Selective serotonin reuptake inhibitor utilization patterns: consistency across research designs

We examined the impact of commonly applied selection criteria on the ability of patients who are initiating antidepressant therapy to reach a stable pattern, which was defined as receipt of only the initial agent at the initial dose for 90 or more consecutive days. Patients in a large US prescription database who initiated fluoxetine, paroxetine, or sertraline therapy between February and April of 1995 were categorized as with (typical design) and without (relaxed design) commonly applied selection criteria. The percentage of patients achieving a stable pattern was then determined. We found that this percentage was significantly higher with the relaxed design (typical, 28.8%; relaxed, 32.4%) and for patients initiating fluoxetine therapy (>5.5% higher than for those initiating paroxetine or sertraline therapy). The results for fluoxetine were consistent across designs, whereas comparisons between paroxetine and sertraline yielded mixed results. Therefore, the relative relationship of the stable pattern is robust across designs for fluoxetine but not for paroxetine and sertraline. Further, application of commonly applied selection criteria may make a sample less representative and reduce the measured rates of stable antidepressant use, potentially leading to underestimation of the benefits of pharmacotherapy.