Reliability of an interview approach to the Functional Independence Measure

Objective: To establish the reliability of an interview approach to the Functional Independence Measure (FIM). Design: Two raters were present at the same time during the interviews in the home and did independent ratings of the 18 FIM items. The interview procedure was repealed within a week by another two raters in the clinic. Subjects: Sixty-three stroke survivors (median age 63 years, range 18-71 years) were studied approximately two years after onset. Raters: Three occupational therapists and one nurse. Methods: Reliability was calculated as unweighted kappa statistics, percentage agreement (PA), and intraclass correlation coefficient (ICC). Results: Best agreement was found in the motor items of FIM. The kappa statistics showed good to excellent inter-rater values during the same interview except for the Social interaction item. The ICCs based on sum-score for motor (0.92) and social-cognitive items (0.75) respectively, were similar to those reported in the literature. The repeated interview by different raters showed less stable agreement according to kappa values for the items dealing with transfers, locomotion and social-cognition. Conclusions: FIM assessments showed high inter-rater agreement for the same interview setting (home as well as clinic), but the stability of the measure over time with a repeated interview by different raters was somewhat less satisfactory.