HIV and the urban homeless in Johannesburg

South Africa (SA) bears a significant burden of the HIV epidemic, with an estimated 18.8% of the adult population (aged 15 49 years) infected.1 However, the prevalence of HIV within an essentially homeless population, as found in the city centre of Johannesburg, is not known. Research performed predominantly in developed countries suggests a substantially higher HIV and sexually transmitted infection (STI) burden among the homeless and marginally housed.2-4 The relationship between HIV infection, employment status and homelessness is complex. It is well documented that HIV leads to neuropsychological impairment,5 potentially leading to decreased job performance and unemployment, with up to 65% of HIV-infected individuals unemployed, even in developed countries.6,7 Unemployment in turn leads to food insecurity, which has been associated with an increased risk of contracting HIV and lower CD4 counts.8 The homeless are also at risk of psychiatric illness and have a decreased awareness of HIV,3 both of which may affect uptake of prevention and/or treatment interventions.9 In addition, the majority of the homeless in SA urban settings are male, and hence more difficult to reach with HIV interventions.10 To gain a better understanding of these issues and to inform future HIV prevention strategies, we investigated the HIV prevalence and risk factors among urban homeless individuals in Johannesburg.

of which may affect uptake of prevention and/or treatment interventions. 9 In addition, the majority of the homeless in SA urban settings are male, and hence more difficult to reach with HIV interventions. 10 To gain a better understanding of these issues and to inform future HIV prevention strategies, we investigated the HIV prevalence and risk factors among urban homeless individuals in Johannesburg.

Methods
We performed a cross-sectional survey of 136 adults attending a Johannesburg inner-city centre soup kitchen and clinic, operated by a local non-governmental organisation. Participants were sampled conveniently and classified as homeless or marginally housed. Data were collected for 1 year commencing 1 April 2010. HIV testing was performed using 2 rapid finger prick tests from separate manufacturers. A third confirmatory test was used in the case of discordant results. We

Background.
There are few data on HIV prevalence and risk factors among inner-city homeless and marginally housed individuals in South Africa. Methods. We recruited 136 adults from a Johannesburg inner-city homeless clinic; mean age was 32.4 years, 129 (95%) were male, and 90 (66%) were of South African nationality. Participants were tested for HIV and answered a short demographic survey. Descriptive statistics and uni-and multivariate regression analyses were used for data analysis. Results. The HIV prevalence in the cohort was 23.5%. Transactional sex, relationship status, number of concurrent sexual partners, condom usage and history of previously treated sexually transmitted infections (STIs), living on the street, the use of alcohol or drugs, and previous exposure to voluntary counselling and testing (VCT), were not significant risk factors for HIVpositivity. Statistically significant HIV risk factors on multivariate analysis included the presence of an STI (odds ratio (OR) 5.6; p<0.01) and unemployment (OR 6.7; p<0.01). South African nationality was a significant risk factor on univariate analysis (OR 2.99; p<0.05), but not on multivariate analysis (OR 2.2; p=0.17). Conclusion. The HIV prevalence in the sample did not differ appreciably from HIV prevalence estimates in other at-risk populations in similar settings, suggesting that homelessness in a South African city alone may not be a significant risk factor for HIV infection. HIV prevention efforts cannot be restricted to behaviour change programmes, but must be more holistic, recognising the protective role that employment has on HIV incidence. also included 6 individuals who had proof of a previous HIV-positive test result. Exclusion criteria included: (i) earnings in excess of R5 000/ month; and (ii) residing in a house defined on the grounds of a solid permanent structure with basic amenities (water, electricity, sewerage) and that is under the ownership or rental of the participant or family (excluding anything considered to be a form of shelter).
Ethics approval was obtained from the Human Research Ethics Committee of the University of the Witwatersrand. HIV risk factors were identified by uni-and multivariate logistic regression analysis. Data were represented as odds ratios (ORs) with 95% confidence intervals (CIs).

Results
Sample characteristics are summarised in Table 1 6 v. 1.2; p>0.05). Of patients who tested HIV-positive, 36.3% had had a previous positive test result elsewhere, but had never been followed up in terms of CD4 count or antiretroviral therapy (ART) initiation. During post-test counselling, both HIV-positive and -negative groups reported unwillingness to disclose their status to others (59% and 50%, respectively).

Discussion
The associations between homelessness and HIV in this study contrast markedly with international literature, in terms of absolute risk and behaviour patterns. The HIV prevalence in our study was high by international standards, but not substantially greater than that of a general SA population of similar age and gender. 12,13 Furthermore, no association was found between drug use and HIV in our sample, although this is a well-documented HIV risk factor in homeless populations in Europe and North America. [2][3][4] HIV incidence in SA has only recently begun to plateau, with no consensus on the reason for this. It is unclear whether the successes of education and prevention programmes are finally being felt, or if increased access to ART and STI treatment are at play. Evidence suggests that HIV education and prevention programmes may not have had a significant effect on HIV status in our sample. For example, previous exposure to VCT did not appear to be protective against HIV, and the patterns of risk behaviour were similar between HIV-positive and -negative individuals. However, the association between HIV status and unemployment was significant, because it linked unemployment to HIV risk, rather than the more simplistic factor of homelessness.
Overall, the results of our study suggest that the dynamics of HIV among the urban homeless in SA may be different from other settings. High levels of inequality and unemployment, especially in an urban environment, lead to a unique set of risk factors which differ from that of the developed world. In SA, simply being homeless or marginally housed may not put individuals at increased risk of HIV compared with the general population. However, unemployment needs to be addressed to mitigate the effect of the HIV epidemic in this setting.