Sub-Saharan Africa will have 28.1 million people living with HIV by the end of 2001 . Uganda was the first country in the region to acknowledge the HIV/AIDS epidemic as a serious problem, and is the first country in the region to show a decline in HIV prevalence [1,2]. Currently, it is estimated that 820 000 people are living with HIV in Uganda, of whom 51.2% are women aged 15–49 years. The vast majority of new infections (84%) in Uganda occur as a result of heterosexual transmission . Women are particularly vulnerable because of their status in this society, with lower education, higher unemployment and weaker negotiating skills within relationships [4,5].
In addition to the HIV epidemic, sub-Saharan Africa has one of the fastest growing populations. Uganda's population was estimated to be 22.2 million in 2000 with an annual growth rate of 2.5% and is expected to double in the next 25 years [6,7]. Population control has become a public health priority, and methods offered by family planning programs include hormonal contraceptives. However, some studies suggest that hormonal contraceptives may increase susceptibility to, and infectiousness of, HIV . Prospective studies of commercial sex workers in Thailand found that use of injectable contraceptives was associated with an increased risk of HIV acquisition, but no association was observed with oral contraceptives , whereas a prospective study of commercial sex workers in Kenya, reported that both injectable and oral contraceptives were associated with HIV risk . However, other prospective studies did not find statistically significant increased risks of HIV associated with hormonal contraceptives [11–15]. The results from cross-sectional studies are also contradictory; several failed to show an association between hormonal contraceptive use and HIV prevalence [16–18], while others reported an increased risk [19,20]. Hormonal contraceptives, especially those containing progestins, cause thinning of the vaginal and cervical mucosa, possibly increasing the likelihood of trauma and increasing the density of HIV target cells [21,22]. Hormonal contraceptives are also associated with cervical ectopy, which is thought to be a risk factor for HIV and other sexually transmitted infections (STD) . In HIV-infected women, hormonal contraceptive use is associated with increased shedding of HIV into the genital tract, suggesting that hormonal contraceptive users maybe more infectious [24,25]. An association between hormonal contraceptives and increased HIV acquisition or transmission is of public health concern, particularly given the need for HIV-infected women to prevent unwanted pregnancy in order to avoid transmission of HIV to their infants . The purpose of this study was to examine the association between hormonal contraception and HIV acquisition in a rural population-based cohort in Rakai District, Uganda.
This observational study used data from an on-going community-based cohort established in November 1994 for the Rakai STD Control for AIDS Prevention Trial . Rakai is a rural district in southwestern Uganda. This analysis utilizes data from six surveys conducted between November, 1994 and December, 1999. All consenting adults aged 15–59 years, resident in 56 communities on secondary roads were enrolled and followed-up at intervals of 10 months. At each survey, participants were interviewed in their homes and asked to provide biological samples for HIV and STD diagnoses. HIV status was determined by two different enzyme immunoassays (EIA) (Vironostika HIV-1, Organon Teknika, Charlotte, North Carolina, and Cambridge Biotech, Worcester, Massachusetts, USA) with Western blot confirmation of discordant EIA tests and seroconverters (HIV-1 Western Blot, Bio-Merieux-Vitek, St Louis, Missouri, USA).
At each survey visit, respondents who reported ever being sexually active were asked whether, at the time of the visit, they or their partners were currently using any of the following family planning methods: pills, injections (mainly Depo-Provera), spermicides, condoms, or an intrauterine device. Information was obtained on sexual abstinence, use of natural family planning (mainly periodic abstinence or calendar method), and lactational amenorrhea. In addition, information was gathered on sociodemographic characteristics (age, marital status, education, occupation and religion), and sexual behaviors (number of sexual partners in the preceding year, condom use in the 6 months prior to the interview and whether this condom use was consistent or inconsistent.) Study subjects were asked whether they had experienced symptoms suggestive of an STD including genital ulcer, discharge or dysuria, during the preceding intersurvey period. Free condoms were made available by the Project and other methods of contraception were available via government clinics, from private pharmacies and drug shops, and from non-governmental organizations. The Project also provided health education on HIV prevention, STD recognition and prevention, and family planning. Confidential voluntary HIV counseling and testing were provided free of charge by 20 community counselors in the client's home or other selected venues.
This analysis was restricted to sexually active, initially HIV-negative women aged 15–49 years, who were followed up at two or more study visits in order to ascertain HIV seroconversion. There were 8092 sexually active women of whom 7283 (90.0%) provided blood for HIV testing, and 6053 (83.1%) were initially HIV negative. Of the 6053 HIV-negative women, one or more follow-up blood sample to detect seroconversion was available for 5117 (84.5%). The exposures of interest were use of hormonal contraceptives (pills and injections) during each 10-months follow-up interval, ascertained by self-report at the interview following each interval of observation. Condom users were treated as a separate exposure group. If a respondent reported a different contraceptive method at a subsequent visit, she was assumed to have changed methods at mid-interval and ascribed 5 months of exposure during the interval. For this analysis, the reference group was defined as women who reported neither use of hormonal contraceptives nor use of condoms. The person-years of exposure were determined from the intervals during which respondents either used or did not use hormonal contraceptives or condoms. A study subject who seroconverted was assigned an exposure of the last method reported during the interval at which seroconversion was detected. HIV incidence per 100 person-years were estimated for each contraceptive exposure group, using the person time of use or non-use as the denominator.
Statistical analysis used STATA software package version 7.0 (Stata Corporation, College Station, Texas, USA). Differences in exposed and non-exposed groups were analyzed using Mantel–Haenszel chi-square method for univariate and bivariate stratification. Poisson multivariate regression was used to estimate adjusted incidence rate ratios (IRR) and 95% confidence intervals (CI) of HIV seroconversion associated with hormonal contraceptive use, relative to non-users of hormonal contraception or condoms, after adjustment for covariates. Potential confounders included in the models were age, marital status, education, reported number of sexual partners in the preceding 12 months, and reported genital ulcer disease in the intersurvey period. The number of sex partners and genital ulcer disease were considered as time-dependent covariates that could change from one survey interval to another, while age, marital status, and education level were assumed as fixed covariates at baseline. Reported condom use in the 6 months prior to a follow-up visit was used in the analysis as a time-dependent measure of condom use for STD/HIV prevention. To assess which covariates might affect the IRR values associated with hormonal contraception, models were constructed with individual variables and stepwise regression models. Since each individual woman could provide more than one observation in these analyses, the Huber–White sandwich estimator of variance was used to adjust for the clustering at individual level in the estimation of the 95% CI in all the models .
Over the duration of the study, 5117 HIV-negative eligible women (3714 recruited in 1994–1995 and the remainder in subsequent survey visits), provided 13 654 person years of observation. The median duration of follow-up was 2.6 years (mean ± SD, 2.8 ± 1.4). The median age was 25.0 years (mean, 27.8 ± 9.0). Table 1 shows the characteristics of the study population and use of hormonal contraception. Exclusive use of oral contraception was reported by 325 women (6.3%), and exclusive use of injectable contraception was reported by 539 women (10.5%), with 96 women (1.9%) reporting use of both hormonal methods at different times during observation. Therefore, a total of 960 women (18.8%) used some hormonal contraception during observation, of whom 8.3% used oral contraception and 12.4% used injectable contraception at some time during observation.
Hormonal contraceptive use was highest in the 20–24 and 25–34 age groups (19.9% and 23.5%, respectively) and was lowest in adolescents and older women (approximately 10% each). Married women were more likely to report current use of hormonal contraceptives (19.1%) than the divorced, separated, widowed (7.5%) or the never-married women (11.5%) (P < 0.001). The proportion of women using hormonal contraceptives increased with higher levels of education (P < 0.001), and those reporting multiple sex partners were more likely to use hormonal contraception (21.6%) than women with one partner (16.3%) (P < 0.05). Hormonal contraception was more frequently reported by women who reported using condoms during at least one study visit (22.8%) compared with women who reported no condom use (14.7%) (P < 0.001). There was no statistically significant difference in hormonal contraceptive use among women who did or did not report genital ulcer disease.
There were 202 HIV incident cases, of whom 159 (78.7%) were not using any hormonal contraception or condoms at time of seroconversion, 12 (5.9%) were using the pill, 16 (7.9%) injectable contraception and 15 (7.4%) used condoms. Table 2 shows HIV incidence rates and unadjusted IRR values for hormonal contraceptive users relative to non-users of hormonal contraceptives or condoms. The overall HIV incidence was 1.5/100 person-years (95% CI, 1.3–1.70). HIV incidence in non-hormonal contraceptive users was 1.5/100 person-years, while incidence in hormonal contraceptives users was 2.3/100 person-years (unadjusted IRR, 1.56; 95% CI, 1.00–2.33). HIV incidence was higher in the younger age groups; in women who were divorced/separated or widowed; and in those reporting multiple sex partners, condom use or genital ulcer disease. The IRR values for hormonal contraceptive users relative to non-users were elevated for those under age 24 years, but this was not statistically significant and was not observed among older women. The HIV IRR values associated with hormonal contraceptive use were also elevated, but not significantly, in all the women assessed, in the group of women who reported one or more sex partners, and among condom users. The IRR of HIV seroconversion associated with hormonal contraceptive use were increased in respondents who reported genital ulcer disease in the preceding 12 months.
Use of oral contraceptive pills was associated with an HIV IRR of 1.70 (95% CI, 0.85–3.04) while use of injectable contraception was associated with an IRR of 1.47 (95% CI, 0.82–2.45).
The Poisson multivariate regression adjusted IRR for any hormonal contraceptive use was 0.94 (95% CI, 0.53–1.64), and the adjusted IRR for oral contraceptive use and injectable contraception were 1.12 and 0.84, respectively (Table 3). None of these estimates was statistically significant. Age, education and condom use were not significantly associated with incident HIV after adjustment. However, HIV risks were significantly increased in women who reported marital disruption (IRR, 2.17; 95% CI, 1.06–4.43), multiple sex partners (IRR, 3.11; 95% CI, 1.50–6.43) and genital ulcer disease (IRR, 3.00; 95% CI, 1.94–4.65). The IRR was reduced from a crude estimate of 1.56 to 0.94 after adjustment (Tables 2 and 3), suggesting substantial confounding. Therefore, separate models including each of these potential confounding variables were constructed to assess which covariates were most likely to have confounded the estimates for hormonal contraception. The IRR for hormonal contraception was reduced to 1.35 (95% CI, 0.86–2.11) in a model that included only marital status, to 0.95 (95% CI, 0.56–1.62) in a model with number of sex partners, and to 1.05 (94% CI, 0.62–1.79) in a model that only included genital ulcer disease. Using a stepwise regression, the only variables retained were marital disruption (IRR, 2.00; 95% CI, 1.24–3.21), number of sex partners (IRR, 3.16; 95% CI, 1.71–5.85), genital ulcer disease (IRR, 3.19; 95% CI, 2.12–4.81) and age group 35–49 (IRR, 0.67; 95% CI, 0.44–1.00). Assessment of potential interaction between hormonal contraceptive use and any of the other variables did not show any statistically significant results.
Over the duration of the study, 22.8% (1161/5102) of respondents reported using condoms at one or more surveys. Inconsistent condom use was reported by 15.9% of respondents (811/5102), and consistent use was reported by 6.8% (350/5102). Those who reported condom use were more likely to be young (15–24 years), unmarried, more educated and have two or more sex partners in the past year. There were 657.4 person-years of observation accrued for condom use and 15 seroconversions, giving an incidence rate of 2.3/100 person-years, and an unadjusted IRR of 1.52 (95% CI, 0.83–2.57) relative to respondents who reported neither condom nor hormonal contraception use. There were no seroconversions among the 350 women who reported consistent condom use. The incidence rate for inconsistent condom users was 4.6/100 person-years (15/328.2) giving an IRR of 3.6 (95% CI, 1.97–6.23) compared with the reference group of women who reported neither condom nor hormonal contraception use.
This is the first prospective study to examine the association between hormonal contraceptives and HIV acquisition in a general population; no significant association was observed between hormonal contraceptive use and the risk of HIV acquisition in either unadjusted or adjusted analyses (Tables 2 and 3). It is noteworthy that the unadjusted IRR was 1.56, and that this was reduced to 0.94 after adjustment (39.7% decrease), suggesting substantial behavioral confounding between hormonal contraception and higher risk behaviors.
No increased risk of HIV was found to be associated with oral contraception (IRR, 1.12; 95% CI, 0.48–2.56). Previous prospective studies that had reported increased risks of HIV among oral contraceptive users were conducted among commercial sex workers [8,10], who have a higher frequency of contraceptive use, high turnover of sexual partners and more frequent contacts with high-risk sexual partners than the general population. However, one prospective study in Kenyan prostitutes found no statistically significant association between use of oral contraceptives and HIV acquisition  and a prospective study of family planning attendees in Tanzania reported similar findings . Case–control studies in prostitutes have shown similar results . Although these observational studies have limitations, the majority suggest that oral contraceptives are not associated with an increased risk of HIV acquisition.
Use of injectable contraception was not associated with HIV seroconversion (adjusted IRR, 0.84; 95% CI, 0.41–1.72) in the present study, which is consistent with a prospective study of family planning attendees in Tanzania  and a cross-sectional study in Kenya . Two studies that found a statistically significant increase in risk of HIV acquisition associated with use of injectable contraception were conducted among commercial sex workers in Thailand  and Kenya . It is likely that these investigations of high-risk women may have been affected by uncontrolled confounding. Our data also suggest confounding between injectable contraception and higher risk behaviors. For example, the unadjusted IRR was 1.47, but after multivariate adjustment the IRR was 0.84 (a 43.2% reduction), suggesting that the model controlled for differential risk behaviors.
We found no overall protective effect against HIV acquisition in women who reported condom use, largely because of the inconsistent nature of condom use. This is supported by other studies, which suggested that only consistent condom use is protective against HIV and STD .
This observational study is subject to limitations. The women self-selected to use hormonal contraceptives and these women may have different risk behaviors to those in women who chose not to use hormonal contraceptives. For example, hormonal contraception was significantly more common in currently married women, and in women reporting multiple partners (Table 1), and hormonal contraceptive users reported higher rates of genital ulcer disease. The regressions that included only hormonal contraception and each potentially confounding covariate, as well as the stepwise regression, suggest that marital status, multiple sex partners and genital ulcer disease are associated with HIV risk (with IRR values greater than twofold), and confounded the estimates for hormonal contraception. The multivariate adjustment in the Poisson models (Table 3) appears to have compensated for such differentials and resulted in non-significant estimates of risk associated with hormonal contraception. Ascertainment of hormonal contraceptive exposure was based on self-reported use and could not be verified from medical records. However, our findings with respect to HIV risk were in agreement with a Kenyan study that used clinic records to assess contraceptive exposure , which suggests that self-reported exposure ascertainment was unlikely to have resulted in misclassification bias. In addition, no seroconversions were observed among women who reported consistent condom use at each survey visit, which suggests that respondents reliably reported condom practice. Another problem related to exposure assignment is the assumption that there was no carry-over effect of hormonal contraceptives after the cessation of use. The hormonal exposure was assigned by current contraceptive use at the last visit at which respondent was confirmed to be HIV negative. If the effects of hormones persisted for some time after the cessation of use, then our exposure assignment may underestimate the effect of the hormonal contraceptives.
In conclusion, our data and those of other investigations suggest that hormonal contraceptive use is not associated with increased HIV acquisition, and that these methods of family planning can be promoted as effective contraception even in areas with high HIV prevalence. In Rakai, the prevalence of hormonal contraceptive use, particularly injectables, has increased over time . Family planning providers should be trained to provide HIV/AIDS counseling to clients and should consider promotion of both hormonal methods in conjunction with condoms. Also, HIV counselors should be encouraged to provide family planning services or referral.
The authors would like to thank all members of the Rakai Project Study for their contribution to this study and S. K. Sempala, the Director of Uganda Virus Research Institute, for his support.
Sponsorship: Supported by grants RO1 A134826 and RO1 A13426S from the National Institute of Allergy and Infectious Diseases; grant 5P30HD06826 from the National Institute of Child Health and Development; the World Bank STI Project, Uganda; and The Bill and Melinda Gates Institute for Population and Reproductive Health.
1.WHO/UNAIDS. Global situation of HIV/AIDS pandemic, end 2001.Weekly Epidemiol Rec
2.WHO/UNAIDS. Global Estimates of the HIV Epidemic as of end 1999.
[Part of the Report on the Global HIV
/AIDS Epidemic.] Geneva: UNAIDS; 2000. http://www.unaids.org/epidemic_update/
report/glo_estim.pdf Downloaded 14 April 2000.
3.Ministry of Health, Uganda. HIV/AIDS Surveillance Report. STD/ACP
; March 1998.
4.Nyanzi S, Pool R, Kinsman J. The negotiation of sexual relationships among school pupils in southwestern Uganda.AIDS Care
5.Quigley MA, Morgan D, Malamba SS, Mayanja B, Okongo MJ, Carpenter LM, et al.Case-control study of risk factors for incident HIV infection in rural Uganda.J Acquir Immune Defic Syndr
6.Statistics Bureau, Ministry of Finance and Economic Planning, Uganda. Statistical Abstract of Uganda
7.United States Census Bureau, US Department of Commerce. IDB Population Pyramids.http://www.census.gov/ipc/www/idbpyr
. html Downloaded 14 April 2001.
8.Stephenson JM. Systematic review of hormonal contraception and risk of HIV transmission: when to resist meta-analysis.AIDS
9.Ungchusak K, Rehle T, Thammapornpilap P, Spiegelman D, Brinkmann U, Siraprapasiri T. Determinants of HIV infection among female commercial workers in north eastern Thailand: results from a longitudinal study.J Acquir Immun Defic Syndr Hum Retrovirol
10.Martin HL, Nyange PM, Jackson DJ, Mandaliya K, Holmes KK, Ngugi E, et al
. Risk factors for HIV seroconversion in commercial sex workers in Mombasa, Kenya: role of hormonal contraception and STDs.X International Conference on AIDS
. Yokohama, August 1994 [abstract 394C].
11.Laga M, Manoka A, Kivuvu M, Malele B, Tuliza M, Nzila N, et al
. Non-ulcerative sexually transmitted diseases as risk factors for HIV-1 transmission in women: results from a cohort study.AIDS
12.Bulterys M, Chao A, Habimana P, Dushimimana A, Nawrocki P, Saah A. Incident HIV-1 infection in a cohort of young women in Butare, Rwanda.AIDS
13.Kapiga S, Lyamuya EF, Lwihula G, Hunter DJ. Sexually transmitted diseases, contraceptives use and the risk of HIV infection among women in Dar-es-Salaam, Tanzania: a prospective study.XI International Conference on AIDS.
Vancouver, July 1996 [abstract TuC563].
14.Saracco A, Musicco M, Nicolosi A, Angarano G, Arici C, Gavazzeni G, et al
. Man-to-woman sexual transmission of HIV: longitudinal study of 343 steady partners of infected men.J Acquir Immune Defic Syndr
15.de Vincenzi I,for the European Study Group on Heterosexual Transmission
. A longitudinal study of human immunodeficiency virus transmission by heterosexual partners.N Engl J Med
16.Gray JA, Dore GJ, Li Y, Supawitkul S, Effler P, Kaldor JM. HIV-1 infection among female commercial sex workers in rural Thailand.AIDS
17.Nagachita T, Duerr A, Garguillo PM, Yutabutra Y, Wannarat A, Tovanabutra S, et al
. HIV infectivity by contraceptive method from a partner study in northern Thailand.XI International Conference on AIDS.
Vancouver, July 1996 [abstract MoC572].
18.Nicolosi A, Leiti MLC, Musicco M, Arici C, Gavazzeni G, Lazzarin A. The efficiency of male-to-female and female-to-male sexual transmission of the human immunodeficiency virus: a study of 730 stable couples.Epidemiology
19.Guimares MD, Munoz A, Boschi-Pinto C,for the Rio de Janeiro Heterosexual Study Group. HIV infection among female partners of seropositive men in Brazil.Am J Epidemiol
20.Hitti J, Walker CK, Nsubuga PSJ, Grant RM, Tager IB, Mbidde EK. Oral contraceptive use and HIV infection.VIII International Conference on AIDS.
Amsterdam, July 1992 [abstract PoC4309].
21.Alexander NJ. Sexual transmission of human immunodeficiency virus: virus entry into the male and female genital tract.Fertil Steril
22.Hall R, Anderson J, Smart GA, Besser M. Ovary.
In: Fundamentals of Clinical Endocrinology.
Edited by Hall R, Besser, M. London: Pitman; 1980:282–290.
23.Jacobson DL, Peralta L, Graham NM, Zenilman J. Histologic development of cervical ectopy: relationship to reproductive hormones.Sex Transm Dis
24.Mostad SB, Kreiss JK, Ryncarz AJ, Mandaliya K, Chohan B, Ndinya-Achola J, et al
. Cervical shedding of herpes simplex virus in human immunodeficiency virus-infected women: effects of hormonal contraception, pregnancy, and vitamin A deficiency.J Infect Dis
25.Duerr A, Curtis K, Shelton JD, Meirik O. Hormonal contraception and genital-tract shedding of HIV-1-infected cells.Lancet
26.Lert F. Advances in HIV treatment and prevention: should treatment optimism lead to prevention pessimism?AIDS Care
27.Wawer MJ, Sewankabo NK, Serwada D, Quinn TC, Paxton LA, Kiwanuka N, et al
. Control of sexually transmitted diseases for AIDS prevention in Uganda: a randomized community trial.Lancet
28.Diggle P, Liang KY, Zeger SL. Analysis of Longitudinal Data.
Oxford: Oxford University Press; 1994.
29.Martin HL Jr, Nyange PM, Richardson BA, Lavreys L, Mandaliya K, Jackson DJ, et al
. Hormonal contraception, sexually transmitted diseases, and risk of heterosexual transmission of human immunodeficiency virus type 1.J Infect Dis
30.Taneepanichskul, S, Phuapradit W, Chaturachinda K. Association of contraceptives and HIV-1 infection in Thai female commercial sex workers.Aust N Z J Obstet Gynaecol
31.Mati JKG, Hunter DJ, Maggwa BN, Tukei PM. Contraceptive use and the risk of HIV infection in Nairobi, Kenya.Int J Gynaecol Obstet
32.Detels R, Visscher BR, Jacobson LP, Kingsley LA, Chmiel JS, Eldred LJ, et al
. Sexual activity, condom use, and HIV-1 seroconversion.
In AIDS and Sex: An Integrated Biomedical and Biobehavioral Approach.
Edited by Voeller BR, Reinisch JM, Gottlieb M. New York: Oxford University Press; 1990:13–19.
33.Lutalo T, Kidugavu M, Wawer MJ, Serwadda D, Zabin LS, Gray RH. Trends and determinants of contraceptive use in Rakai District, Uganda, 1995–98.Stud Fam Plann
The Rakai Project Study members are Noah Kiwanuka, Godfrey Kigozi, Fred Nalugoda, Michael Chen, Joseph Mabirizi and George E. Mondo.