A Novel Cluster Analysis Scheme for Determining Levels of Upper Extremity Functional Capacity in People with Chronic Stroke

Introduction

Outcomes or assessments in stroke are typically performed in silos of measuring impairment or dexterity. Recently, Woodbury et al. have developed a functional capacity measure that seeks to combine different types of assessments (Upper Extremity Fugl-Meyer, UEFM-impairment measure and Wolf Motor Function Test, WMFT-functional ability measure) to provide a comprehensive understanding of stroke patients' functional status. However, how functional capacity varies across different severity ranges in stroke has never been identified to build targeted upper extremity (UE) therapies. With this goal, our rehabilitation group, paired with statistical modeling experts, developed a classification scheme for the first time to define cut-off scores that differentiate levels of UE functional capacity in people with chronic stroke.

Methods

We performed cluster analysis of the UEFM+WMFT short form to identify groups of participants with similar levels of functional capacity (Range: 0-57). We further evaluated the consistency of the clusters through Silhouette analysis.

Results

We pooled seventy-one stroke survivors' baseline data from three funded interventional studies. Participants had a mean (standard deviation, SD) age of 63.6 (SD 10.4), UEFM Score of 30.1 (SD 17.5), and WMFT Functional Ability Score of 36.8 (SD 21.1). Results from our cluster analysis support a three-group classification scheme of UE functional capacity: 0-26 (Low), 27-41 (Moderate), and 42-57 (High). The average silhouette width was 0.6, indicating reasonable consistency of clusters.

Conclusion

Severity-specific functional capacity characterized using our classifier scheme can facilitate the design of targeted UE rehabilitation interventions for people with chronic stroke.