Trusting medical science: humanists’ teachings medically annotated

19.06.2020

Olli S. Miettinnen, Johann Steurer

This article first outlines humanists’ teachings on medicine and medical science, and then posit some annotations on these from the vantage of the authors’ critical theory of medicine and medical research. These commentaries can be thought of as providing topical “continuing education” for statistico-medical researchers, both clinical and epidemiological, and guidance for serious readers of their reports.

Introduction

Published last year was a book which, drawing from historians’ and philosophers’ ideas about science, advocated trust in consensus beliefs of medical (and other) scientists [1]. And published last year also was a book – ours – which, in a way, advocated the opposite, namely distrust of even consensus beliefs of today’s experts on “patient-oriented” clinical research; and accordingly, it called for radical transformation of patient-level, statistical-type clinical research [2].

The development of that trust-advocating book was prompted by its author’s displeasure with the now-common lay questioning of consensus beliefs of scientists, most notably as to the now-ongoing global climate change (especially its causal origin and collective manageability). Our book, in turn, arose from our displeasure with the enormous quantity of “patient-oriented” original research that already has been published and also “systematically reviewed” – its failure to have substantially advanced the requisite knowledge-base of clinical medicine. More specifically, it arose from our recognition that the lack of relevance of this research has been due to it not having addressed that which clinicians really need to know – namely, probabilities of relevant but hidden truths about their patients’ health, specifically the way these probabilities depend, jointly, on accessible and relevant facts on the cases.

Here, we first outline those humanists’ teachings on (medicine and) medical science and then posit some annotations on these from the vantage of our critical theory of (medicine and) medical research [2]. These commentaries can be thought of as providing topical “continuing education” for statistico-medical researchers, both clinical and epidemiological, and guidance for serious readers of their reports.

Humanists’ teachings on the trustworthiness of medical science

In the first of those two books, Naomi Oreskes – a professor of the history of science, and also an affiliated professor of Earth and planetary sciences, at Harvard University – makes, from the vantage of the history and philosophy of science, a case for general trustworthiness of consensus beliefs of (natural) scientists. This takes up the first two chapters of the book. Then, some colleagues of hers discuss those teachings (and some related matters) in chapters 3–6, generally endorsing them. Oreskes responds to these commentaries in the closing, seventh chapter.

The main point of Oreskes – which she advances in chapter 1, entitled “Why trust science? Perspectives from the history and philosophy of science” – is that, among the divergent claims on any given matter of science, most trustworthy generally are those of scientific experts on the matter – insofar as their individual beliefs on it have converged into a consensus, into scientific knowledge in this meaning of it. Oreskes couples this (unsurprising) point of hers with the (familiar) caveat that such consensus, too, is fallible, subject to being in error.

The main basis for the trustworthiness of a scientific consensus is, so Oreskes explains in this opening chapter, the very essence of the scientific inquiry behind it:

it has been a collective enterprise of the communityof scientists concerned with the matter;
it has been aimed at unveiling the truth(s) in question;
the scientists’ (individual) beliefs about the truth(s) in question have been based on their interpretation of the entirety of the available-and-relevant evidence;
the interpretations of the evidence have been derived rationally;
and in all of this, the scientists have been free from conflicting interests.

The rest of her teachings about the trustworthiness of the claims of science are implicit in a single paragraph of chapter 2 (pp. 143–144), reflecting the qualities of a scientific inquiry that Oreskes sees as making a scientific consensus particularly trustworthy: “If we can establish that there is a consensus among the community of qualified experts, then we may also want to ask:

Do the individuals in the community bring to bear different perspectives? Do they represent a range of perspectives in terms of ideas, theoretical commitments, methodological preferences, and personal values?

Have different methods been applied and diverse lines of evidence been considered?
Has there been ample opportunity for dissenting views to be heard, considered, and weighed?
Is the community open to new information and able to be self-critical?
Is the community demographically diverse: in terms of age, gender, race, ethnicity, country of origin, and the like?”

Oreskes’s call for addressing these questions in the context of a scientific consensus, notably when in principle trusting it, harkens back to the book’s opening page. There, the only content is the (pithy) Russian proverb “Trust, but verify” together with Oreskes’s attribution of this to an American (Ronald Reagan).

Medical annotations on those teachings: part 1

In line with these teachings of hers, Oreskes presumably gave the book’s advance readers “ample opportunity for dissenting views” on those teachings “to be heard, considered, and weighed.” But no real dissent is expressed by her colleagues in chapters 3–6.

We here do present some dissenting views, specific to the medical content of the book.

Such content is found mostly in the examples taken up in the book. In the beginning of chapter 1, Oreskes gives three examples of what scientists have recently been telling us, and then poses the questions, “But how do they know these things? How do we know they’re not wrong?”

The first among those examples is medical: “Immunologists tell us that vaccines are generally safe for most people, have protected millions of people from deadly and disfiguring diseases, and do not cause autism” (p. 15). But in our view, immunologists are not experts on the effects of vaccines; for they are bio-medical scientists, working in laboratories (rather than on people). Study of the effectiveness and safety of (the use of) vaccines does not belong in their domain of research; it is a concern in statistico-medical research, clinical and, especially, epidemiological [2].

The second example is “atmospheric physicists” telling us about climate change, and the third one is again medical: “Dentists tell us to floss our teeth.” But dentists – practitioners of dental medicine – are not (developers of and) authorities on the scientific knowledge-base for the (preventive and other) practices in this discipline of medicine, such as telling their clients to floss their teeth. Relevant here would be consensus opinions of dental scientists (on the effects of dental floss).

The content of the implied scientific consensus behind these two medical “tellings” is not presented in this chapter, nor is the nature of the research and the evidence from it. And no other examples of (purported) claims of science are presented in this main chapter of the book. Instead, this chapter addresses, almost exclusively, the general history of “the philosophy of science” (i,e., particular philosophers’ cogitations about science, rather than of scientists’ principles guiding their work).

In chapter 2, entitled “Science awry”, Oreskes addresses “the problem of science gone awry on its own accord ... of scientists coming to conclusions that were later overturned, ... [which] has been the central question of [her] research career” (p. 74). She does this in terms of “some examples in which scientists clearly went astray” (p. 75). The examples are ones Oreskes has “come to know well, and they might help us answer the question of ex ante trust, by helping us recognize cases where it might be appropriate to be skeptical, to reserve judgment, or to ask with good reason for more research.”

The first of these examples, five in number, is medical. It addresses an idea of a physician, a professor at Harvard Medical School in the 19th century. He took the thermodynamic principle of conservation of energy (in any closed system) to apply to the human organism; and on this basis he argued that higher education should not be available to women, as it would take away from their reproductive “energies.” Although this physician’s thinking obviously went (very) awry, science did not; for there was no science behind that idea.

The second example is about continental drift (which was an early, major interest of Oreske’s), but the third example is, again, medical. It addresses, very extensively, the eugenics movement of the early 20th century, and to an extent also the science of inheritance related to it.

About the state of this science at the turn of that century, Oreskes says that “Mendel’s findings” – already decades old, but only then taken note of – “seemed to rule out Lamarckian notions that individual improvements could be effected via environmental improvement” (p. 92). But: Mendelian genetics hardly was able to call into question something so obviously true; and Lamarck actually (and very questionably) held that some acquired characteristics of people can be inherited (biologically) by their offspring.

As is well known, the eugenics movement (which first developed in the United States and Britain in the earliest decades of that century) went very awry in Germany in the 1930s and 40s. But this is no justification for Oreskes’s implication that its underlying science went awry. Indeed, Oreskes presents no criticism of this science, notably in the framework of her principles of judging the trustworthiness of scientists’ consensual beliefs. She merely recounts some old criticisms by others, most notably the one that was directed against genetic determinism, the doctrine that a person’s “outcomes” are determined by their genome alone, with no role for the environment (or behaviour, for that matter). Such determinism, however, was not a scientific predicate of the eugenics movement (any more than it has ever been, or is now, a premise of improvement-seeking breeding of animals or plants).

Oreske’s fourth example of “Science awry” is research on the development of mental depression as a side-effect of hormonal contraception. She first quotes a commentary of a practicing physician, which recently appeared in the (non-scientific) Harvard Gasette. In this, the doctor praises a recent, very large, registry-based study (from Denmark) on the matter and criticises the reliance on “iffy methods” in the prior studies on the topic. From these (non-expert) opinions she proceeds to presentation of her own critical ideas about the research on a topic like that.

At issue is research on the causal origin – aetiology/aetiogenesis – of an illness, or the health effects of an intervention (preventive or therapeutic). Oreskes explains that, in this (causal) type of medical research, “the conventional approach” is that of “assuming no effect and requiring statistical proof of a specific level to say that an effect had been detected – and was therefore known” (pp. 114–115). But: actual researchers on aetiogenetic or intervention effects do not, and should not, think in terms of “statistical proofs”, nor do they, or should they, take an effect measure’s “statistically significant” empirical deviation from its null value to mean that the effect in question is “known.”

She goes on to teach that “There are two problems with this approach (i.e., with that purported conventional methodology of these studies). The first ... is that a negative finding is often taken as indicating “no effect.” And she explains that “Many negative studies see effects, but not ones that pass the bar of statistical significance at the 95% level.” But: these researchers – in line with the teachings of Immanuel Kant – understand that causation is not a phenomenon, and thus not subject to being seen or otherwise observed. They thus understand that, among instances of the outcome of interest (mental depression, say) occurring in association with the potentially causal antecedent at issue (hormonal contraception, say), the causal ones of these, if any, are not distinguishable from the non-causal ones. (This understanding secures their freedom from the post hoc, ergo propter hoc fallacy.) For this reason, the issues of these studies’ design are very challenging [2]. And in these studies, the researchers never entertain “statistical significance” anywhere near the 95% level. (Common are 5% and 2.5%.)

The fifth, final example addresses, so Oreskes says, “a very grave public health issue: dental floss” (p. 118) – its effectiveness in the prevention of dental caries and periodontal disease. But in point of fact, nothing about this procedure is being held as a grave issue in the teachings of schools of public health, nor does Oreskes substantiate her assertion of such gravity. And again, she doesn’t outline the ways in which she sees the research to have gone awry in the (implied) consensus derived from it, notably in terms of deviations from her teachings on trustworthy science.

In the context of this last example of “science gone awry”, curiously, Oreskes takes up “The most well-known and respected source of information on the state of the art in biomedicine.” But there is no “biomedicine” segment within medicine, much less a “well-known and respected source of information” specific to this. All of medicine deals with (the health of) a biological entity, the human soma [2]. In medicine-serving science/research, however, there eminently is, as we noted above, the specifically biomedical segment, very distinct from the statistico-medical one.

The “source of information” Oreskes is referring to is not “the Cochrane group” – there is no such (singular) group – but one of the numerous self-appointed groups that have contributed “systematic reviews” of medical studies (of the statistical variety) to the Cochrane Collaboration programme. The source she actually refers to is the group that recently reported on their review of randomised trials on the effectiveness of dental floss.

That review did not address a case of “science gone awry” in its consensus; it merely produced a synthesis of the available evidence from randomised trials on the preventive effectiveness of dental flossing. The main result of that review Oreskes quotes as, “Flossing plus tooth-brushing showed a statistically significant befit compared to tooth-brushing in ...”

Oreskes does not comment on this review’s (questionable) focus on experimental studies on the matter, nor does she point out, much less correct, the overt flaw in that purported finding from that “most well-known and respected source.” She does not point out that the treatments in the studies did not “show” anything (about the causal secrets of nature that were at issue), nor did the studies; that instead, the evidence from the studies indicated something. It did not indicate a “statistically significant benefit”: it merely indicated (the existence of) some benefit, the indication of this being the “statistically significant” difference in the (rates of adverse) outcomes between the trials’ two arms. (This flaw is, by the way, remarkably commonplace in reports of medical studies on intervention effects, often made even worse by writing about “significant benefit” or the like.)

Medical annotations on those teachings: part 2

Even though almost all of Oreskes’s science-related examples in the book are medical, she could not have been aware of our conception of the essence of modern medicine vis-à-vis science [2]:

that, contrary to its prevailing definitions in eminent dictionaries of medicine, medicine has not become a science (but is now, and will always be, the aggregate of its constituent arts, service professions);
that even though medicine now deploys, extensively, products and processes developed (technologically) on the basis of knowledge from laboratory-level (“basic”) medical sciences, doctors’ operational interface with science is in their pursuit of insights into important but hidden truths concerning the health of their clients;
that the species of these insights – this esoteric, ad-hoc knowing, medical gnosis – are diagnosis, aetiognosis and prognosis;
and that the requisite knowledge-base of clinical medicine is of the form of gnostic probability functions (GPFs) – diagnostic, aetiognostic, and prognostic – for suitably-defined domains of case presentation (for gnosis).

And in particular, she presumably hasn’t been concerned about the prevailing state of contemporary medicine vis-à-vis the ideal of scientific medicine:

that in “patient-oriented” clinical research up to the present, those patient-relevant GPFs have not been addressed (as the objects of study);
and that, therefore, people still cannot receive trustworthy science-based teaching– ‘doctoring’ in this etymological meaning of the word – from their doctors on the probabilities of important possibilities in their health.

Example: When a case of melanoma has been detected – rule-in diagnosed – and also resected, the first-order patient-relevant medical questions are again diagnostic, now in a second-stage meaning of this. There is the need to know, for one thing, whether the patient now is free of any residue of the melanoma at the site of the resection; and for another, whether there already are metastases of the resected lesion (in other sites). To answer these questions, the doctor needs access to knowledge about the probabilities of these possibilities – conditional on the set of all relevant and known facts about the case from the most trustworthy source of this (i.e., from the relevant GPFs, in cyberspace).

Whereas clinicians do not yet have the requisite knowledge-base for the gnostic probabilities that are of pivotal relevance in their practices, nor will this knowledge ever be produced by the prevailing melange of varieties of “patient-oriented” clinical research, needed now is the development of a comprehensive programme of suitably transformed clinical research [2] – one that continually draws from the collective clinical experience of a network of first-rate facilities of clinical care, deploying a shared system of clinical informatics designed to serve high-quality gnostic research in addition to the ad-hoc needs of the care [2].

In the genuinely scientific medicine that such a programme will provide for (in the fullness of time), the trustworthiness of medical science underpinning the teachings of clinical doctors (about their clients’ personal health) will continue to be a concern of their patients and other clients, but in a new way: As the teachings will be about probabilities (of hidden truths) specific to the case profiles (in terms of the available and relevant facts) and are read from GPFs representing the frontiers of gnostic clinical science, these probabilities will be trusted (as being preferable to any subjective beliefs) by the doctors themselves and then, through their suitably educated teachings, by their clients just the same.

Johann Steurer, Horten Centre for Patient-Oriented Research and Knowledge Transfer, University of Zurich, Zürich, Switzerland

References

Oreskes N. Why trust science? Princeton: Princeton University Press; 2019
Miettinen OS, Steurer J, Hofman A. Clinical Research Transformed. Cham: Springer Nature Switzerland AG; 2019.