First, 
and I really want to like this paper, but I find it lacking (at least in commentary and discussion if not for their somewhat dubious assumptions):
The word "mask" appears exactly 0 times in the manuscript.
For a study on self-reported behaviors and their link to infection risk, I would think the bare minimum is asking participants at least a few basic questions about their masking behavior. Even if there isn't a standardized survey they could pull from another work's methods section, it should be pretty easy to have respondents report how often they mask when in an enclosed space with people who don't live in their household. Without a reason to discount potential associations between diet and mask-wearing, this seems like a major oversight.
There's no effort to rule out socioeconomic variability as a confounding factor.