Abstract
The impact of cryptic relatedness (CR) on genomic association studies is well studied and known to inflate false-positive rates as reported by several groups. In contrast, conventional epidemiological studies for environmental risks, the confounding effect of CR is still uninvestigated. In this study, we investigated the confounding effect of unadjusted CR among a rural cohort in the relationship between environmental risk factors (body mass index, smoking status, alcohol consumption) and systolic blood pressure. We applied the methods of population-based whole-genome association studies for the analysis of the genome-wide single nucleotide polymorphism data in 1622 subjects, and detected 20.2% CR in this cohort population. In the case of the sample size, approximately 1000, the ratio of CR to the population was 20.2%, the population prevalence 25%, the prevalence in the CR 26%, heritability for liability 14.3% and prevalence in the subpopulation without CR 26%, the difference of estimated regression coefficient between samples with and without CR was not significant (P-value = 0.55). On the other hand, in another case with approximately >20% heritability for liability, we showed that confounding due to CR biased the estimation of exposure effects.