A latent class model to identify city/town chronic disease patterns.

Abstract

OBJECTIVE: Most departments of health grapple with how to most effectively allocate resources to address chronic diseases. We adapted a model created by Massachusetts to create customized city/town profiles in order to identify the patterns of chronic disease among 39 cities/towns in Rhode Island. METHODS: We used four data sources to identify 20 indicators of four domains: demographics and socioeconomic status; health behaviors and chronic diseases prevalence; no regular provider and non-emergent emergency department visits; and chronic disease-related hospitalizations. A latent class model was used to group cities/towns into distinct latent class memberships based on similar patterns of indicators. Data were analyzed in 2014. RESULTS: The latent class model differentiated three distinct classes of city/town, reflecting three levels of economic and health indicators. CONCLUSIONS: Our model was a simplified version of one constructed by Massachusetts that larger states can also use to understand chronic disease patterns among cities/towns. Chronic disease programs and policies can use the findings to direct resources toward targets not always identified by more traditional analyses.