. 2015 Mar 13;123(10):985–992. doi: 10.1289/ehp.1408909

Table 2.

Estimates of the percent variance explained (PVE) by genetic factors for DMA% obtained from linear mixed regression models.

HEALS participants	Covariate adjustment	All genotyped SNPs (n = 257,747)			All genotyped and imputed SNPs (n = 1,211,988)
HEALS participants	Covariate adjustment	PVE (%)	SE	p‑Value	PVE (%)	SE	p‑Value
All participants^a (n = 2,053)	No adjustment	13	10	0.09	16	12	0.08
	Adjusted for covariates^b	10	10	0.15	12	12	0.16
	Further adjusted for PCs^c	11	11	0.16	15	12	0.10
	Adjusting for two 10q24.32 SNPs	3	10	0.36	5	12	0.34
All participants, defining distant relationships as “unrelated”^d (n = 2,053)	No adjustment	48	13	0.0004	63	16	0.0002
	Adjusted for covariates^b	42	14	0.002	54	17	0.001
	Adjusted for two 10q24.32 SNPs	35	14	0.007	41	17	0.01
PCs, principal components. ^aUsing the full GRM, K_IBS on all individuals. The PVE is in between the full narrow-sense heritability and the heritability due to measured SNPs. ^bCovariates including sex, age (continuous), concentration of water arsenic (quartiles), genotyping batch, BMI, and smoking status. ^cTwenty principal components as additional covariates to minimize inflation in significance testing caused by population stratification. ^dUsing a modified GRM, with K_IBS set as 0 if K_IBS < 0.05 (i.e., ignoring distant relationships); this approximates the K_IBD for all individuals. The PVE corresponds to the full narrow-sense heritability. After eliminating close relative pairs from the data set (K_IBS > 0.05), our sample size was too small (n = 1, 338) to generate a non-zero heritability estimate using GCTA.