Description of included studies
The flow diagram of study selection is shown in Figure 1. Through an electronic search of three databases (PUBMED, EMBASE and CENTRAL) by keywords, 4317 citations were screened after 505 duplicates removed. Full-text review was performed in 47 studies, of which 26 studies did not meet the inclusion criteria and leaving 21 studies in the final qualitative and quantitative analysis.
The baseline characteristic of individual studies and patients are presented in Table 1 and Supplementary Table 1, respectively. In consideration of different studies from the same research team or database reporting various endpoints, we selected one with largest sample size and calculated that a total of 69919 patients were included in this meta-analysis, and the mean follow-up duration varied from 1.8 to 5 years. All studies included were observational design, in which 5 studies were prospective and the others were retrospective. The majority of included studies are from centers in European and the USA, and all patients have received CRT-P or CRT-D between 1999-2017. Trials were quite variable in association with sample size. The smallest of the studies included only 135 subjects, and the largest trial enrolled 50084 subjects.
Of those included studies, 17 of them reported adjusted HR of all-cause mortality [6,8-10,12-14,16,17,19-27], and 4 studies reported HRs after propensity score matching [8,14,17,19]. Adjusted covariates of each studies are listed in Supplementary Table 2. CRT-D implantation as a primary prevention therapy for sudden-cardiac death was clearly stated in 5 studies [6,12,14,19,25]. We assessed the quality of studies by using NOS, and only four studies had a score of ≥7 (Supplementary Table 3). Meanwhile, we pooled adjusted HR of all-cause mortality in patients with NICM, ICM and older age (≥75 years) to explore the optimal selection of individual device.