Risk Factor Prediction and Categorization for Glenohumeral Osteoarthritis: A Classification and Regression Tree (CART) Analysis.

Abstract

Objective: This study aimed to apply Classification and Regression Tree (CART) analysis to determine factors associated with glenohumeral osteoarthritis (GH OA) and establish specific cut-off points for risk factors based on this methodology.

Design: The cross-sectional study included 3,383 participants with shoulder pain. Cases were selected for GH OA. Patients with other shoulder pathologies were included as controls. 33 potential risk factors were assessed. The CART analysis was used to determine the highest-ranked risk factors associated with GH OA. Multivariable logistic regression analysis was then performed using the cut-off points obtained from the CART analysis.

Results: The CART analysis showed that age and body mass index (BMI) were the two most significant risk factors for GH OA. Multivariable logistic regression revealed that age categories ≥31- < 58 years (OR = 8.92), ≥58- < 64 years (OR = 20.20), and ≥ 64 years (OR = 42.20), and BMI categories ≥25-30 kg/ m2 (OR = 1.47) and ≥ 30 kg/ m2 (OR = 1.71) had higher odds of developing GH OA compared to age < 31 years and BMI <25 kg/m2.

Conclusion: This was the first study to use CART analysis to evaluate significant risk factors for GH OA and establish cut-off points for increased risk. The findings present age categories that are distinct from the arbitrary age groups used in previous studies.