It is perhaps not surprising that the present study was unable to identify genome-wide signals for this quantitative trait. The genome-wide significance level of 5 × 10−8 is conservative for large EA families such as those in COGA, where the number of independent tests was small due to extended linkage disequilibrium (LD). Our power calculation showed that in the combined COGA and SAGE sample we had good power (~90%) to detect SNPs explaining approximately 1% of the variance.
