| Objective To develop a triple-stage osteoporosis (OP) and osteopenia (OST) prediction model (TOOP) based on health examination data for primary screening and assessment of bone mineral density (BMD) status in middle-aged and elderly population. Methods In this retrospective study, 8021 individuals who were over 40 years old and completed dual-energy X-ray absorptiometry examination were collected in the health management center of People’s Hospital of Qiannan from January 2019 to December 2021, and were 1:1 randomly divided into a fitting dataset and a validation dataset. According to the principle of conditional probability, the triple-stage prediction for BMD status was equally converted into independent prediction on OP and OST. Two prediction indices were then integrated to construct the TOOP index. The accuracy, sensitivity, specificity, and detection rate were used for model validation and further comparison with OSTA index. Results Gender, age, and weight were predictors of OP, and age and weight were predictors of OST. The accuracy of TOOP was 0.70 (95% CI: 0.68-0.71), which was significantly higher than that of OSTA (P<0.001). The sensitivity of TOOP was 0.46 for OP (95% CI: 0.40-0.53) and 0.56 for OST (95% CI: 0.53-0.60), which was significantly higher than those of OSTA (OP: P<0.001, OST: P<0.001). The specificity of TOOP was 0.96 for OP (95% CI: 0.95-0.97) and 0.79 for OST (95% CI: 0.77-0.81), which was slightly lower than those of OSTA (OP: P<0.001, OST: P<0.001). The detection rate (42.0%) and detection rate (32.7%) of TOOP index for people with abnormal bone mass were both higher than OSTA index (P<0.001), and the missed detection rate (12.4%) was lower than OSTA index (P<0.001). Conclusion TOOP index provides a good prediction tool for the risk screening of osteoporosis and osteopenia in middle-aged and elderly people. |