Variation of the definition of "survival" in reverse total shoulder arthroplasty: a systematic review

Although studies have been published on long-term outcomes of reverse total shoulder arthroplasty (rTSA), various definitions of "survival" and "revision" are used, making comparisons of the findings difficult. Arthroplasty registries also vary in their definitions of these terms. The goal of this systematic review was to determine how variation in definitions of these terms affects the reported long-term results of rTSA.

We systematically searched the PubMed and Embase databases to identify studies reporting ≥10-year follow-up after primary rTSA. Nine studies (493 shoulders) met our selection criteria. The included clinical studies were predominantly Level IV retrospective cohort studies. Definitions of survival and revision were extracted from these studies. Previously published lists of national or regional arthroplasty registries were used to identify publicly available registry reports. When available, 10-year survival data for primary rTSA were extracted from these reports. Of the 11 registries reviewed, 3 had substantive 10-year survival data and definitions and were included in our analysis. Using standardized definitions of survival, we calculated pooled 10-year survival rates from the included studies and compared them with those from the arthroplasty registries. We defined entire-construct survival as the percentage of shoulders in which neither the metal components (humeral stem and/or glenoid baseplate) nor the polyethylene component had been revised. Similarly, we assessed 10-year survival of the metal components (no revision of metal components, but shoulder underwent polyethylene exchange/revision). Using these definitions, we recalculated survival rates for both entire-construct survival and metal component survival.

Definitions of survival varied among clinical studies. The weighted average 10-year survival rates were 82.7% for the entire rTSA construct (polyethylene, humerus, and glenoid components), and 89.5% for the metal components only. The rate of isolated polyethylene liner exchange was 7%. Registry-reported 10-year survival rates for the entire construct were consistently higher (approximately 94%-95%) than the weighted averages from the included clinical studies.

With standardized definitions, 10-year revision-free survival was 82.7% for the entire construct and 89.5% when isolated polyethylene liner exchange was excluded. Considerable inconsistency in the definition of long-term survival across clinical series and arthroplasty registries limits direct comparison and supports standardized reporting of survival endpoints.