Table 2. Comparison by province of genome-wide significant signals for fasting glucose.

Province	N	SIX3-SIX2			G6PC2
		rs895636-T			rs34177044-A			rs2232326-C
		MAF	Beta	P-value	MAF	Beta	P-value	MAF	Beta	P-value
Heilongjiang	736	0.35	0.134	0.007	0.35	0.194	0.0009	0.02	-0.199	0.22
Liaoning	564	0.37	0.036	0.48	0.36	0.172	0.006	0.02	-0.365	0.03
Shandong	625	0.36	0.152	0.003	0.40	0.156	0.01	0.03	-0.232	0.16
Jiangsu	723	0.38	0.125	0.004	0.39	0.086	0.11	0.04	-0.007	0.95
Henan	732	0.36	0.120	0.009	0.37	0.076	0.14	0.03	-0.405	0.0009
Hubei	349	0.43	0.086	0.25	0.38	0.130	0.14	0.04	-0.151	0.42
Hunan	660	0.44	0.057	0.21	0.36	0.084	0.13	0.05	-0.305	0.003
Guizhou	461	0.43	0.128	0.07	0.36	0.188	0.03	0.07	-0.178	0.22
Guangxi	936	0.47	0.057	0.14	0.36	0.157	0.001	0.09	-0.277	3.5 x 10−5
All Subjects	5,786	0.40	0.099	2.3 x 10−8	0.40	0.145	6.9 x 10−12	0.04	-0.252	1.8 x 10−9

Provinces are ordered based on geographical location from north to south. Trait values were adjusted for age, age2, BMI, PC1, and sex, and the residuals were then inverse normal transformed. Beta estimates reflect per allele effects of variants on inverse normal transformed traits. Only non-diabetic subjects were included in the analyses. Global tests for heterogeneity between the provinces were not statistically significant (P>0.05). CHNS, China Health and Nutrition Survey; MAF, minor allele frequency.