Because billions of dollars can be made by selling pharmaceuticals, only good evidence of established harm can overcome financial incentives. An example of good evidence of established harm is what had existed when the anti-inflammatory drug, rofecoxib (Vioxx), was pulled off the market.
In a trial of a couple thousand people taking either rofecoxib (Vioxx) or placebo to prevent colon cancer, they discovered a 92% increase in the rate of thrombotic events. Here are personal notes on their findings:
Row 4 is the treatment group and row 5 is the placebo group.
At right, a 90% confidence interval was formed around the rate of thrombotic events per 1,000 person-years (KPY). The highlighted cells indicate statistical significance, as even the lower bound rate for the treatment group (orange cell) is still higher than the upper bound for placebo group (green cell).
For each 139 people taking Vioxx, it would be expected that a new thrombotic event (myocardial infarction, ischemic stroke, etc.) would be caused every 12 months.
If 139,000 people would take Vioxx, then a thousand new thrombotic events would be expected to occur each year. But, by the time the drug was pulled in late 2004, there were 80 million people who had taken it.
That’s worth many thousands of caused heart attacks and strokes, even though the absolute increase in yearly risk was still less than a 1% increase (1.5% vs. 0.8%).
The upshot is:
If a drug seriously harms every 139th person who takes it, then it is time to pull it off of the market (even if it is making billions of dollars in revenue every year).
Reference
Bresalier RS, Sandler RS, Quan H, Bolognese JA, Oxenius B, Horgan K, Lines C, Riddell R, Morton D, Lanas A, Konstam MA, Baron JA; Adenomatous Polyp Prevention on Vioxx (APPROVe) Trial Investigators. Cardiovascular events associated with rofecoxib in a colorectal adenoma chemoprevention trial. N Engl J Med. 2005 Mar 17;352(11):1092-102. doi: 10.1056/NEJMoa050493. Epub 2005 Feb 15. Erratum in: N Engl J Med. 2006 Jul 13;355(2):221. PMID: 15713943. https://pubmed.ncbi.nlm.nih.gov/15713943/