Differential functioning of items: can the FIM stand up to being measured?

摘要

Background: The Functional Independence Measure (F1M) is a widely used measure of activity in subacute rehabilitation settings. This study aims to evaluate the scaling properties of the FIM using Rasch analysis in a large sample of patients participating in multidisciplinary inpatient neurological rehabilitation. Method: Patients' ability was measured on admission and discharge using the FIM by consensus of the multi-disciplinary rehabilitation team. The FIM was examinedfor unidimensionality using RUMM 2020 software. Age, sex and diagnosis were used to examine for differential item functioning (DIF).Results: Over 10 years, 1495 patients had a median (IQR) age of 48 (36-58) years and a median length of stay of 18 (18-39) days. Initial analysis revealed poor fit of the FIM motor scale (chi-square < 0.000). Both motor and cognitive scales had large ceiling effects with poor targeting in the upper scores. Ten items had disordered thresholds and 14 items displayed DIF. Discussion: It is not legitimate to derive summated scores from the FIM motor or cognitive scales. It is possible to construct unidimensional scales of patients' motor and cognitive ability from FIM raw data, but this indicates redundancy in the data our teams collect and introduces extra layers of complexity. Conclusion: It would be advantageous to develop a new item bank for rehabilitation based on qualitative interviewing of patients, with an interval scale developed iteratively using Rasch analysis, and administered using computer adaptive testing. This would produce an outcome measure fit for purpose for the future.