Design: Single-centre retrospective cohort
Population: 791 consecutive anatomic total shoulder arthroplasties, 2003-2015
Primary outcome: Implant survival and clinically important ASES outcomes at 10 years
Effect: Survival 96.1% at 10 years, 93.7% at 15 years; 69.5% reached acceptable symptom state

A single centre reviewed 791 consecutive anatomic total shoulder arthroplasties performed from 2003 to 2015 (JBJS, 9 October 2026). Kaplan-Meier implant survival was 96.1% at 10 years and 93.7% at 15 years. Younger age and a Walch B (posteriorly eroded) glenoid were independently associated with revision.

The authors also asked how many patients were still doing well, not just unrevised. At 10 years, 69.5% of shoulders reached the patient acceptable symptom state on the ASES score (threshold 68.1), 79.9% had improved by at least the minimal clinically important difference and 55.8% had a substantial clinical benefit. Function improved early and then gradually declined. Younger patients and those with primary osteoarthritis were more likely to have substantial benefit, even though younger patients were also more likely to need revision.

For counselling, the useful framing is that most implants last and most patients keep a meaningful gain at 10 years, but roughly 3 in 10 will not feel their shoulder is in an acceptable state by then.

Quote about 96% implant survival at 10 years and 94% at 15 years after anatomic shoulder replacement.
Tell patients that about 7 in 10 had an acceptable shoulder at 10 years, and function tends to decline slowly.
Flag a Walch B posteriorly eroded glenoid as a revision risk factor in planning.
Discuss the trade-off for younger patients: more benefit, but a higher chance of revision.

Why it matters

It shows that revision-free survival overstates success, because a shoulder can survive without the patient being satisfied with it.

The statistics, in plain English

Kaplan-Meier survival estimates the chance of avoiding revision over time, accounting for patients followed for different lengths. The acceptable-symptom threshold is anchored to what patients themselves said was satisfactory.