Statistical analysis
Study features and patients’ characteristics were reported. Data are
presented as number (percentage) or mean and 95% confidence interval
(CI), as appropriate. The pooled estimated prevalence of each outcome
was calculated utilizing a random-effect model, which is more
generalized when potential heterogeneity among studies is present. All
values were presented with 95% confidence intervals (CI), and the
weighted prevalence of outcomes is depicted as a forest plot. We
assessed the homogeneity of incidence of events across studies using
Cochran Q for each of the outcomes. Also, to test for heterogeneity,
Higgins I2 was calculated.I2 represents the percentage of total variation
across studies that can be attributed to heterogeneity rather than
chance. The I2 value 0% indicates no
heterogeneity, 25% low heterogeneity, 50% moderate heterogeneity, and
75% high heterogeneity (9). Funnel plots were drawn for all extracted
outcomes to assess publication bias. Visual inspection of asymmetry was
assessed to check for publication bias, and Egger’s test was also
conducted to evaluate small-study effects on the pooled estimated
outcome (10). Moreover, Moment base univariate meta-regression was
performed to assess the effects of hypothermic circulatory arrest (HCA)
time on neurologic outcomes (stroke and SCI), depicted as bubble plot.
All data were analyzed using STATA software (StataCorp, TX, USA).