A Comparison of Individual Change using Item Response Theory and Sum Scoring on the Patient Health Questionnaire-9: Implications for Measurement-Based Care

Abstract

We examined change over time in depression with standard sum vs. Item Response Theory (IRT) scoring. Patient Health Questionnaire 9 item responses were extracted from the electronic health records of 5,405 people receiving depression treatment at the start of treatment and 30 to 180 days later. We used four methods to classify change: the Reliable Change Index (RCI), the 5-point change and 50% change from baseline for sum scores and the z-test for IRT scoring. The 5-point change and 50% change from baseline are both Health Effectiveness Data and Information Set measures. The z-test mostly agreed with the RCI, 5-point change or 50% change. More people had change using 5-point change or 50% change but not IRT scoring than no change using 5-point or 50% change but change using IRT scoring. Kappas between changes on IRT and sum scores ranged from 0.620 to 0.813. This difference in agreement is likely meaningful at the individual, patient level. People classified differently between IRT and sum scoring had moderate symptom change. Differences in conclusions from IRT and sum scoring may be most relevant in challenging clinical situations such as small or moderate symptom change.