Web table B shows details of the 20 single nucleotide polymorphisms typed, together with risk estimates for type 2 diabetes from published meta-analyses (web table A). All genotypes were in Hardy-Weinberg equilibrium. In line with previous reports, FTO rs1421085 was associated with body mass index (P<0.001).30 CDC123/CAMK10 rs10811661 (P=0.001), KCNJ11 rs5219 (P<0.01), and TCF7L2 rs7901695 (P<0.002) were associated with two hour glucose after an oral glucose tolerance test. CALPN10 rs3792267 was associated with significant differences in concentrations of total cholesterol and low density lipoprotein cholesterol (P=0.01). We found no other associations between genotypes and phenotypes included in either of the non-genetic risk models at phase 3 or phase 7 (web tables D-J). The point estimates of the effect sizes were consistent with previous meta-analyses involving many thousands of cases of type 2 diabetes (web table A). However, as would be expected in a prospective study with fewer cases of diabetes and in agreement with the previous report from a prospective study of similar size,35 most associations were not significant at P<0.05 in this dataset.
