Forty-eight studies (13,216 patients, range 11 to 4,299) were included in the review: 47 retrospective studies and one prospective study.
Immediate postoperative mortality was 6.7% (95% CI 5.8 to 7.5; 47 studies; substantial heterogeneity Ι²=48.2%). Pooled late survival rates were 87.6% (at one year), 78.7% (at two years), 65.4% (at three years) and 29.7% (at 10 years). The funnel plot did not demonstrate evidence of publication bias. Subgroup analysis suggested that immediate postoperative mortality was lower (5.8%, 95% CI 4.8 to 6.9) in 18 studies with a mid-date from 2000 to 2006. Similar proportions were reported for early and late mid-dates in relation to five-year survival.
The pooled result for postoperative stroke was 2.4% (95% CI 2.0 to 2.9; 21 studies; no substantial heterogeneity Ι²=18.9%). Similar pooled risks were reported for earlier and later mid-dates. The pooled proportion at risk for postoperative dialysis was 2.6% (95% CI 1.6 to 3.8; 10 studies; significant heterogeneity) and for pacemaker implantation was 4.7% (95% CI 3.4 to 6.1; six studies; no significant heterogeneity). Pooled mean length of stay in intensive care was 3.5 days (95% CI 2.8 to 4.3; six studies; significant heterogeneity) and mean length of in-hospital stay was 13.3 days (95% CI 11.1 to 15.6; nine studies; no significant heterogeneity).
Further results for subgroup analyses were reported in the paper.