Pooled outcome
Early (30 day) mortality occurred in 339 patients, corresponding to a weighted early mortality of 5.5% (95% CI: 4.3 – 7.1%). The linearized occurrence rates (LOR) of mortality, reintervention on the aortic root, hemorrhage, thromboembolism, endocarditis and major adverse valve-related events are presented along with a measure of statistical heterogeneity in Table 2.
Late mortality occurred in 1,037 patients (4.6%/patient-year); in 41% the cause was unknown or not reported. The main cause of late mortality was cardiac (52%), of these were 51% valve-related and 49% non-valve-related cardiac death.
Publication BiasAnalysis of the funnel plots revealed evidence of underreporting of late mortality, reintervention on the aortic root, and thromboembolism in studies with smaller patient numbers. For other variables, no evidence of publication bias was found (Supplementary file 3).
Sensitivity analyses revealed that this potential publication bias did not have a substantial effect on pooled outcomes, as these remained generally unchanged after temporary exclusion of smallest quartile of the studies.