IJE Advance Access originally published online on December 14, 2006
International Journal of Epidemiology 2007 36(1):166-174; doi:10.1093/ije/dyl251
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Prospective study of hormonal contraception and women's risk of HIV infection in South Africa
1 Infectious Diseases Epidemiology Unit, School of Public Health & Family Medicine, University of Cape Town, South Africa.
2 Department of Epidemiology, Mailman School of Public Health, Columbia University, New York, USA.
3 Department of Obstetrics & Gynaecology, University of Cape Town, South Africa.
4 Department of Pathology, College of Physicians & Surgeons, Columbia University, New York, USA.
5 Sergievsky Center, College of Physicians & Surgeons, Columbia University, New York, USA.
* CorrespondIing author. Infectious Diseases Epidemiology Unit, School of Public Health & Family Medicine, University of Cape Town, Anzio Road, Observatory 7925, Cape Town, South Africa. E-mail: lmyer{at}cormack.uct.ac.za
 |
Abstract |
|---|
 Top Abstract IntroductionMethodsResultsDiscussionAcknowledgementsReferences |
|---|
Background Many women using hormonal contraceptives are also at risk of sexually transmitted HIV infection, but data are mixed on whether hormonal contraception increases women's risk of HIV. We investigated associations between HIV incidence and use of combined oral contraceptives (COC), norethindrone enanthate (NET-EN) or depot medroxyprogesterone acetate (DMPA) in a cohort of South African women.
Methods Participants were 4200 HIV-negative women aged 35–49 years enrolled into a cervical cancer screening trial. At enrollment, women were tested for sexually transmitted infections and reported on their sexual behaviour and contraceptive use. During the 24 months of follow-up, women reported on their sexual behaviours and contraceptive use and underwent repeat HIV testing.
Results During the 5010 person-years of follow-up, 111 incident HIV infections were observed (HIV incidence, 2.2 infections/100 person-years). At enrollment, 21% of women reported using hormonal contraception, primarily DMPA (14% of all women) or NET-EN (5%). After adjusting for sexual risk behaviours and sexually transmitted infections, the incidence of HIV was similar among women using COC, NET-EN or DMPA compared with women not using any hormonal method [incidence rate ratios and 95% confidence intervals, 0.65, 0.16–2.66; 0.79, 0.31–2.02 and 0.96, 0.58–1.59, respectively]. There was also no association between increased duration of DMPA use and HIV incidence (P-value for trend, 0.51).
Conclusions These findings contribute to the evidence from general population cohorts of women that hormonal contraceptive use is not associated with increased risk of HIV acquisition. Nonetheless, family planning services are an important venue for HIV prevention activities.
Keywords Human immunodeficiency virus (HIV), contraception, risk factors, HIV prevention, women's health, South Africa
Accepted 21 October 2006
|
Introduction |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionAcknowledgementsReferences |
|---|
With an estimated 17 million women globally currently infected with HIV, and 1.5 million incident infections occurring among women annually,1 identification of strategies to reduce sexual transmission of HIV presents an ongoing challenge. The possibility that hormonal contraceptives may increase women's risk of infection is generating increasing concern.2,3 Over 100 million women use hormonal methods of contraception, including combined oral contraceptives (COC) and the injectable progestin-only methods norethindrone enanthate (NET-EN) or depot medroxyprogesterone acetate (DMPA).4 Use of hormonal contraception is common in several countries where HIV is prevalent. In South Africa, where the national antenatal HIV seroprevalence was 30% in 2004,5 an estimated 43% of women aged 15–49 years use hormonal contraceptives (30% injectable methods and 13% COC).6 In this and many other settings, definitive evidence that hormonal contraception increases male-to-female HIV transmission would have major implications for HIV prevention and family planning programmes.7
There are plausible biological mechanisms that may underlie such an association. For progestin-only methods such as DMPA and NET-EN, long-term use may cause thinning of cervico–vaginal mucosa;8 although the clinical significance of this thinning remains uncertain, it may lead to abrasions that facilitate access of HIV to CD4+ target cells in submucosal tissue. This possibility is supported by primate data showing that treatment with progestins leads to decreased thickness of cervico–vaginal epithelium9 and increased simian immunodeficiency virus acquisition.10 Several studies in humans have found significant thinning of cervico–vaginal epithelium with progestin use,8,11 although the clinical relevance of observed changes is unclear and not all studies support this.12–14 Progestin-only methods may also alter dominant vaginal flora and promote bacterial vaginosis, which may in turn pre-dispose women to HIV infection.15 COC and other estrogen-containing contraceptive methods may increase cervical ectopy (the proportion of the cervix covered by simple columnar epithelium),16–18 which is thought to present a more direct route of HIV infection compared with stratified squamous vaginal epithelium.19 Several studies have suggested that women with greater cervical ectopy have increased risk of HIV acquisition.20,21
Despite these potential mechanisms, prospective epidemiological evidence for hormonal contraception's role in increasing women's risk of HIV is highly mixed. Research among sex workers in Thailand and Kenya have observed associations between use of hormonal contraceptive methods and HIV acquisition.22–26 One prospective analysis of Kenyan sex workers found that DMPA [hazard ratio (HR): 1.8; 95% confidence interval (CI): 1.4–2.4] and COC (HR: 1.5; 95% CI: 1.0–2.1) were independently associated with increased risk of HIV.25 However, several studies in non-sex worker populations have not observed these associations.27–29 For example, a large population-based cohort from Uganda found that no evidence of increased risk of HIV acquisition among women using either COC or DMPA compared with women not using any hormonal method [rate ratio (RR): 1.12: 95% CI: 0.48–2.56 and RR: 0.84: 95% CI:0.41–1.72, respectively].30 The reasons for these variations remain unclear, and may include methodological differences (in particular, in the measurement of contraception and sexual risk behaviours) as well as differences in the frequency and type of exposure to HIV infection across different populations.
Given the widespread use of hormonal contraception among sexually active women in settings where HIV is prevalent, additional data are required on the putative role of different hormonal contraceptive methods in increasing susceptibility to HIV infection. We conducted a prospective study of these associations among women enrolled into a cervical cancer screening trial in South Africa.
|
Methods |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionAcknowledgementsReferences |
|---|
This study is based on a cohort of women enrolled into a trial to evaluate cervical cancer screening approaches conducted in the African community of Khayelitsha in Cape Town, South Africa.31 Volunteers were eligible if they were aged 35–65 years, had not been screened or treated for cervical cancer or its precursors; had no evidence of cervical cancer; and were suitable for cryotherapy (e.g. had a cervix and did not have cervical polyps or prolapse). These criteria ruled out 6% of volunteers, primarily because they were unsuitable for cryotherapy. The trial randomized women to one of three approaches to cervical cancer screening and treatment (screening and immediate treatment with cryotherapy on the basis of either direct visual inspection of the cervix or human papilloma virus (HPV) DNA testing, or observation for 6 months).
At enrollment, women underwent a pelvic examination and were administered a questionnaire to collect demographic and sexual risk information. Women reported on their past and present use of COC or injectable methods, with clear distinction between the 2-month (NET-EN) and 3-month (DMPA) injectable. Samples for Neisseria gonorrhoeae and Chlamydia trachomatis were collected using endocervical cone brushes and tested using the Hybrid Capture GC/CT DNA Assay (Digene Corporation, Gaithersburg, MD, USA). Endocervical cone brush samples were also tested for HPV DNA using the Hybrid Capture II HPV DNA Assay (Digene Corporation). Microscopic assessment of wet mount specimens was performed to identify Trichomonas vaginalis. Bacterial vaginosis (BV) was assessed based on at least three of the following four signs: moderate or severe vaginal discharge; positive whiff test after application of a 10% potassium hydroxide solution; presence of clue cells on wet mount exam; and vaginal pH > 4.5.32 Blood specimens taken at enrollment and each subsequent visit were tested for HIV antibodies (Abbott Laboratories, Abbott Park, IL, USA); positive results were confirmed using a separate antibody test (Organon Teknika, Boxtel, The Netherlands).
Follow-up visits took place 6 months after enrollment for all women and again 12 and 24 months later for a subset (all with positive visual or HPV screening tests at enrollment and all who were enrolled during 2002).31 At each follow-up visit, women were re-questioned about contraceptive use and sexual behaviours in the preceding interval and provided blood samples for HIV testing. Measures for trichomonas, gonorrhoea and chlamydia were not taken during follow-up.
All participants provided written informed consent before enrollment. Ethical approval to conduct the study was granted by the University of Cape Town and Columbia University.
Participant disposition
Of the 6555 women randomized in the trial between June 2000 and December 2002, 5273 were aged 35–49 years. Of these, 718 (14%) were HIV-positive at enrollment and were excluded from this analysis. Of the remaining 4555 HIV-negative women, 4073 were followed at 6 months (89%) and 2582 were scheduled for longer follow-up with follow-up rates at 12 and 24 months of 75 and 68%, respectively. We defined loss to follow-up as failure to attend the 6-month visit for all participants, and failure to attend a visit after 6 months for those women scheduled for longer follow-up. Because 198 participants missed either 6 or 12-month follow-up visits but completed a later visit, a total of 4200 women are included in this analysis of incident HIV infection through 24 months. No woman in this analysis missed more than one follow-up visit.
Data analysis
Data were analysed using Stata version 9.0 (Stata Corporation, USA). We analysed specific hormonal contraceptive methods (NET-EN, DMPA and COC) separately with women not using any hormonal method as the reference group. Bivariate comparisons of baseline characteristics by hormonal contraceptive use employed Wilcoxon rank sum tests, Fisher's exact tests, and chi-square tests, as appropriate.
We conducted separate analyses of HIV incidence among participants followed to 6 months only and of the subset followed to 24 months. To allow for changes in method use over time, we divided follow-up of the 4200 women into 7704 intervals (0–6 months, 6–12 months and 12–24 months) and assigned a hormonal contraceptive exposure status to each interval based on interval-specific reports. Incident HIV infections were defined as a negative result followed by a positive result at the next study visit; we used the midpoint of the interval to estimate the timing of the infection.33 During follow-up, 85 participants became pregnant (2% of the cohort), all women becoming pregnant during follow-up were censored when the pregnancy was detected. All analyses were stratified by trial arm and by whether cryotherapy was received; because the findings did not differ across these strata, we report here on the full cohort.
In multivariate analysis, we modelled the association between HIV incidence and hormonal contraceptive methods adjusting for demographic characteristics, prevalent sexually transmitted infections at enrollment and sexual behaviours. To account for intra-individual correlations of the analysis intervals, we used a population-average log-linear model with an exchangeable working correlation structure and the Huber–White sandwich estimator of variance.34 Variables were included in multivariate analysis if they demonstrated appreciable crude associations with HIV infection, and were removed if they were not associated with HIV and their removal did not alter associations involving other covariates, including hormonal contraceptive use.
|
Results |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionAcknowledgementsReferences |
|---|
The median age of the 4200 HIV-negative women included in the analysis was 40 years, the median number of years of schooling was 6, and most women (72%) lived in informal housing (a marker of low socioeconomic status). Although more than three-quarters (79%) of the participants reported that they were sexually active in the month prior to enrollment, the majority (98%) reported only one sexual partner during this period.
At enrollment, 896 HIV-negative women (21%) reported current use of any form of hormonal contraception. DMPA was the most common (14%), followed by NET-EN (5%) and COC (2%). Among women reporting injectable methods at enrollment, the median duration of use was 5 years (inter quartile range 3–9 years). Among women not using any hormonal method, 36% had tubal ligation. Most women (86%) reported using an injectable hormonal method in the past.
Women reporting hormonal contraception use at enrollment were younger, and had higher levels of education (Table 1). They were also more likely to have positive HPV tests and less likely to have trichomonas or clinical signs of BV. Condom use was low at enrollment (with 1% of women reporting current condom use and 8% reporting having ever used condoms) but increased slightly during the study, with 8% of women at the 6-month follow-up visit reporting using condom use most of the time or always. At follow-up, women who used DMPA were more likely to report sexual activity [odds ratio (OK): 1.21; 95% CI: 1.08–1.37], but there was no difference in the reporting of multiple sexual partners between women using any hormonal contraceptive method and women using no method. There were no associations observed between condom use and either DMPA or NET-EN, but condom use was more common among women who reported COC use (OR: 2.07; 95% CI: 1.03–2.48).
|
A total of 5010 person-years of observation were accrued and 111 new infections were identified during follow-up (HIV incidence, 2.2 infections/100 person-years). The mean follow-up time per participants was 14.3 months; the average duration of study intervals (i.e. time between follow-up visits) was 7.8 months. There were no significant differences in loss to follow-up by demographics, sexual behaviours at enrollment or prevalent sexually transmitted infections. Women lost to follow-up had a similar distribution of hormonal contraceptive use at baseline compared with those retained in the cohort (15% of women lost to follow-up used DMPA compared to 14% of women retained, P = 0.49). Compared with women aged 35–39 years, women aged years 45–49 were approximately half as likely to report multiple sexual partners during the follow-up period [incidence rate ratio (IRR): 0.53; P = 0.05] and consistent condom use (IRR: 0.56; P < 0.001), although there were no differences in sexual risk behaviours between women aged 35–39 years and women aged 40–44 years.
Overall, 86 incident HIV infections were observed among women not using any hormonal contraceptive method compared with 18 incident infections among women using DMPA, 5 incident infections among women using NET-EN and 2 incident infections among women using COC. In unadjusted analyses (Table 2), there was no difference in HIV incidence rates between women using any form of hormonal contraception compared with women not using any form of contraception (IRR: 1.13; 95% CI: 0.72–1.76). The rate of HIV infection was highest among current users of DMPA (2.62 infections/100 person-years) compared with women using NET-EN or COC, but none of these rates were significantly different from that of women not using hormonal methods. Overall, the incidence of HIV infection increased significantly with decreasing age, to 2.88 infections/100 person-years among women aged 35–39 years. Other factors associated with incident HIV infection in crude analyses included being unmarried, consuming alcohol in the month before the enrollment interview and prevalent HPV at enrollment.
|
There was no association between increasing duration of DMPA use and risk of HIV infection (Table 3). Of 603 women using DMPA at enrollment, 15% stopped its use after enrollment, and of 3304 women not using it at enrollment, 2% initiated DMPA use during follow-up. Women who reported DMPA use during the first follow-up interval but stopped DMPA after the 6-month visit had high HIV incidence (6.23 100 person-years), but this was not significant and six of seven sero-conversions in this group occurred 12–24 months post-enrollment. In addition, women who used DMPA throughout the entire follow-up period did not have a higher incidence of HIV infection when compared with women who did not use any hormonal contraceptive during the follow-up period (age-adjusted IRR: 1.11; 95% CI: 0.65–1.87).
|
In multivariate analysis, the association between DMPA use and incident HIV was attenuated after adjustment for demographics, sexual behaviours and baseline sexually transmitted infections (Table 4). Among women of 35–49 years, DMPA users had a similar incidence rate of HIV infection as women not using hormonal contraception (IRR: 0.96; 95% CI: 0.58–1.59). When the multivariate model was restricted to women aged 35–39 years, the association remained consistent with the null (adjusted IRR: 1.35; 95% CI: 0.77–2.37). Other variables significantly associated with HIV incidence included younger age, being unmarried and reporting multiple sexual partners at follow-up. In a separate model restricted to women using injectable methods, DMPA users had a slightly higher rate of HIV infection than NET-EN users, but this association did not approach statistical significance (adjusted IRR: 1.48; 95% CI: 0.49–4.44). These results did not change when analyses were restricted to the 6-month follow-up visit (Table 4) or to women who did not report symptoms of menopause (data not shown). The null association between hormonal contraception and HIV incidence persisted when analyses were stratified by demographic characteristics (including marital status and housing type), sexual risk behaviours (including alcohol consumption, condom use, previous treatment for a sexually transmitted infection and multiple sexual partners), baseline sexually transmitted infections (including trichomonas, BV, HPV and chlamydia and/or gonorrhoea) and the arm of the trial.
|
|
Discussion |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionAcknowledgementsReferences |
|---|
These data suggest that hormonal contraception is not associated with increased risk of HIV infection among women in this setting. The observations from this sample of South African women aged 35–49 years are consistent with findings of previous population-based research in sub-Saharan Africa,30 and contribute to the evidence suggesting that hormonal contraceptive methods are not associated with increased risk of male-to-female transmission of HIV.27–29 Although the confidence intervals for NET-EN and COC are relatively wide due to small numbers of women using these methods, the upper confidence limit of the IRR associated with DMPA use (1.59) excludes the point estimates for this association observed among sex worker cohorts.23–26 In addition, these are the first data to examine associations between the 2-monthly injectable contraceptive, NET-EN and HIV risk; although data are sparse, they suggest that NET-EN does not confer increased risk of HIV infection.
Given unavoidable attrition, loss to follow-up is an important consideration. However, for loss to follow-up to mask an appreciable true association, women lost would need to be more likely to use hormonal contraceptives and to have increased risk of HIV infection. Based on a sensitivity analysis (data not shown), we estimate that women lost to follow-up would need to be four times more likely to use DMPA, and to have a 3-fold higher incidence of HIV infection, to mask a true relative risk of HIV infection of 1.5 due to DMPA use. Based on comparisons of women retained in the cohort vs those lost to follow-up, this seems implausible. However, it is important to note that the analysis has insufficient power to rule out small associations (e.g. RR < 1.5).
This is one of the few studies to examine these associations among older women (age range, 35–49 years) who are at risk of sexually transmitted HIV infection. However the age range of this cohort is somewhat older than has been the focus of previous studies which may limit its generalizability. It is possible that any changes caused by hormonal contraceptives to cervico–vaginal mucosa may be greater among younger women, and that contraceptives have little effect on the mucosa of older women who may have advanced thinning of cervico–vaginal tissues independent of exogenous hormones (due to decreased endogenous estrogens with increasing age). Although previous studies which included both older and younger women found that age did not modify the association between hormonal contraceptive use and HIV infection,30 the role of age as a modifier of this association warrants additional attention in future studies in younger populations, particularly among sexually active adolescents.
We are intrigued by the disparity between findings from general population cohorts (including these data) that typically observe no association between hormonal contraceptive use and HIV risk, and studies among female sex workers that sometimes observe significant associations. One possible explanation is systematic differences in the ability to measure and adjust for sexual behaviours, leading to residual confounding in studies of sex worker cohorts.30 Use of hormonal contraception is highly correlated with sexual risk behaviours, and thus adequate adjustment for sexual behaviours is necessary to correctly determine the true effect of contraceptive methods on HIV acquisition. Measuring sexual behaviours accurately is notoriously difficult, particularly as it is unclear which aspects of sexual activity most accurately reflect sexual exposure to HIV-infected partners.35 We hypothesize that in general population samples, routine measures of sexual risk behaviour (e.g. sexual activity, number of sexual partners, types of sexual partners and condom use), are relatively accurate proxies for increased likelihood of exposure to HIV-infected partners, thus allowing adequate statistical adjustment for this confounding. However, in studies of female sex workers, standard questions may miss salient parameters relevant to the risk of contact with HIV-infected partners. This increased error in the measurement of important confounders would result in reduced ability to adjust for confounding in studies of this association among sex workers, spuriously inflating the observed association between hormonal contraception and HIV infection.36,37 More rigorous examination of the interrelationships between contraceptive use, HIV and different sexual risk behaviours, including measures routinely used in HIV prevention research as well as measures that are unique to contexts of sex work (such as different measures of client's HIV risk), are needed to assess this possibility.
Alternatively, it may be possible that hormonal contraception requires the presence of a third factor, which may be more prevalent among sex workers, to activate deleterious effects on the risk of HIV infection. This effect modifier may be biological (including specific sexually transmitted infections) or behavioural (such as increased sexual activity). To examine this possibility, we stratified our analyses by baseline sexually transmitted infections and different sexual risk behaviours. While this exploration was limited to those factors on which measures were available, no such effect modifiers were identified. However, it is important to note that no measures of sexually transmitted infections were made during follow-up and we did not have any measures of herpes simplex virus type-2 (HSV-2) infection. HSV-2 is an established co-factor in HIV transmission which one study has suggested may be an important modifier of the effect of hormonal contraceptives on male-to-female transmission.38 In addition, the analysis of bacterial vaginosis was restricted to clinical (Amsel) measures since we do not have more definitive microbiological (Nugent) measures of BV on all participants. Previously, we reported significant associations between Nugent-defined BV and women's risk of HIV infection,15 and the links between HIV risk, abnormal vaginal flora and the use of hormonal contraception require further investigation.
In considering the systematic differences in the association between hormonal contraception and HIV in different study populations, it is important to recognize that female sex workers are more likely to be exposed regularly to HIV infection while in general population samples, levels of exposure to HIV are likely to be lower. However, this study was conducted in a high HIV prevalence setting (the local HIV sero-prevalence among antenatal clinic attenders in Khayelitsha is >20%), and thus we anticipate high levels of exposure to sexually transmitted HIV.39
The conflicting evidence regarding the potential role of hormonal contraception in increasing women's risk of HIV infection would appear to demand further epidemiological investigation. However, any true association is likely to be small, requiring large studies to rule out chance findings. Also, all observational research on this question must confront strong likelihood of confounding and, even for measured confounders, the possibility of residual confounding due to chronic issues of measurement error. The problem of distinguishing small causal associations is a significant challenge throughout epidemiology.40,41 In the case of hormonal contraception and HIV infection, it is unclear whether more definitive evidence is likely to emerge from observational epidemiological studies alone. More productive insights may come from basic research on how different forms of hormonal contraception influence potential mechanisms involved in sexual HIV transmission, such as cervical ectopy or mucosal thinning.
Although hormonal contraception does not appear to increase male-to-female HIV transmission in this cohort, it is still important to remember that most women using contraceptive methods are sexually active, and in many settings the use of male or female condoms is less likely when women are using a non-barrier contraceptive method.42,43 In this light, dual method use (combined use of condoms with a non-barrier method) should be actively promoted by family planning services in populations where HIV is prevalent.44 In addition, increasing coverage of effective contraception among HIV-infected women who do not wish to become pregnant may be one of the most cost-effective approaches to reduce paediatric HIV infections.45
In summary, we found no association between use of hormonal contraceptives and women's risk of HIV infection in this general population cohort of women aged 35–49 years in South Africa. While further research is required, there is no indication that the benefits of hormonal contraception for sexually active women should be qualified by concerns regarding increased risk of HIV infection.
|
Acknowledgements |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionAcknowledgementsReferences |
|---|
This research was funded by the Bill & Melinda Gates Foundation and EngenderHealth.
Conflict of interest: None declared.
Key Messages
|
|
Notes |
|---|
The originally published version of this article was incorrect. The first author's affiliations were not shown in full. The publisher apologizes for this error.
|
References |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionAcknowledgementsReferences |
|---|
1 UNAIDS. AIDS Epidemic Update (2005) UNAIDS/05.19E, December 2005. Geneva: UNAIDS.
2 World Health Organization. Hormonal Contraception and HIV: Science and Policy. (2005) Geneva: Department of Reproductive Health and Research, World Health Organization.
3 Morrison C, Best K. Hormonal Contraception and HIV: An Update (2004) Research Triangle Park, NC: Family Health International.
4 United Nations Population Division. World Contraceptive Use (2004) New York: UN Population Division. 2003. E.04.XIII.2.
5 Department of Health of South Africa. National HIV and Syphilis Antenatal Sero-prevalence Survey in South Africa in 2004 (2004) Pretoria: Department of Health, Directorate: Health Systems Research, Research Coordination and Epidemiology.
6 Department of Health of South Africa. South African Demographic and Health Survey 1998 (2001) Pretoria: South African Department of Health, Medical Research Council, Macro International.
7 Daly CC, Helling-Giese GE, Mati JK, Hunter DJ. Contraceptive methods and the transmission of HIV: implications for family planning. Genitourin Med (1994) 70:110–17.[ISI][Medline]
8 Ringrose CA. Effects of medroxyprogesterone acetate employed as a long-term injectable contraceptive. J Reprod Med (1972) 8:75–79.[ISI][Medline]
9 Hild-Petito S, Veazey RS, Larner JM, Reel JR, Blye RP. Effects of two progestin-only contraceptives, Depo-Provera and Norplant-II, on the vaginal epithelium of rhesus monkeys. AIDS Res Hum Retroviruses (1998) 14:s125–30.[ISI][Medline]
10 Marx PA, Spira AI, Gettie A, et al. Progesterone implants enhance SIV vaginal transmission and early virus load. Nat Med (1996) 2:1084–89.[CrossRef][ISI][Medline]
11 Miller L, Patton DL, Meier A, Thwin SS, Hooton TM, Eschenbach DA. Depomedroxyprogesterone-induced hypoestrogenism and changes in vaginal flora and epithelium. Obstet Gynecol (2000) 96:431–39.
12 Kuhn L, Denny L, Pollack AE, Wright TC. Prevalence of visible disruption of cervical epithelium and cervical ectopy in African women using Depo-Provera. Contraception (1999) 59:363–67.[CrossRef][ISI][Medline]
13 Mauck CK, Callahan MM, Baker J, et al. The effect of one injection of Depo-Provera on the human vaginal epithelium and cervical ectopy. Contraception (1999) 60:15–24.[CrossRef][ISI][Medline]
14 Bahamondes L, Trevisan M, Andrade L, et al. The effect upon the human vaginal histology of the long-term use of the injectable contraceptive Depo-Provera. Contraception (2000) 62:23–27.[CrossRef][ISI][Medline]
15 Myer L, Denny L, Telerant R, De Souza M, Wright TC, Kuhn L. Bacterial vaginosis and susceptibility to HIV infection among South African women: a nested case-control study. J Infect Dis (2005) 192:1372–80.[CrossRef][ISI][Medline]
16 Critchlow CW, Wolner-Hanssen P, Eschenbach DA, et al. Determinants of cervical ectopia and of cervicitis: age, oral contraception, specific cervical infection, smoking, and douching. Am J Obstet Gynecol (1995) 173:534–43.[CrossRef][ISI][Medline]
17 Goldacre MJ, Loudon N, Watt B, et al. Epidemiology and clinical significance of cervical erosion in women attending a family planning clinic. Br Med J (1978) 1:748–50.[ISI][Medline]
18 Jacobson DL, Peralta L, Graham NM, Zenilman J. Histologic development of cervical ectopy: relationship to reproductive hormones. Sex Transm Dis (2000) 27:252–58.[ISI][Medline]
19 Moench TR, Chipato T, Padian NS. Preventing disease by protecting the cervix: the unexplored promise of internal vaginal barrier devices. AIDS (2001) 15:1595–602.[CrossRef][ISI][Medline]
20 Plourde PJ, Pepin J, Agoki E, et al. Human immunodeficiency virus type 1 seroconversion in women with genital ulcers. J Infect Dis (1994) 170:313–17.[ISI][Medline]
21 Myer L, Wright TC Jr, Denny L, Kuhn L. Nested case-control study of cervical mucosal lesions, ectopy, and incident HIV infection among women in Cape Town, South Africa. Sex Transm Dis (2006) 33:683–87.[CrossRef][ISI][Medline]
22 Martin H, Nyange P, Richardson B, et al. Hormonal contraception, sexually transmitted diseases, and risk of heterosexual transmission of human immunodeficiency virus type 1. J Infect Dis (1998) 178:1053–59.[ISI][Medline]
23 Rehle T, Brinkmann UK, Siraprapasiri T, Coplan P, Aiemsukawat C, Ungchusak K. Risk factors of HIV-1 infection among female prostitutes in Khon Kaen, Northeast Thailand. Infection (1992) 20:328–31.[CrossRef][ISI][Medline]
24 Plummer FA, Simonsen JN, Cameron DW, et al. Cofactors in male-female sexual transmission of human immunodeficiency virus type 1. J Infect Dis (1991) 163:233–39.[ISI][Medline]
25 Lavreys L, Baeten JM, Martin HL, et al. Hormonal contraception and risk of HIV-1 acquisition: results of a 10-year prospective study. AIDS (2004) 18:695–97.[CrossRef][ISI][Medline]
26 Ungchusak K, Rehle T, Thammapornpilap P, Spiegelman D, Brinkmann U, Siraprapasiri T. Determinants of HIV infection among female commercial sex workers in northeastern Thailand: results from a longitudinal study. J Acquir Immune Defic Syndr Hum Retrovirol (1996) 15:500–07.
27 De Vincenzi I. A longitudinal study of human immunodeficiency virus transmission by heterosexual partners. European Study Group on Heterosexual Transmission of HIV. N Engl J Med (1994) 331:341–46.
28 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 (1994) 8:1585–91.[ISI][Medline]
29 Laga M, Manoka A, Kivuvu M, et al. Non-ulcerative sexually transmitted diseases as risk factors for HIV-1 transmission in women: results from a cohort study. AIDS (1993) 7:95–102.[ISI][Medline]
30 Kiddugavu M, Makumbi F, Wawer MJ, et al. Hormonal contraceptive use and HIV-1 infection in a population-based cohort in Rakai, Uganda. AIDS (2003) 17:233–40.[CrossRef][ISI][Medline]
31 Denny L, Kuhn L, De Souza M, Pollack AE, Dupree W, Wright TC. Screen-and-treat approaches for cervical cancer prevention in low-resource settings: a randomized controlled trial. JAMA (2005) 294:2173–81.
32 Amsel R, Totten PA, Spiegel CA, Chen KC, Eschenbach D, Holmes KK. Nonspecific vaginitis. Diagnostic criteria and microbial and epidemiologic associations. Am J Med (1983) 74:14–22.[CrossRef][ISI][Medline]
33 Clayton D, Hills M. Statistical Models in Epidemiology (1993) New York: Oxford University Press.
34 Diggle PJ, Liang KY, Zeger SL. Analysis of Longitudinal Data (1994) New York: Oxford University Press.
35 Myer L, Morroni C, Link BG. Impact of measurement error in the study of sexually transmitted infections. Sex Transm Infect (2004) 80:318–23.
36 Greenland S. The effect of misclassification in the presence of covariates. Am J Epidemiol (1980) 112:564–69.
37 Davey Smith G, Phillips AN. Confounding in epidemiological studies: why "independent" effects may not be all they seem. Br Med J (1992) 305:757–75.[ISI][Medline]
38 Morrison C, Richardson BA, Celentano DD, et al. The hormonal contraception and risk of HIV-1 acquisition (HC-HIV) study. (2005) Paper TW-002 presented at the International Society for Sexually Transmitted Diseases Research, July 11–13: Amsterdam, Netherlands.
39 Department of Health/Provincial Administration of the Western Cape. HIV prevalence in the Western Cape: Results of the 2002 annual provincial and district survey. (2003) Cape Town: District Health Services and Programmes, Department of Health.
40 Phillips AN, Davey Smith G. The design of prospective epidemiological studies: more subjects or better measurements? J Clin Epidemiol (1993) 46:1203–11.[CrossRef][ISI][Medline]
41 Shapiro S. Bias in the evaluation of low-magnitude associations: an empirical perspective. Am J Epidemiol (2000) 151:939–45.
42 Myer L, Mathews C, Little F. Condom use and sexual behaviors among individuals procuring free male condoms in South Africa: a prospective study. Sex Transm Dis (2002) 29:239–41.[ISI][Medline]
43 Ott MA, Adler NE, Millstein SG, Tschann JM, Ellen JM. The trade-off between hormonal contraceptives and condoms among adolescents. Perspect Sex Reprod Health (2002) 34:6–14.[CrossRef][ISI][Medline]
44 Morroni C, Smit J, McFadyen L, Mqhayi M, Beksinska M. Dual protection against sexually transmitted infections and pregnancy in South Africa. Afr J Reprod Health (2003) 7:13–19.
45 Sweat MD, O'Reilly KR, Schmid GP, Denison J, de Zoysa I. Cost-effectiveness of nevirapine to prevent mother-to-child HIV transmission in eight African countries. AIDS (2004) 18:1661–71.[CrossRef][ISI][Medline]
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  C. S Morrison Commentary: Hormonal contraception and HIV acquisition--current evidence and ongoing research needs Int. J. Epidemiol., February 1, 2007; 36(1): 175 - 177. [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||