Validity and reliability of the FIM instrument in the inpatient burn rehabilitation population

Objective: To provide evidence of construct validity for the FIM instrument in the inpatient rehabilitation burn population. Design: Confirmatory factor analysis and item response theory were used to assess construct validity. Confirmatory factor analysis was performed on a 2-factor model of the FIM instrument and on a 6-subfactor model. Mokken scale analysis, a nonparametric item response theory, was performed on each of the FIM instrument's 2 major factors, motor and cognitive domains. Internal consistency using Cronbach alpha and Molenaar and Sijtsma's statistic was also examined. Setting: Inpatient rehabilitation facilities. Participants: Data from the Uniform Data System for Medical Rehabilitation for patients with an impairment code of burn injury from the years 2002 to 2011 were used for this analysis. A total of 7569 subjects were included in the study. Interventions: Not applicable. Main Outcome Measures: Comparative fit index results for the confirmatory factor analyses and adherence to assumptions of the Mokken scale model. Results: Confirmatory factor analysis provided a comparative fit index of.862 for the 2-factor model and.941 for the 6-subfactor model. Mokken scale analysis showed scalability coefficients of.681 and.891 for the motor and cognitive domains, respectively. Measures of internal consistency statistic gave values of >.95 for each major domain of the FIM instrument. Conclusions: The FIM instrument has evidence of validity and reliability as an outcome measure for patients with burn injuries in the inpatient rehabilitation setting. The 6-subfactor model provides a better fit than the 2-factor model by confirmatory factor analysis. There is evidence that the motor and cognitive domains each form valid unidimensional metrics based on nonparametric item response theory.