中老年男性原发性骨质疏松患者骨折风险预测模型列线图的构建及应用

Construction and application of nomogram of fracture risk prediction model for middle-aged and elderly male patients with primary osteoporosis

目的构建中老年男性原发性骨质疏松骨折的风险预测工具并评价应用效果。方法选择河北省3家医院骨科2019年接诊的中老年男性原发性骨质疏松患者为研究对象,随访患者骨折发生情况。采集患者临床资料,完成数据预处理后,所有因素均进入单因素及多元COX风险因素分析,并构建风险预测模型,采用列线图展示预测模型,ROC曲线评价模型区分度,采用calibrationplot曲线评价模型准确度,采用临床决策曲线DCA评价模型的有效性。结果研究纳入的中老年骨质疏松患者随访结果显示:发生骨折和未发生骨折患组患者骨密度、吸烟、过度饮酒、安眠药用药史、高血压等指标比较差异无统计学意义(P>0.05),其余组间指标年龄、骨折、体质量指数、皮质激素应用、跌倒、饮食因素、饮食异常、缺乏运动、药物、饮茶、牛奶、咖啡、血脂异常、低钠血症等差异均有统计学意义(P<0.05)。将所有因素纳入本COX风险预测模型,构建风险预测模型,结果显示:年龄、家族脆性骨折史、体重指数、皮质激素应用、饮食因素(钙/维生素D缺乏)、饮食异常(神经性厌食症)、药物(免疫抑制剂、双磷酸盐药物)、饮茶、牛奶、血脂异常、低钠血症等指标与中老年男性原发性骨质疏松并发骨折密切相关。采用ROC曲线展示模型的区分度。本次构建的COX风险预测模型的ROC曲线下面积(AUC)为0.965,95%CI为0.9514~0.9672。COX风险模型校准曲线提示本模型构建的准确度较好。结论本次构建的中老年常见病原发性骨质疏松患者并发骨折的风险预测模型可用于预测新诊断中老年男性原发性骨质疏松患者并发骨折的风险,从而指导临床医护人员进行相应的针对性的干预措施,最终避免或降低患者并发骨折的可能性,值得临床推广应用。

Objective To construct a risk prediction tool for predicting the primary osteoporotic fractures in middle-aged and elderly male patients,and to evaluate its application effects.Methods The middle-aged and elderly male patients with primary osteoporosis who were admitted to the orthopedics department of 3 hospitals in Hebei province in 2019 were enrolled in the study,who were followed up for observing the occurrence of bone fracture.The clinical data of patients were collected to complete data preprocessing,and all the factors were included in single factor and multivariate COX risk factor analysis,and a risk prediction model was constructed.A nomogram was used to display the prediction model,the ROC curve evaluation model and the calibrationplot curve evaluation model were used to evaluate the accuracy of model,moreover,clinical decision curve DCA was used to evaluate the effectiveness of model.Results There were no significant differences in the bone mineral density,smoking,excessive drinking,history of sleeping pills and hypertension between the patients with bone fracture and those without fractures(P>0.05).There were significant differences in patient’s age,fracture,body mass index,corticosteroid use,falls,dietary factors,dietary abnormalities,lack of exercise,drugs,tea,milk,coffee,dyslipidemia,hyponatremia among groups(P<0.05).Moreover the patient’s age,family history of fragility fractures,body mass index,corticosteroid application,dietary factors(calcium/vitamin D deficiency),eating abnormalities(anorexia nervosa),drugs(immunosuppressive drugs,bisphosphonates),drinking tea,milk,dyslipidemia,hyponatremia and other indicators were related to primary bone fracture of patients(P<0.05).In addition the area under ROC curve(AUC)of the COX risk prediction model was 0.965(95%CI:0.9514~0.9672),which showed that the accuracy of the model was better.Conclusion The risk prediction model for fractures in middle-aged and elderly patients with primary osteoporosis can be used to predict the risk of fractures in the patients,so as to guide clinical medical staff to take corresponding measures and targeted interventions to avoid or reduce the possibility of occurrence of fractures,therefore,which is worthy of clinical application.