Prolonging Life: Some Legal Considerations

George P. Fletcher I

Much of what follows is an exercise in con­ceptual analysis. It is an effort to devise a test for determining which of two competitive schemes-that for acts or that for omissions ­should apply in analyzing a given question of responsibility for the death of another. It is significant inquiry, if only to add a word to the discussion of the ponderous legal quandaries of physicians who care for terminal patients. The problem is also of wider significance for the theory of tort and criminal liability. The area of liability for omissions bristles with moral. ana­lytic and institutional puzzles. In the course of this inquiry, we shall confront some of these problems and others we shall catalogue in  passing. 

II

The question is posed: Is the physicians' discontinuing aid to a terminal patient an-act or omission? To be sure, the choice of legal track does not yield radically different results. For some omissions, physicians are liable in much the same way as they are for non-permitted oper­ations and negligent treatment. One need only consider the following turn of events. Doctor Brown is the family doctor of the Smith family and has been for several years. Tim Smith falls ill with pneumonia. Brown sees him once or twice at the family home and administers the necessary therapy. One evening, upon re­ceiving a telephone call from the Smith family that Tim is in a critical condition, Dr. Brown decides that he should prefer to remain at his bridge game than to visit the sick child. Brown fails to render aid to the child; it is clear that Brown would be liable criminally and civilly if death should ensue. That he has merely omit­ted to act, rather than asserted himself in­tentionally to end life, is immaterial in assess­ing his criminal and civil liability. Of course, the doctor would not be under an obligation to respond to the call of a stranger who said that he needed help. But there is a difference be­tween a stranger and someone who has placed himself in the care of a physician. The factor of reliance and reasonable expectation that the doctor will render aid means that the doctor is legally obligated to do so.' His failure to do so is then tantamount to an intentional infliction of harm. As his motive, be it for good or ill, is irrelevant in analyzing his liability for assertive killing, his motive is also irrelevant in analyzing his liability for omitting to render aid when he is obligated to do so. 

Thus, it makes no difference whether a doc­tor omits to render aid because he prefers to continue playing bridge or if he does so in the hope that the patient's misery will come quickly to a natural end. A doctor may be crim­inally and civilly liable either for intentionally taking life or for omitting to act and thus per­mitting death to occur. However, the sources of these two legal proscriptions are different. And this difference in the source of the law may provide the key for the analysis of the doctor's liability in failing to prolong life in the case discussed at the outset of this article. That a doctor may not actively kill is an application of the general principle that no man may actively kill a fellow human being. In contrast, the prin­ciple that a doctor may not omit to render aid to a patient justifiably relying upon him is a func­tion of the special relationship that exists be­tween doctor and patient. Thus, in analyzing the doctor's legal duty to his patient, one must take into consideration whether the question involved is an act or an omission. If it is an act, the relationship between the doctor and patient is irrelevant. If it is an omission, the re­lationship is all controlling. 

With these points in mind, we may turn to an analysis of specific aspects of the medical de­cision not to prolong life. The first problem is to isolate the relevant medical activity. The re­current pattern includes: stopping cardiac re­suscitation, turning off a respirator, a pace­maker or a kidney machine, and removing the tubes and devices used with these life sustaining machines. The initial decision of classification determines the subsequent legal analysis of the case. If turning off the respirator is an "act" under the law, then it is un­equivocally forbidden: it is on a par with in­jecting air into the patient's veins. If, on the other hand, it is classified as an "omission," the analysis proceeds more flexibly. Whether it would be forbidden as an omission would de­pend on the demands imposed by the re­lationship between doctor and patient. 

There are gaps in the law; and we are con­fronted with one of them. There is simply no way to focus the legal authorities to determine whether the process of turning off the respira­tor is an act or an omission. That turning off the respirator takes physical movement need not be controlling. There might be "acts" without physical movement, as, for example, if one should sit motionless in the driver's seat as one's car heads toward an intended victim. Surely that would be an act causing death; it would be first-degree murder regardless of the relationship between the victim and his assas­sin. Similarly, there might be cases of omis­sions involving physical exertion, perhaps even the effort required to turn off the respira­tor. The problem is not whether there is or there is not physical movement; there must be another test. 

That other test, I should propose, is whether on all the facts we should be inclined to speak of the activity as one that causes harm or one merely that permits harm to occur. The usage of the verbs "causing" and "permitting" cor­responds to the distinction in the clear cases between acts and omissions. If a doctor injects air into the veins of a suffering patient, he causes harm. On the other hand, if the doctor fails to stop on the highway to aid a stranger injured in an automobile accident, he surely permits harm to occur, and he might be morally blameworthy for that; but as the verb "cause" is ordinarily used, his failing to stop is not the cause of the harm .2

 

As native speakers of English, we are equipped with linguistic sensitivity for the dis­tinction between causing harm and permitting  harm to occur. That sensitivity reflects a com­mon sense perception of reality; and we should employ it in classifying the hard cases arising in discussions of the prolongation of life. Is turn­ing off the respirator an instance of causing death or permitting death to occur? If the pa­tient is beyond recovery and on the verge of death, one balks at saying that the activity causes death. It is far more natural to speak of the case as one of permitting death to occur. It is significant that we are inclined to refer to the respirator as a means for prolonging life; we would not speak of insulin shots for a diabetic in the same way. The use of the term "pro­longation of life" builds on the same percep­tion of reality that prompts us to say that turn­ing off the respirator is an activity permitting death to occur, rather than causing death. And that basic perception is that using the respirator interferes artificially in the pattern of events. Of course, the perception of the natural and of the artificial is a function of time and culture. What may seem artificial today, may be a mat­ter of course in ten years. Nonetheless, one does perceive many uses of the respirator to­day as artificial prolongations of life. And that perception of artificiality should be enough to determine the legal classification of the case. Because we are prompted to refer to the activity of turning off the respirator as activity permitting death to occur, rather than causing death, we may classify the case as an omission, rather than as an act. 

To clarify our approach, we might consider this scenario. A pedestrian D notices that a nearby car, parked with apparently inadequate brakes, is about to roll down hill. P's house is parked directly in its path. D rushes to the front of the car and with effort he is able to arrest its movement for a few minutes. Though he feels able to hold back the car for several more min­utes (time enough perhaps to give warning of the danger), he decides that he has had enough; and he steps to one side, knowing full well that his quarry will roll squarely into P's front yard. That is precisely what it does. What are P's rights against D? Again, the problem is whether the defendant's behavior should be treated as an act or as an omission. If it is act, he is liable for trespass against P's property. If it is an omission, the law of trespass is inap­plicable; and the problem devolves into a search for a relationship between P and D that would impose on D the duty to prevent this form of damage to P's property. Initially, one is inclined to regard D's behavior as an act bringing on harm. Like the physician's turning off a respirator, his stepping aside represents phys­ical exertion. Yet as in the physician's case, we are led to the opposite result by asking whether under the circumstances D caused the harm or merely permitted it to occur. Surely, a news­paper account would employ the latter de­scription; D let the cargo, he permitted it to roll on, but he is no more a causal factor than if he had not initially intervened to halt its forward motion. We deny D's causal contribution for reasons akin to those in the physician's case. In both instances, other factors are sufficient in themselves to bring on the harmful result. As the car's brakes were inadequate to hold it on the hill, so the patient's hopeless condition brought on his death. With sufficient causal factors present, we can imagine the harm's oc­curring without the physician's or the pedes­trian's contribution. And thus we are inclined to think of the behavior of each as something less than a causal force.'           

One might agree that as a matter of common sense, we can distinguish between causing harm and permitting harm to occur and yet balk at referring to the way people ordinarily de­scribe phenomena in order to solve hard prob­lems of legal policy. After all, what if people happen to describe things differently? Would that mean that we would have to devise differ­ent answers to the same legal problems? To vindicate a resort to common sense notions and linguistic usage as a touchstone for separating acts from omissions, we must clarify the inter­lacing of these three planes of the problem: (1) the distinction between acts and omissions, (?) the ordinary usage of the terms "causing" and "permitting" and (3) resorting in cases of omissions, but not in cases of acts, to the re­lationship between the agent and his victim in setting the scope of the agent's duties. The question uniting the second and third variables is this: Is there good reason for being guided by the relationship between the parties in cases where the agent has permitted harm to occur, but not in cases where the agent has in­tentionally and directly caused harm to a stran­ger? To answer this question, we need to turn in some detail to the function of causal judg­ments in analyzing liability, whereupon we may clarify the link between the first and sec­ond variables of the analysis, namely between the category of omissions and the process of permitting harm to occur.           

Ascribing liability for tortuous and criminal harm may be looked upon as a two-stage process. The first stage is the isolation of a candi­date for liability. In virtually all dimensions of the law of crimes and torts, we rely upon the concept of causation to separate from the mass of society those individuals who might prove to be liable for the proscribed harm. Upon reduc­ing the number of potentially liable parties to those that have caused the harm, the final stage of analysis demands an evaluation of the facts under the apt rules of liability, e.g., those pre­scribing negligence and proximate cause as conditions for liability. 

The one area of the law where one has difficulty isolating candidates for liability is the area of omissions. When others have stood by and permitted harm to occur, we either have too many candidates for liability or we have none at all. A helpless old woman succumbs to starvation. Many people knew of her condition and did nothing; the postman, her hired nurse, her daughter, the bill collector, the telephone operator--each of them allowed her to die. Could we say, on analogy to causing death, that permitting the death to occur should serve as the criterion for selecting these people as candidates for liability? If we say that all of them are candidates for liability, then the bur­den falls to the criteria of fault to decide which of them, if any, should be liable for wrongful death and criminal homicide. The problem is whether the criteria of fault are sufficiently sen­sitive to resolve the question of liability. What kinds of questions should we ask in assessing fault? Did each voluntarily omit to render aid? Did any one of them face a particular hazard in doing so? Were any of them in a particularly favorable position to avert the risk of death? If these are the questions we must ask in assess­ing fault and affixing liability, we are at a loss to discriminate among the candidates for liability. Each acted voluntarily with knowledge of the peril; none faced personal hazard in offering assistance; and their capacities to avert the risk were equal. Thus, we may use the concept of permitting as we do the notion of causation to narrow the field to those who should be judged on criteria of fault. But if we do. the criteria of fault are useless (at least in the type of case sketched here) for discriminating among the candidates.           

One wonders why this is so. In the arena of caused harms, one may have a large number of candidates for liability. The conventional test of causing harm sweeps wide in encompassing all those but for whose contribution the harm would not have occurred. Yet the criteria of liability-reasonableness of risk, ambit of risk, proximate cause-are effective in further re­ducing the field to those we might fairly hold liable. The reason is that each causal agent is chargeable with a different risk that loss of the given kind would occur. The risks differ in quantum and scope. Some bear a remote re­lationship to the harm; others seem reasonable in light of other circumstances. These differ­ences in the posture of each causal agent to­ward the risk of harm enable us to assess their individual fault with some sensitivity. 

In contrast, those who permit harm to occur do not bear individualized responsibility for the risk of harm. Their status derives not from the creation of the risk, but merely from knowl­edge that the risk exists and from the oppor­tunity to do something about it. One could speak of the likelihood that each could avert the harm. And in some cases, this approach might be useful; a doctor's failing to render aid to a man lying in the street is more egregious than a layman's turning the other way. Yet in the general run of case-the starvation of the old woman discussed above, the Kitty Geno­vese incident'-the risks assignable to passive bystanders are of the same murky order: each could have done something but did not. 

Affixing liability fairly in cases of omission requires a more sensitive filtering mechanism prior to the application of the traditional cri­teria of personal fault. The concept of permit­ting harm sweeps too wide; and the criteria of personal fault tend to be of little avail in nar­rowing the field. Thus one can understand the role of the relationship between the parties as a touchstone of liability. Legal systems, both common law and Continental, have resorted to the relationship between the parties as a device for narrowing the field to those individuals whose liability may be left to depend on per­sonal fault. According to the conventional rules, the old woman's nurse and daughter are candidates potentially liable for permitting death to occur. Liability would rest on personal fault, primarily on the voluntariness of each in omitting to render aid. Thus the conventional rules as to when one has a duty to render aid fulfill the same function as the causal inquiry in its domain: these rules, like the predication of causation, isolate individuals whose behavior is then scrutinized for the marks of negligent and intentional wrongdoing. 

By demonstrating the parallel between the causal concept in cases of acts and the re­lationship between the parties in cases of omis­sions, we have come a long way in support of our thesis. We have shown that in cases of permitting harm to occur, one is required to resort to the relationship between the parties in order fairly to select those parties whose li­ability should turn on criteria of personal fault. In the absence of a causal judgment, with its attendant assignment of differentiated respon­sibility for the risk of harm, one can proceed only by asking: Is this the kind of relationship, e.g.. parent-child, doctor-patient, in which one person ought to help another? And on grounds ranging from common decency to contract, one derives individual duties to render aid when needed. 

One step of the argument remains: the con­clusion that cases of permitting harm are in­stances of omissions, not of acts. This is a step that turns not so much on policy and analysis, as on acceptance of the received premises of the law of homicide. One of these premises is that acting intentionally to cause death is un­conditionally prohibited: the relationship be­tween the defendant and his victim is irre­levant. One may resort to the relationship between the parties only in cases of omissions indirectly resulting in harms With these two choices and no others, the logic of classification is ineluctable. Cases of permit­ting harm, where one must have recourse to the relationship between the parties, cannot be classified as cases of acts: to do so would pre­clude excusing the harm on the ground that the relationship between the parties did not require its avoidance. Thus, to permit recourse to re­lationship of the parties, one must treat cases of permitting harm as cases of omissions. 

To complete our inquiry, we need attend to an asymmetry in the analysis of causing and permitting. As Professors Hart and Honore have shown, some omissions may be the causes of harm. And thus, the category of caus­ing harm includes some cases of omitting as well as all cases of acting to bring on harm. Suppose, for example, that an epileptic reg­ularly takes pills to avert a seizure. Yet on one occasion he omits to take the pills in the hope that he is no longer required to. He has a sei­zure. The cause of his seizure is clear: he omit­ted to take the prescribed pill. In the same way, a physician failing to give a diabetic patient a routine shot of insulin would be the cause of harm that might ensue. The taking of the pill and the giving of the shot are the expected state of affairs They represent normality, and their omission, abnormality. Because we anticipate me opposite, the omission explains what went wrong. why our expectations were not realized. In contrast, if pills to avert epileptic sei­zures had just been devised, we would not say as to someone who had never taken the pills that his failure to do so have brought on his attack. In that case, our expectations would be different, the omission to take pills would not represent an abnormality, and the anticipated omission would not be a satisfying causal ex­planation of the attack.' 

A doctor's failure to give his diabetic patient an insulin shot is a case warranting some atten­tion. By contemporary standards, insulin shots, unlike mechanical respirators, do not interfere artificially in the course of nature; because the use of insulin is standard medical practice, we would not describe its effect as one of prolonging life. We would not say that withholding the shot permits death; it is a case of an omission causing harm. With the pro­hibition against causing death, one should not have to refer to the doctor-patient relationship to determine the criminality of the doctor's omission. Yet in fact, common law courts would ground a conviction for omitting to give the shot on the doctor's duty to render aid to his patient-a duty derived from the doctor-pa­tient relationship. Thus we encounter an ap­parent inconsistency: a case of causing in which one resorts to the relationship of the parties to determine criminality. We can recon­cile the case with our thesis by noting that cases of omissions causing harm possess the criteria -regularity of performance and reliance-that give rise to duties of care. The doctor is clearly under a duty to provide his patient with insulin shot if the situation demands it. And the duty is so clear precisely because one expects an average doctor in the 1960s to use insulin when necessary; this is the same expectation that prompts us to say that his failure to give the shot would be the cause of his patient's death. 

That an omission can on occasion be the cause of harm prompts us slightly to re­formulate our thesis. We cannot say that caus­ing harm may serve as the criterion for an act as opposed to an omission because some in­stances of causation are omissions. But we may claim with undiminished force that permit­ting harm to occur should be sufficient for classification as an omission. Upon analysis, we find that our thesis for distinguishing acts from omissions survives only in part; it works for some omissions, but not for all. Yet, so far as the stimulus of this investigation is con­cerned, the problem of physicians permitting death to come to their terminal patients, the thesis continues to hold: permitting a patient to die is a case in which one appropriately refers to the relationship of the parties to set the scope of the physician's legal duty to his patient; in this sense it functions as an omission in legal analysis. 

III

By permitting recourse to the doctor-patient relationship in fixing the scope of the doctor's duties to his patient, we have at least fashioned the concepts of the common law to respond more sensitively to the problems of the time. We have circumvented the extravagant legal conclusion that a physician's turning off a kid­ney machine or a respirator is tantamount to murder. Yet one critical inquiry remains. How does shunting the analysis into the track of legal omissions actually affect the physician's flexibility in the operating room? We say that his duties are determined by his relationship with his patient; specifically, it is the con­sensual aspect of the relationship that is sup­posed to control the leeway of the physician. Yet there is some question as to where the control actually resides. 

To take a clear case, let us suppose that prior to the onset of a terminal illness, the patient demands that his physician do everything to keep him alive and breathing as long as possi­ble. And the physician responds, "Even if you have a flat EEG reading and there is no chance of recovery?" "Yes," the patient replies. If the doctor agrees to this bizarre demand, he becomes obligated to keep the respirator going indefinitely. Happily, cases of this type do not occur in day-to-day medical practice. In the average case, the patient has not given a thought to the problem; and his physician is not likely to alert him to it. The problem then is whether there is an implicit understanding be­tween physician and patient as to how the physician should proceed in the last stages of a terminal illness. But would there be an implicit understanding about what the physician should do if the patient is in a coma and dependent on a mechanical respirator? This is not the kind of thing as to which the average man has ex­pectations. And if he did, they would be ex­pectations that would be based on the custom­ary practices of the time. If he had heard about a number of cases in which patient had been sustained for long periods of time on respira­tors, he might (at least prior to going into the coma) expect that he would be similarly sustained. 

Thus, the analysis leads us along the fol­lowing path. The doctor's duty to prolong life is a function of his relationship with his patient; and in the typical case, that relationship de­volves into the patient's expectations of the treatment he will receive. Those expectations, in turn, are a function of the practices prevail­ing in the community at the time, and practices in the use of respirators to prolong life are no more and no less than what doctors actually do in the time and place. Thus, we have come full circle. We began the inquiry by asking: Is it legally permissible for doctors to turn off respi­rators used to prolong the life of doomed pa­tients? And the answer after our tortuous jour­ney is simply: It all depends on what doctors customarily do. The law is sometimes no more precise than that. 

The conclusion of our circular journey is that doctors are in a position to fashion their own law to deal with cases of prolongation of life. By establishing customary standards, they may determine the expectations of their pa­tients and thus regulate the understanding and the relationship between doctor and patient. And by regulating that relationship, they may control their legal obligations to render aid to doomed patients.

 Thus the medical profession confronts the challenge of developing humane and sensitive customary standards for guiding decisions to prolong the lives of terminal patients. This is not a challenge that the profession may shirk. For the doctor's legal duties to render aid de­rive from his relationship with the patient. That relationship, along with the expectations im­plicit in it, is the responsibility of the individual doctor and the individual patient. With respect to problems not commonly discussed by the doctor with his patient, particularly the prob­lems of prolonging life, the responsibility for the patient's expectations lies with the medical profession as a whole.

NOTES 

1. Other relationships of reliance giving rise to duties of care are those of carrier and passenger, innkeeper and guest. ship captain and seaman. school master and pupil. W. PROSSER. TORTS 337 (3d ed. 1964). 

2. For the sake of exposition, the thesis is put simply at this stage; it receives some adjustment below. See text at pp. 299-300. 

3. This conclusion is supported by the German theory of conditions (Bedingungs theory), which holds that a factor is not casual if one can imagine the same sequence of events in the absence of that factor. H. L. A. HART & A. M. HONORE, CAUSATION IN THE LAW 391-92 (1959). 

4. Thirty-eight people in New York City watched and listened as Kitty Genovese was murdered outside their apartment building. 198 NATION 602-04 (1964). 

5. E.g., Rex v. Smith. 2 Car & P. 448. 172 E.R. 203 (Gloucester Assizes 1826) (The analysis of criminality of D for failing to care for an idiot brother turns on whether keeping the brother locked up was an act or omission. Finding the latter. the court held that the defendant bore no duty to aid his brother and directed an acquittal)....

 6. H. L. A. HART & A. M. HONORE, CAUS­ATION IN THE LAW 35-36 (1959). 

7. The relationship between expectations and causation is developed more fully in HART & HONORE, ibid., ch. 2. 

From 42 WASHINGTON LAW REVIEWS 999-1016, 1967. Reprinted by permission from Fred B. Rothman & Co., and the Washington Law Review. 

 

Table of Contents for the Online Textbook

© Copyright Philip A. Pecorino 2002. All rights reserved.

Web Surfer's Caveat: These are materials intended for class use.  They are not intended for publication or general distribution.