Measuring quality of pharmacotherapy for depression in a national health care system

Objectives: This article examines the use of antidepressant medication in the treatment of major depression in the Department of Veterans Affairs (VA) during the 2001 fiscal year and considers the relationship of patient, drug, provider, and facility to adherence to medication treatment guidelines. Methods: Prescription drug records for all VA outpatients diagnosed with major depression (International Classification of Diseases, 9th edition, code 296.2 or 296.3) were collected for October 2000 through September 2001. Indicators were constructed that noted whether patients newly treated with antidepressants (ie, with no prescription in the previous 8 weeks) received at least 180 days (continuation phase) of antidepressant drug treatment (84- and 140-day measures were also considered). Logistic regression with and without center fixed effects and generalized estimation equations were used to identify patient, drug, and facility characteristics that were associated with these treatment quality indicators. Results: Of the 27,713 patients in the final sample, 54% received at least 181 days of treatment. This is higher than recent rates reported by Health Employer Data and Information Set (HEDIS) for the general population, although our measures and those used by HEDIS are not exactly the same. Women, married patients, older patients, and whites were more likely than others to have higher-quality antidepressant drug treatment. Contrary to previous research, we find few significant differences among specific antidepressant agents prescribed in this large sample. Comorbid substance abuse was associated with fewer days of treatment, whereas other psychiatric comorbidities increased the length of treatment. We found few differences resulting from provider type. Although significant differences among facilities were found in the unadjusted rates (similar to those used by HEDIS), these diminished greatly after controlling for relevant covariates. Conclusions: In the nation's largest mental health system, quality of pharmacotherapy for depression, at least by one standard measure, is relatively good. We found the specific antidepressant drug used has little impact on quality. In considering differences among facilities, we found that it is critical to control for relevant patient characteristics.