Predicting Hip Osteoporosis with Routine Demographic and Biochemical Data: The Shao HipOsteoRisk Model

To develop and externally validate a simple, accessible prediction model for identifying individuals at risk of hip osteoporosis using routine demographic and laboratory data.

This retrospective study included 7686 adult patients who underwent hip dual-energy X-ray absorptiometry (DXA) at two medical centers in northern Guangdong, China. A total of 4638 patients were used for model development and 3048 for external validation. Predictors were selected using appropriate imputation and regularized regression techniques to ensure stability across datasets. Model performance was evaluated using discrimination, calibration, and clinical utility metrics.

Four routinely available variables-age, sex, body mass index, and the serum albumin-to-alkaline phosphatase ratio-were identified as the key predictors. The final logistic regression model demonstrated strong discrimination, with an area under the curve of 0.9107 in the development cohort and 0.8286 in the external validation cohort. Sensitivity and specificity were both favorable, and calibration showed good agreement between predicted and observed risk across most probability ranges. Decision curve analysis indicated meaningful net clinical benefit across a wide range of threshold probabilities, supporting the model's potential to improve risk stratification in practice.

We developed and validated a practical predictive model for hip osteoporosis based entirely on information commonly obtained during routine clinical care. Because it requires no specialized testing beyond standard laboratory panels, the model offers a low-cost, scalable screening tool-particularly valuable in settings where DXA access is limited. Its strong performance and ease of application suggest that it may help clinicians identify high-risk patients earlier, guide referral for confirmatory DXA scanning, and support more proactive osteoporosis prevention strategies.