What does it mean to know something in medicine? In this episode, we’ll explore this question by developing a historical framework of medical epistemologies in a journey that involves King Nebuchadnezzar, citrus fruit, leeches, water pumps, ICD-10, Socrates, skepticism, and 1970’s computer programs designed to replace doctors. This is a version of a Grand Rounds given at BIDMC on October 25, 2019.
- Bothwell LE et al, “Assessing the Gold Standard — Lessons from the History of RCTs,” NEJM June 2, 2016.
- Khushf G, “A Framework for Understanding Medical Epistemologies,” Journal of Medicine and Philosophy, 38: 461-486, 2013.
- Guyatt G and Tonelli M, Med Roundtable Gen Med Ed.;June 13, 2012 1(1): 75 – 84.
- Morabia A, A History of Epidemiologic Methods and Concepts, 2004.
- Tonelli MR, “Integrating evidence into clinical practice: an alternative to evidence-based approaches,” Journal of Evaluation in Clinical Practice, 12(3) 248-256.
Music from https://filmmusic.io
“Tango de Manzana” and “Return of the Mummy” by Kevin MacLeod (https://incompetech.com)
License: CC BY (http://creativecommons.org/licenses/by/4.0/)
This is Adam Rodman, and you’re listening to Bedside Rounds, a monthly podcast on the weird, wonderful, and intensely human stories that have shaped modern medicine, brought to you in partnership with the American College of Physicians. This episode I called “I Know Nothing,” and it’s an edited version of the Grand Rounds I I gave on October 25th at Beth Israel Deaconess Medical Center. If you’ve been listening to the show for a while, you know that one of my overarching interests in medical history is epistemology — how we know things in medicine, and how this simple question — how do I know how to best take care of the human being in front of me? — has done so much to unconsciously shape and constrain our actions today. This Grand Rounds is fundamentally an attempt to provide a historical framework for a lot of the themes that I talk about all the time, and it has nice tie-ins to the evidence based medicine community, and to common critiques of EBM. So let’s get going!
Let’s do a little philosophy 101 housecleaning first. Epistemology might be a word that *I* love to say, but I don’t think it pops up in most people’s day-to-day speech. Like a lot of -ologies, the word actually refers to a number of different though related concepts. The classical definition is the study of knowledge, in particular, how we know. In the Western tradition, this unsurprisingly starts with Socrates and Plato, from the dialogue with Theaetetus. Essentially, in order for something to be knowledge, you must first believe a proposition to be correct, the proposition has to be actually correct, and you have to be justified in this belief. So to give a medical example, I need to actually believe a patient has heart failure, pathophysiologically, they need to have poor “squeeze” of their heart leading to venous congestion, and my diagnosis needs to be based on something real — a justified belief.
So now I want to calm everyone down, because this is not going to be a philosophy lecture. Philosophers of science like to call this level of thinking “meta-epistemology”’. At the end of the day, I don’t spend a lot of time thinking about the underlying nature of knowledge, because I have sick people in front of me who need help now. That’s why I really like this quote:
“Diagnosis is not knowledge for knowledge’s sake. It is knowledge for the sake of action. Medicine exists in order to cure, to care, to intervene, or in limiting cases, to know when not to intervene. Medicine is not a contemplative science,” which is from a philosopher about the ethics of diagnosis.
So instead I’m doing to talk about lower case e epistemologies — specific ways that we as physicians know what’s going on with your patients, and the best way we can intervene. In particular, I’m going to give an historical framework, which I hope will make it much easier to understand debates that come up on the wards.
So before I start, I have to address skepticism. So as any of my residents here know, I’m a skeptic with a lower case s, when it comes to medicine, or really when it comes to life. Like Socrates, I feel, essentially, that beliefs need to be justified, and that our default position when doing something as important as caring for other human beings should be to demand justification, and when it’s not there be honest about it. But, and this is the last I’ll talk about meta-epistemology, there is also a fundamental problem with epistemology generally called philosophical skepticism. In fact, in the Theaetetus, Socrates points it out — what does it take to be justified in a belief? Justification is ultimately meta-recursive. Ah, I know beta blockers reduce heart failure mortality because of the MERIT-HF trial. But how do you know there weren’t negative trials that weren’t published? Or that there was some fundamental problem in randomization? Or that the Lancet didn’t just fabricate the trial wholesale? Philosophical skepticism ultimately leads to fundamental unknowing, as in this famous, though paraphrased and taken way out of context quote from Socrates: “I am the wisest man in Athens, for I know one thing, and that is that I know nothing.”
But I prefer the version from Rose Leslie’s character from Game of Thrones, in part because of epistemological connection of having the hero named after the most famous epidemiologist of all time: You know nothing, Jon Snow.
All of this is to say, while I’m not clever enough to have a meta-epistemological reason against philosophical skepticism, I’m going to take the pragmatic approach and just shrug it off because, well, we all have patients to take care of.
So let’s talk about lower case e epistemologies. So we have a mixture of training levels in this presentation, some of whom actually trained before the phrase “evidence-based” started to become a meaningless prefix added to most everything. The EBM movement is complex, and still changing, but one thing it did do, especially in the early days, is provide a coherent epistemology. Probably everyone here has been some example of this pyramid — on the bottom is expert opinion, then case reports, then cross-section studies, case-control studies, cohort studies, leading to randomized controlled trials and at the top systematic reviews and meta-analyses of randomized controlled trials. This hierarchy, even if the major proponents of evidence-based medicine have largely stopped promoting it, underlies the modern approach to medical knowledge. I mean, I doubt there’s any single person in this room who would try to suggest that their approach to medicine was not “evidence-based.”. There’s only one way of knowing, it suggests — there’s just low quality and high quality, just as, say, an Impala and a Tesla are both cars, but there’s very little to disagree about in terms of quality difference.
But it pains me when I see this taught to medical students and residents, because, I mean, it’s self-evidently not true, which we all know from our experience in treating patients.
My goal in this lecture is to show that there are, in fact, multiple complementary epistemologies, and that far from existing in a hierarchy, they complement — and sometimes contradict — each other. And because this is me, I’m going to illustrate this with an historical framework, because I think it’s useful to know what particular problems our forbearers were struggling with as they developed new methods to know.
And one final aside before I get started. I want to challenge everyone here to think about this question: is medicine a science? The way that this is usually phrased is the science-art dichotomy.
- I want to start with a case, which I have personally designed to be as mundane as possible. A 63 year-old woman with a history of coronary artery disease comes to the hospital with shortness of breath, orthopnea, and weight gain. You take a look at her, she she’s breathing fast and shallow — 22 times a minute, with her oxygen saturation hanging around at 89%. Her neck veins are distended, she has swelling in her legs, and when you tap on her lungs, there’s dullness at the bases. You guys pretty much know the diagnosis before you’ve gotten a chest x-ray, labs, or a heart ultrasound.
But how do you know that diagnosis? Now, thousands of med-ed pages have been spilled about what constitutes diagnostic epistemology, but I think the approach that we fundamentally is one of the oldest in medicine — observation, or empiricism.
Let’s have a little aside about the word empiricism, which might be the dirtiest word in medicine. What does empiric mean to you? Like, if you give a patient “empiric antibiotics”? It means a lack of evidence, right? Pretty much that you’re guessing — let’s cover ALL the bugs! But the root word is actually “experience,” and that would be the ancient definition of the world — knowledge that comes from personal observation and experience. Of course, by the middle ages, “empiric” actually became an insult, essentially a synonym for “quack,” meaning someone who didn’t care for evidence, which is kind of the usage when we talk about “empiric antibiotics.” Then you have the movement of “skeptical empiricism” in the early 19th century, where the word is again redefined to mean rejection of unjustified theory, and use of scientific observation — hence the modern usage of the word “empirical.” So depending on context in medicine, empiric means observation, guessing, quackery, or using scientific evidence. This is why I love our field.
Observation is the oldest extant method to know in medicine, so old, in fact, that it dates to at least 5,000 years ago. This image, which I like to use a lot, is from the Edwin Smith papyrus, which is surgical and gynecology treatise from 1500 BCE, though apparently the language is archaic so it’s likely copied from a much older text. It’s traditionally attributed to Imhotep, so he gets to be considered both a father of medicine and architecture. This is essentially the table of contents of the papyrus. And what I think is no remarkable is, well, it looks like of like a modern surgical text, right? We only have part of it, so it’s going from head wounds down to the ribs.
I’ve selected Case 45, because if this patient were seen at the BI, they’d be on my service. I have the original in hieroglyphics on the left for reference. This is Bulging Tumors of the Breast. As the language here should be strikingly familiar to any of us: the examination: “if you put your hand upon these tumors, they are cool, without granulation. They form no fluid, nor generate secretions for fluid, and they are bulging. Touching them is like touching a ball of wrappings, similar to a green hemat fruit.” The diagnosis: “One has tumors,” and treatment, “there is no treatment.”
This is actually the first description of cancer that we have. And one reason that I always present this case is because of the resonance it has — now, I’m not sure exactly what a hemat fruit is, but I think everyone here knows the emotional weight of examining a breast, and feeling a cool, firm mass underneath. It isn’t just the content of this case that makes it relevant to medicine today, but the form. This is a case presentation! This is basically what our residents do every single morning! The most famous texts of the Hippocratics, which were studied and memorized until the 19th century, are essentially just long lists of medical cases. I mean, I even felt compelled to make up a case to explain my thoughts in this lecture. And that is because at a fundamental level, our own personal experience — and then the experiences of colleagues and predecessors — make up our fundamental, and first, epistemology. In fact, many types of medical knowledge are solely based on empiricism. Most famous is probably diagnosis. For all our attempts to ground diagnosis in, say, Bayesian theory, at its core, constructing a differential still operates from a mixture of cases that we’ve experienced ourselves, or that we’ve studied from others. The idea of an “illness script” would be fundamentally familiar to the Hippocratics.
But even in everyday cases on the wards, empiricism still leads the day. Take our heart failure patient here. How should I diurese her? What diuretic does should I start with? My residents always tell me 1-2 liters were day, and that I need to check their electrolytes twice daily. I tell them that in many patients, you can take off a whole lot more, and in the vast majority of people, there is not need for twice daily electrolytes (which the ACCF agrees with). But at the end of day, fundamentally this is based on both of our personal experiences.
This brings us to our second epistemology — theory. And by this, I’m referring to the older definition of the world, which means an overarching set of facts and principles that constrain reality. The beginnings of a theoretical approach to medicine are lost in time, but presumably it’s somewhat newer than observation. My friend Tony Breu likes to joke that I can’t make it through a lecture without mentioning the four humors, and I wish he was here so he could see that he is, once again, right. For well over two thousand years, the idea that all health and disease were related to imbalances in four fundamental fluids underpinned much of medical thought. In fact, the empirics were originally scorned because they abandoned all theory — they only cared about what they observed, and whether or not a theory worked. There have been other theories in medicine of course, but all of them — humoralism, Paracelsusianism, vitalism, Broussai-ism — all of the were more or less put to rest by the skeptical empiricists of the 19th century. In fact, at first blush, theory in this sense — a predetermined set of rules and regulations — seems to only persist in a couple of alternative medical therapies that are largely a holdover from the 19th century such as vertebral subluxations in chiropractic, or Hanhemann’s law of similarity in homeopathy.
But wait, aren’t there truly fundamental laws and rules that control and constrain all activity in the human body? Because really, anatomy, physiology, and more recently immunology and biochemistry really are modern humoralism. Early 20th century physiologists recognized this, when they playfully described non-cellular immunity as “humoral.” Their laws are based on experimental empiricism, which is a whole other epistemological beehive that I’m not going to shake, but from the perspective of the individual physician, and especially for the internist, physiology offers us a wonderful epistemological framework . If my heart failure patient becomes light headed after I give her 100 mg of intravenous furosemide, for example, I assume it’s because of intravascular fluid shifts leading to relative cerebral hypoperfusion, and might slow down. Yes, it’s based on “science,” but at a fundamental level, the reasoning is the same as prescribing a purgative for my phlegmatic patient suffering from consumption.
Those are the two ancient epistemologies that are still commonly used today. Now let’s get to the modern era.
So first, let’s talk about experimentation. One of the joys of reading a LOT of medical literature, especially from an historic perspective, is that you can see how footnoting constrains everything on the subject that comes after it. So know that this example is not truly the first medical trial, it’s just fun. I’m going back some 6,000 years to the peer-reviewed publication of the Book of Daniel.
The relevant verses, if you’re following along from home, are Daniel book 1, verses 1-21. Nebuchadnezzar has sacked Jerusalem and besides looting the temple, he’s carried off the finest young royal Israelites to serve in his court. Daniel didn’t want to “defile” himself with the meat and wine from the King’s table, so he proposes an experiment.
For these trials, by the way, I’m using the PICO format — population, intervention, comparison, outcomes, another EBM method for clarifying clinical questions. On the left you can see the PICO question: in children in whom there was no blemish, does eating only pulse (which is lentils and beans) and water, compared to eating the King’s meat and wine, lead to improve “countenances looked upon” by the King. Or if you’re more into quality improvement, you can see a process diagram on the right.
Daniel’s experiment was a success — the young Israelites, with their vegetarian diet, in fact did have better countenances for Nebuchadnezzar to look upon, and the entire court was soon thanking Daniel as they ate lentils and beans. Now I want to be very clear here. While I think this fundamentally is an experiment, it is in no meaningful way a clinical trial. There is not a disease, nor is there an understanding that this diet would be a medical intervention. This is a fun example, that sometimes gets taken too far in the literature. We don’t get the first things approaching clinical trials until the early modern period, as Enlightenment thinking starts to spread into medical practice. There are a few possibilities to get the honor of the first trial — Ambroise Pare and his surgical balms certainly stands out — but what is generally considered the first clinical trial comes from the Scottish surgeon James Lind.
It should be no surprise that Lind was interested in scurvy — during the age of sail, it’s estimated that almost two million seamen died of vitamin C deficiency. Lind’s famous experiment was carried out on board the HMS Salisbury in May of 1747, performing a blockage in the English channel. For something that would fundamentally change our approach to knowledge, it’s very short — the description occupies just four pages in his 430 page Treatise. An unknown number of seamen had fallen ill with scurvy, and rather than treat them all with the standard of care, Lind decided to perform what he called a “fair test” to decide which of the purported therapies for scurvy worked best.
Again, Lind’s question — in 12 sailors on board the HMS Salisbury, does a daily ration of two oranges and one lemon, as compared with either 1.1 liters of cider, 25 mL of dilute sulphuric acid, 18 mL of vinegar with meals, 0.5 pints of sea water increase the ability to return to duty? While he didn’t truly think of a placebo, he did include a comparison group — a medicinal paste which he felt would “open the stomach.” And the answer was a dramatic yes — at the end of six days, one of the citrus group returned to duty, and the other was well enough to act as a nurse for the rest of the sick men. The others did not particularly improve.
For a variety of reasons I’m not going to get into, it took years for citrus to become the standard treatment for scurvy. But a couple of points are important here — by nature, a clinical trial tests knowledge independently of theory. You don’t NEED to understand how citrus works — in fact, Lind’s attempting to understand it within a humoral system partially led to its delay. And while observation led to the selection of therapies to test — each of the five was a previously “standard” therapy — a trial again provides information independent of observation alone.
Second, Lind recognizes from the beginning that his trial needs to be as “fair” as possible. While he doesn’t describe allocation criteria, he makes note that he wants the men to be “as equal as I could make them,” and he treats them all in the same location (the forehold) to make sure there are no environmental factors at play.
And if you look at the evolution of clinical trials through today, you can see increasingly sophisticated attempts to make trials “fair.” Blinding is first introduced by Lavoisier in the Franklin Commission, which investigated mesmerism, as an attempt to minimize what we now call the placebo effect. By the middle of the 19th century, alternative allocation — that is, giving every other patient a treatment — became increasingly used to create two comparison groups. While the word placebo had existed as a treatment since the late 18th century, Austin Flint introduced placebo treatments as the comparator in a very ethically dubious trial on treating rheumatic fever in prisoners in 1863. Rather than alternative allocation, Bullowa introduced random allocation, a coin flip, in 1928 to prevent doctors from unconscious bias. And in 1948, Bradford Hill simply decided that there were far too many chances for physicians to consciously or unconsciously cheat in clinical trials, and introduced double-blinding — of the patients, of the physicians taking care of them, and of the physicians in charge of determining a treatment effect. This double-blinded randomized controlled trial remains essentially the form of clinical trial that we have to this day, though with increasing mathematical and statistical sophistication.
Which brings me to the final of the Types of Knowing. I’m not sure what best to call it. Maybe even a decade ago, we probably would have called it “epidemiology,” but more and more people have settled on “population medicine,” which is the terminology I’ll be using. Whereas experimentation grew out of a desire to test individual therapies, population medicine grew out of a desire to expand and perfect the process of observation. The early 19th century saw many of the centralized bureaucracies of Europe gather increasingly specific information on their citizens. In fact, it was in studying this information — on the chest size in Belgian soldiers, to be specific, that showed an astronomer named Adolphe Quetelet that there was such a thing as a normal distribution. It was only a matter of time before this sort of methodology was mapped to medical care.
Like experimentation, population medicine has arguably a progenitor, though unfortunately he remains obscure today. So let’s talk about the skeptical empiricsts of the early 19th century in Paris — the Paris Clinical School. For the first half of the 19th century, Paris was the center of the medical world. In fact, a number of well-heeld and intelligent young Boston physicians made the journey, studying how disease was located in the tissues from Bayle, localizing disease on the exam, including with a new-fangled device called a stethoscope from Laennec, and important for this discussion,using mathematics to analyze observations from multiple patients to determine underlying truths about disease — the numerical method, by Pierre Louis.
But this was not the only group of the Paris Clinical School, even if they get all the press today. The most prominent member was not, in fact, a skeptical empiricist, but Francois Joseph Victor Broussais, who advocated a system he called physiologic medicine. The details aren’t important, but you should know that it was incredibly popular, both in France, England, and the new United States, and it involved copious amounts of bloodletting, especially with leeches. Bloodletting is an ancient therapy, but is traditionally very mild. A barber might remove essentially what our residents take off during “morning labs.” During this period though advocates of Broussais-ism, and other “heroics” took off increasing amounts of blood, amounts that we would consider dangerous today. The gist of Broussais’ argument against the skeptical empiricists was, it’s great that you can identify disease in the tissues and on the physical exam. Your patients are still dying. At least I’m doing something about that!
The arguments between the skeptical empiricists and Broussais were vehement and political. But they inspired Pierre Louis to try and use his numerical method to study the effects of bloodletting on severe pneumonia. Louis considered himself a Hippocratic, and kept detailed records of all his patients at the Charity Hospital in Paris. He chose pneumonia to study bloodletting because this group of patients was all healthy at the beginning of their illness, as opposed to, say, consumptives. Louis found 77 appropriate patients, and grouped them into two groups — early bloodletting, and late bloodletting. His logic: if bloodletting truly was a lifesaving therapy, certainly the group that got it first should be in a better shape. He compared them in a number of different domains. But the most shocking results was a 44% mortality in the early group, and only 25% in the late, which Louis called “startling and apparently absurd.” Broussais was discredited, and bloodletting largely fell out of favor.
Pierre Louis was never terribly famous, which is when I suspect many of you have never heard of him today. But in his time, he was a sought-after teacher of the Paris Clinical School, and his students and acolytes would spread his numerical method across Europe. You’ve almost certainly heard of some of them — the bookseller Lemuel Shattuck, here in Boston, who would found the American Statistical Society, and William Farr, who would become the registrar-general of the UK, and arguably the first epidemiologist. Now the epidemiologist-historian Alfredo Morabia has written wonderfully about the history and development of population medicine, and unlike clinical trials there’s not nearly as clear a path to follow. But the next two slides shown the most prominent examples from its beginning. The first, of course, is the map of Soho by the anesthesiology John Snow, who mapped out cholera cases surrounding the Broad Street pump during the second cholera epidemic, and then had the pump handle removed. The second slide is arguably the second graph in history — a coxcomb of deaths of the British Army in the Crimea by preventable and unpreventable causes, showing the effectiveness of basic sanitary interventions, made by William Farr and Florence Nightingale. This graph, and the report it accompanied, fundamentally reorganized military medicine, and cut deaths in war by almost 90%.
The evolution of population medicine has come in fits and starts. I have some highlights that I’ve merely chosen because I find them fascinating — the international classification of disease, formalized by Bertillon, but essentially William Farr and Florence Nightingale’s system, the introduction of case-control studies, which became so essentially in identifying risk factors for chronic diseases. The introduction of “cohort” trials, which followed a population through time — the British Doctors Study was the first, by Bradford Hill and Doll, but Framingham soon followed here in Massachusetts. And then we have attempts to use all of this population data to make decisions for individual patients. We have expert systems in the 1970s, essentially computer programs that could “think” like doctors. MYCIN was the first for antibiotic selection, but INTERNIST-I was a fully-fledged internal medicine doctor. Those have largely died out – but decision tools like the Well’s score for VTE, TIMI for heart attack, or 4T for heparin-induced thrombocytopenia have largely become essential and invisible in modern medicine.
So let’s go over the four types of knowing that I’ve outlined again. The oldest is observation — based on our own experiences, or the experiences of others. The analogue to this is “expert opinion”. Next we have theory, a set of underlying rules about how the human body works. In modern parlance, this would be physiological thinking. Getting to the modern era, we start to have experimentation, clinical trials. This is the most obviously “scientific” type of medical knowledge. We ask a clinical question, and try and design a clever way to study it. Finally, we have population medicine, essentially the abstraction of observation. Instead of observations of individuals, we can look at dozens, hundred, or hundreds of thousands of individuals at a time — with increasingly sophisticated tools to minimize the biases that can skew even individual observations.
Let’s look back at the hierarchy of evidence here. Lest anyone think I’m being overly critical of the EBM movement, I want to be clear — this sort of hierarchy was largely abandoned in the early 2000s, to be replaced by the more subjective GRADE criteria is the classic example. Even though, I think this idea of a hierarchy has had a large impact on our medical thought. And one of the biggest effects has been to imagine what we do is a scientific, epistemological whole. As we move up the pyramid, we just get closer to the truth.
What I’m trying to argue here is that this is absolutely not the case. An historical view reveals not progressive improvements in our ability to know, but unique epistemologies designed to answer completely different clinical questions for the benefit of individual patients. Lest you think I’m crazy, I want to point out that I am also not the only person to argue this. While I’ve taken an historical perspective here, Mark Tonelli, a pulm/critical care doc at the University of Washington has been one of the most dogged critics of EBM’s hierarchy, and gives a similar “universe” of epistemologies, as he calls it, dividing medical decisions as coming from clinical experience, pathophysiologic rationate, patient values, results of clinical research, and system features.
Let’s go back to our patient with acute decompensated heart failure, and think about the sort of daily questions that come up in managing her. What sort of diuretic should I give? How much? How much fluid should I take off each day? How often should I check her electrolytes while diuresing her? How should I tell when she is dry? Do heart failure patients even need to be dry to have good outcomes, or just have their symptoms alleviated?
I answer these questions using a variety of different epistemologies. For example, I take off far more fluid daily than my residents do, which I base on an amalgamation of reasoning from an RCT (the DOSE trial) and my own personal experience. I check electrolytes only once daily, based on my physiological understanding of renal pathophysiology, plus the consensus statement from the ACCF — expert opinion. I follow a combination of markers — including the veins of the neck, lower extremity edema, weight, and serum electrolytes, as well as the patient’s symptoms, to figure out when they’re dry. This is based largely on an amalgam of population-based studies, as well as my personal experience.
And this is is just a single case. I use these different forms of “knowing” every day, on every patient that I see. And sometimes epistemologies come into conflict. A simple example — the creatinine “bump” in heart failure. I was always taught, and reasonably enough, that a bump reflects relative intravascular volume depletion, and that I should hold diuresis. There’s a purported physiological reason, but mostly this is expert opinion. But there’s always randomized controlled trial evidence that suggests such a bump doesn’t prolong hospitalization, and some evidence from biomarker studies that it’s not associated with tubular injury. This leads to disagreement between clinicians — but really, how are you best equipped to know what’s right?
I now want to get back to the original question that I asked all of you in the beginning — is medicine a science?
I’m curious to hear your thoughts. But I’d argue that the answer is a resounding no, at least if you think of science in the traditional positivist sense, like physics. We’re not even a social science. Because medicine is, for better or worse, an epistemological minefield. With every patient and every decision, we need to weigh evidence coming from a combination of sometimes contradictory epistemologies, and do our best to apply it to an individual. I’d argue that our field has far more in common with engineering, or the law, or even the clergy than it does with science.
Which is why I want to go back to the ancients to give a new framework to understand medical knowledge — and that’s the concept of phronesis. This comes, by the way, from an excellent book called How Doctors Think, not by Jerome Groopman, which has the same title, but by Kathryn Montgomery, which I highly recommend as a pretty quick and easy read if this topic interests you. Phronesis is an ancient concept of “practical reasoning,” which Aristotle points out is the same that sailors or engineers use.
That’s almost ist for the lecture. I want to point out that I realize that there are weaknesses of my historical framework for epistemology — in particular, that it deals with the decisions only of an individual physician towards a patient, and not inputs from the outside world, or the patients themselves, which I think is a strength of Tonelli’s framework. But there’s an advantage too — if there has been an historical evolution towards new sorts of epistemologies, can we potentially predict new ways of knowing things? It’s a compelling question, because the two newer forms of epistemology — experimentation and population medicine — were greatly inspired by scientific, social, and political movements outside of medicine.
I will offer two possible new epistemologies, with everyone understanding of course that if I could predict the future, I would be very reach indeed, which I most certainly am not. The first is Bayesian reasoning, or conditional probability. This isn’t really NEW per se — it’s existed for the past 50 years — but there are EBM advocates, especially in the emergency medicine world, who would argue that we should be supplanting much of our observation with conditional probability. That is, when a disease or complication comes up, we should figure out the prevalence, then figure out the conditional probabilities of various tests or treatments, and then we’ll have an answer based on “science”. A couple of problems though — we’ve been trying this a half century and it hasn’t caught on, and I don’t see any reason why it would now. Fagan’s nomogram, after all, dates from the 70s. And secondly, this gets into the whole Frequentist versus Bayesian debate that I have been sidestepping, and that I’m super unqualified to talk about, but I’m not really sure that Bayesianism truly is a new epistemology.
The idea of big data, or deep learning, or artificial intelligence, or whatever you want to call it, is fascinating to me, though, and might truly offer a new epistemology. Might there be connections between discrete data points that are otherwise obscure to physicians that we can use to care for the individual patient? And would this offer a new way of knowing in medicine? Whether yes or no, I suspect that we will have the answer to this within our lifetimes.
And of course, I offer my TL;DR at the end of every lecture, because it’s hard to pay attention to me talking for an hour.
- Observation, theory, experimentation, and population medicine are all distinct methods for making decisions about our patients;
- There is not one “better” form of knowing; different circumstances call for different epistemologies – often all four of them;
- Different epistemologies evolved to solve different types of clinical problems, and it is likely we will see new ways of knowing in our careers;
- Medicine is best described not as a science, but as a form of flexible, practical reasoning that often uses science; and
- Medical history and philosophy are relevant to clinical practice!
That is it for the episode! (and lecture). I hope everyone enjoyed it. Preparing this lecture has really been a culmination of the thousands — and distressingly, probably tens of thousands — of pages I’ve gone through over the past several years working on Bedside Rounds. And I’m not done with it either. I have some interesting stuff coming out on evidence-based medicine, the McKeown thesis, and Bayesian expert systems that will deal with very contemporary debates about epistemology. This episode is really summative about why I make Bedside Rounds — because an historical perspective has real effects on how I practice medicine today, and I hope that it does for you guys too!
You can get CME/MOC credit just for listening if you are an ACP member. Go to www.acponline.org/BedsideRounds. All of the episodes are online at www.bedsiderounds.org, or on Apple Podcasts, Spotify, Google Podcasts, or the podcast retrieval method of your choice. I’m also on facebook at /BedsideRounds. The Bedside Rounds Twitter account is @BedsideRounds. And I’m personally @AdamRodmanMD, and occasionally make Tweetorials and nerd out about internal medicine.
All of the sources are in the show notes.
For a long time, my outro has had a disclaimer to “contact your primary care provider” if you have medical questions. I chose the term provider because I wanted to be inclusive of the many non-physicians who provide primary care, including PAs and NPs. I don’t personally like the term provider, since it connotes a sort of commercial relationship that I don’t think truly encompasses the whole or importance of the physician-patient relationship. Like, I have an internet service provider. My professional society, the American College of Physicians advocates AGAINST using the term provider, either using the term physician when talking about MDs and DOs, or practitioner, when talking about a larger group. So as a believer in the importance of phronesis and the practice of medicine, I give you my new disclaimer:
While I am actually a doctor and I don’t just play one on the internet, this podcast is intended to be purely for entertainment and informational purposes, and should not be construed as medical advice. If you have any medical concerns, please see your primary care practitioner.