A question that arises often when discussing the optimal role of science in medicine is the precise role of plausibility, or prior probability. This is, in fact, the central concept that separates (for practical if not philosophical reasons) science-based medicine (SBM) from evidence-based medicine (EBM).
The concept featured prominently in the debate between myself and Dr. Katz at the recent Yale symposium that Kimball Atwood recently discussed. Dr. Katz’s treatment of the topic was fairly typical of CAM proponents, and consisted of a number of straw man derived from a false dichotomy, which I will describe in detail below.
I also recently received (I think by coincidence) the following question from an interested SBM reader:
What would Science Based Medicine do if H. pylori was not known, but a study showed that antibiotics given to patients with stomach ulcers eliminated symptoms? I assume that SBM wouldn’t dismiss it outright saying that it couldn’t possibly be helping because antibiotics don’t reduce stomach acid. I assume a SBM approach would do further studies trying to discover why antibiotics work. But, in the meantime, would a SBM practitioner refuse to give antibiotics to patients because he doesn’t have a scientific explanation as to why it works?
This is the exact type of scenario raised by David Katz during our discussion. He claimed that strict adherence to the principles of SBM would deprive patients of effective treatments, simply because we did not understand how they work. This is a pernicious straw man that significantly misconstrues the nature of plausibility and its relationship to the practice of medicine.
Plausibility is essentially an application of existing basic and clinical science to a new hypothesis, to give us an idea of how likely it is to be true. We are not starting from scratch with each new question – which would foolishly ignore over a century of hard-won biological and medical knowledge. Considering plausibility helps us to interpret the clinical literature, and also to establish research priorities. But plausibility is not the ultimate arbiter of clinical truth – it must be put into context with clinical evidence, just as clinical evidence must be put into the context of scientific plausibility.
One common mistake when considering plausibility is to reduce it to a false dichotomy – a claim is either plausible (which is falsely equated to scientists understanding its precise mechanism of action), or implausible (which is falsely equated to not knowing the mechanism of action). Rather, at least three broad categories need to be considered with regard to plausibility.
The first category are those treatments with a known mechanism or mechanisms of action that should, according to our existing models, produce a certain clinical effect. For example, we know that beta blockers bind and inhibit beta receptors in the heart and on blood vessels and thereby reduce cardiac output and dilate arteries which lowers blood pressure. It is therefore very plausible that beta blockers would have a protective effect against syndromes that result in an excess of catecholamine (adrenalin) production, since catecholamines bind and activate beta receptors.
We may also add to this category treatments for which there is anecdotal or preliminary evidence for efficacy – clinical plausibility.
There is still a range of plausibility within this category, but in such cases there is at least some reason to think that a treatment should work. The core principle of EBM, however, is that even in such situations we still need clinical studies looking at net health effects to show that plausible treatments are safe and effective – plausibility is not enough.
The next broad category is not implausible, but neutral or unknown with respect to plausibility. For such treatments we have no particular reason to think that they should work, but no reason to suspect that they do not or cannot work either. This category would include any pharmacological substance with an unknown mechanism of action, or mechanisms that are not known to interact with the disease or symptom being treated. There is no reason to think that beta blockers would improve memory in dementia, but this is not inherently implausible. Beta blockers are drugs, and may have other effects that have not yet been discovered.
This category applies to the question above – if we had reliable clinical data that showed antibiotics worked for ulcers, even though we had no idea how, we would still accept and even use this treatment (assuming the clinical data was sufficient). This of course would then lead to further investigation – is the beneficial effect due to a pharmacological property of the antibiotic not related to its antibiotic effects, or are some ulcers perhaps caused by or exacerbated by a bacterial infection.
Dr. Katz argued that SBM eliminates the possibility of serendipity – discovering new treatments by accident through clinical observations. But this is simply not true (one of his straw men) – SBM considers all the evidence, clinical and basic science. If clinical evidence is solid, that is enough, and often in the history of medicine lead to discoveries about mechanism and biology.
The arrow of research can go both ways – understanding plausibility can lead to new treatments, but discovering new treatments can lead to discoveries about biology and mechanism. The two play off each other.
But there is a third category in the plausibility spectrum – treatments that are inherently implausible. These are treatments that not only lack a known mechanism of action, they violate basic laws of science. Homeopathy violates the law of mass action (a basic principle of chemistry), the laws of thermodynamics (extreme dilutions maintaining the chemical “memory” of other substances), and all of our notions of bioavailability and pharmacokinetics.
Homeopaths therefore substitute any notion of chemical activity with a vague claim about “energy” – but this just puts homeopathy in the category of energy medicine, which is just as implausible. Invoking an unknown fundamental energy of the universe is not a trivial assumption. Centuries of study have failed to discover such an energy, and our models of biology and physiology have made such notions unnecessary, resulting in the discarding of “life energy” as a scientific idea over a century ago.
Essentially any claim that is the functional equivalent to saying “it’s magic” and would, by necessity, require the rewriting not only of our medical texts, but physics, chemistry, and biology, can reasonably be considered, not just unknown, but implausible.
Dr. Katz and others would like us to believe that this category does not exist, based upon the premise that we do no yet understand enough science to make such judgments. They often invoke vague references to quantum mechanics or the counter-intuitive nature of subatomic physics or cosmology to make their point. But this is an anti-intellectual and unscientific approach – it denies existing knowledge.
The alternative (often another false dichotomy and straw man) is not that we know everything – no one claims that. But not knowing everything is not the functional equivalent of knowing nothing. We do know stuff, and it is folly to deny the accumulated knowledge of the last few centuries of organized science.
Having said that – even the most implausible claim can still prove itself with sufficient clinical evidence. If homeopathy actually worked, it could be demonstrated through repeated rigorous clinical studies (something which has never happened). Admittedly, the bar for such evidence would be as high as the prior implausibility of the claim – which is very high – but if it really worked, that bar of evidence should theoretically be reachable. In that very hypothetical situation, the results would be extremely intriguing – clearly there would be something fundamental missing from our understanding of the relevant areas of science – a situation that often results in Nobel prizes.
SBM is ultimately about achieving the optimal relationship between science and the practice of medicine. SBM requires considering all the science, in its proper context, and does not follow any simplistic algorithm as is often suggested by critics. We look at what is known and what is unknown, at basic science and clinical evidence, and we put it all together, making an individual judgment for each individual claim.
We also are students of history – what claims have prospered or failed in the past, and what patterns predict ultimate success or failure? One pattern that should be obvious is that of highly implausible claims (not merely unknown) that can only produce weak and preliminary evidence, where more rigorous evidence tends to be negative, and positive evidence cannot be replicated, followed by special pleading by proponents. That is the pattern of a treatment that does not work.
We have seen this pattern with treatments that are now not controversial in their failure – phrenology, radioactive tonics, animal magnetism, and Abram’s dynomyzer (turned out to be a black box with loose non-functioning parts). We see the same pattern with homeopathy, therapeutic touch, energy medicine, and acupuncture.
We also see the same pattern for highly implausible (what some would consider pathological) fringe sciences outside of medicine – ESP research, ghost hunting, free energy, and cryptozoology, for example. There are also non-controversial historical examples, such as N-rays.
Of course, those who have not learned the lessons of history are doomed to repeat it.