Rasch analysis of the Catz-Itzkovich spinal cord independence measure

• Published in: Spinal cord 2002-08, Vol.40 (8), p.396-407
• Main Authors: ITZKOVICH, M, TRIPOLSKI, M, ZEILIG, G, RING, H, ROSENTUL, N, RONEN, J, SPASSER, R, GEPSTEIN, R, CATZ, A
• Format: Article
• Language: English
• Subjects: Activities of Daily Living; Adolescent; Adult; Aged; Aged, 80 and over; Biological and medical sciences; Cerebrospinal fluid. Meninges. Spinal cord; Cohort Studies; Data Interpretation, Statistical; Disability Evaluation; Female; Health Status Indicators; Humans; Male; Medical sciences; Middle Aged; Models, Statistical; Nervous system (semeiology, syndromes); Neurology; Reproducibility of Results; Retrospective Studies; Sensitivity and Specificity; Spinal Cord Injuries - rehabilitation; Surveys and Questionnaires - standards; Tropical medicine
• ISSN: 1362-4393; 1476-5624
• DOI: 10.1038/sj.sc.3101315

The Spinal Cord Independence Measure (SCIM) is a new disability scale developed specifically for patients with spinal cord lesions (SCL). Its original and second versions (SCIM and SCIM II) were found to be reliable and more sensitive than the Functional Independence Measure (FIM) to functional changes in SCL patients. To further validate the SCIM II, examining its components on a larger population. Retrospective cohort study. Two rehabilitation centers in Israel. Two hundred and two inpatients with SCL. Routine SCIM assessments by staff nurses. Rasch and accompanying analyses. Unidimensionality of subscales (areas of function); goodness of fit of the tasks to the Rasch model; relationship of total-patient and single-task performance-ability; usability of task categories and the order of threshold locations between them; subscale discrimination of ability and difficulty and hierarchical nature; discrimination of task-categories ability, ie, distribution of thresholds along ability levels; and differential task behavior by age, gender and examination subgroups. Four unidimensional subscales were identified, and an acceptable goodness of fit to the Rasch model was demonstrated in most of their tasks (infit mean square=0.8-1.2, outfit mean square=0.6-1.4). However, some tasks showed overfit (bathing lower body) and some showed misfit (wheelchair-car transfer). Additional analyses performed to check for reasons for less than acceptable fit revealed flaws in a minority of the outcome measures. The findings of this analysis confirm the validity and reliability of the SCIM II. To a large extent we can infer that the SCIM II construct allows for the detection of any level of disability in any patient with SCL. A few item categories, however, should be rephrased or removed.