Anthropometric measures as screening test for carpal tunnel syndrome: ROC curves and accuracy

Objective Some anthropometric measures were reported in association with carpal tunnel syndrome (CTS). Our aim was to detect optimal cut-offs of body, hand/wrist measures to correctly identify CTS patients using receiver-operating characteristic (ROC) curves. Methods We enrolled one CTS case for two controls regardless of age and gender. CTS diagnosis was made according to clinical findings and distal conduction delay of the median nerve. Anthropometric measures were: weight, height, waist and hip circumferences, wrist depth and width, third digit length, palm length and width. Obesity indicators and hand and wrist ratios were calculated. Area under the ROC curves (AUC), sensitivity, specificity, and likelihood ratios were calculated separately for gender. To assess the role of multiple anthropometric measures, we fitted multivariable logistic regression models including age, wrist ratio, shape index, body mass index, and waist-to-hip ratio. Results 1,117 patients were included (250 cases and 474 controls among females; 120 cases and 273 controls among males). Among females, all anthropometric measures presented low accuracy (AUC≤0.64). Among males, hand ratio, shape index, and wrist-palm ratio showed moderate accuracy (AUC 0.75). The estimates from the multivariable models confirmed the well-known associations between the selected variables and the risk of CTS, but the use of multiple predictors did not dramatically improved the diagnostic performance observed for single anthropometric indexes. Conclusion The cut-offs of many anthropometric measures have limited value in clinical practice as a diagnostic tool for CTS