Improving the precision of primary care physician self-report of antidepressant prescribing

Abstract

Background: Methods to select physicians most likely to benefit from educational interventions to improve the outcome of depression have not been adequately developed.

Objective: The purpose of this study was to identify a combination of primary care provider (PCP) self-report questions to improve the precision of PCP estimates of actual antidepressant prescribing as a potential tool for PCP selection.

Methods: The total number of new and refill antidepressant prescriptions written by 124 PCPs and actually filled at pharmacies over a 2-year period were matched with telephone survey results of these PCPs completed before the 2-year period. Multiple regression techniques were used to identify a set of variables that improved upon PCPs' self-report of prescriptions.

Results: The mean for PCP-reported antidepressant prescriptions written in the last week was 7.8 (+/-11.2). The average weekly prescriptions actually filled was 6.72 (+/-5.65). Most survey variables were significantly correlated with antidepressant prescriptions. The final model included 6 variables that explained 52% of the variance in prescriptions. In addition to PCP-reported number of antidepressants prescribed, average number of primary care patients seen per week and number of patients covered by managed care were directly related to the volume of prescriptions. PCP age, percentage of patients referred immediately without treatment, and mental health services being too far away were inversely related.

Conclusions: PCP self-reports on antidepressant prescribing are reasonably accurate proxies of actual prescribing. The precision of estimates of actual prescribing can be improved by considering practice structural and financial characteristics.