Marcadores prognósticos derivados da monitorização eletrocardiográfica de 24 horas (Holter) em pacientes com disfunção sistólica ventricular esquerda: meta-análise

Abstract

Left ventricular systolic dysfunction (LVSD) usually precedes systolic heart failure. Heart failure is a world public health problem with a great impact in costs, morbidity and mortality. Sudden cardiac death is responsible for half of deaths in heart failure. The other half of deaths occurs by pump failure and its complications. Twenty four hour ambulatory electrocardiography (Holter) is frequently done in heart failure patients. The aim of this work is doing a meta-analysis of adverse prognosis predictors obtained from 24h Holter in patients with LVSD, presenting three articles instead of a thesis. The first paper is a review about doing systematic review and meta-analysis of diagnostic and prognostic markers, a specific and different field in relation to metaanalysis of therapeutic interventions. The paper was organized in the following way: a) introduction; b) details of eight steps to be followed; c) rules to present the text; d) conclusion. Methods of systematic review were described in detail with a critical analysis of meta-analysis statistical options. The summary Receiver Operator Characteristic was emphasized. The second paper is a meta-analysis of prognostic performance of non-sustained ventricular tachycardia (NSVT) as a predictor of sudden cardiac death or its equivalents: ventricular fibrillation and ventricular tachycardia. Eleven studies of patients with ischemic and non-ischemic ventricular dysfunction were included. There was threshold effect (Spearmans correlation = - 0.818, p<0.01) related to different definitions of NSVT. Negative likelihood ratio was 0.617 (95%CI 0.550 0.693) without heterogeneity ( 2 = 9.26 (10 degrees of freedom) p = 0.507, Inconsistency (I2) = 0.0%). The area under the sROC curve (AUC) was 0.68±0.02, suggesting moderate discrimination of NSVT as a risk marker of sudden death. True negative rate varied from 89 to 97%. Absence of NSVT means low probability of sudden death. There is statistically significant contribution of NSVT to risk stratification, independently of the ejection fraction. Risk stratification strategies must be evaluated in prospective studies using NSVT and ejection fraction. The third article is a meta-analysis of the prognostic performance of a 24 h Holter heart rate variability index SDNN (standard deviation of normal-normal interbeat intervals) as a risk marker for all cause death, transplantation, cardiac death or heart failure progression. The standardized mean difference (SMD) of SDNN between survivors and non-survivors was used to combine studies results. Several analysis were done separating studies by end-point, by design, by ejection fraction and by quality criteria of studies. Sixteen studies with 2394 patients were included. Patients were predominantly male (70-88%) and the mean age ± standard deviation varied from 40 ± 12 to 70 ± 10 years. SDNN was lower in non-survivors to all-cause-death (SMD 0.457 95%CI 0.226-0.688; heterogeneity test p = 0.0647). Results of subgroup analysis show consistency (all studies: SMD 0.594 IC95% 0.385-0.803; heterogeneity test p < 0.0001). The subgroup analysis of ejection fraction < 30% and 30% and a metaregression including ejection fraction as co-variable suggest no significant influence of ejection fraction in SMD results. Therefore, SDNN can be used as a risk marker. Additional studies are necessary to evaluate if SDNN can be used to decide changes in therapeutics. In conclusion, systematic reviews are useful to evaluate prognostic studies allowing critical and objective analyses of evidences. It is interesting to note that in building simple tables to compare studies we get different conclusions in relation to non-systematic reviews. It is possible to quantify the contribution grade of predictors to risk stratification by doing meta-analysis. Results of the meta-analysis of NSVT suggest that it offers additional predictive information of sudden cardiac death not offered by ejection fraction. Results of the meta-analysis of SDNN suggest that it offers additional information predictive of death, cardiac death or transplantation but not sudden cardiac death. Prospective studies based on these conclusions are necessary to evaluate the possibility of building risk scores discriminating low and high risk in left ventricular systolic dysfunction.