Controversy on the Association of the Critical Shoulder Angle and the Development of Degenerative Rotator Cuff Tears: Is There a True Association? A Meta-analytical Approach

Abstract

Background:

The role of the critical shoulder angle (CSA) as a risk factor for rotator cuff tear (RCT) remains controversial. Studies on the association between the CSA and RCT show considerable differences in design, and this could be responsible for the variation in study results.

Purpose:

To (1) describe the reliability of CSA measurement and (2) evaluate the results of the studies reporting the association between the CSA and RCT using meta-analytical techniques to explore potential sources of variation of study results.

Study Design:

Systematic review and meta-analysis; Level of evidence, 4.

Methods:

MEDLINE, EMBASE, and CINAHL electronic databases were searched through June 30, 2019. Case-control and cross-sectional studies reporting the association of the CSA and RCT were selected. The weighted mean difference in the CSA was estimated using a random-effects model. Prediction interval was computed to better express uncertainties in the effect estimate. Metaregression and subgroup analyses were performed to explore potential sources of heterogeneity.

Results:

A total of 14 studies, including 1154 cases and 1271 controls, were identified. Of these studies, 79% (11/14) assessed the reliability of the CSA measurement, demonstrating an excellent intraobserver (range, 0.91-0.99) and interobserver (range, 0.87-0.99) reliability. Compared with controls, cases with RCT showed larger measurements of the CSA (3.3° [95% CI, 2.3°- 4.4°]). However, there was a high heterogeneity (I² = 93%), and the 95% prediction interval (-0.4° to 7.1°) included no difference in the CSA. Results of the metaregression analysis showed a significant association of several methodological aspects with the heterogeneity. The difference in the CSA tended to be larger when only full-thickness tears were included, when no specific defined criterion for assessing radiographic viewing perspective was used, in studies with smaller sample sizes, and in studies at higher risk of bias.

Conclusion:

While the CSA can be reliably measured, the difference in the CSA between cases and controls varied from very large to modest or almost no difference. Several determinants of heterogeneity were determined. Owing to this heterogeneity, it is difficult to gain an insight into the strength and exact nature of the association between the CSA and RCT with the current evidence.