The Impossible Prudence of the Robotic Doctor

Michel Accad, physician and author of the blog Alert & Oriented—as well as various academic works, including an article on my reading list about hylomorphism and cell theory—writes in a quæstione disputatae with Darrel Francis in The BMJ against the reign of “evidence based medicine.” What is EBM, why does Accad argue against it, and what light might a classically inspired philosophy of science shed upon the subject?

Evidence-based medicine (EBM) is an approach to medical practice intended to optimize decision-making by emphasizing the use of evidence from well-designed and well-conducted research. Although all medicine based on science has some degree of empirical support, EBM goes further, classifying evidence by its epistemologic [sic] strength and requiring that only the strongest types (coming from meta-analyses, systematic reviews, and randomized controlled trials) can yield strong recommendations; weaker types (such as from case-control studies) can yield only weak recommendations.

Accad himself, citing academic literature on EBM published in The BMJ itself, notes that “a well established definition of the EBM is ‘the conscientious, explicit, and judicious use of best evidence in making decisions about the care of individual patients,’” where the key terms are ‘judicious,’ ‘best’ and ‘evidence,’ as well as ‘decisions.’ Accad also brings forward the typical qualification placed upon judgment made by EBM’s promoters, which is that “good doctors use both individual clinical expertise and the best available external evidence, and neither alone is enough.”

All this is enough to make one think: isn’t EBM just a description of the usual, good practice of modern medicine? How could one be against it?

Accad first notes the connection between the notion of judgment and evidence: “Judicious use of best evidence implies that evidence is subject to judgment.” Philosophically implicit here is a regress argument. To guide judgment, one needs prior evidence, in this case of medical practice of scientific research that includes evidence that expands beyond what is possible to glean personally. Yet to recognize the relevance of evidence such that it is applicable to a specific case is an act of judgment. And that judgment must be based upon knowledge, which if it is of the sort of research evidence in question, against requires judgment, etc. This is what Accad zeroes in on: “How can evidence be a check on judgment when judgment is obviously required to appraise the quality of the evidence and its relevance to the patient at hand?”

Accad then adds that EBM tilts medical practice towards practicing according to standardized techniques or (see below) perhaps ‘algorithms’ are the apt term: “But practising according to standards is antithetical to practising according to clinical judgment: standardisation can only identify best practices for an ‘average patient’ under average conditions.” Indeed, either the average patient is considered by the medical rubrics, which are informed by studies done on “clean patients” studied for the fact that extraneous factors do no interfere with the methodical goals of the study. Yet such limpid scientific methodology is not available in the waiting room. As such, Accad concludes that EBM, designed to avoid the personal cognitive biases of individual practitioners, nonetheless “introduces a bias of its own: the tendency to treat according to population norms rather than personal needs, a ‘groupthink’ of sorts.”

Combating this cognitive bias is the very point brought in EBM’s favor by Francis (who is rather batting for the home team in this dispute—or has the home pitch, given the journal). He notes: “The real danger is not obviously nonsensical ideas but seemingly logical ones,” and continues:

EBM protects our patients not from nonsensical therapies but from rational ones that cause more harm than good. The human body is incomprehensibly complex. Unlike complex computer software, in which each component has precisely specified behaviours designed to fit together in a manner comprehensible by human software engineers, human biology underwent natural selection for providing a competitive edge, rather than for ease of describing. Moreover, even the language we use in medicine is almost incapable of describing dependence on more than one variable, never mind thousands.

However, the “clean” nature of patients subject to medical research might seem to be the very reason why a scientific method whose logical structure is of the same form as the computer software thus demeaned is not sufficient in and of itself to provide a theoretically adequate backdrop, or “evidence basis,” for medicine. Yet even were one to grant its theoretical power, this argument (Francis’ central argument) doesn’t refute Accad’s main point, which is the priority of judgment over medical evidence.

The key is that medical judgment is, in one respect, prior to medical evidence (as understood by EBM), and yet medical judgment is not prior to all evidence or knowledge. This allows us to avoid the infinite regress noted above.

The nature of evidence, in this case medical evidence from medical-scientific research, assays the structure of clinical realities and this is rooted in scientific principles. This relates two poles of the central distinction in the modes of human knowledge, which is captured in the following passage in St. Thomas (De Trinitate, q. 5, a. 1, ad 4):

When we divide medicine into theoretical and practical, the division is not on the basis of the end-goal. For on that basis the whole of medicine is practical, since it is directed to practice. But the above division is made on the basis of whether what is studied in medicine is proximate to, or remote from practice. Thus we call that part of medicine practical which teaches the method of healing; for instance, that these particular medicines should be given for these abscesses. On the other hand, we call that part theoretical which teaches the principles directing a man in his practice, although not immediately; for instance, that there are three virtues [medieval medicine: vital, natural, and animal], and that there are so many kinds of fever.

This distinction between the two modes of knowledge is still not the medical art in exercise, the actual application and exercise through habitual knowledge to concrete circumstances. Here, another old medieval distinction is helpful, that between art and prudence. The medievals thought of an art, the Greek techne, as recta ratio factibilium—the right reason about things to be made or worked. Art was the shape and structure which reason brought forth in matter through a “right” or correct consideration of means to a set end. The nature of this rectitude is such that Aristotle comments that art does not deliberate. Once the correct situation arises, the means can be applied by the artist who is skilled enough. The good speller does not deliberate (the art of grammar); Michelangelo did not deliberate about how to hold the brush. As for prudence, it was termed recta ratio agibilium—the right reason about things to be done or acted in the world. This is the structure and moral truth which reason finds between general principles of action, the good end of action one is aiming at, and the “right” or morally correct means needed and perceived in concrete circumstances of human life. Prudence, in this old sense, is moral rectitude in action. The nature of this rectitude is such that prudence must deliberate, since one must sift between the apparent and true good and the apparent and true means which are mired in concrete circumstances and are for all but the best of men are clouded and colored by our biases, passions, emotions, and personal preference. Yet, common to both art and prudence is the note that they flow from their abstracted or general principles to the concrete moment and locus of action. Since this here and now which is the stage of both art and prudence is infinitely variable of its nature—and essentially so, for the individuals realized at different times and places, even if ever so similar in their common modes of being and real features, are yet distinct in being and being known—the difference between the inexperienced and experienced practitioner shows up most clearly at the moment of decision or the work itself. The prudent man sees and decides, while the imprudent vacillates. The expert sets to work easily and swiftly, while the beginner fumbles and stutters.

Thus, we have from the classical view of human knowledge an array of distinctions at our disposal. Because of the distance between these modes of art and prudence in its theoretical, abstract beginnings (which are discovered, taught, researched, and improved  upon by attending to the concrete reality of human nature in need of healing), Accad is on solid ground with his central criticism of EBM: “Individual decision making cannot be based on general evidence, and clinical judgment cannot be specified by methodological formalities.” That is, the nature of the prudent act of the moral man, which in the art of modern medicine is married to the art of healing the patient, comes into its full being in the concrete. The general evidence and methodological formalities live in an abstracted and virtual space where medical acts are mere potencies, Platonic forms of their hypostatized selves. This evidence or knowledge provided by EBM can guide but not replace judgment. And yet, medical judgment is not prior to all evidence or knowledge. Here one must fall back upon the first principles of the medical art, and the first starting points of being a moral human being, the architectonic to which Accad’s arguments implicitly appeal.

Now, EBM as a method is not without its use. Yet note the aside as Francis queries: “Can’t we be trusted to spot a nonsensical therapy or diagnostic test? No. Humans are easily fooled, and doctors are—for now—human.” That “for now” rather gives the game away. It implies that, at some technological limit which we are asymptotically approaching, the human act at the heart of medicine can be cast aside. This is as impossible as putting the act of science into a computer, for the human soul is still required to contain, in its powers, aptitudes, and habits, the very art that we might encode and engineer into future robotic doctors.