Abstract
CONTEXT: The Spinal Cord Independence Measure is a comprehensive functional rating scale for individuals with spinal cord lesion (SCL). OBJECTIVE: To validate the scores of the three subscales of SCIM IV, the fourth version of SCIM, using advanced statistical methods. STUDY DESIGN: Multi-center cohort study. SETTING: Nineteen SCL units in 11 countries. METHODS: SCIM developers created SCIM IV following comments by experts, included more accurate definitions of scoring criteria in the SCIM IV form, and adjusted it to assess specific conditions or situations that the third version, SCIM III, does not address. Professional staff members assessed 648 SCL inpatients, using SCIM IV and SCIM III, at admission to rehabilitation, and at discharge. The authors examined the validity and reliability of SCIM IV subscale scores using Rasch analysis. RESULTS: The study included inpatients aged 16-87 years old. SCIM IV subscale scores fit the Rasch model. All item infit and most item outfit mean-square indices were below 1.4; statistically distinct strata of abilities were 2.6-6; most categories were properly ordered; item hierarchy was stable across most clinical subgroups and countries. In a few items, however, we found misfit or category threshold disordering. We found SCIM III and SCIM IV Rasch properties to be comparable. CONCLUSIONS: Rasch analysis suggests that the scores of each SCIM IV subscale are reliable and valid. This reinforces the justification for using SCIM IV in clinical practice and research.