Radiographic survival in total shoulder arthroplasty.

BACKGROUND: Glenoid component loosening is thought to be a major cause of failure. This study assesses radiographic and clinical failure in shoulder arthroplasty, identifying factors predictive of loosening.
METHODS: Three-hundred two shoulder arthroplasties were implanted utilizing a cemented, keeled glenoid component, mean clinical follow-up 8.6 years. One-hundred fifty one shoulders had preoperative, early postoperative, and most recent radiographs, mean radiographic follow-up 8.0 years, minimum 4 years or less if revision was performed (2 cases).
RESULTS: Fifty-two of 151 glenoid components (34%) showed a shift in position or a complete lucent line ≥1.5 mm. Four humeral components (3%) shifted or showed a 2-mm lucency in 3 zones. Component survival (Kaplan-Meier) free from radiographic failure at 5 and 10 years were 99% (95% CI) (98-100%) and 67% (95% CI) (58-78%). Glenoid components with lines at the keel on initial radiographs were at risk for radiographic failure, hazard ratio 4.6 95% CI 1.2-17.2, P = .02. No associations were found between radiographic survival and age, gender, diagnosis, glenoid erosion, and preoperative or early subluxation. Late subluxation superiorly was associated with the glenoid at risk for radiographic failure (P = .006). Glenoid component survivals free from revision at 5 and 10 years for the 302 shoulders were 99% (95% CI) (97-100%) and 93% (95% CI) (90-97%).
CONCLUSION: Glenoid radiolucencies are seldom seen early, except beneath the faceplate. Glenoid radiolucencies develop, with notable changes 5 or more years following surgery. Humeral components seldom loosen. Revision rates remain low. The high frequency of late radiographic changes dictates the need for innovation.