There's a replication crisis in a handful of more recent fields that use human subjects and didn't have hard rules and restrictions on how to treat human subjects in the early 20th century. Psychology is the field that has had the biggest issue, with many old studies having what we now see as serious methodology issues. It doesn't inherently mean all of those studies are wrong, just that they need to be revised with updated methodology to confirm if their results are accurate.

There's also about 1500 years of scientific study aside from that which doesn't relate to human subjects at all, and by this point has been replicated numerous times, so I would not doubt the claim that most research is replicable and valid. I would expect about 80-90% of our collective scientific knowledge to be accurate.