Thirteen studies (509 wrists in 373 patients) were included.
None of the studies completely described the clinical characteristics of the patients, or the methods used to diagnose CTS. One study only included patients with normal nerve conduction study results. The internal validity score was positively correlated with the year of publication (rs=0.57, P=0.04). Most of the studies described the MRI method well (scores ranged from 3 to 10; median 9).
The specific median nerve MRI findings that were significantly more common in cases than in controls were enlargement of the cross-sectional area (4 studies), flattening (3 studies), the combination of enlargement plus flattening (1 study), and increased T2 signal (7 studies). Other findings that were significantly more common in cases compared with controls were bowing of the flexor retinaculum (6 studies), synovial swelling (1 study) and anatomic variants (absent hamulus, median nerve interposition and double-branching of the median nerve; 1 study).
Nine studies had adequate data for calculating sensitivity and specificity.
For enlargement of the cross-sectional area, the sensitivity was 35% and the specificity was 84%.
For median nerve flattening, the sensitivity was 54% and the specificity was 95%.
For bowing of the flexor retinaculum, the sensitivity was 70% and the specificity was 93%.
For increased T2 signal, the sensitivity was 75% and the specificity was 66%.
The results from the two studies with the highest validity scores were similar to the overall results. The results from the five studies with internal validity scores of 2 or 3 were also similar to the overall results. Three studies reporting different results scored 1 or 2 on internal validity, 2 to 5 for external validity, and 7 to 9 for MRI methods.