I don’t have the patience right now to analyze the meta-analysis, but many times they’re really not conducted properly, and can be heavily biased. Even without the authors seeing it.

Edit: it’s even behind a paywall. I’m not using my institution login for that crap.