Development and Validation of a Diagnostic Prediction Rule for Osteopenia

ObjectivesTo triage patients with a high likelihood of osteopenia before referring them for a standard bone mass density test for diagnosis. IntroductionOsteopenia defined by low bone mineral density, is a precursor for osteoporosis and is primarily associated with aging-linked natural bone loss in adulthood. The model and findings can be used to adopt an inclusive screening and swift treatment model that can work in most settings where resources are limited. MethodsWe developed a diagnostic prediction rule based on clinical characteristics. A retrospective cohort of 798 patients who were going to be diagnosed with osteopenia or osteoporosis, within January-September 2022. The multivariable logistic regression to assess potential predictors. The logistic coefficients were transformed as a risk-based scoring system. The internally validation was performed using a bootstrapping procedure. ResultsThe model initially included seven predictors: sex, age, height, weight, body mass index, diabetes mellitus, and estimated glomerular filtration rate. However, after using backward elimination for model reduction, only three predictors--sex, age, and weight--were retained in the final model. The discrimination performance was assessed with the area under the receiver operating characteristic curve (AuROC); it was 0.779 (95%CI 0.74-0.82), and the calibration plot showed good calibration. For internal validation, bootstrap resampling was utilized, yielding an AuROC of 0.768 (95% CI 0.73-0.81), indicating robust performance of the model. ConclusionsThis study developed and internally validated the Osteopenia Simple Scoring System. This clinical risk score could be one of the important tools for diagnosing osteopenia and allocating resources in resource-limited settings.