Risk Prediction Model And Screening Model For Coronary Heart Disease

BackgroundCardiovascular disease(CVD)is an important disease that threatens human health.Coronary Heart Disease(CHD)is one of the most common types of cardiovascular disease.CHD is caused by coronary artery stenosis and/or occlusion,resulting in coronary artery vascular diameter changes.This will affect the blood circulation,causing myocardial ischemia and hypoxia.The development of CHD is a chronic process.The whole process from health → low-risk state → high-risk state → early cardiovascular disease→clinical symptoms → coronary heart disease often takes several decades.There is ample time to delay or prevent disease progression through health management.In the early stages of cardiovascular disease,it is necessary to construct a risk prediction model of CHD to detect high-risk individuals and to carry out personalized health intervention for their specific risk factors in order to prevent the occurrence of coronary heart disease as soon as possible.For individuals with suspected coronary heart disease symptoms,a simple,inexpensive and appropriate screening tool is needed to assess the coronary atherosclerotic plaque load in order to screen for high-risk individuals for further diagnosis(coronary angiography).This will not only reduce the cost of high-risk individuals,but also significantly reduce the excessive medical care.Based on this,a CHD model was established by establishing a CHD follow-up cohort and using subdistribution risk model.Simple CHD high-risk individual screening models was established by using the routine physical examination biomarkers of suspected coronary heart disease.Materials and MethodsⅠ Construction of prediction model for coronary heart diseaseBased on the "Shandong multi-center health management longitudinal observation cohort",we constructed coronary heart disease risk prediction sub-cohort according to the following inclusion criteria:① Baseline(including within one month after entering the baseline)without coronary heart disease;② With complete physical examination data;③ Age ≥ 20 years old.73386 people(41968 males,31418 females)entered the cohort eventually.On the basis of descriptive analysis,the classical risk factors of coronary heart disease were selected and the statistically significant indexes were used to construct the Fine and Gray model.Area under the ROC curve,O/E and ten-fold cross-validation were used to evaluate the models.Ⅱ Coronary plaque distribution spectrum and screening model of high-risk subjects in suspected CHD patientsWe selected patients who took coronary angiography in July 1,2014-April 29,2016 at Linyi City People’s Hospital.1631 subjects(1030 males,601 females)entered our research.On the basis of descriptive analysis,the discriminant factors were selected based on the logistic regression model.Using the selected discriminant factors and age,hypertension,diabetes,dyslipidemia,we build screening models on the discriminant rules of TPS score,SSS score,CADS score,three grades in parallel and three scores in series.The area of the ROC curve was used to evaluate discriminant ability of models.ResultsI Construction of prediction model for coronary heart disease1.There were 73386 people in the cohort,including 41968 males and 31418 females.The median follow-up time was 3.10 years.1545 cases of CHD occurred during the follow up,including 958 males and 587 females.The incidence density is 5.95 per 1,000 person-years for male and 4.90 per 1,000 person-years for female.2.The risk predictive model of coronary artery disease was established based on the competitive risk model.The predictive factors for male model were age,BMI,systolic blood pressure,total cholesterol,high density lipoprotein cholesterol,smoking and diabetes mellitus with an area under the ROC curve of 0.809(95CI:0.804～0,815)and an O/E value of 0.98.After ten-fold cross validation,the AUC still reached 0.806(95%CI:0.801～0.812).The predictive factors for female model were age,BMI,systolic blood pressure,total cholesterol,and diabetes mellitus with an area under the ROC curve of 0.869(95%CI:0.863～0.874)and an O/E value of 1.02.After ten-fold cross validation,the AUC still reached 0.866(95%CI:0.860～0.872).The predictive model of CHD has good predictive ability.Ⅱ Coronary plaque distribution spectrum and screening model of high-risk subjects in suspected CHD patients1.Coronary arteriosclerosis plaque load was quantitatively characterized by TPS,SSS and CADS,respectively,for 1631 suspected CHD patients(870 males,422 females)who underwent coronary angiography.The results showed that the incidence of coronary stenosis was 79.22%(1292/1631).The incidence of male stenosis was 84.47%(870/1030)and the incidence of female stenosis was 70.22%(422/601).The CADS score in the population showed that the number of obstructive coronary artery disease was 62.42%(1018/1631).The TPS score and SSS score showed a partial distribution in the population.The proportion of people with high scores was relatively low.47.70%(778/1631)subjects SSS score ≤ 5 points,74.00%(1207/1631)subjects TPS score≤ 5 points.There were differences in the scores of the three grades between different sexes and age groups.2.Age,high-density lipoprotein cholesterol,hemoglobin,direct bilirubin,alanine aminotransferase,diabetes mellitus and hypertension were used as the discriminant factors in logistic regression model.To build screening model for high risk individuals of CHD,TPS>5,SSS>5,CADS being obstructive lesions,three scoring methods used in parallel,and three scoring methods used in series were defined as criteria for discrimination.For male models,the AUC was 0.687(95%CI:0.656～0.717)under the TPS criteria,0.662(95%CI:0.630～0.693)under the SSS criteria,0.632(95%Cl:0.601～0.661)under the CADS criteria,0.625(95%CI:0.593-0.656)under the parallel criteria,0.691(95%CI:0.661～0.721)under the series criteria.For female models,the AUC was 0.731(95%CI:0.692～0.768)under the TPS criteria,0.761(95%CI:0.723～0.796)under the SSS criteria,0.740(95%CI:0.703～0.774)under the CADS criteria,0.753(95%CI:0.714-0.788)under the parallel criteria,0.735(95%CI:0.695～0.771)under the series criteria.Conclusion:1.The risk prediction model of CHD based on routine physical examination biomarkers was constructed for health management population.The results showed that the predictive ability was good,which provided a simple tool for early prediction of CHD in healthy management population.2.The distribution of coronary atherosclerotic plaques in suspected CHD patients were described.3.In this study,a simple CHD high-risk individual screening model was established by using the routine physical examination biomarkers based on coronary artery angiography.