Consistency checks to improve measurement with the Hamilton Rating Scale for Depression (HAM-D)

Area of Science:

• Psychiatry
• Clinical Trials Methodology
• Psychometric Measurement

Background:

• Mood disorder symptom presentation can be subtle, complicating accurate treatment response measurement.
• Minor measurement imprecisions can accumulate, impacting the fidelity of clinical trial data.
• Ensuring logical and statistical consistency in rating scales is crucial for reliable assessment.

Purpose of the Study:

• To develop and validate flags for assessing the consistency of Hamilton Rating Scale for Depression (HAM-D17) ratings.
• To improve the accuracy and reliability of depression rating scales used in clinical trials.
• To identify potential sources of measurement error in depression studies.

Main Methods:

• An expert Working Group developed logical and statistical consistency flags for the HAM-D17.
• Flags were applied to a large dataset (95,468 administrations) from antidepressant medication trials (NEWMEDS repository).
• Flags were also tested on Monte Carlo-simulated data to assess performance under known inconsistency.

Main Results:

• Nearly 30% of HAM-D17 administrations showed at least one logical scoring inconsistency flag.
• Approximately 7% of administrations had flags indicating a need for thorough rating review.
• Almost 22% of administrations had at least one statistical outlier flag, with 7.9% having multiple flags.

Conclusions:

• Consistency flags can help detect imprecise measurements in clinical ratings.
• Reviewing and addressing flagged administrations may improve the reliability and validity of clinical trial data.
• These flags offer a systematic approach to enhance data quality in depression research.