Between-study heterogeneity
Between-study heterogeneity ranged from 0 to 98%, and was statistically significant (Q test, p <0.10) in 164 meta-analyses (32%), of which the vast majority (83%) examined the risk of childhood hematological malignancies (Supplementary Table 5). A total of 105 meta-analyses (21%) showed considerable heterogeneity (I2=50-75%). Substantial heterogeneity (I2>75%) was found in 29 associations (6%) which examined different risk factors in relation to childhood hematological malignancies, CNS tumors, neuroblastoma and testicular cancer. To further assess the uncertainty of effect estimates, we calculated the 95% prediction intervals, which included the null value in most associations (n=388, 76%).