急性缺血性脑卒中病人营养风险预测模型的建立和验证

Development and validation of a nutritional risk prediction model in patients with acute ischemic stroke

目的:建立并验证急性缺血性脑卒中病人营养风险预测模型。方法:前瞻性选择2018年1月至2019年12月收治于安徽医科大学第二附属医院神经内科且符合标准的急性缺血性脑卒中病人为研究对象。运用NRS 2002评估病人营养风险状况。依入组先后顺序按7:3的比例分配入模型建模队列与验证队列。采用Logistic回归法建立模型,分别应用ROC曲线、Hosmer-Lemeshow拟合优度检验对构建的预测模型在建模及验证队列中的区分度和校准度进行评价,计算预测的准确率。结果:纳入符合标准的病人837例,营养风险发生率为67.14%。建模队列和验证队列的营养风险发生率分别为65.36%和71.31%。Logistic回归结果显示,NIHSS≥15分、年龄≥62岁、每日服处方药≥3种、存在卒中史、小腿围≥32.4cm、饮酒史超过20年均是营养风险发生的独立危险因素(P均<0.05)。评分系统分值范围为0~24分,预测评分为14分时,Youden指数最大。该模型在建模和验证队列中的ROC面积分别为0.67(95%CI:0.62~0.70)和0.70(95%CI:0.68~0.77),P值均<0.001。Hosmer-Lemeshow拟合优度检验结果显示,模型预测风险与实际发生风险的一致程度较好,模型预测准确率较高(建模队列和验证队列分别为86.02%和87.37%)。结论:本研究构建的风险评估系统具有较好的区分度和校准度,具备一定的预测能力,可作为病人营养风险评估参考工具。

Objective:To develop and validate a nutritional risk prediction model in acute ischemic stroke patients.Methods:From January 2018 to December 2019,patients with acute ischemic stroke who were admitted to the Department of Neurology,the Second Affiliated Hospital of Anhui Medical University were selected.NRS 2002 was used to assess nutritional risk status of the study patients.According to the date of admission,they were assigned to the model development and validation cohorts in a proportion of 7:3.The model was developed by Logistic regression method.ROC curve and the HL goodness-of-fit test were used to evaluate the discriminative ability and calibration of the prediction model,and the accuracy of the prediction was calculated.Results:837 patients were included,and the incidence of nutritional risk was 67.14%.The incidence of nutritional risk in the development and validation cohorts were 65.36%and 71.31%,respectively.The results showed that NIHSS≥15 points,age≥62 years old,daily prescription drugs≥3,stroke history,calf circumference≥32.4cm,drinking history more than 20 years wereindependent risk factors for nutritional risk.The range of the scoring system is 0~24,and the Youden index is largestwhen the predicted score is 14.The area under the ROC curves in the development and validation cohorts were 0.67(95%CI:0.62~0.70)and 0.70(95%CI:0.68~0.77),respectively.The results of the goodness of fit test of the modelshow that the predictive risk of the model is consistent with the actual risk,and the accuracy of the model is high(86.02%and 87.37%,respectively).Conclusion:The risk assessment model has good discrimination and calibration,and can be used as a tool for nutritional risk assessment of patients.