Assuming you are talking about Oxycontin, the problem there was that there was no expert consensus - in fact there was no information at all on addictiveness - but the FDA still allowed that language to be used in the promotional literature. The guy who approved it ended up working for Purdue a few months after leaving the FDA.

The story of Purdue Pharma and the Sackler family definitely exposes issues with the FDA, but I'm not sure it's a good example of expert consensus getting it wrong. (Not to say that expert consensus is always right.)

Doctors were parroting the line and prescribing the drug anyway. Despite their perception as experts apparently a lot of them never bothered to cross check the alleged study with their common sense and previous training.