Cluster analysis of state Medicaid prescription drug benefit programs based on potential determinants of per capita drug expenditure

Background: Drug cost-containment strategies result in varied outcomes in different state Medicaid programs perhaps, because they differ in ways beyond the usual considerations of similarity among the states and their Medicaid populations. Objective: To classify state Medicaid programs into homogenous groups based on potential determinants of their prescription drug expenditures. Methods: Cluster analysis techniques were used to identify clusters of state Medicaid programs, and their characteristics, using variables identified as potential determinants of Medicaid prescription drug expenditures. Results: Five distinct clusters were identified with memberships ranging from 1 to 24 states. Sociodemographic factors like race/ethnicity and elderly persons in Medicaid, severity of diseases, state population covered by Medicaid, population below poverty level, and high-school graduates; access factors like access to hospitals, pharmacies, physicians, and prenatal care; and policy factors like support for publicly funded health care were identified as key characteristics that distinguished among these clusters. Per capita prescription drug expenditure was also found to be significantly different among these clusters. Conclusion: Knowledge of clusters among state Medicaid programs and their characteristics could potentially increase the chances of success for Medicaid administrative policies and interventions at lower costs of targeted implementation.