Psychometric properties of the disorders of consciousness scale

Objective To provide evidence for psychometric properties of the Disorders of Consciousness Scale (DOCS). Design Prospective observational cohort. Settings Seven rehabilitation facilities. Participants Patients (N=174) with severe brain injury. Interventions Not applicable. Main Outcome Measure DOCS. Results Initial analyses suggested eliminating 6 items to maximize psychometrics, resulting in the DOCS-25. The 25 items form a unidimensional hierarchy, rating scale categories are ordered, there are no misfitting items, and differential item functioning was not found according to sex, type of brain injury, veteran status, and days from onset. Person separation reliability (.91) indicates that the DOCS-25 is appropriate for individual patient measurement. Items are well targeted to the sample, with the difference between mean person and item calibrations less than 1 logit. DOCS-25 Rasch measures result in a 62% gain in relative precision over total raw scores. Internal consistency is very good (Cronbach α=.86); interrater agreement is excellent (intracIass correlation coefficient=.90) for both the DOCS-25 and the sensory subscales. The DOCS-25 total measure, but not subscale measures, correlates with the Glasgow Coma Scale and the Coma/Near-Coma Scales and distinguishes significantly between vegetative and minimally conscious states, indicating concurrent validity. Conclusions The DOCS-25 is psychometrically strong. It has excellent measurement precision and captures a broad range of patient function, which is critical for capturing recovery of consciousness. The sensory subscales are clinically informative but should not be reported as separate measures. The Keyform synthesizes clinical observations to visualize response patterns with potential for informing clinical decision-making. Future studies should determine sensitivity to change, examine issues of rater severity, and explore the usefulness of the Keyform in clinical practice.