Abstract
Hispanic immigrant communities across the U.S. experience persistent health disparities and barriers to primary care. We examined whether community-based participatory research (CBPR) and geospatial modeling could systematically and reproducibly pinpoint neighborhoods in Charlotte, North Carolina with large proportions of Hispanic immigrants who were at-risk for poor health outcomes and health disparities. Using a CBPR framework, we identified 21 social determinants of health measures and developed a geospatial model from a subset of those measures to identify neighborhoods with large proportions of Hispanic immigrant populations at risk for poor health outcomes. The geospatial model included four measures-poverty, English ability, acculturation and violent crime-which comprised our Hispanic Health Risk Index (HHRI). We developed a Primary Care Barrier Index (PCBI) to determine (1) how well the HHRI correlated with a statistically derived composite measure incorporating all 21 measures identified through the CBPR process as being associated with access to primary care; (2) whether the HHRI predicted primary care access as well as the statistically-derived composite measure in a statistical model; and (3) whether the HHRI identified similar neighborhoods as the statistically derived composite measure. We collapsed 17 of the 21 social determinants using principal components analysis to develop the PCBI. We determined the correlation of each index with inappropriate emergency department (ED) visits, a proxy for primary care access, using logistic generalized estimating equations. Results from logistic regression models showed positive associations of both the HHRI and the PCBI with the use of the ED for primary care treatable conditions. Enhanced by the knowledge of the local community, the CBPR process with geospatial modeling can guide the multi-tiered validation of social determinants of health and identify neighborhoods that are at-risk for poor health outcomes and health disparities.