Statistical analysis
All analyses were performed using the STATA software (StataCorp, TX,
USA). All extracted data are summarized as mean (±SD) and number
(percentage). All values were concomitantly presented with 95%
confidence interval (CI). Median values for continuous variables were
converted into mean
values.3 The weighted
prevalence of endpoints during follow-up was calculated using the
random-effects model and was presented as forest plots. To estimate the
percentage of variation across studies, the Cochrane Q test was
calculated and reported as the I2 statistic. The
I2 value was classified into low heterogeneity
(I2 = 25% to 49%), moderate heterogeneity
(I2 = 50% to 74%), or high heterogeneity
(I2>75%).4 To
address heterogeneities, subgroup analyses using the Z-test were
conducted. The test compared groups for estimation of the subgroup
effect sizes and subgroup effects on heterogeneity. A meta-regression
analysis was also conducted to find the potential effects of variables
(i.e., age, gender, hypothermic circulatory arrest time, and the
chronicity of surgery) on the pooled outcome. Publication bias was also
estimated via the funnel plot visual inspection and the Begg’s test.