[Development and validation of a clinical predictive model for the risk of malignant ventricular arrhythmia during hospitalization in patients with acute myocardial infarction].

OBJECTIVE

To develop and validate a clinical prediction model for the risk of malignant ventricular arrhythmia in patients with acute myocardial infarction (AMI) during hospitalization, and evaluate the effect of the prediction model.

METHODS

A retrospective study was conducted. A total of 2 649 patients with AMI admitted to cardiology department of Changzhou No.2 People's Hospital of Nanjing Medical University from December 2012 to August 2020 were enrolled. The clinical characteristics including gender, age, medical history, discharge diagnosis, vital signs during hospitalization, electrocardiogram characteristics at admission, laboratory examination indexes, interventional treatment, drug usage, malignant ventricular arrhythmias [mainly included sustained ventricular tachycardia (VT), ventricular flutter or ventricular fibrillation (VF)], and death were recorded. All patients were divided into two groups according to whether VT/VF occurred during their hospitalization. Independent risk factors for VT/VF during hospitalization were evaluated by multivariate Logistic regression analysis, and a clinical prediction model was constructed. The receiver operating characteristic curve (ROC curve) was plotted, and the area under ROC curve (AUC) was calculated to evaluate the accuracy of the prediction model.

RESULTS

A total of 2 649 eligible patients with AMI were enrolled, of whom 134 (5.06%) developed VT/VF during hospitalization. The in-hospital mortality rate in VT/VF group was significantly higher than that in non-VT/VF group (38.1% vs. 1.7%, P < 0.01). Compared with the non-VT/VF group, the patients in the VT/VF group with lower systolic blood pressure [SBP (mmHg, 1 mmHg = 0.133 kPa): 125.9±28.2 vs. 132.0±24.2], higher random blood glucose (mmol/L: 8.6±4.8 vs. 7.4±3.7), worse cardiac function [Killip heart function grade ≥ 3: 36.6% vs. 10.7%, left ventricular ejection fraction (LVEF) < 0.50: 56.7% vs. 33.6%, frequent premature ventricular contractions: 12.7% vs. 1.2%] and more hypokalemia (46.3% vs. 17.3%), with significant differences (all P < 0.05). Multivariate Logistic regression analysis showed that Killip classification of cardiac function ≥ 3 [odds ratio (OR) = 3.540, 95% confidence interval (95%CI) was 2.336-5.363], random blood glucose > 11.1 mmol/L (OR = 1.841, 95%CI was 1.171-2.893), LVEF < 0.50 (OR = 0.546, 95%CI was 0.374-0.797), frequent premature ventricular contractions (OR = 12.361, 95%CI was 6.077-25.144), potassium < 3.5 mmol/L (OR = 4.268, 95%CI was 2.910-6.259), SBP < 90 mmHg (OR = 0.299, 95%CI was 0.150-0.597) and creatinine (Cr) > 100 μmol/L (OR = 2.498, 95%CI was 1.170-5.334) were independent risk factors for VT/VF in patients with AMI (all P < 0.05). The clinical prediction model of VT/VF risk was constructed based on the variables selected by multivariate regression analysis. The ROC curve analysis showed that the AUC of the model in predicting VT/VF was 0.779 (95%CI was 0.735-0.823, P < 0.001); the optimal cut-off value of the model was 17, the sensitivity was 76.1%, the specificity was 67.3%.

CONCLUSIONS

The incidence of VT/VF during hospitalization of AMI patients significantly increases the risk of in-hospital death. The independent risk factors of VT/VF are Killip grade ≥ 3, random blood glucose > 11.1 mmol/L, LVEF < 0.50, frequent ventricular premature beats, potassium < 3.5 mmol/L, SBP < 90 mmHg and Cr > 100 μmol/L. The newly constructed clinical prediction model has certain predictive value for the occurrence risk of VT/VF.