To Explore The Factors Of Overall Pump Function For Patients With Latent Coronary Heart Disease Based On Stepwise Regression Analysis

Objective: To explore the impact of comprehensive factors on cardiac overall pump function for patients with latent coronary heart disease. The comprehensive factors include diastolic and systolic condition and local heart function, which measured by means of acquiring parameters and also taking the effect of age into consideration.Methods: 38 cases of asymptomatic myocardial ischemia as well as no myocardial infarction history subjects were given rest-stress myocardial perfusion tomography imaging, a week before the equilibrium gated radionuclide cardiac blood pool imaging. After getting a clear series of images concerning cardiac cycle, namely ventricle film, delineate the regions of interest(ROI) of left ventricle, and obtain the ventricular volume curves, namely time radioactivity curve by computer data processing, then calculate parameters of cardiac function: left ventricular ejection fraction(LVEF), peak ejection rate(PER), peak filling rate(PFR), time of peak filling(TPF), regional ejection fraction(REF) and radius shortening(RS). Before performing stepwise regression analysis, firstly observe if each independent variable(PER, PFR, TPF, age, REF1-REF5 and RS1-RS5) and dependent variable(LVEF) show linear correlation or not via drawing the scatter plots. The REF1-REF5 represent the regional ejection fraction of the apical, anterior wall, partition wall, side wall and lower-rear wall, respectively. Similarly, RS1-RS5 represent the radius shortening of the apical, anterior wall, partition wall, side wall and lower-rear wall,respectively. Secondly select independent variables which have linear correlation with the dependent variable in turn and then calculate the correlation coefficient matrix so as to performing multiple linear test for the variables. Thirdly, eliminate the influence of multicollinearity by means of stepwise regression analysis method to establish the regression model in terms of the heart overall pump function, and perform goodness of fit test and significance test(F test) for the model. The aim is to comprehensively analyses the influence factors of the overall pump function for patients with latent coronary heart disease as far as possible. Statistics by SPSS Statistics 19 analysis software, P < 0.05 for the difference had statistical significance.Results: 38 cases of subjects show reversible defect or sparse of radioactivity distribution on rest-stress myocardial perfusion tomography imaging, which suggests the presence of myocardial ischemia. Scatter plots showed that PER, PFR, REF1,REF2, REF3, REF5 have linear correlations respectively with the dependent variable(LVEF). Because of the actual data with regression model assumptions(such as normality, equal variance), select Pearson correlation analysis method to calculate the correlation coefficient matrix among selecting independent variables, finding that the collinearity among independent variables is relatively serious, so stepwise regression analysis method should be available to eliminate the influence of multicollinearity.Let PER=X1, PFR=X2, REF1=X3, REF2=X4, REF3=X5, REF5=X6, then obtain the regression model as follows: ?=18.794+3.439X1+3.023X2+23.620X3 +17.546X6.Only PER, PFR, REF1 and REF5 are incorporated into the model and positively correlated with LVEF. The above cardiac pump function regression model also pass the test of goodness of fit(R2=0.63,Adjusted R2=0.586) and significant test(P <0.01).Conclusion : The establishment of the stepwise regression analysis model shows important value on the accuracy of cardiac function measuring by equilibrium gated radionuclide blood pool imaging, improving the diagnostic efficiency, predicting serious cardiac events, guiding treatment as well as evaluating prognosis.