 weeks
were not adjusted for multiplicity, so the inter- of gestation was 75% (1513 of 2025 women) in
vals should not be used to infer definitive treat- the progesterone group and 72% (1459 of 2013
ment effects. women) in the placebo group (relative rate, 1.03;
We analyzed the treatment effect on the pri- 95% CI, 1.00 to 1.07; P = 0.08). The sensitivity
mary outcome in prespecified subgroups defined analysis, in which multiple imputation was used
according to maternal age (<35 years vs. ≥35 for missing data, did not change the findings
years), BMI (<30 vs. ≥30), fetal heart activity (relative rate, 1.03; 95% CI, 1.00 to 1.07; P = 0.08).
(present vs. absent), estimated gestation at pre- The incidence of ongoing pregnancy at 12
sentation (<6 weeks vs. 6 to <9 weeks vs. ≥9 weeks was 83% (1672 of 2025 women) in the
weeks), amount of vaginal bleeding (PBAC score progesterone group and 80% (1602 of 2013
of ≤2 vs. score of ≥3),11 number of previous mis- women) in the placebo group (relative rate, 1.04;
carriages (0 vs. 1 or 2 vs. ≥3), number of gesta- 95% CI, 1.01 to 1.07). The incidence of miscar-
tional sacs (1 vs. ≥2), race (white, black, south riage was 20% (410 of 2025 women) in the pro-
Asian, or other), history of polycystic ovaries (yes gesterone group and 22% (451 of 2013 women)
vs. no), and previous cervical excision (yes vs. no). in the placebo group (relative rate, 0.91; 95% CI,
The effects of these subgroups were examined 0.81 to 1.01). The results of all the other second-
by adding the variables for the interaction of ary outcomes are presented in Table 2, and in
subgroup with trial group to the regression Table S2 in the Supplementary Appendix.
model; a chi-square test was used to determine A significant subgroup effect was identified
whether the effects of progesterone and placebo for only 1 of the 10 prespecified subgroups —
differed in the various subgroups. the subgroup of participants defined according
Interim analyses of principal safety and ef- to the number of previous miscarriages. The inci-
fectiveness outcomes were performed on behalf dence of live births in the subgroup of women
of the data and safety monitoring committee by who had no previous miscarriages was 74% in
the trial statistician (who remained unaware the progesterone group and 75% in the placebo
of the treatment assignments) on two occasions. group (relative rate, 0.99; 95% CI, 0.95 to 1.04);
Because these analyses were performed with the the incidence among women who had one or
use of the Peto principle,14 no adjustment was two previous miscarriages was 76% and 72%,
made in the final P values to determine signifi- respectively (relative rate 1.05; 95% CI, 1.00 to
cance. 1.12); and the incidence among women who had
three or more previous miscarriages was 72%
R e sult s and 57%, respectively (relative rate, 1.28; 95%
CI, 1.08 to 1.51) (P = 0.007 for the interaction
Trial Participants between trial group and the number of miscar-
From May 19, 2015, through July 27, 2017, a total riages) (Fig. 2). The results of two post hoc sub-
of 12,862 women were identified as being eligi- group analyses in which we categorized the
ble for the PRISM trial; of these women, 4153 number of previous miscarriages differently from
were randomly assigned to receive either proges- the subgroup analysis described here are provided
terone (2079 women) or placebo (2074 women) in Figure S1 in the Supplementary Appendix.
(Fig. 1). The percentage of women with available There was no significant between-group dif-
data for the primary outcome was 97% (4038 of ference in the percentage of participants who
4153 women). Demographic and baseline char- had either a maternal or neonatal serious ad-
acteristics were similar in the two trial groups verse event (5% [105 of 2025 participants] in the
(Table 1, and Table S1 in the Supplementary Ap- progesterone group and 5% [98 of 2013 partici-
pendix). Information on the route of administra- pants] in the placebo group), including specifi-
tion was available for 88% (3662 of 4153) of the cally the percentage of babies who had neonatal
women: 99% (3611 of 3662 women) administered congenital abnormalities (3.4% in each group),

1818 n engl j med 380;19 nejm.org  May 9, 2019

The New England Journal of Medicine

Downloaded from nejm.org by ampri yuwana loyra on May 23, 2019. For personal use only. No other uses without permission.
Copyright © 2019 Massachusetts Medical Society. All rights reserved.
 Progesterone in Women with Bleeding in Early Pregnancy

23,775 Patients were assessed for eligibility

10,913 Were not eligible

3373 Had vaginal bleeding stop >4 days
before screening
2756 Had last menstrual period >12 weeks
before screening
1444 Had evidence of ectopic pregnancy
1323 Were >39 yr of age
811 Had current or recent use of progesterone
supplement
530 Were <16 yr of age
476 Were unable to understand English
115 Were participating in another clinical trial
65 Had contraindication to progesterone
therapy
20 Had life-threatening bleeding

12,862 Were eligible

8709 Declined to participate

4153 Underwent randomization

2079 Were assigned to receive 2074 Were assigned to receive

progesterone placebo

10 Withdrew 20 Withdrew
44 Were lost to follow-up 41 Were lost to follow-up

2025 Had data available for analysis of 2013 Had data available for analysis of
primary outcome primary outcome
54 Had missing responses imputed 61 Had missing responses imputed
and were included in sensitivity analysis and were included in sensitivity analysis

Figure 1. Enrollment, Randomization, and Follow-up.

nor was there any significant between-group dif- cantly higher incidence of live births after at
ference in the number of maternal or neonatal least 34 weeks of gestation than placebo. There
serious adverse events. A summary of serious was also no significant difference between the
adverse events is provided in Table S3 in the groups in the incidence of miscarriage or still-
Supplementary Appendix. birth. Although there appeared to be slightly
more ongoing pregnancies at 12 weeks in the
progesterone group than in the placebo group,
Discussion
an inference of benefit cannot be drawn because
Our large multicenter, randomized, double-blind, the confidence interval for the relative rate was
placebo-controlled trial showed that among not adjusted for multiplicity of testing.
women with bleeding in early pregnancy, pro- The large sample size in our trial allowed in-
gesterone therapy administered during the first vestigation of the primary outcome in prespeci-
trimester of pregnancy did not result in a signifi- fied subgroups. Among the 10 subgroup analyses,

n engl j med 380;19 nejm.org  May 9, 2019 1819

The New England Journal of Medicine
Downloaded from nejm.org by ampri yuwana loyra on May 23, 2019. For personal use only. No other uses without permission.
Copyright © 2019 Massachusetts Medical Society. All rights reserved.
 The n e w e ng l a n d j o u r na l of m e dic i n e

Table 1. Demographic and Baseline Characteristics of the Participants.*

Progesterone Placebo
Characteristic (N = 2079) (N = 2074)
Demographic characteristics
Maternal age†
Mean — yr 30.6±5.1 30.5±5.1
Distribution — no. (%)
<35 yr 1604 (77) 1601 (77)
≥35 yr   475 (23)   473 (23)
Body-mass index†
Mean 26.4±6.2 26.5±6.3
Distribution — no. (%)
<30 1589 (76) 1589 (77)
≥30 490 (24)   485 (23)
Race — no. (%)‡
White 1714 (82) 1742 (84)
Black   84 (4)   79 (4)
South Asian 114 (5) 102 (5)
Other 167 (8) 151 (7)
Pregnancy history
Nulliparous — no. (%)   474 (23) 514 (25)
Previous preterm births at ≥24 wk to <34 wk — no. (%)   83 (4) 90 (4)
Previous miscarriages at <24 wk of gestation — no. (%)
0 1145 (55) 1157 (56)
1 or 2   792 (38) 758 (37)
≥3 142 (7) 159 (8)
No. of previous miscarriages — median (IQR) 0 (0–1) 0 (0–1)
Amount of bleeding as assessed by PBAC score†§ — no. (%)
≤2 1913 (92) 1907 (92)
≥3 166 (8) 167 (8)
Fetal heart activity at presentation — no. (%)†¶ 1710 (82) 1701 (82)

* Plus–minus values are means ±SD. There were no significant differences between the two groups in the characteristics
listed. Further details regarding baseline characteristics are provided in Table S1 in the Supplementary Appendix. Per­
centages may not sum to 100 because of rounding. IQR denotes interquartile range.
† This variable was a minimization variable.
‡ Race was reported by the participant.
§ Pictorial blood-loss assessment chart (PBAC) scores range from 1 to 4, with higher scores indicating greater vaginal
blood loss.11
¶ If more than one fetus was present, this variable was characterized as any fetus that had heart activity at presentation.

1 showed differential effects of progesterone: the increase in the loss of chromosomally normal
effect of progesterone in women with bleeding pregnancies (i.e., euploid miscarriages) with in-
in early pregnancy differed according to the creasing number of previous miscarriages.15
number of previous miscarriages, with a sugges- Given that the potential benefit of progesterone
tion of benefit among women who had had therapy would be expected to be specific to eu-
three or more previous miscarriages. Previous ploid pregnancies, an increasing level of benefit
reports have indicated a steep and proportionate in women with increasing number of previous

1820 n engl j med 380;19 nejm.org  May 9, 2019

The New England Journal of Medicine

Downloaded from nejm.org by ampri yuwana loyra on May 23, 2019. For personal use only. No other uses without permission.
Copyright © 2019 Massachusetts Medical Society. All rights reserved.
 Progesterone in Women with Bleeding in Early Pregnancy

Table 2. Primary Outcome and Secondary Outcomes.*

Relative Rate or
Progesterone Placebo Mean Difference
Outcome (N = 2025) (N = 2013) (95% CI)†
Primary outcome — no. (%)
Live birth at ≥34 wk 1513 (75) 1459 (72) 1.03 (1.00 to 1.07)‡
Secondary maternal outcomes — no. (%)§
Ongoing pregnancy at 12 wk 1672 (83) 1602 (80) 1.04 (1.01 to 1.07)
Miscarriage, defined as loss of pregnancy at <24 wk¶   410 (20) 451 (22) 0.91 (0.81 to 1.01)
Live birth at <34 wk   68 (3) 64 (3) 1.06 (0.76 to 1.49)
Ectopic pregnancy 0 2 (<1) —
Stillbirth, defined as intrauterine death at ≥24 wk     5 (<1) 6 (<1) 0.82 (0.25 to 2.66)
Termination of pregnancy‖   34 (2) 36 (2) 0.94 (0.59 to 1.50)
Secondary neonatal outcomes among women with live
births at ≥24 wk§
Gestational age at delivery**
Wk of gestation 38 wk 4 days±2 wk 38 wk 4 days±2 wk 0.11 days (−0 wk 1 day
4 days 3 days to 0 wk 2 days)†
No. of women 1581 1521
Birth weight††
Mean weight — g 3242±656 3261±659 −21 (−67 to 25)†
No. of infants 1604 1539
Death at 28 days of neonatal life — no./total no. (%)‡‡ 8/1605 (<1) 2/1533 (<1) 3.84 (0.80 to 18.40)†

* Plus–minus values are means ±SD. Additional data on outcomes are provided in Table S2 in the Supplementary Appendix.
† Relative rates are shown for the primary outcome, all secondary maternal outcomes, and the secondary neonatal outcome of death at 28
days of neonatal life. The mean difference is shown for the secondary neonatal outcomes of gestational age at delivery and birth weight.
For binary outcomes, a relative rate of less than 1 favors the progesterone group, except for live birth after at least 34 weeks of gestation
and ongoing pregnancy at 12 weeks, for which a relative rate greater than 1 would favor progesterone. For continuous outcomes, a mean
difference of less than 0 favors the progesterone group. The widths of the confidence intervals were not adjusted for multiplicity, so the
­intervals should not be used to infer definitive treatment effects.
‡ P = 0.08.
§ Five women in the progesterone group and three women in the placebo group had both a live birth after at least 34 weeks of gestation and
a miscarriage; one woman in the placebo group had both a termination of pregnancy and a miscarriage; and one woman in the placebo
group had both a live birth before 34 weeks and a stillbirth.
¶ The median gestational age was 8 weeks (interquartile range, 7 to 10) in both trial groups.
‖ The reasons for termination of pregnancy in the progesterone group were social for 13 women and medical for 21 women. The reasons
for termination of pregnancy in the placebo group were social for 12 women and medical for 24 women. The median gestational age was
14 weeks (interquartile range, 12 to 19) in the progesterone group and 15 weeks (interquartile range, 11 to 18) in the placebo group.
** The gestational age at delivery was unknown for the infants of two women in the placebo group.
†† The birth weight was unknown for the infants of 6 women in the progesterone group and 6 women in the placebo group.
‡‡ The neonatal vital status at 28 days of life was unknown for 17 women (5 in the progesterone group and 12 in the placebo group).

miscarriages is consistent with our understand- our statistical analysis plan,17 and multiple com-
ing of the biologic factors associated with risk of parisons were performed (without adjustment
miscarriage. A history of miscarriage is one of for multiplicity); thus, this observation requires
only two stratification or prognostic risk factors validation.
(the other being maternal age) cited in the 2017 Some limitations of our trial should be con-
guideline of the European Society of Human sidered. First, we studied a vaginal preparation
Reproduction and Embryology on recurrent preg- of progesterone, at a dose of 400 mg twice daily,
nancy loss as being useful for identifying high- and it is possible that the results observed with
risk patients.16 However, we did not identify this this regimen are not generalizable to women
subgroup as one of special interest a priori in receiving other doses and preparations by other

n engl j med 380;19 nejm.org  May 9, 2019 1821

The New England Journal of Medicine
Downloaded from nejm.org by ampri yuwana loyra on May 23, 2019. For personal use only. No other uses without permission.
Copyright © 2019 Massachusetts Medical Society. All rights reserved.
 The n e w e ng l a n d j o u r na l of m e dic i n e

P Value for
Subgroup Placebo Progesterone Risk Ratio (95% CI) Interaction
no. of events/total no.
Maternal age 0.70
<35 yr 1148/1555 1184/1558 1.03 (0.99–1.07)
≥35 yr 311/458 329/467 1.05 (0.97–1.13)
Body-mass index 0.31
<30 1128/1541 1155/1548 1.02 (0.98–1.06)
≥30 331/472 358/477 1.07 (0.99–1.15)
Fetal heart activity 0.37
Present 1304/1648 1347/1672 1.02 (0.99–1.06)
Absent 155/365 166/353 1.10 (0.94–1.30)
Estimated gestation at presentation 0.13
<6 wk 202/365 209/358 1.04 (0.91–1.18)
6 wk to <9 wk 832/1162 917/1211 1.05 (1.01–1.10)
≥9 wk 425/486 387/456 0.98 (0.94–1.03)
Amount of bleeding on PBAC 0.61
≤2 1344/1853 1396/1860 1.03 (1.00–1.07)
≥3 115/160 117/165 1.00 (0.88–1.14)
No. of previous miscarriages 0.007
0 840/1127 824/1111 0.99 (0.95–1.04)
1–2 534/738 591/777 1.05 (1.00–1.12)
≥3 85/148 98/137 1.28 (1.08–1.51)
No. of gestational sacs 0.34
1 1432/1975 1480/1971 1.04 (1.00–1.07)
≥2 27/38 33/54 0.90 (0.69–1.19)
Race 0.87
White 1245/1697 1263/1673 1.03 (1.00–1.07)
Black 46/75 58/82 1.08 (0.87–1.35)
South Asian 69/101 76/113 0.97 (0.82–1.15)
Other 99/140 116/157 1.04 (0.91–1.19)
History of polycystic ovaries 0.35
Yes 161/221 157/220 0.98 (0.88–1.10)
No 1297/1791 1356/1805 1.04 (1.00–1.08)
Previous LLETZ 0.30
Yes 68/102 80/109 1.12 (0.95–1.33)
No 1391/1911 1433/1916 1.03 (0.99–1.06)
All Participants 1459/2013 1513/2025 1.03 (1.00–1.07)
0.75 1.00 1.25

Placebo Better Progesterone Better

Figure 2. Subgroup Analysis.

Shown is the analysis of live births after at least 34 weeks of gestation (the primary outcome) in prespecified subgroups. The widths of
the confidence intervals were not adjusted for multiplicity, so the intervals should not be used to infer definitive treatment effects. The
size of each black square is proportional to the total number of women in the subgroup. The body-mass index is the weight in kilograms
divided by the square of the height in meters. The score on the pictorial blood-loss assessment chart (PBAC) indicates the amount of
vaginal bleeding; scores range from 1 to 4, with higher scores indicating greater vaginal blood loss.11 Race was reported by the participant.
LLETZ denotes large loop excision of the transformation zone of the cervix (also known as the loop electrosurgical excision procedure [LEEP]).

routes. Micronized vaginal progesterone has an blastic–decidual interface have been proposed as
identical molecular structure to natural proges- a mechanism whereby progesterone might pre-
terone, whereas other formulations of progesta- vent miscarriage.18 The vaginal route delivers a
tional agents have a different molecular structure greater proportion of drug to the relevant site
and therefore potentially different mechanisms (i.e., the uterus) with the use of the “first uterine
of action and pharmacologic features. Immuno- pass” effect.19,20 Furthermore, trials that have
modulatory effects of progesterone at the tropho- evaluated vaginal progesterone in the prevention

1822 n engl j med 380;19 nejm.org  May 9, 2019

The New England Journal of Medicine

Downloaded from nejm.org by ampri yuwana loyra on May 23, 2019. For personal use only. No other uses without permission.
Copyright © 2019 Massachusetts Medical Society. All rights reserved.
 Progesterone in Women with Bleeding in Early Pregnancy

of preterm birth have shown its effectiveness authors and do not necessarily reflect those of the National
Health Service, the NIHR, the Medical Research Council, the
when administered by this route.21,22 NIHR Central Commissioning Facility, the NIHR Evaluation,
Second, we started progesterone treatment Trials and Studies Coordinating Centre, the NIHR Health Tech-
only in women who had an intrauterine sac; nology Assessment program, or the Department of Health.
A data sharing statement provided by the authors is available
therefore, our trial cannot provide evidence on with the full text of this article at NEJM.org.
the effects of earlier use of progesterone, before a Supported by the United Kingdom NIHR Health Technology
pregnancy sac is visible on an ultrasound exami- Assessment program (project number HTA 12/167/26).
Dr. Lumsden reports receiving advisory fees from PregLem;
nation. Third, the participants discontinued pro- and Dr. Norman, receiving consulting fees, paid to the Univer-
gesterone at 16 weeks of gestation; however, we sity of Edinburgh, from Dilafor and GlaxoSmithKline. No other
consider it to be unlikely that therapy beyond potential conflict of interest relevant to this article was reported.
Disclosure forms provided by the authors are available with
this time would have affected the outcomes re- the full text of this article at NEJM.org.
lated to miscarriage. Finally, although we found We thank the women who participated in this trial; the inves-
no increase in the risk of congenital abnormali- tigators for supervising recruitment and randomization at the
trial centers (Mr. Samson Agwu, Mrs. Rita Arya, Miss Miriam
ties among babies of women treated with pro- Baumgarten, Dr. Catey Bass, Miss Sumita Bhuiya, Prof. Tom
gesterone, the trial was not powered for such Bourne, Mr. James Clark, Mr. Samual Eckford, Mr. Zeiad El-
rare outcomes. Gizawy, Mrs. Joanne Fletcher, Miss Preeti Gandhi, Dr. Mary
Gbegaje, Dr. Ingrid Granne, Mr. Mamdough Guirguis, Dr. Pratima
In conclusion, treatment with progesterone did Gupta, Dr. Hadi Haerizadeh, Dr. Laura Hipple, Mr. Piotr Lesny,
not result in significant improvement in the in- Miss Hema Nosib, Mr. Jonathan Pepper, Mr. Jag Samra, Ms.
cidence of live births among women with vagi- Jayne Shillito, Dr. Rekha Shrestha, Dr. Jayasree Srinivasan, Dr.
Ayman Swidan, and Prof. Derek Tuffnell); the PRISM research
nal bleeding during the first 12 weeks of preg-
nurses who assisted in the collection of data; Leanne Beeson,
nancy. Mary Nulty, and Louisa Edwards for their support in managing
This article presents independent research commissioned by and coordinating the trial; Prof. Siladitya Bhattacharya for
the National Institute for Health Research (NIHR). A monograph chairing the trial steering committee; Prof. Andrew Shennan for
reporting the data collected in this trial is planned for publica- chairing the data and safety monitoring committee; Dr. Javier
tion in the NIHR Journals Library. Further information is avail- Zamora and Dr. Willem Ankum for participating in the data and
able at www​.­journalslibrary​.­nihr​.­ac​.­uk/​­hta/​­. The views and opin- safety monitoring committee; and all those not otherwise men-
ions expressed by the authors in this publication are those of the tioned above who have contributed to the PRISM trial.

Appendix
The authors’ full names and academic degrees are as follows: Arri Coomarasamy, M.B., Ch.B., M.D., F.R.C.O.G., Adam J. Devall,
B.Med.Sci., Ph.D., Versha Cheed, M.Sc., Hoda Harb, M.B., Ch.B., Ph.D., Lee J. Middleton, B.Sc., Ioannis D. Gallos, D.M.S., M.D.,
Helen Williams, B.Sc., Abey K. Eapen, M.D., Ph.D., Tracy Roberts, Ph.D., R.G.N., Chriscasimir C. Ogwulu, Ph.D., Ilias Goranitis,
Ph.D., Jane P. Daniels, M.Med.Sci., Ph.D., Amna Ahmed, M.B., B.S., M.R.C.O.G., Ruth Bender‑Atik, B.A., Kalsang Bhatia, M.B., B.S.,
F.R.C.O.G., Cecilia Bottomley, DM.B.B.Chir., M.R.C.O.G., M.D., Jane Brewin, B.A., Meenakshi Choudhary, M.B., B.S., M.D.,
M.R.C.O.G., Ph.D., Fiona Crosfill, M.B., B.S., M.R.C.O.G., Dip.Med.Ed., Shilpa Deb, M.B., B.S., M.R.C.O.G., D.G.O., Ph.D., W. Colin
Duncan, M.B., Ch.B., M.D., F.R.C.O.G., Andrew Ewer, M.B., Ch.B., M.D., M.R.C.P., F.R.C.P.C.H., Kim Hinshaw, M.B., B.S.,
F.R.C.O.G., Tom Holland, M.B., B.S., M.R.C.O.G., M.D., Feras Izzat, M.B., Ch.B., M.R.C.O.G., Jemma Johns, M.B., B.S., M.R.C.O.G.,
M.D., Kathiuska Kriedt, M.B., B.S., Mary‑Ann Lumsden, M.B., B.S., M.D., F.R.C.O.G., Padma Manda, M.B., B.S., D.G.O., D.F.F.P.,
F.R.C.O.G., Jane E. Norman, M.B., Ch.B., M.D., F.R.C.O.G., F. Med.Sci., F.R.C.P. Edin., Natalie Nunes, M.B., B.S., M.R.C.O.G., M.D.
(Res.), Caroline E. Overton, M.B., B.S., M.D., M.R.C.O.G., F.H.E.A., F.R.C.O.G., Siobhan Quenby, B.Sc., M.B., B.S., M.R.C.O.G.,
C.C.S.T., M.D., F.R.C.O.G., Sandhya Rao, M.B., B.S., M.D., M.R.C.O.G., Jackie Ross, M.B., B.S., F.R.C.O.G., C.C.S.T., Anupama Sha-
hid, M.B., B.S., D.F.S.R.H., M.R.C.O.G., Martyn Underwood, M.B., Ch.B., M.R.C.O.G., Nirmala Vaithilingam, M.D., Linda Watkins,
M.B., Ch.B., M.R.C.O.G., D.F.F.P., Catherine Wykes, M.B., B.S., M.R.C.O.G., M.A., Andrew Horne, M.B., Ch.B., M.R.C.O.G., Ph.D.,
F.R.C.O.G., and Davor Jurkovic, M.B., B.S., M.D., M.R.C.O.G.
The authors’ affiliations are as follows: Tommy’s National Centre for Miscarriage Research, College of Medical and Dental Sciences,
University of Birmingham, Birmingham (A.C., A.J.D., V.C., H.H., L.J.M., I.D.G., H.W., A.K.E., T.R., C.C.O.), the Faculty of Medicine
and Health Sciences, University of Nottingham (J.P.D.), and Nottingham University Hospitals NHS Trust (S.D.), Nottingham, City
Hospitals Sunderland NHS Foundation Trust, Sunderland (A.A., K.H.), the Miscarriage Association, Wakefield (R.B.A.), East Lancashire
Hospitals NHS Trust, Burnley (K.B.), Tommy’s Charity (J.B.), Guy’s and St. Thomas’ NHS Foundation Trust (T.H.), King’s College
Hospital NHS Foundation Trust (J.J., J.R.), University College London Hospitals NHS Foundation Trust (K.K., D.J.), West Middlesex
Hospital, Chelsea and Westminster NHS Foundation Trust (N.N., C.B.), and Barts and the London NHS Trust (A.S.), London, New-
castle upon Tyne Hospitals NHS Foundation Trust, Newcastle (M.C.), Lancashire Teaching Hospitals NHS Foundation Trust, Preston
(F.C.), the MRC Centre for Reproductive Health, University of Edinburgh, Edinburgh (W.C.D., J.E.N., A.W.H.), Liverpool Women’s
NHS Foundation Trust (L.W.) and St. Helens and Knowsley NHS Trust (S.R.), Liverpool, University Hospitals Coventry and Warwick-
shire NHS Trust, Coventry (F.I.), the Department of Medicine, University of Glasgow, Glasgow (M.-A.L.), South Tees Hospitals NHS
Foundation Trust, Middlesbrough (P.M.), University Hospitals Bristol NHS Foundation Trust, Bristol (C.E.O.), the Biomedical Research
Unit in Reproductive Health, University of Warwick, Warwick (S.Q.), Shrewsbury and Telford NHS Trust, Telford (M.U.), Portsmouth
Hospitals NHS Trust, Portsmouth (N.V.), and Surrey and Sussex Healthcare NHS Trust, Redhill (C.W.) — all in the United Kingdom;
the Melbourne School of Population and Global Health, University of Melbourne, Melbourne, VIC, Australia (I.G.); and the Carver Col-
lege of Medicine, University of Iowa Health Care, Iowa City (A.E.).

n engl j med 380;19 nejm.org  May 9, 2019 1823

The New England Journal of Medicine
Downloaded from nejm.org by ampri yuwana loyra on May 23, 2019. For personal use only. No other uses without permission.
Copyright © 2019 Massachusetts Medical Society. All rights reserved.
 Progesterone in Women with Bleeding in Early Pregnancy

References
1. Everett C. Incidence and outcome of 9. American College of Obstetricians loss:​guideline of the European Society of
bleeding before the 20th week of preg- and Gynecologists. ACOG practice bulle- Human Reproduction and Embryology.
nancy: prospective study from general tin — clinical management guidelines for November 2017 (https://www​.eshre​.eu/​
practice. BMJ 1997;​315:​32-4. obstetricians–gynecologists:​early pregnan- Guidelines​-­and​-­Legal/​Guidelines/​Recurrent​
2. Cantwell R, Clutton-Brock T, Cooper cy loss. May 2015 (https://www​.acog​.org/​​ -­pregnancy​-­loss).
G, et al. Saving mothers’ lives: reviewing -­/​media/​Practice​-­Bulletins/​Committee​-­on​ 17. Wang R, Lagakos SW, Ware JH, Hunter
maternal deaths to make motherhood -­Practice​-­Bulletins​-­​-­​-­​-­Gynecology/​Public/​ DJ, Drazen JM. Statistics in medicine —
safer: 2006-2008 — the eighth report of pb150​.pdf). reporting of subgroup analyses in clinical
the Confidential Enquiries into Maternal 10. Practice Committee of the American trials. N Engl J Med 2007;​357:​2189-94.
Deaths in the United Kingdom. BJOG Society for Reproductive Medicine. Evalu- 18. Robinson DP, Klein SL. Pregnancy and
2011;​118:​Suppl 1:​1-203. ation and treatment of recurrent preg- pregnancy-associated hormones alter im-
3. Murphy FA, Lipp A, Powles DL. Follow- nancy loss: a committee opinion. Fertil mune responses and disease pathogene-
up for improving psychological well being Steril 2012;​98:​1103-11. sis. Horm Behav 2012;​62:​263-71.
for women after a miscarriage. Cochrane 11. Bottomley C, Van Belle V, Pexsters A, 19. Bulletti C, de Ziegler D, Flamigni C,
Database Syst Rev 2012;​3:​CD008679. et al. A model and scoring system to pre- et al. Targeted drug delivery in gynaecol-
4. Farren J, Jalmbrant M, Ameye L, et al. dict outcome of intrauterine pregnancies ogy: the first uterine pass effect. Hum
Post-traumatic stress, anxiety and depres- of uncertain viability. Ultrasound Obstet Reprod 1997;​12:​1073-9.
sion following miscarriage or ectopic preg- Gynecol 2011;​37:​588-95. 20. Cicinelli E, Cignarelli M, Sabatelli S,
nancy: a prospective cohort study. BMJ 12. White IR, Horton NJ, Carpenter J, Po- et al. Plasma concentrations of proges-
Open 2016;​6(11):​e011864. cock SJ. Strategy for intention to treat terone are higher in the uterine artery
5. Malassiné A, Frendo JL, Evain-Brion D. analysis in randomised trials with miss- than in the radial artery after vaginal
A comparison of placental development ing outcome data. BMJ 2011;​342:​d40. administration of micronized progester-
and endocrine functions between the hu- 13. Zou G. A modified Poisson regres- one in an oil-based solution to post-
man and mouse model. Hum Reprod Up- sion approach to prospective studies with menopausal women. Fertil Steril 1998;​69:​
date 2003;​9:​531-9. binary data. Am J Epidemiol 2004;​159:​ 471-3.
6. Wahabi HA, Fayed AA, Esmaeil SA, 702-6. 21. Fonseca EB, Celik E, Parra M, Singh M,
Bahkali KH. Progestogen for treating 14. Peto R, Pike MC, Armitage P, et al. Nicolaides KH. Progesterone and the
threatened miscarriage. Cochrane Data- Design and analysis of randomized clini- risk of preterm birth among women with
base Syst Rev 2018;​8:​CD005943. cal trials requiring prolonged observation a short cervix. N Engl J Med 2007;​357:​
7. Coomarasamy A, Williams H, Trucha- of each patient. I. Introduction and de- 462-9.
nowicz E, et al. A randomized trial of pro- sign. Br J Cancer 1976;​34:​585-612. 22. Romero R, Conde-Agudelo A, Da Fon-
gesterone in women with recurrent mis- 15. Ogasawara M, Aoki K, Okada S, Suzu- seca E, et al. Vaginal progesterone for pre-
carriages. N Engl J Med 2015;​373:​2141-8. mori K. Embryonic karyotype of abor- venting preterm birth and adverse perina-
8. Haas DM, Hathaway TJ, Ramsey PS. tuses in relation to the number of previ- tal outcomes in singleton gestations with
Progestogen for preventing miscarriage ous miscarriages. Fertil Steril 2000;​73:​ a short cervix: a meta-analysis of individ-
in women with recurrent miscarriage of 300-4. ual patient data. Am J Obstet Gynecol
unclear etiology. Cochrane Database Syst 16. ESHRE Early Pregnancy Guideline De- 2018;​218:​161-80.
Rev 2018;​10:​CD003511. velopment Group. Recurrent pregnancy Copyright © 2019 Massachusetts Medical Society.

1824 n engl j med 380;19 nejm.org  May 9, 2019

The New England Journal of Medicine

Downloaded from nejm.org by ampri yuwana loyra on May 23, 2019. For personal use only. No other uses without permission.
Copyright © 2019 Massachusetts Medical Society. All rights reserved.