Factors influencing anxiety and depression in post-stroke dysphagia patients based on Logistic regression and decision tree model

Li Mengru; Cui Yan; Mao Erli; Liu Jingjing; Chen Shen

doi:10.3969/j.issn.1009-6574.2025.11.003

Abstract：Objective To explore the prevalence of anxiety and depression in patients with post-stroke dysphagia and to analyze the factors affecting anxiety and depression in patients with post-stroke dysphagia from multiple perspectives.Methods A questionnaire survey of 226 patients with post-stroke dysphagia in 14 Class Ⅲ Grade A hospitals in 8 cities of Jiangsu Province from October 2023 to May 2024 was conducted using convenience sampling method.A total of 221 valid questionnaires were recovered.Logistic regression model and decision tree model were established respectively to obtain the influencing factors of anxiety and depression under the results of the two models,and the comparison of predictive performance and the comparison of receiver operating characteristic curve were conducted.Results The prevalence of anxiety and depression in patients with post-stroke dysphagia was 61.5%(136/221).Logistic regression model showed that intermittent tube feeding[OR=0.307,95%CI(0.110,0.853)],water swallow test result gradeⅣ[OR=2.645,95%CI(1.226,5.703)],type D personality[OR=4.485,95%CI(1.732,11.608)],and high perceived social support[OR=0.522,95%CI(0.277,0.984)]were the influencing factors of anxiety and depression,and the differences were statistically significant(P＜0.05).Decision tree model showed type D personality was the main factor influencing anxiety and depression,followed by eating style,water swallow test results,perceived social support and personality traits,with a statistically significant difference(P＜0.05).The area under the receiver operating characteristic curve was＞0.7 for both models and the difference in predictive performance was not statistically significant(P＞0.05).Conclusions The prevalence rate of anxiety and depression in patients with post-stroke dysphagia is high.Combining Logistic regression model with decision tree model to guide clinical practice enables early identification and prevention of anxiety and depression in patients,and late targeted intervention.