Development of Models for Assessing First Incident Fragility Fracture Risk in Postmenopausal Women: Data from the CLSA

Abstract

Objective: To develop models for assessing first incident fragility fracture risk in postmenopausal women which could be used as a screening tool in primary care and similar settings. Design: Cohort study. Methods: Outcome was defined as first incident fragility fracture reported at either year 3 or year 6. Model development was conducted using logistic regression with multiple imputation as sensitivity analysis. Model performance was assessed through AUC, sensitivity, and specificity. Results: Analysis included 10930 female participants (1048 events) aged 45 to 85 years old without a history of fragility fracture before baseline from the Canadian Longitudinal Study on Aging (CLSA). The final model consisted of 11 factors including age, alcohol consumption, antidepressants, balance, epilepsy, hand grip strength, height, osteoarthritis, parent hip fracture after age 50, fall in the past 12 months, and smoking. The model outperformed BMD T-score total hip alone showing moderate discrimination with an AUC of 0.63 [0.61, 0.65]. With a threshold of a fracture probability at 7.30%, sensitivity was 80.49% and specificity was 34.61%. After adjusting for BMD T-score total hip, antidepressants, balance, epilepsy, hand grip strength, height, osteoarthritis, parent hip fracture after age 50, and fall in the past 12 months remained statistically significant. Conclusion: This model uses routinely collected factors and shows reasonable ability to distinguish between individuals at higher and lower risk of first incident fragility fracture. It demonstrates good sensitivity capturing the majority of true cases. Although its specificity is relatively limited, the model still has potential as a screening tool to help identify those at high risk who might benefit from further examinations and early intervention. Further studies are needed to validate the model performance in external populations