Measurements and recovery patterns in a multicenter study of acute spinal cord injury

Study Design. Post hoc, secondary analysis of data from 1992 to 1998 in the trial of Sygen(R) in acute spinal cord injury. Objectives. Quasi-epidemiologic understanding of measurement tools and of recovery patterns. No drug efficacy results. Summary of Background Data. Many authors have studied individual scales for measuring the severity of spinal cord injury. Methods. Emphasis on descriptive, rather than inferential, statistics: consistent with secondary analysis. Results. Of the 760 patients, 43 died within 365 days. The rate was higher for complete injuries (7.1% vs. 3.2%, P = 0.017). Marked recovery at 26 weeks was more frequent in those with better baseline American Spinal Injury Association (ASIA) Impairment Scale (AIS) scores, but was not different for methyl prednisolone within versus after 3 hours. Light touch scores improved at each visit, more so in those with higher scores at baseline. Bladder control similarly improved. Motor and sensory scores exhibited departures from assumptions underlying normal-theory statistical techniques: t test and analysis of variance. Furthermore, they were mixtures of differing distributions from different study strata, so that overall conclusions depend on the mixture of patients seen. Conclusions. The prognosis of these patients with spinal cord injury seen at 28 centers in North America during the mid-1990s appears better than was often assumed earlier. The general patterns are similar across different measurement scales, although there are intriguing differences. The patterns in different strata are different in specifies, and complete injuries do less well. Pooling data from different strata may result in probability distributions that depart from normal-theory assumptions and give misleading results depending on recruitment patterns.