1 Objective:1.1 To collect the basic data of patients with hypertension with left ventricular hypertro-phy(LVH)and summarize the influencing factors,and to construct the risk prediction model of hypertension with left ventricular hypertrophy by Logistic regression analysis combined with TCM elements such as constitution,tongue and pulse,basic characteristics,laboratory indexes and other modern medical methods,and to detect the prediction efficiency of the model and evaluate the prediction value of the model.1.2 Analysis of different LV geometry in patients with gender,age,physical fitness,cardiac ultrasound index differences,while observing the difference of cardiac function in patients with hypertension of different physical fitness.2 Methods:2.1 A total of 500 participants who met the diagnostic criteria were included as the training set.The basic information,medical history,tongue and pulse,constitution,physical and chem-ical examination and other information were collected.Univariate analysis showed that there were several factors with statistically significant differences,and then the variables with statis-tically significant differences(P<0.05)were substituted into binary Logistic regression analysis to obtainβcoefficient,which was assigned to the final variables to construct the risk prediction model of hypertension with LVH.The ROC curve and the validation set data was used to test the prediction results.2.2 According to the relative wall thickness and left ventricular mass index,500 patients with hypertension were divided into four groups to observe the basic characteristics,constitu-tion and echocardiography indexes of the 4 groups3 Results:3.1Baseline data500 participants include 212 males(42.2%),206 LVH patients(41.2%).The age was 63.37years old(average).There were 173 smokers(34.6%)and 122 drinkers(24.4%).The average age of menopause in women was 49.32 years old,and there were 10 non-menopausal women,accounting for 2%.The average duration of hypertension was 12.02 years,of which the patients with 1-10 years accounted for the most(50.8%),and the hypertension grade was mostly grade3(69.6%).The history of coronary heart disease accounts for the most,accounting for 22.8%.In the medication history,statins(67.4%),calcium channel blockers(56.8%),antiplatelet in-hibitors(54.8%)andβreceptor blockers(50.0%)were the most frequently used drugs.3.2 Analysis of influencing factors of hypertension with LVHIn this study,a total of 97 influencing factors in 22 categories of hypertension with LVH were collected,including gender,age,smoking and drinking history,menopause age,family history,hypertension level and course of disease,past history,vital signs,BMI,medication history,blood routine and urine routine,liver function and kidney function,four items of blood lipid,blood lipids,D-dimer,erythrocyte sedimentation rate,tongue color,tongue shape,coating color,coating quality,pulse condition and constitution.Analysis 24 variables with statistically significant differences,including gender,age,history of drinking,family medical history(heart failure),past medical history(coronary heart disease,diabetes).history of ARB drug use,history of statins,history of nitrate drug use,history of aldosterone receptor antagonist use,course of hypertension,height,BMI,hemoglobin,lymphocytes,red blood cell distribution width SD and CV,hematocrit,urinary protein,urea,D-2 aggregates,bleak tongue,greasy fur and yin defi-ciency.When the platelet content of the two groups was analyzed,the P value was 0.051.Alt-hough it was higher than 0.05,it was extremely close.Therefore,platelets and 24 variables with statistical differences were substituted into the multivariate analysis.3.3 Independent predictors of risk for hypertension with LVHA total of 7 variables with significant difference(P<0.05)were screened out from 25influencing factors,which were BMI,urinary protein,platelet,ARB medication history,greasy coating,dark tongue,yin deficiency.The results showed that the above 7 variables were inde-pendent predictors of hypertension with LVH.Theβcoefficient of platelet and ARB medication history is negative,indicating that the two are protective factors for hypertension with LVH,and the fiveβcoefficients are positive,indicating that they are risk factors for hypertension with LVH.3.4 Risk prediction formula of hypertension with LVHP=1/[1+e(-2.117+1.248X1+0.899X2+0.588X3+0.463X4+0.062X5-0.003X6-0.543X7)]3.5 The test of prediction probability of training set by modelHosmer-Lemeshow test was performed on the risk prediction model of hypertension with LVH,and the chi-square value was 5.205,P=0.735,indicating that the consistency between the model prediction of LVH risk in patients with hypertension and the actual observation value was good,and the model fitting degree was good.The ROC curve was plotted with LVH as the state variable and the prediction probability obtained by binary logistic regression as the test variable.The ROC curve(AUC)was 0.713,P<0.001.3.6 Test of risk prediction model for hypertension with LVHAccording to the risk prediction model of LVH in hypertension constructed by binary lo-gistic regression,33 cases of data in the validation set were substituted by SPSS software score wizard function.The true positive rate was 94.1%and the true negative rate was 93.8%.How-ever,when the AUC was used to describe the prediction probability of the model in the valida-tion group,the results showed that AUC=0.57,95%CI:0.37–0.769,P=0.494,indicating that the prediction effect of this model on the validation group was not high.3.7 The test of prediction probability of validation set by modelAccording to the risk prediction model of LVH in hypertension built by binary logistic regression,the scoring wizard function of SPSS software was used to plug in the data of 20cases in the validation,the positive predictive value was 50%and the negative predictive value was 42.9%.Calculate the ROC curve area of the prediction probability of the validation group model.The results showed that AUC=0.727,95%CI:0.495~0.960,P=0.087,AUC>0.7,indi-cating that the model has good predictive efficacy for validation group.3.8 Comparison of ultrasound indexes in patients with hypertension of different constitutions.In this study,the proportion of the top three physical analysis of pairwise comparison,analysis of the correlation between physical and ultrasonic indicators,the results show that phlegm in LVMI,LVEDD,RA,RV four indicators are higher than yin deficiency and qi defi-ciency,phlegm in LA,IVST,LVEDD,RA,LVPWd and qi deficiency compared with statistical difference(P<0.05).4 Conclusion:4.1 In this study,the prevalence of LVH in hypertension was 41.2%.The correlation anal-ysis showed that a total of 24 factors,including gender,age,height,BMI,history of drinking,course of HT,past medical history,history of medication,hemoglobin,lymphocytes,SD,CV,hematocrit,urinary protein,urea,D2 aggregates,dark tongue,greasy fur,and yin deficiency,were related to LVH in hypertension.4.2 Multivariate analysis showed that BMI,urinary protein,platelet,history of ARB drug use,greasy fur,dim tongue,and yin deficiency were independent predictors of hypertension complicated with LVH.4.3 The risk prediction model is great,but still need to increase quantity of validation set for further evaluation.4.4 The female group had the highest proportion of hypertrophy and the male group had the highest proportion of remodeling.By pairwise comparison of the top three physical,it was found that the cardiac ultrasonic indexes of phlegm-dampness changed more seriously than those of yin-deficiency and qi-deficiency. |