As SAGE is an ethnically diverse sample, as described in the ‘Methods’ section, our primary analyses used covariates to account for ethnicity. However, for the top SNPs listed in Table 2, we conducted secondary analyses for the primary CD symptom count variable on the European American and African American subsets of the sample to test whether the evidence for association was evident in both groups. These results are presented in Table 3. Most SNPs yielded evidence for association in both the European-American and African-American samples, with the exception of four SNPs (rs11838918, rs8179116, rs13398848 and rs2720508), in which the evidence for association was largely limited to the African-American subgroup (P > 0.20 in the European American sample). One additional SNP, rs10776612, provided stronger evidence of association in the African-American subgroup, with P = 0.14 in the European-American subgroup. In all cases, the direction of effect was consistent in both samples, including for four SNPs in which the minor allele differed between the groups (indicated in Table 3).
