Comparing Structural Determinants, Psychosocial Factors, and Individual-Level Risk Indicators for Women's HIV Prevention in the Urban South.

Abstract

Background: Cisgender women, particularly Black women in the Southeastern U.S., experience substantial HIV risk. While current prevention strategies emphasize individual-level risk indicators, understanding the relative contributions of structural determinants and psychosocial factors is essential. This study aimed to evaluate the effects of structural determinants, psychosocial factors, and individual indicators on HIV risk among women in a high-incidence U.S. region.

Methods: Retrospective cohort of adults assigned female at birth, with ≥1 encounter at Grady Health, a safety-net system serving two of Atlanta's Ending the HIV Epidemic counties (2012-2022). New HIV diagnoses were identified using state surveillance. Logistic regression estimated associations between structural, psychosocial and individual-level indicators and HIV. Population attributable fractions (PAFs) were derived from multivariable models.

Results: Of 333,263 women, 617 (0.19%) acquired HIV; 88% were Black, mean age was 41. Structural determinants were common and strongly associated with HIV, specifically above-median SVI (aOR=1.7 [1.5-2.0]) and housing insecurity (5.2 [3.9-6.7]). Psychosocial factors such as substance use disorder (3.8 [3.1-4.6]) and individual-level indicators like STIs [3.1 [2.2-4.4]) were also associated. PAF analyses indicated that structural determinants, specifically residence in high-prevalence HIV zip codes (PAF=63% [57%-70%]) and above-median SVI (21% [14%-27%]), accounted for a larger share of population-level HIV risk than individual-level indicators like STIs (5% [3%-7%]).

Conclusion: Structural determinants were more strongly associated with HIV and explained a greater share of population-level risk than individual-level indicators. Integrating community context into risk tools may better identify PrEP candidates and support prevention in high-incidence settings.