Exploring the association between sexual behavior, socio-demographic and biological factors with HIV infection using data from the 2011 Uganda Aids Indicator Survey (UAIS)

New annual HIV infections have been increasing steadily in Uganda despite a decreasing HIV prevalence. This thesis aimed to identify sexual behavior, socio-demographic, biological and other factors associated with HIV infection. Data extracted from the 2011 UAIS, a nationally representative two-staged stratified sample, for sexually active respondents aged 15-59 who provided a blood sample for HIV and syphilis testing were used for modeling (n=18,395). The first model, a design-based logistic regression, was used to examine the association between gender and self-perceived risk of getting HIV interaction effect with HIV infection. The second model, a classification tree, was used to identify important complex higher level interaction effects. Overall, the prevalence of HIV in this study population was 8.15%. Results from the unadjusted analyses show males with a low compared to those with a high self-perceived risk of getting HIV were more likely to engage in high-risk sexual behaviors. After adjusting for other factors, self-perceived risk of getting HIV was significantly associated with HIV infection for uncircumcised male respondents only (p-value = 0.0018). The odds of HIV infection among uncircumcised males with a high self-perceived risk of getting HIV were 65% higher compared to uncircumcised male respondents with a low self-perceived risk of getting HIV (adjusted (adj.) OR = 1.65, 95% CI: 1.21 – 2.26). Also, circumcised males with a high self-perceived risk of contracting HIV had 49% lower odds of HIV infection when compared with uncircumcised males with a high self-perceived risk of contracting HIV (adj. OR = 0.51, 95% CI: 0.28 –0.92). Other factors that were significantly associated with HIV infection included: being married (adj. OR = 1.55, 95% CI: 1.08 – 2.24) or divorced/separated (adj. OR =2.10, 95% CI: 1.35 – 3.26) or widowed (adj. OR =3.74, 95% CI: 2.24 – 6.23) compared to never married, alcohol consumption during sex (adj. OR =1.28, 95% CI: 1.03 – 1.58), engaging in commercial or exchange sex (adj. OR =2.07, 95% CI: 1.36 – 3.17), having an STI (adj. OR =1.60, 95% CI: 1.33 – 1.93), having 2-3 (adj. OR =1.56, 95% CI: 1.17 – 2.08) or 4 or more (adj. OR =1.98, 95% CI: 1.47 –2.68) total lifetime number of sex partners compared to having one, and having 1-2 (adj. OR =1.42, 95% CI: 1.15 – 1.77) or 3-4 (adj. OR =1.32, 95% CI: 1.01 – 1.73) biological children away from home compared to not having a child away from home. The classification tree identified four potential interaction patterns. However, only having an STI × total lifetime number of sex partners was found to be statistically significant after evaluation in the design-based logistic regression model. Overall, the findings of this study provide an insight into Uganda’s HIV epidemic. A number of factors, and interactions amongst them, were found to be associated with HIV infection in 2011. Outstandingly, this data provided evidence that self-perceived risk of getting HIV is significantly associated with HIV infection for uncircumcised males only. Confirmation of these findings in prospective studies may be needed.