Share this article on:

Determinants of Differential HIV Incidence Among Women in Three Southern African Locations

Mavedzenge, Sue Napierala MPH*; Weiss, Helen A MSc, PhD*; Montgomery, Elizabeth T MHS, PhD; Blanchard, Kelly MSc; de Bruyn, Guy MBBCh, MPH§; Ramjee, Gita PhD; Chipato, Tsungai MBChB; Padian, Nancy S MSc, PhD#; Van Der Straten, Ariane MPH, PhD

JAIDS Journal of Acquired Immune Deficiency Syndromes: September 1st, 2011 - Volume 58 - Issue 1 - p 89-99
doi: 10.1097/QAI.0b013e3182254038
Epidemiology and Prevention

Introduction: We explored factors associated with differential HIV incidence among women participating in a HIV prevention trial in Harare, Durban, and Johannesburg. The trial had shown no effect of the intervention (diaphragm and lubricant gel) on HIV incidence.

Methods: A prospective cohort analysis was conducted of trial participants followed for 12-24 months. Sociodemographic, biological, and behavioral data were collected at baseline and at quarterly visits. Factors associated with HIV incidence were estimated using multivariable Cox regression models, stratified by study location. Attributable risk was calculated from the adjusted hazard ratios (AHR).

Results: There were 309 incident HIV infections among the 4948 women in the analysis. HIV incidence was highest in Durban [6.75/100 person-years; 95% confidence interval (CI): 5.74 to 7.93], lower in Johannesburg (3.33/100 person-years; 95% CI: 2.51 to 4.44), and lowest in Harare (2.72/100 person-years; 95% CI: 2.26 to 3.26). Sexually transmitted infections were important risk factors in Harare [prevalent herpes simplex virus type 2 (HSV2) AHR = 2.56, 95% CI: 1.61 to 4.06; incident HSV2 AHR = 12.60, 95% CI: 2.13 to 21.87; Neisseria gonorrhoeae AHR = 6.82, 95% CI: 2.13 to 21.87] and in Durban (prevalent HSV2 AHR = 1.64, 95% CI: 1.07 to 2.51; N. gonorrhoeae AHR = 4.40, 95% CI: 2.07 to 9.39). In Durban, having multiple partners (adjusted odds ratio (AOR) = 1.78 95% CI: 1.11 to 2.85) and sex although a partner was under the influence of alcohol/drugs (AOR = 1.51 95% CI: 1.05 to 2.16) significantly increased risk, whereas in Johannesburg, sexual debut <16 years (AOR = 2.60 95% CI: 1.30 to 5.17) was a strong predictor of HIV acquisition.

Discussion: Important differences were seen in drivers of HIV incidence at the 3 study locations. Results from this analysis imply that targeted HIV programing could have a large impact on incident HIV infection in women, and that the most effective approach will likely vary based on knowledge of the local situation/epidemiology.

From the *Department of Infectious Disease Epidemiology, London School of Hygiene and Tropical Medicine, London, United Kingdom; †RTI International, San Francisco, CA; ‡Ibis Reproductive Health, Cambridge, MA; §Perinatal HIV Research Unit, University of the Witswatersrand, Johannesburg, South Africa; ‖HIV Prevention Research Unit, Medical Research Council, Durban, South Africa; ¶Department of Obstetrics and Gynecology, University of Zimbabwe, Harare, Zimbabwe; and #School of Public Health, UC Berkeley, Berkeley, CA.

Received for publication January 18, 2011; accepted May 18, 2011.

Supported by funding from the Bill and Melinda Gates Foundation.

Each author individually contributed to the preparation of this article and reviewed and approved the final version. S.N.M. conducted the statistical analysis and wrote the first draft of the article. H.A.W. provided study design and statistical support. E.T.M. was protocol director for the main trial. G.D.B., G.R. and T.C. were site Investigators for the main trial. N.P. was the Principal Investigator for the main trial. K.B. and A.V.D.S. were co-Principal Investigators for the main trial and provided scientific input toward the study design.

The authors have no conflicts of interest to disclose.

Correspondence to: Sue Napierala Mavedzenge, MPH, Department of Infectious Disease Epidemiology, London School of Hygiene and Tropical Medicine, Keppel Street, London WC1E 7HT, United Kingdom (e-mail:

Back to Top | Article Outline


An estimated 33.3 million people were living with HIV in 2009, with two-thirds of all HIV-infected individuals and 69% of all new infections in sub-Saharan Africa.1 Women are disproportionately affected by HIV in this region, where they comprise 61% of all infections.2 Southern Africa is the region worst affected by the HIV epidemic, with >10% HIV prevalence in 9 countries.1 Within this region, there is substantial national and subnational variation, with a decline in HIV prevalence in some districts and/or countries whilst others have relatively stable or increasing HIV prevalence.

Zimbabwe is one of the most severely affected countries, with an estimated adult (15-49 years) HIV prevalence of 14.3% in 2009.3 The HIV epidemic in this country began early, with an adult prevalence of almost 15% by 1990.4 The epidemic seems to have peaked in Zimbabwe in 1998,5 and national adult prevalence was estimated to be 26.5% in 2001, 23.2% in 2003, 19.4% in 2005, and 15.3% in 2007.5,6 Zimbabwe is the only southern Africa country with clear evidence of a decline in both prevalence and incidence of HIV.7,8 In addition to mortality, this decline has been attributed to change in sexual behavior patterns, specifically an increase in condom use and decline in number of sexual partners.7,8

The HIV epidemic erupted in South Africa in the 1990s, with antenatal survey data indicating an exponential increase in adult HIV prevalence from <1% to more than 20% from 1990 to 1998.9 A moderate but steady increase was seen from 1998 to 2003, followed by a trend toward stabilization beginning in 2004.10 Overall adult HIV prevalence in South Africa was 17.8% in 2009,1 with provincial estimates ranging from 5.3% to 25.8%.11 The HIV epidemic in Gauteng province, where Johannesburg is located, increased from <1% among antenatal clinic attendees in the early 1990s to more than 20% in the span of just 10 years.12 HIV prevalence in this province seems to have peaked around 2002 at 20.3% among adults, and a decrease to 15.2% was seen by 2008.11 KwaZulu-Natal province, whose largest city is Durban, has the highest HIV prevalence in the country. An estimated 25.8% of adults were HIV infected in 2008, up from 15.7% in 2002.11

Cross-sectional studies have been conducted in Zimbabwe and South Africa to evaluate risk factors associated with HIV,13-17 but few studies have looked at risk factors for incident HIV infection in women. A better understanding of specific factors (or combinations of factors) that contribute to the site-specific HIV incidence rates will add to our knowledge of the dynamics of HIV in these 3 locations, and can provide a nuanced understanding of local context to develop targeted interventions to reduce male to female HIV transmission. We explored factors associated with incident HIV infection among women in Harare, Zimbabwe, and Johannesburg and Durban, South Africa, respectively, to gain a better understanding of the underlying factors driving the epidemic in these 3 locations.

Back to Top | Article Outline


Between September 2003 and September 2005 women in Harare, Durban and Johannesburg were enrolled in the Methods for Improvement of Reproductive Health (MIRA) study, a randomized controlled trial to evaluate the diaphragm plus lubricant gel for the prevention of HIV acquisition (, NCT00121459). The Harare location included 2 sites: Chitungwiza, a peri-urban municipality 30 km outside Harare, and Epworth, a less urbanized suburb of Harare. The Durban location included 2 sites: Umkomaas and Bothas Hill, peri-urban and semirural areas around Durban, respectively. The Johannesburg site was located in the urban township of Soweto. Women were recruited from the general population at family planning, well-baby and general health clinics, and from community groups, using flyers and radio advertising. Approval was granted from the affiliated ethical review boards of all collaborating organizations, and all women provided written informed consent before initiating study procedures. Details of eligibility criteria and study procedures are provided elsewhere.18 Briefly, sexually active HIV-negative women aged 18-49 were randomized to receive either a diaphragm and Replens lubricant gel plus condoms or condoms alone. Exclusion criteria included sensitivity to latex, being pregnant or planning to become pregnant in the following 24 months, and having had a hysterectomy. Women were followed quarterly for 12-24 months (median 21 months) follow-up. Women received product adherence and risk-reduction counselling, treatment of curable sexually transmitted infections (STI), and resupply of study products at each visit.

Demographic data were collected at baseline, medical, and sexual history, and sexual behavior data were collected at baseline and all quarterly visits. Most demographic and medical information were collected by face-to-face interview, while sensitive information, including sexual behavior, was collected using audio computer-assisted self-interviewing (ACASI).

Testing was conducted for HIV, Chlamydia trachomatis, Neisseria gonorrhoeae, and Trichomonas vaginalis at each quarterly visit. Women positive for treatable STIs were treated based on local protocols.19 HIV testing was conducted using 2 rapid tests, Determine HIV-1/2 (Abbott Laboratories, Tokyo, Japan) and Oraquick (Orasure Technologies, Bethlehem, PA) on samples obtained by finger prick or venipuncture. Discordant and concordant positive results were confirmed using enzyme-linked immunosorbent assay (ELISA) (Vironostika, Biomerieux, Durhan, NC; BioRad, Redmond, WA; or AxSYM HIV Ag/Ab Combo, Abbott Laboratories, Abbott Park, IL). Retrospective HIV polymerase chain reaction (PCR) testing was conducted on baseline samples for women who tested positive at their first quarterly visit to exclude prevalent infections at enrolment.18 A urine sample was provided for C. trachomatis and N. gonorrhoeae testing using DNA PCR (Roche Pharmaceuticals, Branchburg, NJ), and for T. vaginalis DNA detection using an adaptation of this commercial assay, as described elsewhere.20

A blood sample was provided at baseline for syphilis testing (rapid plasma reagin and Treponema pallidum hemagglutinin, Randox Laboratories, Crumlin, United Kingdom) and for herpes simplex virus type 2 (HSV2) testing using ELISA (FOCUS Diagnostics, Cypress, CA). If HSV2 negative at baseline, HSV2 testing was conducted at the closing visit. For participants HSV2 positive at closing visit testing, retrospective testing was conducted on stored sera to determine timing of infection.21 Women were classified into 3 mutually exclusive categories: HSV-positive at baseline, HSV newly infected (incident), or HSV-negative.

Timing of HIV seroconversion was defined as halfway between the last negative visit and the first visit with a positive HIV ELISA test. Demographic variables and alcohol and drug use were measured at baseline only, and syphilis infection was measured at baseline and exit visits. All other exposure variables were assessed in a time-dependant manner at the last HIV-negative visit.

Back to Top | Article Outline

Statistical Analysis

Statistical analysis was conducted using STATA 11 (StataCorp LP, TX). Differences in distributions of categorical variables across study locations were tested using a χ2 test. Differences in distribution of continuous variables were tested using the Kruskal-Wallis test to allow for non-normal distributions.

Previous analyses showed no significant difference in HIV incidence by study arm,18 and we therefore conducted analyses for both arms combined, but stratified by study location. A sensitivity analysis controlling for study arm did not change our findings (data not shown). Women were censored midway between last HIV-negative and first HIV-positive test, or (if HIV-negative) at their exit visit. Univariable and multivariable Cox proportional hazard regression models were used to estimate hazard ratios and 95% confidence intervals (CIs) for associations with incident HIV infection. Likelihood ratio tests were used to determine the strength of associations. The assumption of proportionality was tested based on Schoenfeld residuals.22 A conceptual framework was used to model proximate and distal determinants using a hierarchical (3-stage) analysis model. Sociodemographic factors were initially included, then behavioral, and finally STI-related variables, as they may be on the causal pathway between more distal sociodemographic and behavioral factors, and HIV.23 All variables found to be associated with incident HIV infection (P ≤ 0.10) were included in multivariable analysis, and those significant at the P ≤ 0.10 level at each hierarchical stage were retained in our final model. Population attributable fraction (PAF) of HIV acquisition was calculated for predictors in the final model using adjusted hazard ratios (AHR).24

Back to Top | Article Outline


Of 5045 women randomized into the MIRA trial, 97 women (1.9%) were not included in our analysis (19 were found to be HIV infected at baseline, 6 were discontinued as they did not meet other eligibility criteria, and 72 had no postenrollment HIV data). The remaining 4948 women contributed data to the HIV incidence endpoint (2455 from Harare, 1485 from Durban, 1008 from Johannesburg). There were 309 incident HIV infections in 7799 person-years at risk, including 114 infections in 4197 person-years in Harare, 148 infections in 2193 person-years in Durban, and 47 infections in 1409 person-years in Johannesburg. The overall HIV incidence rate was 3.96/100 person-years (95% CI: 3.54 to 4.43), with rates of 2.72/100 person-years (95% CI:2.26 to 3.26), 6.75/100 person-years (95% CI:5.74 to 7.93), and 3.33/100 person-years (95% CI:2.51 to 4.44) in Harare, Durban, and Johannesburg, respectively.

Back to Top | Article Outline

Descriptive Analyses

Many baseline characteristics differed between sites (Table 1). Women in Harare were more likely to live with their partner, be employed, and have more live births. They were less likely to consume alcohol/drugs and more likely to wipe inside their vagina. Overall, they had later sexual debut and fewer partners, but reported more sex in exchange for money, food, drugs, or shelter in the past 3 months (transactional sex) compared with the other locations. In Durban, women tended to be less educated and to have had earlier sexual debut than women in Harare and Johannesburg. In Johannesburg, women were more likely to have consumed alcohol in the past 3 months, reported multiple sexual partners, and reported more sex while under the influence of alcohol/drugs. There was a higher burden of nonviral STI at baseline in Johannesburg, and women were more likely to report that a partner had tested positive for HIV compared with Harare or Durban.





Back to Top | Article Outline

Individual Analyses

Table 2 shows univariable associations with incident HIV infection. The only factor associated with HIV incidence across all 3 locations was having >1 partner in the past 3 months. Younger age, lifetime number of partners, partner having other partners, sex while under the influence of alcohol/drugs, prevalent HSV2, and N. gonorrhoeae were associated with incident HIV in both Harare and Durban. Wiping inside the vagina was associated with HIV in both Harare and Johannesburg. Living with a partner was protective, and sexual debut <16 years and C. trachomatis increased risk of HIV incidence in both Durban and Johannesburg. Additionally, there were factors associated with incident HIV unique to each location as follows: alcohol consumption, condom use, incident HSV2, testing positive for any nonviral STI, syphilis infection, and genital sores/ulcers in Harare; being non-Christian, condoms as contraceptive method, sex while under the influence of alcohol or drugs, and ever using male condoms in Durban; and transactional sex and a partner having spent ≥1 month away from home in the previous year in Johannesburg.





Because different factors were significant by location, we conducted separate multivariable models for each location, adjusted for the variables in bold (Table 3). The results which are not bolded are presented to enable comparison of exposures across locations and are adjusted for the appropriate (bolded) confounding factors in each location. In multivariable analysis, no exposures were associated with HIV incidence across all locations. Independent factors associated with HIV incidence in Harare included the following: younger age, alcohol consumption at baseline, wiping inside the vagina, HSV2 (prevalent and incident), N. gonorrhoeae, and syphilis infection at baseline. In Durban, younger age, living with a partner, >1 lifetime partners, >1 partner in the past 3 months, sex while a partner was under the influence of alcohol/drugs, condoms as contraceptive method, prevalent HSV2, and incident N. gonorrhoeae were significant risk factors. In Johannesburg, factors associated with incident HIV included living with a partner, wiping inside the vagina, sexual debut <16 years, and transactional sex. Adjusting for age in Johannesburg had little impact on effect estimates of the multivariable model.



Back to Top | Article Outline

Attributable Fractions

Table 4 presents the attributable fraction of HIV cases for behavioral and reproductive health risk factors identified at each location. In Harare, we estimated that 42.4% of incident HIV infections were attributable to prevalent HSV2 infection and 20.2% to wiping inside the vagina. In Durban, we estimated that 36.1% of HIV infections were attributed to having >1 lifetime partner, 29.3% to prevalent HSV2 infection, 19.4% to using condoms as contraceptive method, and 11.9% to having sex while a partner was under the influence of alcohol/drugs. In Johannesburg, an estimated 22.3% of HIV infections were attributable to wiping inside the vagina, and 14.8% were attributable to sexual debut at <16 years.



Back to Top | Article Outline


In this study we explored factors associated with incident HIV infection among sexually active women in Harare, Durban, and Johannesburg. We compared the distribution of risk factors among women in these locations, which are at different epidemic phases, and saw different patterns of risk factors for HIV at each site.

A striking feature of the Zimbabwe sample was that more than 96% of women in this study were living with their partner, median lifetime number of partner was 1.3, and only 2.3% of women reported changing sexual partners during the study period (data not shown). This implies that nearly all HIV infection was acquired from the live-in partner. However, only 2.7% of women reported having had a partner test positive for HIV, and 25% did not know their partner's status. Given this, measures to increase HIV testing and counselling among couples should be a high priority, particularly in Harare.

In Harare, STIs including incident and prevalent HSV2, N. gonorrhoea and syphilis played an important role in HIV acquisition. None of the partner-related risk factors were associated with HIV after controlling for STIs. Alcohol consumption and wiping inside the vagina were associated with HIV, both of which have been associated with HIV in studies conducted in African populations.25-29 Although only 3.1% of women in Harare reported drinking alcohol at baseline, it was nevertheless strongly associated with incident HIV infection. This could be an important area for intervention in this location, particularly among sub-populations where alcohol consumption is common. Over 60% of women reported intravaginal wiping ≥1 per week in Harare. The biological plausibility for increased risk of HIV with vaginal wiping has been well described,30 and we estimated that more than 20% of HIV cases could be attributable to this practice. In a previous analysis of vaginal practices among MIRA participants, wiping inside the vagina was independently associated with decreased condom use.31 In site-stratified analysis, we found wiping inside the vagina was associated with decreased condom use in Harare (data not shown), which may also partially explain the association observed here.

In contrast to Harare, partner variables played an important role in increasing risk of HIV infection in Durban. Living with a partner, lifetime number of partners, >1 partner in the past 3 months, and sex while a partner was under the influence of alcohol/drugs were all significantly associated with HIV acquisition at this location. Calculation of the joint PAF indicated that addressing these primary partner sexual behavior risk factors (Tables 1-3) has the potential to prevent roughly half of new HIV infections among women in Durban. STI variables including prevalent HSV2 and N. gonorrhoea were risk factors for HIV in Durban. Reported condom use for contraception was also strongly associated with incident HIV in Durban, implying that in this location, condom use may be a marker of HIV risk rather than protection. Women who used condoms for contraception did not report higher levels of risky sexual behaviors measured. However they were younger and less likely to be cohabitating (data not shown), and therefore, condoms for contraception may be a proxy for not being in a stable or long-term relationship. Condoms used consistently and correctly are considered the gold standard for HIV prevention, however, their use is notoriously over-reported,32-34 and this may have occurred here.

In Johannesburg, living with a partner was highly protective, but no other partner-related variables were associated with HIV incidence. Participant risk behavior including vaginal wiping, early sexual debut (though not young age), and transactional sex elevated risk of HIV acquisition. Addressing individual behavioral risk remains important in any intervention to prevent the sexual transmission of HIV, and the effectiveness of such interventions will be dependent on how well social drivers of behavioral risk are addressed.

Previous analyses of MIRA data that combined study locations highlighted the importance of STIs on HIV acquisition.35,36 Prevalent and incident HSV2, and N. gonorrhoea, were associated with a 2.1-fold, 4.4-fold, and 6.9-fold increased risk of HIV, respectively. In this analysis stratified by location, STIs played an important role in HIV acquisition in Harare and Durban. Despite a high prevalence of HSV2 (65%), and elevated risk associated with incident HSV2 (AHR = 5.95) in Johannesburg, neither this nor other STIs played a significant role in predicting HIV acquisition in this location. Notably, the sample size in Johannesburg was smaller than other locations, and there were just 24 incident HSV2 infections here versus 81 and 46 in Harare and Durban, respectively. In a previous analysis of MIRA data, incidence of C. trachomatis was found to be nearly 3 times higher than N. gonorrhoea in all sites combined.19 C. trachomatis was a more common infection than N. gonorrhoea across locations, and although there was an association with HIV in univariable analysis in Johannesburg, it was not independently associated after adjusting for confounders. N. gonorrhoea, however, was independently associated with HIV acquisition in Harare and Durban. N. gonorrhoea has been associated with a more acute inflammatory response, and therefore, a potentially greater biological susceptibility to HIV, as compared with C. trachomatis.37,38 It has been associated with more frequent partner change as compared with C. trachomatis in other research, which may also explain this association, at least in Durban where partner change appeared more common.39-41

A major strength of this research is the longitudinal study design, allowing us to collect data on incident HIV infection and other cofactors. To address the temporal nature of the association with HIV, time-dependent cofactors were examined at the visit immediately before HIV detection. We used highly sensitive methods for measurement of biological predictors. Sensitivity of N. gonorrhoeae detection may have been improved, however, through the use of urogenital samples.42 The large sample size, with 12-24 months of follow-up per woman, provides ample statistical power to assess relationships between exposures and HIV at each location. Sexual behavior data were collected using ACASI to reduce biased reporting of sensitive information.43 Different behavioral factors were associated with HIV in univariable versus multivariable analysis in each location, which provides some confidence that we adequately controlled for these.

A limitation of this research is that it was conducted among a population of clinical trial participants, and therefore, stringent eligibility criteria were applied. Both syphilis and bacterial vaginosis have been shown to be associated with risk of HIV acquisition in previous studies.44,45 Data were not collected at regular quarterly visits for these infections, and therefore, we could not control for these in our analysis. Incidence of syphilis, however, was low, with only 8 cases detected at study exit. Partner circumcision status was evaluated in this trial, however, a large proportion of women reported not knowing whether their partner was circumcised, and these data are therefore presented in univariable analysis only. Estimates of PAF should be interpreted with caution, as they are based on the assumption of a simple causal association free from confounding and bias. Finally, although we evaluated sociodemographic, behavioral, and biological factors that could potentially influence HIV acquisition, we cannot rule out the possibility that some of the differences by site could be due to additional unmeasured factors, for example access to STI testing and treatment services in the public sector, or exposure to HIV prevention messages.

Each location had distinct profiles of sociodemographic, behavioral, and biological risk factors. On an individual level, we also found quite diverse patterns of risk factors and their relative importance in terms of attributable risk for HIV acquisition. HSV2 and bacterial STIs were important, particularly in Harare and Durban, emphasizing the importance of continued STI control for HIV prevention. Sexual behavior and alcohol use played a large role in determining risk of HIV acquisition overall and in explaining differences in incident infection between sites.

As an epidemic matures, more transmission occurs within stable partnerships, and we may see this as the epidemic continues to mature in South Africa.9,10 As the epidemic wanes, as seems to be the case in Zimbabwe,7,8 we may begin to see a different trend, where HIV transmission among young people and high risk core groups become increasingly important drivers of the epidemic.46,47 In this evaluation of risk factors for HIV acquisition, important differences were seen in drivers of HIV incidence at the 3 study locations. Results from this analysis imply that targeted HIV programming could have a large impact on incident HIV infection in women, and that the most effective approach will likely vary based on knowledge of the local situation/epidemiology.

Back to Top | Article Outline


We would like to acknowledge the MIRA study teams at UZ-UCSF, MRC Durban, RHRU in Johannesburg, Ibis Reproductive Health, and UCSF. We also thank the women who participated in the MIRA trial.

Back to Top | Article Outline


1. UNAIDS. 2010 Report on the Global AIDS Epidemic. Geneva, Switzerland: Joint United Nations Programme on HIV/AIDS (UNAIDS); 2010.
2. WHO. AIDS epidemic update: December 2007. Geneva, Switzerland: Joint United Nations Programme on HIV/AIDS (UNAIDS); World Health Organization (WHO); 2007.
3. UNAIDS. 2008 Report on the Global AIDS Epidemic. Geneva, Switzerland: Joint United Nations Programme on HIV/AIDS (UNAIDS); 2008.
4. UNAIDS/WHO. Epidemiological Fact Sheet on HIV and AIDS: Core Data on Epidemiology and Response, Zimbabwe. Geneva, Switzerland: UNAIDS/WHO Working Group on Global HIV/AIDS and STI; 2008.
5. Zimbabwe Ministry of Health and Child Welfare. National HIV Estimates Processes for 2001, 2003, 2005, and 2007. Harare, Zimbabwe: Zimbabwe Ministry of Health and Child Welfare; 2002-2008.
6. UNAIDS. 2009 AIDS Epidemic Update. Geneva, Switzerland: Joint United Nations Programme on HIV/AIDS (UNAIDS); 2009.
7. UNAIDS. Evidence for HIV Decline in Zimbabwe: a Comprehensive Review of the Epidemiological Data. Geneva, Switzerland: UNAIDS; 2005.
8. Gregson S, Gonese E, Hallett TB, et al. HIV decline in Zimbabwe due to reductions in risky sex? Evidence from a comprehensive epidemiological review. Int J Epidemiol. 2010;39:1311-1323.
9. UNAIDS/WHO. Epidemiological Fact Sheet on HIV and AIDS: Core Data on Epidemiology and Response, South Africa. Geneva, Switzerland: UNAIDS/WHO Working Group on Global HIV/AIDS and STI; 2008.
10. Republic of South Africa: Progress report on Declaration of Commitment on HIV and AIDS 2005. Republic of South Africa: UNAIDS country progress reports; 2006.
11. Shishana O, Rehle T, Simbayi LC, et al.Third South African National HIV, Behaviour and Health Survey 2008: Outcome Indicators for Assessing Mid-Term Progress on the NSP 2007-2011. Report 1: A Turning Tide Among Teenagers? Cape Town, South Africa: HSRC Press; 2008.
12. Buve A, Bishikwabo-Nsarhaza K, Mutangadura G. The spread and effect of HIV-1 infection in sub-Saharan Africa. Lancet. 2002;359:2011-2017.
13. Zuma K, Gouws E, Williams B, et al. Risk factors for HIV infection among women in Carletonville, South Africa: migration, demography and sexually transmitted diseases. Int J STD AIDS. 2003;14:814-817.
14. Mbizvo MT, Mashu A, Chipato T, et al. Trends in HIV-1 and HIV-2 prevalence and risk factors in pregnant women in Harare, Zimbabwe. Cent Afr J Med. 1996;42:14-21.
15. Coffee MP, Garnett GP, Mlilo M, et al. Patterns of movement and risk of HIV infection in rural Zimbabwe. J Infect Dis. 2005;191(suppl 1):S159-S167.
16. Gomo E, Chibatamoto PP, Chandiwana SK, et al. Risk factors for HIV infection in a rural cohort in Zimbabwe: a pilot study. Cent Afr J Med. 1997;43:350-354.
17. Boerma JT, Gregson S, Nyamukapa C, et al. Understanding the uneven spread of HIV within Africa: comparative study of biologic, behavioral, and contextual factors in rural populations in Tanzania and Zimbabwe. Sex Transm Dis. 2003;30:779-787.
18. Padian NS, van der Straten A, Ramjee G, et al. Diaphragm and lubricant gel for prevention of HIV acquisition in southern African women: a randomised controlled trial. Lancet. 2007;370:251-261.
19. Ramjee G, van der Straten A, Chipato T, et al. The diaphragm and lubricant gel for prevention of cervical sexually transmitted infections: results of a randomized controlled trial. PLoS ONE. 2008;3:e3488.
20. Van Der Pol B, Kraft CS, Williams JA. Use of an adaptation of a commercially available PCR assay aimed at diagnosis of chlamydia and gonorrhea to detect Trichomonas vaginalis in urogenital specimens. J Clin Microbiol. 2006;44:366-373.
21. de Bruyn G, Shiboski S, van der Straten A, et al. The effect of the vaginal diaphragm and lubricant gel on acquisition of HSV-2. Sex Transm Infect. 2011;87:301-305.
22. Cleves M, Gould W, Gutierrez R, et al. The Cox model: diagnostics. In: StataCorp LP, ed. An Introduction to Survival Analysis Using Stata. College Station, TX: Stata Press Publications; 2008.
23. Victora CG, Huttly SR, Fuchs SC, et al. The role of conceptual frameworks in epidemiological analysis: a hierarchical approach. Int J Epidemiol. 1997;26:224-227.
24. Bruzzi P, Green SB, Byar DP, et al. Estimating the population attributable risk for multiple risk factors using case-control data. Am J Epidemiol. 1985;122:904-914.
25. Hilber AM, Francis SC, Chersich M, et al. Intravaginal practices, vaginal infections and HIV acquisition: systematic review and meta-analysis. PLoS One. 2010;5:e9119.
26. Fisher JC, Cook PA, Kapiga SH. Alcohol use before sex and HIV risk: situational characteristics of protected and unprotected encounters among high-risk African women. Sex Transm Dis. 2010;37:571-578.
27. Fisher JC, Bang H, Kapiga SH. The association between HIV infection and alcohol use: a systematic review and meta-analysis of African studies. Sex Transm Dis. 2007;34:856-863.
28. Dixon-Mueller R. Starting young: sexual initiation and HIV prevention in early adolescence. AIDS Behav. 2009;13:100-109.
29. Leclerc-Madlala S. Age-disparate and intergenerational sex in southern Africa: the dynamics of hypervulnerability. AIDS. 2008;22(suppl 4):S17-S25.
30. Kilmarx PH, Limpakarnjanarat K, Supawitkul S, et al. Mucosal disruption due to use of a widely-distributed commercial vaginal product: potential to facilitate HIV transmission. AIDS. 1998;12:767-773.
31. van der Straten A, Cheng H, Chidanyika A, et al. Vaginal practices and associations with barrier methods and gel use among Sub-Saharan African women enrolled in an HIV prevention trial. AIDS Behav. 2010;14:590-599.
32. Agnew CR, Loving TJ. The role of social desirability in self-reported condom use attitudes and intentions. AIDS Behav. 1998;2:229-239.
33. Fenton KA, Johnson AM, McManus S, et al. Measuring sexual behaviour: methodological challenges in survey research. Sex Transm Infect. 2001;77:84-92.
34. Morisky DE, Ang A, Sneed CD. Validating the effects of social desirability on self-reported condom use behavior among commercial sex workers. AIDS Educ Prev. 2002;14:351-360.
35. Mavedzenge SN, Pol BV, Cheng H, et al. Epidemiological synergy of Trichomonas vaginalis and HIV in Zimbabwean and South African women. Sex Transm Dis. 2010;37:460-466.
36. Venkatesh KK, van der Straten A, Cheng H, et al. The relative contribution of viral and bacterial sexually transmitted infections on HIV acquisition in southern African women in the Methods for Improving Reproductive Health in Africa study. Int J STD AIDS. 2011;22:218-224.
37. Kaul R, Pettengell C, Sheth PM, et al. The genital tract immune milieu: an important determinant of HIV susceptibility and secondary transmission. J Reprod Immunol. 2007;77:32-40.
38. Holmes KK, Sparling PF, Mardh P, et al. Sexually Transmitted Diseases. 3rd ed. USA: McGraw-Hill Press; 1999.
39. Hook EW III, Reichart CA, Upchurch DM, et al. Comparative behavioral epidemiology of gonococcal and chlamydial infections among patients attending a Baltimore, Maryland, sexually transmitted disease clinic. Am J Epidemiol. 1992;136:662-672.
40. Ghani AC, Swinton J, Garnett GP. The role of sexual partnership networks in the epidemiology of gonorrhea. Sex Transm Dis. 1997;24:45-56.
41. Stoner BP, Whittington WL, Aral SO, et al.Variation in frequency of sex partner change wihtin sexually transmitted disease networks (abstract 5). Presented at: 33rd Interscience Conference on Antimicrobial Agents and Chemotherapy (New Orleans). Washington, DC: American Society for Microbiology; 1993.
42. Cook RL, Hutchison SL, Ostergaard L, et al. Systematic review: noninvasive testing for Chlamydia trachomatis and Neisseria gonorrhoeae. Ann Intern Med. 2005;142:914-925.
43. Langhaug LF, Sherr L, Cowan FM. How to improve the validity of sexual behaviour reporting: systematic review of questionnaire delivery modes in developing countries. Trop Med Int Health. 2010;15:362-381.
44. Atashili J, Poole C, Ndumbe PM, et al. Bacterial vaginosis and HIV acquisition: a meta-analysis of published studies. AIDS. 2008;22:1493-1501.
45. Rottingen JA, Cameron DW, Garnett GP. A systematic review of the epidemiologic interactions between classic sexually transmitted diseases and HIV: how much really is known? Sex Transm Dis. 2001;28:579-597.
46. Way P, Schwartlander B, Piot P. The Global Epidemiology of HIV and AIDS. Sexually Transmitted Diseases. 3rd ed. USA: McGraw-Hill; 1999.
47. Wasserheit JN, Aral SO. The dynamic topology of sexually transmitted disease epidemics: implications for prevention strategies. J Infect Dis. 1996;174(suppl 2):S201-S213.

epidemiology; HIV incidence; risk factors; southern Africa; women

© 2011 Lippincott Williams & Wilkins, Inc.