You're assuming studies conducted today with minimal or no replication of results are evidence of health benefits. If 5000 of scientifically less-than-perfect evidence is not worth anything, then this study is also worth nothing.
I'm not citing it as a substitute, rather I'm saying that it isn't something you can entirely discount either. And we know for a fact that a lot of published "scientific" studies turn out to be not reproducible too, but people will continue to believe them for a long time regardless.