The Journal of Arthroplasty, Volume 35, Issue 5, 1268 - 1274

Does Very High Surgeon or Hospital Volume Improve Outcomes for Hemiarthroplasty Following Femoral Neck Fractures?

Clement, R. Carter et al.


This study evaluates whether very high-volume hip arthroplasty providers have lower complication rates than other relatively high-volume providers.


Hemiarthroplasty patients ≥60 years old were identified in the New York Statewide Planning and Research Cooperative System 2001-2015 dataset. Low-volume hospitals (<50 hip arthroplasty cases/y) and surgeons (<10 cases/y) were excluded. The upper and lower quintiles were compared for the remaining “high-volume” hospitals (50-70 vs >245) and surgeons (10-15 vs ≥60) using multivariable Cox proportional hazards regression. Multiple sensitivity analyses were performed treating volume as a continuous variable.


In total, 48,809 patients were included. Very high-volume hospitals demonstrated slightly less pneumonia (6% vs 7%, hazard ratio [HR] 0.77, 95% confidence interval [CI] 0.68-0.88, P < .0001). Very high-volume surgeons experienced slightly higher rates of inpatient morality (3% vs 2%, HR 1.30, 95% CI 1.06-1.60, P = .01), revision surgery (3% vs 3%, HR 1.24, 95% CI 1.02-1.52, P = .03), and implant failure (1% vs <1%, HR 1.80, 95% CI 1.10-2.96, P = .02). Sensitivity analyses did not significantly alter these findings but suggested that inpatient mortality may decline as surgeon volume approaches 30 cases/y before gradually increasing at higher volumes.


A clinically meaningful volume-outcome relationship was not identified among very high-volume hemiarthroplasty surgeons or hospitals. Although prior evidence indicates that outcomes can be improved by avoiding very low-volume providers, these results suggest that complications would not be further reduced by directing all hemiarthroplasty patients to very high-volume surgeons or facilities. Future research investigating whether inpatient mortality changes with surgeon volume (particularly around 30 cases/y) in a different dataset would be valuable.

Level of Evidence

Prognostic Level III.

Download article