Relationship between antidepressant medication possession and treatment response

Abstract

OBJECTIVE: To estimate the correlation between antidepressant medication possession ratios (MPR) measured from administrative pharmacy data and changes in self-reported depression symptoms. METHODS: The sample includes 360 primary care patients enrolled in a randomized trial of collaborative care in the Department of Veterans Affairs. Treatment response at 6 months was defined as a 50% improvement in symptoms as measured by the Hopkins Symptom Checklist (SCL-20). MPRs were calculated from administrative pharmacy data. Logistic regression analysis (controlling for intervention status and casemix) was used to test the hypothesis that MPR was significantly associated with treatment response. RESULTS: Seventy percent of the patients filled an antidepressant prescription and the average MPR was 0.46. A fifth (19.2%) of the patients responded to treatment. Having an MPR > or = 0.9 was significantly correlated with treatment response (OR=2.43, CI(95)=1.29-4.57, P=.006). CONCLUSIONS: If the predictive validity of antidepressant MPR measured from administrative pharmacy data is validated in other patient populations, it could be used to estimate treatment response rates whenever it is not feasible to collect symptom data directly from patients. Thus, the effectiveness of quality improvement programs designed to increase rates of antidepressant initiation and adherence could potentially be evaluated routinely at the population or system level