急性脑卒中患者衰弱风险预测模型的构建及验证

Construction and validation of frailty risk prediction model in patients with acute stroke

目的分析急性脑卒中患者衰弱的影响因素,构建并验证其列线图预测模型。方法采用简单随机抽样选取2022年9月至2023年5月在苏州市某三级甲等医院住院的急性脑卒中患者为研究对象,采用一般资料调查表、中文版Tilburg衰弱量表(TFI)、健康问卷抑郁症状群量表(PHQ-9)、Barthel指数(BI)、Morse跌倒评估量表(MFS)、Braden压疮风险评估量表、洼田饮水试验等进行调查。采用单因素和多因素logistic回归分析急性期脑卒中患者衰弱的影响因素,构建列线图预测模型。采用Bootstrap法对预测模型进行内部验证,Hosmer-Lemeshow检验评价拟合优度,C-index指数评价区分度,校准曲线验证一致性。结果共纳入急性脑卒中患者193例,其中发生衰弱70例,发生率为36.3%。单因素分析结果显示,BMI、美国国立卫生研究院卒中量表(NIHSS)评分、PHQ-9评分、BI评分、Braden评分、年龄、糖尿病史、跌倒风险、改良Rankin量表(mRS)评分≥2分及吞咽障碍与急性脑卒中患者衰弱有关(P<0.05)。预测模型变量包括年龄、糖尿病史、PHQ-9评分及Braden评分,其受试者工作特征(ROC)曲线下面积为0.839(95%CI:0.781~0.897),截断值为0.328,灵敏度为78.6%,特异度为75.6%。Hosmer-Lemeshow检验显示拟合度较好(χ^(2)=12.281,P=0.139),C-index为0.839(95%CI:0.837~0.841),预测模型具有良好的区分度,校准曲线显示模型预测衰弱发生风险与实际发生风险一致性较好。结论构建的列线图模型可预测急性期脑卒中患者衰弱的发生风险,为护理人员早期筛查和干预提供了依据。

Objective To analyze the influencing factors of frailty in the patients with acute stroke,and to construct and verify its nomogram prediction model.Methods The simple random sampling method was adopted to select the inpatients with acute stroke admitted to a class 3A hospital of Suzhou City from September 2022 to May 2023 as the study subjects.The general information questionnaires,Chinese version of Tilburg Frailty Scale(TFI),Health Questionnaire Depressive Symptom Group Scale(PHQ-9),Barthel Index(BI),Morse Fall Assessment Scale(MFS),Braden Pressure Ulcer Risk Assessment Scale and Kubota drinking water test were used to conduct the survey.The univariate and multivariate logistic regression were adopted to analyze the influencing factors for frailty occurrence in the patients with acute stroke The nomogram prediction model was constructed.The Bootstrap method was adopted to conduct the internal verification of prediction model.The goodness of fit was evaluated by the Hosmer-Lemeshow test.The discrimination was evaluated by the C-index.The consistency was verified by the calibration curve.Results A total of 193 patients with stroke were included,among them 70 cases developed frail with an incidence rate of 36.3%.The univariate analysis results showed that BMI,NIHSS score,PHQ-9 score,BI score,Braden score,age,history of diabetes,fall risk,modified Rankin scale(mRS)score≥2 points and dysphagia were related with the frailty occurrence in the patients with acute stroke(P<0.05).The variables of the prediction model included the age,history of diabetes,PHQ-9 score and Braden score.The area under the receiver operating characteristic(ROC)curve was 0.839(95%CI:0.781-0.897),the best cutoff value was 0.328,the sensitivity was 78.6%and the specificity was 75.6%.The Hosmer-Lemeshow test showed the goodness of fit was good(χ^(2)=12.281,P=0.139).C-index was 0.839(95%CI:0.781-0.897),the prediction mode had good discrimination.The calibration cure showed that the frailty occurrence risk predicted by the model had good consistency with the actual risk occurrence.Conclusion The constructed nomogram model could predict the occurrence risk of frailty in the patients with acute stroke,which provides a basis for the nursing staff condcuting early screening and intervention.