Statistical analysis
Odds ratios (ORs) were calculated, and forest plots were constructed using RevMan software (RevMan version 5.3 Windows; The Cochrane Collaboration, Oxford, UK) by either fixed or random effects model based on the level of heterogeneity. The level of heterogeneity in the forest plot was measured by the Cochrane chi-square-based Q-test and was considered significant if p -value<0.1 as described elsewhere [44]. However, I2 statistic was used to test the heterogeneity of included studies in which I2<25%, I2=25–50% and I2>50% indicates low, moderate and high level of heterogeneity, respectively [47]. A random effects model was applied to estimate ORs if I2>50% and considered that the study had high level of heterogeneity. In contrast, a fixed effects model was used to estimate ORs if I2<50%. Sensitivity analysis was carried out to assess the impact on any individual studies on measured pooled risk. Publication biasness was detected by visual inspection of the funnel plot where symmetrical distribution of the plot indicated no publication bias [46]. All the calculatedP -values were considered statistically significant if the values were<0.05.