South Africa is currently facing one of the world's worst HIV epidemics, with nearly 5 million people or more than 16% of the adult population being HIV infected.1 Understanding the factors that influence unprotected sex among HIV-positive individuals in South Africa is an important precursor to developing effective HIV prevention interventions specifically designed to help HIV-positive individuals reduce sexual risk behavior and thus help stop the spread of HIV.2,3
It is commonly believed that alcohol contributes to the occurrence of unprotected sexual behavior. Qualitative and quantitative data from around the world support an association between alcohol consumption and unprotected sexual behavior among a wide variety of populations.4-10 The majority of these studies have been conducted in the United States among presumed HIV-negative populations. However, there is a growing literature on the association between alcohol use before sex and unprotected sexual behavior in Africa and between alcohol use before sex and risk of HIV acquisition.11,12 However, few have studied the role of drinking before sex in unprotected sexual behavior among HIV-positive populations in Africa.6,13,14
Worldwide, the association between alcohol consumption and unprotected sex has been studied using different methodologies, which answer different questions regarding the nature of the ostensible relationship between alcohol consumption and unprotected sex. Two types of descriptive cross-sectional studies are commonly used: those which correlate global self-reports of overall frequency of alcohol consumption or the situational frequency of drinking before sexual intercourse with self-reported frequency of unprotected sex. These types of studies answer questions such as: “Do individuals who drink more frequently also engage in unprotected sex more frequently?” and “Do individuals who frequently drink before they have sex have more unprotected sex?” These cross-sectional methodologies do not permit the researcher to determine if drinking before sex increases the likelihood of unprotected sex or if an individual drinks equally frequently before protected and unprotected sex acts.
Often the question of greatest interest regarding the relationship between alcohol consumption and unprotected sex is: “Does drinking alcohol before sexual intercourse make people more likely to have unprotected sex?” This question implies a causal relationship between consuming alcohol before sex and subsequent unprotected sex. Although it is not ethical to conduct a controlled experimental study of the causal effects of alcohol consumption on sexual behavior, a viable alternative is a prospective cohort study using daily diary methods to collect event-level data that temporally sequence discrete instances of alcohol consumption as occurring before discrete events of unprotected sex.15 However, there have been no such studies among HIV-positive individuals in Africa. The present prospective cohort study collected event-level data using daily diary methods among a sample of HIV-positive individuals in South Africa to document the levels of unprotected sexual behavior in this population and test the hypothesis that consuming alcohol before sex increases the likelihood and number of subsequent unprotected sex acts.
Participants and Procedure
Eighty-two HIV-positive individuals (58 women and 24 men) participated in an intensive, longitudinal (42 days) structured daily phone interview, which yielded a possible 3396 data points (1 wave of data collection with 12 participants was shortened to 38 days in duration because of overlap with religious holidays). Using purposeful sampling, we recruited participants from 5 HIV service organizations in Cape Town, South Africa, between May and November 2006. Eligibility for the study required the following: (a) being older than 18 years of age, (b) being HIV positive, (c) having vaginal or anal sex in the prior 30 days, (d) consuming alcohol in the prior 30 days, and (e) having access to a phone where they could receive calls every afternoon. Only 1 member of a couple was allowed to participate. Of those screened, 5 did not meet the eligibility criteria (4 for sexual inactivity and 1 for no alcohol in prior 30 days). Two eligible individuals declined participation. The study was approved by institutional review boards in the United States and in South Africa.
Interviewers called participants each day between 1 pm and 6 pm and conducted a structured interview in Xhosa. Participants were compensated with up to 660 Rand (about $US 95, including transportation reimbursement), the amount of which depended on the number of interview days completed.
Based on our pilot work with this population and a conservative estimate of the magnitude of effects, we estimated the effect size for a difference between when alcohol is consumed and is not consumed before sex as Cohen d = 0.3 for the proportion of sex events that would be unprotected and the number of unprotected sex events that occurred each day. With this expected difference in these 2 outcomes, an α level of 0.05, a sample size of 80 participants, and 3500 unprotected sexual events would provide 0.80 power to detect a difference in our 2 outcomes that is attributable to alcohol consumption.
Each day participants reported on their drinking and sexual behavior for last night (after yesterday's interview, approximately 5 pm yesterday until going to sleep) and for today (since waking up until now). For these 2 time periods, participants reported the following: (a) how many drinks (drink sizes standardized to equate alcohol content across types of alcoholic drinks, 1 drink = 12 g of alcohol) they consumed, (b) how many times they had vaginal and anal sex, (c) how many times a condom was used during each type of sex, (d) the perceived HIV serostatus of their partner (positive, negative, or unknown), (e) the type of partner (main/steady, secondary/casual), (f) if the alcohol consumption occurred before/during or after the sexual event(s), and (g) if their partner consumed alcohol before the sexual event (yes, no). The measures were adapted from measures validated in South Africa and in daily diary studies.14,16-18 Oral sex was not assessed because of the low frequency of this behavior among HIV-positive samples in South Africa.14 The number of unprotected sex events during each time period (daytime and evening) was calculated by taking the sum of the number of vaginal and anal sex events minus the sum of the number of times a condom was used during vaginal and anal sex. The proportion of sex events that were unprotected during each time period was also calculated.
Data Analysis Approach
Generalized estimating equation (GEE) analyses using SPSS 15.0 software19 were conducted to examine the event-level association between alcohol consumption before sex and unprotected sex. Also of interest were the effects of the time-varying event-level variable of partner type (main/steady, casual/secondary) and its interaction with alcohol consumption before sex. Time-invariant sociodemographic variables of age, education (primary or less vs high school or more), marital status (unmarried vs married), socioeconomic status (self-rated on a 4-point continuum ranging from “I do not have enough money to buy food” to “I have most of the important things, but not enough money to pay for things like my children's education”),1 months known of HIV positiveness, disclosed HIV status to current partner(s) (yes vs no), HIV stigma,20,21 sexually transmitted infection (STI) in prior 6 months (yes vs no), and currently taking antiretrovirals (ARVs) (yes vs no); time-varying event-level variables of partner's perceived HIV serostatus (positive, negative, or unknown) and days in study (1-42 days) were also included in each model. Gender was included as a potential moderator of each event-level association. Alcohol before sex was operationalized as a predictor of unprotected sex both as a categorical time-varying variable representing 4 drinking categories: neither partner, only the female partner, only the male partner, or both partners drank any before sex, and as a categorical time-varying variable representing the risk level of the alcohol consumption based on the amount of alcohol the individual consumed before sex. We adopted World Health Organization categories for level of risk from specific quantities of alcohol (5 categories ranging from no risk to very high risk), with gender-adjusted correspondence between alcohol quantity and risk level (see Table 3 for definitions of risk levels).22
In our event-level models, the outcome of unprotected sex events can be represented as a proportion and as a count of events that occurred each evening. Generalized estimating equation (GEE) models, using robust SEs, which account for clustering because of repeated measurement of participants over time, were used for all analyses. These models also account for differences in the number of data points between participants by weighing data from participants with more data points more heavily. To examine predictors of the log odds of unprotected sex, given the total number of sex acts that occurred per evening over time, we used logistic GEE models with a binomial distribution and logit link, which provide odds ratios (ORs). To examine predictors of the number of unprotected sex acts that occurred per day over time, we used GEE models with a Poisson distribution and log link, which provide a ratio of the estimated mean number of unprotected sex acts compared with overall mean number of unprotected sex acts.
Both unadjusted and adjusted (including all variables) models were estimated. Results did not differ between these models, and therefore, we report only the adjusted models. In Tables 2 and 3, using the model-generated coefficients for each factor included in the model and plugging these into the model regression equation, we calculated the adjusted estimated proportion and number of unprotected sex acts based on the levels of the factor of interest controlling for all other factors in the model. Sociodemographic and situational time-varying variables were retained in the final models if they were statistically significant (P < 0.05). Our results focus on the data from drinking and sexual behavior that occurred during the evening because results from the evening and daytime data were nearly identical.
All participants were of Xhosa ethnicity, average age was 32.23 years (SD 7.20, range 21-55 years), 46.7% had completed primary school or less and 53.4% had completed high school, 32% were married or in the process of getting married, 22.7% were living with a partner but were not married, 25.3% were single, and 20% were widowed. Nearly half (43.8%) reported not having enough money for food. On average, participants had known that they were HIV-positive for 36.45 months (SD 30.81, range 1-144), 41.3% reported having an STI in the last 6 months. More than half (59.2%) were eligible for and were currently taking ARVs, indicating that at ARV initiation, their CD4+ count was <200 per millimeter.3,23
Of the possible 3396 data points during the study, participants completed 89.37% of the possible days, yielding 3035 data points. The mean number of days completed was 37.04 (SD 7.56, range 8-42 days).
There were 4848 vaginal and 79 anal sex events reported, of which 80.36% occurred during evening hours. Of the total events, 3904 vaginal and 46 anal sex events were unprotected. On evenings when participants had sex, they reported an average of 2.30 sex events (SD 1.23, range 1-11) and 2.01 unprotected sex events (SD 1.10, range 1-11). On evenings when participants drank alcohol, women drank an average of 5.91 drinks (70.92 g of alcohol; SD 3.59), and men drank an average of 6.74 drinks (80.88 g of alcohol; SD 3.99) (range 1-11 for both men and women).
Overall, 80.1% of sexual events when the participant did not drink alcohol before sex were unprotected compared with 83.0% when the participant drank alcohol. In Table 1, we present the total number of protected and unprotected sex events over the duration of the study by partner type and partner's HIV serostatus and the proportion of sex events that were unprotected when the participant did not and did consume alcohol before sex. From these aggregated data, which do not control for within-subject correlation resulting from repeated observations over time, for women, a greater proportion of the total sex acts were unprotected when alcohol was consumed before sex (0.82) than when it was not (0.79), χ2 = 3.95, P = 0.05. We also observed that among men and women, respectively, for sex events with casual partners, a greater proportion of the total sex acts were unprotected when alcohol was consumed before sex (0.86 men, 0.77 women) than when it was not (0.73 men, 0.71 women), χ2 = 8.76, P = 0.01, men, χ2 = 6.18, P = 0.01, women. When women drank before sex, a greater proportion of sex acts with partners with unknown HIV status were unprotected than when they did not drink before sex (0.60 vs 0.79), χ2 = 29.85, P < 0.001. No other statistically significant differences were observed in the aggregated data.
Effects of Time-Invariant and-Varying Factors
No gender differences were observed in the proportion or number of unprotected sex acts that occurred per day. The proportion and number of unprotected sex acts per day or the number of drinks consumed per day did not decrease over time in response to completing the daily interview for 42 days.
As presented in Table 2, individuals who had completed high school engaged in more unprotected sex events per day over time than did those who had less education [adjusted ratio of means 1.25, confidence interval (CI) 1.09 to 1.43]. Age, marital status, socioeconomic status, length of time knowing HIV-positive status, disclosure, HIV stigma, STI history, or currently taking ARVs did not predict the proportion or the number of unprotected sex acts that occurred per day over time. Regarding time-varying variables, when the sex event was with a main/steady partner, it was more likely to be unprotected compared with those with a casual/secondary partner (adjusted OR 1.46, CI 1.08 to 1.96). Compared with HIV-positive partners, sex was less likely to be unprotected if it was with a partner of unknown HIV serostatus (adjusted OR 0.70, CI 0.53 to 0.93).
Association Between Alcohol Consumption Before Sex and Unprotected Sex
Controlling for all other factors, consumption of 1 or more drinks of alcohol before sex by only the female partner or only the male partner, or by both partners increased the proportion of sex events that were unprotected per day and also increased the number of unprotected sex events that occurred per day (Table 2). The proportion and number of sex acts that were unprotected was largest when only the male partner drank before sex (adjusted OR 3.81, CI 1.08 to 9.38; adjusted ratio of means 1.52, CI 1.27 to 1.82), followed by when both partners drank before sex (adjusted OR 3.04, CI 1.59 to 5.81; adjusted ratio of means 1.43, CI 1.19 to 1.72). However, the effect of drinking before sex varied based on partner type. As illustrated in Figure 1, for women, when they had sex with both casual and steady partners and when only their male partner drank before sex, the sex event was more likely to be unprotected than when only they themselves drank before sex or when neither partner drank before sex. In addition, when women had sex with casual/secondary partners, unprotected sex was most likely to occur when both they and their partner drank. The same pattern of results was found for men (Fig. 1), suggesting that for both men and women, drinking by the partner before sex influences the likelihood that sex will be unprotected more than does their own drinking. Furthermore, when having sex with casual partners, the likelihood of sex being unprotected is even greater when both partners drank before sex.
Effect of the Amount of Alcohol Consumed
Controlling for all other factors, the amount of alcohol, corresponding to a gender-adjusted risk level,22 consumed before sex affected the proportion and the number of unprotected sex events that occurred. As shown in Table 3, compared with not drinking before sex, low-risk drinking did not increase the proportion or number of subsequent sex acts that were unprotected. For moderate-risk drinking and above (men >40 g of alcohol and women >20 g of alcohol), consuming an increasing number of drinks before sex led to incremental increases in the proportion and number of sex acts that were unprotected; however, the trajectory varied based on partner type. As illustrated in Figure 2, the effect of the number of drinks on the likelihood of unprotected sex was stronger when sex was with casual/secondary partners than when it was with main/steady partners. These findings also suggest that the effect of alcohol consumption before sex on unprotected sex may be limited to instances in which moderate or higher risk drinking occurs.
Unprotected sex with HIV-negative partners, partners with unknown HIV status, and HIV-positive partners was reported with high frequency in our sample of HIV-positive individuals in Cape Town, South Africa. Over 42 days, the 82 HIV-positive participants in our study engaged in nearly 4000 unprotected sex acts, more than half of which were with 87 partners who were perceived to be HIV negative and 110 with unknown HIV status, resulting in possible incident HIV infections as discussed subsequently. All but 2 of the participants reported sex with HIV-negative partners or with partners with unknown HIV status, and all but 1 participant reported sex with both regular/steady and casual/secondary partners. Our findings of high levels of unprotected sex among HIV-positive individuals are consistent with those reported in 2 recent studies that were carried out in the same city.24,25
Using the present data, we calculated an estimate of how many incident HIV infections may have occurred during our 42-day study. Taking the number of unprotected sex acts that occurred with each of the 87 HIV-negative partners and 113 partners with unknown HIV status (assuming that, based on South Africa's HIV prevalence rate of 16%,1 84% of partners with unknown HIV status were HIV negative) of study participants and multiplying the number of sex acts with each partner thought to be HIV-negative by the average 0.0012 per act probability of HIV infection estimate from the Rakai, Uganda data,26,27 we estimated that 2.95 incident HIV infections occurred among partners of study participants during the 42-day study. We also ran this calculation by including only the unprotected sex acts that involved alcohol, and we calculated that 0.98 of the 2.95 estimated incident HIV infections were attributable to alcohol consumption before sex. Assuming the same rate of unprotected sex with a participant's current partners over time, we calculated that these estimates correspond to an incidence rate of 30.34 per 100 person-years, with 10.11 attributable to drinking before sex.
Alcohol was frequently consumed before sexual events and consumed in large quantities; an average of more than 5 drinks per sitting for both men and women, which meets the World Health Organization definition of high-risk drinking for men and very high-risk drinking for women.22 Our analyses for the effect of the number of drinks consumed revealed that an individual's drinking affected behavior only in instances where at least moderate-risk (>3 drinks for men and >1.8 drinks for women) drinking behavior occurred. Our findings regarding a general positive association between alcohol consumption before sex and sexual risk behavior are contrary to 1 study13 but consistent with 2 other studies in Africa6,14 and elsewhere.7,8,28-30 Notably, our study is the first to examine the association at an event level over time in an HIV-positive population in Africa. Using daily diary methods reduced recall bias and afforded us greater confidence in ruling out alternative explanations for the effect of drinking before sex on increasing the likelihood and amount of subsequent unprotected sexual behavior.
Finding that engaging in moderate-risk drinking before discrete sex events almost doubled the likelihood that those sex events were unprotected compared with instances in which drinking did not occur before sex allows us to rule out the competing explanation for an association between alcohol consumption and unprotected sex-that individuals drink equally frequently before protected and unprotected sex. Interestingly, when the data were aggregated (Table 1) and clustering due to repeated measurement of subjects over time was not accounted for, only a slightly greater proportion of sex events were unprotected when preceded by alcohol use than when not. However, these aggregated data do not permit a comparison between sexual events that did and did not involve alcohol within individuals, nor do they permit us to rule out a competing hypothesis that the effect of drinking before sex on increasing the likelihood of unprotected sex is actually caused by individual differences such as personality characteristics.
The daily diary methodology also allowed us to examine the effect of situational variables including partner type and partner drinking before sex at the event level over time. Drinking before sex by either partner increased the proportion and number of unprotected sex events, except for with steady partners among men when only the individual himself drank before sex. In all cases, except for women in sex events with main/steady partners, compared with when only the individual drank before sex, if both partners drank, this further slightly increased the likelihood (Fig. 1) and number of unprotected sex events. In a recent review of the literature on the association between alcohol consumption and unprotected sex in sub-Saharan Africa, Kalichman et al11 concluded that women's unprotected sexual behavior is often the result of their partner's drinking. In contrast, our data suggest that women's unprotected sexual behavior can be attributed to their own and their partner's drinking. Furthermore, women and men consumed alcohol before sex at similar rates. Consistent with prior event-level research,7,17,18,31 we found that drinking before sex influenced subsequent sexual behavior with casual/secondary partners more than it did with main/steady partners.
Gender-related power differentials in sexual behavior decision making, especially in the context of alcohol consumption, are believed to be an important factor in understanding unprotected sexual behavior in South Africa.32-34 Our null findings regarding gender differences may be in part as a result of lack of power to detect gender differences because the sample was composed of more women than men. Although every effort was made to recruit equal numbers of men and women into the study, it is more difficult to recruit HIV-positive men for participation in research studies because of the fact that fewer men than women generally attend HIV services in South Africa. This is because women frequently receive voluntary HIV testing as part of antenatal care, and thus, more women than men are aware of their HIV-positive status.
A limitation of the present study is that the sample consisted of individuals who were seeking HIV services. Therefore, it is unknown to what extent the findings would generalize to HIV-positive individuals who have not sought treatment and care services and may experience greater fear of being stigmatized. Our sample consisted of HIV-positive individuals in an urban area who were both sexually active and had recently consumed alcohol. This limits the generalizability of our findings beyond HIV-positive individuals with these characteristics. However, a recent study that sampled more than 1000 HIV-positive individuals in Cape Town found that for the prior 3 months, 90% of men and 81% of women reported having sex, and 64% of men and 43% of women reported consuming alcohol.24 This suggests that our findings are generalizable to a significant proportion of HIV-positive individuals in urban areas of South Africa.
As with all studies with multiple assessment points, participant burden and reactivity are possible limitations. Analyses for the effect of time indicated that the participants did not change their sexual or alcohol use behavior during the study, which suggests that measurement reactivity was not a problem. The accuracy of self-reports of sensitive behaviors such as sexual behaviors is still to some extent unknown, but methodologies such as those used in the present study increase the likelihood of veridical responses.35-37 Finally, although our methodology allowed us to rule out competing explanations for our findings such as that individuals who drink alcohol may be more likely to have unprotected sex and that those who have unprotected sex may be more likely to drink alcohol, explain the observed effects, we cannot infer causality per se because it was a descriptive not a randomized study.
Despite these limitations, the present prospective cohort study using daily diary methods provided the first evidence that at the event-level engaging in moderate or higher risk drinking before sex increases the likelihood and rate of unprotected sex among HIV-positive individuals in South Africa. As the first study we are aware of in South Africa to collect daily diary data and to do so using phone interviews, we demonstrated that this data collection method is both a feasible and a valuable methodology for studying unprotected sexual behavior among HIV-positive individuals in Africa.
Most surprising were the sheer numbers of unprotected sex events that occurred during the brief course of this study and especially the large percentage of these events that were with presumed HIV-negative partners and partners with unknown HIV serostatus, which likely led to incident HIV infections. From a public health perspective, these data indicate a critical need for the rapid and widespread implementation of HIV prevention efforts for HIV-positive individuals that consider the role of alcohol use in precipitating unprotected sexual behavior. Such interventions could help individuals who are living with HIV reduce their unprotected sexual behavior to protect their own health and to avoid transmitting the virus to others. Speaking the importance of reducing alcohol-involved unprotected sex, extrapolating further from our calculations of incident HIV infections attributable to alcohol consumption among our sample, our data suggest that an intervention that reduced alcohol-involved unprotected sex by 50% among urban HIV-positive individuals in South Africa who drink alcohol would result in 5 less incident infections per 100 person-years. Such prevention efforts would help avoid a ballooning of the already devastating HIV epidemic in South Africa.
We thank Siyavuya Gudula and Thanduxolo Kamati for their role in data collection and Michael D. Stein for his helpful comments on the manuscript.
1. Shisana O, Rehle T, Simbayi LC, et al. South African National HIV Prevalence, HIV Incidence, Behavior and Communication Survey, 2005
. Cape Town, South Africa: Human Sciences Research Council Publishers; 2005.
2. Global HIV Prevention Working Group. HIV Prevention in the Era of Expanded Treatment Access
. Seattle, WA: Bill and Melinda Gates Foundation and Kaiser Family Foundation. Available at: http://www.kff.org/hivaids/hiv061004pkg.cfm
; 2004. Accessed January 11, 2008.
4. Morojele NK, Kachieng'a MA, Mokoko E, et al. Alcohol use and sexual behaviour among risky drinkers and bar and shebeen patrons in Gauteng province, South Africa. Soc Sci Med
5. Kalichman SC, Weinhardt L, DiFonzo K, et al. Sensation seeking and alcohol use as markers of sexual transmission risk behavior in HIV-positive men. Ann Behav Med
6. Moatti JP, Prudhomme J, Traore DC, et al. Access to antiretroviral treatment and sexual behaviours of HIV-infected patients aware of their serostatus in Cote d'Ivoire. AIDS
. 2003;17(Suppl 3):S69-S77.
7. Fortenberry JD, Orr DP, Katz BP, et al. Sex under the influence. A diary self-report study of substance use and sexual behavior among adolescent women. Sex Transm Dis
8. Irwin TW, Morgenstern J, Parsons JT, et al. Alcohol and sexual HIV risk behavior among problem drinking men who have sex with men: an event level analysis of timeline followback data. AIDS Behav
9. Malta M, Bastos FI, Pereira-Koller EM, et al. A qualitative assessment of long distance truck drivers' vulnerability to HIV/AIDS in Itajai, southern Brazil. AIDS Care
10. Rhodes T, Cusick L. Accounting for unprotected sex: stories of agency and acceptability. Soc Sci Med
11. Kalichman SC, Simbayi LC, Kaufman M, et al. Alcohol use and sexual risks for HIV/AIDS in sub-Saharan Africa: systematic review of empirical findings. Prev Sci
12. 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
13. Olley BO, Seedat S, Gxamza F, et al. Determinants of unprotected sex among HIV-positive patients in South Africa. AIDS Care
14. Kiene SM, Christie S, Cornman DH, et al. Sexual risk behaviour among HIV-positive individuals in clinical care in urban KwaZulu-Natal, South Africa. AIDS
15. Tennen H, Affleck G. Daily process in coping with chronic pain: methods and analytic strategies. In: Ziedner M, Endler NS, eds. Handbook of Coping: Theory, Research, Applications
. New York, NY: Wiley; 1996:151-177.
16. Simbayi LC, Kalichman SC, Cain D, et al. Alcohol and risks for HIV/AIDS among sexually transmitted infection clinic patients in Cape Town, South Africa. Subst Abus
17. Kiene SM, Barta WD, Tennen H, Armeli S. Alcohol, helping young adults have sex with casual partners: Findings from a daily diary study of alcohol use and sexual behavior. J Adolesc Health
. In Press.
18. Barta WD, Portnoy DB, Kiene SM, et al. A Daily process investigation of alcohol-involved sexual risk behavior among economically disadvantaged problem drinkers living with HIV/AIDS. AIDS Behav
19. SPSS [computer program]. Version 15.0. Chicago, IL: SPSS Inc; 2006.
20. Berger BE, Ferrans CE, Lashley FR. Measuring stigma in people with HIV: psychometric assessment of the HIV stigma scale. Res Nurs Health
21. Kalichman SC, Simbayi LC, Jooste S, et al. Development of a brief scale to measure AIDS-related stigma in South Africa. AIDS Behav
22. Lopez A, Stockwell T, Chikritzhs T. International Guide for Monitoring Alcohol Consumption and Related Harm
. Geneva, Switzerland: World Health Organization; 2000.
23. Republic of South Africa Department of Health. National Antiretroviral Treatment Guidelines
. Pretoria, Republic of South Africa; 2004.
24. Simbayi LC, Kalichman SC, Strebel A, et al. Disclosure of HIV status to sex partners and sexual risk behaviours among HIV-positive men and women, Cape Town, South Africa. Sex Transm Infect
25. Eisele TP, Mathews C, Chopra M, et al. High levels of risk behaviour among people living with HIV initiating and waiting to start antiretroviral therapy in Cape Town, South Africa. AIDS Behav
26. Wawer MJ, Gray RH, Sewankambo NK, et al. Rates of HIV-1 transmission per coital act, by stage of HIV-1 infection, in Rakai, Uganda. J Infect Dis
27. Gray RH, Wawer MJ, Brookmeyer R, et al. Probability of HIV-1 transmission per coital act in monogamous, heterosexual, HIV-1-discordant couples in Rakai, Uganda. Lancet
28. Ehrenstein V, Horton NJ, Samet JH. Inconsistent condom use among HIV-infected patients with alcohol problems. Drug Alcohol Depend
29. Purcell DW, Mizuno Y, Metsch LR, et al. Unprotected sexual behavior among heterosexual HIV-positive injection drug using men: associations by partner type and partner serostatus. J Urban Health
30. Stein M, Herman DS, Trisvan E, et al. Alcohol use and sexual risk behavior among human immunodeficiency virus-positive persons. Alcohol Clin Exp Res
31. Vanable PA, McKirnan DJ, Buchbinder SP, et al. Alcohol use and high-risk sexual behavior among men who have sex with men: the effects of consumption level and partner type. Health Psychol
32. Meyer-Weitz A, Reddy P, Van Den Borne HW, et al. Determinants of multi-partner behaviour of male patients with sexually transmitted diseases in South Africa: implications for interventions. Int J Mens Health
33. Wojcicki JM.“She drank his money”: survival sex and the problem of violence in taverns in Gauteng province, South Africa. Med Anthropol Q
34. Wojcicki JM, Malala J. Condom use, power and HIV/AIDS risk: sex-workers bargain for survival in Hillbrow/Joubert Park/Berea, Johannesburg. Soc Sci Med
35. Schroder KE, Carey MP, Vanable PA. Methodological challenges in research on sexual risk behavior: II. Accuracy of self-reports. Ann Behav Med
36. McAuliffe TL, DiFranceisco W, Reed BR. Effects of question format and collection mode on the accuracy of retrospective surveys of health risk behavior: a comparison with daily sexual activity diaries. Health Psychol
37. Orr DP, Fortenberry JD, Blythe MJ. Validity of self-reported sexual behaviors in adolescent women using biomarker outcomes. Sex Transm Dis