Development and validation of a novel risk stratification tool for patients 65 years and older undergoing total shoulder arthroplasty for primary osteoarthritis

Abstract

Abstract

Purpose

Current preoperative risk tools assess frailty, nutrition, and acute illness separately, limiting accuracy. This study aimed to create and validate a composite risk score for adults ≥65 undergoing total shoulder arthroplasty for primary osteoarthritis.

Methods

A retrospective study of 10,061 patients from the ACS NSQIP (2015–2021) was conducted. A composite score combining ASA class, Risk Analysis Index, and acute conditions was developed via multivariable logistic regression. Outcomes included 30-day mortality, complications, readmission, reoperation, extended stay, and non-home discharge. Model performance was assessed using ROC curves and internal bootstrap validation.

Results

The composite score demonstrated higher discrimination for major complications (area under the curve 0.747), readmission (0.669), extended length of stay (0.660), and non-home discharge (0.700) compared to individual indices. In multivariable analysis, the composite score outperformed the modified frailty index, Risk Analysis Index, and nutritional risk index. Internal validation confirmed robust predictive accuracy with minimal optimism bias across all outcomes.

Conclusions

This composite risk score provides a validated framework for perioperative risk stratification among older adults undergoing total shoulder arthroplasty. Its integration of chronic frailty, physiologic reserve, and acute severity may aid clinical decision-making and resource allocation in short-term perioperative care. However, the score is limited by the absence of psychosocial factors and disease-specific characteristics (e.g., preoperative stiffness), which are important determinants of functional outcomes and are not captured in the ACS-NSQIP database.