Quantitative testing in spinal cord injury: overview of reliability and predictive validity

Object. The objective of this study was to identify commonly used physiological outcome measures and summarize evidence on the reliability and predictive validity of quantitative measures used in monitoring persons with spinal cord injury (SCI). Methods. A systematic search of PubMed through January 5, 2012, was conducted to identify publications using common outcome measures in persons with SCI and for studies that were specifically designed to evaluate the reliability and predictive validity of selected quantitative measures. Quantitative measures were defined as tests that quantify sensory and motor function, such as amount of force or torque, as well as thresholds, amplitudes, and latencies of evoked potentials that might be useful in studies and monitoring of patients with SCI. Reliability studies reporting interclass correlation coefficients (ICCs) or weighted kappa coefficients were considered for inclusion. Studies explicitly evaluating correlation between measures and specific functional outcomes were considered for predictive validity. Results. From a total of 121 potentially relevant citations, 6 studies of reliability and 4 studies of predictive validity for quantitative tests met the inclusion criteria. In persons with incomplete SCI, ICCs for both interrater and intrarater reliability of electrical perceptual threshold (EPT) were >= 0.7 above the sensory level of SCI but were less reliable below the sensory level. Interclass correlation coefficients for interrater and intrarater reliability of the Graded Redefined Assessment of Strength, Sensibility, and Prehension (GRASSP) components ranged from 0.84 to 0.98. For electromyography, the ICC was consistently high for within-day tests. The overall quality of reliability of the majority of studies was poor, due to the potential for selection bias and small sample sizes. No classic validation studies were found for the selected measures, and evidence regarding the predictive validity of the measures was limited. Somatosensory evoked potentials (SSEPs) may be correlated with ambulatory capacity, as well as the Barthel Index and motor index scores, but this correlation was limited for evaluation of bladder function recovery in 3 studies that assessed the correlation between baseline or initial SSEPs and a specific clinical outcome at a later follow-up time. All studies used convenience samples and the overall sample quality was low. Conclusions. Evidence on the reliability and validity of the quantitative measures selected for this review is limited, and the overall quality of existing studies is poor. There is some evidence for the reliability of the EPT, dermatomal SSEPs, and the GRASSP to suggest that they may be useful in longitudinal studies of patients with SCI. There is a need for high quality studies of reliability, responsiveness, and validity for quantitative measures to monitor the level and degree of SCI. (http://thejns.org/doi/abs/10.3171/2012.5.AOSPINE1296)