A predictive algorithm for perioperative complications and readmission after ankle arthrodesis.

Area of Science:

• Orthopedic Surgery
• Medical Informatics
• Health Services Research

Background:

• Ankle arthrodesis is a common procedure for ankle arthritis.
• Accurate pre-operative risk stratification for ankle arthrodesis is currently limited.
• Developing a reliable predictive model is crucial for patient management.

Purpose of the Study:

• To develop and validate a predictive model for major perioperative complications or 30-day readmission following ankle arthrodesis.
• To identify key risk factors associated with adverse outcomes after ankle arthrodesis.
• To provide a tool for improved pre-operative patient risk assessment.

Main Methods:

• Retrospective cohort study of 1084 adult patients undergoing ankle arthrodesis (2015-2017).
• Development of logistic regression and machine learning (ML) models, including XGBoost.
• Assessment of model discrimination (AUC) and calibration; feature importance analysis.

Main Results:

• 12.1% of patients experienced major complications or readmission.
• The XGBoost model showed the highest discrimination (AUC=0.707) and good calibration.
• Key predictors included diabetes, peripheral vascular disease, teaching hospital status, obesity, and infection history.

Conclusions:

• A well-calibrated algorithm for predicting complications and readmission after ankle arthrodesis has been developed.
• This predictive tool can aid in accurately risk-stratifying patients.
• Implementation may help reduce the incidence of major adverse events.