Determinants of prescription drug utilization in U.S. counties: Testing for flexible functional models with habit formation

The rise in US healthcare expenditures slowed recently but prescription drug spending and utilization are growing due to fewer expiring patents and greater use of specialty drugs. Due to changing health policies, this timely study uses more recent data of the US counties and the flexible Box-Cox transformation model to probe the correlates of prescription drug utilization. Also, we test statistically the rarely-entertained hypothesis on the potential role of inertia as a determinant of prescription drug use (e.g., for treating chronic diseases including diabetes, hypertension, and mental illnesses). Moreover, to account for heterogeneities, separate regression models were fitted to 2012 data partitioned into low and high utilization counties. Maximum Likelihood Estimates of the empirical models suggest that: (1) the optimal functional form model depends on the utilization intensity; (2) habit formation is the most important correlate of current prescription use; and (3) primary care physician density, and lagged income and education are additional correlates. Given the negative correlation between income and prescription drug utilization, the Box-Cox income elasticities (-0.35 and -2.25 respectively in low and high utilization groups of counties) suggest that the mostly generic prescription drugs behave as inferior goods. Therefore, relevant policies should consider differences in utilization intensities.