Calibration of Impairment Severity to Enable Comparison across Somatosensory Domains

Comparison across somatosensory domains, important for clinical and scientific goals, requires prior calibration of impairment severity. Provided test score distributions are comparable across domains, valid comparisons of impairment can be made by reference to score locations in the corresponding distributions (percentile rank or standardized scores). However, this is often not the case. Test score distributions for tactile texture discrimination (n = 174), wrist joint proprioception (n = 112), and haptic object identification (n = 98) obtained from pooled samples of stroke survivors in rehabilitation settings were investigated. The distributions showed substantially different forms, undermining comparative calibration via percentile rank or standardized scores. An alternative approach is to establish comparable locations in the psychophysical score ranges spanning performance from just noticeably impaired to maximally impaired. Several simulation studies and a theoretical analysis were conducted to establish the score distributions expected from completely insensate responders for each domain. Estimates of extreme impairment values suggested by theory, simulation and observed samples were consistent. Using these estimates and previously discovered values for impairment thresholds in each test domain, comparable ranges of impairment from just noticeable to extreme impairment were found. These ranges enable the normalization of the three test scales for comparison in clinical and research settings.

Automated summary

Comparison of impairment severity across somatosensory domains is important, but the distributions of test score vary greatly, making it difficult to calibrate impairment. An alternative approach is to determine comparable locations in the psychophysical score ranges. Simulation studies and theoretical analysis were conducted to establish the expected score distributions for completely insensate responders in each domain. Comparable ranges of impairment were found, enabling the normalization of the three test scales for comparison in clinical and research settings.