Determining county-level counterfactuals for evaluation of population health interventions: A novel application ofK-means cluster analysis

Objectives Evaluating population health initiatives at the community level necessitates valid counterfactual communities, which includes having similar complexity with respect to population composition, healthcare access, and health determinants. Estimating appropriate county counterfactuals is challenging in states with large inter-county variation. We present and discuss an application of K -means cluster analysis for determining county-level counterfactuals in an evaluation of a county perinatal system of care for Medicaid-insured pregnant women. Materials and Methods Counties were described using indicators from the American Community Survey, Area Health Resources Files, University of Wisconsin Population Health Institute County Health Rankings, and vital records for Michigan Medicaid-insured births for the year intervention began (or the closest available year). We ran analyses of 1,000 iterations with random starting cluster values for each of a range of number of clusters from 3 to 10 and used standard variability and reliability measures to identify the optimal number of clusters. Results One county was grouped with the intervention county in all solutions for all iterations and thus considered most valid for 1:1 population county comparisons. Two additional counties were frequently grouped with the intervention county. However, no county was ideal for all subpopulation analyses. Practice Implications Although the K-means method was successful at identifying a comparison county, concerning intervention-comparison differences remained. This limitation of the method may be specific to this county and the constraints of a within-state study. This method could potentially be more useful when applied to other counties in and outside of Michigan.