The patient categorisation tool: psychometric evaluation of a tool to measure complexity of needs for rehabilitation in a large multicentre dataset from the United Kingdom

Purpose: This first psychometric evaluation of the Patient Categorisation Tool examined its properties as an instrument to measure complexity of needs in a mixed population of patients presenting for specialist neurorehabilitation. Materials/methods: Analysis of a large multicentre cohort of patients (n = 5396) from the national clinical dataset representing 63 specialist rehabilitation services across England. Structural validity was examined using exploratory and confirmatory factor analysis. Concurrent and criterion-validity were tested through a priori hypothesized relationships with other validated measures of resource requirements and dependency. Results: All but two items loaded strongly onto a single principal component with Cronbach's alpha 0.88. A total score of >= 30 identified patients with complex (category A) needs with sensitivity 76% and specificity 75%. However, confirmatory factor analysis provided a better fit when the scale was split into two subscales - a 'Cognitive/psychosocial' and a 'Physical' sub-scale (alpha 0.83 and 0.84, respectively). Moderate convergent and discriminant correlations were consistent with hypothesized relationships. Conclusions: The findings provide some overall support for the Patient Categorisation Tool as a unidimensional tool for measuring complexity of needs for neurorehabilitation, but the subscales may be more suitable for certain groups of patients. Further analysis is now required to evaluate its performance in different conditions.