Objective: To determine the interrater reliability of the Donovan system for classification of pain in spinal cord injury (SCI) as well as the clinician-based usefulness of each of the classification criteria used in the Donovan system.
Design: Information pertinent to the Donovan system was provided incrementally by videotape for each pain site. After each additional piece of information, the 3 raters classified the pain site into 1 of 5 types and gave a confidence rating (5-point Likert scale) regarding the accuracy of their classifications. Thus, each pain site was classified 6 separate times, each with an associated confidence rating.
Setting: Academic rehabilitation hospital.
Participants: Twenty-eight persons with traumatic onset SCI reported 60 pain sites.
Interventions: Not applicable.
Main outcome measure: The short-form McGill Pain Questionnaire.
Results: Interrater agreement ranged from 50% to 70%. Interrater agreement did not change as additional information was provided. In contrast, confidence ratings significantly increased as additional information was provided.
Conclusions: There was considerable variability between raters using the Donovan system for classifying SCI pain. Additional clinical information increased the rater's confidence in the accuracy of their ratings but did not improve interrater agreement.