How well do clinicians and patients agree on depression treatment outcomes? Implications for personalized medicine.

OBJECTIVE

In order to inform outcomes assessments in personalized medicine research, we evaluated the level of agreement between self-reported (SR) and clinician-rated (CR) measures of depression severity before and after treatment with an antidepressant medication.

METHODS

We pooled data from three trials (totaling 2075 patients) assessing the efficacy of antidepressant monotherapy in major depressive disorder. Differences between CR (17-item Hamilton Rating Scale for Depression [HAM-D17]) and SR (30-item Inventory of Depressive Symptomatology-Self-Rated) scale scores were used to determine concordance between CR-SR ratings. The effect of anxiety (HAMD17 anxiety-somatization subscale score ≥7) on SR-CR agreement was also assessed.

RESULTS

The CR-SR scale agreement was good for response (κ = 0.64) and moderate for remission (κ = 0.57). Patients who rated their depression as less severe than the clinician were significantly more likely to respond to treatment than over-reporters (odds ratio = 1.62; 95% confidence interval: 1.17-2.25). Although anxiety did not impact the level of agreement, among patients with SR-CR discordance, high anxiety was associated with over-reporting of depression severity.

CONCLUSION

The levels of disagreement for response and remission were too high for CR and SR scales to be considered interchangeable for research on patient-level outcomes. Anxiety does not meaningfully impact SR-CR agreement.