Question Four—What Are the Probabilities of Success of Various Treatment Options?
In the above cases, judgments about diagnosis and treatment reflect a certain level of certainty or uncertainty. Given the nature of medical science and the particularities of each patient, clinical judgment is never absolutely certain. Clinical medicine was described by Dr. William Osler as “a science of uncertainty and an art of probability.” The central task of clinicians is to reduce uncertainty to the extent possible by using clinical data, medical science, and reasoning to reach a diagnosis and propose a plan of care. The process by which a clinician attempts to make consistently good decisions in the face of uncertainty is called clinical judgment.
The inevitable uncertainty of clinical judgment can be reduced by the methods of evidence-based medicine, using data from controlled clinical trials, and by the development of practice guidelines, which assist the physician's reasoning through a clinical problem. Although evidence-based medicine and practice guidelines aim to reduce the “uncertainty” and the “probability” of which Osler spoke, some degree of uncertainty always remains, because these methods reach general statistical conclusions that may not fit the real patient who is before the physician.
In addition to uncertainty about data and their interpretation, there will be uncertainty about what action to take in any particular case. This is reflected in such questions as “Now that we have medical evidence about what is possible, what should we do?” “Given all the possibilities, what goals are appropriate for this patient?” These questions cannot be solely answered by clinical data. The ethical principles of beneficence and nonmaleficence reduce the scope of this sort of uncertainty by directing intention and effort away from the wide range of possible diagnoses and treatments and toward the more narrow range most likely to help this patient in these circumstances. However, the ethical principles do not dictate particular clinical decisions. These decisions must be confronted in candid, realistic discussions among clinicians, the patient, and the family. This is the shared decision making that constitutes an appropriate professional relationship; see Chapter Two, “Patient Preferences.”
Feinstein AR.Clinical Judgment. New York, NY: Krieger; 1974.
Goodman KW.Ethics and Evidence-Based Medicine. Fallibility and Responsibility in Clinical Science. Cambridge, MA: Cambridge University Press; 2003.
An important ethical problem is closely associated with the probabilistic nature of medical judgment. The question is whether a high probability that a particular treatment will be unsuccessful justifies withholding or withdrawing that treatment. This is often called the futility problem, or “medically ineffective or nonbeneficial treatment.” A long, hotly contested debate over “futility” has been inconclusive. One definition at the center of the debate states: “futility designates an effort to provide a benefit to a patient, which reason and experience suggest is highly likely to fail and whose rare exceptions cannot be systematically produced.” In The Dying Patient, we have seen the term “physiologic futility,” that is, an utter impossibility that the desired physiologic response can be affected by any intervention. However, futility more properly is a judgment about probabilities, and its accuracy depends on empirical data drawn from clinical studies and from clinical experience. Because clinical studies that demonstrate this sort of futility are rare, and because clinical experience is so varied, clinicians make widely different estimates of futility: physicians' judgments that various procedures should be called futile range from 0% to 50% chance of success, clustering about 10%. Some ethicists and clinicians deny the utility of the concept of futility because of its confused meaning and frequently inappropriate application. Others, including ourselves, consider it a useful term when applied thoughtfully to treatment decisions about interventions with low likelihood of success.
Beauchamp TL, Childress JF. Conditions for overriding the prima facie obligation to treat. In: Beauchamp L, Childress JF, eds. Principles of Biomedical Ethics. 6th ed. New York, NY: Oxford University Press; 2009:167–169.
Lo B. Futile interventions. In: Lo B, ed. Resolving Ethical Dilemmas. A Guide for Physicians. 4th ed. Philadelphia, PA: Lippincott Williams & Wilkins; 2009:61–66.
Schneiderman LJ, Jecker NS, Jonsen AR. Medical futility: its meaning, and ethical implications. Ann Intern Med
Three main questions about futility are debated: (1) What level of statistical or experiential evidence is required to support a judgment of futility? (2) Who decides whether an intervention is futile, physicians or patients? (3) What process should be used to resolve disagreements between patients (or their surrogates) and the medical team about whether a particular treatment is futile?
(1) Statistical probability. Clinical futility requires a probabilistic judgment that an intervention is highly unlikely to produce the desired result. This judgment comes from general clinical experience and from clinical studies that demonstrate low rates of success for particular interventions, such as CPR for certain types of patients, or continued ventilatory support for patients with adult respiratory disease syndrome. Even the data that are available may prove deceptive in a particular case because studies apply to groups rather than individuals. Further, a lack of agreement exists about how low a level of probability would justify calling a treatment futile. One group has suggested that if soundly designed clinical studies reveal less than a 1% chance of success, intervention should be considered futile.
A study of 865 patients who required mechanical ventilation after bone marrow transplantation showed no survivors among the 383 patients who had lung injury or hepatic or renal failure and who required more than 4 hours of ventilator support. This study suggests that it would be probabilistically futile to intubate patients with these conditions or to continue ventilation after 4 hours.
Rubenfeld GD, Crawford SW. Withdrawing life support for medically ventilated recipients of bone marrow transplantation: a case for evidence-based qualitative guidelines. Ann Intern Med.
A large clinical study examined hospital discharge records of more than 5000 patients from eight U.S. cities, who suffered cardiac arrest out of hospital, were resuscitated by emergency teams, and were transported to hospital for further care. The investigators applied rules they had developed earlier for stopping CPR in the field and then tried to predict which of the resuscitated patients would survive to be discharged from the hospital. Their study was designed to validate the rules for predicting CPR futility. None of the 1192 patients who did not meet Advanced Life Support criteria for termination of CPR survived to discharge; of 776 patients who met Basic Life Support criteria, 4 (0.5%) survived to discharge.
Sasson C, Hegg AJ, Macy M, et al. Prehospital termination of resuscitation in cases of refractory out-of-hospital cardiac arrest. JAMA. 2008;12:1432–1438.
The first study was done in 1996. It clearly illustrates probabilistic futility: not a single patient from a large cohort left the ICU alive. A decade later, these data have not been contradicted. The second study was a retrospective cohort study developed to predict when it would be futile to continue resuscitation in cases of refractory out-of-hospital cardiac arrest. Applying these rules to the data accurately predicted 99.9% of the patients who did not survive to hospital discharge. Therefore, these rules predicted probabilistic futility in out-of-hospital cardiac arrests with great accuracy.
(2) Who decides? It is relatively rare that carefully designed clinical studies such as the previous reports provide hard data for determination of futility. Inevitable debates will ensue about the level of probability that should represent futility. Who has the authority to establish the goals of the intervention and to decide the level of probability for attaining such goals? Some ethicists argue that physicians have the right to refuse care that they believe is highly unlikely to produce beneficial results; other ethicists maintain that futility must be defined in light of the subjective views, values, and goals of patients and their surrogates.
A 75-year-old woman is brought to the ER by paramedics after suffering massive head trauma, with extrusion of brain tissue, as a result of a vehicular accident. She had been intubated by the paramedics. After careful evaluation, the ER physicians judged that her injuries were so severe that no intervention could retard her imminent death. When her grieving family gather in the ER, they demand that the woman be admitted to the ICU and be prepared for operation by a neurosurgeon. The physicians state that further treatment is futile.
Helga Wanglie was an elderly Minnesota woman who suffered irreversible brain damage from strokes and slipped into a chronic vegetative state. She required mechanical ventilation. Physicians and family agreed that she had no hope of regaining the ability to interact with others. However, Mrs. Wanglie's husband refused to authorize discontinuing the ventilator, saying that his goal (and, he asserted, hers) was that her life should not be shortened, regardless of her prospects for neurologic recovery. Physicians requested court intervention to authorize withdrawal of ventilatory support.
A 72-year-old man with late-stage emphysema is admitted to the ICU with fever, respiratory failure, and hypoxemia. While he is being intubated, he has a cardiac arrest. He is resuscitated in the unit, but remains unconscious after resuscitation. He is found to have had a large anterior wall myocardial infarction, requiring pressors to maintain blood pressure. The laboratory calls to say that blood culture data drawn in the ER are growing gram-negative bacteria. Because of his multisystem organ failure and sepsis, the physicians decide to write a DNR order, believing that a second attempt at CPR would be futile.
In Case I, the physicians are speaking of futility in the sense used in The Dying Patient, that is, physiological futility. The issue here is not the likelihood but the impossibility of continued life regardless of any intervention. They are ethically justified in refusing to pursue treatment. In Case II, continued ventilatory support and other interventions can extend Mrs. Wanglie's life. These interventions, employed for this purpose, cannot be judged physiologically futile. However, physicians judge that there is a vanishingly low probability of restoring Mrs. Wanglie's health and a low probability also that her life will be extended very long, even with support. They also judge that Mrs. Wanglie's life, if extended, will be of very low quality. Physicians may recommend termination of the intervention on the grounds of medical futility, but they lack the ethical authority to define the benefit of continued life even without consciousness. This is a matter for the patient and her surrogate to decide (as the Minnesota court determined). Some contextual features, such as scarcity of resources, might be relevant to this case (see Chapter Four, “Contextual Features,” Allocation of Scarce Health Resources).
In Case III, the patient's multiorgan system failure, dependence on pressors, and sepsis make it highly unlikely that a second resuscitation will succeed. A DNR order should be recommended to appropriate surrogates.
(3) Dispute Resolution. What process should be used to resolve disputes about futility? Institutions should design a policy for conflict resolution. These policies should prohibit unilateral decision making by physicians, except in cases of physiological futility. For judgments of futility based on low probability of successful treatment, policy should stress the need for valid empirical evidence, provide for consultation with outside experts and with ethics committees, and, above all, create an atmosphere of open negotiation or mediation rather than confrontation. The policy should allow physicians to withdraw from cases in which they judge continued treatment futile and should provide for transfer of patients to other institutions willing to accept them. Futility arguments should be moved into court only after all other reasonable attempts to resolve the disagreement fail. Elements of a model hospital policy on nonbeneficial care can be found in the AMA Code of Medical Ethics 2008, 2.037 (www.ama-assn.org).
Despite continued debates about the concept of futility, we believe it is useful in medical ethics, because it highlights the necessity to make decisions about treatments that are of questionable benefit. It introduces a note of realism into excessive medical optimism by inviting physicians and families to focus on what realistically can be done for the patient under the circumstances and which goals, if any, can be realized. It provides the opportunity to open an honest discussion with patients and their families about appropriate care. It calls for a careful investigation of the literature about the efficacy of proposed treatments in particular situations.
Physicians should never invoke futility, except in the sense of physiologic futility, to justify unilateral decision making or to avoid a difficult conversation with patient or family. A physician's judgment that further treatment would be futile does not justify a conclusion that treatment should cease; instead, it signals that discussions of the situation with patient and family are mandatory. Futility should never be invoked when the real problem is a frustration with a difficult case or a reflection of the physician's negative evaluation of the patient's future quality of life; see Chapter Three, “Quality of Life.” Also, a futility claim by itself does not justify rules or guidelines devised by third-party payers to avoid paying for care; see Chapter Four, “Contextual Features.” Further, even when the facts of the case support a judgment of futility, we suggest that it may be advisable to avoid the actual word “futility” in discussions with patients or their families. Many persons may interpret this word as an announcement that the physician is “giving up” on the patient or that the patient is not worth further attention. At this point, rather than explicitly using futility language, clinicians should raise the question of redirecting the efforts of clinical care to palliation and comfort, because the burdens of more aggressive care far exceed the chances for benefit. Ethicists sometimes refer to this reasoning as proportionality (see The Ethical Principle of Proportionate Treatment).
Finally, we acknowledge that a physician has the moral right to withdraw from a case in which he or she has reached an honest judgment of futility, even though continued care is demanded by others. Such a judgment would be based on the belief that nothing is being done to benefit the patient, while continued interventions actually are harming the patient. Should a physician reach this conclusion, proper steps to inform the family should be taken. Hospital policy should support physician's judgments in this regard.