Improving the Accuracy and Reliability of Ratings on the Hamilton Depression Rating Scale via a Video-Based Training Program.

Introduction: The clinician-rated 17-item Hamilton Depression Rating Scale (HAM-D17) allows for a systematic severity assessment of depressive symptoms. Applying the HAM-D17 in clinical practice requires that staff members' ratings on the HAM-D17 are accurate and reliable. Here, we aimed to investigate whether such accuracy and reliability can be achieved through a brief video-based training program.

Methods: One-hundred-and-ten psychiatric hospital staff members (psychologists, medical doctors, nurses, health care workers, physio-/occupational therapists, and social workers) performed baseline HAM-D17 ratings after watching a videotaped patient interview. Subsequently, a theoretical introduction video was displayed, followed by five successive videotaped patient interviews. After watching each interview, individual ratings were conducted before a video providing the gold standard rating was displayed. Accuracy was estimated by calculating the proportion of participants whose ratings did not display a deviation from the gold standard of > 1 point on all individual HAM-D17 items and > 6 points on the HAM-D17 total score. Reliability was calculated using Gwet's agreement coefficient (AC1).

Results: At baseline and after the sixth rating session, 43% versus 70% of the staff members, respectively, rated within the acceptable deviation of the gold standard (p < 0.001). At the HAM-D17 item level, baseline reliability indices were highest for item 6 (Late Insomnia) and lowest for item 14 (Sexual Interest) (AC1 = 0.97 vs. 0.47), but both improved following training (AC1 = 0.99 vs. 0.84 at the sixth rating session).

Conclusions: Most staff members conducted accurate and reliable HAM-D17 ratings after participating in a brief video-based training program.