RESUMEN
BACKGROUND: Identifying patients at risk of sudden cardiac death (SCD) remains a challenge. AIM: To evaluate the performance of non-sustained ventricular tachycardia (NSVT) from 24 hour ambulatory electrocardiography as a predictor of SCD in patients with heart failure or non-ischaemic dilated cardiomyopathy with left ventricular systolic dysfunction (LVSD). METHODS AND RESULTS: Study search and selection were performed by independent reviewers using a validated strategy. Eleven prognostic studies with >100 patients with good quality data and multivariate analysis of predictors of SCD were included. Publication bias was evaluated by funnel plot with Kendall's tau b test. A summary ROC (sROC) curve was built to evaluate predictive performance of NSVT. There was threshold effect (Spearman's correlation between sensitivity and specificity=-0.818, p<0.01) which indicates that combining sensitivity and specificity was not appropriate. The area of 0.68+/-0.02 under the sROC curve indicates a statistically significant contribution of NSVT in the prediction of SCD. The true negative rate varied from 89 to 97%. Multivariate analysis and meta-regression suggested that the contribution of NSVT to risk stratification is independent of ejection fraction. CONCLUSIONS: Absence of NSVT indicated a low probability of SCD in patients with LVSD. A risk score including NSVT should be evaluated in prospective studies.