In the GIANT meta-analysis for height and BMI3,4, the summary statistics were adjusted by the genomic control method20 in each of the participating studies, and the test statistics were adjusted by the genomic control method for a second time in the combined analysis of all studies, which is sometimes called ‘double-GC’ correction. In the present study, we did not perform the second genomic control correction (although we have provided an option to implement this in our software tool) for two reasons. First, the purpose of genomic control correction is to adjust for the effect of population stratification, but, in the absence of population stratification and presence of polygenic inheritance, genomic inflation is expected21; therefore, double-GC correction might be too conservative and overkill. Second, if the genomic inflation is due to stratification, there is no reason to find additional associated SNPs at known loci, whereas under the hypothesis that the genomic inflation is consistent with polygenic inheritance and that there are multiple variants at the same loci segregating in the population, we would expect to see what we found empirically. The
