Learn more: PMC Disclaimer | PMC Copyright Notice
Use of injectable hormonal contraception and women’s risk of herpes simplex virus type 2 acquisition: a prospective study of couples in Rakai, Uganda
Associated Data
Abstract
Background
The injectable hormonal contraceptive depo-medroxyprogesterone acetate (DMPA) has been associated with increased risk of HIV acquisition, but findings are inconsistent. It is unknown if DMPA increases risk for other sexually transmitted viral infections. We assessed the relationship between DMPA and incident herpes simplex virus type 2 (HSV-2) in women.
Methods
682 HIV and HSV-2 negative women whose HIV-negative male partners were enrolled in a trial of male circumcision in Rakai, Uganda were assessed for incident HSV-2 infection annually over two years. HSV-2 seroconversion was detected using HSV-2 ELISA confirmed by Western blot. The primary exposure was consistent DMPA use reported at two consecutive annual visits.
Findings
There were 70 HSV-2 seroconversions. Incidence was 13.5/100 person-years (py) among women consistently using DMPA (n=9/66.5 py), 4.3/100 py in pregnant women (n=18/423.5), and 6.6/100 py in women who were neither pregnant nor using hormonal contraception (HC) (n=35/529.5). Compared to women who were neither pregnant nor using HC, women consistently using DMPA had a 2.26 adjusted hazard ratio for HSV-2 seroconversion (95%CI: 1.09–4.69, p=0.029). Among 132 women with HSV-2 seropositive partners, seroconversion was 36.4/100 py in consistent DMPA users (n=4/11 py) and 10.6/100 py in women who were neither pregnant nor using HC (n=11/103 py; adjHR: 6.23; 95%CI: 1.49–26.3, p=0.012).
Interpretation
Consistent DMPA use may increase the risk of HSV-2 seroconversion; however, study power was limited. These findings should be evaluated in larger populations with more frequent follow-up.
Funding
Bill & Melinda Gates Foundation, Doris Duke Charitable Foundation, and National Institutes of Health.
Introduction
Hormonal contraception (HC) use is becoming increasingly common in Sub-Saharan Africa1, and long-acting injectable HCs are among the most effective and frequent contraceptive methods used by African women.2,3 Over the last two decades, depo-medroxyprogesterone acetate (DMPA), the most commonly used injectable contraceptive method3, has been associated with increased risk of female HIV acquisition in some but not all studies.4 A possible association between DMPA and HIV risk has prompted debate over its safety and relative reproductive health benefits with some calling for randomized controlled trials of various contraceptive methods including DMPA.5 It is unknown whether DMPA increases risk for other sexually transmitted viral infections such as herpes simplex virus type 2 (HSV-2).
Numerous observational studies have assessed the association between injectable and oral HC use and female HIV acquisition4, though the interpretation of results has been complex.5,6 In a systematic review, Polis et al. found limited evidence for increased HIV risk with either oral HC or norethisterone enanthate (NET-EN) injectable HC use4; however, results were conflicted with regard to whether DMPA was linked to HIV acquisition. Of seven high quality epidemiological studies that reported DMPA-specific estimates and were included in the review4, three studies found statistically significant increased HIV risk7–9, two studies found elevated but non-significant increased HIV risk10,11, and two studies found non-significant decreased HIV risk in DMPA users.12,13 A meta-analysis including these and three other studies estimated an 40% increased HIV risk with DMPA use among all women (pooled HR: 1.40; 95%CI:1.16–1.69), though this risk was lower among women in the general population (pooled HR: 1.31, 95%CI: 1.10–1.57).14
The pathophysiology of how DMPA, a progestin-only HC method, might theoretically increase HIV risk is unclear. Thinning of the vaginal epithelium and endometrium, increased immune activation in the genital tract, and heightened risk for other sexually transmitted infections, including HSV-2, have been postulated as mechanisms15,16. Vaginal thinning following administration of DMPA has been observed in rhesus macaques17,18; however, studies in women have found little evidence for vaginal thinning or cervical ectopy with DMPA use.19–21 Others studies suggest that progestins may modulate HIV susceptibility via increased expression of CCR5 co-receptors on CD4+ T cells and recruitment of HIV target cells (e.g. dendritic cells) to the vaginal and stromal epithelial tissues.22–24
HSV-2 is a common viral STI and its transmission is associated with substantial morbidity worldwide.25,26 The global burden of HSV-2, like that of HIV, is disproportionately concentrated in Sub-Saharan Africa 27. HSV-2 infection is a frequent cause of genital ulcer disease 28,29, which is a risk factor for HIV.30,31 The primary HSV-2 target cells in the genital tract are the epithelial cells, though HSV-2 can also infect dendritic cells 32–35. Challenge studies in mice have found increased HSV-2 susceptibility and impaired immunologic responses to HSV-2 in the presence of DMPA.36,37 Despite high rates of HIV and HSV-2 coinfection31,38, only one epidemiologic study has examined the association between use of oral or injectable HC and incident HSV-2 infection in women.39 This study of HIV-negative sex workers in Kenya found no association between use of injectable or oral HC and HSV-2; however, the HSV-2 serostatus of the women’s partners was unknown. Here, we assessed whether use of DMPA or oral HC was associated with increased risk of HSV-2 acquisition among 682 HIV-negative women in long-term sexual partnerships with HIV-negative men of known HSV-2 serostatus in Rakai, Uganda.
Methods
Study design and participants
From September 2003 to September 2005, the Rakai Health Sciences Program, in Rakai, Uganda enrolled 4996 HIV-negative men into a randomized trial of male circumcision (MC) for HIV and STI prevention.40,41 Men were eligible for enrollment if they were HIV-negative, uncircumcised, aged 15–49, and provided written informed consent to be randomly assigned to receive immediate MC (intervention arm) or MC delayed for 24 months (control arm). Consenting female partners of male trial participants who were married or in long term consensual relationships were invited to participate in a separate follow-up study.42
At enrollment and annual follow-up visits, women were interviewed to ascertain socio-demographic characteristics, sexual risk behaviors, and health status. Men were similarly interviewed. Women were also asked about pregnancy status and intention and use of family planning methods, including oral, injectable, or implantable HC. Though women were not asked to specify which form of injectable contraception they were using, DMPA is the primary injectable HC method available in Rakai, Uganda (>98% of users). At each visit, men and women were asked to provide blood samples which were maintained at 4–10°C for less than 6 hours and then at −80C until assayed.
Our analysis was restricted to HIV and HSV-2 negative women, aged 15–49, who were enrolled concurrently with their HIV-negative male partner. Female participants were also required to have at least one follow-up visit at which her HSV-2 status was ascertained and her partner’s HIV status was also known. We excluded women if either they or their male partners HIV seroconverted (n=10) since there is an association between HIV and HSV-2 acquisition30,31,43, and these events were too infrequent for stratified analyses. Women who self-reported using intrauterine devices (n=2) or implants (n=7) were also excluded since small numbers limited power.
The trials were approved by the Uganda National Council for Science and Technology, the Science and Ethics Committee of the Uganda Virus Research Institute (Entebbe, Uganda), the Committee for Human Research at Johns Hopkins University Bloomberg School of Public Health (Baltimore, MD, USA), and the Western Institutional Review Board (Olympia, WA, USA).41,42 The written informed consent included the participant’s permission to use samples for future research. The trials were registered with ClinicalTrials.gov (NCT00425984 and NCT00124878).
Laboratory assays
HSV-2 infection was determined by HSV-2 ELISA (Kalon Biological Ltd, Guilford, UK).38,40,44 As previously described, an HSV-2 seroconversion was defined as a negative enrollment serology (optical density ≤0.9) followed by a positive follow up serology (optical density index value ≥ 1.5). All incident HSV-2 positive cases by ELISA were confirmed by Euroimmun Western blot (Euroimmun, Lubeck, Germany).38,40 HIV status was determined using two separate ELISAs, and discordant results were confirmed by HIV-1 Western Blot. HSV-2 and HIV-1 infection status was determined for women and their partners41.
Classification of exposure to hormonal contraceptives
Self-reported current HC use was assessed at each study visit (Figure 1A). HC use was treated as a time-varying exposure, and use during a given visit interval was based upon report of HC use at the current (t) and prior study visits (t−1) (Figure 1B). HC use during a visit interval was defined as either DMPA or oral HC use. More than 80% of all HC use was DMPA, so consistency of HC use was analyzed among DMPA users. DMPA use was classified as consistent when a woman reported using DMPA at both her current and prior study visit. When a woman reported using DMPA at her current but not prior visit, she was classified as having initiated use during that interval. Women who self-reported using DMPA at the prior but not current visit were classified as having discontinued use. Women who were pregnant at the current or prior visit and who reported no HC use, DMPA or oral, were defined as a separate exposure group.
Figure (A) shows individual-level patterns of hormonal contraceptive use between enrollment and year 2 study visits for a random sample of 50 hormonal contraceptive users. At a given visit, women were classified (based on self-report) as using DMPA (purple), using oral HC (blue), pregnant (green), not using HC and not pregnant (Red), or as having no data (gray). Figure (B) shows how exposure to hormonal contraception was classified for a visit interval using the contraceptive/pregnancy statuses from the current and prior study visits.
The reference group was women who were neither pregnant nor using any HC method at the current or prior study visit, and included women who were using condoms, natural family planning methods, spermicides, or who had a hysterectomy or tubal ligation. In primary analyses, we analyzed condom use as a time-varying confounder, and in a sensitivity analysis we excluded women who self-reported condom use at either enrollment or follow-up visits.
Statistical analyses
Enrollment demographic, sexual behaviors, and health characteristics of women and their male partners were tabulated. Significant differences between women who ever reported oral or DMPA use and those who never reported HC use were determined using Chi-square and Wilcoxon rank-sum tests for categorical and continuous variables, respectively.
The primary study outcome was HSV-2 seroconversion. Women who did not acquire HSV-2 were administratively censored at year two or at loss to follow-up. All women were included in the primary analysis regardless of their spouse’s HSV-2 status. In a separate analysis, the association between DMPA use and HSV-2 acquisition was assessed among women whose male partner was HSV-2 seropositive at her current study visit (n=132), including women whose partners acquired HSV-2 during the same visit interval (n=6).
Time at risk was calculated as the time elapsed between annual visits. Women who seroconverted between visits were assumed to seroconvert at the midpoint of the visit interval. Incidence of HSV-2 was calculated per 100 person-years. Given the discrete nature of the data (only two follow-up visits) and the large amount of time between annual visits, complementary log-log regression models with generalized estimating equations and robust variance estimators were used to estimate hazard ratios (HR) and 95% confidence intervals (95% CI). The complementary log-log model is a discrete-time survival model and similar to the continuous-time Cox proportional hazards model; however, the discrete model assumes constant hazard within an interval, though the hazard may vary between intervals.45 Generalized estimating equations (GEE) and robust variance estimators were used to obtain population-averaged effect estimates and account for correlation between observations from the same woman over time.46 The complementary log-log regression models were used for all primary analyses and the sensitivity analysis excluding women who self-reported condom use. Models were implemented using the xtcloglog command with the pa and robust options in STATA Version 11.0 (StataCorp LP, College Station, TX).
Variables were included in adjusted analyses if they were (1) identified as confounders in prior observational studies of HC use and HIV acquisition and not a marker of HSV-2 or HIV-1 infection6 or (2) were associated with HC exposure and HSV-2 seroconversion in the study population at p<0.1. Time-invariant covariates included age and educational status of female participants and their male partners, and female self-reported number of lifetime sexual partners at enrollment. Time-varying confounders included male circumcision (MC), female self-reported coital frequency, and female and male partner self-report of any condom use and non-marital partners in the past year. Coital frequency and male circumcision status were dropped from the final adjusted model because they did not significantly alter the primary effect estimates (>10%) or improve the model likelihood in nested comparisons as determined by quasi-likelihood under independence criterion, or QIC.47
Role of the funding source
The study sponsors had no role in the design, analysis, interpretation of results, or manuscript preparation.
Results
A total of 740 HSV-2 and HIV-negative females were concurrently enrolled with their HIV-negative male spouse. After excluding 41 women with insufficient follow-up (n=41, 5.5%), women who used intrauterine devices (n=2, 0.3%) or implantable contraception (n=5, 0.7%), and women who HIV seroconverted (n=6, 0.8%) or whose partners HIV seroconverted (n=4, 0.5%), there were 682 women with 660 male partners in the final analysis population. The number of female participants exceeded the number of male partners because of polygamous relationships. There were equal numbers of women with male partners in the intervention (n=341) and control (n=341) arms of the male circumcision trial. Female participants contributed 1,207 person-visits to the analysis and were followed for an average of 1.7 person-years. Retention was 95% at year 1 (n=650) and 82% at year 2 (n=557). Follow-up between year one and year two did not statistically differ between oral HC users (82%), DMPA users (consistent use, 72%; initiated use, 67%; discontinued use, 83%), or pregnant women (84%) when these groups were compared to women who were neither pregnant nor using HC (79%).
Twenty-five percent (n=173/682) of women self-reported ever using HC at enrollment or a follow-up visit. Among women using HC, 147 (85%) reported DMPA use only, 22 (13%) reported oral HC use only, and 4 (2%) reported DMPA and oral use at different study visits. Women frequently initiated (n=52) and discontinued DMPA use (n=99) during the study (Figure 1A). Pregnancy was common: 50% (n=338/682) of female participants experienced a pregnancy at either an enrollment (n=163, 24%) or follow-up visit (n=218, 32%).
Table 1 shows participant characteristics at enrollment. The median age of female participants was 24 years (IQR: 21–28) and the median age of their male partners was 28 years (IQR: 24–32). Only 16% (n=112) of women reported using condoms at enrollment. Self-report of extra-marital partners was 3% among women (19/682) and 33% in men (225/682). There were differences in enrollment characteristics between women who did and did not use DMPA or oral HC (Table 1). Women who reported DMPA or oral HC use were slightly older, more educated, and less likely to be pregnant. Their male partners were also older, more educated, and more likely to self-report using condoms in the last year. At enrollment, both DMPA and oral HC users and non-users had comparable proportions of HSV-2 seropositive partners, self-reported condom use, or self-reported genital ulcer disease. Partners of DMPA and oral users and non-users also had comparable self-reported genital ulcer disease and assignment to the intervention arm of the MC trial.
Table 1
Enrollment characteristics of 682 female participants stratified by use of any oral or injectable hormonal contraceptives (HC) at enrollment, year 1 or, year 2 study visits.
| No HC | Any Oral HC† | Any DMPA† | |||
|---|---|---|---|---|---|
|
| |||||
| Participant characteristic | N (%) | p value* | N (%) | p value* | |
| Overall | 509 (74.6) | 26 (682) | 151 (22.1) | ||
| Demographics at enrollment | |||||
| Female Age, years (median, IQR) | 24 (21–27) | 25 (21–27) | 0.874 | 25 (22–29) | 0.023 |
| Male partner’s age, years (median, IQR) | 27 (24–32) | 30 (25–33) | 0.633 | 29 (25–33) | 0.085 |
| Polygamous male partner | 55 (10.8) | 4 (15.4) | 0.467 | 19 (12.6) | 0.543 |
| Female Education status | |||||
| None | 67 (13.2) | 2 (7.7) | 13 (8.6) | ||
| Primary | 379 (74.5) | 16 (61.5) | 98 (64.9) | ||
| Secondary | 57 (11.2) | 6 (23.1) | 30 (19.9) | ||
| Postsecondary | 6 (1.2) | 2 (7.7) | 0.011 | 10 (6.6) | <0.001 |
| Male partner’s educational status | |||||
| None | 51 (10.0) | 1(3.9) | 6 (4.0) | ||
| Primary | 374 (73.5) | 12 (46.2) | 97 (64.2) | ||
| Secondary | 65 (12.8) | 8 (30.8) | 29 (19.2) | ||
| Postsecondary | 19 (3.7) | 5 (19.2) | <0.001 | 19 (12.6) | <0.001 |
| Sexual behaviors at enrollment | |||||
| Female Lifetime sex partners | |||||
| One | 234 (46.0) | 9 (34.6) | 61 (40.4) | ||
| Two | 177 (34.8) | 9 (34.6) | 53 (35.1) | ||
| Three | 65 (12.8) | 5 (19.2) | 23 (15.2) | ||
| Four or more | 33 (6.5) | 3 (11.5) | 0.488 | 14 (9.3) | 0.455 |
| Non-marital relations in the last year | 17 (3.3) | 0 (0.0) | 0.344 | 2 (1.3) | 0.193 |
| Condom use in the last year** | 79 (15.6) | 4 (15.4) | 0.978 | 30 (19.9) | 0.214 |
| Male partner, sexual behaviors at enrollment | |||||
| Male partner, non-marital relations in the last year | 163 (32.0) | 5 (19.2) | 0.170 | 58 (38.4) | 0.144 |
| Male partner, condom use in the last year** | 159 (31.2) | 6 (23.1) | 0.379 | 68 (45.0) | 0.002 |
| Health characteristics | |||||
| Ever pregnant during study | 269 (52.9) | 11 (42.3) | 0.294 | 61 (40.1) | 0.007 |
| Female self-reported GUD at enrolment | 42 (8.3) | 0 (0.0) | 0.127 | 12 (8.0) | 0.905 |
| Male partner, self-reported GUD at enrolment | 23 (4.5) | 0 (0.0) | 0.268 | 8 (5.3) | 0.691 |
| Male partner in trial intervention arm | 256 (50.3) | 14 (53.9) | 0.724 | 73 (48.3) | 0.674 |
| Male partner’s HSV-2 status at enrolment | |||||
| Negative | 366 (71.9) | 22 (80.8) | 116 (76.8) | ||
| Indeterminate | 61 (12.0) | 2 (7.7) | 12 (8.0) | ||
| Positive | 80 (15.7) | 3 (11.5) | 22 (14.6) | ||
| Unknown | 2 (0.4) | 0 (0.0) | 0.792 | 1 (0.7) | 0.496 |
HC=hormonal contraception, oral or DMPA; IQR=Interquartile range; HSV-2=herpes simplex virus-2;
There were 26 instances where a woman reported using DMPA or oral HC at the start of a visit interval and was found to be pregnant at her following visit (Figure 1B). Sixty-one percent (n=16/26) of these women reported planning to become pregnant during the interval. Of those who experienced an unplanned pregnancy (n=10), one-reported using oral HC at the time she became pregnant and one reported using DMPA. The remaining 8 women reported using no form of contraception. In those instances where a woman became pregnant and was not using HC at her prior visit (n=199; Figure 1B), 54% (n=108/199) experienced an unplanned pregnancy. Of these women, none reported using hormonal contraception, DMPA or oral, at the time she became pregnant.
As shown in Table 2, there were 70 incident HSV-2 infections. Thirty-nine infections occurred among women with HSV-2 seronegative male partners (56%), 23 (33%) among women with HSV-2 seropositive partners, and eight (11%) among women who had partners of unknown or indeterminate HSV-2 status. Nineteen percent (n=13/70) of women who seroconverted reported genital ulcers in the last year (HR=2.2, 95%CI: 1.2–4.0). Overall, HSV-2 incidence was 5.8/100 person-years (py). We observed no increased risk of HSV-2 acquisition during pregnancy (adjHR: 0.66, 95% CI: 0.37–1.19) or among oral HC users (adjHR=0.49; 95%CI: 0.08–3.01). Risk of HSV-2 was significantly increased among women consistently using DMPA (adjHR 2.26, 95%CI: 1.09–4.69, p=0.029); however, there was no indication of heightened risk among those who recently initiated or discontinued DMPA use (Table 2). In a sensitivity analysis, results were similar in women who never reported condom use (Supplemental Table 1).
Table 2
Risk factors for incident HSV-2 infection, including use of hormonal contraception (HC), among 682 HSV-2 and HIV-negative women in Rakai, Uganda.
| Incident infections/person-years | Incidence | Unadjusted hazard ratios | Adjusted hazard ratios† | |||
|---|---|---|---|---|---|---|
|
| ||||||
| No./py | per 100 py | HR (95%CI) | p value | HR (95%CI) | p value | |
| No HC use, not pregnant‡ | 35/529.5 | 6.6 | 1.00 (Ref.) | - | 1.00 (Ref.) | - |
| No HC use, pregnant | 18/423.5 | 4.3 | 0.64 (0.36–1.12) | 0.119 | 0.66 (0.37–1.19) | 0.169 |
| Oral HC use | 1/35.5 | 2.8 | 0.44 (0.06–3.22) | 0.418 | 0.49 (0.08–3.01) | 0.441 |
| DMPA use | ||||||
| Discontinued use | 2/51 | 3.9 | 0.57 (0.14–2.37) | 0.437 | 0.58 (0.13–2.51) | 0.535 |
| Initiated use | 5/96.5 | 5.2 | 0.75 (0.30–1.92) | 0.554 | 0.75 (0.29–1.92) | 0.632 |
| Consistent use | 9/66.5 | 13.5 | 2.02 (0.96–4.26) | 0.066 | 2.26 (1.09–4.69) | 0.029 |
| Female Age (years) | - | - | 0.95 (0.90–1.00) | 0.043 | 0.95 (0.86–1.04) | 0.247 |
| Male Age (years) | - | - | 0.98 (0.94–1.03) | 0.400 | 1.00 (0.92–1.09) | 0.992 |
| Female Education status | ||||||
| None | 5/151.5 | 3.3 | 0.49 (0.20–1.21) | 0.165 | 0.66 (0.29–1.50) | 0.323 |
| Primary | 58/88.5 | 6.8 | 1.00 (Ref.) | - | 1.00 (Ref.) | - |
| Secondary | 6/164 | 3.7 | 0.55 (0.24–1.28) | 0.165 | 0.41 (0.16–1.05) | 0.065 |
| Postsecondary | 1/31.5 | 3.2 | 0.50(0.07–3.49) | 0.488 | 0.15 (0.02–1.27) | 0.083 |
| Male, educational status | ||||||
| None | 4/104 | 3.8 | 0.82 (0.29–2.30) | 0.688 | 1.09 (0.39–3.11) | 0.863 |
| Primary | 42/854.5 | 4.9 | 1.00 (Ref.) | - | 1.00 (Ref.) | - |
| Secondary | 18/171 | 10.5 | 2.19 (1.26–3.81) | 0.005 | 1.92 (1.09–3.41) | 0.024 |
| Postsecondary | 6/73 | 8.2 | 1.71 (0.74–3.97) | 0.212 | 2.85 (1.13–7.18) | 0.026 |
| Female Lifetime sex partners | ||||||
| One | 24/549 | 4.4 | 1.00 (Ref.) | - | 1.00 (Ref.) | - |
| Two | 22/415.5 | 5.3 | 1.23 (0.69–2.20) | 0.486 | 1.27 (0.71–2.29) | 0.414 |
| Three | 11/161.5 | 6.8 | 1.58 (0.77–3.26) | 0.212 | 1.41 (0.66–3.01) | 0.371 |
| Four or more | 13/76.5 | 17.0 | 3.99 (2.03–7.82) | <0.001 | 3.45 (1.72–6.92) | <0.001 |
| Female Non-marital relations | ||||||
| No | 65/1162 | 5.6 | 1.00 (Ref.) | - | 1.00 (Ref.) | - |
| Yes | 5/40.5 | 12.3 | 2.26(0.96–5.29) | 0.062 | 1.28 (0.46–3.58) | 0.633 |
| Male, non-marital relations | ||||||
| No | 46/918.5 | 5.0 | 1.00 (Ref.) | - | 1.00 (Ref.) | - |
| Yes | 24/284 | 8.5 | 1.68 (1.03–2.76) | 0.039 | 1.08 (0.61–1.90) | 0.802 |
| Female Condom use | ||||||
| No | 54/973 | 5.6 | 1.00 (Ref.) | - | 1.00 (Ref.) | - |
| Yes | 16/216.5 | 7.4 | 1.33 (0.77–2.33) | 0.337 | 0.89 (0.49–1.63) | 0.715 |
| Male, condom use | ||||||
| No | 36/822.5 | 4.4 | 1.00 (Ref.) | - | 1.00 (Ref.) | - |
| Yes | 33/371.5 | 8.9 | 2.00 (1.25–3.21) | 0.004 | 1.70 (0.98–2.97) | 0.060 |
HC=hormonal contraception, oral or DMPA; HR=hazard ratio; adjHR=adjusted hazard ratio;
Other factors that were significantly associated with female HSV-2 seroconversion after adjustment included a higher numbers of female lifetime sexual partners at enrollment, and male partner’s higher educational status (Table 2). Few women reported non-marital relationships during follow-up (n=35/682, 5 %). Among the 39 HSV-2 seroconverting women who had HSV-2 seronegative partners, 5% (2/39) reported non-marital relationships during the interval between the last negative and first positive visits.
There were 132 women who had an HSV-2 seropositive spouse at enrollment or follow-up. Of these women, 27 (20 %) had partners who HSV-2 seroconverted after enrollment including 6 HSV-2 female incident cases whose spouses seroconverted during the same visit interval. The rate of female HSV-2 acquisition was 10.6/100 py (23/218 py) among women with an HSV-2 infected partner, which was significantly higher than incidence among women with an HSV-2 negative partner (4.8/100 py; 39/814.5 py, p=0.001). Among women with HSV-2 infected partners, incidence of HSV-2 was 36.4/100 py (n=4/11 py) in women who consistently used DMPA, whereas incidence was 10.7/100 py (n=11/103) in women who were neither pregnant nor using HC. The increased risk of incident HSV-2 among consistent DMPA users was statistically significant (adjHR=6.23, 95%CI=1.49–26.3; p=0.012).
Discussion
We found consistent use of DMPA was associated with greater than two-fold increased risk of HSV-2 seroconversion. In an analysis restricted to women with known exposure to HSV-2 seropositive partners, the risk of incident HSV-2 with consistent DMPA use was over 6-fold. The pregnancy rate was high, with 50% experiencing pregnancy during the two year follow-up, predominately among women not using any form of contraception (99.5%). Some, but not all, observational studies have observed increased risk for HIV during pregnancy48, including one prospective study in Rakai, Uganda.49 We observed no such analogous risk for HSV-2 acquisition during pregnancy in this study.
Most women who reported using an HC method used DMPA. DMPA use was highly dynamic: 85% (n=129) of DMPA users initiated or discontinued use during the study and only 38% (n=57) consistently used DMPA over a one year interval. DMPA use has been associated with HIV acquisition in several studies including a meta-analysis, but the data are inconsistent4. Two studies assessed the association between DMPA use and HIV in Rakai; however, both found no statistically significant increased HIV risk in DMPA users.50,51 In contrast to DMPA, there is little evidence that oral HC methods augment HIV risk4. We had very limited power to detect an association between oral HC use and HSV-2 acquisition, though our results suggest there is no increased risk of HSV-2 acquisition among oral HC users.
We cannot rule out that the possibility that unmeasured or poorly measured confounding affected our results.6 For example, if women with high risk sexual behaviors were more likely to use DMPA but these behaviors were not reported accurately, the findings could be biased. Indeed, women substantially underreported extra-marital partnerships as evidenced by 50% of all HSV-2 infections occurring among women who reported no external relationships but had HSV-2 seronegative partners. However, the findings were stronger when analyses were conducted among women with known HSV-2 infected partners. Moreover, women with incident HSV-2 infection were no more likely to have an HSV-2 seronegative spouse if they were using DMPA (p=0.341).
Consistent condom use can reduce HIV acquisition by 80%52; however, meta-analyses suggest that condom efficacy is only 30% for HSV-2 acquistion.53 This is because HSV-2 can be transmitted to and from genital areas that are not protected by male condoms.25 Though condoms have been identified as important time-dependent confounders in observational studies of HIV and HC use6, we found no association between female reported condom use and HSV-2 acquisition. We also found no significant differences in our effect estimates in a sensitivity analysis excluding female participants who ever reported using condoms.
There are other limitations to this study. Information on contraceptive exposures was based on self-reported use of DMPA or oral HC use at the time of a study visit, and we inferred consistency of DMPA use by reported use at successive visits. The trial from which these data were derived was not designed to assess associations between contraception and HSV-2 acquisition, and the interval between follow up visits (one year) cannot accurately assess the timing of HSV-2 acquisition in relation to contraceptive exposures. The long follow-up intervals could also affect women’s recall of their risk behaviors. However, the prospective design, laboratory methods, and inclusion of information on both female participants and their stable male partners were study strengths. Notably, the main finding of increased HSV-2 seroconversion risk associated with consistent DMPA use was based on only 9 incident cases and the confidence intervals around the point hazards estimate of 2.29 are wide. Nevertheless, other studies of HIV acquisition and injectable contraception were also based on comparably small numbers of seroconversions54.
In conclusion, this study found that consistent use of DMPA may increase women’s risk for HSV-2 acquisition. The findings may be relevant to the associations between HIV risk and use of DMPA in some prior studies, and might indicate that DMPA broadly affects the risk of viral sexually transmitted infections. However, these results are based on only a small number of cases and need to be confirmed in other populations with more frequent follow-up and validated contraceptive exposure histories. Any future randomized trials of HC should consider assessing incidence of HSV-2 in addition to HIV as potential study outcomes.5 The possibility of an increased risk for HIV and HSV-2 associated with DMPA must be weighed against its benefits, including reduced risk of maternal mortality and unwanted pregnancy. Increasing access to other forms of highly effective long-acting, low- dose contraceptive methods, including intrauterine devices and implants is needed in Sub-Saharan Africa.10
Table 3
Association between hormonal contraception (HC) and HSV-2 acquisition among 132 HIV-negative women with HSV-2 seropositive partners†.
| Incident infections/person-years | Incidence | Unadjusted hazard ratios | Adjusted hazard ratios‡ | |||
|---|---|---|---|---|---|---|
| No./py | per 100 py | HR (95%CI) | p value | HR (95%CI) | p value | |
| No HC use, not pregnant* | 11/103 | 10.7 | 1.00 (Ref.) | - | 1.00 (Ref.) | - |
| No HC use, pregnant | 3/71.5 | 4.2 | 0.38 (0.11–1.35) | 0.133 | 0.27 (0.06–1.13) | 0.075 |
| Oral HC use | 0/6.0 | - | - | - | - | - |
| DMPA use | ||||||
| Discontinued use | 1/7.5 | 13.3 | 1.19 (0.16–9.01) | 0.863 | 2.42 (0.29–20.1) | 0.411 |
| Initiated use | 4/19 | 21.1 | 1.86 (0.59–5.78) | 0.283 | 1.42 (0.25–8.08) | 0.691 |
| Consistent use | 4/11 | 36.4 | 3.31 (0.99–11.0) | 0.051 | 6.23 (1.49–26.3) | 0.012 |
HC=hormonal contraception, oral or DMPA; HR=hazard ratio; adjHR=adjusted hazard ratio;
Acknowledgments
We are most grateful to the study participants and the Rakai Community Advisory Board whose commitment and cooperation made this study possible.
Funding
This study was supported by the Doris Duke Charitable Foundation (#2011036), Bill and Melinda Gates Foundation (22006.03), the National Institutes of Health (#U1AI51171), the Division of Intramural Research, National Institute of Allergy and Infectious Diseases and the Fogarty International Center, 1D43TWOO9578-01. M.K.G was supported by NIH T32AI102623 and the Doris Duke Charitable Foundation. A.A.R.T. was supported by the NIH 1K23AI093152-01A1 and the Doris Duke Charitable Foundation Clinician Scientist Development Award.
Footnotes
Author Contributions
M.K.G., R.H.G, F.M, and A.A.R.T conceptualized and designed the study. G.K., F.N., J.K, S.J.R., D.S., R.H.G., M.J.W. oversaw survey data collection. A.A.R.T., T.C.Q, and A.R.R. oversaw and performed laboratory assays and contributed reagents. M.K.G. conducted statistical analyses. F.M and T.L. assisted with data management and organization. All authors contributed to the writing of the manuscript and agreed with the manuscript results and conclusions.Declaration of Interests
The authors declare that they have no potential conflicts of interest relevant to this report.