PART-O Properties

PART-O

The original validation study of the 24-item PART-O was based on a sample of 400 persons who received inpatient rehabilitation from one of 8 TBI Model Systems sites. Rating scale analysis indicated that the frequency of item endorsement fell along a continuum. Infit statistics were generally within the range of acceptable fit (.70 to 1.40). Person separation and reliability, as well as item separation and reliability, were satisfactory. Principal components analysis of the residuals suggested a unitary construct.

As expected, stronger correlations were obtained between the PART-O total score and measures of disability obtained during the same follow-up when compared to correlations with measures taken at discharge, though all were significant. No significant relationships were found between PART-O scores and GCS score or length of PTA, though a weak relationship was found with length of coma. No differences in scores were observed for gender, cause of injury or years post-injury, however significant differences were found for age, education, marital status, and ethnicity.

PART-O-17

The properties of PART-O-17 were investigated with three groups of participants (Bogner et al., 2011). Two groups were drawn from a sample of Coloradoans contacted for completion of the Behavioral Risk Factor Surveillance Survey (BRFSS). One of the groups was comprised of persons who identified themselves as having a disability, while the other group was comprised of persons who did not self-identify as having a disability. The third group was comprised of persons recruited from three inpatient rehabilitation centers (Craig Hospital, Rehabilitation Institute of Chicago, and Ohio State University Medical Center). Analyses were conducted only with persons under the age of 65, based on the assumption that working age participation differs substantively from participation in later years.

The distributions for both total scores, as well as the Out and About subscale score, were found to be normal for groups of persons with and without disability. The Productivity score was deviated from the normal distribution for the groups with disability. The Social Relations score was normally distributed for the rehabilitation sample, but negatively skewed for both of the BRFSS groups.

Significant differences in the expected direction were found between the groups for the total and domain scores. The two total scores are strongly correlated with each other, but low correlations were observed among domain scores.