A Multidimensional Approach to Measure Structural Racism and Its Association with Fatal and Nonfatal Community Firearm Violence at the Census Tract Level in New York City.

Structural racism is an important determinant of health inequities. Recent studies have operationalized structural racism using latent construct approaches to capture its multidimensionality instead of relying on single indicators as structural racism proxies. However, such studies have only captured multidimensional structural racism at larger geographies, not census tract level. A latent profile model used unidimensional indicators across the domains of criminal justice, economic status, education, employment, and housing/segregation to construct a multidimensional measure of structural racism (MMSR) for 2219 NYC census tracts from 2018-2022. The indicators were derived from publicly available datasets and reflect differences between Black and non-Hispanic White New Yorkers. We assessed the MMSR's utility by testing its association with fatal and nonfatal community firearm violence in 2018-2022, using a global Wald test and pairwise comparison z-tests. The analysis identified five multidimensional structural racism profiles in NYC. Two profiles were more advantaged across the indicators, while the other three were more disadvantaged. Census tracts in the advantaged profiles had above the NYC average of non-Hispanic White residents and above median income, while disadvantaged profiles had above average non-Hispanic Black residents and below median income. Community firearm violence significantly differed across the five profiles-the disadvantaged profiles had the highest prevalence of firearm violence. Maps visualized that disadvantaged profiles co-occurred in census tracts that also experience high levels of violence. This analysis confirms the need to capture the complexity of structural racism across multiple domains, including the heterogeneity of structural racism at a small geography in a large city.