The Threshold for Favorable Outcome should be Set According to Functional Independence Measure Score at Admission in Logistic Regression Analysis

Abstract

Background: Logistic regression analysis setting Functional Independence Measure (FIM) score of 80 points or above at discharge as favorable rehabilitation outcome is reported. But this method is problematic because whether or not FIM score at discharge becomes greater than the fixed score depends on the FIM scores at admission. Therefore, we stratified FIM scores at admission and set thresholds in each stratified group. Methods: In total, 290 patients with stroke hospitalized in a convalescent rehabilitation ward were included in this study. Their FIM scores at admission were all less than 80 points. Logistic regression analysis was performed on six explanatory variables including age, sex, type of stroke, number of days from onset to admission, motor FIM score at admission, and cognitive FIM score at admission. The objective variable was FIM at discharge (0 or 1). We defined favorable outcome in two ways. In the conventional method, FIM scores at discharge of 80 points and above were defined as the favorable outcome. In the new method, FIM scores at admission were stratified into seven groups and the median value and above of FIM scores at discharge in each group were defined as the favorable outcome. Results: FIM score at discharge was favorable if FIM scores at admission were higher in the conventional method, but independent from FIM scores at admission in the new method. Conclusion: To investigate factors influencing FIM improvement using logistic regression analysis, the threshold for favorable outcome should be set according to FIM scores at admission.