This emergence of DNR orders was based on patient or surrogate consent. In fact, when the public realized that DNR orders were sometimes enacted secretly in New York State, the government responded with specific legislation to assure that consent was obtained [5]. Nonetheless, individual health care providers continued to limit what they regarded as inappropriate care by using slow codes and show codes [6]. In short, CPR held a unique position. Unlike all other therapies, one had to get consent to withhold CPR, not to administer it.

In 1991, the Council on Ethical and Judicial Affairs of the American Medical Association [7] wrote, "When efforts to resuscitate a patient are judged by the treating physician to be futile, even if previously requested by the patient, CPR may be withheld". Akin to the 1974 proposal by the American Medical Association to document DNR orders, this 1991 statement leads us to the next phase of the debate: "May physicians write unilateral DNR orders?" Unilateral DNR orders highlight the idea that the decision to offer CPR can be a medical decision [8]. Like all therapies, CPR is sometimes inappropriate or not indicated. Thus, policies permitting unilateral DNR orders can be considered as CPR-not-indicated policies, implying that the decision to offer CPR can be a medical decision.

Why are CPR-not-indicated policies developing now? Physicians may be frustrated by the glare of autonomy [8] compelling them to provide, in the physician's mind, inappropriate treatment [9]. Because the patient's autonomy may not allow the physician to give what he considers the best possible care, physicians may feel hindered in fulfilling their fiduciary relationship [10]. This requirement to give unsuccessful and inappropriate care can be demoralizing to the caregiver [11]. Tomlinson and Brody [12] have argued that the current primacy of the patient's wishes does not provide true autonomy. By instituting a therapy in a futile situation, one falsely offers hope to the family and patient that undermines the patient's ability for rational judgment and autonomy. Cardiopulmonary resuscitation-not-indicated policies may be a way to increase the patient's or surrogate's autonomy by limiting their choices to only those treatments that offer achievable goals [13]. Finally, the pressure to control costs is greater now than it has been in the past, and providers may be looking for ways to ration care; expensive and marginally effective care may be a place to start [14].

A CPR-not-indicated policy is predicated on the idea that in some situations CPR is inappropriate for the patient; that is, doing CPR is futile. The state of futility implies that an action has little or no likelihood of achieving a specified goal. Although different interpretations referring to futility and CPR have been proposed, none is entirely satisfactory; in fact, it is questionable whether futility, as it relates to CPR, can be defined at all [6, 15].

Despite these problems, some distinctions are helpful. The futility of an action cannot be discussed without a defined goal. For example, antibiotics are ineffective if they do not eradicate the infection; reevaluating the infection after the antibiotics is the standard by which success or failure is judged. Futility definitions also have quantitative and qualitative aspects [16]. Quantitative assessments place a percentage value on the likelihood of something occurring, whereas qualitative assessments define the worth of what may be achieved. In addition, the effect with which the percentage of quantitative futility can influence the performance of an action may depend on the qualitative goal. For example, most people consider playing a state-run lottery quantitatively futile because the odds of winning are low. However, a million-dollar lottery is not qualitatively futile because, to most people, a million dollars is a substantial amount of money. If the lottery prize were less valuable, for example, $100, people might consider buying a ticket qualitatively futile.

Similarly, if the prize money increases and the odds of winning stay the same, more people will play the lottery. Because the reward is greater than it was before, or more qualitatively substantial, playing the lottery is considered less quantitatively futile. Thus, the power of the quantitative assessment may be influenced by the definition and value of the qualitative assessment, or the odds with which one takes a chance depends on the reward. Similarly, the decision to do CPR is sometimes based not only on the quantitative likelihood of survival but also on whether the qualitative aspects of survival are considered worthwhile. Quantitative and qualitative futility are intertwined, and, occasionally, the determination of quantitative futility becomes a value judgment or a qualitative assessment of futility.

Risk perspective also colors futility judgment. Patients and physicians may each individually favor either a risk-taking optimistic approach or a risk-avoiding pessimistic approach. Consider two patients with the same chronic disease. An operation with an 80% complete cure rate and a 20% mortality rate is offered. Other considerations aside, the optimistic patient may have this operation because he or she likes the 4 in 5 success rate, whereas the pessimistic patient may refuse, influenced by the 1 in 5 mortality rate. An a priori view of whether a certain percentage of success represents a good or a bad chance, regardless of the specific outcome, is perhaps the most powerful component of the influence of the quantitative aspect [17].

Futility definitions, therefore, hinge on values; the question is: whose values? In the traditional autonomy paradigm, the patient's values are prized above all [18]. Cardiopulmonary resuscitation-not-indicated policies modify these values; depending on the view, the selected values can be either the physician's values or the community's values. Further, when a physician makes a decision for a patient, the physician can make a choice on the basis of his or her own values, that is, "What would I want in this situation?" or on the basis of his or her interpretation of the patient's values, or "What do I think the patient would want?" Both methods are necessarily contaminated by the physician's beliefs [19].

The following two sections provide specific examples of CPR-not-indicated policies. The first section is a review of policies from four hospitals. These policies were selected only because we were aware of them; in no way is this intended as a conclusive review. These policies do, however, cover many important issues; they can be evaluated in light of quantitative futility, qualitative futility, and value orientation. The second section reviews three distinct, previously published proposals that may act as templates for future hospital policies.

Hospital Policies

Allegheny General Hospital, Pittsburgh, Pennsylvania

"When the attending physician believes that life-sustaining treatment may be potentially "futile," that is physiologically unable to work, then it is not necessary to initiate this treatment. The physician does not need permission to forego such treatment. Other definitions of futility are ambiguous because they involve making value judgements about the quality of life. determinations of futility in this context (quality of life), made unilaterally by the physician, are not appropriate" [20].

Allegheny General Hospital endorses a policy that is based on the ability to achieve certain physiologic parameters. Although these specific parameters are not defined, the hospital's proposal is clarified by contrasting physiologic futility with other kinds of futility, such as a treatment with a high likelihood of failure or one incapable of resolving the patient's underlying condition. Furthermore, the policy notes that "very few life-sustaining treatments are going to meet this narrow definition of futility". Using a physiologic basis to define futility limits the need for qualitative and quantitative judgments. Allegheny General Hospital's policy is an excellent example of an attempt to appreciate the concerns of potentially futile care while acknowledging the difficulty of implementing a more sweeping proposal.

Veterans Affairs Medical Center, Seattle, Washington

"A medical judgment is made that attempted CPR would be futile when either of the following (are fulfilled): (a) Quantitative futility:. a very low or rare probability of achieving the return of vital organ function and survival beyond a short period of time or (b) Qualitative futility: CPR might be effective in sustaining life but the patient's quality of life falls well below the threshold considered minimal by general professional judgement. In making a judgement of futility, the physician relies on personal experience, medical literature, and on observations in the context of caring for the individual patient. When a patient (or surrogate decision-maker) disagrees with the physician's judgement of medical futility and would like to receive CPR, the physician should not write a DNAR (do not attempt resuscitation) order until resolution of the disagreement" [21-23].

This policy attempts to avoid the inherent problems of vague definitions by adopting a proposal by Schneiderman and colleagues [8]; they define a treatment as futile if "physicians conclude (either through personal experience, experiences shared with colleagues or consideration of reported empirical data) that in the last 100 cases, a medical treatment has been useless". This would mean that a clinician would be "95% confident that no more than 3 successes would occur in each 100 comparable trials" [8]. Useless is defined as death, permanent unconsciousness, or permanent dependence on intensive medical care. When a situation is judged to be futile, recovery of the patient would be "highly improbable and. cannot be systematically reproduced" [7].

The precision of quantitative futility is diminished in the Veteran Affairs policy (compared with the proposal of Schneiderman and colleagues [8]); however, "very low or rare" does have the admirable quality of admitting imperfection in the ability to predict outcomes. The process for defining futility is severely hampered by inconsistency, memory, and medical prejudices. For example, relying on subjective and anecdotal information allows physicians to unintentionally make decisions, perhaps through selective recollection, that may not be equitable [24, 25]. Even if an exact definition of quantitative futility were possible, empirical data may not provide the answers. A recent study [26] indicated the difficulty of defining which clinical variables predict survival after CPR: "Successful resuscitation was not associated with gender, comorbidities, age or interval from admission". Worthwhile comorbid conditions somewhat predictive of survival to hospital discharge in one hospital were not predictive in another hospital [26]. Further, defining what determines a cohort group for treatment analysis of the last 100 patients can be difficult. The qualitative definition of life—"below the threshold considered minimum by professional judgement"—and length of time—"short periods of time"—requires value assessment and is open to interpretation. This policy, too, requires the resolution of disagreements among patient, family, and caregiver before a DNR order can be written; it is therefore not a unilateral CPR-not-indicated policy.

Beth Israel Hospital, Boston, Massachusetts

"The attending physician may enter the ‘CPR Not Indicated’ order only in the following circumstances: (a) The patient is dying with no chance of recovery; or (b) There is no reasonable likelihood that CPR efforts would be successful in restoring cardiac and pulmonary function. and no therapeutic alternatives could reverse or significantly slow the downhill course of illness. and the proposed treatment would increase or prolong the patient's suffering and cannot be justified on medical grounds. The attending physician should discuss with the patient or family the medical basis for not attempting resuscitation. the discussion is not for the purposes of securing permission (emphasis added) but rather an opportunity to develop as full an understanding of the patient's situation ..."^[27].

Although the policy states "dying with no chance of recovery," Beth Israel Hospital uses Schneiderman and colleagues' [8] definition of quantitative futility; this is stated in a footnote. This policy also asks for "concurrence by other involved physicians". Although they do not define qualitative futility directly, they imply that when the physician believes that the patient's suffering outweighs the advantage of being alive, the treatment is qualitatively futile. This qualitative assessment is broad and certainly subject to personal or institutional biases. This policy places the question of "whose values" squarely in the physicians' corner. In this sense, Beth Israel Hospital has an unequivocal CPR-not-indicated policy, implying that the offering of CPR can, in some circumstances, be an absolute medical decision.

Johns Hopkins Hospital, Baltimore, Maryland

"Any course of treatment may be regarded as futile if it is highly unlikely to have a beneficial outcome, or if it is highly likely merely to preserve permanent unconsciousness or persistent vegetative state or require permanent hospitalization in an intensive care unit. Substantiation of (the reasons why requested options are futile) through recording in the chart literature references and the opinions of medical consultants is encouraged. It is the policy of The Johns Hopkins Hospital that attending physicians are not required to offer life-sustaining intervention, and may refuse a request for the same, if the intervention is medically futile and will not offer meaningful benefit to the patient. When conflicts arise. attempts at resolution should proceed. If the conflict remains irreconcilable, the attending physician or the patient may seek an alternative attending physician" [28].

This policy uses "highly unlikely" and "beneficial" to define quantitative and qualitative futility, but, more importantly, this policy defines quality of life as a function of location. This presupposes that any patient who is permanently hospitalized in the intensive care unit has a poor quality of life. Although it is unlikely that they intended to cut such an extensive swath, precision in policies is important. Until these concerns can be answered, qualitative definitions that hinge on location are questionable. Such a proposal gives a physician exceedingly broad rights.

The Johns Hopkins policy solves the value-orientation problem by allowing the patients or surrogates to seek out alternative physicians whose views may be similar to their own. This is consistent with the fact that patients and surrogates have always been free to go to another physician if they chose. By allowing the patient to control options by choosing a physician with similar values, this policy permits the patient's values to be honored without subjugating the physician's values.

Three Different Proposals

The following three proposals present discrete conceptual views of CPR-not-indicated policies. Physiologic futility [29] promotes a nearly value-free policy, whereas Schneiderman and colleagues' [8] proposal favors the physician's values. Murphy and Finucane [30] attempt to operationalize futility by championing society's preeminence in determining the appropriateness of a person's therapy.

Physiologic Futility

Physiologic futility asks whether a treatment can achieve its physiologic objective [29]. If a treatment cannot, then it is futile. For example, CPR is futile only if it is impossible to do cardiac massage and ventilation. As long as circulation and gas exchange are occurring, CPR is not futile, even if no one expects improvement in the patient's condition. If a treatment cannot achieve a physiologic objective, and, thus, no benefit is being offered to the patient, then a physician is not obligated to offer this treatment. Physiologic futility appears to have the least risk for unilaterally imposed physician value judgments [31]. Admittedly, problems may occur in determining precise definitions of physiologic functions (such as circulation and ventilation), but these are more technical in nature and do not involve substantial value judgments.

The adoption of physiologic futility cannot substantially change practices. If the decision to halt therapy must wait until the physiologic objective cannot be achieved, then the patient, for all practical purposes, will be dead nearly immediately after the declaration of physiologic futility. This strength of physiologic futility minimizes the chance of a physician error in quantitative assessment limiting the length of life. Thus, physiologic futility has the important symbolic value of recognizing the legitimate desire to limit futile care while implicitly acknowledging the inherent difficulties in designing a broader CPR-not-indicated system [31].

Schneiderman and Colleagues' Policy

The policy developed by Schneiderman and colleagues [8] promotes standardized but arguable criteria for determining qualitative futility, such as permanent dependence on intensive medical care. This policy is reviewed earlier in the paper. As noted, the quantitative aspect heavily depends on personal experience and knowledge. Most importantly, this type of policy accepts physician errors in determining quantitative futility by allowing a therapy to be declared futile if the physician is 95% confident that a therapy would be successful no more than 3 times out of 100. The imprecision of the certainty of futility in this proposal is disturbing. Presumably, if we are willing to accept that a percentage of patients can have therapy inappropriately withheld in the name of futility, then we accept the idea that, at some percentage, the benefit of appropriately limiting therapy outweighs the maleficence of inappropriately limiting therapy. We are not comfortable defining such an Equation forCPR-not-indicated policies at this time.

Murphy and Finucane: Operationalizing Futility

Murphy and Finucane [30] reference the definition of futility to society's values and empirical data. Futility is defined as "treatment that is so unlikely to succeed that many people—professional and lay persons—would consider it not worth the cost" [30]. They argue that this operationalizes the concept of futility and precludes individual caregivers from having to make qualitative or quantitative value judgments. Table 1 lists characteristics of several groups of patients who have been shown to have less than a 3% chance of survival to discharge after CPR. For these patients, Murphy and Finucane [30] claim "CPR should not be considered standard medical care ... . Physicians would not offer this treatment to patients and patients would not have the right to demand it". Local communities would be able to determine for which disease processes therapy would be considered futile through "a consortium of representatives ( ... medicine, nursing, social work, administration, and chaplaincy) (that) would work with ... professional groups, lay groups and government agencies to propose new DNR policies. Hospital administrators, legal councilors and ethics committees would decide whether to adopt the DNR policy" [30]. Murphy and Finucane's radical proposal has unusual implications and is worth further examination.

As indicated by the title of their article [30] "New Do-Not-Resuscitate Policies: A First Step in Cost Control," "The major rationale for this policy change is cost control". If this is true, then perhaps this should be called a rationing policy rather than a futility policy. Truly futile care should never be offered no matter what the cost. When care is limited for economic reasons, the justification for the policy is the need for rationing. Referring to this type of policy as a "futility" policy may make it more politically palatable, but this is done at the cost of misrepresentation.

Murphy and Finucane [30] suggest two reasons why end-of-life care should be rationed. One argument is that recognizing and acting on marginally effective care is a symbolic action necessary to "change ... society's expectations". The second argument is that rationing end-of-life care can result in substantial cost savings.

With regard to their first argument, the public substantially mistrusts physician and government institutions (which is what the community consensus panel would represent): "There's a slimy image out there ... . People don't feel they can trust ... medical associations" [32]. This mistrust suggests that a more appropriate action might be to garner society's trust with less controversial limitations of treatment, and then, if economically necessary, tackle the more volatile issues.

As for their second argument, whether substantial cost savings are available by modifying end-of-life care is questionable [33, 34]. If a consensus group can develop a proposal with which most of the community agrees, the policy is not really needed because, presumably, most people will already be making decisions in accordance with the policy. The ability to educate patients about end-of-life statistics should not be considered a hindrance. Patients can easily understand information about the probability of survival to discharge after CPR, and such information does influence patients' desires to receive treatment [35]. Murphy and Finucane note that patients' wishes expressed though advance directives are poorly followed and, thus, rarely effective. The response to this observation, however, should be renewed attempts to increase communication, not the development of rule-based policies.

With this policy, Murphy and Finucane [30] contend that physician-patient discussions can be directed toward more meaningful issues (such as feedings, surgery, and antibiotics) because physicians will not feel "that they have adequately communicated with the patient and family once they have established code status". The opposite seems more likely. Physicians comfortable with discussing end-of-life issues pursue them with their patients and families. On the other hand, those uncomfortable with discussing end-of-life issues avoid the subject entirely. Without this policy, physicians at least have to broach the issue of DNR status, which may lead to discussion of these other issues at the patient's or surrogate's behest. With this policy, a physician can enact the DNR order without discussion or consensus, and the physician can, through accident or intention, foreclose the opportunity to discuss other issues. This is of particular concern if the patient and family do not have a regular and trusted physician.

Conclusions

Cardiopulmonary resuscitation-not-indicated policies may be an appropriate way to limit what some perceive as an excessive glare of autonomy. A desirable CPR-not-indicated policy, however, must include clear and definitive thresholds for quantitative and qualitative determinations of futility. This ideal policy would engage broad community support, and the rules would be consistent across socioeconomic strata. Economic factors should be considered only if 1) this is clearly stated, 2) the policy actually saves money, and 3) a reasonable guarantee exists that the money saved with this policy benefits other medical causes [36]. Money can be saved in many places without cutting potentially life-saving therapy; the benefit of starting with the most controversial areas for cost savings does not seem to outweigh the disadvantages.

The attempts of the reviewed proposals to define these issues and advance them into practice and public debate should be applauded. These concerns are noted in the hope that the problems can be resolved through thought and public negotiation. At this time, however, we believe that policies based on anything other than physiologic futility are indefensible because of imposed value judgments, imprecise definitions of quantitative and qualitative futility, inexact data, lack of certitude of economic benefit, and the role of autonomy for the patient and physician. Allegheny General Hospital's policy comes the closest to meeting our objectives and could provide a model for other hospitals to emulate. Although physiologic futility does little to advance current practice management, it is a symbolic statement of our commitment to acknowledge that some care is indeed futile, without endangering either patients' and physicians' values.