Increasing HIV prevention initiatives are focusing on “positive prevention” to reduce secondary HIV transmission from individuals who know they are HIV-infected to HIV-uninfected individuals.1-4 A central assumption of current “test and treat” strategies for HIV prevention is that HIV-infected individuals who are linked to care and treatment do not engage in sexual risk behaviors that could spread the virus to their uninfected sex partners.5,6 Reducing sexual risk behaviors, including increasing condom use and decreasing one's number of sex partners, can be effective behavioral strategies to prevent the acquisition and transmission of HIV/AIDS and other sexually transmitted infections (STIs).7,8 However, condom-based interventions, a mainstay of HIV/STI prevention, have yielded mixed results in various regional settings, and consistent usage can vary over time.9,10
The continued high prevalence and incidence of HIV in southern Africa, with an estimated incidence in young women of 5.5 per 100 woman-years,11 is likely driven by a range of sexual behaviors, including low levels of condom use and multiple sex partners.12-14 Despite emerging data that many HIV-infected adults continue to engage in sexual risk behaviors following knowledge of their HIV status,15-17 to date few studies have prospectively assessed sexual risk behaviors before and after acquisition of HIV among women in southern Africa. The few prospective studies suggest that women in this region report more consistent condom use after HIV seroconversion.18-20 More data are needed to understand risk-taking behaviors after HIV seroconversion to develop secondary prevention interventions for women in the increasingly feminized HIV epidemics of southern Africa.
In the current study, we examined changes in sexual risk behaviors before and after HIV seroconversion among southern African women enrolled in the multisite randomized controlled Methods for Improving Reproductive Health in Africa (MIRA) trial. We examined the question of whether notification of HIV status at the time of testing positive led to decreased sexual risk behaviors.
Study Setting and Participants
The MIRA study was a phase III randomized, controlled, open-label trial of the effectiveness of the diaphragm and lubricant gel to prevent heterosexual HIV/STI acquisition in women conducted at 3 study sites in southern Africa: Harare, Zimbabwe, and Durban and Johannesburg, South Africa. Detailed trial methods, procedures, and primary results have been published elsewhere.21,22 Participants were followed quarterly for 12-24 months between 2003 and 2006. All participants received a comprehensive HIV prevention package including HIV/STI pretest and post-test counseling, treatment of curable laboratory-diagnosed STIs, condom promotion, and risk reduction counseling. All women provided written informed consent at screening and enrollment into the study. The study protocol was reviewed and approved by the ethics review committees at all participating institutions.
Women were recruited from the general community and health facilities that provided services to women. Briefly, to be eligible, women had to be sexually active (≥4 sex acts per month); be aged 18-49 years; have a healthy cervix; test HIV negative; test negative for pregnancy and cervical and vaginal STIs (or treatment thereof); and be able to demonstrate the ability to correctly insert the diaphragm before randomization.
At screening, consenting women received HIV-1 counseling and testing. Women were tested for cervical and vaginal STIs and were treated if they tested positive for a curable STI. At enrollment, women completed a baseline audio computer-assisted self-interviewing (ACASI) questionnaire in their native language, including demographics and sexual behavior. Women were then randomized into 2 study groups and provided with latex diaphragm, lubricant gel, and condoms (intervention) or with condoms alone (control).
At each quarterly follow-up visit, women completed an ACASI that included questions on sexual behavior and received the following: a clinical exam when indicated, product adherence and risk reduction counseling, a resupply of male condoms, and study gel (in the intervention group only). Because the effect of the diaphragm on HIV/STI acquisition was unknown, all women were counseled to use condoms for every sex act. Use of effective contraception was encouraged, and hormonal contraceptives were provided free of charge. Women were provided with free male condoms, but not female condoms due to possible interference with diaphragm use; however, women were allowed to use female condoms if they chose to do so.
Women who seroconverted were encouraged to continue regular study participation, and these women received ongoing one-on-one psychosocial support and standard MIRA trial clinical care, including counseling, regular STI testing and treatment, provision of free hormonal contraception, and referral for treatment of opportunistic infections. The study either sponsored on-site psychosocial support groups or referred seroconverters to local agencies that offered this service. They also were referred to relevant treatment facilities in the public sector based on pre-established contacts with those institutions. Toward the end of the trial, a Standard of Care program was developed to link women who acquired HIV to long-term HIV-related care and treatment, and details have been published elsewhere.23 Trial follow-up did not vary between women who acquired HIV infection compared with women who remained seronegative.24 Correlates of HIV acquisition in this cohort can be found in an earlier publication.25
The following sociodemographic variables were assessed at screening: age, education, employment, currently living with primary male partner, and number of past live births. Self-reported contraception status at enrollment was categorized based on the most effective method reported (in increasing order of effectiveness): (1) no method/other, (2) male or female condoms, (3) progestin-only oral contraceptives/combined estrogen and progestin oral contraceptives, (4) injectables, and (5) long-acting methods (including intrauterine devices [IUDs], implants, male and female sterilization). At enrollment, the following risk behaviors were assessed: lifetime number of male partners, alcohol use in the last 3 months; we also constructed a composite index of high versus low participant risk behavior, with high-risk defined as a positive response for ≥1 of the following variables: exchanged sex for money, food, drugs, or shelter in the last 3 months; ≥2 sex partners in the last 3 months; having sex under the influence of alcohol/drugs in the last 3 months; ever injecting drugs; and ever having anal sex. We also constructed a composite index of high versus low partner risk behavior, with high-risk defined as a positive response for ≥1 of the following variables: having any sex partners known to ever test positive for HIV; suspecting or knowing that a regular partner had other sex partners in the last 3 months; ever had vaginal sex when the regular partner was under the influence of drugs or alcohol in the last 3 months; and regular partner was away from home for ≥1 month/year.
To assess HIV status at each study visit, 2 HIV rapid tests were done on whole blood samples from finger prick or venipuncture by use of Determine HIV-1/2 (Abbott Laboratories, Tokyo, Japan) and Oraquick (OraSure Technologies, Bethlehem, PA). Confirmatory laboratory testing was done for women with double-positive or inconsistent HIV rapid results with HIV enzyme-linked immunosorbent assay (Vironostika, Biomerierux, Durham, NC; BioRad, Redmond, WA; or AxSYM HIV Ag/Ab Combo assay, Abbot Laboratories, Abbot Park, IL). To confirm that participants with HIV infection identified during follow-up were negative at enrollment, we tested for viral nucleic acid by HIV DNA polymerase chain reaction (PCR) (Amplicor HIV-1 DNA v1.5, Roche Molecular Systems, Branchburg, NJ) from stored dried blood spot samples. At the South African sites, HIV RNA PCR tests (Amplicor HIV-1 MONITOR, v1.5, Roche Molecular Systems) were also done on stored plasma from the enrollment visit. The PCR based results were used in sensitivity analyses to exclude those women who were found to be HIV infected at enrollment.
Sexual Behavior Definitions
Variables that influence the spread of HIV can be divided into factors that affect transmissibility of HIV per sex act (ie, condom use, type of sex) and factors that influence the number of sex acts during which exposure to HIV may occur (ie, coital frequency).26 All of these sexual behaviors were assessed utilizing ACASI, which has been shown to be a particularly valid measure among Africans populations.27,28 Being sexually active was defined as reporting having had vaginal sex since the last quarterly visit. Women who reported to not having had vaginal sex were asked to confirm their response. Coital frequency was assessed by inquiring about the typical number of sex acts per week, and the outcome of “>4 sex acts per week” was defined based on the underlying distribution of responses. Male and female condom use were assessed separately. Compared with earlier studies, we examined a more rigorous measure of male condom use: reporting always using a condom over the past quarter “and” use at last sex act.29 Consistent condom use was defined as a positive response to both measures, which involved 2 recall periods. Women who reported discrepant responses (ie, reporting always use, but not at last sex act, or use at last sex act but inconsistent or no use during the quarter) were categorized as inconsistent condom users. Reporting any use of a female condom, having had anal sex, and having >1 sex partner were all assessed since the last quarterly visit.
To assess the effect of knowledge of HIV infection status on potential changes in sexual behaviors, we compared sexual behaviors during pre-HIV seroconversion follow-up visits to behaviors reported after HIV seroconversion; the visit at which women tested HIV positive was used in analysis as a post-seroconversion visit. Because not all women reported being sexually active, coital frequency and condom use were analyzed only among those women who reported being sexually active since the last quarterly visit. We employed generalized estimating equations with a logit link and exchangeable correlation structure to assess the impact of HIV seroconversion on sexual risk behaviors utilizing PROC GENMOD in SAS,30 which is consistent with earlier analyses examining longitudinal patterns of sexual behavior.31,32 In multivariable analyses, we assessed the following measures of sexual behavior: (1) being sexually active (Model I); (2) coital frequency in the past week defined as >4 sex acts versus 4 sex acts or less (Model II); (3) consistent male condom use (IIIa) and used any female condoms (Model IIIb); (4) having >1 sex partner (Model IV); and (5) anal sex (Model V). After introducing HIV infection based on antibody testing—the primary predictor variable—into the models, each covariate was introduced to assess potential confounding based either on a change of >10% of the nonlog transformed beta coefficient, or a priori confounders based on prior published analyses of MIRA (ie, age, study site, and study arm). Sexual behaviors were assessed as time-varying outcomes. HIV serostatus at each visit was included as a time-varying covariate (0 = HIV negative and 1 = HIV positive). The baseline status of all other sociodemographic and behavioral covariates were adjusted for in the models, namely study site, study arm, age, contraception, living with primary male partner, alcohol use in the last 3 months, number of past live births, education, employment, participant high-risk behavior, and partner high-risk behavior. We also controlled for time-dependent changes in sexual behaviors by adjusting for months enrolled in the trial.
In sensitivity analyses, we also assessed whether the study outcomes differed after excluding 19 women who had a positive test for HIV DNA PCR, HIV RNA PCR (>400 copies/uL), or both, at enrollment (ie, prevalent infection). We also assessed effect modification (ie, interaction), to understand whether the association between HIV seroconversion and sexual risk behaviors differed by (1) living with primary male partner at baseline, (2) study arm, and (3) study country. We first introduced the interaction term into the multivariable models, and those interaction terms which were significant were further examined in stratified models by the effect modifier.33,34 All analyses used STATA (STATACORP, version 10.0, College Station, TX) and SAS (SAS Institute, version 9.2, Cary, NC) software.
Characteristics of Women
Among 5039 enrolled seronegative women, 328 women acquired HIV by the end of the trial (4.0 of 100 woman-years); 1 woman who acquired HIV lacked available sexual behavior data and was excluded. These 327 women who seroconverted were the focus of this analysis. They contributed a total of 1828 quarterly follow-up visits, of which 718 (39%) were pre-HIV visits and 1110 were post-HIV visits (61%). Women were followed for a median of 6 quarterly follow-up visits or 18 months overall (range: 1-8 quarterly visits).
Table 1 describes the baseline characteristics of the 327 women who seroconverted. About half of the women (53%) were assigned to the diaphragm/gel and condoms arm (ie, intervention) and the remaining 47% to the condoms only arm (ie, controls). Over two-fifths of the women (47%) were enrolled in the Durban study site, 15% in Johannesburg, and 38% in Harare. About half of the women (51%) were ≤24 years of age and 47% lived with their primary male partners. Two-thirds of the women (64%) reported >1 lifetime male sex partner. A third of the women (32%) were classified as engaging in high-risk behavior and 81% of their male partners were classified as engaging in high-risk behavior.
Frequency of Sexual Behaviors Before and After HIV Seroconversion
Table 2 presents the frequency of each of the measured sexual risk behaviors overall at all visits and then stratified at pre-HIV seroconversion visits compared with post-HIV seroconversion visits. Overall, women reported being sexually active at 93% of visits. Women were more likely to be sexually active at visits before HIV seroconversion compared with visits after HIV seroconversion (94% vs. 91%; P = 0.002). Similarly, women were more likely to report a higher coital frequency in the last week before HIV seroconversion compared with visits after HIV seroconversion (>4 sex acts: 50% vs. 41%; P < 0.0001). Reporting consistent male condom use significantly increased at visits after HIV seroconversion compared with before (50% vs. 42%; P = 0.001), although reporting anal sex decreased (6% vs. 4%; P = 0.011). There were not significant changes in the frequency of reporting >1 sex partner and female condom use.
Figure 1 shows the changes in each of the sexual risk behaviors over time at visits after HIV seroconversion to assess whether initial changes in behavior persisted at follow-up. In general, initial decreases in sexual risk behaviors after HIV seroconversion persisted at follow-up visits. We noted statistically significant trends over time for a decrease in being sexually active (P = 0.017) and an increase in consistent condom use (P = 0.019).
Analyses Comparing Sexual Behaviors at Pre-HIV Visits Compared With Post-HIV Visits
Table 3 presents the unadjusted and adjusted analyses of the associations between HIV seroconversion and each of the sexual behavior outcomes. In multivariable analyses, women were somewhat less likely to report being sexually active at visits after HIV seroconversion compared with pre-HIV seroconversion visits [Model I; adjusted odds ratio (AOR): 0.63 (95% CI: 0.39 to 1.02)]. Women were significantly less likely to report >4 sex acts over the past week after HIV seroconversion compared with preseroconversion visits [Model II; AOR: 0.74 (95% CI: 0.60 to 0.91)]. Women were significantly more likely to report consistent condom use at visits after HIV seroconversion compared with visits before HIV infection [Model IIIa; AOR: 1.36 (95% CI: 1.11 to 1.67)]. Similarly, women were less likely to report >1 sex partner at visits after HIV seroconversion compared with visits before HIV infection [Model IV; AOR: 0.66 (95% CI: 0.48 to 0.91)] and reported anal sex less frequently [Model VI; AOR: 0.58 (95% CI: 0.36 to 0.95)]. Reporting use of any female condoms (Model IIIb) did not vary at visits before and after HIV seroconversion.
In sensitivity analyses, we assessed the robustness of the above associations between HIV seroconversion and each of the sexual behavior outcomes excluding 19 women (comprising 88 study visits) who had a positive PCR test at enrollment. These results were similar to the associations shown above, and hence are not shown. The impact of HIV status on consistent male condom use (Model IIIa) varied between those women assigned to the diaphragm/condoms intervention arm [AOR: 1.45 (95% CI: 1.15 to 1.82)] compared with those in the condoms only control arm [AOR: 1.02 (95% CI: 0.80 to 1.29)] (multiplicative interaction P value = 0.029). We did not note any significant differences in the associations between HIV seroconversion and being sexually active (Model I), coital frequency (Model II), use of female condoms (Model IIIb), having >1 sex partner (Model V), and anal sex (Model VI) by living with male partner, study arm, and study country.
This prospective study conducted among urban southern African women before and after HIV seroconversion demonstrated significant decreases in multiple sexual risk behaviors. Initial reductions in sexual risk behavior after HIV seroconversion persisted over time at follow-up visits. However, even though the decreases in sexual risk behaviors were significant, they were relatively modest (ie, <10% change in the frequency), suggesting the need for more robust HIV secondary prevention interventions. Timely notification of HIV status coupled with consistent prevention messages can contribute to reductions in sexual risk behaviors.
These findings documenting decreased sexual risk behavior after notification of HIV status are concordant with 2 similarly conducted longitudinal cohort studies following women before and after HIV seroconversion in sub-Saharan Africa18,19 and earlier data from men who have sex with men in the United States.3 A study in Zimbabwe reported similar findings in a smaller condom promotion trial conducted among married women, in which condom use increased postseroconversion despite the fact that most women were monogamous and were likely infected from their husbands.35 A recent mixed-methods study conducted in Malawi and South Africa noted a reduction in risk behaviors after acute HIV infection.36 Notification of HIV status through HIV voluntary counseling and testing has also been shown to reduce HIV risk behavior in this region.20,37,38 However, population-based and clinic-based studies in sub-Saharan Africa have shown significant levels of unprotected sex after diagnosis with HIV.17,39,40 Comparing the findings of different studies assessing sexual behavior associated with notification of positive HIV status is complicated by differing treatment and care programs, regional settings, measurement techniques used to assess outcomes, and types of study designs. The current trial took place within a care program in which women received comprehensive risk reduction counseling and free male condoms quarterly at each study visit. Toward the end of the trial, the MIRA Standard of Care program enabled women who acquired HIV during the trial to be linked to long-term HIV-related care and treatment.23
Despite reductions in sexual risk behaviors after seroconversion, we continued to document a substantial level of sexual risk behavior throughout the study, suggesting the need for further HIV secondary prevention. Possible reasons for decreasing sexual risk behavior after HIV seroconversion include awareness of the infection and the potential for transmission to male sex partners. The lack of further behavior change after seroconversion could be because the majority of women only had a single male partner throughout the duration of the trial, and hence it was likely many women acquired HIV from their primary partner. Women may have had little incentive to increase condom use or modify risky sexual practices after HIV seroconversion if their primary sex partner was already HIV infected. Data from this region have suggested that having a primary male partner can negatively impact on a woman's ability to negotiate sex and condom use.36 Earlier studies have consistently shown lower levels of consistent condom use among cohabiting or married African couples, and that a substantial number of HIV infections occur within such long-term relationships.41-43 When analyzing data separately by whether women lived with their primary partner, we did not identify significant differences in the association between HIV infection with sexual risk behaviors. Understanding precisely why sexual behaviors may change in association with HIV infection can be difficult to elucidate with epidemiological methods and separate from the benefits of being enrolled in an ongoing HIV prevention trial, further qualitative and behavioral studies are warranted to describe reasons motivating these changes.
The documented reductions in sexual risk behaviors are likely a conservative estimate as the visit at which women tested HIV positive was used as a postseroconversion visit in analyses and reported risk behaviors were higher before knowledge of HIV status. Among women who acquired HIV, we did collect neither data on CD4 cell count nor plasma viral load, and therefore we could not examine the relationship with sexual risk behaviors. We did not corroborate changes in self-reported sexual behaviors with biomarker data. The current study included women enrolled in a HIV prevention trial in which they received ongoing counseling and prevention messages, and hence these women may have been less likely to report risky sexual behaviors. There is a potential for recall bias of self-reported risk behaviors,44 which we attempted to minimize by utilizing ACASI and repeatedly measuring these variables at quarterly prescheduled follow-up visits. A number of studies have questioned the validity of using self-reported measures of sexual behaviors among African populations and have recommended utilizing ACASI and other biomarkers to assess sexual behaviors.27,28 Additional variables that are important determinants of sexual risk behavior that we were unable to assess include desire for children and primary partner HIV status. Women who already knew their male partners were HIV positive would have less incentive to reduce HIV transmission-related behaviors upon learning their own positive status. The duration of follow-up after HIV seroconversion was limited in the current study, and further data are needed to understand long-term changes in sexual behaviors after acquiring HIV. Because women in the intervention arm seemed to have increased their condom use after seroconversion, it is possible that in the absence of an intervention, the increase in condom use may not be present. Although the current study utilized data from a clinical trial with stringent inclusion criteria, such as excluding those who wanted to be pregnant, women were recruited from the general population in 3 large peri-urban centers of southern Africa. Prior studies have often utilized less generalizable subpopulations of women identified as belonging to high-risk groups.
Strengths include prospective data collection before and after HIV seroconversion, which avoids the bias that may be present when comparing different groups of HIV-infected and HIV-uninfected individuals. The current study assessed behaviors for most women for >6 months after HIV seroconversion, which allowed us to assess whether initial changes were sustained over time. Women received the same prevention and counseling services before and after HIV seroconversion, reducing potential information bias. A large sample size of women who seroconverted allowed us to examine a range of confounding variables in our multivariable models. We also assessed sexual behavior using a variety of relevant outcome measures and demonstrated consistent changes in behavior. The close frequency of follow-up visits every 3 months on average allowed us to utilize relatively short recall periods for questions about sexual behaviors.
The current prospective analysis assessing women before and after HIV seroconversion demonstrates a decrease in sexual risk behaviors following their knowledge of being HIV positive and emphasizes the need for HIV secondary prevention programs for women in southern Africa. HIV secondary prevention must be a component of test and treat strategies in which HIV testing and prevention is delivered at the population level, and those found to be infected are promptly linked to treatment and care. The favorable changes in sexual risk behaviors taking place with knowledge of HIV status could be further reinforced with test and treat programs in the hyperdemic settings of Southern Africa.
1. Kiene SM, Fisher JD, Fisher WA. Interventions in clinical settings. In: Kalichman SC, ed. Positive Prevention: Reducing HIV Transmission Among People Living With HIV/AIDS
. New York, NY: Kluwer; 2005.
2. Fisher J, Smith L. Secondary prevention of HIV infection: the current state of prevention for positives. Current Opin HIV AIDS
3. Colfax G, Buchbinder SP, Cornelisse PG, et al. Sexual risk behaviors and implications for secondary HIV transmission during and after HIV seroconversion. AIDS
4. Kelly JA, Kalichman SC. Behavioral research in HIV/AIDS primary and secondary prevention: recent advances and future directions. J Consult Clin Psychol
5. Dodd P, Garnett GP, Hallett TB. Examining the promise of HIV elimination by 'test and treat' in hyperendemic settings. AIDS
6. Granich R, Gilks CF, Dye C, et al. Universal voluntary HIV testing with immediate antiretroviral therapy as a strategy for elimination of HIV transmission: a mathematical model. Lancet
7. Coates T, Richter L, Caceres C. Behavioural strategies to reduce HIV transmission: how to make them work better. Lancet
8. Crepaz N, Lyles CM, Wolitski RJ, et al, HIV/AIDS Prevention Research Synthesis (PRS) Team. Do prevention interventions reduce HIV risk behaviours among people living with HIV? A meta-analytic review of controlled trials. AIDS
9. Foss A, Hossain M, Vickerman PT, et al. A systematic review of published evidence on intervention impact on condom use in sub-Saharan Africa and Asia. Sex Transm Infect
10. Ross D. Behavioural interventions to reduce HIV risk: what works? AIDS
. 2010;24(suppl 4):S4-S14.
11. Rehle T, Hallett TB, Shisana O, et al. A decline in new HIV infections in South Africa: estimating HIV incidence from three national HIV surveys in 2002, 2005, and 2008. PLos ONE
12. Pettifor A, Kleinschmidt I, Levin J, et al. A community-based study to examine the effect of a youth HIV prevention intervention on young people aged 15-24 in South Africa: results of the baseline survey. Trop Med Int Health
13. Eaton L, Flisher AJ, Aarø LE. Unsafe sexual behaviour in South African youth. Soc Sci Med
14. Hargreaves J, Bonell CP, Morison LA, et al. Explaining continued high HIV prevalence in South Africa: socioeconomic factors, HIV incidence and sexual behaviour change among a rural cohort, 2001-2004. AIDS
. 2007;21(suppl 7):S39-S48.
15. Matovu J, Gray RH, Makumbi F, et al. Voluntary HIV counseling and testing acceptance, sexual risk behavior and HIV incidence in Rakai, Uganda. AIDS
16. Matovu J, Gray RH, Kiwanuka N, et al. Repeat voluntary HIV counseling and testing (VCT), sexual risk behavior and HIV incidence in Rakai, Uganda. AIDS Behav
17. Bunnell R, Opio A, Musinguzi J, et al. HIV transmission risk behavior among HIV-infected adults in Uganda: results of a nationally representative survey. AIDS
18. McClelland R, Hassan WM, Lavreys L, et al. HIV-1 acquisition and disease progression are associated with decreased high-risk sexual behaviour among Kenyan female sex workers. AIDS
19. Turner A, Miller WC, Padian NS, et al. Unprotected sex following HIV testing among women in Uganda and Zimbabwe: short- and long-term comparisons with pre-test behaviour. Int J Epidemiol
20. Sherr L, Lopman B, Kakowa M, et al. Voluntary counselling and testing: uptake, impact on sexual behaviour, and HIV incidence in a rural Zimbabwean cohort. AIDS
21. Padian N, van der Straten A, Ramjee G, et al, MIRA Team. Diaphragm and lubricant gel for prevention of HIV acquisition in southern African women: a randomised controlled trial. Lancet
22. van der Straten A, Cheng H, Chidanyika A, et al, MIRA Team. Vaginal practices and associations with barrier methods and gel use among Sub-Saharan African women enrolled in an HIV prevention trial. AIDS Behav
23. Clouse K, Montgomery ET, Milford C, et al. Establishing a continuum of care between HIV prevention trials and public healthcare systems: the MIRA Standard of Care program. Clin Trials
24. Gappoo S, Montgomery ET, Gerdts C, et al, MIRA Team. Novel strategies implemented to ensure high participant retention rates in a community based HIV prevention effectiveness trial in South Africa and Zimbabwe. Contemp Clin Trials
25. Venkatesh K, 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 MIRA study. Int J STD AIDS
26. Guthrie B, de Bruyn G, Farquhar C. HIV-1-discordant couples in sub-Saharan Africa: explanations and implications for high rates of discordancy. Curr HIV Res
27. Langhaug L, 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
28. Langhaug L, Cheung YB, Pascoe SJ, et al. How you ask really matters: randomised comparison of four sexual behaviour questionnaire delivery modes in Zimbabwean youth. Sex Transm Dis
29. Van der Straten A, Shiboski, S, Montgomery ET, et al.Patterns and predictors of adherence to diaphragm use in a phase III trial in sub-Saharan Africa: a trajectory analysis. J Acquir Immune Defic Syndr
30. Allison P. Logistic Regression Using the SAS System: Theory and Application
. Cary, NC: SAS Institute; 1999.
31. Venkatesh K, de Bruyn G, Lurie MN, et al. Decreased sexual risk behavior in the era of HAART among HIV-infected urban and rural South Africans attending primary care clinics. AIDS
32. McClelland R, Graham SM, Richardson BA, et al. Treatment with antiretroviral therapy is not associated with increased sexual risk behavior in Kenyan female sex workers. AIDS
33. Szklo M, Nieto FJ. Epidemiology Beyond the Basics
. 2nd ed. Sudbury, MA: Jones and Bartlett Publishers; 2006.
34. Van Hess P, Allore, HG. Using SAS to investigate effect modification. Paper presented at: SUGI 31 Proceedings; March 26-29, 2006; San Francisco, CA.
35. Callegari L, Harper CC, van der Straten A, et al. Consistent condom use in married Zimbabwean women after a condom intervention. Sex Transm Dis
36. Pettifor A, Macphail C, Corneli A, et al. Continued high risk sexual behavior following diagnosis with acute HIV infection in South Africa and Malawi: implications for prevention. AIDS Behav
. 2010; [Epub ahead of print].
37. Weinhardt LS, Carey MP, Johnson BT, et al. Effects of HIV counseling and testing on sexual risk behavior: a meta-analytic review of published research, 1985-1997. Am J Public Health
38. The Voluntary HIV-1 Counseling and Testing Efficacy Study Group. Efficacy of voluntary HIV-1 counseling and testing in individuals and couples in Kenya, Tanzania, and Trinidad: a randomized trial. Lancet
39. Kalichman S, Ntseane D, Nthomang K, et al. Recent multiple sexual partners and HIV transmission risks among people living with HIV/AIDS in Botswana. Sex Transm Infect
40. Lurie M, Pronyk P, de Moor E, et al. Sexual behavior and reproductive health among HIV-infected patients in urban and rural South Africa. J Acquir Immune Defic Syndr
41. Hugonnet S, Mosha F, Todd J, et al. Incidence of HIV infection in stable sexual partnerships: a retrospective cohort study of 1802 couples in Mwanza Region, Tanzania. J Acquir Immune Defic Syndr
42. Dunkle K, Stephenson R, Karita E, et al. New heterosexually transmitted HIV infections in married or cohabiting couples in urban Zambia and Rwanda: an analysis of survey and clinical data. Lancet
43. Lurie M, Williams BG, Zuma K, et al. Who infects whom? HIV-1 concordance and discordance among migrant and non-migrant couples in South Africa. AIDS
44. Schroder K, Carey MP, Vanable PA. Methodological challenges in research on sexual risk behavior: II. Accuracy of self-reports. Ann Behav Med