Complications of Measuring Outcomes of Depression

TO THE EDITOR: Coley et al. (1) reported various methods to calculate the outcome of depression treatments using routinely collected nine-item Patient Health Questionnaire (PHQ-9) scores in two large integrated health care systems. Their conclusion was that a response, defined as 50% or greater reduction in PHQ-9 score, was the best outcome measure because it was not influenced by baseline severity. We are wondering whether other options should also be considered.

The PHQ-9 focuses on symptoms, and perhaps a scale to measure social functioning, such as the Work and Social Adjustment Scale (WSAS), should be added because social functioning is very important for patients. The WSAS has been used in randomized controlled trials (2), and its five questions can easily be combined with the PHQ-9.

If one does not want to add another questionnaire, the available data could be investigated further. The dropout rate was high. In one center the analytic cohort was 4,120 episodes, with no follow-up for 1,561. No longer attending appointments not only may indicate that somebody is feeling better but also could imply that a patient is feeling worse (3). We agree with the authors that one should think of ways to encourage patients who do not show for appointments to fill in the questionnaire from home. However, the authors could also consider looking at the number of patients whose PHQ-9 scores indicated worsening symptoms on one or more occasions during the treatment.

The authors mention that response to the PHQ-9 is not influenced by baseline severity, unlike the other measures, and for the authors this was the main reason to prefer this measure. However, we are not sure whether this reason is sufficient to be the main determinant. Service providers and insurance companies may well prefer the providers to concentrate on more severely ill people and hence prefer an effect size greater than or equal to 0.8 as a criterion.