Development of a Multidimensional, Multigroup Measure of Self-Care for Inpatient Rehabilitation

Objective: To develop and evaluate a measure of clinician-observed and patient-performed self-care function for use during inpatient rehabilitation. Design: Retrospective analysis of self-care assessments collected by therapists using confirmatory factor analysis (CFA) followed by multidimensional item response theory (MIRT). Setting: Freestanding inpatient rehabilitation hospital in the Midwestern United States. Participants: Inpatients (N=7719) with stroke, traumatic brain injury, spinal cord injury, neurologic disorders, and musculoskeletal conditions. Interventions: Not applicable Main Outcome Measures: A total of 19 clinician-selected self-care measures including the FIM and patient-performed, clinician-rated measures of balance, upper extremity function, strength, changing body position, and swallowing. Clinicians completed assessments on admission and at least 1 interim assessment. Results: CFA was completed for 3 patient groups defined by their highest level of balance (sitting, standing, walking). We reduced the number of items by 47.5% while maintaining acceptable internal consistency; unidimensionality within each item set required development of testlets. A recursive analysis defined a self-care measure with sensitivity (Cohen d_max−min =1.13; Cohen d_last−first.=0.91) greater than the FIM self-care items (d_max−min.=0.94; d_last−first .=0.83). The CFA models provided good to acceptable fit (root mean square error of approximations 0.03-0.06). Most patients with admission FIM self-care ratings of total assistance (88%, 297 of 338) made improvements on the MIRT self-care measure that were undetected by the FIM; the FIM detected no change for 26% of these patients (78 of 297). The remaining 74% (219 of 297) improved on the MIRT-based measure an average of 14 days earlier than was detected by the FIM. Conclusions: This MIRT self-care measure possesses measurement properties that are superior to the FIM, particularly for patients near its floor or ceiling. Methods assure accommodation for multidimensionality and high levels of sensitivity. This self-care measure has the potential to improve monitoring of self-care and manage therapy effectively during inpatient rehabilitation.