Intended for healthcare professionals

Letters

Uncertainty about clinical equipoise

BMJ 2001; 322 doi: https://doi.org/10.1136/bmj.322.7289.795 (Published 31 March 2001) Cite this as: BMJ 2001;322:795

Clinical equipoise and the uncertainty principles both require further scrutiny

  1. Fred Gifford (gifford{at}msu.edu), professor of philosophy
  1. Michigan State University, East Lansing, MI 48824, USA
  2. H Lee Moffitt Cancer Centre and Research Institute at the University of South Florida, Division of Blood and Bone Marrow Transplant, Tampa, FL 33612, USA
  3. Trout Research and Education Centre at Irish Lake, RR 1, Markdale, Ontario, Canada N0C 1H0

    EDITOR—The exchange between Weijer et al and Enkin addresses the question of under what circumstances and for what reasons entering patients in clinical trials can be morally justified.1 It is important to see, however, that the issues are a good deal more complicated. There are problems on both sides, but I will focus on clinical equipoise.

    This concept inadvertently conflates two distinct concepts, and neither one provides a convincing resolution of the moral dilemma posed by clinical trials. 2 3 Most of the essay by Weijer et al focuses on just one of these, which should really be termed “community equipoise” (the situation where not all within the community of “experts” have come to agreement that one treatment is superior to another). Enkin raises one problem with this criterion: that it fails to take seriously the clinical and moral judgment of the individual physician. But a closer look at community equipoise shows in addition that, once we understand that a policy decision (to stop the trial, announce the results, approve the drug, etc) requires a greater amount of evidence than does an individual decision to choose what is best for one's present patient, then community equipoise will typically be disturbed long before we obtain the predetermined level of statistical significance required to support the policy decision.

    The concluding suggestions made by Weijer et al, concerning their preferred criterion embodying a pragmatic approach, involve instead a distinct contrast—clinical (as opposed to theoretical) equipoise. Thus these comments will not help make the case for community over individual equipoise. For it is one thing to distinguish two kinds of questions, a theoretical question about whether a drug has a causal effect on the incidence of a certain simple, well-defined outcome, and a practical or clinical question about whether that drug is a better treatment overall for a certain set of patients than is another drug. But it is a different matter to distinguish two modes of assessment of either one of these questions: “What do I think concerning whether there is evidence for the claim?” or, “Is there community agreement concerning this?” For there to be hope of attaining community agreement on these matters, both clinical equipoise and the uncertainty principle will require further scrutiny.

    References

    1. 1.
    2. 2.
    3. 3.

    Equipoise and uncertainty principle are not mutually exclusive

    1. R J Lilford, professor of research of public health and epidemiology,
    2. Djulbegovic Benjamin (djulbebm{at}moffitt.usf.edu), associate professor of oncology and medicine
    1. Michigan State University, East Lansing, MI 48824, USA
    2. H Lee Moffitt Cancer Centre and Research Institute at the University of South Florida, Division of Blood and Bone Marrow Transplant, Tampa, FL 33612, USA
    3. Trout Research and Education Centre at Irish Lake, RR 1, Markdale, Ontario, Canada N0C 1H0

      EDITOR—The debate about the usage of equipoise versus the uncertainty principle as an entry criterion for a randomised trial is misplaced.1 These are not two mutually exclusive concepts, and equipoise simply represents the point (or distribution) of maximum uncertainty. 2 3 In decision analysis, this is the situation where a patient is indifferent between treatment options. 3 4 The question would be better phrased as, “How much uncertainty can we accept before entering a patient into a trial and by whom (patients, physicians, and community)?” Intertwined with this question is the question of a relation between not knowing, uncertainty, and equipoise, previously discussed by one of us.4 This relation was also noted by Bradford Hill, who in 1963 wrote that we should accept randomisation “only in our state of ignorance, the treatment given [being] a matter of indifference.”5 It is surprising to witness that little empirical work has been done to resolve issues that were put forward before the clinical community almost 40 years ago.

      References

      1. 1.
      2. 2.
      3. 3.
      4. 4.
      5. 5.

      There is another exchange on equipoise and uncertainty

      1. David L Sackett (sackett{at}bmts.com), director
      1. Michigan State University, East Lansing, MI 48824, USA
      2. H Lee Moffitt Cancer Centre and Research Institute at the University of South Florida, Division of Blood and Bone Marrow Transplant, Tampa, FL 33612, USA
      3. Trout Research and Education Centre at Irish Lake, RR 1, Markdale, Ontario, Canada N0C 1H0

        EDITOR—With reference to the article by Weijer et al,1 there is another Canadian exchange on equipoise, between Shapiro, Glass, and myself. 2 3 My response included a passage that might be relevant here. If a term is to do more good than harm in human affairs, it must pass at least the following three tests. First, consistency: it must mean roughly the same thing to everybody who uses it. Second, reality: it must describe something that is real. Third, utility: it must be frequently used to aid and justify decisions.

        The term equipoise fails all three tests.

        Consistency—Published definitions of equipoise vary, and new, often conflicting, ones are still being generated that defeat attempts to distinguish any theoretical versus clinical distinction. Some users define it as a perfect balance of evidence and would take odds of 1:1 on a bet, only to be contradicted by others to whom it means that the data suggest but do not prove efficacy and safety. Some permit its ownership by individual clinicians and patients, but a letter in this issue insists that equipoise, unlike uncertainty, can never be possessed by individual trialists. Shapiro and Glass define equipoise as uncertainty that rests with the expert clinical community as a whole. By using my transparent, old fashioned term (uncertainty) to define their opaque, new one (equipoise) they render things wonderfully clear, but leave me wondering why on earth they cling to such an arcane, confusing word. None the less, we seem to be in agreement that, at the community level, uncertainty over the efficacy and safety of a treatment provides a proper basis for conducting a randomised controlled trial.

        Reality—A recent report to the health technology assessment programme of the British NHS has summarised it best. There is some ingenuity in the equipoise theory, although its constraints seem bizarre if one tries to apply the theory in practice.

        Utility—The term equipoise just has not been found useful at the coalface. A search I conducted last October identified only 52 hits for equipoise (a text word that maps to no MeSH terms or trees at all), and none of them came from the reports of trials. A similar search yielded 292 860 hits for uncertainty, and it was commonly used in primary reports of actual randomised controlled trials as a justification for their execution. Moreover, uncertainty maps to the MeSH tree of probability, the first branch of which is Bayes's theorem (a formula for reassessing uncertainty in the face of new evidence).

        References

        1. 1.
        2. 2.
        3. 3.