So, I see no attempt to delineate covariables in this metastudy. Right off the bat, that's a huge red flag. You don't do univariate studies when the correlation you are studying has multiple independent variables. For example, people wearing masks are more likely to have received one or more vaccine doses. You have to control for that, along with behavioral differences. Mask-wearing undoubtedly has a positive correlation with other preventative measures. Studying only one of those variables (and likely the least effective one) will show a correlation that is NOT causally related.

The selection criteria for the studies chosen for inclusion should be disclosed, particularly how 48 of the 61 that "met the criteria" were eliminated from consideration. Without that information, this result could be entirely due to cherry-picking the set that confirms the authors' a priori conclusion.

This is likely another study that will never muster peer review, at least not in any reputable journal. It'll languish in pre-print while propagandistic sources tout it as a meaningful result.