Fig. 2

Forest plot meta-analysis of three independent breast cancer case-control studies. The effect estimates are derived from the “per 1 SD odds ratio” and presented as a log of odds ratio. The p value for heterogeneity is p = 0.01 indicating significant heterogeneity in the populations. The number of subjects (cases and controls) in each study is reported. Data from Severi et al. [23] have been reported elsewhere