Does Patient Adherence to Antidepressant Medication Actually Vary Between Physicians?

Abstract

Objective: Previous research and improvement efforts have presumed that patients' nonadherence to antidepressant medication reflects physicians' quality of care. We used population-based health records to examine whether adherence to antidepressant medication actually varies between prescribing physicians.

Methods: Electronic health records and insurance claims data from 5 integrated health systems in Washington, Idaho, Minnesota, Colorado, Hawaii, and California were used to identify 150,318 adults starting new episodes of antidepressant treatment for depression between January 1, 2010, and December 31, 2012. Early adherence was defined as any refill or dispensing of antidepressant medication in the 180 days following an initial antidepressant prescription. Patient-level demographic and clinical characteristics potentially associated with adherence were identified from health system records.

Results: Average probability of early adherence was 82% for psychiatrists and 74% for primary care physicians. Among individual physicians, the range of raw or unadjusted early adherence rates (5th to 95th percentiles) was from 33% to 100% for psychiatrists and from 0% to 100% for primary care physicians. After accounting for sampling variation and case mix differences, the range of adjusted early adherence rates (5th to 95th percentiles) was from 72% to 78% for psychiatrists and from 64% to 69% for primary care physicians.

Conclusions: After accounting for sampling variation and case mix differences, early adherence to antidepressant medication varies minimally among prescribing physicians. Early discontinuation of antidepressant treatment is not an appropriate measure of individual physician performance, and efforts to improve adherence should emphasize system-level interventions rather than the performance of individual physicians.

Figure 1 –

Distribution of number of patients per provider, stratified by provider-level rate of initial prescription refill and provider specialty. Boxes show range from range from 25^th to 75^th percentiles (with bar for median). Lines show range from 5^th to 95^th percentiles.

Figure 2 –

Variation among physicians in refill rates before and after accounting for sample size and differences in patient characteristics.