A predictor model for discharge destination in inpatient rehabilitation patients

Objective: The aims of this study were to conduct an exploratory factor analysis on admission data, identify key variables that may predict discharge to home, and create and test a predictor model using confirmatory factor analysis and structured equation modeling. Design: A secondary data analysis was conducted using a data set of 176,419 cases. Statistical analyses used included an exploratory factor analysis, confirmatory factor analysis, and structural equation modeling using variables collected upon admission. Results: The hypothesis model included ten items that loaded on a latent factor of physical performance and five items that loaded on a latent factor of cognitive performance. The final predictor model resulted in three physical performance items (grooming, toileting, and chair transfers) and four cognitive performance items (comprehension, expression, problem solving, and memory) with results of χ² (df) of 44,708.630 (11), root mean square error of approximation of 0.152, and comparative fit index/Tucker-Lewis index of 0.957/0.918. Conclusions: Four factors (admit cognitive scores, admit physical scores, age, and diagnosis category) were identified and tested. The latent factors admit cognitive performance scores and admit physical performance scores were shown to be strong predictors for discharge to home, whereas diagnosis categories and age were weak predictors in this model.