By Mayli M.
Consider Chris, an unconscious coma patient in intensive care. Suppose that, according to tests of Chris’s brain activity, he is predicted to have a ‘poor outcome’, which could be death or a prolonged disorder of consciousness like vegetative or minimally conscious state. This prognosis informs treatment decisions about this patient, specifically the decision to withdraw life-sustaining treatment. When life-sustaining treatment is withdrawn based on the prediction of ‘poor outcome’, Chris dies as a result. This makes the prediction true regardless of whether continued treatment would have led to significant recovery. Thus, the prognosis is a self-fulfilling prophecy.
Self-fulfilling prophecies (SFPs) have long fascinated scholars and practitioners in many fields. In medicine, the placebo effect is one of the most commonly acknowledged SFPs and it has become the basis for standard quality assurance in clinical trials. Because the placebo effect results from optimistic predictions, the effect is positive, improving the patient’s health. In contrast, pessimistic predictions that self-fulfill have a negative effect. This is particularly worrisome in prognostic practices and, increasingly, in early-diagnostics when pessimistic SFPs ensure negative outcomes. Whether their effects are positive or negative, however, SFPs have received the kind of attention they have precisely because of their ability to change things, for better or worse. If the mere making of a prediction can make it so, when not making the prediction would not make it so, how can we be anything but fascinated?
Unfortunately, the focus on self-fulfilling prophecies that change the outcome—thus leaving aside those that do not—has led practitioners as well as scholars to misunderstand both the mechanism of SFPs and the normative and epistemic issues that SFPs really entail.
I don’t mean to suggest that the effect of SFPs that cause an outcome that would not otherwise have occurred isn’t distinctively fascinating. Clearly, if Chris died unnecessarily just because his prognosis was based on a false positive test result, that outcome is tragic in ways that Chris’ inevitable death would not be. It is thus vital to distinguish between transformative SFPs that do change the outcome and operative SFPs that do not. However, to properly evaluate the ethics of SFPs, we must include all self-fulfilling prophecies, even the seemingly less remarkable operative ones.
Not acknowledging operative SFPs as genuine SFPs is the same as not acknowledging that the genuine cause of a boxer’s death is the accidental lethal blow of his opponent, just because—counterfactually—the boxer would have collapsed from heart failure soon thereafter. We have to acknowledge them as real SFPs because, even if the outcome would have occurred for some independent reason, the outcome now occurs because of the prediction and the way the prediction was used—the same way a transformative outcome occurs. In practice, distinguishing between operative SFPs and transformative SFPs can therefore be extremely difficult, if not impossible. Once Chris dies after our actions based on the best available evidence caused his death, how do we really know if our evidence was correct?
Medical practice is marred by uncertainty and practitioners are often working with probabilities rather than certainties. Indeed, this uncertainty often drives prognostic research and innovation. In order to give the patient the best possible care, clinicians are often forced to employ state-of-the-art or even experimental evidence. Thus, it is not always easy to uphold the research-practice distinction. This need not be a problem in itself, however. Typically, retrospective studies can later test for the quality of the evidence used by matching the eventual outcomes with the original prognostic information and learn from the discrepancies between prediction and reality. Not so with self-fulfilling predictions, though, which are always true.
In short, once SFPs occur, even if none of them are transformative, we cannot check whether positive tests results were actually true or false test results. False positives keep seeping through our quality assurance tests because the feedback we get is always affirmative and because that affirmative feedback is always due to the way the prediction was used.
Prognostic innovation practices aim to improve our predictive capacities through learning. Yet SFPs inhibit such learning. That is why all self-fulfilling prophecies matter. Whether transformative or operative, self-fulfilling prophecies make it impossible to distinguish our successes from our failures.
Authors: Mayli Mertens [1,2] Owen C. King  Michel J.A.M. van Putten [3,4] and Marianne Boenink [2,5]
 Center for Medical Science and Technology Studies, Department of Public Health, University of Copenhagen, Denmark
 Department of Philosophy, University of Twente, the Netherlands
 MIRA-Institute for Biomedical Technology and Technical Medicine, University of Twente, the Netherlands
 Department of Neurophysiology, Medisch Spectrum Twente, the Netherlands
 Department IQ Helathcare, RadboudUMC/Radboud University, the Netherlands
Competing interests: None declared
Social media accounts of post author(s): Mayli M’s Linkedin