ABC | Volume 111, Nº5, November 2018

Original Article Rattanawong et al PR interval and chronic resynchronization therapy Arq Bras Cardiol. 2018; 111(5):710-719 Figure 1 – Search methodology and selection process. 498 articles identified from MEDLINE 82 articles identified from EMBASE 376 articles reviewed after duplicates removed 13 full text articles reviewed 6 cohorts included in our meta analysis 363 articles were excluded • not cohort study • not related to our topic • animal studies • not conducted in CRT patients 7 articles were excluded • outcome not our main interest • not appropriate representatives • not appropriate comparators HF hospitalization is available in two studies [2, 4] involving 16,152 normal PR and 3,031 prolonged PR. Both studies achieved statistical significance. HF hospitalization pooled risk ratio is 1.30 (95 % confidence interval: 1.16-1.45, p < 0.01). The statistical heterogeneity was low with I 2 of 6.6%. Forest plot of this meta-analysis is shown in Figure 2B. Composite outcome (all-cause mortality and HF hospitalization) is available in four studies 2,4,8,15 involving 17,001 normal PR and 3,866 prolonged PR. All four studies revealed an increased death rate among patients with prolonged PR interval with two achieving statistical significance. In composite outcome, the pooled analysis also demonstrated a statistically significant increased composite outcome in CRT patients with prolonged PR interval compared to participants without prolonged PR interval with the pooled risk ratio of 1.21 (95% confidence interval: 1.13-1.30, p < 0.01). The statistical heterogeneity was low with I 2 of 0%. Forest plot of this meta‑analysis is shown in Figure 2C. Sensitivity analysis To assess the stability of the results of the meta-analysis, we conducted a sensitivity analysis by excluding one study at a time. None of the results was significantly altered, indicating that our results were robust (supplementary document 2). However, after exclusion of Freidman et al., 2 the heterogeneity decreased from 57.0% to 0% (supplementary document 3). Given moderate heterogeneity (I 2 = 57.0%) among all-cause mortality meta-analysis results, meta-regression (supplementary document 3) showed non-significant changes in all-cause mortality in PR interval > 230 msec compared with PR interval > 200 msec with risk ratio of 0.73 (95% confidence interval: 0.43-1.23, p = 0.123). Publication bias To investigate potential publication bias, we examined the funnel plot with pseudo 95% confidence limits of the included studies in assessing change in log risk ratio of death or composite outcome (Figure 3). The vertical axis represents study size (standard error) while the horizontal axis represents effect size (log risk ratio). From this plot, bias is present because there is asymmetrical distribution of studies on both sides of the mean. The Egger's test was significant (p < 0.05). However, using the trim and fill methods in the random‑effects model, there was no difference of the imputed risk ratio and its 95% confidence interval. Discussion The evidence provided in this systematic review and meta-analysis shows that a prolonged PR interval is significantly associated with an increased risk for all-cause mortality, composite outcome, and HF hospitalization of patients with CRT. Prolongation of PR interval, also known as first-degree atrioventricular block, is independently associated with increased risk for mortality and atrial fibrillation in the general population. 17 Even though correlation of PR interval with CRT response was conflicted in previous studies, our meta‑analysis confirms the negative effect on clinical outcome in patients with prolonged PR interval. According to the Comparison of Medical Therapy, Pacing and Defibrillation in Heart Failure (COMPANION) trial, around 50% of patients with CRT have prolonged PR interval. In addition, patients with CRT and prolonged PR interval are more likely to have ischemic cardiomyopathy, wider QRS complexes, more 712