There is an entire section of the popular research establishment devoted to "meta-studies" that find, consolidate, and review past research of a topic, judging each for rigor, reproducibility, elimination of bias, etc. Since these studies are separately funded, there is less risk of bias or exaggeration of results. If experimental results hold up to this scrutiny, broader, more dramatic conclusions can be made. The nuances of this (and potential complications) were discussed in Carroll's Healthcare Triage video about the worms and replication [1]. These are easier to trust (and safer to report on), since the heavy lifting is done by the meta-researchers rather than the time-pressured, under-paid pop-sci writer who is trying to fill a word-count quota.

[1] https://www.youtube.com/watch?v=9SCFlYlNlLQ