OBJECTIVE: To determine the most robust dimensional structure of the International Classification of Functioning, Disability and Health (ICF) categories relevant to spinal cord injury (SCI) across subgroups of lesion level, health care context, sex, age, and resources of the country.

DESIGN: A multidimensional between-item response Rasch model was used. The choice of the dimensions was conceptually driven using the ICF components from the functioning chapters and splits of the activity and participation component described in the ICF.

SETTING: Secondary analysis of data from an international, cross-sectional, multicentric study for the Development of ICF Core Sets for Spinal Cord Injury.

PARTICIPANTS: Persons with SCI (N=1048) from the early postacute and long-term living context from 14 middle/low- and high-resource countries.

INTERVENTIONS: Not applicable.

MAIN OUTCOME MEASURE: Ratings of categories of the ICF relevant for SCI were analyzed.

RESULTS: Five models were tested on the complete sample and 5 subgroups. The overall reliability of all models and reliability within dimensions of the unidimensional and 2-dimensional models were good to excellent. The ICF categories spread well along the disability scale. The model fit improvement from the unidimensional to the 2-dimensional and from the 2-dimensional to the 3-dimensional model was significant in all groups (P<.0001). The improvement, however, from a unidimensional to a 2-dimensional structure was markedly better than from a 2-dimensional to a 3-dimensional one.

CONCLUSIONS: We propose that a 2-dimensional structure separating body functions and body structures from the activity and participation categories should serve as a basis for developing clinical measures in SCI in the future.