LAUSR

We thank the editors for the opportunity to respond to the issues raised by Folley et al. [1]. We also thank Folley et al. for their interest in our paper and for helping to provide further insight into our findings. In our paper, we sought to identify genetic correlates of habitual physical activity (PA) by examining five PA measures: three based on self-report, and two based on accelerometry data [2]. Folley et al. hypothesized that individuals with poor cognitive function are more likely to over-estimate and over-report their level of physical activity. They support their hypothesis by showing that those with lower memory, lower education, and older age tend to self-report higher levels of PA, but generally exhibit lower levels of accelerometry-measured PA – an objective and presumably more accurate measure of actual PA (although not without its own limitations). The trends that they show are relatively consistent, and do seem to suggest that there is over-reporting of PA by these individuals. This could suggest that some of the loci identified for moderate-tovigorous PA (MVPA), and perhaps for some of the other self-report measures, may be related, at least in part, to relatively poorer cognitive performance rather than to MVPA. To determine whether the associations between the nine SNPs and self-reported MVPA described in Klimentidis et al. [2] are unrelated to education and/or memory, we performed analyses with and without adjustment for educational attainment (College/University degree, other qualification, no qualification), and then for both educational attainment and memory (pairs-matching test), both categorized in a similar fashion as Folley et al. We find that the SNP associations with MVPA remain relatively unchanged after these adjustments (see Table 1), suggesting that these genetic associations are not mediated by educational attainment or memory. Folley et al. also mention that this information bias may be particularly exemplified by the fact that the Alzheimer’s (and heart disease) ε4 risk allele at APOE was found to be associated with significantly greater levels of MVPA. In our paper, we suggested that this may be the result of selection/ survival bias due to an enrichment, at older ages, of APOE ε4 positive individuals with healthy lifestyles (individuals with risk allele and an unhealthy lifestyle are less likely to be alive or healthy enough to participate). We reported that the frequency of this allele in the sample decreases with age, and its strength of association with MVPA increases with age. Folley et al. mention there was not a significant