| |
Hormonal contraception and the risk of HIV acquisition
|
| |
| |
AIDS: Volume 21(1) 2 January 2007 p 85-95
".....In summary, this large, multisite study found no overall increased risk of HIV acquisition associated with hormonal contraceptive use. This provides reassurance for women in a setting of moderate and high HIV prevalence who need effective contraception; that any increased overall risk associated with hormonal contraception is, at most, modest. Among women who are HSV-2 negative, DMPA and COC users may be at increased risk of HIV acquisition. Additional research should confirm and interpret this finding. Regardless, our findings should reinforce efforts to counsel all women at risk of HIV infection to use condoms consistently and correctly to prevent HIV acquisition...... Our finding of no overall association between hormonal contraceptive use and HIV acquisition suggests that the inconsistent results of previous studies would not support a strong association between hormonal contraception and HIV. Previous studies had important methodological shortcomings: few hormonal contraceptive users and low study power, poor measurement of hormonal contraceptive use and potential confounding variables, lack of adjusted analyses, and poor follow-up. Nevertheless, one methodologically strong study conducted among Kenyan sex workers found an association between both DMPA and COC use and HIV acquisition [15,17]. As our study participants were seeking family planning services, our study cannot rule out the possibility that hormonal contraception might increase HIV susceptibility among some highly exposed subpopulations..... our study results are not inconsistent with a modest increase in HIV risk associated with hormonal contraceptive use, particularly for DMPA. Nevertheless, we found no evidence of higher HIV risk associated with hormonal contraception among high-risk subgroups in our study, including among participants with STI...."
Authors: Morrison, Charles Sa; Richardson, Barbra Ab; Mmiro, Francisd; Chipato, Tsungaie; Celentano, David Df; Luoto, Joanneg; Mugerwa, Royd; Padian, Nancyh; Rugpao, Sungwali; Brown, Joelle Mh; Cornelisse, Peterc; Salata, Robert Aj; for the Hormonal Contraception and the Risk of HIV Acquisition (HC-HIV) Study Group
From the aClinical Research Department, Family Health International, Research Triangle Park, North Carolina, USA
bDepartment of Biostatistics, University of Washington and the Statistical Center for HIV/AIDS Research & Prevention (SCHARP), Fred Hutchinson Cancer Research Center and the
cStatistical Analysis Unit, Cystic Fibrosis Therapeutics Development Network, Children's Hospital and Regional Medical Center, Seattle, Washington, USA
dFaculty of Medicine, Makerere University, Kampala, Uganda
eDepartment of Obstetrics and Gynecology, University of Zimbabwe, Harare, Zimbabwe
fDepartment of Epidemiology, Bloomberg School of Public Health, Johns Hopkins University, Baltimore, Maryland, USA
gNational Institute of Child Health and Human Development, National Institutes of Health, Department of Health and Human Services, Bethesda, Maryland, USA
hDepartment of Obstetrics, Gynecology and Reproductive Sciences, University of California at San Francisco, San Francisco, California, USA
iResearch Institute for Health Sciences (RIHES), Chiang Mai University, Chiang Mai, Thailand
jDepartment of Medicine, Case Western Reserve University, Cleveland, Ohio, USA.
Abstract
Background: Combined oral contraceptives (COC) and depot-medroxyprogesterone acetate (DMPA) are among the most widely used family planning methods; their effect on HIV acquisition is not known.
Objective: To evaluate the effect of COC and DMPA on HIV acquisition and any modifying effects of other sexually transmitted infections.
Methods: This multicenter prospective cohort study enroled 6109 HIV-uninfected women, aged 18-35 years, from family planning clinics in Uganda, Zimbabwe and Thailand. Participants received HIV testing quarterly for 15-24 months. The risk of HIV acquisition with different contraceptive methods was assessed (excluding Thailand, where there were few HIV cases).
Results:
HIV infection occurred in 213 African participants (2.8/100 woman-years).
Use of neither COC [hazard ratio (HR), 0.99; 95% confidence interval (CI), 0.69-1.42] nor DMPA (HR, 1.25; 95% CI, 0.89-1.78) was associated with risk of HIV acquisition overall, including among participants with cervical or vaginal infections.
While absolute risk of HIV acquisition was higher among participants who were seropositive for herpes simplex virus 2 (HSV-2) than in those seronegative at enrolment, among the HSV-2-seronegative participants, both COC (HR, 2.85; 95% CI, 1.39-5.82) and DMPA (HR, 3.97; 95% CI, 1.98-8.00) users had an increased risk of HIV acquisition compared with the non-hormonal group.
Conclusions: No association was found between hormonal contraceptive use and HIV acquisition overall. This is reassuring for women needing effective contraception in settings of high HIV prevalence. However, hormonal contraceptive users who were HSV-2 seronegative had an increased risk of HIV acquisition. Additional research is needed to confirm and explain this finding.
Introduction TOP
Over 100 million women worldwide use hormonal contraception, primarily combined oral contraceptives (COC) containing estrogen and progestin, and the progestin-only injectable contraceptive depot-medroxyprogesterone acetate (DMPA) [1]. Use of hormonal contraception, and especially DMPA, is increasing rapidly in many resource-poor countries where the prevalence of HIV and other sexually transmitted infections (STI) is high [2-4].
Globally, about 18 million women are infected with HIV; an estimated 2 million women became infected during 2005, most at risk of pregnancy and infected heterosexually [5]. Understanding whether hormonal contraceptive use alters the risk of HIV acquisition among women is a critical public health issue.
Twelve prospective studies have examined this issue. However, previous studies were not specifically designed to assess the hormonal contraception-HIV relationship and most had important methodological shortcomings. Eleven examined COC use and HIV acquisition [6-17]; two found a statistically significant increased risk [7,17], and nine found no association [6,8-14,16]. Six studies addressed DMPA use and HIV acquisition [11,13-18]; two found a statistically significant increased risk [11,17], and four found no association [13,14,16,18]. In addition, two reviews have been published with opposite conclusions concerning the risk of HIV acquisition among COC users [19,20]. Most studies reporting an increased HIV risk involved high-risk populations, such as sex workers [7,11,13,15,17] or women attending STI clinics [6].
This prospective study was specifically designed to represent the general populations in Uganda, Zimbabwe, and Thailand and examine the association between COC and DMPA use and HIV acquisition. Secondary objectives were to determine if the rate of HIV infection among hormonal contraceptive users was modified by other STI and to determine whether the type of hormonal contraception (COC versus DMPA) had a differential impact on HIV acquisition.
Methods
The research was approved by the institutional review boards of collaborating institutions in the United States, Uganda, Zimbabwe, and Thailand. All participants provided written informed consent.
Study population and procedures
Between November 1999 and January 2004, women were recruited and followed from those seeking reproductive and general healthcare services at three sites in Kampala (Uganda), four sites in Harare and Chitungwiza (Zimbabwe), and seven sites in Chiang Mai, Hat Yai, Khon Kaen, and Bangkok (Thailand). Participants were aged 18-35 years, uninfected with HIV, sexually active, not pregnant, had not injected drugs or had a blood transfusion in the prior 3 months, and had been using no hormonal method (non-HC), COC (low-dose pills containing 30 μg ethinylestradiol and 150 μg levonorgestrel) or DMPA (150 mg administered every 12 weeks) for at least 3 months and intended to continue use for the following year. Women were ineligible if they had a hysterectomy or had used an intrauterine device or had a spontaneous or induced abortion in the previous month. Each site attempted to enrol equal numbers of women into the three contraceptive groups.
A smaller group of 'high-risk referral' women were also recruited in Uganda (398 women) and Thailand (488 women) where there was a low initial HIV incidence. These women were recruited from STI or primary healthcare clinics (women with STI symptoms), sex worker networks, or military bases.
At screening, women received pretest HIV counseling and provided blood for testing for HIV, syphilis, and herpes simplex virus 2 (HSV-2). Women returned within 15 days for their test results and possible enrolment. HIV-infected women were counseled and referred for on-going medical and social support.
At enrolment, eligible women who had given consent received a standardized interview in the local language to collect demographic, sexual behavior, reproductive health, and contraceptive history data. Contraceptive, HIV risk reduction, condom use counselling, and free contraceptives and condoms were provided. Study clinicians conducted a standardized physical (including pelvic) examination and collected specimens for Chlamydia trachomatis, Neisseria gonorrhoeae, Trichomonas vaginalis, Candida albicans, and bacterial vaginosis testing. Participants were treated on-site for vaginal infections while those diagnosed with chlamydial infection, gonorrhea, or syphilis were recalled for treatment.
Follow-up visits were conducted every 12 weeks for 15-24 months. Follow-up procedures were similar to those at enrolment and included testing for HIV, HSV-2, and others STI (except semi-annual syphilis testing and annual or exit-only Pap smears). Women with abnormal Pap smears were referred for colposcopy.
Analysis population and variable definition
The analysis population consisted of participants with at least one follow-up visit with valid HIV results. Thailand was excluded from the primary analysis because of few (four) incident HIV infections. Data were also excluded from visit segments when participants used non-study contraceptive methods (Table 1).
The outcome was the number of days from baseline to the earlier of either the diagnosis of HIV infection or the last study contact. To estimate the effects of time-varying contraceptive exposure, a participant's time was divided into segments corresponding to the period between visits. This allowed changes in contraceptive use, sexual behaviours, and STI between study visits to be captured and taken into account in the analyses.
Contraceptive exposure was defined by COC or DMPA use since the last visit. When a woman switched from a hormonal to the non-HC group, contraceptive exposure was calculated as 180 days from last DMPA injection and 90 days from last pill use [21,22].
Laboratory methods
At screening, serum was tested for HIV using enzyme-linked immunosorbent assays (ELISA) [Recombigen HIV-1/HIV-2 (Cambridge Biotech, Galway, Ireland), Organon Vironostika (Organon Teknika, Durham, North Carolina, USA), Abbott Murex (Abbott Park, Illinois, USA), Sanofi (Sanofi Diagnostics Pasteur, Redmond, Washington, USA)] with positive results confirmed with HIV rapid tests [HIV SAV1 or SAV2 (Savyon Diagnostics, Ashdod, Israel), Capillus HIV-1/HIV-2 (Trinity Biotech USA, Jamestown, New York, USA) or Determine (Abbott)]. Polymerase chain reaction (PCR) tests for HIV DNA (Amplicor HIV-1 DNA test, version 1.5, Roche Diagnostics, Branchburg, New Jersey, USA) were conducted on dried blood spots [23] to adjudicate discrepant results.
At follow-up, HIV was measured with ELISA, with positive results confirmed with rapid testing then with Western blot (BioRad, Hercules, California, USA) or PCR; PCR results were the final arbiter of infection status. Incident infections were confirmed by a subsequent blood draw using an ELISA test. For confirmed incident HIV infections, HIV PCR was performed serially on previous visit specimens. The final date of HIV acquisition was defined as the date of the first positive PCR result.
PCR testing for gonorrhea and chlamydia (Amplicor CT/NG, Roche Diagnostics) was conducted on endocervical specimens. Commercial RPR tests [Shield (Shield Diagnostics, Dundee, Scotland, UK), Human (Human GmbH, Wiesbaden, Germany), Randox (Randox Laboratories, Antrim, Northern Ireland, UK)] were used for T. pallidum antibodies, with positive results confirmed by TPHA/TPPA (Serodia; Fujirebio, Tokyo, Japan). Serum was tested for HSV-2 antibodies utilizing an ELISA assay (Focus Diagnostics, Cypress, California, USA) [24]. Vaginal infections were diagnosed by microscopy: trichomonas by motile flagellates and yeast infection by yeast and pseudohyphae. Bacterial vaginosis was diagnosed according to Amsel criteria [25].
Statistical methods
The 213 incident HIV infections diagnosed provided 80% power to detect a 1.6-fold difference between the non-HC comparison group and COC or DMPA users.
Follow-up rates among the contraceptive groups were compared using Kaplan-Meier survival analysis. Comparisons of baseline characteristics among groups were assessed using ƒÔ2 tests. The association of time-varying covariates (e.g., condom use) with contraceptive exposure was tested using ƒÔ2 test statistics adjusted for clustering [26].
The primary endpoint was defined as the date of the first positive HIV result. It was assumed that, with relatively low incidence and frequent follow-up, analysis results would be similar using time of detection and estimated infection date (e.g., midpoint between last negative and first positive test) [27]. The relationship between COC or DMPA use and incident HIV infection was assessed using a Cox proportional hazards regression analysis of time to HIV infection. Variables considered a priori to be important (site, living with partner, condom use) were retained in the primary model. Participant demographic characteristics, baseline reproductive health history, sexual behavior, and primary partner characteristics were assessed as potential confounding factors. Variables were defined as confounders if the hazard ratio (HR) for COC or DMPA exposure changed by ≥ 10% when the variable was added to the primary model and these confounders were retained in final multivariate models. It was decided a priori to test four variables for effect modification of the hormonal contraception-HIV relationship: time-varying vaginal and cervical infections, HSV-2 infection status at enrolment, and study site. Because interaction terms for HSV-2 status and site were statistically significant, strata-specific results are reported.
Finally, to assess the robustness of the results, sensitivity analyses were performed on the final multivariate models: (a) changing the wash-out effect for COC and DMPA to 0 and 4 months after last use or injection [28], (b) censoring women at the time of first contraceptive switch or pregnancy, and (c) adjusting for pregnancy by including a time-varying pregnancy variable.
Results
The study screened 10 082 women in Uganda, Zimbabwe, and Thailand. Of these, 2425 women (24%) were infected with HIV (16.4% in Uganda, 38.1% in Zimbabwe, 2.1% in Thailand) and thus ineligible. An additional 1548 women were ineligible for other reasons (Table 1). Therefore, 6109 women participated in the study: 2235 in Uganda, 2296 in Zimbabwe, and 1578 in Thailand.
After excluding data from Thailand and participants without follow-up or those exclusively using non-study methods, 4439 African participants contributed 31 197 segments to the analysis (Table 1). At baseline, 34.7% participants used COC, 34.2% used DMPA, and 31.1% were in the non-HC group (84% used condoms, 13% practiced withdrawal, 10% used the 'rhythm' method, 3% were sterilized, and 5% used another non-HC method).
Participant characteristics at enrolment
Study participants were evenly divided between Uganda and Zimbabwe. Median age of study participants was 25 years; median education was 10 years, and most participants lived with a partner. Study participants had a median of two lifetime pregnancies and 28% breastfed at enrolment. Few participants reported multiple sex partners, commercial sex, or sex while using alcohol or drugs, and less than half reported recent condom use.
Participants from Zimbabwe were somewhat more likely to use hormonal contraception at baseline than participants from Uganda (Table 2). Compared with the non-HC participants, more participants using hormonal contraceptives were older (25-35 years), lived with a partner, and had two or more previous pregnancies. COC and DMPA users had used their respective methods for a mean of 10.2 and 8.5 months, respectively, prior to study enrolment.
Non-HC participants reported higher levels of sexual risk characteristics including STI symptoms, multiple sex partners, and having a primary partner who spent nights away from home. However, more non-HC than hormonal users reported consistent condom use (63% versus 5%; P < 0.001) during the previous 3 months. No important differences were found in STI prevalence at enrolment between contraceptive groups (Table 2).
Characteristics of participants completing versus not completing the study
The 24-month retention rate was 92%: 96% in Uganda and 88% in Zimbabwe. Retention was similar among the contraceptive groups (91% COC, 93% DMPA, 91% non-HC; P = 0.053). Mean follow-up was 21.9 months; median time between study visits was 81 days.
At baseline, older age (P = 0.002), lower education (P = 0.002), multiple sex partners (P = 0.016), less frequent sex (P < 0.001), inconsistent condom use (P = 0.024), and STI symptoms were associated with higher retention. Frequency of unprotected sex, commercial sex, a new sex partner, alcohol or drug use during sex,and having a chlamydial or gonococcal infection were not associated with retention.
Sexual risk factors among contraceptive groups during follow-up
Most study participants (67%) did not switch contraceptive methods during the study (Table 1). Pregnancy was more common among non-HC participants than among COC or DMPA users (17%, 5%, and 0.4% of visit segments, respectively; P < 0.001), while breastfeeding was more common among DMPA users than among non-HC and COC users (24%, 18%, and 8%, respectively; P < 0.001). STI symptoms and risky sexual behaviors were higher among non-HC participants than among hormonal contraceptive users but were generally rare. More non-HC participants than hormonal users had a primary partner with high-risk characteristics (48% versus 36%; P < 0.001). While hormonal contraceptive users reported higher coital frequency than non-HC participants (36% versus 28% reporting ≥ 15 acts per month; P < 0.001), more non-HC participants than hormonal users used condoms consistently (51% versus 13% of visit segments; P < 0.001).
Incident HIV infections by site and contraceptive exposure group
There were 213 incident HIV infections, giving an incidence rate of 2.75/100 woman-years (Table 3): 4.07/100 woman-years in Zimbabwe and 1.55/100 woman-years in Uganda. HIV incidence among the COC, DMPA, and non-HC group participants, respectively, was 2.59, 3.11, and 2.55/100 woman-years.
Neither COC nor DMPA use was significantly associated with risk of HIV acquisition in multivariate analysis (COC: HR, 0.99; 95% CI, 0.69-1.42; DMPA: HR, 1.25; 95% CI, 0.89-1.78; Table 4). Covariates significantly associated with HIV acquisition included Zimbabwe site and Uganda high-risk referral group, not living with partner, young age, participant behavioral risk, primary partner risk, and coital frequency (15-29 versus 0-14 acts per month). Neither higher coital frequency (≥ 30 acts per month) nor condom use were significantly associated with HIV acquisition.
The full multivariate model was also examined while restricting the analysis to sexually active participants reporting no condom use. While the risk estimates were raised for each hormonal method, neither COC (HR, 1.47; 95% CI, 0.78-2.80) nor DMPA (HR, 1.61; 95% CI, 0.85-3.06) use was significantly associated with HIV acquisition.
None of the sensitivity analyses had an important impact on the hormonal contraception-HIV acquisition relationship. The use of 0- and 4-month washout periods for COC and DMPA, respectively, yielded multivariate HR values of 1.02 (95% CI, 0.71-1.46) for COC and 1.27 (95% CI, 0.90-1.78) for DMPA. Inclusion of a time-varying pregnancy variable yielded a multivariate HR for COC use of 0.93 (95% CI, 0.64-1.34) and for DMPA use of 1.15 (95% CI, 0.80-1.64). When women were censored at first contraceptive switch or pregnancy, the multivariate HR values were 0.80 (95% CI, 0.51-1.24) for COC use and 0.88 (95% CI, 0.57-1.34) for DMPA use.
The effect of DMPA or COC use on the risk of HIV acquisition were also directly compared. No statistically significant difference was found in the risk of HIV acquisition (HR, 1.26; 95% CI, 0.92-1.74) between the two hormonal methods adjusting for covariates.
There was a significant interaction between hormonal contraception and site (P = 0.015). For COC use, the multivariate HR was 0.78 (95% CI, 0.51-1.20) in Zimbabwe and 1.43 (95% CI, 0.71-2.86) in Uganda. For DMPA use, the HR was 0.97 (95% CI, 0.65-1.45) in Zimbabwe and 1.81 (95% CI, 0.93-3.56) in Uganda.
The effect of sexually transmitted infection on the hormonal contraception-HIV link
Neither the presence of vaginal (trichomonas, bacterial vaginosis, yeast) nor cervical (chlamydia, gonorrhea) infections modified the hormonal contraception-HIV relationship. However, HSV-2 infection status at enrolment significantly modified the effect of hormonal contraception on HIV acquisition (P = 0.003).
Analyses stratified by HSV-2 status at enrolment
HIV incidence was higher among HSV-2-positive participants (52%) than HSV-2-negative participants (48%) at enrolment (Table 5). This relationship was consistent across contraceptive groups but was especially strong for the non-HC group. Among HSV-2-positive participants at enrolment, neither COC nor DMPA use increased HIV risk, after adjusting for covariates. However, among HSV-2-negative participants, HIV acquisition risk was significantly higher for both COC (HR, 2.85; 95% CI, 1.39-5.82) and DMPA (HR, 3.97; 95% CI, 1.98-8.00) users than for the non-HC group. Among HSV-2-negative participants at enrolment, there was no difference in the modification effect on the hormonal contraception-HIV association according to whether the participant remained HSV negative or whether she seroconverted during follow-up (P = 0.335). Also, using a time-varying HSV-2 variable, risk of HIV acquisition remained higher for both COC (HR, 2.48; 95% CI, 0.82-7.53) and DMPA (HR, 4.96; 95% CI, 1.73-14.20) users who were HSV-2 seronegative.
The association between hormonal contraception and HIV acquisition among HSV-2-negative participants at enrolment was not altered in sensitivity analyses. An important impact on the hormonal contraception-HIV relationship was not seen with a change in washout periods for COC and DMPA, the inclusion of a pregnancy variable, or censoring participants at first contraceptive switch or pregnancy.
Discussion
In this study designed to evaluate prospectively a possible association between hormonal contraceptive use and HIV acquisition, we found no overall association between use of hormonal contraception and the risk of HIV acquisition.
Our study had a number of important strengths. Hormonal contraceptive use was carefully measured and self-reports were validated against clinic records. The study was large and had 80% power to detect a 1.6-fold difference in HIV risk between hormonal contraceptive and comparison groups. Retention rates were high and similar among exposure groups. HIV infection was defined using a standardized testing algorithm that minimized errors and accurately timed the infection. We carefully evaluated potential confounding factors for their impact on risk estimates for hormonal contraception. We conducted sensitivity analyses that confirmed the robustness of our study results under different assumptions. Finally, the study was conducted primarily among a general population, thus enhancing the general applicability of the results.
Our finding of no overall association between hormonal contraceptive use and HIV acquisition suggests that the inconsistent results of previous studies would not support a strong association between hormonal contraception and HIV. Previous studies had important methodological shortcomings: few hormonal contraceptive users and low study power, poor measurement of hormonal contraceptive use and potential confounding variables, lack of adjusted analyses, and poor follow-up. Nevertheless, one methodologically strong study conducted among Kenyan sex workers found an association between both DMPA and COC use and HIV acquisition [15,17]. As our study participants were seeking family planning services, our study cannot rule out the possibility that hormonal contraception might increase HIV susceptibility among some highly exposed subpopulations. Also, the upper bound of the CI values around our risk estimates were very close to the risk estimates from the Kenyan study (HR values of 1.5 for COC, 1.8 for DMPA). Therefore, our study results are not inconsistent with a modest increase in HIV risk associated with hormonal contraceptive use, particularly for DMPA. Nevertheless, we found no evidence of higher HIV risk associated with hormonal contraception among high-risk subgroups in our study, including among participants with STI.
An important secondary objective was to assess whether other STI modified the hormonal contraception-HIV relationship. While vaginal and cervical infections did not modify the relationship, we did find a significantly increased risk of HIV acquisition associated with both COC and DMPA use among those HSV-2 negative at enrolment. Moreover, this relationship did not change whether the participant seroconverted for HSV-2 or remained HSV-2 negative. This finding was largely unexpected as HIV acquisition risk was markedly higher both among participants who were HSV-2 positive at enrolment and among those who seroconverted for HSV-2. Although we cannot rule out chance, this finding was strong, occurred with both COC and DMPA use and was robust in all sensitivity analyses.
The mechanism(s), if any, by which hormonal contraception might facilitate HIV infection are unknown. Possibilities include increased cervical ectopy associated with COC use [29,30], increased cervical chlamydial infection (and associated purulence) [6,28,29,31], the hypoestrogenic effect associated with DMPA use and reduction in hydrogen peroxide-producing lactobacilli and irregular uterine bleeding [32,33], suppression of the local (cell-mediated) immune response [15,34-36], increased recruitment of inflammatory and other target cells to the genital tract [15,35-37], or a direct effect on the infecting virus inoculum by upregulation of HIV gene expression and associated viral replication [34]. However, a solid biological explanation for our finding among the HSV-2-negative women is elusive. Because HSV-2 infection has such an important impact on HIV acquisition risk, the effect of HSV-2 could overshadow any impact of hormonal contraception. Our findings could be explained if both HSV-2 infection and hormonal contraception were associated with disruption of the genital epithelium but the effect of HSV-2 infection is greater than that of hormonal contraceptive use. Alternatively, HSV-2-negative status may be a marker for other variables that may interact with hormonal contraception, placing certain women at increased risk of HIV infection.
Our findings concerning HSV-2-negative participants have not been evaluated by other studies. Inconsistent results across studies might be explained by differences in study populations and their exposure to HIV and HSV-2 infections. The modifying role of HSV-2 serostatus reported here merits further investigation.
There was a marginally higher HIV risk associated with hormonal contraception in Uganda than in Zimbabwe. Possible explanations include chance, differences in the accuracy of measurement of potential confounders by country, or that the effect of hormonal contraception differs by HIV subtype (primarily A and D in Uganda, C in Zimbabwe) [38].
Our study has several limitations. The study was not randomized, and selection and confounding biases cannot be excluded. However, the study's prospective cohort design allowed women to continue using their chosen method. We believe that randomizing women to less effective contraception (such as condoms) is unethical. Moreover, use of a preferred method results in higher contraceptive continuation [39]. Consequently, we chose not to randomize participants. Second, the study relied on self-reported sexual behavior, and the accuracy of such data is unknown. Inaccurate or incomplete measurement of important covariates could result in residual confounding and could bias study results. However, it is unlikely that such confounding would differentially affect our risk estimates in the overall analyses compared with the HSV-2-stratified analyses. Finally, while the study had sufficient power to detect a hormonal contraception-HIV acquisition relationship overall, subgroup analyses had limited power.
In summary, this large, multisite study found no overall increased risk of HIV acquisition associated with hormonal contraceptive use. This provides reassurance for women in a setting of moderate and high HIV prevalence who need effective contraception; that any increased overall risk associated with hormonal contraception is, at most, modest. Among women who are HSV-2 negative, DMPA and COC users may be at increased risk of HIV acquisition. Additional research should confirm and interpret this finding. Regardless, our findings should reinforce efforts to counsel all women at risk of HIV infection to use condoms consistently and correctly to prevent HIV acquisition.
|
|
| |
| |
|
|
|
