Twenty-one studies met the inclusion criteria (982 patients, 982 wrists; range 11 to 125). Seventeen studies assessed MRI and eight assessed magnetic resonance arthrography. Seventeen of the 21 studies reported recruiting a representative patient spectrum, six avoided progression bias, 15 avoided partial verification bias, 19 avoided differential verification bias, 19 avoided incorporation bias, 13 reported blinding of interpreters of the index test, three reported blinding of interpreters of the reference standard and 11 reported uninterpretable/intermediate results.
MRI (17 studies): The summary estimate of sensitivity for detecting acetabular labral tears was 75% (95% CI 70 to 79) and specificity was 81% (95% CI 76 to 86). Technology using 3.0 Tesla magnets (sensitivity 86%, 95% CI 65 to 97 and specificity 100%, 95% CI 87 to 100; one study) had higher accuracy than those using 1.5 Tesla (sensitivity 70%, 95% CI 64 to 75 and specificity 79%, 95% CI 72 to 85; 11 studies).
Magnetic resonance arthrography (eight studies): The summary estimate of sensitivity was 84% (95% CI 79 to 89) and specificity was 95% (95% CI 92 to 98). Technology using 3.0 Tesla magnets (sensitivity 100%, 95% CI 79 to 100 and specificity 100%, 95% CI 82 to 100; one study) had higher accuracy than those that used 1.5 Tesla (sensitivity 83%, 95% CI 78 to 89 and specificity 95%, 95% CI 91 to 98; three studies).