ABSTRACT
OBJECTIVE: To determine whether response shift (a change in the self-perceived meaning of health-related quality of life [HRQL]) was present in a model of physical function over time poststroke. DESIGN: Secondary data analysis of a longitudinal observational study. SETTING: Community. PARTICIPANTS: A consecutive sample of stroke survivors (N=677) at 1, 3, 6, and 12 months poststroke was included. Sixty-seven individuals were approached, but refused. Sixty-seven percent completed the study at 12 months. Mean age was 68 years; 45% of the participants were women. INTERVENTIONS: Not applicable. MAIN OUTCOMES MEASURES: The Medical Outcomes Study 36-Item Short-Form Health Survey, Euroqol, Stroke Impact Scale, Preference-Based Stroke Index, and the Health Utilities Index. RESULTS: Structural equation modeling was used to identify response shift. A chi-square difference test between constrained and unconstrained longitudinal models suggested the presence of response shift in the data. Reprioritization response shift, a change in relative importance of domains, was observed for physical activites. Recalibration response shift, a change in internal standards of measurement, was observed in physical activities, stairs, walking, and hand function. CONCLUSIONS: Response shift has implications for the measurement of change in physical function. Measures that focus on difficulty in task performance may be sensitive to response shift, resulting in a change in perceived HRQL over time. This has implications for choosing self-perceived or performance-based measures to detect change in physical function.
Subject(s)
Hemiplegia/psychology , Hemiplegia/rehabilitation , Quality of Life/psychology , Stroke Rehabilitation , Stroke/psychology , Activities of Daily Living , Aged , Aged, 80 and over , Female , Health Status , Hemiplegia/etiology , Humans , Longitudinal Studies , Male , Middle Aged , Motor Activity , Physical Therapy Modalities , Stroke/complicationsABSTRACT
OBJECTIVE: To develop and apply a framework that uses structural equation modeling to identify response shift (RS) in data with more than two time points. STUDY DESIGN AND SETTING: The framework addresses key issues that arise when analyzing data with multiple time points using a model-based approach to test for RS: model validation, correction for multiple testing, and adoption of an exploratory or theory-driven approach to identify the type and timing of RS. Data from an observational study of 678 individuals at 1, 3, 6, and 12 months poststroke are used to demonstrate the application of the framework to a model for mental health. RESULTS: Uniform and nonuniform recalibration was identified at 6 and 12 months poststroke. CONCLUSION: Studies that identify the type and timing of RS in certain client populations are useful for planning the timing of treatment and the methods to measure RS clinically. Validation of the model and adjusting for the effects of multiple testing increases confidence in the mental health model and the resulting identification of RS.
