Interrater reliability of the Neuromuscular Recovery Scale for spinal cord injury.

Abstract

To determine the interrater reliability of the Neuromuscular Recovery Scale (NRS), an outcome measure designed to classify people with complete or incomplete spinal cord injury (SCI) into 4 phase-of-injury groups by assessing motor performance based on normal preinjury function and disallowing use of compensation for 4 treadmill-based items and 6 overground/mat items. Masked comparison, multicenter observational study. Outpatient rehabilitation. Raters (N=14) and a criterion standard expert assigned scores to 10 video NRS assessments of persons with SCI. The raters were volunteers from the NeuroRecovery Network. Not applicable. Interrater reliability measured with the Kendall coefficient of concordance (W). Interrater reliability was generally strong (W=.91-.98; 95% confidence interval [CI], .65-.99), while lower reliability occurred for treadmill stand retraining (W=.87; 95% CI, .06-1) and seated trunk extension (W=.82; 95% CI, .28-.94). Less experienced raters assigned slightly lower scores than the expert for most items, but the difference was less than half a point and did not weaken concordance. NRS had strong interrater reliability, a necessary first step in establishing its utility as a clinical and research outcome measure.