The interobserver reliability of the revised American Spinal Injury Association standards for neurological classification of spinal injury patients.

To test the interobserver reliability of clinicians using the American Spinal Injury Association's 1989 revised "Standards for Neurological Classification of Spinal Injury Patients," two quizzes were given to 15 house officers and physician faculty members of a department of Physical Medicine and Rehabilitation. The first quiz consisted of five spinal cord injury cases of varying degrees of complexity. The participants were asked to classify each case with respect to sensory level, motor level, zone of injury and Frankel classification by using the original standards. Two months later, after a brief explanation of the significant changes in the revised standards, the same group was given five slightly different cases to classify. With the use of the 1989 revision of the standards, the percent correct improved for Frankel grade (65-81%), motor level (59-85%), sensory level (71-93%) and zone of injury (31-89%). Many of the same type of errors were repeated in both quizzes, i.e., T-12 and L-1 sensory level distinction and the classification of thoracic motor levels. We conclude that the interobserver reliability for the revised ASIA standards, although improved, continues to be less than optimal. We recommend that changes clarifying sensory levels near the inguinal ligament and motor level classification with very incomplete injuries be made in the standards. In addition, training methods should be developed to improve the interobserver reliability of the standards when they are used by clinicians and researchers.