Evaluation of comorbidity measures for predicting mortality and revision surgery after elective primary shoulder replacement surgery based on data from the National Joint Registry and Hospital Episode Statistics for England: population based cohort study.

Abstract

Objective: To determine the importance of comorbidity measures when predicting mortality and revision surgery after elective primary shoulder replacement surgery.

Design: Population based cohort study.

Setting: Linked data from the National Joint Registry and NHS Hospital Episode Statistics were used to identify all elective primary shoulder replacements in England, 6 January 2012 to 30 March 2022.

Participants: 37 176 consenting patients, aged 18-100 years, who had elective primary shoulder replacement surgery.

Main outcome measures: Risk of mortality at 90 and 365 days, and risk of long term revision surgery after the primary surgery.

Results: 37 176 primary shoulder replacement procedures were included; 102 patients died within 90 days and 445 within 365 days of the primary surgery. 1219 patients had revision surgery over a maximum follow-up period of >10 years. The addition of comorbidity measures derived from Hospital Episode Statistics (Charlson comorbidity index with summary hospital mortality index weights, Elixhauser comorbidity index, and hospital frailty risk score) to simpler models resulted in little improvement in predictive performance. Optimism adjusted performance (C index) of the models that included age, sex, American Society of Anesthesiologists (ASA) grade, and main surgical indication was 0.76 (95% confidence interval (CI) 0.72 to 0.81) for 90 day mortality, 0.74 (0.71 to 0.76) for 365 day mortality, and 0.64 (0.63 to 0.66) for revision surgery. The best performing models that included a comorbidity measure had an optimism adjusted C index of 0.77 (95% CI 0.73 to 0.81) for 90 day mortality, 0.76 (0.74 to 0.78) for 365 day mortality, and 0.65 (0.63 to 0.66) for revision surgery. Heterogeneity in model performance across regions of England was low, and decision curve analysis showed minimal improvement in net benefit when including comorbidity measures.

Conclusions: In this study, patient comorbidity scores added little improvement to simpler models that included age, sex, ASA grade, and main surgical indication for predicting mortality and revision surgery after elective primary shoulder replacement surgery. This improvement needs to be balanced against the additional challenges of routine data linkage to obtain these scores.