HIV risk assessment models use multiple risk factors to build composite index scores to evaluate population level HIV risk. In this report, four risk assessment models were applied to a dataset with demographic, biological, and behavioral risk factors from 927 individuals in high and low HIV burden zip code groups in metro Atlanta, GA. Predictive ability of the risk assessment models were evaluated by comparing their sensitivity and specificity, area under the ROC curve, and mean score difference between high-burden and low-burden zip code area. The results show that the proportion of study participants who scored high in the risk assessment method are significantly greater in high-HIV burden zip code area than in low-HIV burden zip code area in all four risk assessment models. The Clinical Decision Rule risk-scoring model showed the best predictive ability of HIV risk and Binary Risk Indicator model showed the best predictive ability in predicting the residence zip code area.
Identifer | oai:union.ndltd.org:GEORGIA/oai:scholarworks.gsu.edu:iph_theses-1498 |
Date | 13 May 2016 |
Creators | Renfroe, Joshua |
Publisher | ScholarWorks @ Georgia State University |
Source Sets | Georgia State University |
Detected Language | English |
Type | text |
Format | application/pdf |
Source | Public Health Theses |
Page generated in 0.0008 seconds