脑卒中患者发生吞咽障碍的列线图风险预测模型的构建与验证

Construction and validation of a nomogram risk prediction model for dysphagia in stroke patients

目的构建并验证脑卒中患者发生吞咽障碍的列线图风险预测模型。方法研究方法为回顾性分析,观察对象为2022年1月至2023年1月西南医科大学附属简阳医院·简阳市人民医院收治的入院的120例脑卒中患者。收集患者一般资料[性别、年龄、体重指数、发病至入院时间、蒙特利尔认知评估量表(MoCA)评分、焦虑自评量表(SAS)评分、美国国立卫生研究院卒中量表(NIHSS)评分、舒张压、收缩压、气管插管时间、脑卒中类型、基础病症(有无伴发糖尿病、高脂血症、高血压)]。以吞咽障碍发生与否分为两组:42例脑卒中患者出现吞咽障碍,设定为发生组,78例脑卒中患者未出现吞咽障碍,设定为未发生组。脑卒中患者出现吞咽障碍的危险因素予以多因素Logistic回归分析,列线图风险预测模型选取R 4.1.0软件包及rms程序包构建,经Bootstrap法予以抽样;绘制准曲线对脑卒中患者发生吞咽障碍预测中该列线图模型的可靠性予以评估,同时通过受试者工作特征(ROC)曲线分析脑卒中患者发生吞咽障碍预测中该列线图模型的应用价值。结果与未发生组比较,发生组的年龄[(65.08±3.85)岁vs.(62.93±3.79)岁]、SAS评分[(52.15±1.34)分vs.(48.03±1.17)分]更高,气管插管时间[(47.58±5.54)h vs.(44.02±4.36)h]更长,MoCA评分[(20.17±0.52)分vs.(22.41±0.48)分]更低,差异均有统计学意义(P<0.05)。脑卒中患者发生吞咽障碍的危险因素有年龄、MoCA评分、SAS评分及气管插时间(P<0.05)。据此构建的列线图模型预测吞咽障碍发生率与实际发生率基本一致。脑卒中患者发生吞咽障碍预测中该列线图模型的ROC曲线下面积(AUC)为0.817(95%CI:0.782~0.851),约登指数为0.561,灵敏度为0.866,特异度为0.694。结论脑卒中患者发生吞咽障碍的危险因素有年龄、MoCA评分、SAS评分及气管插时间,以上述因素为基础构建的列线图风险预测模型可辅助预测脑卒中吞咽障碍。

Objective To construct and verify the nomogram risk prediction model of dysphagia in stroke patients.Methods The study method was retrospective analysis,and the observation objects were 120 stroke patients admitted to Jianyang Hospital Affiliated to Southwest Medical University,Jianyang People's Hospital from January 2022 to January 2023.General data[sex,age,body mass index,time from onset to hospital admission,montreal cognitive assessment scale(MoCA) score,self-rating anxiety scale(SAS) score,national institutes of health stroke scale(NIHSS) score,diastolic blood pressure,systolic blood pressure,time to intubation,stroke type,underlying disease(whether accompanied with diabetes,hyperlipidemia and hypertension)] of patients were collected.They were divide into two groups based on the occurrence of swallowing disorders,42 patients with dysphagia were set as the occurrence group,and 78 patients without dysphagia were set as the non-occurrence group.Multivariate Logistic regression analysis was performed to analyze the risk factors of dysphagia in stroke patients.The risk prediction model of the column graph was constructed using R 4.1.0 software package and rms program package,and sampled by Bootstrap method.The reliability of the model was evaluated by drawing a quasi curve,and the value of the model was analyzed by receiver operating characteristic(ROC) curve.Results Compared with the non occurrence group,the age of the occurrence group [(65.08±3.85) years vs.(62.93±3.79 years],SAS score [(52.15±1.34) points vs.(48.03±1.17) points] were higher,tracheal intubation time [(47.58±5.54) hours vs.(44.02±4.36) hours] was longer,MoCA score [(20.17±0.52) points vs.(22.41±0.48) points] was lower,and the differences were statistically significant(P<0.05).The risk factors of dysphagia in stroke patients were age,MoCA score,SAS score and tracheal insertion time(P<0.05).Based on this,the nomogram model predicted the incidence of dysphagia,which was basically consistent with the actual incidence.In the prediction of dysphagia in stroke patients,the ROC area under curve(AUC) of this nomogram model was 0.817(95%CI:0.782-0.851),Jordan index was 0.561,sensitivity was 0.866,and specificity was 0.694.Conclusion The risk factors for dysphagia in stroke patients include age,MoCA score,SAS score and tracheal insertion time.The risk prediction model based on the above factors can help predict dysphagia in stroke patients.