There is an entire medical-industrial complex devoted to discovering and testing drugs to treat cardiovascular disease. Unfortunately there is a near complete absence of efforts to figure out if and when these same drugs should be discontinued.
Consider Desmond Julian. The very distinguished cardiologist has been nearly killed by his profession on multiple occasions. In a State-of-the-Art review article in JACC he writes that he
developed an episode of severe hypotension during exercise, and developed another of sinoatrial block, which led to syncope after being on a beta-blocker for >15 years, cough from an ACE inhibitor after being on ramipril for 10 years, and aspirin-induced gastrointestinal bleeding after being on the drug for 20 years.
Julian and his co-authors, Xavier Rossello and Stuart J. Pocock, write that they “suspect such risks are not uncommon.” The authors analyze the lack of evidence for four widely used cardiovascular drugs– aspirin, statins, beta-blockers, and ACE inhibitors– but note that “the issues raised apply more broadly to all long-term medications across cardiovascular diseases and the whole of medicine.” They write:
The key limitation is a gap of knowledge between the short-term evidence and the long-term use of these drugs. Although the average follow-up in RCTs is limited, these medications are often administered open-endedly over many years
Further, they note, long-term use of drugs is based on the untested assumption that the relative “benefits continue (and stay constant) over the long-term.” As well, the evidence supporting the use of some drugs may reflect an earlier era and may be less or no longer applicable in a later era. Also contributing to the problem is a rapidly aging population taking multiple drugs, increasing the likelihood of interactions and side effects.
Clinical trials testing the withdrawal of drugs– deprescribing– are much needed, the authors write. But, of course, industry has little incentive to fund such trials.
Here is the authors’ summary of their main points: