Which requires research.Truly 'scientific' studies (with actual, non-bought-and-paid-for) peer review will control for such variables. (Often, it's a 'meta-analyis' of multiple papers and studies.) But this is also the reason why broad-based statistical measures are informative (as was "all-cause mortality").
A six- or ten-sigma even will stand out, and be undeniable (except to the Totally Blue-Pilled.)