As part of the grant I’m on at work, I am expected to attend “continuing ethics training” each year. Last Wednesday was the first of two sessions, each a little over an hour long, and I ended up presenting a case study to the other folks in the room regarding the way science is conducted and how it is perceived by the general public. This past Wednesday, however, we had a guest speaker in the form of Stephen Lefrak, a pulmonary physician that also has research interests in medical ethics.
He covered a range of subjects, but he specifically highlighted a series of studies he was involved with over 10 years ago, studies published in the New England Journal of Medicine, among other high profile journals. Studies funded by the NIH and carried out by the National Emphysema Treatment Trial Research Group (NETT). These studies involved a surgical procedure for patients with emphysema, where portions of the lung with damaged tissue would be removed, and the rest of the lung (presumably healthy tissue) would be restructured to form a better-functioning respiratory organ. Lefrak and his colleague here at Wash U were involved early on with the trial, but left after they had serious ethical concerns, one of which centered on the idea of a “randomized controlled trial (RCT).”
For the sake of simplicity, an RCT is essentially the idea that you apply one of two (or more) potential treatments to a given individual, and that individual is selected at random from a given group. In this case, the treatment was the surgical removal of lung tissue (presumably damaged) in order to refashion a healthier lung, and the group was emphysema patients. However, and importantly, it was known at the time that you can’t just do this to someone that has lung damage spread throughout the lung: it only works if there is healthy tissue still in there to salvage.
Lefrak knew it wouldn’t work if the trials were carried out at random (i.e. paying no attention to the quality of the patients lungs, or whether they had healthy lung tissue remaining, or whether they had a “homogeneous” mix of damaged and undamaged tissue). However, when this concern was raised in the pages of NEJM, he was essentially told that he couldn’t “know” it because an RCT had not been done to prove it.
As a result, almost 50% of the patients it was tried on ended up dying, for the very reason Lefrak and colleagues warned them about.
Which brings us to the title of this post: empiricism vs rationalism. “Empiricism” is what drives the belief that an RCT is essential to making the claim that this kind of lung surgery is “dangerous” to a subset of individuals. “Rationalism” is behind the idea that we actually know things about how the body works and can make an informed inference as to what the outcome would be without having to do the RCT to “prove” it.
The example Lefrak gave is that an RCT to prove that you need a parachute to jump out of a plane would be silly. We already know the answer.
As Lefrak talked about his experience, it got me thinking about where our knowledge comes from and how we build upon it. Whether I concern myself, personally, with “evidence” more than I should, without thinking rationally about a particular subject in order to come to a conclusion. I’d consider myself to be a “rational” person, but perhaps not. Then again, as he described what the surgery was seeking to do, my physiology training assured me that I would have been on his side from the beginning, rather than advocating the continuation of the NETT work.
It’s just something we, as scientists, ought to consider more often than we typically do, I guess.