Share this article on:

Oral and injectable contraception use and risk of HIV acquisition among women in sub-Saharan Africa

McCoy, Sandra I.a; Zheng, Wenjinga; Montgomery, Elizabeth T.b; Blanchard, Kellyc; van der Straten, Arianeb,d; de Bruyn, Guye; Padian, Nancy S.a,f

doi: 10.1097/QAD.0b013e32835da401
Epidemiology and Social

Objective: To evaluate the effect of oral and injectable hormonal contraception on the risk of HIV acquisition among women in South Africa and Zimbabwe.

Design: Secondary data analysis of 4913 sexually active women aged 18–49 years followed for up to 24 months in the Methods for Improving Reproductive Health in Africa (MIRA) phase III effectiveness trial of the diaphragm and lubricant gel for HIV prevention.

Methods: Participants were interviewed quarterly about contraception and sexual behavior and were tested for pregnancy, HIV, and other sexually transmitted infections. We used a Cox proportional hazards marginal structural model, weighted by the inverse probability of hormonal contraception use, to compare the risk of HIV acquisition among nonpregnant women reporting use of combined oral contraceptive pills (COC), progestin-only pills (POP), and/or injectable hormonal contraception to women not using these methods.

Results: During the study, 283 participants seroconverted. Use of oral contraceptives (POP or COC) was not associated with HIV risk [adjusted hazard ratio (HRa) = 0.86, 95% confidence interval (CI) 0.32, 1.78]. Injectable hormonal contraception was associated with a small nonsignificant risk of HIV infection (HRa = 1.34, 95% CI 0.75, 2.37). The effect of injectable hormonal contraception was similar in the unweighted site-adjusted only (HRa = 1.32, 95% CI 1.00, 1.74) and baseline factor adjusted models (HRa = 1.27, 95% CI 0.94, 1.72).

Conclusions: In this study, oral contraceptives were not associated with HIV acquisition. There is substantial uncertainty in the effect of injectable hormonal contraception on HIV risk. These findings underscore the importance of dual protection with condoms and the need for diverse contraceptive options for women at risk of HIV infection.

Supplemental Digital Content is available in the text

aUniversity of California, Berkeley, California

bWomen's Global Health Imperative, RTI International, San Francisco, California

cIbis Reproductive Health, Cambridge, Massachusetts

dUniversity of California San Francisco, California, USA

eUniversity of the Witwatersrand, Johannesburg, South Africa

fUS Department of State, Washington, DC, USA.

Correspondence to Sandra I. McCoy, MPH, PhD, Assistant Adjunct Professor, Division of Epidemiology, School of Public Health, 1918 University Avenue, Suite 3B, University of California, Berkeley, CA 94704, USA. Tel: +1 510 642 0534; fax: +1 510 642 5018; e-mail:

Received 14 September, 2012

Revised 29 November, 2012

Accepted 6 December, 2012

Supplemental digital content is available for this article. Direct URL citations appear in the printed text and are provided in the HTML and PDF versions of this article on the journal's Website (

Back to Top | Article Outline


More than 14 million women in sub-Saharan Africa (SSA) use hormonal contraception, of which nearly 60% rely on injectable hormonal contraception as their primary method [1]. A growing body of research has examined the effect of hormonal contraception use on HIV acquisition, an issue of great public health significance given women and girls’ vulnerability to HIV infection in SSA. More women than men are living with HIV in the region, and young women become susceptible to HIV and unintended pregnancy at an early age [2–4]. However, the effect of hormonal contraception on HIV risk remains unresolved despite nearly two decades of scientific inquiry [5,6].

Although laboratory and challenge studies in rhesus macaques suggest the biologic plausibility of an association between hormonal contraception and HIV infection [7–9], only a subset of observational studies have found an increased HIV risk among women using oral and injectable hormonal contraception [10–15]. Other studies have found no association [5,16,17]. The health, social, and economic benefits of effective contraception, including the reduction of unintended pregnancies and maternal morbidity and mortality, have warranted cautious interpretation of these inconsistent findings [18,19]. According to the WHO, women at high risk of HIV can continue to use all existing hormonal contraceptive methods without restriction; however, women using progestogen-only injectables are strongly advised to use condoms [20,21]. In response to a call for more research in this area [19], we evaluated the effect of oral and injectable hormonal contraception on the risk of HIV acquisition in a cohort of women in South Africa and Zimbabwe.

Back to Top | Article Outline


Study population

We conducted a secondary data analysis of nonpregnant women in the Methods for Improving Reproductive Health in Africa (MIRA) study, a phase III effectiveness trial of the diaphragm and lubricant gel for HIV prevention [22]. Eligible participants were sexually active women (reporting an average of at least four sexual acts per month), aged 18–49 years, living in Durban and Johannesburg, South Africa, and Harare, Zimbabwe.

Back to Top | Article Outline

Study design

The study has been previously described [22]. In brief, participants were randomized to receive a latex diaphragm, inactive lubricant gel, and condoms (intervention) or condoms alone (control). Participants received a comprehensive HIV prevention package including counseling and testing for HIV and sexually transmitted infections (STIs), treatment of curable, laboratory-diagnosed STIs for women and partners, risk-reduction counseling, and partner(s) HIV testing. Participants attended quarterly visits for up to 24 months and were interviewed about contraception and sexual behavior and were tested for pregnancy, HIV, and other STIs. The study found no added protective benefit of the diaphragm and lubricant gel against HIV infection [22].

We included all eligible women with at least one postenrollment HIV test who were not discontinued for eligibility reasons or HIV-positive at enrollment. A small number of women reported using hormonal implants (n = 53); we therefore excluded baseline implant users from the analysis (n = 35) or censored participants at the first report of implant use (n = 18). Data were organized into quarterly intervals corresponding to study visits. For this study, a participant was censored at the current visit (visit t) if she was permanently lost to follow-up (LTFU) at t, reported use of a hormonal implant at the previous visit (visit t − 1), had a positive pregnancy test at visit t − 1, or at study end.

Back to Top | Article Outline

Outcome assessment

HIV status was measured at baseline and each quarterly study visit with two simultaneous rapid tests (Determine HIV-1/2; Abbott Laboratories, Tokyo, Japan) and Oraquick (Orasure Technologies, Bethlehem, Pennsylvania, USA) on blood samples from finger prick or venipuncture. Discordant and concordant positive results were confirmed with ELISA (Vironostika, Biomerieux, Durham, North Carolina, USA; BioRad, Redmond, Washington, USA; or AxSYM HIV Ag/Ab Combo, Abbott Laboratories, Abbott Park, Illinois, USA). Retrospective HIV PCR testing was conducted on baseline samples for women who tested positive at their first quarterly visit to exclude women with prevalent infections at enrollment. The time of seroconversion was assigned to the visit interval with the first positive test if there were no missed visits between the last HIV-negative visit and the first ELISA positive visit. In cases where there were missed visits prior to seroconversion, the time of seroconversion was assigned to the visit interval representing the midpoint between the last HIV-negative visit and the first ELISA positive visit [22].

Back to Top | Article Outline

Exposure assessment

At each study visit, women were asked about contraceptive use, including oral contraceptives [combined oral contraceptive pills (COC) or progestin-only pills (POP)] and injectable hormonal contraception. Contraceptive use patterns have been previously described [23]. Midway through the trial, hormonal contraception was offered for free to participants.

We created three binary indicator variables representing reported use of COC, POP, and injectable hormonal contraception at each visit; women could therefore switch methods throughout the study (51.6% of women used the same method throughout and 86.8% of women using no method at baseline adopted a method during the study [23]) or report use of multiple products in a given visit interval (although rare). When all three indicators were at the referent level (0), it indicated nonuse of hormonal contraception; women may have been using nonhormonal contraceptive methods (e.g. condoms, traditional methods, withdrawal, nonhormonal intrauterine devices) or no contraception. This reference group was selected to facilitate comparison to previous studies. For a subset of injectable users, we differentiated between reported use of depot medroxyprogesterone acetate (DMPA) and norethisterone enantate (Net-En) using self-reported concurrent medication data. Contraceptive data for a given participant was only carried forward across missed study visits when the date of HIV seroconversion was assigned to a midpoint interval and the hormonal contraception type reported at adjacent intervals was the same (n = 20). A binary indicator was used to denote missing hormonal contraception data.

Back to Top | Article Outline


At each visit, participants provided a urine specimen for PCR testing for Neisseria gonorrhoeae, Chlamydia trachomatis and Trichomonas vaginalis (Roche Pharmaceuticals, Branchburg, New Jersey, USA) and pregnancy testing. Blood samples were collected at enrollment for syphilis testing [rapid plasma reagin (RPR) with confirmation of positives by Treponema pallidum hemagglutinin (TPHA); Randox Laboratories, Crumlin, UK] and herpes simplex virus 2 testing (HSV2; ELISA, FOCUS Diagnostics, Cypress, California, USA). Incident HSV-2 seroconversions detected at endline were assigned to a visit interval by back testing stored blood samples. Syphilis testing was additionally conducted at endline and when clinically indicated.

The following time-fixed covariates were measured at baseline: site, randomization (study) arm, marital status (married versus steady partner), cohabitation with spouse or steady partner, lifetime sexual partners, education, housing status, and lifetime live births. Consistent with the parent study [22], we created a composite binary indicator of baseline behavioral risk indicating one or more of the following behaviors: transactional sex in the last 3 months, two or more sexual partners in the last 3 months, ever had anal sex, or ever used a needle to inject drugs. We also created a composite indicator variable for baseline partner risk indicating one or more of the following characteristics: ever had a partner test positive for HIV infection, knowledge or suspicion that the regular partner has had other partners in the past 3 months, or regular partner away from home at least 1 month in the past year.

The following time-varying covariates were measured at each quarterly visit: age, diaphragm use over the past 3 months (never, sometimes, always, or no sex), vaginal washing or wiping in the past 3 months, circumcision status of the regular male sexual partner, new sexual partners, coital frequency (≤3 times per week, >3 times per week), condom use at last sex, condom use frequency over the past 3 months (never, sometimes, always, or no sex), and incident STIs.

Back to Top | Article Outline

Statistical analysis

Our primary analysis was to determine the effect of oral and injectable hormonal contraception use on risk of HIV acquisition using an inverse-probability-of-treatment weighted (IPTW) marginal structural model (MSM). Secondary data analyses included a conventional Cox proportional hazards model, examination of effect measure modification, and sensitivity analyses.

Back to Top | Article Outline

Primary analysis

We first examined the site-adjusted association of hormonal contraception use (reported at t − 1 visit) with HIV acquisition (measured at time t) using an unweighted Cox proportional hazards model with separate baseline hazards for each of the three study sites, and hormonal contraception type as the only independent variables. We evaluated injectables separately from a combined category of oral contraceptives, as well as oral contraceptives disaggregated into COC and POP. We also examined a baseline factor adjusted association (‘baseline-adjusted model’) by adding covariates measured at baseline to the model along with the time-varying hormonal contraception use variables and separate baseline hazards for site. These two models were fitted on the subsample of patient visits with hormonal contraception information that occurred prior to an HIV-positive test result or censoring.

To adjust for time-dependent confounding, we then constructed an IPTW Cox proportional hazards MSM, which summarized the hazard of HIV acquisition at time t as a function of hormonal contraception use at time t − 1 [24]. Each person visit was treated as an observation; its weight was determined by the product of three stabilized weights: treatment weights corresponding to the probability of hormonal contraception use at t − 1; hormonal contraception availability weights corresponding to the probability of not missing hormonal contraception data at t − 1 (unknown hormonal contraception use mostly results from missed visits) [24]; and censoring weights corresponding to the probability of not having been censored by time t. The time ordering of the variables and the predictors in each weight model were prespecified from substantive knowledge and/or existing literature. The weighted Cox proportional hazards was then fitted on the subsample of patient visits not missing hormonal contraception information at t − 1 that occurred prior to an HIV-positive test result or censoring. In this model, we evaluated injectables separately from a combined category of oral contraceptives (POP users were a homogeneous group and the lack of experimental support precluded disaggregation into COCs and POPs [25]). We report hazard ratios (HRs) and 95% confidence intervals (CIs), which were obtained using the nonparametric bootstrap percentile method (see Supplemental Digital Content,

Back to Top | Article Outline

Secondary analyses

Due to previous reports [11,16] of effect modification by age and HSV-2 status, we a priori specified a secondary analysis to evaluate effect modification by age (≤24, >24 years) and baseline HSV-2 status. In the Cox MSM, we included both the main term (for age or HSV-2) and an interaction term with hormonal contraception, and changed the stabilized weights accordingly. We then performed a Wald test (using the bootstrapped covariance matrix) for the null hypothesis that the interaction terms jointly equal zero.

For comparison with previous studies, we examined the association of hormonal contraception use with HIV acquisition using a conventional Cox proportional hazards model adjusted for time-varying confounders. Covariates were individually examined for effect modification, confounding, and their effect on HIV acquisition. The adjusted model includes separate baseline hazards for each site and all relevant confounding and modifying covariates. A single measure of condom use was included (frequency in the past 3 months). We present adjusted hazard ratios and 95% CIs with robust standard errors to account for repeated measurements.

Back to Top | Article Outline

Sensitivity analyses

To determine the potential effect of an extended exposure period after COC, POP, or injectable discontinuation, we created three additional binary exposure variables that included a 90-day washout period into the next visit interval. We re-ran the primary effect analysis with these new exposure variables. Recognizing that hormone exposure after discontinuation (or since the last injection) may be longer with injectable hormonal contraception than oral contraceptives, we repeated this analysis with the extended exposure period for injectable hormonal contraception alone. Finally, we added splines to the model to determine sensitivity of the primary effect estimates to weight model specifications.

Back to Top | Article Outline


Participant characteristics at enrollment

Between September 2003 and September 2005, 4948 eligible women enrolled in the study and were included in the parent study analysis [22]. Of these, 35 women using hormonal implants at baseline were excluded from the current analysis, leaving 4913 participants contributing 6972 woman-years of follow-up. The median follow-up time was 17.9 months; 1340 (27%) of participants had at least one missed visit during follow-up (i.e. before study end, seroconversion, or they were LTFU). Approximately half of the participants were from Zimbabwe, 51% reported one lifetime sexual partner, 58% were HSV-2 positive at baseline, and 11% had a curable STI at screening or enrollment (Table 1).

Table 1-a

Table 1-a

Table 1-b

Table 1-b

At baseline, 21% of participants reported COC use, 14% reported POP use, 26% reported injectable use, and 38% were using nonhormonal methods of contraception or no contraception. Oral contraceptive users were largely from Zimbabwe (88% of all oral contraceptive users), whereas injectables were most commonly reported among women in Durban (47%). Oral contraceptive users were more likely to report being married, living with their spouse or steady partner, having only one lifetime sexual partner, and had fewer baseline STIs than injectable or non-hormonal contraception users. There was little difference between reported condom use at last sex or condom use in the past 3 months among COC, POP, or injectable users; however, women reporting non-hormonal contraception methods or no contraception were more likely than hormonal contraception users to report condom use at last sex (75 versus 68%; P < 0.01) and ‘always’ in the past 3 months (40 versus 26%; P < 0.01).

Back to Top | Article Outline

HIV incidence by contraceptive group

A total of 283 seroconversions occurred, representing 92% of the 309 seroconversions reported in the parent study (two seroconversions among women using hormonal implants and 24 seroconversions occurring after pregnancy were not included in this analysis). HIV incidence was 4.7, 6.3, 2.6, and 2.7 per 100 woman-years in the non-hormonal contraception, injectable, COC, and POP groups, respectively (Table 2). For 6091 (86%) of 7119 visit intervals with injectable use, we were able to distinguish between DMPA or Net-En use. HIV incidence was 5.6 per 100 woman-years in the DMPA group and 6.3 per 100 woman-years in the Net-En group.

Table 2

Table 2

Back to Top | Article Outline

Effect of hormonal contraception use

In the unweighted site-adjusted Cox model, oral contraceptive use overall (Table 2; hazard ratio 0.82, 95% CI 0.58, 1.15) or disaggregated (COC: hazard ratio 0.78, 95% CI 0.53, 1.12; POP: hazard ratio 0.91, 95% CI 0.49, 1.50) was not associated with an increased risk of HIV acquisition. In contrast, use of injectable contraception was associated with an increased risk of HIV infection (hazard ratio 1.32, 95% CI 1.00, 1.74). When separated, neither DMPA (hazard ratio 1.18, 95% CI 0.84, 1.62) nor Net-En (hazard ratio 1.40, 95% CI 0.72, 2.35) was significantly associated with HIV infection in the subset of women with this information.

After adjustment for baseline characteristics only, the effect of injectable contraception on HIV acquisition was attenuated [adjusted hazard ratio (HRa) = 1.27, 95% CI 0.94, 1.72). After adjustment for both baseline and time-dependent confounders using an IPTW MSM, there was no effect of oral contraceptives (overall) on HIV risk (HRa = 0.86, 95% CI 0.32, 1.78) and a small nonsignificant increased risk among injectable users (HRa = 1.34, 95% CI 0.75, 2.37). There was no evidence of a statistically significant statistical interaction between baseline age and oral and injectable hormonal contraception use (P interaction = 0.60) or between baseline HSV-2 status and oral and injectable hormonal contraception use (P interaction = 0.21).

In the conventional Cox model adjusted for baseline time-varying covariates, neither COC nor POP use was associated with an increased risk of HIV acquisition (COC: HRa = 0.86, 95% CI 0.58, 1.28, POP: HRa = 0.98, 95% CI 0.56, 1.73). In contrast, use of any injectable contraception was associated with an increased risk of HIV infection (HRa = 1.37, 95% CI 1.01, 1.85).

Back to Top | Article Outline

Sensitivity analyses

The site-adjusted and baseline-adjusted effect estimates were relatively insensitive to the extended 90-day washout period. When all three exposure periods were extended, use of oral contraceptives was not associated HIV acquisition, and the risk among injectable users was attenuated to HR = 1.26 and adjusted hazard ratio 1.18 for the site-adjusted and baseline-adjusted association, respectively. Using a washout period for injectable users only did not significantly change these estimates. However, the effect estimates from the IPTW MSM model were sensitive to the washout periods; when the exposure periods for both oral contraceptives and injectables were extended, the effect of injectables was attenuated to HRa = 0.73 and when only the exposure period for injectables was extended, the hazard ratio for injectables was HRa = 0.82. Finally, we reran the IPTW analysis using restricted cubic splines in the weight models and the results were qualitatively the same as presented above.

Back to Top | Article Outline


In sub-Saharan Africa, women and girls are highly vulnerable to HIV infection, comprising nearly 60% of people living with HIV infection in the region [2]. In addition to women's increased biological susceptibility to infection, poverty and other structural factors such as entrenched sex inequities reduce their ability to control HIV-related risk behaviors [26,27]. Any further increase to their vulnerability is concerning. This secondary data analysis suggests that there may be a small increased risk of HIV acquisition among nonpregnant women reporting use of injectable hormonal contraception versus non-hormonal contraception methods. However, the large CIs indicate substantial uncertainty in this conclusion. Some studies have reported an increased HIV risk associated with injectables among both higher-risk [13,28] and lower-risk women [11,12,29] and among women in serodiscordant partnerships [10]. Other studies have not found an increased HIV risk with injectables (see reviews [5,30]) [16,17,31]. Given that 41 million women worldwide use injectable hormonal contraception, including 8.7 million women in SSA, this is a critical issue [1]. In contrast, we found no increased risk among oral contraceptive users in any of our analyses, methods used by more than 100 million women globally [1].

Hormonal contraception prevents unintended pregnancies, reduces maternal and infant morbidity and mortality, and has other significant social and economic benefits [32]. In addition, pregnancy itself may independently increase HIV risk [33]. Appropriately balancing the trade-offs between increased HIV risk and the consequences of reducing the availability and/or demand of reliable, effective contraception is therefore critical. Modeling studies suggest that the increase in HIV risk from injectables would have to be higher than most studies have reported to warrant changes to current recommendations, especially given the prevention benefits of antiretroviral therapy [34,35]. In addition, the WHO technical consultation did not recommend changes to current policy recommendations [21].

This study is subject to important limitations. This was a secondary data analysis of an HIV prevention trial that was not designed to determine the effect of hormonal contraception on HIV acquisition risk. We were unable to examine the effect of other types of hormonal contraception such as implants. Further, contraceptive use and risk behavior were self-reported, and the long visit interval period (3 months) increased the chance of measurement error. Our study also has significant strengths, including a large sample size, large number of incident infections, and high rates of oral contraceptive use, which allowed us to determine separate measures of effect for COC and POP users in some of our models. We were also able to differentiate between DMPA and Net-En for a subset of injectable users, although this disaggregation substantially reduced power. Our results remained relatively robust in several sensitivity analyses.

Laboratory studies in nonhuman primates have suggested that progesterone-based hormonal contraception could increase susceptibility to HIV infection by thinning the cervico-vaginal epithelium, decreasing colonization with H2O2-positive Lactobacillus that may kill free virus, increasing the frequency of target cells and CCR5 expression in the genital tract, and immunosuppressive effects (see review [7]). The evidence for biologic plausability is weaker in human studies, and the physiologic effects exerted by different progestins may vary [7]. However, the nature of injectables themselves – a hormonal contraception method that can be used clandestinely – underscores the potentially important role of confounding. For example, a fundamental challenge is disaggregating the direct effect of hormonal contraception on HIV acquisition from the indirect effect mediated by behavior. Differential condom use across contraceptive method groups [36,37] complicates statistical analyses insofar as condom use is a confounder of future hormonal contraception use (and should be adjusted for) but is also affected by earlier hormonal contraception use (and hence should not be included in models). Conventional statistical models are biased in these situations where time-dependent confounding is present [38].

We used an IPTW Cox proportional hazards MSM to estimate the causal effect of hormonal contraception use on HIV acquisition, a model that appropriately handles time-dependent confounding. In this model, we found a small nonsignificant increased HIV risk among injectable users. However, causal interpretation of the parameters relies on the following assumptions: no unmeasured confounding of the relationships between hormonal contraception use and HIV infection, censoring and HIV infection, missing data on hormonal contraception (missed visits or failure to report) and HIV infection [25,38]; within every covariate strata, there is a nonzero probability of observing any given type of hormonal contraception (positivity) [25]; our time-ordering assumptions are correct; and correct specification of the weight models. For the last assumption, future work might use more sophisticated doubly robust estimators such as targeted maximum likelihood estimation.

Some have argued that a randomized trial is warranted to definitively resolve the association between hormonal contraception use and HIV risk [6]. However, randomizing women to various hormonal contraception methods may not be feasible or acceptable and the results of such a trial would not be available for several years. The way forward will depend on whether a consistent effect of injectables on HIV risk emerges, including data from meta-analyses, modeling and simulation studies, and carefully conducted observational studies with good measurement of confounders that use more sophisticated analysis techniques that can assess the direct effect of hormonal contraception on HIV risk. In addition, the continued emphasis on dual protection with condoms for women at risk for HIV infection is especially important for women living in areas with generalized HIV epidemics [21]. Finally, affordable access to a wider range of contraception options, including long-acting methods that are more effective for preventing unintended pregnancy and do not increase the risk of HIV infection, is imperative.

Back to Top | Article Outline


We are grateful to the women who participated in the MIRA study. We are also indebted to Ms Helen Cheng for data management, Ms Karen Shiu for help with evaluating DMPA and Net-En exposure histories, and to Ms Lauren Ralph for critical review of the manuscript.

Back to Top | Article Outline

Conflicts of interest

Sources of support: S.M. is supported by Award Number K01MH094246 from the National Institute of Mental Health. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institute of Mental Health or the National Institutes of Health. The MIRA trial was funded by the Bill & Melinda Gates Foundation (number 21082).

Back to Top | Article Outline


1. United Nations Department of Economic and Social Affairs. World contraceptive use 2011. New York: United Nations; 2011.
2. UNAIDS. Report on the global AIDS epidemic. Geneva, Switzerland: UNAIDS; 2010.
3. Pettifor AE, Rees HV, Kleinschmidt I, Steffenson AE, MacPhail C, Hlongwa-Madikizela L, et al. Young people's sexual health in South Africa: HIV prevalence and sexual behaviors from a nationally representative household survey. AIDS 2005; 19:1525–1534.
4. Boonstra HD. Learning from adolescents to prevent HIV and unintended pregnancy. New York:Guttmacher Institute; 2007.
5. Morrison CS, Turner AN, Jones LB. Highly effective contraception and acquisition of HIV and other sexually transmitted infections. Best Pract Res Clin Obstet Gynaecol 2009; 23:263–284.
6. Morrison CS, Nanda K. Hormonal contraception and HIV: an unanswered question. Lancet Infect Dis 2012; 12:2–3.
7. Hel Z, Stringer E, Mestecky J. Sex steroid hormones, hormonal contraception, and the immunobiology of human immunodeficiency virus-1 infection. Endocrine Rev 2010; 31:79–97.
8. Trunova N, Tsai L, Tung S, Schneider E, Harouse J, Gettie A, et al. Progestin-based contraceptive suppresses cellular immune responses in SHIV-infected rhesus macaques. Virology 2006; 352:169–177.
9. Marx PA, Spira AI, Gettie A, Dailey PJ, Veazey RS, Lackner AA, et al. Progesterone implants enhance SIV vaginal transmission and early virus load. Nature Med 1996; 2:1084–1089.
10. Heffron R, Donnell D, Rees H, Celum C, Mugo N, Were E, et al. Use of hormonal contraceptives and risk of HIV-1 transmission: a prospective cohort study. Lancet Infect Dis 2012; 12:19–26.
11. Morrison CS, Chen PL, Kwok C, Richardson BA, Chipato T, Mugerwa R, et al. Hormonal contraception and HIV acquisition: reanalysis using marginal structural modeling. AIDS 2010; 24:1778–1781.
12. Wand H, Ramjee G. The effects of injectable hormonal contraceptives on HIV seroconversion and on sexually transmitted infections. AIDS 2012; 26:375–380.
13. Baeten JM, Benki S, Chohan V, Lavreys L, McClelland RS, Mandaliya K, et al. Hormonal contraceptive use, herpes simplex virus infection, and risk of HIV-1 acquisition among Kenyan women. AIDS 2007; 21:1771–1777.
14. Watson-Jones D, Baisley K, Weiss HA, Tanton C, Changalucha J, Everett D, et al. Risk factors for HIV incidence in women participating in an HSV suppressive treatment trial in Tanzania. AIDS 2009; 23:415–422.
15. Plummer FA, Simonsen JN, Cameron DW, Ndinya-Achola JO, Kreiss JK, Gakinya MN, et al. Cofactors in male-female sexual transmission of human immunodeficiency virus type 1. J Infect Dis 1991; 163:233–239.
16. Morrison CS, Skoler-Karpoff S, Kwok C, Chen PL, van de Wijgert J, Gehret-Plagianos M, et al. Hormonal contraception and the risk of HIV acquisition among women in South Africa. AIDS 2012; 26:497–504.
17. Myer L, Denny L, Wright TC, Kuhn L. Prospective study of hormonal contraception and women's risk of HIV infection in South Africa. Int J Epidemiol 2007; 36:166–174.
18. Radloff S. USAID communication to the field (October 7, 2011): USAID updated response to publication of findings on hormonal contraception and HIV acquisition in uninfected women and HIV transmission from infected women to men. United States Agency for International Development (USAID); 2011.
19. World Health Organization. Statement on the Heffron et al. study on the safety of using hormonal contraceptives for women at risk of HIV infection. Geneva, Switzerland: Department of Reproductive Health and Research, World Health Organization; 2011.
20. World Health Organization. Medical eligibility criteria for contraceptive use. 4th ed. Geneva, Switzerland: Department of Reproductive Health and Research, World Health Organization; 2010.
21. Department of Reproductive Health and Research. Hormonal contraception and HIV. Technical statement. Geneva, Switzerland: World Health Organization; 2012.
22. Padian NS, van der Straten A, Ramjee G, Chipato T, de Bruyn G, Blanchard K, et al. Diaphragm and lubricant gel for prevention of HIV acquisition in southern African women: a randomised controlled trial. Lancet 2007; 370:251–261.
23. Blanchard K, Bostrom A, Montgomery E, van der Straten A, Lince N, de Bruyn G, et al. Contraception use and effectiveness among women in a trial of the diaphragm for HIV prevention. Contraception 2011; 83:556–563.
24. Seaman SR, White IR. Review of inverse probability weighting for dealing with missing data. Stat Methods Med Res 2011 [Epub ahead of print].
25. Chen PL, Cole SR, Morrison CS. Hormonal contraception and HIV risk: evaluating marginal-structural-model assumptions. Epidemiology 2011; 22:877–878.
26. Comparison of female to male and male to female transmission of HIV in 563 stable couples. European Study Group on Heterosexual Transmission of HIV. BMJ 1992; 304:809–813.
27. Krishnan S, Dunbar MS, Minnis AM, Medlin CA, Gerdts CE, Padian NS. Poverty, gender inequities, and women's risk of human immunodeficiency virus/AIDS. Ann N Y Acad Sci 2008; 1136:101–110.
28. Ungchusak K, Rehle T, Thammapornpilap P, Spiegelman D, Brinkmann U, Siraprapasiri T. Determinants of HIV infection among female commercial sex workers in northeastern Thailand: results from a longitudinal study. J Acquir Immune Defic Syndr Hum Retrovirol 1996; 12:500–507.
29. Kumwenda JJ, Makanani B, Taulo F, Nkhoma C, Kafulafula G, Li Q, et al. Natural history and risk factors associated with early and established HIV type 1 infection among reproductive-age women in Malawi. Clin Infect Dis 2008; 46:1913–1920.
30. Polis C, Curtis K. Hormonal contraception and HIV acquisition in women: a systematic review of the epidemiological evidence. XIX International AIDS Conference. Washington, DC, 2012.
31. Lutalo T, Musoke R, Polis C, Serwadda D, Makumbi F, Nalugoda F, et al. Effects of Hormonal Contraceptive Use on HIV Acquisition in Women and Transmission to Men among HIV-discordant Couples, Rakai, Uganda. 19th Conference on Retroviruses and Opportunistic Infections. Seattle, 2012.
32. Singh S, Darroch JE, Ashford LS, Vlassoff M. Adding it up: the costs and benefits of investing in family planning and maternal and newborn health. New York: Guttmacher Institute and United Nations Population Fund; 2009.
33. Mugo NR, Heffron R, Donnell D, Wald A, Were EO, Rees H, et al. Increased risk of HIV-1 transmission in pregnancy: a prospective study among African HIV-1-serodiscordant couples. AIDS 2012; 25:1887–1895.
34. Cohen MS, Chen YQ, McCauley M, Gamble T, Hosseinipour MC, Kumarasamy N, et al. Prevention of HIV-1 infection with early antiretroviral therapy. N Engl J Med 2011; 365:493–505.
35. Butler AR, Smith JA, Polis CB, Gregson S, Stanton D, Hallett TB. Modelling the global competing risks of a potential interaction between injectable hormonal contraception and HIV risk. AIDS 2013; 27:105–113.
36. Pazol K, Kramer MR, Hogue CJ. Condoms for dual protection: patterns of use with highly effective contraceptive methods. Public Health Rep 2010; 125:208–217.
37. Sangi-Haghpeykar H, Posner SF, Poindexter AN 3rd. Consistency of condom use among low-income hormonal contraceptive users. Perspect Sex Reprod Health 2005; 37:184–191.
38. Robins JM, Hernan MA, Brumback B. Marginal structural models and causal inference in epidemiology. Epidemiology 2000; 11:550–560.

depot medroxyprogesterone acetate; HIV infection; hormonal contraception; oral contraceptives; prevention; women

© 2013 Lippincott Williams & Wilkins, Inc.