Measuring Outcome of Depression: It Is Complicated

IN REPLY: We appreciate the thoughtful response to our study comparing measurements of depression treatment success in a real-world setting (1). We agree that there are many questionnaires for measuring depression symptom burden that can inform patient care. Our analysis focused on the nine-item Patient Health Questionnaire (PHQ-9) because it is the most commonly used in the United States and most likely to be available in electronic health record data.

We also agree that informative loss to follow-up is an important, albeit challenging, problem to address when analyzing observational data on treatment outcome. Treatment dropout, which may be due to symptom improvement or decline, frequently occurs early (2). In our study, dropout occurred before the start of follow-up at 14 days after treatment initiation, and, as such, we had no useful symptom trajectories to compare with episodes with follow-up.

The response also proposes that some payers and providers may prefer a measure of treatment success that favors improvement in episodes with the greatest initial symptom severity. Our analysis was limited to episodes with baseline PHQ-9 scores ≥10 and, thus, excluded less substantial treatment response for episodes with mild initial symptom severity. We suggest that a fair measure of treatment success, one that neither favors nor penalizes episodes with high baseline PHQ-9 scores, is preferable. Payers and providers could then decide post hoc to attach more value to improvement in different groups.