Psychometric Properties and Sex Differences on the Mayo-Portland Adaptability Inventory Participation Subscale (M2PI) in Veterans With Traumatic Brain Injury

Abstract

ObjectiveTo evaluate the structural validity of the Mayo-Portland Adaptability Inventory Participation Index (M2PI) in a sample of veterans and to assess whether the tool functioned similarly for male and female veterans. DesignRasch analysis of M2PI records from the National Veterans Traumatic Brain Injury Health Registry database from 2012-2018. SettingNational VA Polytrauma System of Care outpatient settings. ParticipantsVeterans with a clinically confirmed history of traumatic brain injury (TBI) (N=6065; 94% male). InterventionsNot applicable. Main Outcome MeasuresM2PI, a 5-point Likert-type scale with 8 items. For this analysis, the 2 employment items were treated individually for a total of 9 items. ResultsThe employment items misfit the Rasch Measurement model (paid employment mean square [MnSq]=1.40; other employment MnSq=1.34) and were removed from subsequent iterations. The final model had eigenvalue 1.87 on the first contrast, suggesting unidimensionality of the remaining 7 items. Item order from least to most participation restriction was transportation, self-care, residence management, financial management, initiation, leisure, and social contact. Wright's person separation reliability for nonnormal distributions was 0.93, indicating appropriateness of M2PI for making individual-level treatment decisions. Mean person measure was −0.92±1.34 logits, suggesting that participants did not report restrictions on most items (item mean=0 logits). A total of 3.8% of the sample had the minimum score (no impairment on all items), and 0.2% had the maximum score. Four items had different item calibrations (≥0.25 logits) for female compared with male veterans, but the hierarchy of items was unchanged when the female sample was examined separately. ConclusionsThese findings suggest that, although employment is a poor indicator of participation restrictions among veterans with TBI, the M2PI is unidimensional. Because of subtle differences in scale function between male and female participants, M2PI should be part of a more thorough clinical interview about participation strengths and restrictions.