Mapping indicators of sexually transmitted infection services in the South African public health sector.

BACKGROUND: Prevention and early treatment of sexually transmitted infections (STIs) is a high public health priority in South Africa. In 2002 a national survey of public health care (PHC) facilities was conducted to develop measurable indicators for monitoring and evaluation of the National STI programme. In this paper we present maps of key indicators obtained from the survey, and discuss their programmatic implications. We also address some methodological issues that arise in the context of producing appropriate maps. METHODS: A national sample, stratified by subdistrict/municipality, of 962 PHC facilities was randomly selected. In each facility the senior nurse was telephonically contacted and interviewed to answer questions related to the implementation of STI prevention and management from a structured questionnaire. Responses were validated through a second phone call, and inconsistencies recorded. The following key variables were mapped: stock-outs of drugs and condoms, knowledge of correct treatment procedures, consistent record keeping, number of STI clients seen per month per 1000 adult population, number of condoms distributed per adult male, and number of trained nurses per 1000 population. Using conditional autoregressive models and Markov Chain Monte Carlo (MCMC) simulation, smoothed subdistrict level clinic responses were computed and 95% confidence limits estimated. RESULTS AND DISCUSSION: Drug stock-outs were reported by 13% [95% CI 10-15%] and condom stock-outs were reported by 4% [95% CI 2-5%] of facilities. Underlying geographical patterns of risk were more clearly observed when maps had been smoothed and were not dominated by sampling error. Smoothed maps show that there is a finite, low risk of drug stock-outs in all areas with higher risk regions more clearly identified. The maps of indicators of STI services at PHC facilities show that there are important differences in quality of service within South Africa and underscore the usefulness of facility level routine data both for local programme monitoring and planning and for providing a national "bird's eye view" of programme performance.