The aim of this study was to identify risk factors associated with HIV infection among young people aged 15-24 years, which included individual and sexual behavioural factors in relation to the individual’s geographic locality. It was found that the association between the some demographic and behavioural variables was significant in the univariate analysis but non-significant in the multivariable analysis. It is of interest to note that the effect of behavioural variables on the risk of HIV infection in the multivariable logistic models is reduced after controlling for the demographic factors. Efforts to control the spread of HIV/AIDS among young persons should focus on eradicating behavioural factors which have been known to propagate the epidemic .
A Bayesian framework was applied that allowed for estimating association at individual and cluster level in an integrated framework . Spatial effects, calculated through this framework, greatly influenced the distribution of HIV/AIDS infection in Uganda and signified underlying factors that may not necessarily be captured by data collection tools in many surveys but are specific to certain locations which may increase or decrease the association with HIV/AIDS . Therefore spatial analysis is pertinent to the understanding of disease variations in different locations.
This study supports the known perceptions that HIV/AIDS continues to be a significant public health issue in Sub-Saharan Africa and is largely propagated by high-risk sexual behaviour [4, 9, 11, 41–43]. The estimation of HIV prevalence among young people from this study is similar to that reported in the AIS in Uganda  . The overall results showed that HIV risk was higher among currently and previously married individuals, low circumcision rates and high-risk sexual behaviours. These findings are consistent with those observed in previous studies based on DHS data [12, 14, 17, 39, 41, 44]. A possible explanation for increased likelihood for HIV infections among divorced/separated individuals could be that previously married individuals tend to have more sexual partners than single or married individuals . In addition, HIV/AIDS and associated unsafe sexual practices might contribute to a marriage breaking down, contributing to the observed association . A study by Clark et al (2006) in Africa and Latin America also observed that married young persons aged 15–24 years had a higher risk of HIV infection, approximately five times when compared with their sexually active unmarried peers . These young persons could most probably have married young and transitioned from virginity to frequent unprotected sex, which they would likely continue to engage in after the end of marriage. A possibility that some of the young persons could have been infected while married and could have been separated or divorced at the recognition that they were infected with HIV could also arise [39, 46]. Programmes and interventions for the control of HIV/AIDS should also focus on young widowed and divorced young persons as well as promoting approriate prevention strategies such as condom use and abstinence from sexual activities in order to prevent contracting HIV or other STIs [31, 39, 41, 46].
The results also confirmed the significant association between early sexual debut and increased likelihood of being infected with HIV. Young persons who engage in early sexual relationships are likely to have more sexual partners than their counterparts whose sexual debut occurs later predisposing them to contracting HIV and other STIs. This observation is consistent with findings from Zimbabwe which observed that a younger coital debut was associated with an increased likelihood in HIV infection compared to a later coital debut .
The results further confirmed that male circumcision and condom use reduced the risk of HIV infection in the study population. These findings are supported by three randomized clinical trials conducted in Kenya , Uganda  and South Africa  which showed an average reduction in HIV infection by 60% in circumcised compared to uncircumcised males. A recent study in Uganda revealed that although male circumcision lowered the HIV prevalence, it significantly increased high-risk sexual behaviour .
Our study employed methods of spatial analysis to evaluate the relations between spatial distributions and HIV prevalence among young people in Uganda. The importance of cluster analysis in epidemiology is the detection of aggregates of diseases as well as testing for the presence of available significant clusters by ascertaining whether diseases found in the same geographical location may be explained by chance or random occurrences [20, 49]. This is because most risk and health-promoting behaviours are clustered in specific communities as opposed to whole communities . The results of this study provide useful information on the existing epidemiological situation of HIV/AIDS in Uganda by highlighting the geographic differences of HIV infection in Uganda. Knowledge on the presence of HIV clusters at regional level can assist regional authorities in strengthening measures that effectively control the spread of HIV/AIDS among young people and mapping out future strategies . Although HIV/AIDS in Uganda is generalized, this study revealed the Central and Eastern regions as being the most probable regions for new HIV infections. Health authorities should therefore investigate why these regions are more affected and identify the geographic factors propagating HIV/AIDS among young people.
Strengths and limitations
The major strength of this study was use of spatial analytical techniques had advantages over standard statistical techniques to identify geographical variations of HIV prevalence in Uganda. This may be of public health significance in the fight against the spread of HIV/AIDS not only in Uganda but in other countries gravely affected by this scourge. Cluster analysis using the Scan Statistic method adjusts for population density and reduces selection bias as the clusters are explored without subsequent knowledge of their location, size or time period . The use of the Bayesian approach, by adding a spatial random effect, reduces bias and inaccurate conclusions that would arise from ignoring spatial auto-correlation present in the associated factors . As inherent with all cross-sectional studies, this study could neither establish temporality nor causality of the observed associations with the outcome. Self-reporting of sexual behaviours could have introduced recall or social desirability bias. Performing cluster analysis in areas of Uganda where the survey communities were further apart may possibly render the spatial scan statistic less robust. Finally, lack of geocodes in sampled areas failed to depict a correct representation of cluster analysis results with the Western most area of Uganda devoid of any spatial HIV clusters.