A lot less than it takes to believe that it doesn’t work, apparently.
Once again, Aaron Carroll of The Incidental Economist has a quick but incredibly important point on how people interpret evidence in medicine.
Following on last week’s publication of the 25-year follow-up of an RCT of screening mammography, he links to the new JAMA paper showing just how little evidence it takes for the FDA to approve a new treatment. You should read the whole post, but the immediate take-away is that the majority of treatments were approved based on only one or two trials. More than half of approvals were based on studies that included fewer than a thousand patients.
That probably has something to do with why the BMJ’s Clinical Evidence site has categorized only 35% of the more than 3,000 treatments they’ve reviewed as “beneficial” or “likely to be beneficial,” while half are of “unknown effectiveness.”