Examining the Unidimensionality of the Disability Rating Scale

Research Objectives: To describe the unidimensionality, rating scale structure, and precision of the Disability Rating Scale (DRS) items in adults with a Traumatic Brain Injury (TBI).

Design: Retrospective Rasch Analysis of the DRS

Setting: Post Acute Care Rehabilitation Facility

Participants: 7,106 adults (15,135 records) from three existing studies. The Severe TBI trial included 12 participants (51 records),18+ years, diagnosis of disordered consciousness not the result of cancer, tumor, or encephalopathy.

The Traumatic Brain Injury Model System data included 6,910 participants (6,910 records), moderate to severe TBI, admitted to system’s hospital within 72 hours of injury.

The Amantadine Trial included 184 participants (1,267 records), 14+ years, diagnosis of disorders of consciousness from a nonpenetrating TBI.

Interventions: Not Applicable

Main Outcome Measures: Disability Rating Scale

Results: A calibration sample of 7,106 records was randomly selecting the first or last record of each participant. Rasch Analysis was applied using WinstepsTM version 5.3.3.1. For this sample, the item hierarchy indicated 3 groups of items (body functions, self-care, and participation) with different difficulty levels. Items within each group had similar logit (l) calibrations . Group 1(range 0.25 l) - eye opening (-6.01 l) and motor response (-5.75 l), Group 2 (range 2.0 l) – communication (-2.49l), feeding (-1.01 l), grooming (-.53 l) and toileting (-.47 l) and Group 3 (range 2.7 l)-level of functioning (6.72 l) and employment (9.44 l). The mean person measure was 2.64+/- 4.07 logits. One-third of people (n=2951) were cognitively independent with self care performance but remained unemployable. An unexpectedly wide logit range min=-7.6 l, max=14.51 l (total range 21 l), an unusually high amount of variance (92%) in the residuals was explained by the Rasch dimension. A high percentage (78.9%) of misfitting participants were also found. During an additional analysis (6 Items), we removed the communication and level of functioning items to examine their role in maintaining scale integrity. Results confirmed 3 largely disconnected subsets of items. The Communication item’s rating scale categories 3 and 4 connected item group 1 to item group 2. The Level of functioning item’s rating scale categories 4 and 5 connected item group 2 to item group 3.

Conclusions: Two items, communication and level of functioning, connect 3 distinct groups of items. Items within a group evaluate a similar level of disability. For most study participants, differences in scores largely reflect differences on these 2 connecting items.

Author(s) Disclosures: Sara Lewis, MS OTR/L
Patricia Grady-Dominguez, PhD, OTR/L
Jennifer Weaver, PhD, OTR/L
Trudy Mallinson, PhD, OTR/L, FAOTA, FACRM