key: cord-0009185-cpidk3nj
authors: Wears, Robert L
title: The limits of techne and episteme()
date: 2003-12-29
journal: Ann Emerg Med
DOI: 10.1016/j.annemergmed.2003.08.007
sha: de73cb5959ef98692db693d16b4740ff771205b4
doc_id: 9185
cord_uid: cpidk3nj

nan

The ancient Greeks identified 4 different kinds of knowledge. Two of them, episteme and techne, are dominant in medical research. Episteme involves abstract reasoning about idealized, disembodied, universal concepts. Its methods are deduction and generalization; it is the basis for modern science and appears in medicine as the quest for methodological rigor. Its most muscular manifestation is the evidence-based medicine movement. Techne involves a mix of the abstract and the practical. Its goals are the development and testing of "recipes"; it is the basis for modern engineering and appears in medicine largely in the form of numeric reasoning, such as applied statistical inference.

Techne and episteme have dominated Western thought since the Renaissance, and no one would deny the benefits that modern science and engineering have produced in health care. In fact, they have become so dominant it is sometimes hard for us to conceive, much less admit, that they may be limited. To some, no other forms of legitimate knowledge even exist. Taken together, techne and episteme are judgments about how knowledge is to be verified, codified, and expressed, once it has been discovered. 1 In this issue of Annals, 2 seemingly disparate articles combine to illustrate the limits of techne and episteme. Reed's 2 article on crossover trials presents a paradox, where the methods of techne demonstrate the limitations of techne, and Tham's 3 article describing the emergency care system's response to the severe acute respiratory syndrome (SARS) outbreak in Singapore illustrates the shortcomings of both.

Crossover trials, where each subject participates in each arm of a study, thus serving as his or her own control subject, have a strong intuitive appeal. Indeed, Reed 2 found 84 crossover trials in a 7-year search of 3 prominent emergency medicine journals (including Annals)-about 1 per month across all 3 journals. Reed notes that, in a sample of these studies, none explicitly examined the possibility of carryover effects, the most important bias peculiar to crossover designs. We then expect a technical explanation of how to do a statistical test for carryover, accompanied by exhortations to be more careful next time, but instead a review of the main methods for such testing reveals that none of them work. Even worse, the current opinion of experts in the field is that no successful method will be devised. 4 The careful application of techne to this issue leads to the conclusion that techne will not be able to help us.

Tham's 3 narrative about Singapore's response to the SARS outbreak exemplifies a situation where both techne and episteme are inadequate. If Tham's article had been a scientific study of the development of a clinical decision rule to identify patients at high risk for SARS, reviewers would have found it seriously flawed. Their screening tool was put into use 4 days after the first nurses became ill; at that time, there had been only about 10 cases of confirmed SARS at their hospital, not nearly enough to support development of a screening test. It was repeatedly modified during the course of its use. And, even if they had been lucky enough to come up with a decision rule that was highly sensitive, the confidence limits on sensitivity would have been ludicrously wide.

Similarly, if the article were an evaluation of their interventions, it would also have been flawed. There is no comparison group, the outcome measures are unclear, and the opportunities for bias in assessing the outcome are great. In the world of techne and episteme, there seems to be little that can be learned here.

But, we don't live in the world of techne and episteme. We live in the "real world" of changeable and unpredictable situations, in a fluid and chancy reality, with limited time, too much information but not enough of break illustrates the value of metis in uncertain situations. It also raises important, hard-to-answer questions. In the face of a new, unknown threat, relaxing the evidentiary standards for taking action seems easy, but what about situations that are less clear, times that are less than a crisis, but are still ambiguous and disconcerting?

Both articles illustrate the value of knowledge that is not captured in the objectified worlds of episteme and techne. We have subordinated this knowledge for so long that we risk its disappearance. Revaluing it and finding the proper balance among these types should be one of the central intellectual questions medicine faces. How can we cultivate and value metis while simultaneously avoiding the slide into romanticism, anecdote, and superstition? We may not see answers even far off, but agreeing on the importance of the question would be a good first step.

The Center for Safety in Emergency Care is funded in part by a grant from the Agency for Healthcare Research and Quality. the right kind, and one where waiting for a better decision can be riskier than acting on hunches and inadequate information.

What is needed in these situations is a third type of knowledge, which the Greeks called metis . Metis is frequently translated as "cunning," but that does not do it justice. What it refers to is a kind of oblique, conjectural, often tacit, unaware, and intuitive knowledge drawn from long practical experience. 1,5 Metis is a difficult-toexpress knowledge "… of short cuts, of sagacious envisioning, of perspicuous intervention, even more mutable than the situation it has to cope with." 5 It is more involved with producing new conjectures and new knowledge than establishing the truth of a given conjecture. We are familiar with this type of knowledge in the clinical world, although we have come progressively to distrust it for a variety of reasons. 6 Tham 3 and colleagues were able to use this sort of knowledge ( metis) about how diseases act, how people react to them, what resources were at hand, and how social systems work to quickly devise an adaptive response to a dynamic, uncertain, and changing situation. We can never know for sure whether their responses were "correct," but it seems likely that waiting for a scientifically sound evidentiary base would have been a worse strategy.

Returning to crossover trials, what should we do about them? The easy and obvious answer is simply to ban them. Reed 2 points out that the US Food and Drug Administration has, in effect, done that, although some disagree with their stance. 4 A more enlightened approach would be to require that authors (and editors and reviewers) explicitly examine the question of carryover effects. If, on the basis of their knowledge of the domain they believe carryover can be neglected, then they should clearly indicate that the results are contingent on the validity of that assumption; if not, then the crossover design must be abandoned. This will require a degree of subjective insight and judgment (that is, metis) that some may find uncomfortable. This discomfort is unwarranted, however. All studies, whether crossover trials or not, contain innumerable unverified and unverifiable assumptions. They may seem less bothersome because many are less obvious, and many we have simply gotten used to, but they are there nonetheless. Reed's article on crossover trials illustrates what an important role this sort of knowledge plays in the research world, even though it is seldom spoken of there.

Similarly, Tham's 3 story of how workers in Singapore reacted adaptively to the unknown in the SARS out-

Seeing Like a State: How Certain Schemes to Improve the Human Condition Have Failed

Analysis of two-treatment, two-period crossover trials in emergency medicine

An emergency department response to severe acute respiratory syndrome: a prototype response to bioterrorism

Cross-over Trials in Clinical Research

Oblique knowledge: the clandestine work of organizations