Abstract
Clinical equipoise (CE) has been proposed as an ethical principle relating uncertainty and moral leeway in clinical research. Although CE has traditionally been indicated as a necessary condition for a morally justified introduction of a new RCT, questions related to the interpretation of this principle remain woefully open. Recent proposals to rehabilitate CE have divided the bioethical community on its ethical merits. This paper presents a new argument that brings out the epistemological difficulties we encounter in justifying CE as a principle to connect uncertainty and moral leeway in clinical ethics. The argument proposes, first, that the methodology of hypothetical retrospection (HR) is applicable to the RCT design and that it can accommodate uncertainty. As currently understood, however, HR should give up its reliance on the assumption of uncertainty transduction, because the latter assumes the principle of indifference, which does not accommodate uncertainty in the right way. The same principle is then seen to distort also the received interpretations of CE.
Similar content being viewed by others
Notes
Stanev (2015) shows that the current practice in adopting early stopping rules may be due to mistaken interpretations of the relevant statistical results that have their roots in uncritical acceptance of error statistics underlying the decisions to implement these rules.
In the light of this issue, IE might be better expressed using alternative methods, such as confidence intervals or non-additive probabilities, rather than standard probability measures.
Freedman (1987b) clarified that beyond scientific validity, an important requirement for medical ethics is the clinical value of a hypothesis. It is specifically on the clinical value that there may be disagreement in the community of experts as indicated by CE.
For a classical analysis of bounded rationality in which the rationality of decisions is shaped by the limitations of time, information and other contextual factors see, for instance, Tversky and Kahneman (1981). Their view, which is “heuristics as errors” in reasoning, takes optimal and analytic procedures to be supplanted in human economizing decisions by suboptimal but fast procedures of reasoning that achieve practical goals with less expenditure on time, money and cognitive effort. For a critical analysis of this paradigm, see (Gigerenzer et al. 1991), among others.
The confusion of “conditions of risk” with “conditions of severe uncertainty” is called the “Tuxedo Fallacy” (see e.g. Hansson 2009).
Even such continued and completed trials are unlikely to achieve the desired sufficiency of evidence, or evidence that is stronger than the initial evidence from early phases of a trial. Such a phenomenon adds to the reasons why systematic reviews and meta-analyses are also required in order for a research to achieve the evidential status concerning it effect sizes and the respective clinical conditions, hospital practices and proposed treatments. The well-known demand for meta-analyses is due to the problems of triangulations of different sources of evidence and to the persistent lack of replicability of findings on specific clinical questions. This, in turn, is a symptom of underlying biases involved at all stages of experimentation and publication of results (Ioannidis 2005). Our paper could be read as adding to this observation the further consequence that even the multi-site, fixed-effect meta-analyses may not solve the problem of the reliability of evidence, whenever the only mode of variance between the trials is within-studies. We thank the reviewer for raising this important point. What the precise implications of these epistemological considerations are to the validity of random-effect meta-analyses we leave for further investigation.
This follows from such designs being biased towards a likelihood that the effect sizes of those trials between competing research teams working on the same problem are exaggerated by those who are the first to report their positive findings. This is an instant of the bias known as the Winner’s Curse.
The logical model for such hypothetical retrospections is a branching-time model.
Exitus may or may not be ruled out, depending on the patients’ desires, wishes, wills and testimonies, severity of conditions and the gravity of suffering, among others.
Common examples of the failure of UT can be drawn from game theory. Let us mention two: (1) games that use the “burning bridges” strategies: they involve actions that intentionally limit the range of one’s own options in order to signal a credible commitment to the opponent and in which way to avoid a conflict, and (2) games in which the epistemic states of the agents are also taken into account in the decisions (see, Chiffi and Pietarinen 2017).
Typically those in which it is unlikely that some unforeseen clinical conditions may crop up, such as those aimed at corroboration of another large study that already has investigated the same clinical question. But even in these cases the effect sizes may vary as the patient populations, cultures and clinical practices are far from identical.
For a discussion of another essential problem of RCTs, i.e. their external validity, see (Cartwright 2007).
There is also the important wider question is whether even the random-effect meta-analyses can be free from appeals to the PoI under residual fundamental uncertainty.
There are some wider theoretical issues lurking at the back of these arguments. We mention two. First, any uncritical or unrecognized presupposition of the PoI and the related confounding of risk-laden situations with fundamental uncertainty is likely to have a negative impact on what the positive predictive values of large experimental studies are. This adds grist to Ionnadis’s mill (Ioannidis 2005). Second, the practice of current clinical research is mainly frequentist. Would a Bayesian perspective on the design of RCTs be more suitable in handling (some variants of) CE, so much so that it could turn it into an epistemic, quantifiable component obtained by prior elicitation of a Bayesian assessment of evidence? For instance, the notion of “admissibility of treatments” in a Bayesian trial may determine the restriction of possible treatments to a certain subgroup of subjects in order to ensure that a specific patient is not administered an inferior treatment (as judged by a committee of experts) merely for the reason that it would facilitate the completion of a trial (see Sedransk 1996). On the other hand, there are concerns on the epistemological and ethical validity and limits to such Bayesian approaches (see e.g. Teira 2010).
On the analysis of the role of uncertainty and future scenarios for clinical reasoning, see (Chiffi and Zanotti 2017b).
References
Annoni M, Miller FG (2014) Placebos in clinical practice: an ethical overview. Douleur Analg 27(4):215–220
Bluhm R (2010) The epistemology and ethics of chronic disease research: further lessons from ECMO. Theor Med Bioeth 31(2):107–122
Broderick JP (2013) Devices and clinical trials. Overview and equipoise. Stroke 44[Suppl 1]:S3–S6
Carnap R (1955) Statistical and inductive probability. In: Carnap R (ed) Statistical and inductive probability. inductive logic and science. The Galois Institute of Mathematics and Art, Brooklyn
Carnap R (1966) Introduction to the philosophy of science. Dover, New York
Cartwright N (2007) Are RCTs the gold standard? BioSocieties 2(1):11–20
Chiffi D, Pietarinen A-V (2017) Fundamental uncertainty and values. Philosophia 45(3):1027–1037
Chiffi D, Zanotti R (2017a) Knowledge and belief in placebo effect. J Med Philos 42(1):70–85
Chiffi D, Zanotti R (2017b) Fear of knowledge: clinical hypotheses in diagnostic and prognostic reasoning. J Eval Clin Pract 23:928–934
Diamond AD, Geffen D (2014) Randomized trials, observational registries, and the foundations of evidence-based medicine. Am J Cardiol 113:1436–1441
Djulbegovic B (2007) Articulating and responding to uncertainties in clinical research. J Med Philos 32(2):79–98
Emanuel EJ, Wendler D, Grady C (2000) What makes clinical research ethical? JAMA 283(20):2701–2711
Freedman B (1987a) Equipoise and the ethics of clinical research. N Engl J Med 317(3):141–145
Freedman B (1987b) Value and validity as ethical requirements for research: a proposed explication. IRB 9(6):7–10
Fried G (1974) Medical experimentation: personal integrity and social policy. American Elsevier Publishing Co., Inc, New York
Gifford F (2007) So-called “clinical equipoise” and the argument from design. J Med Philos 32(2):135–150
Gigerenzer G, Hoffrage U, Kleinbolting H (1991) Probabilistic mental models: a Brunswikian theory of confidence. Psychol Rev 98:506–528
Hansson SO (2007) Hypothetical retrospection. Ethical Theory Moral Pract 10(2):145–157
Hansson SO (2009) From the casino to the jungle. Synthese 168(3):423–432
Hansson SO (2013) The ethics of risk: ethical analysis in an uncertain world. Palgrave Macmillan, New York
Howick JH (2011) The philosophy of evidence-based medicine. Wiley, London
Ioannidis JP (2005) Why most published research findings are false. PLoS Med 2(8):e124
Karlawish JHT, Lantos J (1997) Community equipoise and the architecture of clinical research. Camb Q Healthc Ethics 6:385–396
Keynes JM (1921/1963). A treatise on probability. Macmillan, London
Knight FH (1921) Risk, uncertainty, and profit. Schaffner & Marx; Houghton Mifflin Company, Boston
London AJ (2006) Reasonable risks in clinical research: a critique and a proposal for the integrative approach. Stat Med 25:2869–2885
London AJ (2007) Clinical equipoise: foundational requirement or fundamental error? In: Steinbock B (ed) The Oxford handbook of bioethics. Oxford University Press, Oxford, pp 571–596
Mill JS (1874) A system of logic. Harper & Brothers, New York
Miller FG (2011) Equipoise and the dilemma of randomized clinical trials. N Eng J Med 346(5):476–480
Miller FG (2012) Is anything lost if we give up clinical equipoise? Clin Trials 9:632–633
Miller FG, Brody H (2007) Clinical equipoise and the incoherence of research ethics. J Med Philos 32(2):151–165
Miller PB, Weijer C (2003) Rehabilitating equipoise. Kennedy Inst Ethics J 13(2):93–118
Miller PB, Weijer C (2006) Fiduciary obligation in clinical research. J Law Med Ethics 34(2):424–440
Sackett DL (2000) Why randomized controlled trials fail but needn’t: 1. Failure to gain “coal-face” commitment and to use the uncertainty principle. CMAJ 162(9):1311–1314
Sedransk N (1996) Admissibility of treatments. In: Kadane JB (ed) Bayesian methods and ethics in a clinical trial design. Wiley, New York, pp 65–113
Smithson M (2009) How many alternatives? Partitions pose problems for predictions and diagnoses. Soc Epistemol 23(3–4):347–360
Spector R, Vesell E (2006) Pharmacology and statistics: recommendations to strengthen a productive partnership. Pharmacology 78:113–122
Stanev R (2015) Early stopping of RCTs: two potential issues for error statistics. Synthese 192:1089–1116
Teira D (2010) Frequentist versus Bayesian clinical trials. In: Gifford D (ed) Philosophy of medicine. Elsevier, Amsterdam, pp 255–297
Tversky A, Kahneman D (1981) The framing of decisions and the psychology of choice. Science 211:453–458
van Fraassen B (1989) Laws and symmetry. Oxford University Press, Oxford
Veach RM (2007) The irrelevance of equipoise. J Med Philos 32:167–183
Worrall J (2008) Evidence and ethics in medicine. Perspect Biol Med 51(3):418–431
Yazici Y (2008) Some concerns about adverse event reporting in randomized clinical trials. Bull NYU Hosp Jt Dis 66(2):143–145
Acknowledgements
Supported by the Estonian Research Council (“Abduction in the Age of Fundamental Uncertainty”, PUT 1305). Earlier version of this paper has been presented at Biom Conference in Pistoia, Le humanitas in ambito biomedico. We thank Cristina Amoretti, Mattia Andreoletti, Marco Annoni, Paola Berchialla, Federico Boem, Giovanni Boniolo, Raffaella Campaner, Pierdaniele Giaretta, Alessandro Pagnini and Federica Russo for their suggestions. We thank the two referees for their insightful and substantial remarks on the earlier version of the present paper.
Funding
Estonian Research Council, Research Grant, Abduction in the age of fundamental uncertainty (PUT 1305).
Author information
Authors and Affiliations
Corresponding author
Rights and permissions
About this article
Cite this article
Chiffi, D., Pietarinen, AV. Clinical Equipoise and Moral Leeway: An Epistemological Stance. Topoi 38, 447–456 (2019). https://doi.org/10.1007/s11245-017-9529-x
Published:
Issue Date:
DOI: https://doi.org/10.1007/s11245-017-9529-x