Calibration of an item bank for work capacity in cardiological rehabilitation patients

Abstract

Prevention of work loss is a core objective of cardiovascular rehabilitation. However, comprehensive and economic diagnostic instruments on work limitation are missing. The present study describes the calibration of an item bank for work capacity in cardiovascular rehabilitation patients. This study was based on an item pool (166 items) for assessing work capacity in orthopaedic rehabilitation patients. A total of 349 cardiovascular rehabilitation patients were recruited from 14 German rehabilitation inpatient centres. Exploratory factor analysis was used to determine underlying dimensions of work capacity. Rasch analysis was applied to test for fit of the model, item fit, unidimensionality, particularly local independence and differential item functioning using RUMM2030. Two independent factors (cognitive work capacity/physical work capacity) were found. Fitting of the data to the Rasch model was initially poor (significant item-trait interaction for both factors; p < 0.001). After rescoring disordered categories, excluding misfitting items and items with differential item functioning, 20 and 18 items remained for cognitive work capacity and physical work capacity. Item-trait interaction (cognitive work capacity p = 0.24; physical work capacity p = 0.95) and person separation index (cognitive work capacity = 0.81; physical work capacity = 0.85) were satisfactory. Unidimensionality was given for both factors. The calibrated item bank provides the basis to assess work capacity in cardiovascular rehabilitation patients, fulfilling high psychometric standards. Further, the item bank could be either used to develop a computer adaptive test or an economic short-form questionnaire.