作者：曹芳，周三连，翟佳佳，徐倩，张春梅，严雨薇，李胜利

单位：

226000江苏省南通市第一人民医院康复科 通信作者：周三连，E-mail：123645728@qq.com

Units：

Department of Rehabilitation, Nantong First People's Hospital, Nantong 226000, China Corresponding author: ZHOU Sanlian, E-mail: 123645728@qq.com

关键词：

卒中; 吞咽障碍; 影响因素分析; 列线图模型;

Keywords：

Stroke; Deglutition disorders; Root cause analysis; Nomograms model

CLC：

R 743

DOI：

10.12114/j.issn.1008-5971.2023.00.010

Funds：

江苏省老年健康科研项目（LK2021046）；南通市科技计划指导项目（MSI21077）；南通市卫生健康委科研课题 （指导性）立项课题项目（QNZ2022021）

摘要：

目的 探讨脑卒中患者发生吞咽障碍的影响因素，构建其风险预测列线图模型，并进行内部验证和外部验证。方法 回顾性选取2016年10月至2021年10月南通市第一人民医院收治的550例脑卒中患者为内部验证组；收集同时期南通市中医院、南通市第三人民医院收治的1 100例脑卒中患者为外部验证组。收集患者临床资料：年龄、性别、BMI、发病至入院时间、基础疾病（是否合并高血压、高脂血症、糖尿病）、脑卒中类型、收缩压（SBP）、舒张压（DBP）、美国国立卫生研究院卒中量表（NIHSS）评分、焦虑自评量表（SAS）评分、蒙特利尔认知评估量表（Mo CA）评分、气管插管时间。采用多因素Logistic回归分析探讨脑卒中患者发生吞咽障碍的影响因素，采用R 4.1.0软件包及rms程序包建立脑卒中患者发生吞咽障碍的风险预测列线图模型，采用Bootstrap法重复抽样1 000次，计算一致性指数；绘制校准曲线以评估该列线图模型预测脑卒中患者发生吞咽障碍的可靠性；采用ROC曲线分析该列线图模型对内部验证组及外部验证组脑卒中患者发生吞咽障碍的预测价值。结果 内部验证组中，发生吞咽障碍196例（35.64%）（发生亚组），未发生吞咽障碍354例（64.36%）（未发生亚组）。发生亚组年龄大于未发生亚组，SAS评分高于未发生亚组，Mo CA评分低于未发生亚组，气管插管时间长于未发生亚组（P<0.05）。多因素Logistic回归分析结果显示，年龄、SAS评分、Mo CA评分、气管插管时间是脑卒中患者发生吞咽障碍的影响因素（P<0.05）。该列线图模型的一致性指数为0.818。校准曲线分析结果显示，该列线图模型预测脑卒中患者吞咽障碍的发生率与实际发生率基本吻合。ROC曲线分析结果显示，该列线图模型预测内部验证组脑卒中患者发生吞咽障碍的AUC为0.818[95%CI(0.783,0.852)]，特异度为0.695，灵敏度为0.867，约登指数为0.562；该列线图模型预测外部验证组脑卒中患者发生吞咽障碍的AUC为0.823[95%CI(0.797,0.848)]，特异度为0.743，灵敏度为0.730，约登指数为0.473。结论 年龄、SAS评分、Mo CA评分、气管插管时间是脑卒中患者发生吞咽障碍的影响因素，基于上述因素构建的列线图模型对脑卒中患者发生吞咽障碍具有一定预测价值。

Abstract：

Objective To analyze the influencing factors of neglutition disorders in stroke patients, construct nomogram model for predicting its risk, and conduct internal and external validation. Methods A total of 550 stroke patients hospitalized in Nantong First People's Hospital from October 2016 to October 2021 were analyzed retrospectively as internal validation group; 1 100 stroke patients hospitalized in Nantong Hospital of Traditional Chinese Medicine and Nantong Third People's Hospital in the same period as external validation group. Clinical data of patients, including age, gender, BMI, time from onset to admission, basic diseases (hypertension, hyperlipidemia, diabetes mellitus) , stroke type, systolic blood pressure (SBP) , diastolic blood pressure (DBP) , National Institutes of Health Stroke Scale (NIHSS) score, Self-rating Anxiety Scale (SAS) score, Montreal Cognitive Assessment (MoCA) score and time of tracheal intubation were collected. The multivariate Logistic regression analysis was used to analyze the influencing factors of neglutition disorders in stroke patients. The nomogram model for predicting the risk of neglutition disorders in stroke patients was constructed by using the R 4.1.0 software package and rms package. Bootstrap method was used to repeatedly sample 1 000 times for internal verification, and the consistency index (CI) was calculated. Calibration curve was used to evaluate the reliability of the nomogram model for predicting neglutition disorders in stroke patients, and the ROC curve was used to analyze the predictive value of the nomogram model for neglutition disorders in stroke patients in internal validation group and external validation group. Results In the internal validation group, 196 (35.64%) patients had neglutition disorders (occurrence subgroup) , and 354 (64.36%) patients had no neglutition disorders (non-occurrence subgroup) . The age of the occurrence subgroup was older than that of the non-occurrence subgroup, the SAS score was higher than that of the non-occurrence subgroup, the MoCA score was lower than that of the non-occurrence subgroup, and the time of tracheal intubation was longer than that of the non-occurrence subgroup (P < 0.05) . Multivariate Logistic regression analysis showed that age, SAS score, MoCA score and time of tracheal intubation were the influencing factors of neglutition disorders in stroke patients (P < 0.05) . The CI of the nomogram model was 0.818. The results of calibration curve analysis showed that the incidence of neglutition disorders in stroke patients predicted by the nomogram model was basically consistent with the actual incidence of neglutition disorders in stroke patients. The results of ROC curve analysis showed that the AUC of the nomogram model for predicting the occurrence of neglutition disorders in stroke patients in internal validation group was 0.818 [95%CI (0.783, 0.852) ] , the specificity was 0.695, the sensitivity was 0.867, and Youden index was 0.562; the AUC of the nomogram model for predicting the occurrence of neglutition disorders in stroke patients in external validation group was 0.823 [95%CI (0.797, 0.848) ] , the specificity was 0.743, the sensitivity was 0.730, and Youden index was 0.473. Conclusion Age, SAS score, MoCA score and time of tracheal intubation are the influencing factors of neglutition disorders in stroke patients. Based on the above factors, the nomogram model has a certain predictive value for the risk of neglutition disorders in stroke patients.

ReferenceList：