Item Bank Development to Assess Cognitive Impairment in Patients with Acquired Brain Injury

Purpose: The aim of this study was to develop an item bank to assess cognitive impairment in patients with acquired brain injury. Methods: 294 patients with acquired brain injury answered some questions from 3 cognitive assessment questionnaires (the Mini-Mental State Examination (MMSE), the Middlesex Elderly Assessment of Mental State (MEAMS), and the Rivermead Behavioural Memory Test (RBMT)). An exploratory factor analysis was used to identify cognitive impairment dimensions which were then subjected to Rasch analysis. Reliability of the item bank was tested by internal consistency and person separation index, external construct validity by correlations with cognitive disability (Functional Independence Measure (FIM) cognitive scale). Results: Factor analytic techniques identified one dimension named "cognitive impairment". Except for one item from MMSE, all items loaded on the dimension with a factor loading 0.40 or above. Because of the disordered thresholds, 5 items were rescored. After rescoring, only one RBMT item did not fit to the Rasch model. Most items were free of Differential Item Functioning for age, gender and education. Reliability exceeded 0.86 for the dimension. The correlation between the item bank and the FIM cognitive scale was 0.57, confirming construct validity with an expected moderate association between and cognitive disability and impairment. Conclusion: Using a combination approach of exploratory factor analysis (EFA) and Rasch analysis, this study has shown that it is possible to calibrate items to assess cognitive impairment onto a single metric in a way that can be used to provide the basis of a Computerized. Adaptive Testing application.