Wednesday, January 17, 2018

There are few parachute-like practices in family medicine

I try my best to practice evidence-based medicine on a daily basis. When I know that the test or intervention that I am recommending for my patient is based on expert opinion rather than reliable data on patient-oriented outcomes that matter, I invariably make a point of saying so. It has been my position for several years that despite the impressive effectiveness of newer antiviral medications for hepatitis C at producing a sustained virologic response (SVR), there are still not enough data to be certain that SVR always represents a "cure," and therefore not enough data to warrant age cohort-based screening of adults without known risk factors for the infection. In a recent Medscape commentary, I went one step further, mentioning a famous 2003 BMJ paper on "Parachute use to prevent death and major trauma related to gravitational challenge," which asserted that the health benefits of some interventions are so glaringly obvious that, like parachutes, they do not need to be evaluated in randomized, controlled trials (RCTs). Screening for hepatitis C, I contended, should not be considered a "parachute" for clinical research purposes. As some colleagues and I argued a few years ago, a randomized trial of screening versus usual care would not only be ethical, but logistically feasible and well worth the investment.

In an ingenious analysis published today in CMAJ Open, a team of researchers that included my friend, colleague, and prolific tweeter Dr. Vinay Prasad used Google Scholar to identify articles that cited the BMJ parachute paper to argue that a medical practice was analogous to a parachute - or in other words, so obviously beneficial that RCTs were not needed. The team then searched the literature for previous or subsequent RCTs that tested the practice in question. Of the 35 practices, 22 have, in fact, been tested in one or more RCTs. Guess how many of these practices ended up being backed up by trials that showed a statistically significant benefit? Only 6 out of 22, barely edging out the 5 "obviously beneficial" practices that were actually found to be ineffective in RCTs (the remaining 11 had mixed results or halted or ongoing trials). The investigators concluded: "Most parachute analogies in medicine are inappropriate, incorrect or misused."

Although some interventions that were refuted by RCTs lie outside of the scope of family medicine, I took note of two that not only sounded familiar (because I had once been told by an "expert" that they were true), but where I could personally make an impact on decreasing ineffective, potentially harmful care. Compared to medical therapy, stenting for renal artery stenosis does not reduce cardiovascular events. Compared to standard hemoglobin A1c targets, tighter control of blood glucose levels in persons with type 2 diabetes does not reduce cardiovascular deaths. In particular, I have inherited several adult patients with type 2 diabetes whose previous physicians tried to push their hemoglobin A1c levels to 6.5% or lower by adding expensive second or third drugs that increased their risk for hypoglycemia, based on the faulty assumption (parachute!) that these would prevent a heart attack or stroke somewhere down the line. But I practice evidence-based medicine, not parachute-based medicine. I discontinued those unnecessary medications to prevent further injury to these patients or their pocketbooks.