|Bioethics and Medical
Causation and Moral Responsibility for Death
William E. Stempsey
ABSTRACT: The distinction between killing and letting die has been a controversial element in arguments about the morality of euthanasia and physician-assisted suicide. The killing/letting die distinction is based on causation of death. However, a number of causal factors come into play in any death; it is impossible to state a complete cause of death. I argue that John Mackies analysis of causation in terms of inus factors, insufficient but non-redundant parts of unnecessary but sufficient conditions, helps us to see that moral responsibility for death cannot rest on causation alone. In specifying the cause of death, some factors can be considered alternatively as either causal factors or merely parts of the presupposed background conditions. If a factor is moved from the background field into the causal field, the result is a changed background field. Comparisons of cases of killing and letting die often do just this; hence, the cases depend on different presuppositions and the causation cannot be directly compared. Moral judgments determine how to apportion factors to the causal and background fields.
The distinction between killing and letting die has been used by many to condemn euthanasia and assisted suicide while giving approbation to withdrawing life-support systems in at least some patients. In the recent United States Supreme Court decision which denies a right to physician-assisted suicide, Chief Justice Rehnquist writes that "when a patient refuses life sustaining medical treatment, he dies from an underlying fatal disease or pathology; but if a patient ingests lethal medication prescribed by a physician, he is killed by that medication." (1) It is doubtful, however, that a moral conclusion can be drawn from the distinction. James Rachels gives the example of Smith and Jones, who stand to gain money if a child should die. Smith actively holds the childs head under water; Jones passively stands by and watches the child drown. If both the act and the omission are equally reprehensible, as they do seem to be, then the mere difference between killing and letting die is not sufficient to decide the morality of the case. Rachels concludes that if letting die is sometimes permissible then killing might also sometimes be permissible. (2)
There are several responses one might make. One might reply that it is the underlying intention to bring about death for monetary gain that makes both Smith and Jones wrong, and not whether death results from action or omission. (3) However, one's intention for either killing or letting die might also be to bring about some good such as the relief from suffering. One's intention alone does not determine the morality of an act. No objective system of morality could sustain such a claim.
Dan Brock argues that at least part of the problem with the use of the killing/letting die distinction in common morality is that the notion of causality on which the arguments depend has never been sufficiently clarified. (4) Howard Brody, however, suggests that neither intention nor causation will do the moral work we ask of them; they are just elements to be considered in a casuistic analysis. (5)
I suspect Brody is right, but what I want to focus on here is the element of causation. An analysis of causation that is more complex than usually given helps us to see that our assessments of moral responsibility for death cannot rest on causation alone. Rather, judgments about the cause of death depend in an important way on prior judgments of value and morality. Even so, a more complex analysis of causation can show that there is an important difference between the killing of euthanasia and assisted suicide on the one hand, and the letting die of withdrawing life-support systems on the other.
Causes of Death
One major source of confusion in debates about killing vs. letting die is a failure to distinguish different types of causal explanations. As Martin Benjamin points out, when we ask about the cause of death, we might be asking either for a pathological explanation or a social explanation. (6) The former addresses the biological issue of why some x causes death in all people; the latter addresses the issue of the social and environmental conditions that led a particular person to come into contact with the x that caused death. Omission of treatment is a social cause of death; diseases are pathological causes of death. In saying that when one refuses life-sustaining treatment, death is not caused by the omission of treatment but by an underlying fatal condition, Justice Rehnquist conflates the two senses of causation.
In making assessments of moral responsibility, it is primarily the social causal explanations that are of interest. Both types of causation may be relevant, but we are usually interested in pathological causal explanations only insofar as they are relevant to appraising the social causal explanations.
Any particular death will have a complex of causes of both types. What I want to suggest is that John Mackies analysis of causation (7) can serve as a useful tool for sorting out the various causes of death, and for helping us understand where moral evaluations enter into the analysis.
Mackies Inus Conditions
John Mackie builds on John Stuart Mills analysis of causality and shows that it is seldom, if ever, possible to state a single cause of any effect. (8) Any event might include a number of factors that play a role in the cause. When an event P occurs regularly after the occurrence of some set of factors, A, B, and C, it is the conjunction of those factors, symbolized ABC, that Mackie calls a "minimal sufficient condition" for P. The absence of an event C (symbolized ) can also serve as one or more of these factors. So, the conjunction of the events of drinking a lethal dose of poison, failing to take an antidote, and not having ones stomach pumped () constitutes a minimal sufficient condition for death.
Any minimal sufficient condition ABC may be sufficient to bring about P, but it is usually not necessary. There may be other conjunctions of factors, say DGH or JKL, that are also by themselves sufficient to bring about P. The disjunction of all the conjunctions that represent minimal sufficient conditions, then, would be both a necessary and sufficient condition to bring about P. This complex disjunction (ABC or DGH or JKL) is what Mackie calls the "full cause." No one factor alone, nor any conjunction of just two, is sufficient to bring about P. A, B, and C are individually insufficient, but they are non-redundant parts which, when conjoined, do constitute a sufficient condition ABC. Mackie calls A, B, and C inus factors; each is an insufficient but non-redundant part of an unnecessary but sufficient condition.
One further point is necessary to understand the notion of "full cause." It is virtually impossible to specify all the possible factors that go into causing an event. Causal determinations are made against a background field F which is constituted by many factors of the ordinary environment that are not specified as inus factors. In a consideration of the cause of death of some person, for example, the facts that we are concerned with human beings and that we are living on earth do not get specified as causal factors, but remain covert in F. Having been born might be taken as an inus factor for death, but it is usually more useful to simply include this as a background assumption in F. Mackie states the typical form of a causal regularity as follows:
This is a full description of the cause of P.
As Mackie realizes, it is impossible to state the full cause of death. We simply do not know all the possible combinations of inus factors that are jointly sufficient for death, nor do we know all the possible antidotes that might counteract some sufficient condition. We do, however, know a considerable number of inus factors for death. Usually, we say that some such factor may cause death, given the proper combination of other factors and background conditions. Thus, we can construct "gappy" full causes of death as follows, where A is a known inus factor, X and Y are unknown inus factors (or conjunctions of inus factors), and F is the background causal field:
As Mackie points out, even this incomplete information still permits us to make causal inferences, albeit tentative ones. We can often infer, with probability, an effect from one known inus factor. If A is an inus factor for death, and we observe that death follows, we assume that the other inus factors X and Y, although they remain unknown to us, are also present. The catch is that because these factors are unknown, they are indistinguishable from many of the unspecified factors in the background field F. As experience accumulates, some of the presently unknown factors will become explicit. What I would like to suggest is that this process of stating new knowledge moves previously unknown factors out of the background field F and into specific minimal sufficient conditions. In doing so, it alters the background field itself. But, as we will see, this may make it impossible to directly compare causation in the past and present states. If comparison is not possible, then moral judgments cannot depend on solely on differences in causation.
Causal and Moral Responsibility
I agree with those who argue that what distinguishes killing from letting die is the causal role the agent plays in the death; (11) in the former the agent initiates a course of events that leads to death, and in the latter the agent refrains from intervening in a course of events, not of the agents making, that together with the non-intervention lead to death. (12) Helga Kuhse distinguishes causing death by omission from refraining from preventing death. We fail to prevent death, and so are causes of death in all sorts of cases. But if we are not in a position to prevent the death, even though we are a cause we are not causally responsible. Only when we are in a position to prevent death, and do not, are we causally, and prima facie morally, responsible for the death. Kuhse thinks that moral responsibility for death does not depend on particular social obligations, such as those of the medical profession. That is, it is not the professional role that gives physicians a moral obligation to avoid refraining from preventing death. Rather, causal responsibility is primary and gives rise to the professional obligation. Kuhse says:
I think Kuhse is right in claiming that ability, opportunity and awareness are what give rise to moral responsibility. However, notice how moral judgments have crept into the analysis. How much ability does one need before one is responsible for preventing death? How much is one obliged to risk one's own safety to be judged as having the opportunity to prevent death? Are health care professionals morally required to be aware of the suffering of people who are not their own patients? To what extent do they have responsibility for the health needs of people in developing countries? Settling these types of questions requires recourse to moral values, and so the notion of causal responsibility is not prior to, but in fact depends on ethical judgments. But these sorts of issues are just what are at stake in establishing the professional obligations of professionals. The professional does, in many circumstances, have unique ability, opportunity, and awareness to prevent death, and so may have moral duties that go beyond those of non-professionals.
Killing and Letting Die
Consider the following three scenarios of a typical case. A patient has an advanced carcinoma which is completely blocking the esophageal lumen, and so is being fed through a surgically-placed gastric tube. The patient has no hope for recovery, and is resigned to the fact that death is near.
Scenario 1 Withdrawing life-sustaining treatment: The gastric tube is removed, and all attempts to provide artificial nutrition and hydration are terminated. The patient is given only "comfort care," and dies several days later.
Scenario 2 Active Euthanasia: The patient is given an intravenous dose of a barbiturate followed by a lethal dose of potassium chloride, and dies within seconds.
Scenario 3 Physician-Assisted Suicide: A physician prescribes a lethal dose of a barbiturate, which the patient introduces into the feeding tube. The patient dies in a short time.
Scenario 1 is typically seen as letting die, whereas Scenarios 2 and 3 are taken to be killing, with suicide being a case of killing oneself. The difference hangs on the cause of death, and our interest is in saying how this causal difference makes a moral difference.
Let us symbolize the disease, esophageal carcinoma, as D. The relevant inus factors for death in these scenarios are the patients not eating (N), the failure to intervene with artificial nutrition and hydration (), the lethal injection (L), the prescription of a lethal drug (P), and the patients self-administration of the drug (S). The question is whether L, P, and S are morally worse than .
In Scenario 1, we have a patient with esophageal carcinoma who is unable to eat. When the gastric tube is removed, death ensues. Justice Rehnquist would say that the patient died from esophageal carcinoma. In fact, the patient died from lack of nutrition and hydration (presuming that no other consequence of the cancer, such as a massive hemorrhage and consequent failure to maintain blood pressure, brings about death first). The minimal sufficient condition for death in Scenario 1 is N. It is important, for purposes of moral assessment, to remember that D is included in F, the background causal field. N is caused directly by D and insofar as one is not responsible for D, one is not responsible for N. A person could choose to stop eating and starve to death. The minimal sufficient condition for death would be the same, but moral assessment would be quite different if D were not in F. In that case we would hold the person responsible for not eating.
In Scenario 2, the minimal sufficient condition for death is L. As in Scenario 1, D is still an important background condition for our moral assessment. In Scenario 2, however, N becomes merely a background condition and is no longer an inus factor. N is again important in the moral assessment of the case, but it plays no role in the minimal sufficient condition for death. An injection of a high enough dose of potassium chloride is sufficient to kill anyone, sick or well, eating or unable to eat. Once we put N into the background field, we change the F which serves as the background for comparison of minimal sufficient conditions. Hence, the causation of death in scenarios 1 and 2 cannot be directly compared.
In Scenario 3, the minimal sufficient condition of death is PS. Again D and N are important background conditions for a moral assessment, but they play no role in the minimal sufficient condition for death. Even in the absence of factors like D and N, PS is still a minimal sufficient cause of death. Since F is the same in scenarios 2 and 3, the causation of death can be directly compared. But because F in Scenario 3 is different from the F in Scenario 1, the causation cannot be directly compared in those two cases.
This explains why causation does not do the moral work asked of it in the typical cases of killing vs. letting die, such as that of Rachels. The moral assessment still hinges on whether N, , L, P, and S are themselves morally justified. Our moral assessments about these types of factors will influence how we analyze any particular case. It is possible to formulate minimal sufficient conditions for death in different ways by adding or subtracting factors from the background field. In most cases, I suspect that we make such manipulations in light of our already-existing moral intuitions about the case. The Presidents Commission has explicitly recognized this.
I conclude that euthanasia and assisted suicide are of one causal type. The difference lies in the distribution of responsibility. In euthanasia, only the injector is causally responsible for the death; in assisted suicide, the one who supplies the drugs and the one who takes them share causal responsibility. Withholding or withdrawing life sustaining treatment is of a different causal type. While the one who removes the life-sustaining therapy bears moral responsibility for that act, no one bears responsibility for the other inus factor, the fatal condition. There are good reasons to believe that physicians ought not engage in euthanasia and assisted suicide, but sometimes ought to withhold or withdraw treatment. Those reasons do not depend on the causation, but rather determine the way we decide questions of causation.
(1) Vacco v. Quill, 117 S.Ct. 2293 (1997).
(2) James Rachels, "Active and Passive Euthanasia," New England Journal of Medicine 292 (1975): 78-80.
(3) Thomas D. Sullivan, "Active and Passive Euthanasia: An Impertinent Distinction?" Human Life Review 3 (1977): 40-46.
(4) Dan W. Brock, "Taking Human Life," Ethics 95 (1985): 851-65.
(5) Howard Brody, "Causing, Intending, and Assisting Death," Journal of Clinical Ethics 4 (1993): 112-17.
(6) Martin Benjamin, "Death, Where Is Thy Cause?" Hastings Center Report 6, no.3 (1976): 15-16.
(7) John Mackie, The Cement of the Universe: A Study of Causation, paperback edition (Oxford: Clarendon Press, 1980).
(8) Ibid., 60-62.
(9) Ibid., 63.
(10) Ibid., 66-67. This statement of a "gappy" universal is fundamentally the same as Mackies, but I have altered the formulation for consistency.
(11) O.H. Green, "Killing and Letting Die," American Philosophical Quarterly 17 (1980): 195-204, and Helga Kuhse, The Sanctity-of-Life Doctrine in Medicine: A Critique (Oxford: Clarendon Press, 1987).
(12) Kuhse, 50-51.
(13) Ibid., 67-68.
(14) Ibid., 68.
(15) Presidents Commission for the Study of Ethical Problems in Medicine and Biomedical and Behavioral Research, Deciding to Forego Life-Sustaining Treatment: Ethical, Medical, and Legal Issues in Treatment Decisions (Washington: U.S. Government Printing Office, 1983), 69.