Abstract
Objectives:
Glenoid and humeral bone loss are both known risk factors for failure after anterior shoulder stabilization. Several models have previously been proposed to characterize bone loss and its association with recurrent instability. The purpose of this study was to determine the association of failure after arthroscopic anterior stabilization with three different models of shoulder instability.
Methods:
128 individuals who underwent primary anterior shoulder stabilization between 2007-2015 were included in this study. Subjects were excluded on the basis of prior shoulder surgery, multidirectional instability, presence of connective tissue disorder, and concomitant rotator cuff pathology. Glenoid bone loss and Hill-Sachs lesion size were measured using previously reported methods. Cases were defined as individuals who sustained either a subluxation or dislocation event after the index procedure, while controls were defined as individuals who did not. The association between failure and three different models of instability were investigated. In the first model, a receiver-operating curve was constructed to determine the amount of glenoid bone loss alone that would best predict failure. In the second model, lesions were classified as on-track or off-track based on the glenoid track concept. In the third model, a distance-to-dislocation was determined based on the glenoid track concept. Combinations of glenoid bone loss and distance-to-dislocation were varied to determine the combination that was most closely associated with failure. T-test, chi-square, and Fisher exact testing was conducted to determine associations between categorical and continuous variables with failure.
Results:
There were 44 cases and 80 controls. There was no difference in age (p =.72) and sex (p= .69) between the two cohorts. Receiver-operator analysis of glenoid bone loss alone demonstrated that a threshold value of approximately 12% could best predict failure (AOC = 0.72). 21/25 (84%) of individuals with at least 12% bone loss failed surgery compared to 23/103 (22.3%) who had less than 12% bone loss (p < .001). Using the glenoid track model, 4/4 (100%) of individuals with off-track lesions failed surgery compared to 40/124 (32%) with on-track lesions (p < .001). Using a bipolar bone loss model, the combination thresholds of 9% glenoid bone loss and 11 mm distance-to-dislocation was associated with failure in 18/20 (90%) of subjects while 26/108 (24.1%) of subjects who did not meet this threshold failed surgery (p < .001). Intra-rater reliability for glenoid bone loss (ICC = 0.79) and Hill-Sachs lesion size (ICC = 0.75) were excellent.
Conclusion:
In our study, a threshold value of 12% glenoid bone loss predicted failure following an arthroscopic Bankart repair and all off-track Hill-Sachs lesions failed as well. Further, when both glenoid bone loss and glenoid track were subcritical, the combined use of a bipolar bone loss model accurately predicted failure. This model may be particularly useful for lesions felt to be on the cusp but not beyond threshold values for critical glenoid bone loss and off-track Hill-Sachs defects.