As the HIV epidemic in sub-Saharan Africa has matured, so too has the search for social behaviors that are significant risk factors for the disease. Alcohol use and its deviant manifestations such as binge and symptomatic problem drinking are such behaviors. A large and growing body of evidence from the region indicates that alcohol use is associated with HIV prevalence and incidence and that the effect operates directly and indirectly through other known pathways such as increased risk of concomitant sexually transmitted infections (STIs), notably HSV-2, and patterns of risky sexual behavior.
For instance, a systematic review and meta-analysis by Fisher et al1 found a strong relationship between alcohol use and HIV infection across 20 studies conducted in Africa. Literature reviews by Cook and Clark2 and Kalichman et al 3 reported similar links between alcohol use and other STIs and high-risk sexual behaviors, respectively. In the sample used for this study, alcohol use has been shown to be related to the prevalence of HIV infection,4,5 and this relationship exhibits dose-response characteristics.6 It has also been shown to be associated with HSV-2 prevalence and incidence7 and engagement in high-risk sex.6 Other studies have reported similar findings in the surrounding area of northern Tanzania.8–13
Morojele et al14 proposed a conceptual model to describe how alcohol use might operate in sexual situations. In their model, alcohol use is structured by a set of social, cultural, economic, and intrapersonal factors, analogous to what is known in the alcohol literature as differential association.15 Alcohol use, in turn, is moderated by the same factors and a person's predisposition and history, differential reinforcement. Once the choice to drink is made, alcohol has psychoactive effects, e.g., sexual arousal, disinhibition, etc., which can lead to sexual risk-taking behavior.
Other researchers, notably Kalichman et al,16,17 have adopted this model to explore the psychoactive aspects surrounding alcohol use in sexual situations, specifically sensation-seeking and expectancies of sexual arousal, performance, and enjoyment. Their findings suggest that sensation-seeking is associated with the feeling of expectancy that alcohol use will increase pleasure and performance.18,19 In another study, expectancies and enhanced sexual experience were associated with drinking before sex for men, but not for women. However, expectancies for women were associated with whether her partner was drinking before sex.19a Interestingly, disinhibition was inversely related to alcohol use before sex. Finally, the researchers used their findings to design and test what proved to be a moderately successful intervention to reduce expectations of increased sexual enjoyment through alcohol use.20
One way that drinking before sex might affect HIV transmission is if its influence on the participants interacted with their use of condoms. Drinking in conjunction with sex might affect the mechanics of condom use, increasing the likelihood of improper use, and hence a failure in protection from disease transmission. At least one study found evidence to support this notion; condom failure was more than twice as likely if alcohol was used before sex, and condom failure was reported 4 times more often among problem drinkers.21
Additionally, drinking may promote greater involvement in high-risk sex and/or less frequent use of condoms than they might otherwise be if no alcohol consumption was involved. Although the relationship between alcohol use and various high-risk behaviors such as forced or coercive sex, transactional sex, or having multiple partners is consistent and well established, the relationship between drinking and condom use is not clear. Several studies have noted that condom use is more likely when the participants have been drinking or are problem drinkers.22–27 Conversely, other studies report an increase in protected sex among drinkers.28–30 If the evidence were not conflicting enough, a third set of studies found that alcohol consumption is associated with inconsistent or situation-specific condom use or that there was no relationship at all.31–35
The goal of our study was to investigate the interplay between condom use and drinking before sex in a different manner, by comparing the characteristics of protected and unprotected sexual encounters in which alcohol is used by the woman participant within the preceding 2 hours, with events in which the woman did not drink before sex.
We start from the perspective that alcohol use before sex and condom use takes place in a sexual milieu characterized by situational characteristics, each carrying some contingent risk of HIV infection, such as where the event took place, partner characteristics, etc. Some types of encounters carry more inherent risk than others. Our objective is then to determine if the situational profiles of sexual encounters differ regardless of whether the event is protected, the woman has been drinking before sex, and whether condom use and alcohol consumption with sex are independent or “interact,” that is whether the situational profile of protected and unprotected sexual encounters change when alcohol use before the event is considered.
This study is a continuation of our ongoing investigation into the relationship between alcohol use and HIV infection in Africa. It builds on the findings of our systematic review and meta-analysis on this topic,1 and our article that reported dose-response relationships between alcohol use and problem drinking and HIV infection and HIV risk in this sample of women.6 The study reported here takes the investigation a step further by exploring the relationship at the level of individual sexual events.
Sample and Data Collection
The study was conducted in Moshi town, the administrative capital of the Kilimanjaro region in northern Tanzania. Moshi, the main center of commercial activity in the area, has a large number of bars and hotels that cater to tourist traveling to Mt. Kilimanjaro, nearby game parks, and the Serengeti. In addition, the town is bisected by a major highway and represents a convenient overnight stop for long-haul truckers. Many of the women who work in the surrounding bars and hotels engage in commercial sex work, and hence are at elevated risk to acquire and transmit HIV-1 and other sexually transmitted diseases.
To promote public health, the Tanzanian Ministry of Health requires bar and hotel workers to receive a physical examination and screening twice a year. Treatment, if necessary, is provided at no cost. The Kilimanjaro Reproductive Health Program, a collaborative program of the Kilimanjaro Christian Medical Center, Moshi Municipal Council, and the Harvard School of Public Health, established a clinic to provide these services. The clinic was also the point of intercept for recruitment of women to take part in a prospective study designed to measure HIV-1 incidence and to identify risk factors for seroconversion. Data for our study were taken from the baseline assessment at enrollment of women in the prospective study.
All registered bars and hotels in the 15 local administrative wards of Moshi town were enumerated initially. Sample recruitment began from the ward having the largest number of hotels and progressed to the ward having the next highest number, and so on, until a sample of 1050 women were enrolled. The sample size was determined to be the number needed to detect an HIV-1 incidence of 3.4 per 100 person-years with 80% confidence.
Later, the sample was augmented with the recruitment of an additional 579 participants to increase the statistical precision of HIV incidence and risk estimates. Only 64 (3.8%) women who were approached refused to take part in the study. The final sample for the study, therefore, consisted of 1629 women aged 14 or older who were enrolled in the prospective study between December 2002 and December 2005.
A lengthy face-to-face survey was administered at baseline, which collected information regarding socio-demographic, reproductive health, sexual behavior, and attitudes. Subsequently, the women were tested for HIV and other STIs. A more complete description of the study procedures, in particular, the clinical testing procedures and laboratory methods can be found in Ao et al.4 The Ethics Committee of the Tanzanian Institute of Medical Research and Kilimanjaro Christian Medical Center and the Institutional Review Board of Harvard School of Public Health approved the study protocol.
As part of the baseline survey, participants were asked about their sexual activity over the past 5 years and whether they used male or female condoms with their partners. Respondents were categorized as consistent in their condom use if they always or never used condoms and inconsistent users if they often or sometimes used them. Overall, 1573 women provided information about past condom use of whom 805 (51.2%) reported never having used condoms in the past 5 years, 160 (10.2%) always did, and the remaining 608 women (38.7%) were inconsistent condom users.
Subsequently, the inconsistent condom users were asked a series of questions about the context of their sexual encounters and the characteristics of their partners. A woman was first asked questions about the last time she had protected sex. Then, the questions were repeated and she was asked to answer them again with respect to the last time she had unprotected sex. One of these questions was whether the woman had been drinking within 2 hours before the sexual encounter. Thirty percent of protected encounters involved alcohol use before sex, whereas 24% of unprotected events did. Another question of interest in the analysis was whether the condom used during the last protected sex experience failed, i.e., broke or fell off.
As a consequence of the survey procedure, each woman provided a set of paired responses to the sexual behavior questions. The relationship between pairs of responses was preserved in the analysis such that the characteristics of the encounter when the woman had protected sex were matched with the same characteristics when the sex was unprotected. In this manner, each woman acted as her own control and the variability in responses as a result of potential confounders, such as socio-demographic, cultural and attitudinal variables, was minimized.
Initially, we examined the relationship between drinking before sex and the likelihood that the condom used in protected sexual encounters either broke or fell off. Drinking before sex is a multicondition variable in this analysis, i.e., the situation could be characterized in one of the following four ways: (1) no one drank before sex, (2) only the woman drank, (3) only the man drank, or (4) both man and woman drank within 2 hours before sex. Condom failure was multifaceted as well. The respondents could report that the condom (1) failed (broke or fell off), (2) did not fail, or (3) they were unsure if it failed or not.
For purposes of the analysis, we compared the recalled experience of women in each drinking conditions with that of women who did not drink before sex. Additionally, we created super drinking categories as follows: (1) woman drinking (woman only + woman and man drinking), (2) man drinking (man only + woman and man drinking), and (3) someone drinking (woman only + man only + woman and man drinking). We compared the condom failure experience of these groups with the likelihood of condom failure experience by women whose last protected sexual encounter involved no drinking by either partner. The analysis was done first using the condition that the woman was certain the condom broke or fell off and was repeated if she knew the condom failed or if she was uncertain that it did.
Next, we analyzed the paired response questions. In this instance, we created a 2 × 2 covariance model for matched categorical data, in which one independent variable was whether the woman had been drinking 2 hours before sex (yes or no), the second was whether condoms were used (yes or no), and the covariate was the time since the event (<2 months or ≤2 months). A separate statistical model was tested for each level of an event descriptor, e.g., where the encounter occurred (partner's house, bar/hotel, or guest house each vs. the participant's own house or apartment) or if force was involved (yes vs. no), etc.
In all, there were 22 dependent variables, and hence 22 individual models were created. The models tested the main effects of the two independent variables and their interaction, holding constant time since the event. However, the interaction term was not significant in any analysis, and so it was dropped and the models were re-estimated with the 2 main effects and covariate only. All analyses were conducted with the GENMOD subroutine in SAS.37
A total of 608 women answered questions about their last protected sexual event. Among the events described, 606 (99.7%) involved the use of a male condom and in the other 2 cases (0.3%) a female condom was used. Women reported successful condom use in 569 instances (93.6%), whereas in 29 cases (4.8%) the condom broke or fell off. In another 10 instances (1.6%), the women were not certain if protection afforded by condom use failed or not.
Table 1 shows the likelihood of condom failure for the various conditions of alcohol use before sex. Regarding instances in which the respondent was certain about whether her partner had been drinking before sex and whether the condom had broken or fallen off, the risk of condom failure increased significantly if 1 or both of the partners had been drinking within 2 hours before sex. If neither partner had been drinking before sex, condoms failed 1.7% of the time. By contrast, if someone had been drinking before sex, the woman only, the man only, or both the man and woman, the likelihood of condom failure increased 5 times to 8.2% (OR, 5.19; 95% CI, 2.05–15.46).
The greatest risk of condom failure occurred when the woman alone had been drinking. Although the number of events in which the woman alone was drinking were few (36 or 5.9% of all events), the likelihood of condom failure was very high, 19.4% or 14 times greater than if no one was drinking (OR, 14.05; 95% CI, 4.03–50.41). In fact, nearly a quarter of all condom failures (24.1%) occurred during events in which the woman alone was drinking before sex. By comparison, when the man alone had been drinking before sex, the chance of condom failure was increased but by a lesser amount. Condom failure in these cases occurred 6.1% of the time (OR, 3.81; 95% CI, 1.15–13.33). A similar rate of condom failure (7.0%) occurred if both the man and woman had been drinking before sex, and the risk of condom failure in these events was more than 4 times greater than if neither party had been drinking within 2 hours before sex (OR, 4.38; 95% CI, 1.47–14.37).
The overall likelihood of condom failure increases when events that were more ambiguous with respect to partner drinking and condom failure are included. For instance, the number of condom failures, if all ambiguous events are considered failures, increases to 39 or 6.4% of the 608 total events described, and to 2.3% in the 298 events in which neither party was drinking before sex. The likelihood of condom failure as a percentage of events increases for all drinking combinations as well. However, the risk estimates (odds ratios) are little changed when ambiguous events are included. Thus, the risk of condom failure when someone was drinking before sex increases from 4 to 6 times when compared with instances where neither party drank before sex.
Turning to the case crossover results, Table 2 provides a summary of the characteristics of the sexual events overall, by whether the woman had used alcohol 2 hours before sex and whether a condom was used (protected) or not (unprotected).
Considering alcohol use, drinking by the woman 2 hours before sex was associated with sexual situations that are generally deemed to be high risk. For instance, the sexual encounter was more likely to take place in unfamiliar surroundings or neutral locations such as a bar, hotel, or guest house. Nearly half (49.4%) of the events that involve alcohol use by the woman before the sexual encounter took place in these venues versus 25% of the time when a woman had not been drinking before. Gifts or money were exchanged in half the cases (51.2%) when a woman had been drinking and only a third of the time when she had not (33.1%). If a woman drank before the event, she was less likely to ask her partner his HIV status (34.1% vs. 40.8%).
With respect to partner characteristics, the encounter was more likely to involve a partner with whom the woman had a limited relationship. A total of 30.2% of the time the woman had sex with a man for the first time she had been drinking, as compared with only 20.4% if she had not. Partners of drinking women were more often bar customers or acquaintances (37.8% vs. 26.9%) rather than a person with whom they had an ongoing relationship, such as a husband, regular partner, or someone they lived with. The partners of these women also tended to be older than 35 (23.5% vs. 13.8%) years or have indeterminate age (26.8% vs. 17.1%). The women whose partners drank before sex were more than 3 times more likely to have been drinking themselves (79.9% vs. 26.1%).
Similar high-risk patterns emerge when the characteristics of protected events are compared with that of unprotected ones. But in this case, women, regardless of whether they had been drinking before the encounter, were more apt to use a condom if the encounter was high risk. So, protected events took place more often in unfamiliar surroundings (34.2% vs. 18.9%), with casual (40.3% vs. 19.6%) or first-time partners (32.9% vs. 13.2%) or if the sex was transactional (44.9% vs. 31.1%). Consistent with other studies, the women seem to perceive their risk in some situations and attempt to protect themselves by having the partner wear a condom.
One barrier to this attempt at self-protection is the extent to which the woman can control the decision to have the partner wear a condom. In this sample, the partner made the decision in a quarter of the case (25.9%), and in 267 of these 315 cases (84.8%) a condom was not used. Conversely, the woman made the decision in 379 cases (or 31.2% of the time) and 356 (93.9%) of these events were protected. Viewed from another perspective, 92.1% of the time when an encounter was protected the woman was involved in decision, either having made it alone or in conjunction with her partner. For unprotected events, the woman was involved in the decision only 56.1% of the time, and almost all of these (93.2%) in negotiation with her partner.
Patterns of Risk
Table 3 tests the statistical significance of the relationships seen in Table 2. As a matter of interpretation, the model was fit such that not drinking before sex was the reference category. Risk estimates (odds ratios) above 1 mean then that a characteristic is more likely in situations in which a woman had been drinking before sex. With condom use, the reference category is unprotected sex. Hence, risk estimates for condom use are above 1 if the characteristic is more likely when an event is protected and below 1 when a condom is used less often in an encounter having that particular characteristic.
The interaction between alcohol use before sex and condom use was not significant in any of the analyses. Consequently, the effects seen for alcohol use before sex are independent of condom use and vice versa. In other words, the risk relationships observed for alcohol use before sex will be the same whether the event is protected or not and the risk estimates seen for condom use will be the same whether the woman drank before sex or not.
Alcohol use before sex was significantly associated with encounters that took place in bars or hotels (OR, 1.72; 95% CI, 1.06–2.79) or guest houses (OR, 1.97; 95% CI, 1.36–2.85), as compared with ones that took place in the woman's residence; when it involved a regular partner (OR, 1.51; 95% CI, 1.00–2.29) or casual acquaintance (OR, 1.94; 95% CI, 1.31–2.88) as opposed to instances when the partner was a husband or someone with whom she lived; if it was the first time the woman had sex with the man (OR, 1.66; 95% CI, 1.21–2.27) and if money or gifts were exchanged (OR, 1.94; 95% CI, 1.45–2.59). Age of the partner did not evince a consistent or progressive relationship to alcohol use before sex.
Not surprisingly, drinking before sex by the woman was 10 times more likely if the man was drinking as well (OR, 10.37; 95% CI, 7.44–14.46) and nearly twice as likely (OR, 1.73; 95% CI, 1.10–2.58) if the partner was or may have been using drugs. Interestingly, the decision to use condoms was half as likely to be decided jointly if the woman had been drinking before sex (OR, 0.52; 95% CI, 0.37–0.73) and she was significantly less likely to be involved in the decision at all in these situations (OR, 0.61; 95% CI, 0.44–0.84).
The pattern of condom use paralleled that for alcohol use before sex. Sexual encounters were more likely to involve condoms if they took place away from the woman's residence in a bar or hotel (OR, 1.47; 95% CI, 1.05–2.50) or guest house (OR, 2.53; 95% CI, 1.90–3.35), if it involved a first time partner (OR, 2.88; 95% CI, 1.90–3.35) or one with whom she had a casual relationship (OR, 4.22; 95% CI, 3.23–5.50), if sex was transactional (OR, 1.72; 95% CI, 1.44–2.05), and if the partner was or may have been using drugs before the event (OR, 1.50; 95% CI, 1.08–2.07). Condom use was 10 times more likely (OR, 9.76; 95% CI, 6.96–13.70) if the woman was involved in the decision, either making it alone or jointly with her partner, as opposed to instances when the partner made the decision alone.
In our study, drinking before sex is associated with sexual contacts that have higher risk of HIV infection, such as when the woman's relationship with the partner is impermanent, transitory or sex is transactional, the location is unfamiliar and less under her control, and the partner is drinking or using drugs before sex. Although care was taken to question the women about alcohol use preceding sex, the role alcohol consumption plays in these situations, whether it increases a woman's inclination to take more sexual risks or facilitates interaction once the decisions have been made, for example, cannot be conclusively determined from our cross-sectional analysis.
Condom use too is more likely in precisely the same types of encounters. In this respect, our findings are consistent with others that have noted greater condom use in high-risk situations and that condom use may be a marker for such contacts. From a risk perspective, drinking before sex and condom use tend to cancel each other out. But the two conditions are also independent. So, if alcohol consumption does encourage sexual risk-taking, a fact that has yet to be established, it does not affect the likelihood that a condom will be used during the event. A condom is as likely to be used as not when a woman has been drinking before sex, and a woman is as likely as not to drinking before sex in situations where a condom is used.
The use of a condom appears to be more a function of situational negotiation and the woman's control over the outcome. When a woman participates in the decision, either making it alone or in conjunction with her partner, condom use is much more likely. Hence, it is significant then that when a woman has been drinking before the event she is less likely to participate in the decision-making. Alcohol use before sex could increase risk then by reducing a woman's ability to advocate or negotiate for condom use by her partner.
The one meaningful connection between alcohol use before sex and condom use is that condoms fail 5 to 14 times more often when one or both of the parties involved have been drinking, a finding that is consistent with ones reported by Simbayi et al.21 Condom failure works both ways, increasing the risk for the man as well as the woman, an important consideration in our sample in which 19% of the women were HIV+. This fact alone suggests that alcohol use before sex can put the participants at risk for HIV infection even when steps have been taken to prevent it.
The results are interesting as well for what we did not find. Most importantly, condom use and drinking before sex are do not interact. When a woman acts as her own control as in this study, there is no evidence to suggest the characteristics of sexual encounters in which condoms are used will differ from those for unprotected events depending on whether she has been drinking before sex. Alcohol use before sex, in other words, does not increase the likelihood that a sexual encounter will be unprotected if it is high risk. In short, our findings suggest that condom use and alcohol use before sex are distinct and not contingent risk factors.
Additionally, unlike other studies that have reported high rates of forced or coercive sex, most notably in South Africa, and have documented a relationship between these experiences and drinking behavior,26,27,29,38 threats of force were comparatively rare among the encounters recalled by our sample, occurring only 3.5% of the time, and we found no evidence that incidents of coercion were related to drinking before sex or condom use when they did occur. Similarly, we found no relationship between the 2 conditions and a woman asking the HIV status of her partner or a consistent relationship with the partner's age.
The pattern that emerges from the findings of this study is sexual events consisting of a number of contingent decisions, each carrying some measure of aleatory risk of infection. Whether to engage in sex with an unfamiliar partner, in an unfamiliar setting, to drink before, use a condom, accept gifts, and so forth, add an element of chance to each encounter. Even when all controllable risks are minimized, chance can still play a part; condoms can still break or fall off.
Over a lifetime of sexual activity, myriad of such decisions, each with its associated small risk of infection, set up a cascade of contingent probabilities that will result, when summated, in the combined probability that infection will occur more often among women who make such choices. At some point, a confluence of small probabilities will conspire and the woman will simply be unlucky. In a cross-sectional study, such as this, the summation of separate probability chains across numerous events engaged in by a sample of individuals is evidenced by higher rates of infection among a group of women who more often made a similar set of choices.
If our notion of contingent probability chains is correct, then older women should have higher rates of HIV infection on average because the greater the number of chances, the greater the likelihood of infection. In fact, this is precisely the case. Ao et al4 reported a strong linear relationship between age and the rate of HIV infection in this group of women, an increase of 5 to 6 times from the youngest age group to the oldest.
Although we believe our findings advance the understanding of the disease at the level of human interaction, the data and analysis have limitations. The usual caveats apply, which include that the primary data are based on self-reports of past events and the study is cross-sectional so no temporal priority or causality can be determined. Thus, in case of drinking before sex, it remains unclear whether alcohol consumption causes women to engage more readily in risky sex or once those decisions are made drinking ensues.
Perhaps more telling is the constraint imposed by the paired comparison method of questioning. The tight control afforded by the approach put the differences between protected and unprotected events in sharp relief. But the limitation of the questionnaire design was that women had to be inconsistent condom users before they were asked about drinking and sex. Hence, we do not know how the prevalence of alcohol use before sex varied among consistent condom users or if the rates of condom failure and the characteristics of sexual encounters involving drinking among these women paralleled those seen for inconsistent condom users. This could be a fruitful area of future investigation.
Our findings suggest that the most direct effect of alcohol use before sex is on rates of condom failure and the diminished capacity of a woman to demand a partner use a condom or to successfully negotiate for when the decision is made jointly. Interventions that address alcohol use generally and in these situations may have immediate benefits. The independence we observed between alcohol use before sex and condom use suggests that interventions directed at both behaviors may yield benefits but would not overlap, so separate initiatives will be needed for maximal effect.
1. 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.
2. Cook RL, Clark DB. Is there an association between alcohol consumption and sexually transmitted diseases? A systematic review. Sex Transm Dis 2005; 32:156–164.
3. 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 2007; 8:141–151.
4. Ao TT, Sam NE, Masenga EJ, et al. Human immunodeficiency virus type 1 among bar and hotel workers in northern Tanzania: The role of alcohol, sexual behavior, and herpes simplex virus type 2. Sex Transm Dis 2006; 33:163–169.
5. Kapiga SH, Sam NE, Shao JF, et al. HIV-1 epidemic among female bar and hotel workers in northern Tanzania: Risk factors and opportunities for prevention. J Acquir Immun Defic Syndr 2002:1; 29:409–417.
6. Fisher JC, Cook PA, Sam NE, et al. Patterns of alcohol use, problem drinking, and HIV infection among high-risk African women. Sex Transm Dis 2008; 35:537–544.
7. Tassiopoulos KK, Seage G, Sam NE, et al. Predictors of herpes simplex virus type 2 prevalence and incidence among bar and hotel workers in Moshi, Tanzania. J Infect Dis 2007; 195:493–501.
8. Mnyika KS, Klepp KI, Kvåle G, et al. Risk factors for HIV-1 infection among women in the Arusha region of Tanzania. J Acquir Immun Defic Syndr Hum Retrovirol 1996; 11:484–491.
9. Mnyika KS, Klepp KI, Kvåle G, et al. Determinants of high-risk sexual behaviour and condom use among adults in the Arusha region, Tanzania. Int J STD AIDS 1997; 8:176–183.
10. Mmbaga EJ, Hussain A, Leyna GH, et al. Prevalence and risk factors for HIV-1 infection in rural Kilimanjaro region of Tanzania: Implications for prevention and treatment. BMC Public Health 2007; 7:58.
11. Tengia-Kessy A, Msamanga GI, Moshiro CS. Assessment of behavioural risk factors associated with HIV infection among youth in Moshi rural district, Tanzania. East Afr Med J 1998; 75:528–532.
12. Kapiga SH, Sam NE, Mlay J, et al. The epidemiology of HIV-1 infection in northern Tanzania: Results from a community-based study. AIDS Care 2006; 18:379–387.
13. Watson-Jones D, Weiss HA, Rusizoka M, et al. Risk factors for herpes simplex virus type 2 and HIV among women at high risk in northwestern Tanzania: Preparing for an HSV-2 intervention trial. J Acquir Immun Defic Syndr 2007; 46:631–642.
14. 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 2006; 62:217–227.
15. Fisher JC. Advertising, Alcohol Consumption and Abuse: A Worldwide Survey. Westport, CT: Greenwood Press, 1993: 15–17.
16. Kalichman SC, Simbayi LC, Vermaak R, et al. Randomized trial of a community-based alcohol-related HIV risk-reduction intervention for men and women in Cape Town, South Africa. Ann Behav Med 2008; 36:270–279.
17. Kalichman SC, Simbayi L, Jooste S, et al. Sensation seeking and alcohol use predict HIV transmission risks: Prospective study of sexually transmitted infection clinic patients, Cape Town, South Africa. Addict Behav 2008; 33:1630–1633.
18. Kalichman SC, Simbayi LC, Jooste S, et al. Sensation seeking, alcohol use, and sexual behaviors among sexually transmitted infection clinic patients in Cape Town, South Africa. Psychol Addict Behav 2006; 20:298–304.
19. Kalichman SC, Cain D. A prospective study of sensation seeking and alcohol use as predictors of sexual risk behaviors among men and women receiving sexually transmitted infection clinic services. Psychol Addict Behav 2004; 18:367–373.
19a. Kalichman SC, Simbayi LC, Cain D, et al. Alcohol expectancies and risky drinking among men and women at high-risk for HIV infection in Cape Town, South Africa. Addict Behav 2007; 32:2304–2310.
20. Kalichman SC, Simbayi LC, Vermaak R, et al. HIV/AIDS risk reduction counseling for alcohol using sexually transmitted infections clinic patients in Cape Town, South Africa. J Acquir Immun Defic Syndr 2007; 44:594–600.
21. Simbayi LC, Kalichman SC, Jooste S, et al. Alcohol use and sexual risks for HIV infection among men and women receiving sexually transmitted infection clinic services in Cape Town, South Africa. J Stud Alcohol 2004; 65:434–442.
22. Coldiron ME, Stephenson R, Chomba E, et al. The relationship between alcohol consumption and unprotected sex among known HIV-discordant couples in Rwanda and Zambia. AIDS Behav 2008; 12:594–603.
23. Olley BO, Seedat S, Gxamza F, et al. Determinants of unprotected sex among HIV-positive patients in South Africa. AIDS Care 2005; 17:1–9.
24. Zachariah R, Spielmann MP, Harries AD, et al. Sexually transmitted infections and sexual behaviour among commercial sex workers in a rural district of Malawi. Int J STD AIDS 2003; 14:185–188.
25. Weiser SD, Leiter K, Heisler M, et al. A population-based study on alcohol and high-risk sexual behaviors in Botswana. PLoS Med 2006; 3:e392.
26. Chersich MF, Luchters SMF, Malonza IM, et al. Heavy episodic drinking among Kenyan female sex workers is associated with unsafe sex, sexual violence and sexually transmitted infections. Int J STD AIDS 2007; 18:764–769.
27. 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 2006; 27:37–43.
28. Twa-Twa JM, Oketcho S, Siziya S, et al. Prevalence and correlates of condom use at last sexual intercourse among in-school adolescents in urban areas of Uganda. East Afr J Public Health 2008; 5:22–25.
29. Smit J, Myer L, Middelkoop K, et al. Mental health and sexual risk behaviours in a South African township: A community-based cross-sectional study. Public Health 2006; 120:534–542.
30. Hoffman S, O'Sullivan LF, Harrison A, et al. HIV risk behaviors and the context of sexual coercion in young adults' sexual interactions: Results from a diary study in rural South Africa. Sex Transm Dis 2006; 33:52–58.
31. Kongnyuy EJ, Wiysonge CS. Alcohol use and extramarital sex among men in Cameroon. BMC Int Health Hum Rights 2007; 7:6.
32. Zablotska IB, Gray RH, Serwadda D, et al. Alcohol use before sex and HIV acquisition: A longitudinal study in Rakai, Uganda. AIDS 2006; 20:1191–1196.
33. Yadav G, Saskin R, Ngugi E, et al. Associations of sexual risk taking among Kenyan female sex workers after enrollment in an HIV-1 prevention trial. J Acquir Immun Defic Syndr 2005; 38:329–334.
34. Palen L, Smith EA, Flisher AJ, et al. Substance use and sexual risk behavior among South African eighth grade students. J Adolesc Health 2006; 39:761–763.
35. Myer L, Mathews C, Little F. Condom use and sexual behaviors among individuals procuring free male condoms in South Africa: A prospective study. Sex Transm Dis 2002; 29:239–241.
36. Kapiga SH, Lyamuya EF, Lwihula GK, et al. The incidence of HIV infection among women using family planning methods in Dar es Salaam, Tanzania. AIDS 1998; 12:75–84.
37. GENMOD SAS. Version 9.3. Cary, NC: SAS Institute Inc.
38. Zablotska I, Gray R, Koenig M, et al. Alcohol Use, intimate partner violence, sexual coercion and HIV among women aged 15–24 in Rakai, Uganda. AIDS Behav. 2007; 13:225–233.