老年人群发生髋部骨折的危险因素分析及Nomogram预测模型的构建

Risk factors for hip fractures in the elderly and construction of a nomogram prediction model

目的分析老年人群发生髋部骨折的危险因素并构建Nomogram预测模型。方法前瞻性研究2019年12月—2021年9月河南省第三人民医院收治发生髋部骨折老年患者80例为病例组,另选取同期同医院体检为健康老人80例为对照组。所有患者受试者工作特征曲线(receiver operating characteristic curve,ROC)分析获取各因素的最佳截断值,多因素Cox比例风险模型分析影响老年人群发生髋部骨折的独立风险因素,R软件“rms”包构建预测老年人群发生髋部骨折高风险的Nomogram模型,校正曲线用来对Nomogram模型进行内部验证,决策曲线用来评估Nomogram模型的预测效能。结果病例组患者年龄、吸烟史、糖尿病比例、既往骨折史比例均高于对照组,而维生素D水平及骨密度水平均低于对照组(P<0.05);年龄、维生素D水平、骨密度、中性粒细胞/淋巴细胞计数、血小板计数/淋巴细胞计数、单核细胞/淋巴细胞比值的曲线下面积分别为0.891、0.828、0.789、0.875、0.836、0.820,最佳截断值分别为70岁、115μg/L、1.1g/m^(2)、4、180、0.55;年龄[≥70岁,1.909(1.215~3.000)]、吸烟史[是,2.294(1.493~3.524)]、糖尿病[是,2.294(1.493~3.524)]、既往骨折史[是,1.193(1.108~2.086)]、维生素D水平[<115μg/L,1.901(1.023~3.532)]、骨密度[<1.1g/m^(2),0.445(0.337~1.075)]是老年人群发生髋部骨折的独立危险因素(P<0.05);决策曲线结果显示,当风险阈值>0.37时,此预测模型所提供的临床净收益均高于单个独立危险因素并且在预测老年人群发生髋部骨折的高风险方面可以提供显著额外的临床净收益。结论年龄、吸烟史、糖尿病、既往骨折史、维生素D水平、骨密度是老年人群发生髋部骨折的独立危险因素,本研究基于此构建的Nomogram模型为老年人群发生髋部骨折的防治提供重要策略指导。

Objective The risk factors for hip fractures in the elderly were analyzed and a nomogram prediction model was constructed.Methods This prospective study included 80 elderly patients with hip fractures admitted to the Third People’s Hospital of Henan Province from Dec.2019 to Sep.2021 as the case group,and another 80 healthy elderly patients who underwent physical examination in the same hospital during the same period as the control group.The receiver operating characteristic(ROC)curve of all patients was analyzed to obtain the optimal cut-off value of each factor,and the independent risk factors affecting the incidence of hip fractures in the elderly were analyzed by Cox proportional risk model.R software"rms"was adopted to construct a nomogram model to predict the high risk of hip fractures in the elderly.Calibration curves were used to validate the nomogram model internally while decision curves to evaluate the prediction efficacy of this model.Results Age,smoking history,proportion of diabetes patients and previous fracture history were much higher in case group than in control group,but the vitamin D level and bone density level were much lower(all P<0.05).The area under the ROC curve of age,vitamin D level,bone density,neutrophil/lymphocyte count,platelet count/lymphocyte count and monocyte/lymphocyte ratio were 0.891,0.828,0.789,0.875,0.836 and 0.820,respectively.The optimal cut-off values were 70 years old,115μg/L,1.1g/m^(2),4,180 and 0.55,respectively.Cox proportional risk model showed that the hazard ratio and 95%CI were respectively 1.909(1.215-3.000)for age≥70 years,2.294(1.493-3.524)for smoking history,2.294(1.493-3.524)for diabetes,1.193(1.108-2.086)for previous fracture history,1.901(1.023-3.532)for vitamin D level<115μg/L,and 0.445(0.337-1.075)for bone mineral density<1.1g/m^(2),all of which being independent risk factors for hip fractures in the elderly(all P<0.05).Decision curve results showed that the prediction model provided a higher clinical net benefit than a single independent risk factor and can provide significant additional clinical net benefit in predicting the higher risk of developing hip fractures in the elderly when the risk threshold was>0.37.Conclusion Age,smoking history,diabetes mellitus,previous fracture history,vitamin D level,and bone mineral density are independent risk factors for hip fractures in the elderly.The nomogram model established in this study can provide important strategic guidance for the prevention and treatment of hip fractures in the elderly.