Articles Use of hormonal contraceptives and risk of HIV-1 transmission: a prospective cohort study Renee Heff ron, Deborah Donnell, Helen Rees, Connie Celum, Nelly Mugo, Edwin Were, Guy de Bruyn, Edith Nakku-Joloba, Kenneth Ngure, James Kiarie, Robert W Coombs, Jared M Baeten, for the Partners in Prevention HSV/HIV Transmission Study Team* Background Hormonal contraceptives are used widely but their eﬀ ects on HIV-1 risk are unclear. We aimed to assess Published Online the association between hormonal contraceptive use and risk of HIV-1 acquisition by women and HIV-1 transmission DOI:10.1016/S1473- from HIV-1-infected women to their male partners.
3099(11)70247-X See Online/Comment Methods In this prospective study, we followed up 3790 heterosexual HIV-1-serodiscordant couples parti cipating in DOI:10.1016/S1473- two longitudinal studies of HIV-1 incidence in seven African countries. Among injectable and oral hormonal 3099(11)70254-7 contraceptive users and non-users, we compared rates of HIV-1 acquisition by women and HIV-1 transmission from *Members listed at end of paper women to men. The primary outcome measure was HIV-1 seroconversion. We used Cox proportional hazards Department of Epidemiology regression and marginal structural modelling to assess the eﬀ ect of contraceptive use on HIV-1 risk.
(R Heff ron MPH, Prof C Celum MD, J M Baeten MD), Global Health (C Celum, N Mugo MBChB, Findings Among 1314 couples in which the HIV-1-seronegative partner was female (median follow-up J Kiarie MBChB, J M Baeten), 18·0 [IQR 12·6–24·2] months), rates of HIV-1 acquisition were 6·61 per 100 person-years in women who used Medicine (C Celum, hormonal contraception and 3·78 per 100 person-years in those who did not (adjusted hazard ratio 1·98, 95% CI Prof R W Coombs MD,
J M Baeten), and Laboratory 1·06–3·68, p=0·03). Among 2476 couples in which the HIV-1-seronegative partner was male (median follow-up Medicine (R W Coombs), 18·7 [IQR 12·8–24·2] months), rates of HIV-1 transmission from women to men were 2·61 per 100 person-years in University of Washington, couples in which women used hormonal contraception and 1·51 per 100 person-years in couples in which women Seattle, WA, USA; Statistical did not use hormonal contraception (adjusted hazard ratio 1·97, 95% CI 1·12–3·45, p=0·02). Marginal structural Center for HIV/AIDS Research and Prevention (D Donnell PhD) model analyses generated much the same results to the Cox proportional hazards regression. and Vaccine and Infectious Disease Division (D Donnell), Interpretation Women should be counselled about potentially increased risk of HIV-1 acquisition and transmission Fred Hutchinson Cancer Research Center, Seattle, WA, with hormonal contraception, especially injectable methods, and about the importance of dual protection with USA; Wits Reproductive Health condoms to decrease HIV-1 risk. Non-hormonal or low-dose hormonal contraceptive methods should be considered and HIV Institute for women with or at-risk for HIV-1.
(Prof H Rees MBBChir) and Perinatal Health and HIV Research Unit Funding US National Institutes of Health and the Bill & Melinda Gates Foundation.
(G de Bruyn MBBCh), University of the Witwatersrand, Introduction
contraceptive use and risk of HIV-1 acquisition by Johannesburg, South Africa;
Safe and eﬀ ective family planning services are central to women and HIV-1 transmission from HIV-1-infected Department of Obstetrics and Gynaecology, Kenyatta
initiatives to reduce unintended pregnancies, promote women to their male partners.
National Hospital, Nairobi,
economic development, and improve the health of
Kenya (N Mugo, J Kiarie,
women and children worldwide. Among women with Methods
K Ngure MPH); Department of
and at-risk for HIV-1, the prevention of unintended Study design and participants Obstetrics & Gynaecology, University of Nairobi, Nairobi,
pregnancy is a key component of strategies to reduce From 2004–10, we did two prospective studies of HIV-1 Kenya (N Mugo, J Kiarie);
incidence in African HIV-1-serodiscordant couples (ie, Department of Reproductive
Hormonal contraceptive methods, including daily oral one partner with HIV-1 infection and one partner Health, Moi University, Eldoret,
pills and long-acting injectables, are used by more than without). The Partners in Prevention HSV/HIV Kenya (E Were MBChB); School of Public Health, Makerere
140 million women worldwide.3 During the past two Transmission Study was a randomised, placebo- University College of Health
decades, epidemiological and laboratory studies have controlled, trial of daily acyclovir herpes simplex virus Sciences, Kampala, Uganda
suggested that hormonal contraception could alter the type 2 (HSV-2) suppressive therapy given to 3408 people (E Nakku-Joloba MBChB); and
risk of HIV-1 acquisition in women.4–8 However, results infected with HIV-1 and HSV-2 as an intervention to STD Clinic, Mulago Hospital, Kampala, Uganda
have been inconsistent.9 Only one study has addressed reduce HIV-1 transmission to their heterosexual HIV-1- (E Nakku-Joloba)
the eﬀ ect of hormonal contraception and risk of HIV-1 seronegative partners (Clinicaltrials.gov #NCT00194519); Correspondence to:
transmission from women to men.10 Increased risk acyclovir did not signifi cantly reduce HIV-1 transmission.13 Dr Jared M Baeten, Department
related to hormonal contraceptive use would be of Couples were from seven countries in east and southern of Global Health, University of
importance to global public health because of the large Africa (Botswana, Kenya, Rwanda, South Africa, Washington, Box 359927,
number of women using such methods. WHO has called Tanzania, Uganda, and Zambia) and were followed for WA 98104, USA
for high-quality studies to assess the potential role of up to 24 months. In a parallel study at two of the clinical email@example.com
hormonal contraception in increased HIV-1 risk.11,12 We trial sites (Kampala, Uganda, and Soweto, South Africa), aimed to assess the association between hormonal we enrolled an additional 485 HIV-1 serodiscordant
www.thelancet.com/infection Published online October 4, 2011 DOI:10.1016/S1473-3099(11)70247-X Articles
couples into an observational study of immune correlates enrolment from HIV-1- infected women in the clinical trial of HIV-1 protection and followed them for up to cohort, as previously detailed.21 The lower quantifi cation 12 months. For both studies, eligible participants were limit for HIV-1 RNA testing was 240 copies. aged 18 years or older and sexually active, HIV-1-
At each quarterly study visit, women were asked about
seropositive partners had no history of AIDS-defi ning their current contraceptive method with a standard disorders and were not using antiretroviral therapy questionnaire. Women were classifi ed as exposed to (ART), the HIV-1-seropositive partners in the clinical trial hormonal contraception for each quarterly period if they had CD4 counts of 250 cells per μL or higher, were reported hormonal use at the quarterly visit. seropositive for HSV-2, had no known history of adverse Contraceptive use was analysed as a time-dependent reactions to acyclovir, and were not pregnant. Couples exposure, with women assumed to have used the same were recruited through study-initiated community method during the 3 months that elapsed between study outreach activities and referrals from HIV-1 testing and visits. Analyses were done for exposure to any hormonal care centres, antenatal clinics, and non-governmental contraception and then separately for injectable and oral organisations.14 The main reasons couples did not enrol contraception; the comparison group was women not were that they did not meet the CD4 count, HSV-2, using hormonal contraception, which included women pregnancy, or sexual activity eligibility criteria.15
who had had a hysterectomy or tubal ligation, used
HIV-1-uninfected partners were seen quarterly for HIV-1 condoms only, or used no contraception. Visits at which
serological testing. For HIV-1-infected partners, CD4 women reported use of implantable hormonal methods counts were measured every 6 months, and participants or an intrauterine device were rare (<2% of visits) and eligible for ART initiation during follow-up were referred therefore excluded. Many women reported condom use, to local HIV-1 care clinics. All participants received either with or without another method for contraception; comprehensive HIV-1-prevention services, including condom use was thus included in analyses as a potential individual and couples counselling, free condoms, and confounder. HIV-1-uninfected men were classifi ed as treatment of sexually transmitted infections. Contra-
exposed to hormonal contraception if their HIV-1-
ceptives were oﬀ ered by referral or on-site. Diﬀ erences in infected female partner reported using an injectable or contraceptive use occurred between sites.16,17
oral method at her corresponding study visit. For 4% of
We excluded enrolled patients who were sub sequently male follow-up time, missing contraceptive data from
reported to not have HSV-2 or HIV-1 infection and their female partners were imputed to be the method couples for whom the HIV-1 uninfected participant did consistently reported at adjacent study visits; data were not complete any follow-up visits for assessment of HIV-1 not imputed if methods during adjacent periods were seroconversion. We also censored visits for couples for inconsistent. whom the HIV-1 infected partner started ART, because such treatment eliminated HIV-1 risk in the study Statistical analysis population.18 The protocols were approved by institutional The primary outcome measure was HIV-1 seroconversion. review boards at the University of Washington and We did separate analyses of the association of hormonal collaborating institutions at each study site. Participants contraception with HIV-1 acquisition by women (male- provided written informed consent.
to-female transmission) and HIV-1 transmission from women to men (female-to-male transmission). For
female-to-male transmission, only genetically linked
Rapid HIV-1 antibody tests were used for HIV-1 serological seroconversions were included as outcomes to minimise testing and positive results were confi rmed by ELISA.13 For misclassifi cation of HIV-1 transmissions from outside HIV-1 seroconverters, analysis of HIV-1 env and gag partners with unknown hormonal contraceptive use, and sequences from both members of a couple was used to follow-up time was censored for those men at the time establish whether transmission was linked within the they acquired HIV-1 from a partner other than the partnership.19 Nucleic-acid-amplifi cation testing for HIV-1-infected partner with whom they enrolled. bacterial sexually transmitted infections (STIs) was done
We compared participant characteristics during periods
on samples collected from both partners at study of hormonal contraceptive use and non-use with enrolment.15 All participants were tested for HSV-2 with generalised estimating equations. To assess the eﬀ ect of HerpeSelect-2 EIA (Focus Technologies, Cypress, CA, contraceptive method on HIV-1 risk, we used time-USA) or by HSV-specifi c western blot.20 CD4 quantifi cation dependent Cox proportional hazards regression with was done with standard fl ow cytometry. Plasma HIV-1 robust standard errors to account for within subject RNA concentrations were quantifi ed from a sample correlation with repeated measurements.22 Models were collected at study enrolment and 6 months later with the adjusted for variables that had confounded the COBAS TaqMan real-time HIV-1 RNA assay, version 1.0 contraception–HIV-1-risk relation in previous analyses7,8—(Roche Diagnostics, Indianapolis, IN, USA). Con-
age and time-dependent pregnancy and any sex without
centrations of endocervical HIV-1 were quantifi ed with the condoms—and plasma HIV-1 concentrations in HIV-1-COBAS assay of swab samples collected 6 months after infected partners, a strong predictor of HIV-1
www.thelancet.com/infection Published online October 4, 2011 DOI:10.1016/S1473-3099(11)70247-X Articles
transmission.23 We also assessed several additional
We repeated our analyses with marginal structural
variables for potential confounding: region (east Africa vs modelling, a technique to adjust for time-dependent southern Africa), marital status of couples and the confounding.25,26 We computed stabilised-inverse-probability number of children they had together, HSV-2 status of the weights with logistic regression to predict the probability of HIV-1-uninfected partner, circumcision status of the male hormonal contraceptive use at each visit (by concentrations partner, and STI in either partner, all recorded at study of plasma HIV-1, age, region, and number of children), as enrolment, and time-dependent measures of sexual described by Cole and colleagues;27 the weights adjusted frequency (with and without condoms), sex with additional for time-dependent measures of pregnancy and partners, CD4 count of the HIV-1-infected partner, and unprotected sex. Weights for the eﬀ ect of any hormonal genital-ulcer disease in either partner. None of these contraception on HIV-1 risk (mean 1·00, range 0·82–1·34) additional variables substantially (>10%) changed the were computed separately from the weights to assess the eﬀ ect estimates and thus they were not included in the separate eﬀ ects of injectable and oral contraception on fi nal multivariate models. For analysis of HIV-1 acquisition HIV-1 risk (mean 1·07, range 0·19–4·56). These weights by women, we tested for eﬀ ect modifi cation by baseline were then used in a pooled logistic-regression model of HSV-2 status and age with a likelihood-ratio test, because hormonal contraception versus HIV-1 risk. previous results have shown that the hormonal
Finally, we assessed the prevalence and quantity of
contraception–HIV-1-risk relation was stronger for genital HIV-1 RNA in women using hormonal contra cep-women who were HSV-2 seronegative or who were aged tion versus those who did not by logistic and linear less than 25 years.24
regression. All analyses were done with SAS 9.2. Analysis of HIV-1 acquisition by women Analysis of HIV-1 transmission from women (N=1314 couples) to men (N=2476 couples)
HIV-1-uninfected men HIV-1-infected women
Demographic characteristics Age, years Couple characteristics* Married Sexual behaviour, month before enrolment Number of sex acts with study partner Medical characteristics Sexually transmitted infection† HIV-1 characteristics Plasma HIV-1 RNA, (log copies per mL) at enrolment Contraceptive use (women) Any hormonal contraceptive use at enrolment
Any hormonal contraceptive use during follow up
Any injectable contraceptive use during follow up
Any oral contraceptive use during follow up
Data are number (%), n/N (%), or median (IQR). HSV-2=herpes simplex virus type-2. ART=antiretroviral therapy. *Data shown in the uninfected group apply to couples. †Neisseria gonorrhoeae, Chlamydia trachomatis, or Trichomoas vaginalis; 678 (75%) of 904 participants with a sexually transmitted infection were infected with T vaginalis only, 79 (9%) participants had N gonorrhoea, and 161 (18%) had C trachomatis.
Table 1: Participant characteristics
www.thelancet.com/infection Published online October 4, 2011 DOI:10.1016/S1473-3099(11)70247-X Articles Follow-up intervals for analysis of HIV-1 acquisition by Follow-up intervals for analysis of HIV-1 transmission from women (N=1314 HIV-1-seronegative women) women to men (N=2476 HIV-1-seropositive women) Demographic characteristics Age of HIV-1 seronegative partner, years Sexual behaviour, HIV-1-uninfected partner Any unprotected sex with study partner, past month
Any sex with an outside partner, past month
Medical characteristics CD4 count (cells per μL) in the HIV-1 infected partner
Plasma HIV-1 concentration (log copies per mL) in the
HIV-1-infected partnerPregnant, female partner†
Data are n/N (%) or median (IQR). *Comparisons among contraceptive-exposure groups are adjusted for correlation by multiple measures from the same woman with generalised estimating equations. The number of data points assessed for each cell is total number of visits with each covariate characteristic during study follow-up. †Contraceptive use during pregnancy intervals was either contraceptive failures documented at the time of pregnancy detection or contraceptive uptake during the early postpartum period. Table 2: Participant characteristics during quarterly follow-up intervals with and without hormonal contraceptive use Role of the funding source
women did not switch contraceptive methods during
The sponsor of the study had no role in study design, follow-up (1085 [83%] of 1314 HIV-1-seronegative women data collection, data analysis, data interpretation, or and 1909 [77%] of 2476 HIV-1-seropositive women). writing of the report. The corresponding author had full However, among 1321 women who ever used hormonal access to all the data in the study and had fi nal con traception during the study, 634 (48%; 448 [48%] of responsibility for the decision to submit for publication.
945 HIV-1-seropositive women and 186 [49%] of 376 HIV-1 seronegative women) were not using such methods at
For most of the 3790 HIV-1-serodiscordant couples
Median follow-up for HIV-1-seronegative women was
included in the analysis, the HIV-1-infected partner was 18·0 (IQR 12·6–24·2) months and for seronegative men female (table 1). Most couples were married with children. was 18·7 months (IQR 12·8–24·2). Retention at The median age was in the mid-30s, and 321 (24%) of 12 months for HIV-1-seronegative women was 93% (1153 1314 uninfected women were younger than 25 years. In of 1238 women) and for seronegative men was 90% HIV-1-seropositive participants, the median CD4 count (2098 of 2331 men). Retention at 24 months for was 455 cells per μL (IQR 337–626) and median plasma HIV-1-sero negative women was 87% (423 of 484 women) HIV-1 RNA concentration was 4·10 log copies per mL and for seronegative men was 84% (812 of 970 men).
(IQR 3·37–4·73). More than a quarter of women were HIV-1-seronegative partners accrued 5157·9 person-pregnant during study follow-up (table 1).
years of follow-up for assessment of HIV-1 seroincidence,
27 couples enrolled in the randomised trial13 were during which 167 HIV-1 seroconversions occurred. Of
subsequently reported not to have HSV-2 or HIV-1 the 73 infections in women, 62 (85%) were determined infection and were excluded from the analysis, as were by viral sequencing to be genetically linked within the 76 couples in which the HIV-1-uninfected participant did partnership, and of the 93 infections in men, 59 (63%) not complete any follow-up visits for assessment of HIV-1 were linked. seroconversion. For 350 couples (151 with an HIV-1-
During follow-up, hormonal contraceptives were used
uninfected woman and 199 with an HIV-1-uninfected more frequently by couples with young HIV-1-uninfected man) in which the HIV-1-infected partner started ART, partners and couples who did not experience pregnancy subsequent visits were censored.18
(table 2). Sexual behaviours did not diﬀ er for HIV-1-
At enrolment, 194 (15%) of 1314 HIV-1-seronegative and uninfected women during periods when they were using
430 (17%) of 2476 HIV-1-seropositive women used hormonal contraception compared with when they were hormonal contraception; injectable contraception was not using hormonal contraception. Unprotected sex was more commonly used than oral pills (477 [13%] of more likely and sex with an external partner was less 3790 women used injectable vs 147 [4%] who used oral likely for HIV-1-uninfected men, during periods when contraception). 275 (21%) of 1314 HIV-1-seronegative their female partner was using hormonal contraception women and 815 (33%) of 2476 HIV-1-seropositive women than it was when their partner was not taking hormonal used hormonal methods during study follow-up. Most contraception. Concentrations of plasma HIV-1 RNA and
www.thelancet.com/infection Published online October 4, 2011 DOI:10.1016/S1473-3099(11)70247-X Articles Number of HIV-1 Incidence per Unadjusted Cox proportional Adjusted Cox proportional Adjusted marginal structural seroconversions/person-years 100 person-years hazards regression analysis hazards regression analysis* models analysis†
*Multivariate Cox proportional hazard regression model, adjusted for age, concentrations of plasma HIV-1 in the HIV-1-infected partners, and time varying unprotected sex and pregnancy. Further adjustment for additional factors did not substantially change the fi ndings. †Weighted marginal structural model is adjusted for age, region, number of children, concentration of plasma HIV-1 RNA in the HIV-1-infected partner, and month of visit (5-knot cubic spline with knots at the 5th, 25th, 50th, 75th, and 95th percentiles) and contraceptive history; weights are truncated at the 1st and 99th percentiles. Table 3: Hormonal contraceptive use and risk of HIV-1 acquisition by women Number of genetically linked Incidence per Unadjusted Cox proportional Adjusted Cox proportional Adjusted marginal structural HIV-1 seroconversions/ 100 person-years hazards regression analysis hazards regression analysis* model analysis† person-years
*Multivariate Cox proportional hazard regression model, adjusted for age, plasma HIV-1 levels in the HIV-1 infected partner, and time varying unprotected sex and pregnancy. Further adjustment for additional factors did not substantially change the fi ndings. †Weighted marginal structural model is adjusted for age, region, number of children, plasma HIV-1 RNA concentration in the HIV-1 infected partner, and visit month (5-knot cubic spline with knots at the 5th, 25th, 50th, 75th and 95th percentiles) and contraceptive history; weights are truncated at the 1st and 99th percentiles. Table 4: Hormonal contraceptive use and risk of HIV-1 transmission from women to men
CD4 counts were similar for hormonal-contraception multivariate analysis adjusted for age, pregnancy, exposed versus unexposed periods.
unprotected sex, and concentrations of plasma HIV-1 in
Ratesof HIV-1 acquisition were higher in women using HIV-1-infected partners, men’s HIV-1 risk was increased
hormonal contraception than in those who were not two times when their partners were using hormonal (table 3). In multivariate Cox proportional hazards contraception (adjusted hazard ratio 1·97, 95% analysis adjusted for age, pregnancy, unprotected sex, CI 1·12–3·45; table 4). Both injectable and oral and concentrations of plasma HIV-1 in HIV-1-infected contraceptive use by female partners were associated with partners, use of hormonal contraceptives was associated increased HIV-1 risk for men, although the eﬀ ect was with a two times increased risk of HIV-1 acquisition signifi cant only for injectable contraception (table 4). The (adjusted hazard ratio 1·98, 95% CI 1·06–3·68). Increased marginal structural model analyses generated similar risk was reported for both injectable (adjusted hazard results to the Cox proportional hazards regression. ratio 2·05, 95% CI 1·04–4·04) and oral contraceptive use
To account for the potential persistent biological eﬀ ects
(1·80, 0·55–5·82), although the analysis of oral of hormonal contraception on HIV-1 risk when women contraceptive use included only 50·5 person-years and switched contraceptive methods, we assessed the eﬀ ect of was not statistically signifi cant. The results from the extending the exposure window for 3 months after last marginal structural models were generally in agreement hormonal contraceptive use (thus, women could be with the Cox regression models. No evidence showed exposed to more than one method during one study visit that the eﬀ ect of hormonal contraception on HIV-1 risk window). This assessment aﬀ ected 32·1 (2%) of the 1782·8 was diﬀ erent for HSV-2-seronegative women (195 [15%] person-years and one seroconversion event for the HIV-1 of 1283) versus seropositive women (adjusted hazard acquisition analysis and 70·9 (2%) of the 3375·1 person-ratio 1·56 vs 2·00, p
=0·82) or for women younger years and one event for the female-to-male transmission
than 25 years (321 [24%] of 1314) versus those 25 years or analysis. The results of these analyses were not older (adjusted hazard ratio 1·96 vs 2·21, p
substantially diﬀ erent than those shown in table 3
The rate of HIV-1 transmission from women using and table 4 (data not shown). When we limited the analysis
hormonal contraceptives to their male partners was of HIV-1 acquisition by women to those 62 outcomes that higher than was the rate of transmission from women were genetically linked to their male partners, the eﬀ ect who did not use hormonal contraceptives (table 4). In estimates were not substantially changed (for any
www.thelancet.com/infection Published online October 4, 2011 DOI:10.1016/S1473-3099(11)70247-X Articles Detection of any genital HIV-1 RNA Quantity of genital HIV-1 RNA detected (log copies/swab)
contraceptionAny hormonal 230/358 (64·3)
2·96 (2·08 to 3·65) –0·07 (–0·28 to 0·14) 0·53
*Average diff erence in HIV-1 RNA concentration. †Adjusted for concentration of plasma HIV-1 RNA and CD4 count. Table 5: Endocervical concentrations of HIV-1 RNA in HIV-1 seropositive women, by contraceptive method
hormonal contraceptive use, Cox regression adjusted HIV-1 risk and our results are insuﬃ
hazard ratio 2·06, 95% CI 1·05–4·03 and marginal defi nitive conclusions about oral contraceptive use and structural model odds ratio 2·01, 95% CI 1·02–3·95). In a HIV-1 risk. Our results were robust when we adjusted for third sensitivity analysis, we censored observations during multiple potential confounding factors, undertook pregnancy and adjusted our Cox model for age, unprotected diﬀ erent analytical approaches, and did sensitivity sex, and concentrations of plasma HIV-1 in HIV-1-infected analyses. partners. We did not see substantial diﬀ erences in the
Previous studies of HIV-1-acquisition risk related to
eﬀ ect estimates (Cox regression adjusted hazard ratio contraceptive use have had inconsistent results, partly 1·84, 95% CI 0·97–3·49, for the association of hormonal because of variable methodological quality.9 As a result, contraception and HIV-1 acquisition among women and public
1·86, 1·04–3·32, for the association of hormonal counselling and strategies to promote alternative contraception and HIV-1 transmission to men) for this contraceptive methods for women with or at risk of approach compared with our primary study models.
HIV-1—have not been implemented. Our fi ndings
We measured endocervical concentrations of HIV-1 provide new data that show that contraception might
RNA from a single timepoint in 1691 HIV-1-infected increase a woman’s risk of acquiring HIV-1, and they are women (table 5). Women using injectable contraception consistent with longitudinal studies of sex workers in at the time of endocervical-sample collection were more Kenya and family planning attendees from Uganda and likely to have genital HIV-1 RNA detected than were those Zimbabwe.7,24 Moreover, to our knowledge, ours is the not using hormonal contraception. Concentrations of fi rst prospective study to show increased HIV-1 risk in genital HIV-1 RNA were also higher in those using male partners of HIV-1-infected women using hormonal injectable contraception than in those not using hormonal contraception. We noted raised concentrations of HIV-1 contraception, by an average of 0·19 log copies per swab, RNA in endocervical secretions from HIV-1-infected
after adjustment for plasma HIV-1 concentrations and women using injectable methods, suggesting a potential CD4 cell count (table 5). No association was identifi ed mechanism for increased risk of HIV-1 transmission. between contraception and concentrations of plasma Studies of HIV-1 transmission from women to men are HIV-1 RNA collected at the same time as endocervical urgently needed to confi rm or refute our fi ndings. samples (median 3·91 log copies per mL [IQR 3·07–4·50]
Hormonal contraceptives might have physiological
for injectable users vs 4·03 log copies per mL actions beyond pregnancy prevention, including possible
[IQR 3·22–4·65] for non-users; p=0·10), suggesting a risks of bone-density loss, cervical cancer, and Chlamydia localised eﬀ ect of hormonal contraception on increased trachomatis.28–30 Clinical and laboratory studies have concentrations of HIV-1 in the female genital tract.
suggested possible mechanisms by which hormonal contraception could infl uence HIV-1 susceptibility and
infectiousness including changes to vaginal structure,
Use of hormonal contraceptives was associated with a cytokine regulation, CCR5 expression, and cervicovaginal two-times increase in the risk of HIV-1 acquisition by HIV-1 shedding.31women and HIV-1 transmission from women to men.
Our analyses controlled for age, pregnancy, condom
Injectable methods were the most common form of use, and HIV-1 concentrations in the infected partner; hormonal contraception used by our study population controlling for additional demographic, clinical, and and subgroup analyses showed signifi cantly increased behavioural factors did not alter our results. Only a HIV-1 risk associated with injectable use. Few women clinical trial with random assignment of women to used oral contraceptives in our study; oral contraceptive eﬀ ective hormonal contraception versus non-hormonal use was associated with a non-signifi cant increase in contraception could defi nitively assess HIV-1 risk from
www.thelancet.com/infection Published online October 4, 2011 DOI:10.1016/S1473-3099(11)70247-X Articles
heightened HIV-1 risk. The benefi ts of eﬀ ective
Panel: Research in context
hormonal contraceptive methods are unequivocal and
must be balanced with the risk for HIV-1 infection. Our
We searched PubMed up to July, 2011, to identify studies
fi ndings argue for policies to counsel women about the
relating use of hormonal contraceptives to HIV-1 risk, with
potential for increased HIV-1 risk with hormonal
the search terms “hormonal contraception”, “hormonal
contraceptive use, especially injectable DMPA use, and
contraceptive”, “HIV-1”, and “HIV-1 acquisition or transmission”
the importance of dual protection with condoms to
in diff erent combinations. Additional y, systematic reviews and
one meta-analysis published on this topic were reviewed.
Our data do not provide estimates of HIV-1 risk related
to other hormonal contraceptives, such as implants,
patches, or combination injectables. Data on HIV-1 risk
Several studies show—with similar magnitude of their eff ect
associated with these methods and non-hormonal
estimates—the potential for hormonal contraception to
contraceptive methods, such as intrauterine devices, are
increase women’s risk for acquiring HIV-1, even after
urgently needed, and strategies to improve the accessibility
controlling for sexual behaviour. Our study is the fi rst with
and uptake of these lower-dose and non-hormonal
adequate power to assess and show the potential for
methods should be prioritised. Contraceptive counselling
hormonal contraceptive use by HIV-1-seropositive women to
should be combined with HIV-1 counselling and testing,
increase risk of transmitting the virus to their male partners.
with joint scale-up of both approaches essential for
These fi ndings have important implications for family
optimisation of reproductive health and HIV-1-prevention
planning and HIV-1-prevention programmes, especially in
choices for women and couples. Additionally, as national
HIV-1-prevention programmes begin to incorporate antiretroviral pre-exposure prophylaxis,33–35 this new HIV-1-prevention method could be oﬀ ered to women using
diﬀ erent contraceptive methods with certainty that bias contraceptives or their partners. in contraceptive choice and bias due to unmeasured Contributors
confounding did not infl uence the results. Such a study RH, DD, and JMB designed the study and RH and DD did the analysis. might be diﬃ cult to implement because of women’s RH and JMB wrote the initial draft. All authors contributed to the data preferences for diﬀ erent contraceptive methods and the collection and writing of the report and approved the fi nal draft. likelihood of contraceptive switching that could Partners in Prevention HSV/HIV Transmission Study Team undermine randomisation.
*Principal investigator. University of Washington Coordinating Center and Central Laboratories, Seattle, WA, USA: C Celum*, A Wald (protocol
Limitations of our study were that contraceptive use co-chair), J Lingappa (medical director), J M Baeten, M Campbell, L Corey,
was determined by self-report—we did not gather data R W Coombs, J P Hughes, A Magaret, M J McElrath, R Morrow, on adherence to contraception, and we did not record the J I Mullins. Cape Town, South Africa (University of Cape Town): D Coetzee*; specifi c brand of contraception and thus cannot comment Eldoret, Kenya (Moi University, Indiana University): K Fife*, E Were*;
Gaborone, Botswana (Botswana Harvard Partnership): M Essex*,
on HIV-1 risks from specifi c exogenous hormones. J Makhema*; Kampala, Uganda (Infectious Disease Institute, Makerere
During the study period, low-dose combination hormonal University): E Katabira*, A Ronald*; Kigali,Rwanda (Rwanda Zambia HIV oral contraceptives and long-acting injectable depot Research Group, and Emory University): S Allen*, K Kayitenkore*, E Karita*; medroxyprogesterone acetate (DMPA) were the most Kisumu, Kenya (Kenya Medical Research Institute, University of California San Francisco): E Bukusi*, C Cohen*; Kitwe, Zambia (Rwanda Zambia
commonly used methods in national family planning HIV Research Group, and Emory University): S Allen*, W Kanweka*;
programmes; few studies have assessed HIV-1 risk from Lusaka, Zambia (Rwanda Zambia HIV Research Group, and Emory other injectable methods (eg, norethisterone enanthate).12 University): S Allen*, B Vwalika*; Moshi, Tanzania (Kilimanjaro Christian Most participants in our study were participating in an Medical Col ege, Harvard University): S Kapiga*, R Manongi*; Nairobi, Kenya (University of Nairobi, University of Washington): C Farquhar*,
HIV-1-prevention randomised trial and were recruited G John-Stewart*, J Kiarie*; Ndola, Zambia (Rwanda Zambia HIV Research broadly from HIV-1 testing and care centres. Nearly all Group, and Emory University): S Allen*, M Inambao*; Orange Farm, South HIV-1-infected partners were co-infected with HSV-2; Africa (Reproductive Health Research Unit, University of the Witwatersrand): however, HSV-2 seroprevalence is more than 80% in S Delany-Moretlwe*, H Rees*; Soweto, South Africa (Perinatal HIV Research Unit, University of the Witwatersrand): G de Bruyn*, G Gray*,
HIV-1-infected people in sub-Saharan Africa.32 Thus, J McIntyre*; Thika, Kenya (University of Nairobi, University of Washington): these factors are unlikely to limit the generality of our N R Mugo*. Data management was provided by DF/Net Research (Seattle, fi ndings. We censored follow-up for those couples in USA) and site laboratory oversight was provided by Contract Lab Services
(University of the Witwatersrand, Johannesburg, South Africa).
which the HIV-1-infected partner initiated ART. Future studies with long post-ART follow-up should assess Confl icts of interest
CC has received research grant support from GlaxoSmithKline, which
whether increased risk of HIV-1 acquisition and did not include salary support, and has served on an advisory board for
transmission occurs in the context of ART use.
this company. RWC has received research grant support from the
Several observational studies have shown increased US National Institutes of Health (AI-27757 and AI-38858) and Roche
HIV-1 risk for women using hormonal contraceptives;7,24 Molecular and has served as a consultant for Abbott Molecular. JMB, RH,
and DD have received research support from the US National Institutes
our fi ndings suggest that male partners of HIV-1-
of Health. JMB, CC, GdB, RH, JK, NM, EW, and DD have received grant
infected women using hormonal contraception also face support from the Bill & Melinda Gates Foundation.
www.thelancet.com/infection Published online October 4, 2011 DOI:10.1016/S1473-3099(11)70247-X Articles Acknowledgments
17 Heﬀ ron R, Were E, Celum C, et al. A prospective study of
Funding was provided by the US National Institutes of Health (grants
contraceptive use among African women in HIV-1 serodiscordant
R03 HD068143, R01 AI083034, and P30 AI027757, and support for RH
partnerships. Sex Transm Dis 2010; 37: 621–28.
through T32 AI007140) and the Bill & Melinda Gates Foundation (grants
18 Donnell D, Baeten J, Kiarie J, et al. Heterosexual HIV-1
26469 and 41185). We thank the couples who participated in this study,
transmission after initiation of antiretroviral therapy:
the teams at the study sites, and at the University of Washington for
a prospective cohort analysis. Lancet 2010; 375: 2092–98.
work on data and sample collection and management; and Renee Ridzon
19 Campbell MS, Mullins JI, Hughes JP, et al. Viral linkage in HIV-1
from the Bill & Melinda Gates Foundation for study oversight.
seroconverters and their partners in an HIV-1 prevention clinical
trial. PLoS One 2011; 6: e16986. References
20 Ashley-Morrow R, Nollkamper J, Robinson NJ, Bishop N, Smith J.
WHO. Strategic approaches to the prevention of HIV infection in
Performance of focus ELISA tests for herpes simplex virus type 1
(HSV-1) and HSV-2 antibodies among women in ten diverse
StrategicApproachesE.pdf (accessed Dec 15, 2010).
geographical locations. Clin Microbiol Infect 2004; 10: 530–36.
Reynolds HW, Janowitz B, Wilcher R, Cates W. Contraception to
21 Baeten J, Kahle E, Lingappa J, et al. Genital HIV-1 RNA levels
prevent HIV-positive births: current contribution and potential
predict risk of heterosexual HIV-1 transmission. Sci Transl Med
cost savings in PEPFAR countries. Sex Transm Infect 2008;
2011; 3: 77ra29. 84 (suppl 2): ii49–53.
22 Kleinbaum DG, Klein M. Survival analysis: a self-learning text,
Population Division, UN Department of Economic Social Aﬀ airs.
2nd edn. New York, NY: Springer, 2005.
World Contraceptive Use 2009. http://www.un.org/esa/population/
23 Quinn TC, Wawer MJ, Sewankambo N, et al. Viral load and
heterosexual transmission of human immunodefi ciency virus
type 1. Rakai Project Study Group. N Engl J Med 2000; 342: 921–29.
Plummer FA, Simonsen JN, Cameron DW, et al. Cofactors in
24 Morrison CS, Chen PL, Kwok C, et al. Hormonal contraception and
male-female sexual transmission of human immunodefi ciency
HIV acquisition: reanalysis using marginal structural modeling.
virus type 1. J Infect Dis 1991; 163: 233–39. AIDS 2010; 24: 1778–81.
Kiddugavu M, Makumbi F, Wawer MJ, et al. Hormonal
25 Robins JM, Hernan MA, Brumback B. Marginal structural
contraceptive use and HIV-1 infection in a population-based
models and causal inference in epidemiology. Epidemiology
cohort in Rakai, Uganda. AIDS 2003; 17: 233–40.
2000; 11: 550–60.
Kleinschmidt I, Rees H, Delany S, et al. Injectable progestin
26 Hernan MA, Brumback B, Robins JM. Marginal structural models
contraceptive use and risk of HIV infection in a South African
to estimate the causal eﬀ ect of zidovudine on the survival of
family planning cohort. Contraception 2007; 75: 461–67.
HIV-positive men. Epidemiology 2000; 11: 561–70.
Baeten JM, Benki S, Chohan V, et al. Hormonal contraceptive use,
27 Cole SR, Hernan MA. Constructing inverse probability weights for
herpes simplex virus infection, and risk of HIV-1 acquisition among
marginal structural models. Am J Epidemiol 2008; 168: 656–64.
Kenyan women. AIDS 2007; 21: 1771–77.
28 Burkman RT Jr. Noncontraceptive eﬀ ects of hormonal
Morrison CS, Richardson BA, Mmiro F, et al. Hormonal
contraceptives: bone mass, sexually transmitted disease and pelvic
contraception and the risk of HIV acquisition. AIDS 2007;
infl ammatory disease, cardiovascular disease, menstrual function,
and future fertility. Am J Obstet Gynecol 1994; 170: 1569–75.
Baeten JM, Lavreys L, Overbaugh J. The infl uence of hormonal
29 Appleby P, Beral V, Berrington de Gonzalez A, et al. Cervical cancer
contraceptive use on HIV-1 transmission and disease progression.
and hormonal contraceptives: collaborative reanalysis of individual
Clin Infect Dis 2007; 45: 360–69.
data for 16 573 women with cervical cancer and 35 509 women
10 de Vincenzi I. A longitudinal study of human immunodefi ciency
without cervical cancer from 24 epidemiological studies. Lancet
virus transmission by heterosexual partners. European Study Group
2007; 370: 1609–21.
on Heterosexual Transmission of HIV. N Engl J Med 1994; 331: 341–46.
30 US Food and Drug Administration. Black box warning added
11 WHO. Hormonal contraception and HIV: science and policy. http://
concerning long-term use ofDepo-Provera Contraceptive Injection.
12 WHO. Review of priorities in research on hormonal contraception
31 Hel Z, Stringer E, Mestecky J. Sex steroid hormones,
and IUDs and HIV infection. http://whqlibdoc.who.int/hq/2010/
hormonal contraception, and the immunobiology of human
WHO_RHR_10.21_eng.pdf (accessed June 17, 2011).
immunodefi ciency virus-1 infection. Endocr Rev 2010; 31: 79–97.
13 Celum C, Wald A, Lingappa JR, et al. Acyclovir and transmission of
32 Strick LB, Wald A, Celum C. Management of herpes simplex virus
HIV-1 from persons infected with HIV-1 and HSV-2. N Engl J Med
type 2 infection in HIV type 1-infected persons. Clin Infect Dis 2006;
2010; 362: 427–39. 43: 347–56.
14 Lingappa JR, Lambdin B, Bukusi EA, et al. Regional diﬀ erences in
33 Grant RM, Lama JR, Anderson PL, et al. Preexposure
prevalence of HIV-1 discordance in Africa and enrollment of HIV-1
chemoprophylaxis for HIV prevention in men who have
discordant couples into an HIV-1 prevention trial. PLoS One 2008;
sex with men. N Engl J Med 2010; 363: 2587–99. 3: e1411.
34 Abdool Karim Q, Abdool Karim SS, Frohlich JA, et al. Eﬀ ectiveness
15 Lingappa JR, Kahle E, Mugo N, et al. Characteristics of HIV-1
and safety of tenofovir gel, an antiretroviral microbicide, for the
discordant couples enrolled in a trial of HSV-2 suppression to
prevention of HIV infection in women. Science 2010; 329: 1168–74.
reduce HIV-1 transmission: the partners study. PLoS ONE 2009;
35 Baeten J, Celum C, on behalf of the Partners PrEP Study team.
antiretroviral pre-exposure prophylaxis for HIV-1 prevention among
16 Ngure K, Heﬀ ron R, Mugo N, Irungu E, Celum C, Baeten J.
heterosexual African men and women: the Partners PrEP Study.
Successful increase in contraceptive use among Kenyan HIV-1-
International AIDS Society 2011. Rome, Italy; July 17–20, 2011. Oral
serodiscordant couples enrolled in an HIV-1 prevention trial.
AIDS 2009; 23 (suppl 1): S89–S95.
www.thelancet.com/infection Published online October 4, 2011 DOI:10.1016/S1473-3099(11)70247-X
http://online.wsj.com/article/SB10001424052748704538404574542010. Dow Jones Reprints: This copy is for your personal, non-commercial use only. To order presentation-ready copies for distribution to your colleagues, clients orcustomers, use the Order Reprints tool at the bottom of any article or visit www.djreprints.comhttp://online.wsj.com/article/SB10001424052748704538404574542010. The Foo
THE GLOBAL PHARMACEUTICAL INDUSTRY Sarah Holland* and Bernardo Bátiz-Lazo** The case looks at the development of the ethical pharmaceutical industry. The various forces affecting the discovery, development, production, distribution and marketing of prescription drugs are discussed in terms of their origins and recent developments. Readers are then invited to consider trends for the futur