Study On The Prediction Model Of The Severity Of Spleen Deficiency And Phlegm Turbidity Syndrome In Stable Angina Pectoris Of Coronary Heart Disease

Purpose:To predict the order of severity of patients with stable angina pectoris and syndrome of spleen deficiency and phlegm turbidity.To explore the pathological characteristics of progression of the disease.In order to discover the disease early and detection and intervene early,which is the key to prevent and treat coronary heart disease.This study combined the indicators of western medicine with the syndromes of traditional Chinese medicine.It provided a simple and reliable method for predicting the order of severity of the disease in clinic by establishing the predictive model,validating the model and checking the model.It could promote tertiary prevention of diseases,which was similar to the concept of"preventive treatment of disease"in traditional Chinese medicine.Material and method:Systematic random sampling and cross-sectional observation were used in this study.Six tertiary and first-class hospitals in China were selected as sub-centers.240 patients who met the inclusion criteria were selected as the study subjects.Collected data of patients and established database.192 patients（80%）were randomly selected to establish the model,and 48 patients（20%）were selected to validate the model.According to the severity classification of CCS angina pectoris,192 patients were divided into two groups（gradeⅡ-Ⅲ）to establish the model.Used univariate analysis to screen the independent variables of the prediction equation.To determine the variables of the prediction equation,the binary logistic regression analysis was used to establish the predictive model.The model was evaluated from a comprehensive point of some important test methods.Calculated the predictive critical value of the model based on ROC curve.Another 48 patients were used to validate the model.Results:Through the binary logistic regression analysis,four risk factors were screened out,including age,diastolic blood pressure,palpitation,tongue texture.Age was most closely related to the risk of the disease,and the OR value of age from 51 to 60 years old was5.602（95%CI=0.476-65.964）,the OR value of age from 61 to 70 years old was0.863（95%CI=0.089-8.381）,the OR value of age over 71 years old was0.897（95%CI=0.089-9.005）.The next was tongue texture,the OR value of pale tongue was0.087（95%CI=0.009-0.858）,the OR value of teeth-marked tongue was1.120（95%CI=0.185-6.795）,and the OR value of bluish purple tongue was1.190（95%CI=0.191-7.408）.The next was DBP with OR value of 1.079（95%CI=1.017-1.145）.The last was palpitation,the OR value of mild palpitation was 0.020（95%CI=0.001-0.270）,the OR value of medium palpitation was 0.228（95%CI=0.059-0.886）.Predictive model was ln（P/1-P）=-6.385+1.723x₁-0.148x₂-0.109x₃+0.076x₄-3.916x₅-1.480x₆-2.444x₇+0.114x₈+0.174x₉.In the predictive model,x₁=age from 51 to 60 years old,x₂=age from 61 to 70 years old,x₃=age over 71 years old,x₄=DBP,x₅=mild palpitation,x₆=medium palpitation,x₇=pale tongue,x₈=teeth-marked tongue,x₉=bluish purple tongue.The Hosmer-Lemeshow goodness-of-fit test of the equation showed that P=0.356,indicating that the fitting effect of the model was general.The area under the ROC curve was 0.798,P<0.001,indicating that the identification validity of the model was general.The Cox&Snell determinant R² of the model is 0.166,and Nagelkerke determinant R² is 0.303.The sensitivity,specificity,accuracy,and Youden index of the model were respectively 89.13%,75.00%,88.50%and 0.64.Its positive likelihood ratio was 3.57 and negative likelihood ratio was 0.14,which showed that the model had some predictive value.According to the ROC curve,the predictive critical value of the model was P=0.1298.The sensitivity,specificity,accuracy and AUC of small samples verification were 80.00%,0,75.00%and 0.252,which might be related to the few samples and uneven distribution of cases of CCSⅡandⅢ.Conclusion:1.Age,DBP,palpitation,tongue texture were potential risk factors of the disease progression.2.The predictive model wasln（P/1-P）=-6.385+1.723x₁-0.148x₂-0.109x₃+0.076x₄-3.916x₅-1.480x₆-2.444x₇+0.114x₈+0.174x₉.In the predictive model,x₁=age from 51 to 60 years old,x₂=age from 61 to 70 years old,x₃=age over 71 years old,x₄=DBP,x₅=mild palpitation,x₆=medium palpitation,x₇=pale tongue,x₈=teeth-marked tongue,x₉=bluish purple tongue.The predictive critical value of the model was P=0.1298,it was CCSⅡwhile P<0.1298,and it was CCSⅢwhile P≥0.1298.