Articles

Programmatic human papillomavirus testing in cervical

cancer prevention in the Jujuy Demonstration Project in
Argentina: a population-based, before-and-after
retrospective cohort study
Silvina Arrossi, Melisa Paolino, Rosa Laudi, Juan Gago, Alicia Campanera, Oscar Marín, Cristina Falcón, Verónica Serra, Rolando Herrero,
Laura Thouyaret

Summary
Background Human papillomavirus (HPV) testing for cervical cancer prevention was introduced in Argentina through Lancet Glob Health 2019;
the Jujuy Demonstration Project (2011–14). The programme tested women aged 30 years and older attending the 7: e772–83

public health system with clinician-collected HPV tests. HPV self-collection was introduced as a programmatic See Comment page e688
strategy in 2014. We aimed to evaluate the effectiveness of programmatic HPV testing to detect cervical intraepithelial Centro de Estudios de Estado y
Sociedad, Buenos Aires,
neoplasia (CIN) of grade 2 or worse (CIN2+) in comparison with cytology-based screening.
Argentina (S Arrossi PhD,
M Paolino PhD, J Gago MD);
Methods We did a population-based, before-and-after retrospective cohort study using data from the National Cervical Consejo Nacional de
Cancer Prevention Program for the Jujuy province in northwest Argentina. We obtained data for the cytology-based Investigaciones Científicas y
Técnicas, Buenos Aires,
screening period from Jan 1, 2010, until Dec 31, 2011, and for the HPV-based screening period from Jan 1, 2012, until
Argentina (S Arrossi, M Paolino);
Dec 31, 2014. The primary outcome was detection of histologically diagnosed CIN2+ among women aged 30 years Hospital Ramos Mejía,
and older. To assess the outcomes in all individuals included in the study, we used multivariable logistic regression Buenos Aires, Argentina
and propensity score matching. The reach, effectiveness, adoption, implementation, and maintenance (RE-AIM) (R Laudi, MD); Programa
Nacional de Prevención de
framework was used for the before-and-after analysis of programmatic dimensions.
Cáncer Cervicouterino,
Instituto Nacional del Cáncer,
Findings Of the 29 631 women who underwent cytology-based screening in 2010–11, CIN2+ was detected in Buenos Aires, Argentina
236 (0·8%) individuals. Of the 49 565 women HPV tested in 2012–14 (clinician-collected tests, n=44 700; self-collection (J Gago, L Thouyaret MSc);
Ministerio de Salud de la
tests, n=4865), 693 (1·4%; 658 clinician-collected tests; 35 self-collection tests) were found to have CIN2+ after the
Provincia de Jujuy,
first round of screening. Compared with cytology-based screening, the odds ratio of being diagnosed with a San Salvador de Jujuy,
CIN2+ lesion was 2·34 (95% CI 2·01–2·73; p<0·0010) with clinician-collected tests, and 1·08 (0·74–1·52; p=0·68) Argentina (A Campanera MD,
when screened with self-collection tests, after controlling for age and health insurance status. Screening coverage was V Serra MD); Hospital Pablo
Soria, San Salvador de Jujuy,
similar in both periods (52·7% vs 53·2%); improvements of programmatic indicators were observed in the HPV Argentina (O Marín MD,
testing period in relation to laboratory centralisation, lower overscreening (6·6% vs 0·0%), higher adherance to age C Falcón MD); and International
recommendations (79·3% vs 98·8%), and a decrease of inadequate samples (3·6% vs 0·2%). Agency for Research on Cancer,
Lyon, France (R Herrero PhD)

Interpretation HPV testing in middle-income settings increases detection of CIN2+ lesions and allows for Correspondence to:
Dr Silvina Arrossi, Centro de
improvement of programmatic indicators. Evidence suggests that the introduction of HPV testing will accelerate the
Estudios de Estado y Sociedad,
reduction of cervical cancer burden. Buenos Aires 1193, Argentina
[email protected]
Funding Argentinian National Cancer Institute and National Council of Scientific and Technologic Research.

Copyright © 2019 The Authors. Published by Elsevier Ltd. This is an Open Access article under the CC BY-NC-ND
4.0 license.

Introduction testing has a high sensitivity and negative predictive

Cervical cancer is a major problem because of its high value, thus women without HPV infection are at very
incidence and mortality in low-income and middle- low risk of cervical cancer and do not need additional
income settings.1 Developed countries have shown that screening for at least 5 years.5 HPV testing accurately
the disease can be controlled with cytology-based identifies women at higher risk of cervical cancer; these
screening2 done within organised programmes,3 but women can be followed up, diagnosed, and treated with
these programmes have been difficult to implement in a more specific protocol than if they had been tested
health systems of low-income and middle-income with cytology-based screening.6 In addition, HPV
countries with limited resources.4 testing allows for sample self-collection, which is
In the past two decades, human papillomavirus effective in increasing screening uptake.7,8 These
(HPV) testing has been developed as an alternative features have made HPV testing the preferred tool for
screening method for cervical cancer prevention. HPV cervical cancer screening. In combi­nation with HPV

www.thelancet.com/lancetgh Vol 7 June 2019 e772

 Articles

Research in context
Evidence before this study Implications of all the available evidence
The initial search to define the study protocol covered PubMed Results from the Jujuy Demonstration Project study are very
between Jan 1, 2002, and Dec 31, 2010. The search was extended important for similar settings implementing or considering
to Jun 30, 2018, when writing the manuscript. Relevant implementing HPV testing. The study showed that, compared
publications with the key terms “cervical cancer”, “HPV testing”, with cytology, programmatic HPV testing doubled detection
“cervical cancer screening”, “clinical trials”, and “self-collection” of cervical intraepithelial neoplasia of grade 2 or worse (CIN2+)
were reviewed for quality and relevance. Only studies written in lesions, confirming available evidence from randomised
English or Spanish were considered. We found that much of the controlled trials. In addition, introduction of HPV testing can
evidence about the effect of human papillomavirus (HPV) testing facilitate the programme and health service reorganisation
comes from randomised controlled trials, and to our knowledge needed to improve programme indicators. Therefore,
no study has assessed how HPV testing has performed in real- although the positive predictive value was lower and the
world programmatic conditions of middle-income settings colposcopy referral frequency was higher in the HPV testing
compared with cytology-based screening. group compared with the cytology group, adherence to
colposcopy was similar in both groups. Further analysis
Added value of this study
should provide supportive evidence showing that the
This study analysed results from the Jujuy Demonstration
increased detection of CIN2+ lesions in this first round of
Project, which ran from 2012–14, and was one of the
screening represents early detection and not overdiagnosis,
first population-based HPV testing projects done in a
as shown by studies with longer follow-up.
middle-income setting.

vaccination, HPV testing could accelerate elimination evaluate the effect of HPV testing on detection of cervical
of cervical cancer.9 intraepithelial neo­plasia of grade 2 or worse (CIN2+).
Cervical cancer prevention in Argentina has historically Non-randomised methods are increasingly used to
faced the same organisational problems as most evaluate population health interventions,12 and PSM
countries in Latin America.10 The National Program on ensures that the average characteristics of the intervention
Cervical Cancer Prevention (NPCCP) was relaunched in and comparison groups are similar, which is deemed
2008 and in 2010 a decision was made to introduce HPV sufficient to obtain an unbiased effect.13 Such methods are
testing for screening.11 Jujuy, a province with high cervical particularly suitable when randomisation is not feasible
cancer mortality (11∙8 per 100 000 in 2008–10) was chosen and to produce data for the effect of interventions in real-
for implementation of the Jujuy Demon­stration Project world settings. Additionally, we analysed improvement of
(JDP),11 a 4-year population-based study led by the key indicators related to programme organisation.
Argentinian National Cancer Institute, done during The Jujuy setting has been extensively described else­
2011–14 to evaluate large-scale programmatic introduction where.8,11 The province is located in northwest Argentina
of HPV testing. and has around 673  000 inhabitants; 85% of the
Descriptions of the JDP planning phase (2011), and the population live in urban areas and 32% are poor; its
first year of screening (2012) were previously published.11 public health system includes a tertiary referral hospital,
In this Article, we present the final results of the JDP. We 300 primary health-care centres, 18 diagnostic centres,
aimed to evaluate how effective HPV-based screening and five treatment services. Health services are free for
was in increasing detection of precancerous cervical the population not covered by the social security sector
lesions compared with cytology-based screening. (eg, informal workers and their families).

Methods Cytology-based screening procedures

Study design and participants A situational analysis done in 200714 showed that in Jujuy,
The JDP was population-based and implemented by the cervical screening coverage was low, information systems
NPCCP in the province of Jujuy, Argentina, to evaluate the were unreliable, information on follow-up and treatment
introduction of HPV testing as programmatic, primary was missing, and providers had low adherence to
screening.11 The JDP involved 1 year of planning (from Jan programmatic norms and recommendations. Cytology
1 to Dec 31, 2011) and 3 years of screening (between Jan 1, results were read in six laboratories, which processed in
2012, and Dec 31, 2014). On Jan 1, 2012, all Jujuy public total around 22  000 annual samples without quality
health institutions changed the primary screening method controls.
for cervical cancer prevention from cytology-based Before 2012, cytology-based screening was recom­
screening to HPV testing. We used a before-and-after, mended in Argentina for women aged 25 years and older,
retrospective cohort study, and a pre-post design, every 3 years after two consecutive negative Papanicolaou
combined with propensity score matching (PSM), to (Pap) smear tests, but annual screening was common

e773 www.thelancet.com/lancetgh Vol 7 June 2019

 Articles

practice. Colposcopy, and biopsy if needed, was recom­ (protocol number 1186). De-identification of the data­
mended for women with atypical squamous cells in bases protected the identity of participants. Verbal
whom high-grade or worse lesions could not be excluded informed consent was obtained according to the national
(atypical squamous cells for which high-grade squamous regulations for standard medical practices (Patient’s
intraepithelial lesions cannot be ruled out or worse Rights Act 26.529). Specific consent was not required for
[ASC-H+], high-grade squamous intraepithelial lesion, statistical analysis of aggregated de-identified data.
or cancer). Women with atypical squamous cells of We extracted data from SITAM for the purposes of this
undetermined significance (ASCUS) or low-grade squa­ analysis.16 Colposcopies, biopsies, and treatments not
mous intra­epithelial lesions (LSIL) were recommended registered in SITAM were considered lost to follow-up,
for re-screening in 6 or 12 months. including those done in private services without con­
firmation by the provincial programme. Information on
HPV-based screening procedures health insurance was obtained from the National
In 2012, the JDP introduced HPV testing (Hybrid Database on Health Insurance.
Capture 2; Germantown, MD, USA) for primary
screening11 of women aged 30 years and older, irrespective Outcomes
of previous screening history. JDP protocols have been We compared key programmatic indicators using
described elsewhere.11 Briefly, women who were the reach, effectiveness, adoption, implementation,
HPV-positive were triaged with cytology-based screening. and maintenance (RE-AIM) framework,17 specifically
Individuals whose samples were classified as ASCUS or developed to expand assessment of interventions beyond
worse (ASCUS+) were referred to colposcopy and biopsy efficacy to multiple criteria. This analysis might better
if needed. Women with histologically confirmed CIN2+
were referred for treatment. Women who were HPV 25 000 Proportion of target population
negative were recommended re-screening in 5 years.
HPV testing and cytology triage were collected simul­ 20 000
Number of women screened

taneously, but cytology was read only if the individual

was HPV-positive. Women who were HPV-positive but 15 000
had normal cytology were recommended re-screening in
18 months. Women younger than 30 years continued to 10 000

undergo cytology-based screening.

5000 70% 69% 63% 63% 63%
The Jujuy primary health-care system employs more
than 700 full-time community health workers, who visit
0
around 110  000 households (70% of total provincial 2010 2011 2012 2013 2014
households) twice each year for health-related tasks Year of screening
including promotion of HPV testing.11 On the basis of the
Figure 1: Screened women by year, Jujuy 2010–14
EMA study,8 which showed a four-times increase in
screening uptake due to self-collection, the strategy of self-
collection offered by community health workers during Cytology-based period HPV-based period
home visits was scaled up in 2014.15 The strategy targeted 2010–11 (n=29 631) 2012–14* (n=49 565)
women aged 30 years and older from households visited Age (years) 42·9 (10·3) 43·1 (10·4)
by community health workers, who had not been screened 30–34 7851 (26·5%) 12 582 (25·4%)
in the previous 5 years, and with public health coverage. 35–44 10 478 (35·4%) 17 923 (36·2%)
Women who were HPV-positive who used self-collection 45–54 6624 (22·4%) 10 775 (21·7%)
tests had to attend health centres for cytology triage. 55–64 3711 (12·5%) 6486 (13·1%)
At present, self-collection is in use in four Argentinian 65 and older 967 (3·3%) 1799 (3·6%)
provinces. Health insurance
Private or social 8783 (29·6%) 17 902 (36·1%)
Data sources Public 20 848 (70·4%) 31 663 (63·9%)
Since 2010 in Jujuy, any instance of screening, diagnosis, Target population†
or treatment using public health services has been No 9052 (30·5%) 18 456 (37·2%)
registered in the national screening information system Yes 20 579 (69·5%) 31 109 (62·8%)
(SITAM).16 The HPV laboratory used SITAM to manage
samples at entry; the samples of individuals that did not Data are n (%) or mean (SD). HPV=human papillomavirus. *Clinician-collected and
self-collected tests combined. †Yes=women aged 30–64 years with public health
comply with the recommended age range or screening insurance; No=women aged 30–64 years with private or social health insurance
frequency were not processed. Women were informed and those aged 65 years or older with any health insurance.
about why their test was not analysed and were reminded
Table 1: Sociodemographic characteristics of screened women,
of the date of their next HPV test. The protocol was
Jujuy 2010–14
approved by the CEMIC Institutional Review Board

www.thelancet.com/lancetgh Vol 7 June 2019 e774

 Articles

identify the translatability and public health effect of

Clinician-collected Self-collected Cytology
(2012–14) (2014) (2010–11) health promotion interventions, balancing the emphasis
on internal and external validity.17,18 The RE-AIM frame­
Total screened aged ≥30 years 44 700 4865 29 631
work is particularly appropriate for assessing the public
Test-positive 6111 (13·7%) 633 (13·0%) 1178 (4·0%)
health effect of the intervention as a function of the five
Detection by screening
factors that comprise the RE-AIM acronym, all of which
CIN2+ 552 (1·23%) 35 (0·72%) 236 (0·80%)
are considered necessary for success.17
CIN2 91 (0·20%) 7 (0·14%) 55 (0·19%)
In assessing reach (defined as the proportion of
CIN3* 405 (0·91%) 21 (0·43%) 158 (0·53%)
individuals who receive or are affected by a policy or
CA 56 (0·13%) 7 (0·14%) 23 (0·08%)
programme17), our goal was to evaluate how HPV testing
CIN2+ detection rate 12·3 7·2 8·0 influenced the capacity of the provincial programme to
(per 1000 screened women)
achieve high screening coverage. We defined coverage as
CIN3+ detection rate 10·3 5·7 6·1
(per 1000 screened women) the proportion of women aged 30–64 years with public
Detection by follow-up of HPV-positive women with negative cytology health insurance screened at least once in each period of
CIN2+ 106 (0·24%) 0 NA the estimated number of target women (2-year cytology
CIN2 21 (0·05%) 0 NA
period n=39 000; 3-year HPV period n=58 500, according
CIN3* 81 (0·18%) 0 NA
to the National Census 2010). For the HPV testing period,
CA 4 (0·01%) 0 NA
we measured coverage including women that had
clinician-collected tests and coverage including both
CIN2+ detection rate 2·4 0·0 NA
(per 1000 screened women) women who had a clinician-collected test and those who
CIN3+ detection rate 1·9 0·0 NA submitted a self-collected test.
(per 1000 screened women) Our primary effectiveness outcome was histologically
Overall CIN2+ detection rate 14·7 7·2 8·0 confirmed CIN2+ detection among women aged 30 years
(per 1000 screened women) or older screened between Jan 1, 2010, and Dec 31, 2014.
Overall CIN3+ detection rate 12·2 5·7 6·1 When referring to effectiveness, we followed the
(per 1000 screened women)
definition of Rabin and Brownson,19 which refers to the
Referral to colposcopy 1663 (3·7%) 150 (3·1%) 403 (1·4%) effect of an intervention that has shown efficacy when it
Overall positive predictive value 10·8% 5·5% 20·0% is delivered under real-world conditions. Detection was
Data are n or n (%), unless otherwise specified. HPV=human papillomavirus. CIN=cervical intraepithelial neoplasia. calculated as follows: (1) proportion of women who
CA=carcinoma. NA=not applicable. *CIN3 includes adenocarcinoma in situ. underwent cytology-based screening who had CIN2+
detected of the total number of women who under­went
Table 2: Screening performance indicators by type of tests
cytology-based screening; and (2) proportion of women

29 631 total cytology samples

28 453 negative cytology samples 775 ASCUS or LSIL samples 403 ASC-H+ samples

100 women with ASC-H+ at

follow-up

135 women without 368 women with

colposcopy colposcopy or biopsy

132 women with normal biopsy 236 women with CIN2+

or CIN1 lesions lesions

46 women without 190 women with

treatment treatment

Figure 2: Follow-up of women with ASC-H or worse lesions detected by cytology, Jujuy 2010–11
ASC-H+=atypical squamous cells for which high-grade squamous intraepithelial lesions cannot be ruled out or worse. ASCUS=atypical squamous cells of
undetermined significance. CIN=cervical intraepithelial neoplasia. LSIL=low-grade squamous intraepithelial lesions.

e775 www.thelancet.com/lancetgh Vol 7 June 2019

 Articles

44 700 women had a clinician-collected test

6111 women were HPV-positive

92 women had unsatisfactory cytology

109 women were without cytology

4247 women had normal 1663 women had triage

cytology 1657 women had
abnormal cytology
6 women underwent
coloscopy triage

2509 women had a 423 women without 1240 women with coloscopy
secondary HPV test coloscopy because of loss
to follow-up

56 women were 289 women had 895 women had

without normal results abnormal
available results results

48 women 847 women with

without biopsy biopsy in PHS

295 women 552 women with

with normal CIN2+
biopsy or lesions
CIN1 lesions

89 women 473 women with

without treatment
treatment
data

Figure 3: Follow-up of women who were HPV-positive who had clinician-collected tests, Jujuy 2012–14
CIN=cervical intraepithelial neoplasia. HPV=human papillomavirus. PHS=public health system.

who underwent HPV screening (including both clinician- 1000 screened women) through a descriptive before-and-
collected and self-collected tests) who had CIN2+ detected after analysis using two periods: (1) the 2-year cytology-
of the total number of women who underwent HPV based screening period preceding the introduction of
screening. CIN2+ lesions detected by HPV testing also HPV testing, 2010–11; and (2) the HPV period, 2012–14.
included those detected at 18-month follow-up in women Adoption refers to the intention to use an innovation or
who were HPV-positive but had normal cytology at the evidence-based practice.20 For each period we measured
original screening. The proportion of women who had the following: the proportion of primary health-care
CIN3+ lesions detected was also calculated. The positive centres that provided the screening method of the total
predictive value for HPV testing was calculated at both number of primary health-care centres; the proportion of
baseline and follow-up at 18 months. Histological women screened within the recommended age range
confirmation was considered the gold standard. We (aged 25 years or older for cytology-based screening, and
calculated odds ratios (ORs) and 95% CIs to assess the 30 years or older for HPV screening) of the total number
CIN2+ detection effectiveness of HPV compared of women who were screened; and the proportion of
with cytology, and calculated CIN2+ detection rate (per women who were over-screened of the total number of

www.thelancet.com/lancetgh Vol 7 June 2019 e776

 Articles

4865 women used self-collection tests

633 women were HPV-positive

36 women were 16 women had a 439 women were

negative for colposcopy cytology
HPV on their triaged
second test

48 women were 257 women 134 women

without had had
available normal abnormal
results samples samples

45 women 89 women
without with
colposcopy colposcopy

10 women 6 women had 8 women were 33 women 48 women

had normal abnormal without had normal had
samples samples available results abnormal
results results

6 women had 3 women 45 women

a biopsy without a with biopsy
biopsy

2 women had 4 women had 14 women had 31 women had

normal CIN2+ normal CIN2+
biopsy or lesions biopsy or lesions
CIN1 lesions CIN1
lesions

4 women had 7 women 24 women

treatment without with
treatment treatment

Figure 4: Follow-up of women who were HPV-positive who did self-collected tests, Jujuy 2014
CIN=cervical intraepithelial neoplasia.

women who were screened. Over-screening was defined samples of the total number of cytology samples
as screening done more than once per year for cytology- (2010–11); and the proportion of HPV samples discarded
based screening; and more than once in the 3-year period at the laboratory of the total number of HPV samples
for HPV screening. (2012–14). We additionally measured the proportion of
Implementation refers to the extent to which a inadequate cytology-based triage samples of the total
programme is delivered as intended.20 In each period, we number of cytology triage samples. The third was
measured four outcomes. The first was laboratory completion of follow-up, which was split into four parts:
organisation—ie, the number of laboratories processing (1) the proportion of women with a positive Pap test
the primary screening test. The second was changes in result who had a colposcopy of the total number of
sample quality—ie, the proportion of inadequate cytology women with a positive test result (ie, for the cytology

e777 www.thelancet.com/lancetgh Vol 7 June 2019

 Articles

period, the proportion of women with ASC-H+ cytology

Pre-propensity score matching Post-propensity score matching
that complied with colposcopy of the total number of
women who had ASC-H+; for the HPV period, the OR (95% CI) p value OR (95% CI) p value

proportion of HPV-positive women with ASCUS+ who Screening method

had a colposcopy of the total number of HPV-positive Cytology 1 (ref) ·· 1 (ref) ··
women with ASCUS+); (2) the proportion of HPV- Self-collected HPV test 1·08 (0·74–1·52) 0·68 1·15 (0·73–1·72) 0·52
positive women with normal cytology who complied with Clinician-collected HPV test 2·34 (2·01–2·73) <0·0001 2·31 (1·95–2·74) <0·0001
follow-up at 18 months (HPV period) and the proportion Age, years
of women with ASCUS or LSIL who complied with 30–44 1 (ref) ·· 1 (ref) ··
follow-up at 12 months (cytology period); and (3) for the 35–44 0·89 (0·75–1·05) 0·15 0·89 (0·76–1·05) 0·18
HPV period we also measured the proportion of women 45–54 0·72 (0·59–0·88) <0·0001 0·74 (0·58–0·93) 0·011
with self-collection HPV-positive tests who had triage 55–64 0·96 (0·77–1·19) 0·69 1·06 (0·72–1·51) 0·75
(cytology or colposcopy). The fourth was the proportion ≥65 1·57 (1·13–2·14) 0·0057 2·34 (1·35–3·81) 0·0012
of women who received treatment of the total number of Health insurance
women with CIN2+ lesions. Private 1 (ref) ·· 1 (ref) ··
Maintenance is the extent to which a programme or Public 1·32 (1·14–1·54) <0·0001 1·36 (1·14–1·63) <0·0001
policy becomes institutionalised or part of the routine Constant 0·0058 <0·0001 0·0059 <0·0001
organisation practices and policies.18 We presented the (0·0047–0·0072) (0·0047–0·0074)
number of new women tested for HPV (both with
CIN=cervical intraepithelial neoplasia. OR=odds ratio. HPV=human papillomavirus.
clinician-collected and self-collection tests) in 2015–17,
after the JDP was finalised. Table 3: Logistic regression of CIN+ detection frequency with and without propensity score matching

Statistical analysis [58·8%] of 29 631) and most HPV tested women were
Mean age as a continuous variable was compared screened in 2012 (22 515 [45·4%] of 49 565); figure 1).
between the cytology-based period and HPV-based period Among the 49 565 women who were tested for HPV,
using the Wilcoxon rank-sum test. Multivariable logistic 44 700 (90·2%) had clinician-collected tests and
regression was used to measure the magnitude of the 4865 (9·8%) used self-collection tests. In 2014, when self-
effect of HPV testing on CIN2+ detection compared with collection screening was introduced, the method
cytology-based screening, after adjusting for age and represented 38·1% (4865 of 12 779) of all screening tests
health insurance status. in the study sample.
To account for potential selection into the intervention Compared with women who underwent cytology-based
group and minimise bias, we developed a second model screening, a lower proportion of women who were HPV
using PSM. Included variables were age (in years) and tested (both clinician-collected and self-collection tests)
health insurance status. Functionality of PSM requires had public health insurance and were from the target
datasets with no missing values. To handle missing data, population (table 1). Although the mean age of the
we used average imputation for age (six missing cases) two groups was numerically similar, according to the
and random imputation for health insurance (13 missing Wilcoxon rank-sum test, individuals who underwent
cases). The matching algorithm chosen was the nearest cytology-based screening were significantly younger than
neighbour algorithm, using a caliper value of 0∙1 SD. those who underwent HPV screening (42·9 years vs
We used R statistical software (version 3.5.0) for all 43·1 years; p=0∙013). Among the 4865 women who used
analysis, and Matchit R Package (version 3.0.2) for the self-collection tests, 3520 (72·4%) had public health
PSM. insurance, and 3265 (67·1%) were from the target
population.
Role of the funding source Figures 2–4 show the follow-up of screened women for
The funder had no role in study design, data collection, method of screening. Screening performance indicators
analysis, or writing of the report. The corresponding by type of test are shown in table 2. CIN2+ was detected
author had full access to all data in the study and had in 236 (0·8%) of 29 631 women who had undergone
final responsibility for the decision to submit for cytology-based screening, (figure 2) and the positive
publication. predictive value was 20·0%. CIN2+ was detected in
552 (1·23%) of 44  700 women who had undergone
Results clinician-collected tests in the first round of screening.
Between Jan 1, 2010, and Dec 31, 2014, 79 196 women 2509 (59·1%) of the 4247 women who were HPV-positive
aged 30 years and older were screened; 29 631 women with normal cytology were re-screened (figure 3); CIN2+
underwent cytology-based screening in 2010–11, and was detected in 106 (0·24%) individuals in this group.
49 565 women were tested for HPV in 2012–14. All Overall, 658 CIN2+ lesions were detected through
women were included in the analysis. Most cytology- clinician-collected tests, and the positive predictive value
based screened women were screened in 2011 (17 185 was 10·8%. 35 CIN2+ lesions were identified among

www.thelancet.com/lancetgh Vol 7 June 2019 e778

 Articles

Cytology-based period HPV test overall Clinician-collected test Self-collected test

Reach
Women aged 30–64 years with public health insurance who were screened at 20 579/39 000 (52·7%) 31 109/58 500 (53·2%)* 27 844/58 500 (47·5%) 3225/58 500 (5·7%)*
least once in each period (%)
Effectiveness
CIN2+ detection rate in women aged ≥30 years (per 1000 screened women) 8·0 14·7 12·4 7·2
Odds ratio (95% CI) vs cytology 1 (ref) ·· 2·34 (2·01–2·73) 1·08 (0·74–1·52)
Adoption
Health-care centres that provided screening method in each study period (%) 300/300 (100%) 300/300 (100%) 100% NA
Women of the recommended age screened in each study period (%) 38 043/47 927 (79·3%) 49 565/50 147 (98·8%) 44 700 4865
Women who were over-screened in each period 2500/38 046 (6·6%) 0/49 565 0 0
Implementation
Laboratories processing screening tests for women aged ≥30 years (n) 6† 1 1 1
Inadequate primary test samples (%) 2045/56 709 (3·6%) 121/49 686 (0·2%) 56 65
Inadequate cytology-based triage (%) NA 156/7052 (2·2%) NA NA
Follow-up: triage-positive women who had a colposcopy (%) 380/526 (72·2%) NA 1240/1663 (74·6%) 105/150 (70·0%)
Follow-up: women with CIN2+ who had registered treatment (%) 200/249 (80·3%) NA 463/552 (83·9%) 28/35 (80·0%)
Follow-up: women with ASCUS or LSIL who were followed up at 12 months 345/775 (44·0%) 1169/2509 (46·6%) 1169 NA
(cytology-based period) and women who were HPV-positive with normal
cytology who were followed up at 18 months (HPV-based period, %)
Follow-up: women with HPV-positive self-collected tests who were triaged NA 491/633 (77·6%) NA 491
(cytology or colposcopy, %)
Maintenance (2 years after study end)
HPV-tested women, 2015–17 (n) NA 30 975 19 795 11 180

CIN2+=cervical intraepithelial neoplasia of grade 2 or worse. NA=not applicable. ASCUS=atypical squamous cells of undetermined significance. LSIL=low-grade squamous intraepithelial lesions. HPV=human
papillomavirus. *Self-collected testing was introduced in 2014, so its contribution to reach is based on 1 year only. †In 2010; the number of cytology laboratories was reduced to three in 2012.

Table 4: Reach, effectiveness, adoption, implementation, and maintenance (RE-AIM) measurement

women using self-collection tests (figure 4), and the between method of screening and the variables of age and
positive predictive value was 5·5%. Of the 257 women health insurance status. In these models, the interaction
who used self-collection tests who were HPV-positive terms were not statistically significant, whereas all other
with normal cytology, 41 (16%) were re-screened, but no variables remained statistically significant—as in the
CIN2+ lesions were identified in this group. Overall, model without interaction.
693 (1·4%) of 49 565 women who underwent HPV testing Key programmatic indicators assessed by before-and-
(including both clinician-collected and self-collection after analysis using the RE-AIM framework are shown in
tests) had CIN2+ lesions detected. The proportion of table 4. Regarding reach, estimated coverage was
individuals who were referred for colposcopy was 52·7% for the cytology period (20 579 of 39 000), and
403 (1·4%) of 29  631 for cytology-based screening 47·5% for the HPV period (27 844 of 58 500) when only
(figure 2), 1663 (3·7%) of 44 700 for clinician-collected women who had clinician-collected tests were considered,
tests, and 150 (3·1%) of 4865 for self-collection tests. and 53·2% if both clinician-collected and self-collection
Results from the multivariate logistic regression and tested women are included (31 109 of 58 500).
PSM are shown in table 3. The odds of being diagnosed 100% of health-care centres adopted the screening
with a CIN2+ lesion were higher when using clinician- method in each period. The percentage of screened
collected testing than with cytology-based screening, women within the recommended age range was higher
after controlling for age and health insurance (OR 2·34; in the HPV period: 98·8% (aged 30 years and older)
95% CI 2·01–2·73; p<0·0001). After PSM, the odds of a versus 79·3% in the cytology period (aged 25 years and
CIN2+ result using HPV testing were similar to before older). Over-screening was 0·0% in the HPV period
PSM. The odds of being diagnosed with a CIN2+ lesion and 6·6% in the cytology period.
were similar for women who used self-collection tests The implementation of the programme involved
and those using cytology-based screening, both before six laboratories processing cytology-based screening
and after PSM (table 3). Analysis including only tests in 2009; this number was reduced to three by 2011.
individuals with CIN3+ lesions did not change results In 2012, a central HPV-cytohistology laboratory was
(clinician-collected HPV testing vs cytology, OR 2∙54, created. The percentage of inadequate samples of the
95% CI 2·14–3·03; p<0·0001; self-collection HPV testing primary screening test was reduced from 3·6% (cytology
vs cytology, OR 1·20, 95% CI 0·8–1·75; p=0∙34). We ran period) to 0·2%. The percentage of women with ASC-H+
additional models evaluating the possible inter­ action with colposcopy was similar among women screened

e779 www.thelancet.com/lancetgh Vol 7 June 2019

 Articles

in the HPV period (clinician-collected 74·6% vs factor in explaining the lack of effectiveness of screening
self-collection 70·0%) to those screened in the cytology- in low-income and middle-income countries.4 Problems
based period (72·2%). In total, 491 (77·6%) women with are related to its low to moderate sensitivity, but also to
HPV-positive self-collection tests were triaged. The organisational problems faced by health systems due to
percentage of treated CIN2+ lesions was similar in all the complexity of cytology-based screening. An analysis of
three groups, around 80%. The percentage of re- the programmatic effect of introducing HPV testing
screening of women who were HPV-positive with normal showed that HPV testing is an opportunity to change
cytology at 18 months (46·6%) was similar to the inefficient components of screening programmes.26
percentage of re-screening of women with ASCUS or Therefore, HPV testing has been recommended as a
LSIL at 12 months in the cytology-based period (44·0%). strategy to simplify and improve screening organisation.27
Regarding effectiveness, CIN2+ detection rates were Some concerns exist about the increased CIN2+
12·3 per 1000 screened women for clinician-collected detection frequency representing overdiagnosis of
tests, 7·2 per 1000 screened women for self-collected tests, lesions that would not have progressed to invasive
and 8·0 per 1000 screened women for cytology tests. cancer.28 Although our study did not evaluate this issue,
When compared with cytology, clinician-collected testing evidence has shown that the increased sensitivity of HPV
detected more cases of CIN2+, whereas no significant testing for CIN2+ reflects earlier detection rather than
differences were observed between self-collected tests and overdiagnosis.28–29 Sasieni30 has pointed out that because
cytology tests. HPV testing prevents substantially more cancers than
For maintenance, during 2015–17, 30 975 women who cytology-based screening, even if some of the CIN2+
were new to the screening programme were HPV-tested: lesions will not progress, we should accept a small
19 795 (63·9%) with clinician-collected tests and increase in the numbers of women treated for CIN to
11 180 (36·1%) with self-collection tests. The number of achieve that benefit. In the JDP, only women with
women who were new to the screening programme who histologically confirmed CIN2+ lesions were treated.
were HPV-tested was similar in each year (data not Over-referral to colposcopy has been pointed out as a
shown). major problem in HPV testing, which is related to the
test’s low specificity.31 Over-referral also depends on the
Discussion screening protocol, and is higher when all women who
To our knowledge, these are the first systematic results of are HPV-positive are referred for colposcopy.32 In the
HPV testing introduced as a population-based public JDP, only women who were HPV-positive with abnormal
health policy for cervical cancer screening in a middle- cytology were referred for colposcopy. It has also been
income setting. Our findings advance the existing public shown that, with appropriate protocols, increased referral
health literature by showing that effective screening with is limited to the first round of screening with HPV
HPV testing in real-world programmes of middle- testing28 and that in successive rounds referral will be
income settings is feasible. This is particularly important lower than in cytology-based screening. In Argentina,
given the global call for cervical cancer elimination the reco­ mmendation of the main scientific societies33
launched by WHO.21 for cytology-based screening is referral to colposcopy
Clinician-collected HPV testing significantly increased after ASCUS+ diagnosis, a widespread practice among
detection of CIN2+ lesions when compared with cytology- gynaecologists from all over the country despite
based screening, supporting the results from the first year programmatic recommendations.34 Therefore, the HPV
of the JDP.11 Randomised controlled trials have also testing protocol probably resulted in a more efficient use
showed an increase in CIN2+ detection with HPV testing of colposcopy given that it was provided to women who
compared with cytology-based tests in trials done in high- were high risk (ie, individuals who were HPV-positive
income settings.22–24 In our study, detection of CIN2+ was and had abnormal cytology). However, close monitoring
based on pathological diagnoses done within the Jujuy of the implications for colposcopy services should be
public health system. Therefore, our results showed what done in each setting before introducing HPV testing.
can realistically be achieved by HPV testing in middle- The low specificity of HPV testing can also have a
income settings. The JDP implied improvements in negative psychosocial impact in women.31 To reduce this
programmatic organisation, which might be a possible effect, the JDP communication strategy emphasised the
explanation for the increased CIN2+ detection frequency. fact that HPV infection is a common and prevalent
A refresher course was provided to colposcopists in condition, and that HPV positivity did not mean cancer.35
provincial public health centres, probably increasing their Despite the increase in colposcopy referral, completion
diagnostic accuracy. Also, the fact that women referred for of colposcopy was higher in the clinician-collected testing
colposcopy were all HPV-positive might have increased group (75%) than in the cytology group (72%). Studies
the colposcopist’s alertness. Laboratory centralisation and done in Latin America and the Caribbean have reported
cytologists knowing that slides were from women who a wide range of adherence to colposcopy after abnormal
were HPV-positive might have improved cytological cytology (21–99%).36,37 In Jujuy, a patient navigation
diagnosis.25 Cytology-based screening has been a main programme provides support to women who are

www.thelancet.com/lancetgh Vol 7 June 2019 e780

 Articles

HPV-positive and have abnormal cytology to facilitate The JDP was done over a 3-year period. Given that in
their access to follow-up or treatment.37 Also, the province Argentina the recommended screening frequency for
had capacity to respond to the increase in colposcopies women who are HPV-negative is every 5 years, coverage
resulting from the HPV testing strategy. However, targets might be attainable if the 5-year interval is used
adherence to colposcopy was lower in women who used instead. An estimation of coverage if women HPV-tested
self-collection tests. In addition, only 69% of those in 2015–16 were included supports this. We estimated
women had cytology triage compared with 98% of coverage including HPV tested women during 2015–16
women who underwent clinician-collected tests. This and results showed that 75% of the target population was
result was mainly because clinician-collected tests and screened in the 5-year period.
cytology-based tests were taken simultaneously, whereas HPV testing has introduced a subgroup of women who
women who used self-collection tests needed to undergo are HPV-positive but have normal cytology. In our study,
a subsequent visit to a health centre for triage. This 59% of these women were re-screened and, among them,
additional visit is a major drawback of self-collection less than half had persistent HPV infection at repeated
tests. Low compliance to follow-up among women with testing. Follow-up of these women contributed an
positive self-collection tests has been reported for other additional 15% of CIN2+ lesions detected overall, con­
settings.38 Several studies are evaluating triage alter­ firming the importance of this step of the algorithm. Low
natives (eg, methylation, genotyping, among others) to compliance with repeated testing was common in several
reduce the number of steps in the diagnostic process.25 studies.40 Adherence to follow-up depends on several
Meanwhile, strategies that facilitate women’s access to factors, including type of recommended follow-up. A
triage need to be devised. review of studies showed that around 90% of women
Low coverage is a major problem in middle-income complied with follow-up if they were immediately
settings. Our results showed no effect in coverage after referred for colposcopy based on their screening tests
introduction of HPV testing, probably because health alone.40 Follow-up was considerably lower when women
authorities had already given high priority to increasing who were HPV-positive were recommended to first
coverage in the period preceding HPV testing. However, undergo repeated testing 6–18 months after initial
clinician-collected screening has coverage limitations screening than women who were immediately referred
due to socioeconomic, cultural, and institutional for colposcopy.40 However, colposcopy for women who are
barriers faced by women, and the high number of HPV-positive and have normal cytology is not recom­
women who were screened in the first year of the JDP mended due to its complexity and cost, and the low
could not be replicated in following years. Self-collection sensitivity and specificity of the method.27,41 As most HPV
tests offered by community health workers during infections disappear in 12–24 months,42 re-screening of
home visits were introduced in 2014 to counteract this these women at 18 months seems a reasonable strategy to
coverage decrease.15 In 2014, self-collection represented reduce costs and avoid overtreatment. However, time
38% of total HPV testing and, if we consider the whole elapsed between screening and retesting can constitute a
JDP, 10% of screening in the target population was barrier to rescreening adherence.
achieved through self-collection. Among women who Our results showed that introduction of HPV testing
used self-collection tests, the CIN2+ detection frequency allowed for reorganisation of the laboratory network.
was 7·2 per 1000 screened women, lower than detection A key issue was the installation of the HPV-testing
by clinician-collected tests. This figure is lower than that laboratory as part of central cytology-pathology-HPV-
reported in studies in other countries39 and in the EMA testing. Cytology laboratories in middle-income settings
study in Argentina.8 This decreased detection seen with face problems linked to quality control, decentralisation
self-collection tests is probably due to loss to triage but in small laboratories, and lack of technical staff, among
also to the significantly lower sensitivity of Hybrid others.43 HPV testing not only changed the function of
Capture 2 on self-collection samples than with clinician- cytology from screening to triage, but also facilitated
collected samples.39 When compared with cytology- laboratory centralisation by prompting the political
based tests, self-collection tests did not show significant decisions needed for reorganisation of human resources
differences in CIN2+ detection, but this might be due to and the referral network. These changes would have
the low amount of self-collection tests included in the probably been more difficult to implement if HPV testing
analysis. A study of programmatic self-collection in had not been introduced.
Jujuy showed that self-collection testing allowed for the Low adherence to cytology screening guidelines has
increase in screening uptake among socially vulnerable been widely reported.9 In an HPV screening scenario,
women who were under-screened, a group with this low adherence to guidelines might result in
the highest risk of cervical cancer.15 Thus, the possibility ineffective use of resources, inappropriate screening of
of self-collection testing constituted a substantial young women, overtreating lesions that normally clear
advantage of HPV testing for the increase of coverage, within a few months,6 and potential harm such as adverse
despite CIN2+ detection being hampered by the loss pregnancy outcomes.44 Over-screening was eliminated
to triage. with HPV testing, and screening done on individuals

e781 www.thelancet.com/lancetgh Vol 7 June 2019

 Articles

outside of the recommended age range was greatly the Ministry of Health, or Jujuy’s Ministry of Health. We thank the team
reduced. HPV protocols were established through a of the National Program of Cervical Cancer Prevention for its
participation during the project. We also would like to acknowledge
participatory process with the main scientific societies Mariana Curotto and Julieta Zalacain Colombo for their assistance
of Argentina.11 Also, samples not complying with the during the analysis, and Maria Antonia Rodriguez for her comments on
recommended age range or frequency for screening were the first version of the manuscript.
not processed, which has discouraged screening outside References
of national recommendations. 1 Bray F, Ferlay J, Soerjomataram I, Siegel RL, Torre LA, Jemal A.
Global cancer statistics 2018: GLOBOCAN estimates of incidence
Analysis of maintenance showed that HPV testing and mortality worldwide for 36 cancers in 185 countries.
continued after the JDP and became the standard primary CA Cancer J Clin 2018; 68: 394–424.
screening method. Integrating activities into existing 2 Gustafsson, L, Pontén J, Bergström R, Adami HO. International
incidence rates of invasive cervical cancer before cytological
health systems has been identified in the literature as an screening. Int J Cancer 1997; 71: 159–165.
important factor for successful scale-up.45 On the basis of 3 International Agency for Research on Cancer–WHO.
results from the JDP, a decision was made by the national IARC handbook on cervical cancer prevention. Volume 10. Cervix
Ministry of Health to expand HPV testing. At present, Cancer Screening. Lyon: IARC Press; 2005.
4 Sankaranarayanan R, Budukh A, Rajkumar R. Effective screening
eight of 24 provinces have primary HPV testing, and programmes for cervical cancer in low- and middle-income
six other provinces will introduce it by 2020. developing countries. Bull World Health Organ. 2001; 79: 954–62.
A key limitation of the study concerned cytology-based 5 Sankaranarayanan R, Nene BM, Shastri SS, et al. HPV screening
for cervical cancer in rural India. N Engl J Med 2009; 360: 1385–94.
screening data only being available for the 2 years before
6 Herrero R, Ferreccio C, Salmerón J, et al. New approaches to
introduction of HPV testing. A longer medical history cervical cancer screening in Latin America and the Caribbean.
from cytology-based screening records might have Vaccine 2008; 26 (suppl 11): L49–58.
changed results. Overscreening might be underestimated, 7 Lazcano-Ponce E, Lorincz AT, Cruz-Valdez A, et al. Self-collection of
vaginal specimens for human papillomavirus testing in cervical
as the analysed period was shorter than the recommended cancer prevention (MARCH): a community-based randomised
intervals. Also, our study might be affected by selection controlled trial. Lancet 2011; 378: 1868–73.
bias because PSM controls for measured confounders, 8 Arrossi S, Thouyaret L, Herrero R, et al. Effect of self-collection of
HPV DNA offered by community health workers at home visits on
but, unlike randomisation, does not control for uptake of screening for cervical cancer (the EMA study):
unmeasured or unknown confounders. Finally, our study a population-based cluster-randomised trial. Lancet Glob Health
design did not allow the measurement of detection of 2015; 3: e85–94.
9 Sankaranarayanan R, Qiao YL, Keita N. The next steps in cervical
CIN2+ in subsequent rounds of screening, and we were screening. Womens Health (Lond) 2015; 11: 201–12.
not able to measure the effect of HPV testing on cervical 10 Arrossi S, Paolino M, Sankaranarayanan R. Challenges faced by
cancer incidence. cervical cancer prevention programs in developing countries:
a situational analysis of program organization in Argentina.
In summary, HPV testing resulted in increased Rev Panam Salud Publica 2010; 28: 249–257.
detection of CIN2+ lesions in a middle-income setting. 11 Arrossi S, Thouyaret L, Laudi R, et al. Implementation of HPV
This increased detection was achieved in the context of testing for cervical cancer screening in programmatic contexts:
the Jujuy demonstration project in Argentina. Int J Cancer 2015;
programme reorganisation that included laboratory and 137: 1709–18.
referral network reorganisation, use of self-collection to 12 Tougher S, Dutt V, Pereira S, et al. Effect of a multifaceted social
increase coverage, and development of mechanisms to franchising model on quality and coverage of maternal, newborn,
and reproductive health-care services in Uttar Pradesh, India:
assure adherence to guidelines. Our study provides key, a quasi-experimental study. Lancet Glob Health 2018; 6: e211–221.
real-world evidence for low-income and middle-income 13 Khandker SR, Koolwal GB, Samad HA. Handbook on impact
countries to incorporate HPV testing. evaluation: quantitative methods and practices. World Bank. 2010.
http://bit.ly/1d2Ve8m (accessed April 18, 2019).
Contributors
14 Arrossi S, Paolino M. Proyecto para el mejoramiento del Programa
SA originally conceived the study and secured research support. SA was Nacional de Prevencion de Cancer de Cuello Uterino en Argentina
the principal investigator and study coordinator. MP was the investigator [in Spanish]. Informe final: Diagnóstico de situación del Programa
in charge of monitoring and evaluation. JG and MP did all statistical Nacional y Programas Provinciales. Buenos Aires: Organización
analysis, wrote the description of the statistical analysis, and (in Panamericana de la Salud, 2008. http://www.msal.gov.ar/inc/
consultation with coauthors) produced the figures and tables. RH and RL recursos-de-comunicacion/proyecto-para-el-mejoramiento-del-
made substantial contributions to the conception, design, and analysis of programa-nacional-de-prevencion-de-cancer-de-cuello-uterino-en-
the study. OM is chief of the HPV Laboratory in Jujuy. CF is Chief of the argentina/ (accessed April 18, 2019).
Cervical Pathology Service at Pablo Soria Hospital. AC was the coordinator 15 Arrossi S, Paolino M, Thouyaret L, Laudi R, Campanera A.
of the Provincial Program on Cervical Cancer Prevention. VS was largely Evaluation of scaling-up of HPV self-collection offered by
involved with project implementation. LT made substantial contributions community health workers at home visits to increase screening
to study design and implementation. All authors were involved in among socially vulnerable under-screened women in Jujuy
interpretation of data and critical revision of the manuscript. Province, Argentina. Implement Sci 2017; 12: 17.
16 WHO. WHO Compendium of innovative health technologies for
Declaration of interests low-resource settings, 2011–2014. 2015. http://apps.who.int/
We declare no competing interests. medicinedocs/documents/s22283en/s22283en.pdf (accessed
April 19, 2019).
Acknowledgments
17 Glasgow RE, Vogt TM, Boles SM. Evaluating the public health
This Article presents independent research funded by the Argentinian
impact of health promotion interventions: the RE-AIM framework.
National Cancer Institute. The analysis of follow-up was partly funded by Am J Public Health 1999; 89: 1322–27.
the National Argentinian Council of Scientific and Technologic Research
18 Gaglio B , Shoup JA, Glasgow RE. The RE-AIM framework:
(Project PIP 0337/2013). The views expressed are our own and not a systematic review of use over time. Am J Public Health 2013;
necessarily those of the Argentinian National Cancer Institute, 103: e38–46.

www.thelancet.com/lancetgh Vol 7 June 2019 e782

 Articles

19 Rabin BA, Brownson RC. Developing the terminology for 34 Ministerio de Salud de la Nación. Recomendaciones para el
dissemination and implementation research. In: Brownson RC, tamizaje, seguimiento y tratamiento de mujeres para la prevención
Colditz G, Procter E, eds. Dissemination and implementation del cáncer cervico-uterino en el marco de la incorporación de la
research in health. New York: Oxford University Press, 2012: 23–54. prueba de VPH. 2011. http://www.msal.gob.ar/images/stories/bes/
20 Proctor E, Brownson R. Measurement issues in dissemination and graficos/0000000017cnt-manual_recomendaciones_tamizaje_2015_
implementation research. In: Brownson RC, Colditz G, baja.pdf (accessed April 18, 2019).
Procter E, eds. Dissemination and implementation research in 35 Arrossi S, Curotto M, Thouyaret L, Paolino M, Cuberli M, Laudi R.
health. New York: Oxford University Press, 2012. Manual para la implementación del test de VPH en contexto
21 WHO. WHO Director-General calls for all countries to take action programático. 2016. http://www.msal.gob.ar/images/stories/bes/
to help end the suffering caused by cervical cancer. 2018. graficos/0000000906cnt-2016-11-08-manual-para-la-implementacion-
https://www.who.int/reproductivehealth/DG_Call-to-Action.pdf test-vph-en-contexto-programatico.pdf (accessed April 19, 2019).
(accessed April 19, 2019). 36 Murillo R, Almonte M, Pereira A, et al. Cervical cancer screening
22 Leinonen MK, Nieminen P, Lönnberg S, et al. Detection rates of programs in Latin America and the Caribbean. Vaccine 2008;
precancerous and cancerous cervical lesions within one screening 26 (suppl 11): L37–48.
round of primary human papillomavirus DNA testing: prospective 37 Paolino M, Arrossi S. Analysis of the reasons for abandoning the
randomised trial in Finland. BMJ 2012; 345: e7789. follow-up and treatment process in women with pre-cancerous
23 Ronco G, Giorgi-Rossi P, Carozzi F, et al. Efficacy of human cervical lesions in the province of Jujuy: implications for health
papillomavirus testing for the detection of invasive cervical cancers management. Salud Colect 2012; 8: 247–61 (in Spanish).
and cervical intraepithelial neoplasia: a randomised controlled trial. 38 Sancho-Garnier H, Tamalet C, Halfon P, et al. HPV self-sampling
Lancet Oncol 2010; 11: 249–57. or the pap-smear: a randomized study among cervical screening
24 Arbyn M, Ronco G, Anttila A, et al. Evidence regarding human non-attenders from lower socioeconomic groups in France.
papillomavirus testing in secondary prevention of cervical cancer. Int J Cancer 2013; 133: 2681–87.
Vaccine 2012; 30 (suppl 5): F88–99. 39 Arbyn M, Smith SB, Temin S, Sultana F, Castle P. Collaboration on
25 Wentzensen N, Schiffman M, Palmer T, Arbyn M. Triage of HPV self-sampling and HPV testing. Detecting cervical precancer and
positive women in cervical cancer screening. J Clin Virol 2016; reaching underscreened women by using HPV testing on
76 (suppl 1): S49–55. self-samples: updated meta-analyses. BMJ 2018; 363: 4823.
26 Arrossi S. El impacto de la prueba de VPH en los programas de 40 Rebolj M, Lynge E. Incomplete follow-up of positive HPV tests:
tamizaje en América Latina: el caso de Argentina. overview of randomised controlled trials on primary cervical
Salud Publica Mex 2019; 61: 86–94 (in Spanish). screening. Br J Cancer 2010; 103: 310–14.
27 Herrero R. Cervical cancer screening in low and middle-income 41 Poli UR, Gowrishankar S, Swain M, Jeronimo J. Triage of women
countries. HPV World 2017; 1: 6–9. testing positive with the careHPV Test on self collected vaginal
28 Ronco G, Giorgi Rossi P. Role of HPV DNA testing in modern samples for cervical cancer screening in a low-resource setting.
gynaecological practice. Best Pract Res Clin Obstet Gynaecol 2018; J Glob Oncol 2018; 4: 1–7.
47: 107–18 42 Lazcano-Ponce E, Lörincz AT, Salmerón J, et al. A pilot study of HPV
29 Elfström K, Smelov V, Johansson ALV, et al. Long term duration of DNA and cytology testing in 50,159 women in the routine Mexican
protective effect for HPV negative women: follow-up of primary Social Security Program. Cancer Causes Control 2010; 21: 1693–700.
HPV screening randomised controlled trial. BMJ 2014; 348: g130. 43 Lazcano-Ponce EC, Alonso de Ruiz P, López-Carrillo L, et al.
30 Sasieni P. The difference in sensitivity between HPV testing and Validity and reproducibility of cytologic diagnosis in a sample of
cytology for detecting current and future CIN2+ increases over cervical cancer screening centers in Mexico. Acta Cytol 1997;
time. Evid Based Med 2014; 19: 184. 41: 277–84.
31 Melnikow J, Henderson JT, Burda BU, Senger CA, Durbin S, 44 Arbyn M, Kyrgiou M, Simoens C, et al. Perinatal mortality and
Weyrich MS. Screening for cervical cancer with high-risk human other severe adverse pregnency outcomes associated with treatment
papillomavirus testing: updated evidence report and systematic of cervical intraepithelial neoplasia: meta-analysis. BMJ 2008;
review for the US Preventive Services Task Force. JAMA 2018; 337: a1284.
320: 687–705. 45 Atun R, de Jongh T, Secci F, Ohiri K, Adeyi O. Integration of
32 Lazcano-Ponce E, Lőrincz AT, Torres L, et al. Specimen targeted health interventions into health systems: a conceptual
self-collection and HPV DNA screening in a pilot study of framework for analysis. Health Policy Plan 2010; 25: 104–11.
100,242 women. Int J Cancer 2014; 135: 109–16.
33 Maciel, A and Castaño. Guías de manejo 2015 Citología anormal y
lesiones intraepiteliales cervicales, 2015. 2015. http://www.colpoweb.
org/docs/2015/Guias-de-manejo-2015.pdf (accessed April 19, 2019).

e783 www.thelancet.com/lancetgh Vol 7 June 2019