Abstract
Neighborhood-level structural interventions are needed to address HIV/AIDS in highly affected areas. To develop these interventions, we need a better understanding of contextual factors that drive the pandemic. We used multinomial logistic regression models to examine the relationship between census tract of current residence and mode of HIV transmission among HIV-positive cases. Compared to the predominantly white high HIV prevalence tract, both the predominantly black high and low HIV prevalence tracts had greater odds of transmission via injection drug use and heterosexual contact than male-to-male sexual contact. After adjusting for current age, gender, race/ethnicity, insurance status, and most recently recorded CD4 count, there was no statistically significant difference in mode of HIV transmission by census tract. However, heterosexual transmission and injection drug use remain key concerns for underserved populations. Blacks were seven times more likely than whites to have heterosexual versus male-to-male sexual contact. Those who had Medicaid or were uninsured (versus private insurance) were 23 and 14 times more likely, respectively, to have injection drug use than male-to-male sexual contact and 10 times more likely to have heterosexual contact than male-to-male sexual contact. These findings can inform larger studies for the development of neighborhood-level structural interventions.