Women are disproportionately affected by HIV in sub-Saharan Africa, accounting for 60% of all HIV infections in the region.1 Many women living in areas of high HIV prevalence are of childbearing age, and may desire access to safe and effective contraceptive methods. Access to comprehensive family planning services is critical in sub-Saharan Africa, where rates of maternal mortality are high, with 900 maternal deaths per 100,000 live births.2,3 Contraception can provide control over the spacing and limiting of births, reduce pregnancy-related morbidity and mortality, and reduce the demand for abortion, which is particularly crucial in regions where access to safe and legal abortion services are restricted.4,5 Contraceptive use has also been linked to social and economic benefits for women and their families.6
Hormonal contraceptive methods are among the most effective methods of pregnancy prevention, but evidence assessing whether hormonal contraception (HC) affects events in early HIV infection, such as establishment of viral load (VL) setpoint, is mixed. A high VL setpoint is associated with rapid HIV disease progression.7-10 VL setpoint is thought to reflect the equilibrium between viral replication and immune response. This generally occurs between 2 and 6 months after HIV infection and remains relatively steady throughout the asymptomatic latent phase of HIV infection.11 HIV progression can also be affected by other factors, including viral subtype, viral diversity, coinfections such as sexually transmitted infections and malaria, and older age at seroconversion.12-16
Three studies have examined whether HC use near the time of HIV infection affects VL setpoint. One study of 161 sex workers in Kenya found that use of injectable contraception at seroconversion was associated with a higher VL setpoint, relative to nonhormonal contraceptive users.17 However, in a related analysis of these data which controlled for viral diversity, HC use was not independently associated with higher VL setpoint.18 Another study of 27 women being treated for genital infections in Malawi also found that users of injectable contraception had a higher VL setpoint.19 Conversely, a recent prospective study among 175 Ugandan and Zimbabwean women seeking family planning services found no association between HC use and cervical VL in early infection or plasma VL at setpoint.20 No studies have been conducted in a population-based cohort, and it may be difficult to generalize findings from studies of high-risk groups such as sex workers.
The objective of our study was to assess the relationship between use of HC before HIV seroconversion on VL setpoint in a large, population-based cohort of women with incident HIV infection.
The Rakai Community Cohort Study
Since 1994, the Rakai Health Sciences Program has followed an open cohort of all consenting adults aged 15-49 years living in 50 communities in Rakai District, southwestern Uganda. The Rakai Community Cohort Study has been described in detail,21,22 but in brief, participants provide informed consent and are privately interviewed by same-sex interviewers every 12-15 months, using a standardized questionnaire. Venous blood is collected at each survey round for HIV-1 and sexually transmitted infection serology. More than 90% of eligible individuals have participated in any given survey round. Sexually-experienced female participants who acquired HIV and had VL measurements while under observation in the Rakai Community Cohort Study were eligible for inclusion in this analysis.
Peripheral blood was collected in tubes that contained EDTA (Becton Dickinson), separated within 8 hours of collection by centrifugation, and frozen at −80°C until tested. All specimens were tested in batch for HIV-1 RNA at the Rakai Health Sciences Program laboratory in Uganda. We quantified HIV-1 RNA in human plasma using the Amplicor HIV-1 Monitor Test, Version 1.5 (Roche Diagnostics Corporation, Indianapolis, IN), which has been validated to monitor HIV-1 VL in patients identified as being HIV-1 positive for all group M subtypes of HIV-1, including subtypes A and D, which predominate in Uganda. We quantified HIV-1 viral RNA using the HIV-1 Quantification Standard Procedure. HIV-1 RNA levels in the test were determined by comparing the HIV-1 signal to the quantification standard signal for each specimen. Test results less than 400 copies per milliliter were below the lower limit of quantification of the standard test and were reported as “HIV-1 RNA detected less than 400 (2.6 log10) HIV-1 RNA copies per milliliter.” Such samples were recorded as 399 copies per milliliter to conservatively calculate VL setpoint. CD4 cell counts were performed with FacsCalibur until 2004, and a FacsCount system thereafter.
VL setpoint was estimated by averaging all VL measurements that fulfilled each of the following criteria: (1) collected 6 months or more after the estimated date of seroconversion, (2) collected before the date of disease progression to CD4 ≤250 or WHO Stage 3 or 4 or initiation of antiretroviral therapy, and (3) for individuals who did not experience disease progression to CD4 ≤250 or WHO Stage 3 or 4 while under observation, collected before the median time to CD4 ≤250 or WHO Stage 3 or 4 or death in this population (4.5 years from estimated date of seroconversion).23 These criteria were chosen to exclude VL measurements from samples taken during acute infection or advanced disease. VL setpoint levels were log10 transformed for analysis.
Date of HIV seroconversion was estimated as the midpoint between the last HIV-negative and the first HIV-positive antibody test dates. Seroconversion was detected by 2 independent enzyme-linked immunosorbent assay tests confirmed by Western Blot. For confirmed seroconverters, we performed polymerase chain reaction tests on the prior negative sample. If the subject had acute infection during her study visit (as indicated by polymerase chain reaction positive and enzyme-linked immunosorbent assay negative tests), that visit was considered the date of HIV seroconversion. For individuals who had an indeterminate Western Blot test result (with an RNA positive test) between their last negative and first positive HIV antibody test, the date of this indeterminate test was considered the HIV seroconversion date.
For 75% of the 120 women, disease progression was defined as CD4 cell count ≤250 cells per cubic millimeter (when individuals in Rakai become eligible for antiretroviral therapy), and the onset date was estimated as the midpoint between the last CD4 count above 250 and first CD4 count ≤250. If the first CD4 count was ≤250, that was considered the onset date. For 9 women without available CD4 counts, disease progression was defined as clinical symptoms meeting the WHO clinical stage 3 or 4 criteria,24 with onset date calculated as the date of the symptoms interview. For 21 women with both CD4 information and clinical symptoms information, the earlier of the 2 dates was considered the onset date.
Date of death was ascertained from information collected during an annual census and surveys, and verbal autopsy interviews. A structured verbal autopsy was administered to the individual closest to the deceased at the time of death to obtain information on the date and probable cause of death. We were unable to definitively distinguish between HIV-related and HIV-unrelated deaths, but previous analyses in Rakai have found that more than 90% of all deaths among HIV-positive persons are HIV related.25,26
HC use was defined as self-reported current HC use at the last HIV-negative visit before the visit at which HIV seroconversion was identified, given that HC use in the interval before HIV seroconversion was most likely to accurately represent HC use at the time of HIV infection. Our definition of HC use included use of oral and injectable contraceptives, and after 1999, also included Norplant.
Socioeconomic status (SES) was determined based upon materials used to construct the floor, walls and roof. “Poor” meant that the floor was either mud, wood, or “other”, walls were either thatch, mud/wattle, “other”, or “mobile” (indicating that the individual's family does not own a structure in that community), and that the roof was either made of thatch or “other materials” (such as banana fiber or papyrus), or “mobile.” “Above average” indicates that the roof was either made of iron or tiles, that the floor was cement, and that the walls were cement, brick, or stone. Average includes all other categories.
We calculated VL setpoint by averaging all VL measurements meeting our criteria for setpoint for each individual. We compared VL setpoints between HC users and nonusers using a t test, and we used univariate and multivariate linear regression to examine the effect of HC use before HIV seroconversion on VL setpoint. The following variables, taken from last HIV-negative visit, were considered as potential confounders: age, educational level, marital status, SES, parity, breastfeeding, number of sexual partners in past year, current nonmarital relationship, condom use, pregnancy, and genital ulcer disease. In multivariate analyses, we controlled for all variables associated with both HC use and VL setpoint at P = 0.20 in univariate regressions (socioeconomic status, nonmarital relationship, parity, and breastfeeding), and for variables included for comparability with previous research (age, education, number of sexual partners in past year, and current genital ulcer disease).
We first examined use of any HC (oral contraceptive pills, injectables, or implants), then examined each method separately to assess for the possibility of different biological effects. We performed 2 sensitivity analyses to determine whether risk associated with using HC could have been biased due to initiation or discontinuation of HC by the time of HIV seroconversion. First, we performed an analysis including only women who reported use of HC both before and after HIV seroconversion (ie, women that were most likely to have been on a method at the time of HIV seroconversion), and compared them against women who reported no use of HC either before or after HIV seroconversion. Second, we excluded women in whom ascertainment of contraceptive use was more than 6 months or more than 1 year after the estimated date of seroconversion.
Finally, we dichotomized VL setpoints above or below the median (4.50 log10 copies/mL), and performed univariate and multivariate Cox regression analyses to estimate whether this variable was significantly associated with the hazards ratio for time to death. All analyses were done using STATA.SE, Version 10.1.
We identified 285 seroconverting women who provided 366 blood samples which met our criteria for setpoint estimation. Twenty samples (5%) had below detectable VLs (≤399 copies/mL). VL setpoint was determined using 1 measurement for 75% of women (n = 214), 2 measurements for 22% of women (n = 62), 3 measurements for 3% of women (n = 8), and 4 measurements for 0.4% of women (n = 1). Among women with 1 VL measurement, the majority of those VL measurements (66%) were taken between 6 months and 1 year after HIV seroconversion. A single VL measurement ascertainment date was not possible among women for whom VL setpoint was determined by averaging multiple measurements. Average VL setpoint was 4.47 log10 copies per milliliter (SD: 0.85). Average VL setpoints were similar between women with 1 measurement (4.43 log10 copies/mL) and women with multiple measurements (4.59 log10 copies/mL) [−0.16 log10 copies/mL, 95% confidence interval (CI): −0.38 to 0.07, P = 0.18].
Of the 285 women, 42 (15%) reported using HC at the survey round before HIV seroconversion. Among them, 71.4% (30 of 42) used injectables, 26.2% (11 of 42) used oral contraceptive pills, and 2.4% (1 of 42) used implants. Women who reported using HC before HIV seroconversion had a higher SES, were less likely to have ever been married, reported less condom use, pregnancy, and genital ulcer disease than HC nonusers, and were more likely to have 1 sexual partner in the past year (whereas non-HC users were more likely to report either zero or multiple partners in the past year) (Table 1). We found no univariate association of HC use with VL setpoint (P = 0.88).
No covariates were significantly associated with VL setpoint at P ≤ 0.05 in univariate linear regression, but there was a nonsignificant difference of lower VL setpoint among women with no previous births (P = 0.07) (Table 2).
In multivariate linear regression (Table 3), HC use was not significantly associated with VL setpoint (+0.11 log10 copies/mL; 95% CI: P = 0.47). Above average socioeconomic status was associated with lower VL setpoint as compared with women of poor socioeconomic status (−0.43 log10 copies/mL; P = 0.04), and women with 1 child had a significantly higher VL setpoint (+0.39 log10 copies/mL, P = 0.05) than nulliparous women. Neither OCs nor injectables were significantly associated with VL setpoint in multivariate regression (OC: +0.14 log10 copies/mL, 95% CI: −0.38 to +0.66, P = 0.59; injectable: +0.10 log10 copies/mL, 95% CI: −0.23 to +0.43, P = 0.56). Because only 1 woman reported implant use around the time of seroconversion, we did not assess this method separately.
A sensitivity analysis comparing 15 women who reported using HC both before and after HIV seroconversion against 158 women reporting no HC use either before or after HIV seroconversion did not substantially change our findings (−0.08 log10 copies/mL; 95% CI: −0.57 to +0.42, P = 0.76). Another sensitivity analysis excluding 42 women in whom ascertainment of contraceptive use was more than 1 year after estimated date of seroconversion found no association between HC use and VL setpoint (+0.10 log10 copies/mL, 95% CI: −0.22 to +0.41, P = 0.55). An analysis restricted to 48 women with less than 6 months between contraceptive ascertainment and HIV seroconversion suggested that the 3 women reporting HC use had a significantly lower VL setpoint than 45 women reporting nonuse of HC (−1.72 log10 copies/mL, 95% CI: −3.09 to −0.34, P = 0.02).
Unadjusted analyses suggested that VL setpoints above the median for this population were associated with increased hazards of mortality (Fig. 1). After controlling for age at seroconversion, current age, education, SES, marital status, involvement in a nonmarital relationship, current pregnancy, current breastfeeding, parity, and number of sexual partners in the past year, having a VL setpoint above the median remained strongly associated with increased hazards of mortality: adjHR: 2.54, 95% CI: 1.30 to 4.98, P < 0.01.
We found no evidence that use of HC before HIV seroconversion was associated with an elevated VL setpoint. This observation is consistent with a prior analysis of Rakai data which found no association between HC use and hazards of HIV disease progression or mortality.23 Thus, our findings suggest that HC use does not directly affect VL at setpoint or indirectly affect HIV disease progression via an effect on VL at setpoint.
Our findings differ from 2 previous studies conducted among sex workers or women being treated for genital tract infections.17,19 Studies involving higher risk women may be confounded by other sexually transmitted infections, such as HSV-2, possibly associated with both HC use and with HIV viremia. Our results from a general community cohort corroborate findings from a study conducted among women seeking reproductive and general health care services from facilities in Uganda and Zimbabwe.20 Unlike the latter study, we did not find that pregnancy or breastfeeding near the time of HIV seroconversion were associated with a higher VL setpoint or that younger age was associated with lower VL setpoint.
Our data suggest that higher VL setpoint is associated with more rapid progression to death, as has been previously reported.
This study has limitations. For 75% of women, VL setpoint was estimated using only 1 VL measurement. Although the majority of women had VL measurements, which occurred between 6 months and 1 year after HIV seroconversion, it is possible that for a minority of women, VL measurements were ascertained after the typical time frame used to define VL setpoint. However, VL setpoint is assumed to be relatively stable throughout the asymptomatic latent phase of HIV infection.11 Furthermore, averaging VL measurements may have resulted in a large variance for the VL setpoint estimate, which may have reduced our power to detect a difference in the VL setpoint between HC users and nonusers. However, the mean VL setpoint in our data (4.47 log10 copies/mL) was similar to that reported in 3 other studies of HC and VL setpoint in Kenya (4.46 log10 copies/mL),17 Malawi (4.45 log10 copies/mL),19 and a study in Uganda and Zimbabwe (4.20 log10 copies/mL).20 We were unable to distinguish between different formulations of OCs, which may have changed over time. Additionally, women who reported HC use at the interview before HIV seroconversion may have initiated or discontinued use of HC by the HIV seroconversion date. Although we cannot rule out misclassification of HC exposure, 2 sensitivity analyses conducted on women most likely to have been using HC at the time of HIV seroconversion suggested similar results to our main analysis, limiting the concern that HC initiation or discontinuation biased our results. Restricting the sample to include only women with less than 6 months between HC ascertainment and HIV seroconversion suggested an association between HC use and lower VL setpoint. Although these results provide some reassurance that we did not underestimate risk, the sample size for that sensitivity analysis was small and cannot be interpreted as a protective effect of HC.
In summary, this analysis of a population-based cohort contradicts findings of similar analyses based in higher-risk populations, but corroborates findings from a more generalizable population, in that HC use near the time of HIV seroconversion was not associated with a higher VL setpoint. These findings are reassuring given that HC methods are among the most effective methods of pregnancy prevention.
The authors wish to acknowledge the contributions of Heena Brahmbhatt, Michael Z. Chen, Homayoon Farzadegan, the Rakai Health Sciences Program study teams, and the Rakai Community Cohort Study cohort participants.
1. Joint United Nations Programme on HIV/AIDS (UNAIDS) and World Health Organization (WHO), 2009 AIDS epidemic update. Geneva: UNAIDS/WHO; 2009.
2. Shah IH, Say L. Maternal mortality and maternity care from 1990 to 2005: uneven but important gains. Reprod Health Matters
3. World Health Organization. World Health Statistics, 2010
. Geneva, Switzerland: World Health Organization Press; 2010.
4. Conde-Agudelo A, Belizan JM. Maternal morbidity and mortality associated with interpregnancy interval: cross sectional study. BMJ
5. DaVanzo J, Razzaque A, Rahman M, et al. The Effects of Birth Spacing on Infant and Child Mortality, Pregnancy Outcomes, and Maternal Mortality in Matlab Bangladesh
. Working Paper 198. Santa Monica, CA: RAND Corporations Publications Department; 2004.
6. Singh S, Darroch JE, Ashford LS, et al. Adding It Up: The Costs and Benefits of Investing in Family Planning and Maternal and Newborn Health
. New York, NY: Guttmacher Institute and United Nations Population Fund; 2009.
7. Sterling TR, Vlahov D, Astemborski J, et al. Initial plasma HIV-1 RNA levels and progression to AIDS in women and men. N Eng J Med
8. Lefrere JJ, Roudot-Thoraval F, Mariotti M, et al. The risk of disease progression is determined during the first year of human immunodeficiency virus type 1 infection. J Infect Dis
9. Schacker TW, Hughes JP, Shea T, et al. Biological and virologic characteristics of primary HIV infection. Ann Intern Med
10. Mellors JW, Kingsley LA, Rinaldo CR, et al. Quantitation of Hiv-1 Rna in Plasma Predicts Outcome After Seroconversion. Ann Intern Med
11. Johnston MI, Fauci AS. An HIV vaccine-evolving concepts. N Eng J Med
12. Kiwanuka N, Laeyendecker O, Robb M, et al. Effect of HIV-1 subtypes on disease progression in a cohort of incident cases infected with pure subtypes, recombinants, and multiple subtypes, Rakai, Uganda
. J Infect Dis
13. Lutalo T, Gray RH, Wawer M, et al. Survival of HIV-infected treatment-naive individuals with documented dates of seroconversion in Rakai, Uganda
. 2007;21(Suppl 6):S15-S19.
14. Collaborative Group on AIDS Incubation and HIV Survival. Time from HIV-1 seroconversion to AIDS and death before widespread use of highly-active antiretroviral therapy: a collaborative re-analysis. Collaborative Group on AIDS Incubation and HIV Survival including the CASCADE EU Concerted Action. Concerted Action on SeroConversion to AIDS and Death in Europe. Lancet
15. Phair J, Jacobson L, Detels R, et al. Acquired immune deficiency syndrome occurring within 5 years of infection with human immunodeficiency virus type-1: the Multicenter AIDS Cohort Study. J Acquir Immune Defic Syndr
16. Sagar M, Lavreys L, Baeten JM, et al. Infection with multiple human immunodeficiency virus type 1 variants is associated with faster disease progression. J Virol
17. Lavreys L, Baeten JM, Kreiss JK, et al. Injectable contraceptive use and genital ulcer disease during the early phase of HIV-1 infection increase plasma virus load in women. J Infect Dis
18. Sagar M, Lavreys L, Baeten JM, et al. Identification of modifiable factors that affect the genetic diversity of the transmitted HIV-1 population. AIDS
19. Kumwenda JJ, Makanani B, Taulo F, 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
20. Morrison CS, Demers K, Kwok C, et al. Plasma and cervical viral loads among Ugandan and Zimbabwean women during acute and early HIV-1 infection. AIDS
21. Wawer MJ, Gray RH, Sewankambo NK, et al. A randomized, community trial of intensive sexually transmitted disease control for AIDS prevention, Rakai, Uganda
22. Wawer MJ, Sewankambo NK, Serwadda D, et al. Control of sexually transmitted diseases for AIDS prevention in Uganda
: a randomised community trial. Rakai Project Study Group. Lancet
23. Polis CB, Wawer MJ, Kiwanuka N, et al. Effect of hormonal contraceptive use on HIV progression in female HIV seroconverters in Rakai, Uganda
24. WHO. Antiretroviral Therapy for HIV Infection in Adults and Adolescents: Recommendations for a Public Health Approach
. Geneva, Swizerland: World Health Organization Press; 2006.
25. Wawer MJ, Serwadda D, Gray RH, et al. Trends in HIV-1 prevalence may not reflect trends in incidence in mature epidemics: data from the Rakai population-based cohort, Uganda
26. Sewankambo NK, Gray RH, Ahmad S, et al. Mortality associated with HIV infection in rural Rakai District, Uganda