Christopher S Elliott1, Kai B Dallas2, Dimitar Zlatev2, Craig V Comiter2, James Crew3, Kazuko Shem3. 1. Department of Urology, Stanford University Medical Center, Stanford, California; Division of Urology, Santa Clara Valley Medical Center, San Jose, California. Electronic address: chrsuz@aol.com. 2. Department of Urology, Stanford University Medical Center, Stanford, California. 3. Department of Physical Medicine and Rehabilitation, Santa Clara Valley Medical Center, San Jose, California.
Abstract
PURPOSE: In many individuals with spinal cord injury a return of volitional bladder voiding is considered more important than regaining motor function. Recently a predictive model using only composite bilateral lower extremity motor scores for levels L2-S1 (range 0 to 50) was proposed by the EMSCI (European Multicenter Study about Spinal Cord Injury) group. The model showed exceptional predictive power with an AUC of 0.912. We sought to further validate the EMSCI model in a national spinal cord injury cohort. MATERIALS AND METHODS: We created models of volitional voiding using the United States NSCID (National Spinal Cord Injury Database) for 2007 to 2016. In addition to testing lower extremity motor scores, we evaluated other patient variables that we hypothesized might affect volitional voiding. RESULTS: Volitional voiding was present in 1,333 of the cohort of 4,327 individuals (30.8%) at 1-year followup. While younger age, female gender, increased sacral sparing, improved AIS (American Spinal Injury Association Impairment Scale) classification and a more caudal sensory level predicted volitional voiding, lower extremity motor scores were most predictive (AUC 0.919). Adding the other patient characteristics did little to improve model performance (full model AUC 0.932). Further analysis of the predictive power of lower extremity motor scores suggested that while the AUC appeared to decrease in persons who were most likely to void volitionally, the performance of the predictive model remained outstanding with a combined AIS C and D AUC of 0.792. CONCLUSIONS: Our study verifies the validity of the EMSCI predictive model of volitional voiding after spinal cord injury. The differing performance of lower extremity motor scores in various AIS classifications should be noted.
PURPOSE: In many individuals with spinal cord injury a return of volitional bladder voiding is considered more important than regaining motor function. Recently a predictive model using only composite bilateral lower extremity motor scores for levels L2-S1 (range 0 to 50) was proposed by the EMSCI (European Multicenter Study about Spinal Cord Injury) group. The model showed exceptional predictive power with an AUC of 0.912. We sought to further validate the EMSCI model in a national spinal cord injury cohort. MATERIALS AND METHODS: We created models of volitional voiding using the United States NSCID (National Spinal Cord Injury Database) for 2007 to 2016. In addition to testing lower extremity motor scores, we evaluated other patient variables that we hypothesized might affect volitional voiding. RESULTS: Volitional voiding was present in 1,333 of the cohort of 4,327 individuals (30.8%) at 1-year followup. While younger age, female gender, increased sacral sparing, improved AIS (American Spinal Injury Association Impairment Scale) classification and a more caudal sensory level predicted volitional voiding, lower extremity motor scores were most predictive (AUC 0.919). Adding the other patient characteristics did little to improve model performance (full model AUC 0.932). Further analysis of the predictive power of lower extremity motor scores suggested that while the AUC appeared to decrease in persons who were most likely to void volitionally, the performance of the predictive model remained outstanding with a combined AIS C and D AUC of 0.792. CONCLUSIONS: Our study verifies the validity of the EMSCI predictive model of volitional voiding after spinal cord injury. The differing performance of lower extremity motor scores in various AIS classifications should be noted.