Ao, Trong T. H. ScM*; Sam, Noel E. MD†; Masenga, Elisante J. MD‡; Seage, George R. III ScD§; Kapiga, Saidi H. ScD∥
WOMEN WORKING IN THE bars and hotels in Tanzania are at relatively (high risk) of sexually transmitted diseases (STDs), including human immunodeficiency virus type 1 (HIV-1) infection.1–3 Members of this population are known to practice informal commercial sex work in addition to their regular job of serving food and drink to their customers. Results from a pilot study conducted between June and October 2000 in the town of Moshi, Tanzania, confirm that HIV-1/STDs are a major public health problem among women working in these settings, with an HIV-1 prevalence of 26.3%,1 substantially higher than among women attending primary health care clinics in Moshi (11.5%)4 and female blood donors (13.7%).5
Although formative work has begun to provide insights on the complex behavioral factors associated with increased vulnerability of women working in these settings, there is limited information about the prevalence and incidence of HIV-1/STDs in this population. Such information is required in the design of intervention programs and in monitoring their impact. To obtain this information, we are conducting a prospective cohort study to determine the incidence of HIV-1 and assess the feasibility of implementing long-term HIV-1/STD interventions among women working in the bars and hotels in Moshi. We have enrolled 1050 women working in these settings and the follow-up is ongoing.
In this report, we present results from the analysis of baseline data collected during enrollment of the cohort. The objectives of the report are to estimate the baseline prevalence of HIV-1/STDs and to describe the demographic, behavioral, and socioeconomic risk factors associated with HIV-1 infection in the study population. In addition to confirming our previous results,1 this study involved a larger sample size and use of a standardized screening instrument for measuring problem drinking in this population.
Recruitment of Study Subjects
A major industry in the Kilimanjaro region of northern Tanzania is international and domestic tourism. As a result, a large number of hotels and bars have been established over the years in Moshi, the capital town of this region. As part of the public health measures to ensure the safety of food and drink, bar and hotel workers are required by the Tanzania Ministry of Health to receive routine medical services twice a year. These include a physical examination and testing of stool, urine, and blood samples for parasitic infections and typhoid and provision of treatment if needed at no cost. In December 2002, we established a clinic in Moshi to provide these services to the bar/hotel workers and recruit subjects in a prospective cohort study. This was part of the Kilimanjaro Reproductive Health Program, a collaborative effort between Kilimanjaro Christian Medical Center (KCMC), Moshi Municipal Council (MMC) and Harvard School of Public Health (HSPH).
Before data collection began, we obtained information about the number of registered bars/hotels in all 15 local administrative wards in Moshi. In each ward, bar and hotel owners and/or managers were invited to attend a meeting where information about medical services and the study were provided. Later, outreach workers visited the bars/hotels and met with both male and female workers to provide more information about the medical services and invite them to visit the clinic to receive these services. Female workers received additional information about the study aims and procedures. All women aged ≥14 years, working in the registered bars/hotels, and willing to provide written consent and participate in study procedures were eligible to be enrolled in the study.
We started data-collection activities for the study in the ward with the largest number of bars/hotels and progressed to the ward with the next largest number of bars/hotels until we reached a sample size of 1050 women, which was predetermined based on 80% power to detect a minimum HIV-1 incidence of 3.4/100 person-years.
Women who visited the clinic received the required medical services before meeting with the research staff for more detailed information about the study aims and procedures. Written informed consent was obtained from women who agreed to participate in the study. Consenting women were interviewed in a private room to obtain information about sociodemographic characteristics, employment history and work mobility, sexual behaviors, reproductive health history, knowledge of HIV-1/STDs, and HIV-1 risk perception. We also collected information about the frequency of alcohol use and problem drinking by using a standardized CAGE scale.6 Women were classified to have probable indication of problem drinking if they had 2 affirmative answers and strongly indication of problem drinking if they had 3 or 4 affirmative answers to the 4 CAGE questions.
After the interviews, pretest counseling was done, and 5–10 ml of blood was collected for detection of syphilis, herpes simplex virus type 2 (HSV-2), and HIV-1 infection. Later, a clinical examination was performed, and vaginal fluid pH was measured by using a pH paper touched on the lateral vaginal wall. Genital samples for detection of STDs and other genital infections were collected in the following order: posterior fornix swabs for wet mount and isolation of Candida albicans; and endocervical swabs for isolation of Neisseria gonorrhea and detection of chlamydial infection. Blood and genital samples were transported in cold boxes within 3 to 4 hours of collection to the Department of Clinical Laboratories at KCMC for further processing. Study participants returned to the clinic within 10 to 14 days for results and posttest counseling, and women with STD-related symptoms or laboratory-confirmed infections received free treatment in accordance to the Tanzanian national treatment guidelines. All women enrolled in the study are participating in the ongoing follow-up. The study protocol was approved by the Ethics Committee of KCMC and the institutional review board of HSPH. We ensured participants' confidentiality by storing study documents in a secured location, providing staff training on confidentiality issues, and having the same staff member to provide pretest and posttest counseling for each participant.
HIV-1 infection was determined by using 2 enzyme-linked immunosorbent assays (ELISA). We used Murex 1.2.0 (Murex Biotech Ltd, England, UK) or Enzygnost HIV-1/HIV-2 (Behring, Marburg, Germany) as our first ELISA test kit. All reactive samples were confirmed by using Vironostika HIV Uni-Form II plus ELISA (Organon, Boxtel, The Netherlands). Indeterminate results were resolved by Western blot (Genetic Systems, Bio-Rad Laboratories, Redmond, WA). Active/recent syphilis was diagnosed if serum was reactive on the Rapid Plasma Reagin (RPR) card test (Becton-Dickinson, Cockysville, MD) with a titer ≥1:8 and the Treponema pallidum hemagglutination assay (TPHA) (Murex Biotech Ltd). HSV-2 was detected by the type-specific HSV-2 ELISA according to manufacturer's instructions (Focus Technologies, Cypress, CA).
A wet mount of a vaginal swab was examined microscopically for the presence of clue cells, motile Trichomonas vaginalis and yeast cells. Candidiasis was diagnosed by isolation of the Gram-negative yeastlike cells on Saboraud's dextrose agar and confirmed by using the germ tube test (Remel, Lenexa, KS). We attempted to isolate N gonorrhea by inoculation of the endocervical swab on modified Thayer Martin media, followed by incubation in candle extinction jar with carbon dioxide generating kit (CO2 Gen, Oxoid Ltd, England, UK) at 36 °C for 24 to 48 hours. Chlamydia trachomatis antigen was detected by using an antigen detection enzyme immunoassay (Murex Biotech Ltd). Positive samples were confirmed by a blocking assay from the same manufacturer.
Bacterial vaginosis (BV) was diagnosed on the basis of 4 clinical criteria: (a) a vaginal pH >4.5; (b) an increased homogenous vaginal discharge in women without concurrent trichomoniasis or candidiasis; (c) presence of clue cells in >20% of the vaginal epithelial cells detected by mixing vaginal fluid with a drop of normal saline on a slide and examining under high-power magnification; and (d) a positive amine or whiff test performed by mixing a few drops of 10% potassium hydroxide with vaginal fluid. Women with none of the 4 clinical criteria were classified as having normal vaginal flora. Disturbances of vaginal flora were classified as “mild” if only 1 criterion was met, “moderate” if 2 criteria were met, and “severe” if 3 or more criteria were met. Women with severe disturbance of vaginal flora were classified as having BV, as previously proposed by Amsel et al.7
The questionnaires were first edited at the clinic and doubled entered by using EpiInfo 6.0 program (CDC, Atlanta, GA). Data analyses were subsequently performed in SPSS for Windows 9.0 (Chicago, IL). The associations between HIV-1 and potential predictor variables were summarized by using odds ratios (OR) and corresponding 95% confidence intervals (CI). Spearman's correlation coefficients (r) were used to assess correlations between pairs of continuous variables. To adjust for multiple risk factors simultaneously, we performed analyses by using logistic regression models.8 Significant variables in the univariate analysis and others that were thought to be important based on previous reports were included in our candidate list of variables for this analysis. Stepwise procedures, using both forward selection and backward elimination, were used to determine a final model. We tested for linear trend of ordinal variables by entering them in the models as continuous variables. After the final models were obtained, we assessed potential confounding of variables not retained in the models by entering 1 variable at a time in the model. Variables that changed any of the coefficients of the predictors in the final model by >10% were considered as potential confounders and were retained in the models. The fitness of the final model was checked by using the Hosmer-Lemoshow test.8
Between December 2002 and November 2003, 1050 women were enrolled in the study. In total, we contacted 315 out of 318 bars/hotels in these 7 wards. Out of the 1150 women who were registered in these establishments, 1050 (91.3%) were enrolled in the study. The mean age of the study population was 28 years (standard deviation = 8.2 years, range = 14–64 years). Most women (61%) have been working in the bars/hotels for more than 1 year. Age was highly correlated with duration of work in these settings (r = 0.65, P = 0.01). Residence in Moshi ranged from 1 month to 60 years (mean = 10.2 years, median = 5.8 years). The predominant religion in this population was Catholic (43.0%), followed by Lutheran (27.2%) and Moslem (23.9%). The majority of the participants (57.0%) belonged to the Chagga ethnic group. Fifty-nine percent of the participants reported to have completed 7 to 8 years of formal education, with 3.3% having never attended school.
Of the 1050 women enrolled in the study, 1042 agreed to provide a blood sample for HIV-1 serology. One hundred ninety-eight out of 1042 women were HIV-1 seropositive, resulting in a prevalence of HIV-1 of 19.0% (95% CI = 16.6%–21.4%). The prevalence of other STDs and genital infections were as follows: HSV-2, 54.6% (95% CI = 51.7%–57.7%); BV, 29.8% (95% CI = 26.8%–32.8%); candidiasis, 17.5% (95%; CI = 15.0%–20.0%); trichomoniasis, 8.4% (95% CI = 6.6%–10.2%); C trachomatis, 5.5% (95% CI = 3.9%–7.1%); and active (recent) syphilis, 1.1% (95% CI = 0.5%–1.7%). None of the women in this study had N gonorrhea isolated by culture.
The prevalence of HIV-1 varied by ward, with women working in Kiusa, Bondeni, and Korongoni wards more likely to be HIV-1 infected compared to women working in Mawenzi ward (Table 1). Compared to women under the age of 20 years, older women were more likely to be infected (P value, test for linear trend <0.001). Only 274 (26.3%) women reported to have never used alcohol, while 34.6% were classified to have probable or strong indication of problem drinking. Among women who used alcohol, more than half drank 2 or more days per week, with the highest occurrence among women who drank more than 3 days per week (OR = 4.40). Women who had problem drinking were also more likely to be infected with HIV-1 (P value, test for linear trend <0.001).
Compared to women with at least 9 years of education, women with less education had the highest prevalence of HIV-1 (P value, test for linear trend = 0.001). The risk of HIV-1 was increased among women who were cohabiting, separated, and widowed when compared to married women and among women who were not on regular monthly salary. The risk of HIV-1 increased with increasing number of live births and the duration of working in the bars/hotels (P value, test for linear trend <0.001). Religion, having a husband with another wife, and level of education of the husband/cohabiting partner were not associated with HIV-1 infection in univariate analysis.
The associations between sexual behavior, risk perception, STDs, and HIV-1 are presented in Table 2. Fifty-five (5.3%) women did not respond to questions about sexual behavior. All women who did not respond to sexual behavior questions were HIV-1 seronegative, unmarried, and had no children. They were also relatively younger than other women and were more likely to have at least 9 years of education and to be non-alcohol users. HIV-1 prevalence was associated with increasing number of sex partners during the past 5 years. Nearly 48% of the women never used condoms during the last 5 years, and among the women who did use condoms, the majority did not use them consistently. Compared to women who never used condoms, women who used them occasionally had an increased risk of HIV-1 (OR = 1.61).
Women were more likely to be HIV-1 infected if they met their last sexual partner in the bar/hotel where they were working (OR = 2.17), received or given a gift in exchange for sex (OR = 1.38), or were aware that their male partners had other sex partners (OR = 1.63). The risk of HIV-1 was correlated with the perceived risk of HIV-1, although this was significant only in women who perceived themselves to be at greatest risk of HIV-1 (OR = 2.35).
Laboratory-confirmed infections and self-reported STDs symptoms were strongly associated with HIV-1. HIV-1 infection was strongly associated with HSV-2 (OR = 6.16). Women with abnormal vaginal discharge and genital ulcer, both on clinical examination and in the past 12 months, were more likely to be infected, while the risk of HIV-1 also increased with increasing degree of disturbances of vaginal flora (P value for linear trend <0.001). Compared to women with normal vaginal flora, the risk of HIV-1 was associated with moderate disturbance (OR = 4.02) and severe disturbance or BV (OR = 5.50).
To adjust for potential confounding, multivariate analysis was conducted (Table 3). Women were more likely to be HIV-1 infected if they were older, cohabiting, separated, and widowed. Interestingly, women married to a husband with another wife had reduced risk of HIV-1 (adjusted OR = 0.46, 95% CI = 0.27–0.78). Other independent predictors of HIV-1 were the ward where the hotel/bar was located, number of sex partners in the past 5 years, genital ulcer during clinical examination, HSV-2, and alcohol problem drinking.
Our findings show that HIV-1 and other STDs are major public health problems among female bar and hotel workers in Moshi. The prevalence of HIV-1 was relatively lower than the 26.4% prevalence (95% CI, 21.4%–31.2%) we reported previously in this population.1 This could be due to a number of factors. First, due to lack of effective and affordable antiretroviral therapy in Moshi, women who were previously HIV-1 infected might have become ill and died. Second, the ongoing educational campaigns implemented by the MMC and community-based organizations might have contributed to reduce HIV-1 incidence and subsequently HIV-1 prevalence. However, a substantial proportion of women in our current study reported high-risk sexual behaviors, indicating that these efforts have been less successful. Furthermore, with the exception of trichomoniasis, there were no significant changes in the prevalence of other STDs during the period when these studies were conducted.
One notable result which might be an intervention opportunity is the low rate of reported condom use. Apart from their known effectiveness against HIV-1,9 condoms confer protection against chlamydia, gonorrhea, HSV-2, and syphilis,10 in addition to trichomoniasis in women. Low condom usage might be a result of ineffective condom promotion or, even more likely, related to the underlying low status of women, thus making it extremely difficult for women to negotiate condom use with their male partners.
Genital ulcer and HSV-2 were strongly associated with HIV-1, suggesting the critical importance of preventing genital herpes as an HIV-1 intervention in this population. Most women with a genital ulcer were HSV-2 seropositive. Genital herpes has been recently recognized as the major cause of GUD in developing countries.11 Genital ulcers provide an easy portal of entry for the HIV-1 virus, and increased presence and activation of HIV-1 susceptible cells in the genital tract.12 Subclinical reactivation of HSV-2 is also associated with influx of activated CD4+ cells which could be infected with HIV-1.13 In addition, HSV-2 transregulatory proteins have been shown to promote HIV-1 expression and transactivation.14 The low prevalence of gonorrhea we observed in this study is consistent with our previous results,1 as well as results from another study conducted in Moshi.15 We have also observed very low prevalence of gonorrhea in a community-based study we recently conducted in Moshi (unpublished data). The absence of gonorrhea in this population with high prevalence of other STDs warrants further investigation.
The prevalence of HIV-1 infection in this population was substantially higher than that observed in the community-based studies conducted in this part of Tanzania,4,16 providing further evidence that women working in establishments serving alcohol are at increased risk of HIV-1 than other women in the general population. The use of CAGE, a known standard tool for screening problem drinking, was an improvement from our previous studies in this population. Problem drinking was found to be an independent risk factor for HIV-1 in this population. Alcohol might be a risk factor for HIV-1 because it impairs personal control and diminishes perception of risk from unprotected sex.17 Furthermore, there is a growing body of evidence which implicates chronic alcohol ingestion with increase susceptibility to HIV-1 infection.18–20 Most recently, Chen et al.21 have suggested that alcohol may influence oral HIV-1 transmission by altering the cellular compartmentalization of CXCR4 in the cells of the oral cavity. This is based on in vitro data, and it is unknown whether these results could be generalized to keratinocytes in other body sites. Oral sex is not generally reported in this population, although it might be more commonly practiced than previously postulated.
The risk of HIV-1 was increased among women reporting multiple sex partners. Although we might have underestimated the proportion of women with multiple sex partners due to reporting bias, our study shows that having multiple sex partners is relatively common. Promotion of safer sexual behaviors, particularly consistent condom use and reduction of the number of sexual partners, should be a major focus of HIV-1 prevention efforts in these settings. To be effective, these interventions should involve male partners of these women.
As observed previously, age was independently associated with HIV-1 infection in this population.1 We also observed a strong association between marital status and HIV-1. Similar findings have been reported in other studies conducted in the general populations in Africa.22,23 Women in relatively unstable relationships may be more likely to engage in high-risk sexual behaviors and have limited power to negotiate with their partners about safer sex. In addition, widowed and separated women might have been exposed to HIV-1 in their previous marriages. Surprisingly, women whose husbands had another wife were less likely to be infected with HIV-1 than other women in monogamous marriage. Women whose husbands had another wife were more likely to be Moslem. Thus, cultural and religious factors might account for the relatively low risk of HIV-1 in these women. We also hypothesize that women in these relationships may have fewer sexual encounters with their husband whom they share with other women, and their husbands might be less likely to seek other partners. Information about the frequency of sexual encounters was not collected in this study, and we are unable to confirm this hypothesis.
We found differences in the prevalence of HIV-1 by ward. The risk of HIV-1 was increased among women working in the bars/hotels located in the 3 wards of Kiusa, Bondeni, and Korongoni. Two of these wards (Korongoni and Kiusa) have the highest concentration of guesthouses that have been known to condone covert commercial sex work among its workers. Overall, the large HIV-1 prevalence differences by location of the bar/hotel suggests that geographically focused interventions may be most effective in reducing further expansion of the HIV-1 epidemic in this population. In developing such interventions, it will be helpful to elucidate how women in the most affected wards are connected to their partners by using the sexual network paradigm as delineated by Morris.24 Weir et al.25 have found in South Africa that sexual networks in bars, shops, nightclubs, hotels, and other places where people meet new partners are extensive, diffuse, and have high rates of partnership formation.
Our findings should be interpreted in light of the following potential limitations. First, because of the cross-sectional design of our study, the observed associations may not be causal. Furthermore, OR may overestimate risk estimates in cross-sectional studies. Second, although sexual-behavior questionnaires were administered by trained study staff, our results might have been affected by social desirability bias. In addition, we did not collect information about sexual behavior of male partners of women involved in this study. Third, the validity of CAGE for screening problem drinking has been assessed in only few studies in Africa, and results are not consistent across studies.26,27 Thus, the utility of this tool in our study population is unknown. Last, we recruited women from bars/hotels located in 7 out of 15 administrative wards of Moshi. This might result in selection bias, and our results may not be generalizable to all women working in these settings in Moshi and other women in the general population. On the other hand, we took a careful and systematic sampling approach, identifying all the hotels and bars in these wards, and recruited an exceptionally high proportion (91%) of eligible women.
Based on these findings, the main elements of an effective intervention program in this population should include strategies to reduce STDs, number of sexual partners, and alcohol consumption and an increase in condom usage. Prevention of other STDs, including development of effective HSV-2 control for HIV-1 prevention, should be given the highest priority. This includes increased awareness of STDs, early detection of infections, and proper treatment, including use of antiviral drugs in the GUD syndromic algorithms. In addition, ongoing trials on HSV-2 suppression therapy for prevention of HIV-1 will help to elucidate the effectiveness of this intervention. Efforts aiming at reducing the number of sex partners and promotion of safer sexual practices need to take into account the social context of these women. Female-controlled methods will be especially useful because most women in these settings are not in steady relationships and are often unable to discuss safer sexual practices with their partners, including consistent condom use. Despite their common occupational affiliations, the characteristics of women working in these settings are just as variable as in the general population. Thus, intervention efforts should be focused in most affected geographical areas and target younger women who are just starting to work in these settings.
Finally, due to the nature of their work, women employed in the bars/hotels are expected to sit and drink with their customers in order to sell more drinks. Because of this, the majority of women working in these settings drink alcohol regularly and 17% were classified as problem drinkers. Thus, programs aiming at reducing alcohol use in this population are urgently needed. Individual-level interventions should be combined with structural interventions such as prohibiting women from using alcohol during work hours and HIV/AIDS educational messages targeting male patrons of these establishments.
1. 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; 29:409–417.
2. Riedner G, Rusizoka M, Hoffmann O, et al. Baseline survey of sexually transmitted infections in a cohort of female bar workers in Mbeya Region, Tanzania. Sex Transm Infect 2003; 79:382–387.
3. Mgalla Z, Pool R. Sexual relationships, condom use and risk perception among female bar workers in north-west Tanzania. AIDS Care 1997; 9:407–416.
4. Msuya SE, Mbizvo E, Stray-Pedersen B, et al. Reproductive tract infections and the risk of HIV among women in Moshi, Tanzania. Acta Obstet Gynecol Scand 2002; 81:886–893.
5. United Republic of Tanzania, National AIDS Control Programme. HIV/AIDS/STI Surveillance Report No 16. Dar es Salaam, Tanzania: 2001.
6. Ewing JA. Detecting alcoholism: the CAGE questionnaire. JAMA 1984; 252:1905–1907.
7. Amsel R, Totten PA, Spiegel CA, et al. Nonspecific vaginitis: diagnostic criteria and microbial and epidemiologic associations. Am J Med 1983; 74:14–22.
8. Hosmer DW, Lemeshow S. Applied Logistic Regression. New York, NY: John Wiley & Sons; 1989.
9. Davis KR, Weller SC. The effectiveness of condoms in reducing heterosexual transmission of HIV. Fam Plann Perspect 1999; 31:272–279.
10. Holmes KK, Levine R, Weaver M. Effectiveness of condoms in preventing sexually transmitted infections. Bull World Health Organ 2004; 82:454–461.
11. Corey L, Handsfield HH. Genital herpes and public health: addressing a global problem. JAMA 2000; 283:791–794.
12. Flemming DT, Wasserheit JN. From epidemiological synergy to public health policy and practice: the contribution of other sexually transmitted diseases to sexual transmission of HIV-1 infection. Sex Transm Infect 1999; 75:3–17.
13. Corey L, Wald A, Celum CL, et al. The effects of herpes simplex virus-2 on HIV-1 acquisition and transmission: a review of two overlapping epidemics. J Acquir Immun Defic Syndr 2004; 35:435–445.
14. Laurence J. Molecular interactions among herpes viruses and human immunodeficiency viruses. J Infect Dis 1990; 162:338–346.
15. Klouman E, Masenga EJ, Sam NE, et al. Asymptomatic gonorrhea and chlamydia infection in a population-based and work-site based sample of men in Kilimanjaro, Tanzania. Int J STD AIDS 2000; 11:666–674.
16. Klouman E, Masenga EJ, Klepp KI, et al. HIV and reproductive tract infections in a total village population in rural Kilimanjaro, Tanzania: women at increased risk. J Acquir Immun Defic Syndr Hum Retrovirol 1997; 14:163–168.
17. Mbulaiteye SM, Ruberantwari A, Nakiyingi JS, et al. Alcohol and HIV: a study among sexually active adults in rural southwest Uganda. Int J Epidemiol 2000; 29:911–915.
18. Bagasra O, Kajdacsy-Balla A, Lischner HW. Effects of alcohol ingestion on in vitro susceptibility of peripheral blood mononuclear cells to infection with HIV and of selected T-cell functions. Alcohol Clin Exp Res 1989; 13:636–643.
19. Bagasra O, Whittle P, Kajdacsy-Balla A, et al. Effects of alcohol ingestion on in vitro susceptibility of peripheral blood mononuclear cells to infection with HIV-1 and on CD4 and CD8 lymphocytes. Prog Clin Biol Res 1990; 325:351–358.
20. Bagasra O, Kajdacsy-Balla A, Lischner HW, et al. Alcohol intake increases human immunodeficiency virus type 1 replication in human peripheral blood mononuclear cells. J Infect Dis 1993; 167:789–797.
21. Chen H, Zha J, Gowans RE, et al. Alcohol enhances HIV type 1 infection in normal human oral keratinocytes by up-regulating cell-surface CXCR4 coreceptor. AIDS Res Hum Retroviruses 2004; 20:513–519.
22. Boisier P, Ouwe Missi Oukem-Boyer ON, Amadou Hamidou A, et al. Nationwide HIV prevalence survey in general population in Niger. Trop Med Int Health 2004; 9:1161–1166.
23. Kapiga SH, Shao JF, Lwihula GK, et al. Risk factors for HIV infection among women in Dar-es-Salaam, Tanzania. J Acquir Immun Defic Syndr 1994; 7:301–309.
24. Morris M. Sexual networks and HIV. AIDS 1997; 11(suppl A):S209–216.
25. Weir SS, Pailman C, Mahlalela X, et al. From people to places: focusing AIDS prevention efforts where it matters most. AIDS 2003; 17:895–903.
26. Siegfried N, Parry CDH, Morojele NK, et al. Profile of drinking behaviour and comparison of self-report with the CAGE questionnaire and carbohydrate-deficient transferring in a rural Lesotho community. Alcohol Alcohol 2001; 36:243–248.
27. Claassen JN. The benefits of the CAGE as a screening tool for alcoholism in a closed rural South African community. S Afr Med J 1999; 89:976–979.