Reliability and accuracy of the critical shoulder angle measured by anteroposterior radiographs: using digitally reconstructed radiograph from 3-dimensional computed tomography images.

Background: Accuracy of current standard radiographic measurement of the critical shoulder angle (CSA) is not well established. This study analyzed the reliability and accuracy of the CSA measurements obtained via anteroposterior (AP) radiographs, using a digitally reconstructed radiograph (true AP view) generated from a computed tomography image as the gold standard.

Methods: The CSA was measured on the radiographs and true AP views of 88 consecutive patients who had undergone shoulder arthroscopy for rotator cuff tears. Intraobserver and interobserver reliabilities of the CSA, measured by 2 orthopedic surgeons, were evaluated, and the average deviation of the CSA between radiographs and true AP views was calculated. Moreover, we compared the deviation of CSA between standard AP films (types A1 and C1) and nonstandard AP films (other types) against the Suter-Henninger criteria.

Results: Intraobserver and interobserver reliabilities were almost perfect on radiographs (0.96, 0.86) and true AP views (0.93, 0.85). The average deviation of CSA was 2.1° ± 1.6° for observer 1 and 2.2° ± 1.9° for observer 2. The percentage of cases with deviations of 2° or more when compared with the true AP view was 42% (37 of 88) for observer 1 and 53% (47 of 88) for observer 2. Only 22% (19 of 88) of films were standard AP films. The average deviation of CSA was not significantly different between standard and nonstandard AP films for observer 1 (standard 1.9° ± 1.3°; nonstandard 2.1° ± 1.7°; P = .76) and observer 2 (standard 1.6° ± 1.5°; nonstandard 2.4° ± 1.9°; P = .09).

Conclusion: The CSA measurements using radiography were highly congruent, but a large measurement deviation occurred between radiographs and true AP views. The clinical usefulness and role of CSA in diagnosis require careful consideration.