Development and validation of a predictive model for osteoporosis in middle-aged and older patients with type 2 diabetes mellitus: a multicenter retrospective study

The increasing prevalence of type 2 diabetes mellitus (T2DM) and osteoporosis (OP) among aging populations necessitates early OP risk identification. This study developed and validated a nomogram for predicting OP risk in middle-aged and older T2DM patients, and explored dose-response relationships and model robustness. This retrospective multicenter study enrolled 44,624 T2DM patients from two tertiary hospitals in Southwest China (2018–2025), divided into training ( n = 30,549), internal validation ( n = 13,093), and external validation ( n = 982) sets. LASSO regression and multivariate logistic regression identified independent predictors for nomogram construction. Model performance was assessed using ROC curves, calibration plots, decision curve analysis (DCA), and clinical impact curves (CIC). Restricted cubic spline (RCS) regression examined potential nonlinear dose-response relationships between continuous predictors and OP risk, and subgroup analyses evaluated model stability. Nine independent predictors were identified: age, gender, cardiovascular disease, arthritis, chronic kidney disease, creatinine, uric acid, white blood cell count, and uric acid to high-density lipoprotein cholesterol ratio (UHR). The nomogram demonstrated good discrimination, with AUCs of 0.821 (95% CI: 0.814–0.827), 0.832 (95% CI: 0.822–0.842), and 0.775 (95% CI: 0.740–0.811) in the training, internal validation, and external validation sets, respectively. Calibration curves showed excellent agreement, and DCA confirmed clinical net benefit. RCS revealed nonlinear positive associations for age ( P -nonlinear < 0.001) and nonlinear negative associations for UHR ( P -nonlinear < 0.001) and creatinine ( P -nonlinear < 0.001) with OP risk. Subgroup analyses confirmed robust performance across smoking history, drinking history, and hyperlipidemia strata (AUC range: 0.761–0.822). An online risk calculator based on the nomogram was developed using R Shiny and is freely accessible at https://kfyycxl.shinyapps.io/osteoporosis/ . This nomogram, designed as a primary screening tool for osteoporosis in middle-aged and older T2DM patients, demonstrates good predictive performance and clinical utility. Using a recommended risk threshold of 13.0%, the model effectively identifies individuals warranting confirmatory DXA testing, serving a dual role: screening in DXA-limited primary care settings and triage in DXA-accessible hospitals. The model’s reliance on routine clinical and laboratory data supports its feasibility, though external validation is needed to confirm generalizability.