| Objective:1.To investigate the risk factors of acute posterior circulation ischemic(a PCI)due to acute dizziness.2.To establish a PCI risk warning model of acute vertigo onset and evaluate its diagnostic value;3.To compare the difference between ABCD3 score and this risk warning model in predicting the occurrence of a PCI.Methods:From October 2021 to February 2023,655 inpatients with acute vertigo or dizziness admitted to the Department of Neurology of Shanxi Bethune Hospital and meeting the natrium standard were selected.All kinds of clinical data,such as basic information,past history,symptoms,signs and serological results,were collected when enrolled patients were initially admitted without clear treatment methods,and ABCD3 scores were completed.A total of 655 patients were randomly divided according to the ratio of 7:3 in the modeling group and the verification group,i.e.461 patients in the modeling group and 194 patients in the verification group.The patients in the modeling group and the verification group were then divided into two subgroups,a PCI group and non-APCI group,according to the main diagnosis at discharge.Univariate Logistic regression analysis was used to find the factors of onset of acute vertigo in a PCI patients.Multivariate Logistic regression analysis was used to establish the regression equation,and the risk early warning model was drawn in a line graph,and the ROC(Receiver Operating Characteristic curve)was drawn to obtain the AUC(Area Under Curve),sensitivity,specificity and other information.Besides,the differentiation ability,calibration degree,clinical effectiveness evaluation and external verification of the risk early warning model were carried out.The above risk warning model was compared with each index of ABCD3 score to evaluate the difference between them in diagnosing a PCI.This retrospective study was approved by the Ethics Committee of Shanxi Bethune Hospital(YXLL-2023-046).Results:1.Comparison of clinical data between the modeling group and the verification group:t test was conducted for continuous variables of patients in the two groups.Results showed that there were no statistical significance in age,immediate blood glucose,platelet count,white blood cell count,NLR,and INR between the two groups(p>0.05).χ2 test was performed for the categorical variables of the two groups of patients.The results were as follows:Patients in the modeling group and the verification group were diagnosed with a PCI,ataxia,nausea and vomiting,drinking cough,dysphagia,dysphonia,unilateral limb numbness,unilateral limb weakness,double vision,finger and nose test instability,nystagmus,Babinski sign(+),smoking history,drinking history,coronary heart disease history,history of dyslipidemia、hypertension、diabetes、stroke;and gender.There was no statistical significance in 20 clinical data(p>0.05),so this verification group could be used to verify the risk warning model.2.Comparison of clinical data between the a PCI group and the non-a PCI group in the modeling group:A total of 461 patients in the modeling group were divided into the non-APCI group(86 cases,18.7%)and the a PCI group(375 cases,81.3%)according to the main diagnosis of discharge.The continuous variables of the two groups were tested by t,and there was no statistical significance in age,immediate blood glucose,NLR,platelet and INR(P>0.05).The difference of white blood cell count between the two groups was statistically significant(P<0.05).Secondly,19 categorical variables were tested by chi-square test.Among them,ataxia,cough when drinking water,dysarthria,unilateral limb weakness,unilateral limb numbness,diplopia,finger and nose test instability,nystagmus,Pap sign,smoking history,drinking history,hypertension history,coronary heart disease history,gender,a total of 14 differences were statistically significant(p<0.01 or p<0.05).There were no significant differences in nausea and vomiting,dysphagia,history of diabetes,history of stroke and history of dyslipidemia between the two groups(p>0.05).3.Univariate and multivariate logistic regression analysis of a PCI in the modeling group:univariate Logistic regression analysis showed that smoking history,drinking history,diabetes history,NLR,INR,ataxia were significantly correlated with the occurrence of a PCI(p<0.05).Multivariate logistic regression analysis showed that the above 6 clinical data were all risk factors for the occurrence of a PCI,and the risk warning model was as follows:ln(p/1-p)=1.166+1.133*smoking history+1.011*drinking history+2.100*diabetes history+0.191*NLR+1.117*ataxia-5.3*INR,and a line graph model was drawn.4.Evaluation of differentiation ability,calibration degree and clinical effectiveness of risk early warning model:The ROC curve of the risk warning model was drawn,and the area under the curve(AUC)was 0.819,the standard error was 0.024,the optimal Cut-off value was 0.512,the 95%CI was 0.773～0.866,P=0.000<0,01,the cut-off value was 0.856 points,the sensitivity was 0.640,and the specificity was 0.872.It shows that this risk warning model has high diagnostic value.Hosmer-Lemeshow goodness of fit test,χ~2=7.805,p=0.453>0.05,suggesting that the risk prediction model has good calibration ability.The average absolute error is 0.017,n=461.Based on the clinical decision curve,it is concluded that the threshold probability in the range of 0.07～0.73 is in the line graph to predict the net benefit of the model.5.External verification of risk early warning model:Data from the verification group were inserted into the risk prediction model to obtain corresponding results,and ROC curve was drawn to understand the diagnostic value of risk early warning model for a PCI:AUC was 0.793,standard error was 0.049,95%CI was 0.697～0.889,P=0.000<0.01,indicating that the risk early warning model had certain diagnostic value for a PCI.6.The diagnostic value of ABCD3 for a PCI:The ABCD3 score of 375 patients in the a PCI subgroup and the ABCD3 score of 86 patients in the non-a PCI subgroup in the modeling group were(3.57±2.109)points,and the ABCD3 score of 86 patients in the non-a PCI subgroup was(2.30±1.617)points,indicating a statistically significant difference between the two groups(P=0.024<0.05).The diagnostic value of ABCD3score for a PCI was understood by drawing ROC curve:AUC was 0.673,95%CI was0.607-0.739,P=0.000<0,01,Cut-off value was 2.000 points,sensitivity was 0.660,specificity was 0.635,suggesting that ABCD3 score had a certain significance in the diagnosis of a PCI,although the sensitivity of the two was similar.However,the AUC and specificity of ABCD3 were lower than that of the risk warning model,so the diagnostic significance of a PCI was poor.Conclusions:1.Smoking history,drinking history,diabetes history,NLR and ataxia are positively correlated with the occurrence of a PCI,and diabetes history has the strongest positive correlation with the occurrence of a PCI;2.INR is negatively correlated with a PCI;3.The risk warning model proposed in this study has high diagnostic value;4.In terms of diagnostic prediction of the occurrence of a PCI,ABCD3 score has certain diagnostic prediction ability,but it is lower than the risk warning model. |
PDF Full Text Request