Social Determinant–Based Profiles of U.S. Adults with the Highest and Lowest Health Expenditures Using Clusters

Using only social determinants, we employ an unsupervised clustering methodology that can differentiate high and low expenditure individuals. There are three major implications of this work: (1) clustering algorithms can produce meaningful results; (2) clustering on individuals, not specific variables, can produce predictive clusters; and (3) including comorbidities in cluster formation adds information to better separate the highest expenditure cluster profiles. Using nationally representative data, cluster expenditure distributions are wider for the most expensive clusters and smallest for the least expensive clusters. The clusters using comorbidities show larger separation between the highest two clusters and the remaining clusters than clusters developed excluding comorbidities. Though the profiles designed are representative of U.S. adults, the approach can be applied to any insured population to reveal the impact of the profiles on utilization. Clusters formed using the data without comorbidities can profile new insureds to allow prospective management of certain individuals. The same group profiles can be used in multiple studies with different outcomes, such as inpatient or drug expenditures.