Generalized linear latent mixed modeling of functional independent measures and patient outcomes

Maduranga Kasun Dassanayake, University of Texas at El Paso

Abstract

The Functional Independent Measure (FIM) is one of the most widely accepted functional assessment measures used in the rehabilitation community. Past research studies have investigated the relationship between place of discharge, admission FIM scores or FIM difference scores, and patients' characteristics and found relationships between those variables. However, most of these studies fail to account for the multi-layered multidimensionality of the FIM and the measurement error associated with the FIM items. This study utilizes Generalized Linear Latent Mixed Models (GLLAMM) and Structural Equation Models (SEM) to assess which patient characteristics are associated with FIM difference scores and the structural relationship between admission FIM and place of discharge. With regard to the models predicting place of discharge using FIM, it is found that orthopedic patients have at least a 50% chance of being discharged home if their mean cognition score is higher than 2.5. Similarly, stroke patients have 50% or more chance of being discharged home if their mean cognition score is higher than 4. Both stroke and orthopedic patients have higher odds of being discharged home if they were admitted from home than anywhere else. Now, with regard to the models for the FIM difference, patients that were readmitted tend to increase the motor dimension of FIM by over half a point. Even though the variable age is statistically significant for both motor and cognition FIM difference scores, the change in motor and cognition FIM difference scores due to age is small and thus, practically not meaningful. There is little change in the cognition FIM scores for orthopedic patients. The factors that statistically affect the cognition FIM difference scores, probably do not have strong practical significance (possibly with exception of ethnicity). The variables admission class and ethnicity affect cognition FIM difference scores for stroke patients. Those readmitted have lower cognition FIM difference scores, while Hispanic patients have a higher difference score.

Subject Area

Biostatistics|Statistics|Kinesiology

Recommended Citation

Dassanayake, Maduranga Kasun, "Generalized linear latent mixed modeling of functional independent measures and patient outcomes" (2012). ETD Collection for University of Texas, El Paso. AAI1518194.
https://scholarworks.utep.edu/dissertations/AAI1518194

Share

COinS