Braun T.1,2, Marks D.3, Thiel C.1, Grüneberg C.1
1Hochschule für Gesundheit, Department of Applied Health Sciences, Bochum, Germany, 2University of Cologne, Medical Faculty, Cologne, Germany, 3Rehaklinik Zihlschlacht, Physioherapy, Zihlschlacht, Switzerland
Background: Regaining post-stroke mobility is a primary goal in stroke rehabilitation [Paolucci 2008, Langhorne 2011]. The de Morton Mobility Index (DEMMI) is an established and unidimensional measurement instrument of mobility in acute older medical patients [de Morton 2008, Braun 2015a] and other sub-acute rehabilitation settings [de Morton 2013, Johnston 2013, New 2016]. There is preliminary, yet insufficient evidence for the DEMMI to be valid and reliable in stroke rehabilitation as well [New 2016, Braun 2015b].
Purpose: To analyse the validity and reliability of the DEMMI in patients with a sub-acute stroke seeking inpatient rehabilitation.
Methods: This cross sectional study was conducted in a neurological rehabilitation hospital. We included adult patients within 6 months post stroke. In the first session, the DEMMI and a comprehensive set of motor function measures were assessed. A second rater performed the DEMMI independently in a convenient sub-sample.
Reliability was assessed by means of internal consistency and inter-rater reliability. Rasch analysis was performed to check the DEMMIs hierarchical order, internal validity and logistic item structure [Tennant 2007]. Construct validity was assessed by testing 13 hypotheses about expected correlations between the DEMMI and other constructs (e.g. mobility, ambulation, balance) and expected differences of DEMMI scores between groups defined by mobility related characteristics.
Results: The validity sample included 109 individuals with stroke. The mean age of participants was 66 ± 15 years, 60% were male, 79% had a cerebral infarction, the mean time since stroke was 27 ± 18 days and the median National Institutes of Health Stroke Scale Score was 3 (interquartile range: 1-7) points.
Cronbachs alpha, a measure of internal consistency reliability, was 0.94. Rasch analysis indicated unidimensionality with an overall fit to the model (p = 0.577), no misfitting items, item fit residuals and person fit residuals below ± 2.5 and no Differential Item Functioning (DIF) by age, gender, disease duration or stroke severity.
All hypotheses (100%) on construct validity, including convergent and known-groups validity, were confirmed. The DEMMI showed strong correlations with other measures of mobility and physical functioning (Spearmans rho between 0.76 and 0.96).
The inter-rater reliability sample included 51 patients with stroke. The intraclass correlation coefficient (ICCAGREEMENT) between two physical therapists was 0.95 (95%CI: 0.92 to 0.97). The minimal detectable change with 90% confidence, as an aspect of interpretability, was 13 points on the 100-point DEMMI scale. No floor or ceiling effects were evident.
Conclusion(s): These results indicate excellent [Terwee 2007] internal consistency reliability, inter-rater reliability and construct validity of the DEMMI in patients with a sub-acute stroke. Rasch analysis indicates the DEMMI to be an unidimensional scale of mobility post stroke.
Implications: This study indicated evidence for the DEMMI to be an unidimensional, valid and reliable measurement instrument of mobility in sub-acute stroke rehabilitation. Further studies on the DEMMI in stroke rehabilitation should focus on the DEMMIs responsiveness and other aspects of interpretability.
Funding acknowledgements: None.
Topic: Neurology: stroke
Ethics approval: Swiss Ethics Committee on research involving humans (Kantonale Ethikkommission Thurgau, registration number: 2013/13).
All authors, affiliations and abstracts have been published as submitted.
