Development and validation of vulnerable and enabling indices for hiv viral suppression among people with hiv enrolled in the ryan white program

Abstract

Background: Numerous factors impact HIV care, often requiring consideration of indices to prevent collinearity when using statistical modeling. Using the Behavioral Model for Vulnerable Populations, we developed vulnerable and enabling indices for people living with HIV (PLWH). Methods: We used Ryan White Program (RWP) data and principal component analysis to develop general and gender-and racial/ethnic-specific indices. We assessed internal reliability (Cronbach’s alpha), convergent validity (correlation coefficient), and predictive utility (logistic regression) with non-viral suppression. Results: Three general factors accounting for 79.2% of indicators’ variability surfaced: mental health, drug use, and socioeconomic status (Cronbach’s alpha 0.68). Among the overall RWP population, indices showed convergent validity and predictive utility. Using gender-or racial/ethnic-specific indices did not improve psychometric performance. Discussion: General mental health, drug use, and socioeconomic indices using administrative data showed acceptable reliability, validity, and utility for non-viral suppression in an overall PLWH population and in gender-and racial/ethnic-stratified populations. These general indices may be used with similar validity and utility across gender and racial/ethnic diverse populations.