A clinical prediction rule based on the decision tree model for falls among patients with hip osteoarthritis

Abstract

Background

The factors and thresholds associated with falls in patients with hip osteoarthritis (HOA) have not been clarified. This study aimed to clarify the internal factors associated with falls in patients with HOA and their thresholds using a decision tree analysis.

Methods

The study participants were 203 patients with HOA scheduled to undergo total hip arthroplasty at multiple facilities in Japan. The study evaluated various potential predictors, including sociodemographic factors, medical information, and assessments of motor function. Classification and regression tree techniques were used to develop clinical prediction rules.

Results

In total, 24.1% of the participants had experienced a fall. Age was selected as the first factor for falls in patients with HOA, with age > 79 years found to be highly associated with falls (57.1%). Pain scores on the Japanese Orthopaedic Association Hip-Disease Evaluation Questionnaire (JHEQ) were selected as the second factor. The findings indicated that even among those aged ≤ 69 years, those with a JHEQ pain score ≤ 9 (severe pain) were more likely to experience a fall (40.7%). The model demonstrated a fair level of predictive performance, as indicated by the area under the receiver operating characteristic curve (AUROC) of 0.738.

Conclusion

Being older and having severe pain even at a young age were identified as significant predictors of falls among patients with HOA. These findings could help health-care providers develop more effective interventions to prevent falls in patients with HOA.