Logistic Regression Analysis Was Used To Predict The Risk Of Swallowing Dysfunction After Stroke

Objective:Explore the risk factors related to swallowing dysfunction after stroke,establish a column chart prediction model and validate the model to explore independent risk factors for swallowing dysfunction after stroke.Based on this,provide theoretical basis for clinical identification of high-risk populations with swallowing dysfunction after stroke,and provide reference for further optimizing clinical intervention measures.Method:1.Selection of predictive factors: Based on the literature research method of evidence-based medicine,literature was searched on databases such as China National Knowledge Infrastructure,China Biomedical Literature Database,Wanfang,VIP,Pubmed,Embase,the Cochrane Library,and web of science using established Chinese and English search terms.Literature was screened and extracted according to inclusion criteria.2.Construction and validation of predictive models: Based on the already identified predictive factors,the convenience sampling method was used to retrospectively analyze the relevant medical record information of stroke patients admitted to a tertiary hospital’s brain disease center in Hefei.Single factor analysis and binary logistic regression analysis were used to screen independent risk factors for swallowing disorders after stroke,and then R software was used to construct a column chart prediction model.Then,the predictive model was evaluated for clinical efficacy by calculating the area(AUC)values under the working characteristic curve(ROC)of the modeling and validation groups,drawing calibration curves,and then using clinical decision curves(DCA).Results:1.After searching and screening,21 articles were finally included.Through systematic review,a questionnaire on risk factors of dysphagia after stroke was summarized,mainly including general demography factors,stroke related factors,combined disease factors,treatment factors and general health factors.2.A total of 160 stroke patients were included,of which 51 had swallowing disorders,accounting for 31.9% of the total study subjects.The results of logistic regression analysis showed that older age [OR=2.23,95% CI(1.35,7.73)],NIHSS score [OR=11.4,95% CI(2.63,17.04)],muscle strength [OR=9.39,95% CI(7.51,12.93)],ADL score[OR=3.28,95% CI(2.60,5.58)],lesion site: brainstem [OR=5.98,95% CI(3.58,8.23)],and concomitant hypoproteinemia [OR=7.48,95% CI(3.29,18.99)] were independent risk factors for swallowing disorders after stroke.Based on the selected risk factors,a column chart model was constructed.The areas under the ROC curves of the modeling and validation sets were 0.924 and 0.838,respectively.The calibration curves of the two sets of data models overlapped well with the actual curves,indicating that the column chart model has good predictive ability.Conclusion:After stroke,swallowing disorders are prone to occur.Age,high NIHSS score,poor muscle strength,low ADL score,lesion location in the brainstem,and accompanied by hypoproteinemia are independent risk factors for swallowing disorders after stroke.The column chart prediction model based on the above risk factors has been validated to have good discrimination ability and clinical applicability,and can provide a certain theoretical basis for the prediction of swallowing dysfunction in clinical stroke.