Meta-Analysis on the Performance of Sonography for the Diagnosis of Carpal Tunnel Syndrome

Objective We aimed to review and pool recent large methodological studies evaluating the diagnosis performance of ultrasonography vs electrodiagnostic testing (EDX). Methods Using the keywords: "carpal tunnel syndrome", "ultrasound", and "validity", recent articles evaluating ultrasonography compared with a reference including EDX were selected from 4 databases (PubMed, Embase, Web of science, and BDSP) and from previous review for older articles, after 2 rounds. Relevant data for different thresholds of cross-sectional area of the median nerve were extracted from the articles to calculate the pooled sensitivity, specificity, and likelihood ratios. Different analyses were also performed to study potential sources of heterogeneity, such as calculation of area under the curve, using summary receiver operating characteristic curve. Results Among the 189 articles found, 13 articles were included. A cross-sectional area of the median nerve between 9.5 and 10.5 mm 2 (study included once only), found for 11 studies, gave the pooled sensitivity as 0.84 [0.81 to 0.87] and the likelihood ratio for a negative test as 0.21 [0.17 to 0.27]. Specificity (0.78 [0.69-0.88]) and the likelihood ratio for a positive test (3.74 [2.30-6.10]) were heterogeneous. For a threshold at 7.0 to 8.5 mm 2 , pooled sensitivity was 0.94 [0.87 to 1.00], and for 11.5 to 13.0 mm 2 specificity was 0.97 [0.91 to 1.00]. The only significant variable on potential sources of heterogeneity was the cross-sectional area of the median nerve threshold and area under the curve was 0.87 (asymmetric). Conclusions Pooling recent articles seems to confirm that sonography using cross-sectional area of the median nerve could not be an alternative to EDX for diagnosis of carpal tunnel syndrome but could give complementary results.