The prevalence of infection with the human immunodeficiency virus (HIV) has risen rapidly in Zimbabwe. In 1991, 18% of women attending government antenatal clinics in the Harare were HIV positive; by 1995, this proportion had risen to 30% (1,2). In 1991, AIDS became the leading cause of adult death in the capital city, Harare (3).
In 1993, we began a prospective study of HIV infection in a cohort of seronegative urban factory workers. This cohort will allow examination of interventions (including future vaccines) in a group with known incidence and risk factors for HIV infection. AIDS prevention activities in a workplace setting, which offers easy access to economically and sexually active adults, can also be assessed.
Although the risk factors for heterosexual transmission in Africa have been well described, most studies have been conducted in groups selected for their high risk (e.g. prostitutes, truck drivers, STD clinic attenders) (4-6). Relatively few studies have been done in general population groups (7-9). We report here on baseline data at intake of men who were enrolled in a prospective cohort study in Harare area factories.
Harare is the capital city of Zimbabwe, with a population of ≈1.2 million and a mainly male industrial work force of ≈100,000, from which the cohort was drawn (10). Since 1985, blood donations have been screened for HIV. Between 1985 and 1993, the HIV prevalence among first-time factory donors rose from 3 to 15% (11). This is lower than non-blood donor work place estimates of HIV prevalence, which range as high as 30% (12).
Initial recruitment was among factory workers who came forward for blood donation in 40 Harare area factories (total work force ≈5,000). Both accepted donors and those advised not to donate were eligible for the study. Donor screening excluded men found to be HIV positive at previous blood donation and used a brief questionnaire, which included questions about sexual behavior (13). As enrollment progressed, subjects were also recruited directly (without volunteering to donate blood) at factory visits. Participants agreed to provide blood specimens about three times a year, with the understanding that these would be tested for HIV-1, hepatitis B virus, and syphilis. Incentives included HIV counseling and free STD treatment at the study clinic for participants and their partners. After signed informed consent was obtained, a brief questionnaire was administered, which gathered information on sociodemographic status and sexual behavior, including sexually transmitted diseases (STDs) and condom use, in the past year.
Serum samples were tested for HIV-1 antibodies using a third-generation ELISA (Abbott Laboratories). Specimens that were reactive or borderline were retested with a different third-generation ELISA. Samples not repeatedly positive or borderline reactive on the second ELISA were retested on a separate aliquot of the original serum to confirm ELISA reactivity. If results remained inconsistent, a Western blot was performed.
Subjects were classified as HIV positive when two different ELISA tests were reactive, or by Western blot (if ELISA testing was inconclusive). ELISA testing has been shown in previous studies to correlate well with the Western blot in the Zimbabwean setting (14). Syphilis serology was determined by VDRL and confirmed by TPHA. Hepatitis screening was for hepatitis B surface antigen only.
Statistical analyses were conducted using STATA software (version 3.1) and CART (Systat version 1.01). Associations of HIV seropositivity with age, education, salary, and number of sex partners were examined both as linear and as categorical variables with cutoff points determined by nonparametric analyses in CART. Multivariate logistic regression was used to calculate adjusted odds ratios. All demographic and behavioral variables significant in the univariate analysis at the 0.10 level were entered into multivariate logistic regression models by a forward stepwise process. A backward stepwise process was used to confirm the associations. Cross-product terms were subsequently entered into the multivariate models to explore the significance of anticipated interactions.
A total of 2,717 men and 58 women were recruited between March 1, 1993 and March 1995. Because the sample size for women was small, these data are not presented. Twenty-six men (0.98% of the study population) had indeterminate ELISA results and were omitted from the analyses. The final sample contained 2,691 men, of whom 512 (19.4%) were HIV positive on a screening ELISA, confirmed by a different ELISA or by Western blot.
Acceptance as a blood donor was strongly related to HIV status. Overall, among men accepted as blood donors, 7.7% were HIV positive compared with 35.4% of men not accepted. HIV seroprevalence of potential blood donors and men recruited directly to the study did not differ. Because donor status reflects prior HIV test result and other risk factors, it was not examined as a risk factor here. Donor screening has been examined in a separate analysis (13).
Possible sociodemographic and behavioral risk factors for HIV seropositivity on intake were initially examined by calculating odds ratios in univariate analyses, shown in Table 1. The relationship between age and cumulative prevalence of HIV infection is shown graphically in Fig. 1.
As shown in Fig. 1, the relationship between age and seropositivity was nonlinear, and HIV prevalence rose steeply among young men. In men under 23 years, the unadjusted risk of HIV seropositive status rose two-fold per year of age (OR, 1.95; 95% CI, 1.41-2.71). Compared with men 25-44 years old, both younger men (18-24 years) and older men (≥45 years) were significantly less likely to be HIV seropositive (Table 1). Men who had ever been married had a higher prevalence of HIV infection than single men. Seroprevalence among widowed (60.0%), divorced (38.3%), and married men (22.3%) was significantly higher than in single men (12.6%). Marital living arrangement had a modest effect: HIV prevalence was highest among those living apart from their wives on a seasonal basis (OR, 2.32; 95% CI, 1.80-2.99) compared with single men.
Men completing ≥11 years of education were less likely to be HIV seropositive than those completing ≤10 years (OR, 0.69; 95% CI, 0.57-0.84). No significant differences in HIV prevalence were found between permanent, contract, and casual employment or by monthly salary. HIV prevalence among homeowners was significantly lower than among those renting, lodging, or staying with relatives (OR, 0.51; 95% CI, 0.38-0.70).
Behavioral Risk Factors
Self-reported histories of STDs in the previous year were significantly associated with HIV seropositivity, although the magnitude of the association differed by syndrome (Table 2). The largest odds ratios were for history of genital warts (OR, 9.37; 95% CI, 3.12-28.10) and history of genital ulcer (OR, 6.82; 95% CI, 4.76-9.76). Consistent with self-reported data, serological evidence of syphilis (determined by VDRL) and hepatitis B surface antigenemia correlated to HIV seropositivity.
Men reporting more than one sex partner in the previous year were more likely to be HIV positive than those claiming to be celibate or monogamous in the previous year (OR, 2.66; 95% CI, 2.18-3.24). Self-report of paying for sex in the last year (OR, 3.30; 95% CI, 2.66-4.10) and visiting a beerhall in the last week (OR, 2.27; 95% CI, 1.87-2.77) were also risk factors for HIV infection.
Men who reported that they had used a condom more than once in the past year had a higher risk of HIV infection than men who said they never used a condom or had used one only once (OR, 1.88; 95% CI, 1.55-2.28). Table 3 examines in more detail the relationship between condom use and HIV status. A large proportion of men said they had never used a condom in their lifetime (43.4%), and HIV seroprevalence was lowest for this group (13.9%). Men who reported that they used condoms occasionally or often had a higher prevalence of HIV infection (25.7%). Only a small minority of men reported that they had always used condoms in the past year (5.7%).
Young Men Under 23 Years of Age
Because of the steep rise in prevalent HIV infection with age, we examined separately the youngest group of men in this cohort. There were 571 men 18-22 years old, of whom 43 (7.53%) were HIV positive. As noted, within this age group, the unadjusted increment in risk with age was about two-fold per year of age. Risk factors identified in the bivariate analysis resembled those for the entire group of men. The largest odds ratios found were for genital ulcer (OR, 11.64; 95% CI, 4.90-30.78) and cash for sex (OR, 6.16; 95% CI, 3.15-12.04). Young men did not report more risk behavior than the group as a whole and were less likely to pay for sex or frequent beerhalls. Compared with the whole group, a larger proportion reported that they always used condoms (n = 54, 9.5%). Condom use in the past year was associated with HIV infection (OR, 3.81; 95% CI, 1.79-8.09).
Table 4 presents the results of the multivariate analysis. The models presented include only those variables that were significant, independent predictors of HIV seropositivity. Reported odds ratios are adjusted for the other variables included in each model.
Most of the risk factors identified in the bivariate analysis remained independent risk factors for prevalent HIV infection when adjusted for other risk factors in the multivariate model.
Married men, regardless of HIV status, were more likely than single men to report a history of STD in the previous year (OR, 2.84, 95% CI, 1.52-5.31), multiple partners (OR, 1.74, 95% CI, 1.10-2.75), frequenting a beerhall in the past week (OR, 1.96, 95% CI, 1.02-3.17). Payment for sex and condom use did not differ by marital status. There was a significant interaction between being married and paying for sex in the last year. Payment for sex had a lower adjusted OR among the married men (OR, 2.70) than among the single, divorced, and widowed men (OR, 4.60). Married men who admitted to exchanging money for sex were more likely to report always using condoms (14.3%) than single, widowed, or divorced men (7.1%).
Young Men Under 23 Years
The rise in risk with each year of age persisted when self-reported behavioral risk factors were controlled for (OR, 2.09, 95% CI, 1.41-3.12). The most important independent risk factors were history of genital ulcer (OR, 6.51; 95% CI, 2.10-20.13) and payment for sex (OR, 7.39; 95% CI, 3.26-16.75). The risk of paid sex was modified by marriage, as noted in the entire cohort (Table 5).
This study examined sociodemographic and behavioral risk factors for prevalent HIV-1 infection in a group of 2,691 male factory workers in Harare, Zimbabwe. The overall prevalence of HIV-1 infection in this group was 19.4%. The main risk factors were STDs and multiple partners. In these categories, genital ulcer and paid sex carried the largest risk. These findings are similar to those reported in a 1987 survey of Harare male factory blood donors and elsewhere in sub-Saharan Africa (4-9). Genital ulcer remained important, even after a variety of factors were controlled for. This is a repeated finding in Africa and underscores the importance of strengthening control of all sexually transmitted diseases (15,16). Improved STD treatment has been shown to reduce the incidence of HIV infection (17).
Because we measured prevalent HIV infection of unknown duration, it is difficult to interpret the relationship between HIV serostatus and age. HIV infection occurs in the setting of rising HIV prevalence in the general population. The larger proportion of men >45 years who are infection free may reflect the fact that these men were at highest behavioral risk at a time when HIV was less common in the population, as well as the removal of seroprevalent cases through illness.
Despite our inability to infer incidence from cross-sectional data, the rapid rise of prevalent infection in young men, with the risk of HIV seropositivity increasing two-fold each year, certainly reflects new acquisition of HIV infection. This suggests that young men, many only recently sexually active, have not adopted adequate AIDS prevention strategies and should be targeted in more aggressive education efforts (18-20).
Higher HIV prevalence among married men was observed regardless of age. Similar observations have been made in Uganda and Tanzania (7,8). In Uganda, marriage was associated with higher HIV prevalence, particularly among young men. The authors suggested that marriage at a young age reflected earlier onset of sexual activity. Marriage appears to be associated with more high-risk behavior, which suggests that stable mutually monogamous partnerships may be difficult to achieve. Men who lived apart from their wives were at somewhat higher risk than men whose wives resided with them. Migrant labor may have a role (21). Being a widower was associated with high risk, presumably because some men were widowed by AIDS.
The observation that men who reported using condoms never or once only were less likely to be HIV positive was puzzling. We interpreted this to mean that the decision to use condoms represented a correct personal assessment of high-risk behavior. That condom use is not a risk factor for HIV infection, but a marker of behavior, was supported by the observation that the risk associated with cash payment for sex was modified by marriage. Married men who paid for sex were at lower risk of HIV infection compared with single men, probably because they were more likely to use condoms for paid sex.
Finally, some sociodemographic predictors were important even after control of behavioral variables. Frequenting beerhalls is probably a marker for such behaviors as cash for sex or STDs, which may not be accurately reported. Home ownership was apparently associated with factors strongly protective against HIV infection not measured by other information collected. Men who owned their homes were half as likely to be HIV positive, compared with men who rented or were lodgers. Home ownership may reflect economic security, family stability, and the capacity to plan for the future. These could be strong motivational factors in the adoption of risk reduction strategies.
Acknowledgment: The authors acknowledge the assistance of Magda Kurangwa, Caroline Maposhere, Verna Mzezewa, and Atanasia Mashingaidze for collection of the interview data; Pepukai Zhou for data management; and Ocean Tobaiwa, Luanne Rogers, and Heather Smith for laboratory work. The cooperation of David Mvere and the National Blood Transfusion Service is greatly appreciated. Professors Ahmed Latif and Peter Mason made helpful comments on the manuscript. This research was supported by a grant from the U.S. Public Health Service Award No. AI 33868-02; Preparation for HIV/AIDS Vaccine Evaluation (PAVE).
1. Mahomed K, Kasule J, Makuyana D, Moyo S, Mbizvo M, Tswana S. Seroprevalence of HIV infection amongst antenatal women in greater Harare, Zimbabwe. Centr Afr J Med
2. Mbizvo MT, Chipato T, Mashu A, Makura E, Fottrell P. HIV-1 and HIV-2 prevalence and risk factors amongst antenatal women in greater Harare. Cent Afr J Med
3. City of Harare. Report of the Medical Officer for Health, 1993
. Harare: Government Printers, 1994.
4. Simonsen JN, Cameron DW, Gakinya MN, et al. Human immunodeficiency virus infection among men with sexually transmitted disease. Experience from a center in Africa
. N Engl J Med
5. Van de Perre P, Carael M, Nzaramba D, Zissi G, Kayihigi J, Butzler J. Risk factors for HIV seropositivity in selected urban-based Rwandese adults. AIDS
6. Kreiss JK, Koech D, Plummer FA, et al. AIDS virus in Nairobi prostitutes. Spread of the virus to East Africa
. N Engl J Med
7. Barongo LR, Borgdoff MW, Newell JN, et al. Intake of a cohort of urban factory workers in northwest Tanzania. Risk factors for HIV-1 infection. Trop Geogr Med
8. Nunn AJ, Kengeya-Kayondo JF, Malamba SS, Seeley JA, Mulder DW. Risk factors for HIV-1 infection in adults in a rural Ugandan community: a population study. AIDS
9. Bassett MT, Latif AS, Katzenstein DA, Emmanuel JC. Sexual behavior and risk factors for HIV infection in a group of male factory workers who donated blood in Harare, Zimbabwe. J Acquir Immune Defic Syndr
10. Central Statistical Office. Census 1992
. Central Statistical Office, 1993.
11. National Blood Transfusion Service. Annual report for the year ending 1993,
Havare, Zimbabwe, 1994.
12. AIDS worries ZISCO. The Herald,
November 27, 1990. Harare, Zimbabwe.
13. McFarland W, Kahn JG, Katzenstein A, Shamu R. Deferral of blood donors with risk factors for HIV infection saves lives in Zimbabwe. J Acquir Immune Defic Syndr Hum Retrovirol
14. Emmanuel JC, Smith HJ, Paterson LE. Western blot testing among Zimbabwean HIV patients. Vox Sang
15. Laga M, Nzila N, Goeman J. The interrelationship of sexually transmitted diseases and HIV infection. Implications for the control of both epidemics in Africa
16. Latif AS, Katzenstein DA, Bassett MT, Emmanuel JC, Marowa E. Genital ulcer and transmission of HIV-1 among couples in Zimbabwe. AIDS
17. Grosskurth H, Mosha F, Todd J, et al. Impact of improved treatment of sexually transmitted diseases on HIV infection in rural Tanzania. Randomised trial. Lancet
18. Wilson D, Chiroro P, Lavelle S, et al. Sex worker, client sex behavior and condom use in Harare, Zimbabwe. AIDS Care
19. Moyo IM, Gumbo N, Ray SC, et al. Knowledge and attitudes on AIDS relevant for the establishment of community care in the City of Harare. Cent Afr J Med
20. Campbell B, Mbizvo MT. Sexual behavior and HIV knowledge among adolescent boys in Zimbabwe. Cent Afr J Med
21. Colvin M, Abdool Karim SS, Wilkinson D. Migration and AIDS [letter]. Lancet