Calibration of an item bank in 474 orthopedic patients using Rasch analysis for computer-adaptive assessment of anxiety

Abstract

Objective: To calibrate an item bank of anxiety-related questions for use in orthopedic patients within a computer-adaptive test. Design: This is a psychometric study. Setting: The sample of orthopedic patients was recruited in two orthopedic rehabilitation clinics in Germany. Subjects: A total of 474 orthopedic rehabilitation patients were recruited for this study. Interventions: Not applicable. Main measures: The main measure is an adapted version of an existing anxiety item pool for cardiovascular rehabilitation patients. Results: The results of the confirmatory factor analysis and Mokken analysis confirmed a one-factor structure and double monotonicity. An anxiety item bank (48 items) could be developed and calibrated using Rasch analysis. It fitted to the Rasch model with a non-significant item–trait interaction (χ2(203) = 172.59; P = .94) and was free of differential item functioning. Unidimensionality could be verified and the person separation reliability was .96. The category threshold parameters varied between 4.72 and 3.16 (7.88 logits). Conclusion: The unidimensional anxiety item bank provides the basis for a computer-adaptive test to assess a wide range of anxiety in rehabilitation patients with orthopedic diseases with very good psychometric characteristics.