These results show that family history is a better classifier than current conceptualizations of SNP panels, based on candidate gene and GWAS for AD. Family history is likely a better predictor than this panel of SNPs because it accounts for more of the latent genetic factors contributing to AD, whereas the contribution to risk of the panel of SNPs is less clear. Family history also contains non-genetic predictors, which could account for a significant proportion of the risk as well, as family history could influence to some extent the environment that an individual is exposed to during development. Furthermore, the etiology of AD may be different for one family versus another. Therefore, risk prediction based on an individual’s family history may encompass genetic factors that are more specific to that individual than a general panel of SNPs, which may not explain risk for the particular subgroup to which that individual belongs. We assessed the value of combining information from the candidate gene panel with family history, as family history and the candidate gene sum score were not correlated (r =
