‘Behind walls’: a study of HIV risk behaviours and seroprevalence in prisons in Zambia

Abstract

Inmate populations include a large number of individuals at risk of HIV infection. However, there is insufficient data about HIV/AIDS epidemiology in prisons. Our study, conducted in Zambia, a sub-Saharan African nation with an estimated HIV prevalence of 19% in adults, was designed to address this shortfall. Between November 1998 and June 1999, we visited three of the country's main jails, Mukobeko, Kamfinsa and Solwezi prisons, to determine HIV seroprevalence and related risk behaviors. Mukobeko is a maximum-security prison keeping long-stay inmates, whereas the other two hold prisoners on short sentences. The study was voluntary and was approved by the Zambia Prison Service, the National AIDS/STD/TB and Leprosy Programme and the Zambia Permanent Human Rights Commission. After informed consent, information on HIV-related risk behaviours was obtained using a semi-structured questionnaire. Additional information on risk behaviours was obtained in focus group discussion with inmates and staff. We were particularly interested in assessing the risk of HIV transmission through sex between men, past exposure to sexually transmitted infections, injecting drug use, tattooing and the sharing of shaving equipment. HIV seropositivity was determined using two enzyme immunoassays, whereas Immutrep rapid plasma reagin/ Treponema pallidum haemagglutination assay determined syphilis serology. The study was anonymous but the HIV result was linked to the questionnaire. The proportion of inmates who were positive for HIV antibodies was calculated according to demographic and behavioural characteristics. Cross-tabulation of these variables against HIV status was then carried out and a chi-square test for significance was run to assess the extent of association. All tests significant at the 5% level of significance were then put in a logistic regression model to investigate further the association between prisoner's demographic characteristics and risk behaviours and the prevalence of HIV infection. In all, 1596 inmates out of 1859 (86%) in jail at the time of our visit took part in the study. Participants included 46 women. The mean age of participants was 30.45 (SD 9.08) years whereas the median was 28 years. A total of 1566 (98%) inmates gave samples for HIV testing, out of whom 421 (27%) were HIV seropositive. The mean age for inmates with HIV infection was 30.80 (SD 7.90) years, whereas that for those who were HIV negative was 30.37 (SD 9.51) years (P = 0.403). There was no significant difference in HIV prevalence between male prisoners, 407 out of 1523 (26.7%) and female prisoners, 14 out of 43 (32.6%) (P = 0.395). The HIV prevalence rate by selected risk factors is shown in Table 1. Because of the small number of female prisoners in the study, risk factor analysis was only performed for male inmates. Similarly, as only five inmates reported using heroin in prison, intravenous drug use may be a rare risk behaviour for HIV transmission in Zambian prisons, and was therefore considered but excluded from the table.Table 1: Risk factors for HIV infection among male prisoners in Zambia. Age, marital status, the number of irregular lifetime partners or pick-ups were all found to be significantly associated with HIV status, whereas the duration of sentence and prison of incarceration did not influence the HIV result. Among inmates aged 20 years and below, the HIV prevalence rate was 14.5% compared with 31.5% in those 30–39 years old (P = 0.014). The HIV prevalence among unmarried prisoners was 23.7% among those currently married it was 27.2% and among prisoners who were separated, divorced or widowed it was 36.3% (P = 0.021). The risk of HIV infection increased with increasing lifetime number of casual partners. Among those reporting no casual partner or only one casual partner the HIV prevalence was 25.9 and 24.5% respectively, whereas among those reporting more than nine casual partners it was 38.7% (P = 0.011). Fifty-nine (3.8%) inmates admitted to having sex (penile anal intercourse) with other men in prison. All had heterosexual relationships outside prison. However, when we asked all inmates the question, ‘how common is sex between men in prison?’ the responses included: ‘not practised at all’ 55 (3.4%); ‘a few are involved’ 334 (20.9%); ‘many are involved’ 844 (52.9%); ‘almost all are involved’ 94 (5.9%); and ‘don't know/not sure’ 269 (16.9%). In addition, information from group discussions suggested very high numbers of those involved in sex between men. No association was found between having sex with another man and HIV infection (P = 0.072). A total of 265 (17.4%) inmates had clearly visible tattoos with 69 (26.0%) of them having a positive HIV test. This result was not significantly different in those without tattoos, of whom 336 out of 1245 (27.0%) had HIV antibodies (P = 0.751). A total of 966 (63.4%) inmates admitted to sharing razor blades in prison, with 255 (26.4%) of them being HIV infected. In comparison, 115 out of 427 (26.9%) of those who never shared razors were HIV positive (P = 0.836). Of the 1523 serum samples available, 230 (15.1%) had antibodies for syphilis. Eighty-five (37.0%) samples positive for syphilis were also positive on HIV testing. The corresponding rate in syphilis-negative samples was 322 (24.9%) positive for HIV infection (P = 0.000). Table 2 shows the adjusted odds ratios using logistic regression for characteristics associated with HIV prevalence. The table shows that marital status and syphilis serology were the most important predictors of the HIV status of inmates. The age of the inmate and the number of irregular lifetime partners were found to be insignificant using the stepwise (forward and backward) method. Prisoners who were married, divorced or widowed were one and a half times more likely to have HIV infection than those who were single, whereas inmates with syphilis antibodies were twice as likely to be co-infected with HIV.Table 2: Logistic regression analysis of prevalence of HIV infection in Zambian prisons. Over one in four inmates among the 13–15 000 living in Zambia's prisons are infected with the AIDS virus. These findings are much higher than the HIV seroprevalence rate of 16.1% reported in similar studies conducted in 1988/9 [1], and the estimated HIV prevalence of 19% in adults in Zambia [2], but are comparable with the high infection rates of up to 32% in the big cities [3]. Although we have demonstrated in our study that a large number of inmates have sex with other men, and that tattooing and the sharing of razor blades are common, we believe that for now, the major source of infection among prisoners in Zambia is heterosexual intercourse outside jail. The duration of stay in prison and the prison of confinement were not linked to HIV status, but there was a significant association between HIV infection and syphilis reactivity. This picture is typical of the epidemiology of HIV infection in sub-Saharan Africa, which is common in young and sexually active men and women living in urban areas [4]. On the other hand, tattooing and the sharing of shaving equipment were not associated with infection with HIV. Even though we found no significant relationship between sex among men and HIV seropositivity, we believe that the risk of HIV transmission from penetrative anal sex is high in Zambian prisons. The lack of this association may be attributed to the small number of inmates who reported sex with other men. Although only 3.9% of inmates reported having sex with other men, the figures may be much higher. When we invited inmates to indicate how common sex between men was at their prison, over 50% replied that many were involved, and 6% said that almost all were involved. In the studies of a decade ago we reported that up to 12% of inmates had sex with other men in prison. Condoms are effective for HIV prevention [5], but they are not distributed in Zambian prisons; it is believed that their distribution would encourage homosexuality, a criminal offence in the country [6]. In this situation, and for the interim, we propose that the government should seriously consider other measures that may assist in reducing the risk of HIV transmission in prisons. In particular, the use of non-custodial sentences for young offenders must be encouraged. Urgent attention must also be paid to the poor socioeconomic conditions obtaining in most prisons, which may be contributing to the spread of HIV and other infectious diseases. Acknowledgements The authors are grateful to the Commissioner of Prisons for giving us permission to conduct the study, and to the staff and inmates of the three prisons for their support. Oscar O. Simooyaa Nawa E. Sanjoboa Lovemore Kaetanob Gibson Sijumbilab Fines H. Munkonzea Frank Tailokaa Rosemary Musondab