Effect of sexual behaviour on the spatial distribution of HIV among Kenyan youth in 2003 and 2008

A research report submitted to the Faculty of Health Sciences, the University of the Witwatersrand in partial fulfilment of the requirements for the Degree of Masters in Epidemiology in the field of Infectious Disease Epidemiology. March,2018.Background: Kenya is among the fifteen countries contributing more than 75% of the global HIV prevalence and incidence. In these countries, clustering of risk behaviour across geographical space and population sub-groups such as young people contributes significantly to the disease burden. Low levels of knowledge on HIV and rejecting attitudes towards HIV infected people influence sexual risk behaviour among young people. Objectives: The primary aim of this research study was to determine and compare the relationship between sexual behaviour and the spatial distribution of HIV among young Kenyans across two-time periods, 2003 and 2008. The secondary aim was to investigate the direct and indirect determinants of HIV distribution. Methods: Secondary data analysis was performed using cross-sectional data obtained from two two-stage population based surveys conducted in Kenya. The main outcome investigated was HIV status among young people aged 15-24 years. The study employed Bayesian-based multivariable logistic regression models adjusting for both non-spatial and spatial random effects. A conceptual framework based on knowledge, attitudes and practices (KAP) was the basis of a generalised structural equation model (GSEM) fitted to investigate the complex relationship between sexual risk behaviour and HIV status. Results: Among 2,650 and 2,857 young people aged 15-24 years in 2003 and 2008 respectively, 3.6% and 2.9% were HIV positive with the majority being female. The level of comprehensive HIV knowledge increased from 42% to 78% between 2003 and 2008. In both years, there was county spatial variability in HIV prevalence among young Kenyan people. Having a partner with concurrent sexual partners had the greatest direct effect on the odds of HIV infection (adjusted OR 2.26; 95%CI 1.63-3.12), but this was indirectly affected by knowledge levels during the latter survey year. Conclusion: This study revealed significant spatial variability in the HIV prevalence among young Kenyan youth and significant association between HIV knowledge, high-risk sexual behaviour and HIV infection. Interventions targeting sexual behaviour in this age group should focus on the counties with the greatest disease burden before implementing a wider national approach. Keywords: HIV, young people, spatial analysis, GSEM, Kenya.LG201