Racial and ethnic disparities in antidepressant drug use

Abstract

Background: Little is known about racial and ethnic disparities in health care utilization, expenditures and drug choice in the antidepressant market.

Aims: This study investigates factors associated with the racial and ethnic disparities in antidepressant drug use. We seek to determine the extent to which disparities reflect differences in observable population characteristics versus heterogeneity across racial and ethnic groups. Among the population characteristics, we are interested in identifying which factors are most important in accounting for racial and ethnic disparities in antidepressant drug use.

Methods: Using Medical Expenditure Panel Survey (MEPS) data from 1996-2003, we have an available sample of 10,416 Caucasian, 1,089 African American and 1,539 Hispanic antidepressant drug users aged 18 to 64 years. We estimate individual out-of-pocket payments, total prescription drug expenditures, drug utilization, the probability of taking generic versus brand name antidepressants, and the share of drugs that are older types of antidepressants (e.g., TCAs and MAOIs) for these individuals during a calendar year. Blinder-Oaxaca decomposition techniques are employed to determine the extent to which disparities reflect differences in observable population characteristics versus unobserved heterogeneity across racial and ethnic groups.

Results: Caucasians have the highest antidepressant drug expenditures and utilization. African-Americans have the lowest drug expenditures and Hispanics have the lowest drug utilization. Relative to Caucasians and Hispanics, African-Americans are more likely to purchase generics and use a higher share of older drugs (e.g., TCAs and MAOIs). Differences in observable characteristics explain most of the racial/ethnic differences in these outcomes, with the exception of drug utilization. Differences in health insurance and education levels are particularly important factors in explaining disparities. In contrast, differences in drug utilization largely reflect unobserved heterogeneity across these population groups.

Conclusions: Substantive racial and ethnic disparities exist in all dimensions of antidepressant drug use examined. Observable population characteristics account for most of the differences in the expenditures, with health insurance and education key factors driving differences in spending. Observable characteristics are also important in explaining racial and ethnic disparities in the probability of purchasing generics and new vs old antidepressant drugs used. Differences in total utilization are not well-explained by observable characteristics, and may reflect unobserved heterogeneity such as unobserved physician-patient relationships, mistrust, and cultural factors.

Implications for policy: Reducing differences in observable characteristics such as health insurance and education will mitigate racial and ethnic disparities in expenditures on antidepressant drug use and in the types of antidepressant used (e.g., generics vs. brands; new vs old). But these factors will have less influence in reducing racial and ethnic disparities in overall antidepressant drug utilization. To limit differences in overall antidepressant drug use, policymakers must take into account cultural factors and other sources of heterogeneity.