构建老年四肢骨折患者2年内再骨折的预测模型，并验证其预测效能。

选取2020年1月至2020年12月在我院诊治的老年四肢骨折患者148例为训练组，用于构建列线图预测模型，采用Boot-strap法进行内部验证。选取2021年1月至2021年7月在我院诊治的老年四肢骨折患者86例为验证组，对预测模型进行外部验证。

训练组148例患者随访2年，失访6例，有效随访142例，再骨折27例，发生率为19.01%（27/142）。训练组2年内再骨折患者女性、合并糖尿病、营养不良、合并骨质疏松、固定物松动、使用糖皮质激素、独居多于未发生再骨折患者，年龄大于未发生再骨折患者，四肢骨骼肌指数、早期负荷锻炼、补充钙剂少于未发生再骨折患者（P<0.05）。四肢骨骼肌指数为训练组患者2年内再骨折的独立保护因素，年龄、合并糖尿病、合并骨质疏松、独居为其独立危险因素（P<0.05）。根据Logistic回归分析结果构建患者2年内再骨折的风险预测模型，logit（P）=-14.622+0.220×年龄-0.521×四肢骨骼肌指数+1.724×合并糖尿病+1.485×合并骨质疏松+1.282×独居。经Hosmer-Lemeshow拟合优度检验显示，χ²=6.411，DF=8，P=0.601，拟合度较好。内部验证C-index为0.833，区分度较好，Brier得分为0.107，校准斜率为0.666，校准度良好。外部验证AUC值为0.897，95%CI为0.810~0.953，灵敏度为77.78%，特异度为85.94%，具有良好的辨别能力，经Hosmer-Lemeshow检验显示拟合度较好，Brier得分为0.094，校准斜率为0.793，校准度良好。

老年四肢骨折患者2年内再骨折发生率较高，且影响因素包括年龄、四肢骨骼肌指数、合并糖尿病、合并骨质疏松、独居，医护人员应结合患者特点及高危因素对高危人群进行针对性管理，以降低再骨折发生率。

To construct a prediction model for refracture in 2-year in elderly patients with limb fracture, and verify its prediction efficacy.

A total of 148 elderly patients with limb fracture treated in our hospital from January 2020 to December 2020 were selected as the training group, which was used to construct the nomogram prediction model, and the Boot-strap method was used for internal verification. A total of 86 elderly patients with limb fracture treated in our hospital from January 2021 to July 2021 were selected as the verification group to conduct external verification of the prediction model.

In the training group, 148 patients were followed up for 2 years, 6 cases were lost to follow-up, 142 cases were effectively followed up, and 27 cases were re-fractured, with an incidence of 19.01% (27/142). In the training group, the number of women with refracture in the 2-year, combined with diabetes, malnutrition, combined with osteoporosis, loose fixation, glucocorticoid use, living alone was higher than that of patients without refracture, the age was higher than that of patients without refracture, and the skeletal muscle index of limbs, early load exercise, and calcium supplement were lower than that of patients without refracture (P<0.05). Skeletal muscle index of limbs was an independent protective factor for refracture in the 2-year in the training group, and age, diabetes mellitus, osteoporosis and living alone were independent risk factors (P<0.05). According to the Logistic regression analysis results, a risk prediction model for refracture of patients in the 2-year was constructed. logit (P)=-14.622+0.220×age -0.521×skeletal muscle index of limbs+1.724×diabetes mellitus +1.485×osteoporosis +1.282×solitude. The Hosmer-Lemeshow goodness of fit test showed that χ²=6.411, DF=8, P=0.601, indicating a good fit. The internal verification showed that C-index was 0.833 with good differentiation, Brier score was 0.107, calibration slope was 0.666, and calibration degree was good. External verification showed that AUC value was 0.897, 95%CI was 0.810-0.953, sensitivity was 77.78%, specificity was 85.94%, and it had good discrimination ability. Hosmer-Lemeshow test showed good fit, Brier score was 0.094, and calibration slope was 0.793. Good calibration.

Elderly patients with limb fracture have a high incidence of refracture in the 2-year, and the influencing factors include age, skeletal muscle index of limbs, combined with diabetes, combined with osteoporosis, and living alone. Medical personnel should conduct targeted management of high-risk groups according to the characteristics of patients and high-risk factors to reduce the incidence of refracture.