The truth about proof.

Abstract

I am bothered by proof. The zeal for evidence that so characterizes contemporary medicine leaves me ambivalent and unsettled. On the one hand, outcomes researchers can corral companies motivated by profit to demonstrate product effectiveness before expensive yet marginally helpful therapies are loosed upon the public. On the other hand, evidence-based academicians can focus scientific attention on some aspects of medical practice that seem more like value than experimental therapy. Does new drug A work better than old workhorse B? This question seems reasonable. Does compassion in patient care offer measurable results? This question seems more troubling. Consider the following scenario: a woman with unremitting breast cancer reviews treatment options with her oncologist: “Bone marrow transplantation isn't effective,” the doctor says. “Neither is additional surgery, chemotherapy, or radiation.” “What else is left?” asks the woman. “Well,” the oncologist replies,“There is psychosocial treatment. it's unproven, but I've seen it work.” Lest you think this exchange absurd, let me allay your annoyance: research supports psychosocial “treatment” of women with metastatic breast cancer. The seminal study, published more than 10 years ago,1 shows that women who received weekly support therapy lived longer than those who did not.Relieved? I'm not. The study authors predicted that psychosocial support, although potentially beneficial for psychological wellbeing, would have no effect on the course of their patients' disease. They sought to debunk the business of support, rather than demonstrate its value. The improved survival data took them by surprise. Maybe you find this divergence between expectation and outcome unproblematic. Science, you might contend, is nothing but a challenge to expectation, an objective means of assessing outcome when subjective bias and experience misguide instead of inform decision making. We need to know what we are doing, you might argue. We need to know where we are being wasteful. Suppose, however, that the data had showed something different. Suppose the study of women with breast cancer had showed that psychosocial treatment had no effect on survival or that it had shortened survival instead of extending it. Would we feel comfortable calling psychosocial support wasteful? Would we direct physicians, nurses, and oncology practices not to offer it? We trade on patient data like bankers and brokers and publish research observations like uninvolved neutral parties i recently posed this hypothetical dilemma to a panel of physicians convened to review their ongoing studies of compassion and psychosocial treatment in medicine. Some of the researchers responded with decidedly unscientific and unsympathetic alacrity. “You have to listen to the data,” 1 doctor replied. “Sometimes it speaks to you in ways you aren't appreciating.” Another stressed the practical importance of documenting the role that compassion plays in healing. “Thirdparty payers will only fund treatments with proven efficacy,” he said. “Without the data, people won't get the services.” Listen to the data? Prove efficacy for third party payers? Such jargon sounds more like political gamesmanship than dispassionate scientific study.It subverts the art of healing to the callous rules of the marketplace. It makes me wonder whether uncertainty, not suffering, has become medicine's foremost therapeutic challenge, whether statistical analysis, not compassion, has become the preferred prescription for meaningful care. The truth about proof is that there is none. Quantum physics teaches that where we most search for certainty, we find only probability. What is observed depends on who observes and how. Physicists working in the whirling world of the subatomic particle have learned to accept such uncertainty; physicians working in the supercharged world of human frailty have not. Want to know both the speed and location of an electron? A physicist says you can't. Want to know both the cause and best treatment of a patient's illness? Doctors do it every day. So we deceive ourselves and justify our involvement in patient care through the statistical analysis of symptoms, treatments, and clinical outcomes.Instead of treasuring the participatory and often highly personal nature of healing, physicians reject individual perspective as a threat to the institutional status quo. We press subjectivity through the tired mill of the placebo-controlled trial and distract attention from inexplicable recovery and suffering with complicated methods and analysis. We trade on patient data like bankers and brokers and publish research observations like uninvolved neutral parties. We distance those in need from those most able to help. Proponents of evidence-based medicine can no doubt list an impressive array of useless or unsafe therapies discarded because of careful study and analysis. While I respect such achievements and contributions, I worry that increasing efforts to quantify the qualitative threaten, rather than strengthen, our basic commitment to sick people. Any day now, studies allegedly proving or disproving the effectiveness and efficiency of compassion in health care are due to be published. I, for one, won't feel comforted to see them in print.