Predicting Prolonged Length of Hospital Stay and Identifying Risk Factors Following Total Ankle Arthroplasty: A...

Area of Science:

• Orthopedics
• Data Science
• Health Informatics

Background:

• Ankle osteoarthritis (OA) is a significant cause of chronic pain and functional loss.
• Total ankle arthroplasty (TAA) is a surgical option for end-stage ankle OA.
• Prolonged hospital length of stay (LOS) after TAA increases costs and complications.

Purpose of the Study:

• To utilize machine learning (ML) algorithms to predict hospital LOS in TAA patients.
• To identify key factors contributing to increased LOS in TAA procedures.
• To leverage the American College of Surgeons National Surgical Quality Improvement Program (ACS-NSQIP) database.

Main Methods:

• Employed four supervised ML classification algorithms.
• Analyzed data from adult patients undergoing elective TAA between 2008 and 2018 from the ACS-NSQIP database.
• Assessed the predictive performance using AUC, accuracy, sensitivity, and specificity.

Main Results:

• Identified female sex, ASA Class III, preoperative sodium and hematocrit, diabetes, preoperative creatinine, other arthritis, BMI, preoperative WBC, and Hispanic ethnicity as important predictors of LOS.
• Achieved an average AUC of 0.7257.
• Obtained an average accuracy of 73.98%, with average sensitivity of 48.47% and specificity of 79.38%.

Conclusions:

• ML models can effectively predict prolonged LOS in TAA patients.
• These predictive models can aid in perioperative decision-making.
• Identifying high-risk patients may help mitigate prolonged hospital stays and associated costs.