| Objective:(1)To summarize the clinical features of patients with acute myocardial infarction(AMI),as well as serology,electrocardiogram,cardiac ultrasound,and coronary lesions;(2)To develop a modified GRACE score(Global Registry of Acute Coronary Events)by modifying the traditional GRACE score model to establish a modified GRACE score system to improve its predictive performance;(3)To construct and validate a prediction model for all-cause mortality in AMI patients based on the population in Xinjiang for clinical application.Methods:(1)5512 patients who were hospitalized with a clear diagnosis of AMI in the heart center of the First Affiliated Hospital of Xinjiang Medical University from January 1,2015 to December 31,2020 were collected through the electronic medical record system of the First Affiliated Hospital of Xinjiang Medical University,and a total of 4561 patients were finally enrolled after the inclusion and exclusion criteria.Their general and clinical data were collected,their clinical characteristics as well as the characteristics of coronary lesions were summarized,and the occurrence of major adverse cardiovascular events was observed through a mean follow-up of 51.8±23.4 months.(2)The endpoint events were divided into Death and non-Death groups by endpoint events,and the differences between the two groups were compared by two-sample t-test and chi-square test.The adjusted traditional risk factors as well as Log BNP(Brain Natriuretic Peptide)and modified GRACE scoring system were included in the multi-factor COX regression model,and the modified GRACE scoring system was established by combining Log BNP and traditional GRACE scores,finally the predictive performance of the traditional and modified GRACE scoring system was compared by ROC curve.The modified Grace score was also internally validated,and a comprehensive discrimination improvement index was calculated.(3)The initial variables were screened by a dichotomous logistic regression model among general information,past medical history,infarction complications,serological indicators,inflammatory indicators,electrocardiographic indicators,cardiac ultrasound indicators,and coronary angiography indicators,and the final variables were screened by including all the risk factors in the multifactorial logistic regression model after adjustment of the initial variables to establish a prediction model for all-cause mortality in AMI.By fitting,diagnosing,validating and evaluating the model,and adding different indicators to the base model to establish different models,the predictive performance of the different models was compared,and the best model was selected and confirmed as the final model,and the predictive performance of the new model was compared with the traditional GRACE model and the modified GRACE model.Results:(1)The all-cause mortality rate of AMI patients during hospitalization was 5.4%,the all-cause mortality rate during long-term out-of-hospital follow-up was 13.3%,and the complications of AMI occurred during hospitalization was 7.5%(among which ventricular fibrillation and cardiac arrest were prominent),and the study found that the all-cause mortality rate and complications of AMI patients were higher based on the population of Xinjiang region.(2)The comparison of the number of branches of coronary lesions with the occurrence of major adverse cardiovascular events during hospitalization and follow-up was found by chi-square test,which revealed that three branches of lesions had a higher probability of all-cause mortality,heart failure and cardiogenic shock during hospitalization and follow-up.Patients with infarct-complicated ventricular wall tumors had a higher probability of all-cause death and cardiogenic shock during hospitalization and follow-up.Ventricular fibrillation complicated by infarction had a higher probability of all-cause death,heart failure,bleeding,and cardiogenic shock during hospitalization and follow-up.Cardiac arrest as a complication of infarction had a higher probability of all-cause death,heart failure,bleeding,and cardiogenic shock during hospitalization and follow-up.And there was a correlation between the number of coronary lesion branches,ventricular wall aneurysm,cardiac arrest,ventricular fibrillation and all-cause death in patients with AMI.(3)The modified GRACE score was established by combining BNP and conventional GRACE score.The results were analyzed by multifactorial COX regression model,and after adjusting for confounders,the modified GRACE score was found to be correlated with all-cause mortality in AMI patients.The predictive performance between the traditional GRACE score and the modified GRACE score was compared by ROC curves,and the results revealed that the modified GRACE score(AUC=0.818 P<0.001)had better predictive performance than the traditional GRACE scoring system(AUC=0.801P<0.001).The change in the C statistic was not significant after passing the modified GRACE score through 10-fold crossover internal validation.The integrated discrimination improvement index(IDI)of the traditional and modified models was further calculated as IDI=0.019>0,suggesting that the modified GRACE score was a positive improvement to the traditional GRACE score.(4)The initial variables were screened by dichotomous logistic regression models from general and clinical data,serology,electrocardiogram,cardiac ultrasound,and coronary angiography indicators,and all of the above indicators,as well as those proposed by relevant expert consensus,guidelines,and clinical practice to be associated with all-cause mortality in AMI patients,were included in the multifactorial logistic regression models to screen the final variables: age,creatinine inflammatory index leukocytes,nutritional index albumin,myocardial marker Log BNP,electrocardiographic index ST-segment shift,cardiac ultrasound index left ventricular ejection fraction,coronary angiography index number of coronary lesion branches,cardiac function classification,history of stroke,and cardiac arrest as a complication of AMI.(5)The above indicators were initially used to establish a predictive model for AMI,and fitting of the model revealed that the cardiac function Killip classification was poorly fitted,so the indicator of cardiac function classification was excluded.The variables that confirmed the final model: age,creatinine,leukocytes,albumin,Log BNP,ST-segment shift,left ventricular ejection fraction,number of coronary artery lesion branches,history of stroke,and cardiac arrest.Model diagnosis: 1)diagnosis by strong impact point: there was only one strong impact point,and considering the large sample size,this one strong impact point was not treated;2)covariance diagnosis: the variance inflation factors of all variables in the model were less than 10,so the correlations among the variables included in the model were low.Internal validation of the model: By K-fold cross-validation with K=10,the C-statistic before the internal validation of the model was 0.839 > 0.7,and the C-statistic after 10-fold cross-validation was 0.836 > 0.7.After internal validation,it was found that the C-statistics of the model we built were all > 0.7,showing that this prediction model performed better.Model evaluation: Different models were built from different indicators,and a total of 6 models and a full-variance final model(AMI-DR)were constructed.And the C-statistics of models 1-6 gradually increased,and the differentiation between the models was gradually increasing,and the prediction performance was gradually improving.The comparison between different models revealed that the all-variable model had the largest area under the ROC curve with AUC=0.839,and the model had the best prediction performance,and the comparison between the AMI-DR model and the traditional GRACE model and the modified GRACE model revealed that the prediction ability of the AMI-DR model was better than that of the traditional GRACE model and the modified GRACE model by The delong test found that the comparison between the AMI-DR model and the traditional GRACE model was statistically significant(P=0.002 < 0.05),and the comparison between the AMI-DR model and the modified GRACE model was statistically significant(P=0.028 < 0.05).Therefore,the model with full variance was chosen to create a column line plot and scoring system(AMI-DR Score)and to plot calibration curves,clinical decision curves,and impact images for predicting all-cause mortality in patients with AMI.Conclusion:(1)Based on a single-center,retrospective cohort study of 4561 AMI cases in Xinjiang from 2015 to2020,it was found that AMI is a group of clinical syndromes mostly seen in middle-aged and elderly men,with chest pain as the main symptom,some of which may be accompanied by ST-segment deviation,increased number of coronary lesion branches,inflammatory cells,and elevated myocardial markers.As the number of coronary artery lesions increases,the all-cause mortality rate of AMI patients increases.(2)The modified GRACE scoring system established by combining BNP and the traditional GRACE scoring system was independently associated with all-cause mortality in patients with acute myocardial infarction,with a larger AUC area and higher predictive value compared with the traditional GRACE scoring system.(3)The new AMI-DR scoring model constructed by screening variables(age,creatinine,stroke,white blood cells,albumin,Log BNP,ST-segment excursion,left ventricular ejection fraction,number of coronary lesion branches,and cardiac arrest)by logistic regression model has better predictive performance and can be used to predict all-cause mortality in AMI patients in the short and long term. |
PDF Full Text Request