A diagnostic predictive model for secondary osteoporosis in patients with fragility fracture: a retrospective cohort study in a tertiary care hospital

.Summary Identifying secondary causes among osteoporotic patients is crucial. However, there is no simple tool for screening secondary osteoporosis. A predictive model for screening secondary osteoporosis was constructed using simple clinical and biochemical parameters. This predictive model may provide clinicians with guidance to perform further investigations for specific causes of osteoporosis. Purpose Establishing whether a fragility fracture is secondary to a specific cause of osteoporosis is crucial for treatment outcomes. Therefore, this study aimed to develop a simple screening tool for secondary osteoporosis in the elderly initially presented with fragility fractures. Methods A retrospective cohort study including 456 patients with fragility hip and vertebral fractures that occurred between January 2017 and July 2022 was conducted. Demographic, clinical, biochemical, and final diagnostic data were retrieved. Potential predictors for secondary osteoporosis were determined by multivariable logistic regression analysis, and a predictive model for secondary osteoporosis was subsequently developed using identified potential predictors. Results This study included 343 females and 113 males with a mean age of 76.9 +/- 11.0 years. One hundred and twenty-one patients (26.5%) were diagnosed with secondary osteoporosis. Vitamin D deficiency (71.9%) was the most common cause of secondary osteoporosis, followed by glucocorticoid-induced osteoporosis (23.9%) and primary hyperparathyroidism (9.9%). The potential predictors for secondary osteoporosis included in the predictive model were age, body mass index (BMI), corrected calcium, phosphate, thyroid stimulating hormone, and a 10-year probability of hip fractures calculated by BMI-based FRAX (R). With a cut-off level of 0.22, the proposed predictive model has an AuROC of 0.75 (95% CI 0.69 to 0.81) with a sensitivity of 77%, a specificity of 66%, and an accuracy of 68.9%. Conclusion A predictive model for screening secondary osteoporosis was constructed using simple clinical and biochemical parameters. This newly developed predictive model may provide clinicians with guidance to perform further advanced investigations for secondary causes of osteoporosis.