Lay views in Southern France of the acceptability of refusing to provide treatment because of alleged futility

Abstract

Aim

To carry out a detailed study of existing positions in the French public of the acceptability of refusing treatment because of alleged futility, and to try to link these to people’s age, gender, and religious practice.

Method

248 lay participants living in southern France were presented with 16 brief vignettes depicting a cancer patient at the end of life who asks his doctor to administer a new cancer treatment he has heard about. Considering that this treatment is futile in the patient’s case, the doctor refuses to prescribe it. The vignettes were composed by systematically varying the level of four factors: likelihood of a positive effect, painfulness to the patient of the treatment, cost of the treatment, and attitude of the family.

Results

Five main positions were identified. For 10%, refusing treatment was almost never acceptable. 35% judged acceptability in line with the level of painfulness. 19% judged acceptability consistent with an interaction between the painfulness of treatment and likelihood of positive effect. For 30% it was either almost always acceptable or always acceptable. 5% did not take a position.

Conclusion

A range of positions regarding the acceptability of refusing to provide treatment on the basis of perceived futility was observed. These positions have been analyzed in terms of what physicians and medical ethicists would see as the four principles of medical ethics. This description of lay people’s positions in terms of the principles of medical ethics present clinicians with a conceptual tool to improve communication and shared decision making.

Futile behavior is, in common understanding, behavior that, in a given situation, has little or no chance of achieving the desired objective [1]. Qualifying a medical intervention as futile, therefore, presupposes both that one has accurate knowledge of the objective pursued and that there is empirical evidence that in many situations similar to the one under consideration, the goal has never (or almost never) been achieved as a result of the intervention.

In actual practice, however, it can be complicated to determine and act on futility [2]. Indeed, not only do medical ethicists disagree [3,4,5], but studies of medical practitioners’ personal conceptions of futility show that there is a wide range of definitions and that futility is a matter of judgment [6]. In his definition of medical futility, Schneidermann [7] clearly wishes to show that its two aspects – probabilistic and utilitarian – are inseparable. In his view, “medical futility is the unacceptable likelihood of achieving an effect that the patient has the capacity to appreciate as a benefit” (p. 125). The goal of an intervention and the acceptable probability of success may differ between physicians and their patients [8]. As a result, physicians may agree to prescribe treatments to patients who want them even though the physicians believe them to be futile [9, 10], or they may refuse to prescribe such treatments because they do not understand goals of the patients. In addition, a treatment can be both medically effective and futile if the patient does not see the effects as a benefit [11].

Few studies have examined how lay people perceive the assertion that a given treatment is futile, that is, how they react when a patient is refused treatment because the physician has judged it to be futile [6]. In a large study in Korea that surveyed the attitudes of cancer patients, family, members of the general public, and oncologists toward five types of inventions at the end of life, withdrawal of futile life-sustaining treatment was approved by 89.9% of patients, 87.1% of family caregivers, 89.8% of the general public, and 94.0% of oncologists [12], and a later study by the same group gave almost identical results [13]. In contrast, in the USA, one study reported that 64% of patients’ relatives questioned the physician’s judgment of treatment futility and that 18%, especially those who put forward religious objections, would choose to maintain life support for their relative even if the physician judged the relative’s chances of survival to be zero [14], and another study using a set of focus groups of former hospital patients revealed a lack of knowledge about futility and an insistence that families, not physicians, should be the ultimate decision-makers [15]. In addition, a study in Japan showed that, in various concrete situations (such as a patient with advanced lung cancer who requests another chemotherapy treatment when doctors feel that further treatment would be ineffective and life expectancy is only a few weeks), lay people were about twice as likely as physicians to wish for continuation of treatment [16]. Yet when lay people in the USA were asked about providing chemotherapy or dialysis for an extremely compromised elder patient, a decision to withhold treatment was approved by 80.5% and 73.6%, respectively [17].

It is likely that the different results of these studies of lay people are, at least in part, due to differences in cultures [18] and in the means by which opinions about withholding treatment were obtained. It is clear, nonetheless, that some people are somewhat skeptical about futility judgments and believe that the patient and the family, not the physician, should be the primary decision maker. As in many other areas of health care practice [19,20,21,22,23], however, the positions of lay people on this topic may be even more diverse than suggested by these studies. This possible diversity can be anticipated by applying the interpretive framework offered by the set of four bioethical principles proposed by Beauchamp and Childress [24]: autonomy, non-maleficence, beneficence, and justice. While the adequacy of the “principlism” of Beauchamp and Childress as a guide to ethical decision making by clinicians has been debated [25,26,27,28], their principles can be useful to researchers when considering the importance non-clinicians place on the different features of clinicians’ decisions.

It is very likely that, for some people, the principle of patient autonomy must take precedence over any other consideration. If a patient requests a certain treatment, the physician must, according to these people, prescribe it, unless the law does not allow it. The fact that this treatment is likely to be considered futile by the medical community is not a consideration. For other people, the principle of non-maleficence may prevail. If a patient requests a certain treatment, the physician must, in principle, prescribe it, unless it would cause the patient great hardship. Refusing to prescribe a treatment that is not only futile but would lead to a serious deterioration of the patient’s quality of life would, therefore, be seen as acceptable.

For still others, the principle of beneficence may be primary. If a patient requests a treatment the physician considers futile, the physician must prescribe it if there is a chance, however small, that the treatment will be effective in that particular case, even if the benefit is only a temporary placebo effect or slight prolongation of life. Finally, for some people, the principle of social justice must prevail. If a patient requests a treatment the physician considers futile, the physician must refuse the treatment in light of the limitation of resources available for health care. This position is, moreover, in conformity with the official recommendations in France [29]. It is also close to a position in favor of systematic rationing of care.

What is not known and difficult to anticipate is the frequency of these positions in a given population and the relationships that may exist between a given position and the demographic characteristics of the people who express it. The aim of the present study is, therefore, to carry out the most detailed mapping possible of existing positions in the French public and to link these to people’s age, gender, and religious practice.

Methods

Participants

The 248 lay participants were unpaid volunteers living in southern France. They were approached by one of four research assistants while walking along the main sidewalks in Toulouse. Of the 400 contacted, 248 (62%) agreed to participate. None were health care providers, but fifty-nine (24%) had already faced the issue of treatment refusal either personally or regarding a close friend or family member. Demographic characteristics are shown in Table 1. All participants gave their informed consent.

Table 1 Demographic characteristics of the sample. Composition of the clusters

Material

The material consisted of 16 cards containing a brief vignette, a question, and a response scale. Each vignette depicted the situation of a cancer patient at the end of life who asks his doctor to administer a new cancer treatment that he has heard about. Considering that this treatment is futile in the patient’s case, the doctor refuses to prescribe it. We used the simple dictionary definition of futile (“completely ineffective” [1]) so that the lay participants, as well as medical practitioners would understand. The doctor and patient names were all fictitious.

The vignettes were composed according to a four within-subject factor design: likelihood of a positive effect (no chance of improving the patient’s condition vs. little chance) x painfulness to the patient of the treatment (extremely painful vs. not very painful) x cost of the treatment (very expensive vs. not very expensive) x attitude of the family (family does not request vs. family insists that the patient receive the treatment). The question was, “To what extent do you think that the physician’s decision was acceptable?” The response scale was an 11-point scale with anchors of “not acceptable at all” (0) and “completely acceptable” (10).

One example of a scenario is: “Mr. Georges is suffering from cancer. He is currently in a terminal phase. Everything has been done to ease his physical pain. Mr. Georges insists to the doctor in charge of his care, Dr. Devaux, that a relatively new treatment be administered to him. Mr. Georges’ family is of the opposite opinion. They do not want a new treatment to be initiated. There is absolutely no chance, given the state of Mr. Georges’ health, that this treatment will have any positive effects. Moreover, this treatment is not only very expensive but also extremely painful for the patient. It risks prolonging his agony. Dr. Devaux refuses to administer the treatment. He argues that this treatment is, considering the condition of Mr. George, totally futile.”

Since, in real life, decision making about futility is complex and often involves the consideration of a combination of multiple factors, a scenario methodology is likely to be more informative than asking participants to focus on individual factors. (even though the number of factors in the scenarios is limited). Participants may be incompletely aware of their judgment processes, and some complex judgment processes are difficult to articulate verbally, particularly in cases where the effect on the acceptability judgment of one of the situation’s features, such as pain, depends on the levels of other situational features such as cost and probability, as was the case for the majority of our participants. In the approach we used, participants were presented with various realistic situations and were able almost instantly to judge the level of acceptability of the doctor’s decision. The judgment processes and relative weights given to different factors could then be inferred by means of the statistical analyses [30].

Procedure

The site was a vacant classroom in the local university or the participant’s private home (whichever was most convenient for the participant). Each person was tested individually according to the procedure recommended by Anderson [30,31,32]. The 16 cards were presented in random order to the participants, who made ratings at their own pace. They took 15–20 min to complete the session. The study conformed to the ethical recommendations of the French Society of Psychology. The Ethics Committee of the University of Toulouse approved the study. Participants’ anonymity was respected.

Results

To look for groupings of participants, a cluster analysis was performed on the raw data in accordance with the recommendations of Hofmans and Mullet [33]. Six clusters of participants were identified. They are described in Table 1 and shown in Fig. 1.

Fig. 1

Mean acceptability judgments of the clusters. Mean acceptability (on the y-axis) of refusing treatment, as a function of the pain to the patient (on the x-axis) and as a function of the combination of the chances of improving the patient’s condition and the cost of the treatment (the four curves)

The first cluster (N = 26) was named Almost Never Acceptable for the reason that, as can be observed in Fig. 1 (left-hand panel), all ratings of the acceptability of the physician’s refusal to administer treatment were low (M = 2.14, SE = 0.75). The second cluster (N = 86) was named Mainly Depends on Painfulness because, as can be observed in Fig. 1 (second panel), this factor had by far the strongest effect. Acceptability scores were higher when the level of pain caused by the treatment would be very great (M = 7.99, SE = 0.39) than when it would not be so great (M = 3.44, SE = 0.45), η²_p = .82. In the latter case, the other factors also had an effect. The acceptability rating was higher when the treatment was both expensive and without a chance of producing a positive effect (M = 5.35, SE = 0.35) than in the opposite cases (M = 1.84, SE = 0.21), η²_p = .25.

The third cluster (N = 48) was named Depends on Chances and Painfulness for the reason that, as can be observed in Fig. 1 (third panel), these two factors not only had strong individual effects but also interacted. Acceptability ratings were higher when chances were zero (M = 8.98, SE = 0.36) than when they were very low though not zero (M = 4.54, SE = 0.82), η²_p = .84. Ratings were also higher when the level of pain caused by the treatment would be very high (M = 7.71, SE = 0.61) than when it would not be so high (M = 5.80, SE = 0.48), η²_p = .73. This second effect was strongly apparent only when some benefit was possible even if the chances were very low (6.06–3.02 = 3.04). When the chances were zero, the impact of the level of pain was much reduced (9.36–8.60 = 0.76), η²_p = .57.

The fourth cluster (N = 58) was named Almost Always Acceptable because, as can be observed in Fig. 1 (fourth panel), all acceptability ratings of refusal to treat were high (M = 7.79, SE = 0.28). When the level of pain caused by the treatment would not be extremely high, however, acceptability was somewhat lower when the treatment was not very expensive and had at least a minimal chance of producing a positive effect (M = 5.43, SE = 0.20) than when the treatment was both very expensive and without any chance of success (M = 7.92, SE = 0.21), η²_p = .28.

The fifth cluster (N = 18) was named Always Acceptable for the reason that, as can be observed in Fig. 1 (fifth panel), acceptability ratings were consistently very high (M = 9.41, SE = 0.34). Finally, the sixth cluster (N = 12) was named Indeterminate because, as can be observed in Fig. 1 (right hand panel), all ratings were near the center of the response scale (M = 4.99, SE = 1.25) and no effect of any factor was detected.

In order to test the impact of demographic characteristics, chi-square tests of independence were performed. The threshold was set at p = .05. None of the characteristics of the participants—age, gender, education, religiosity (atheists, believers in God but not church attendees, or regular church attendees), and experience with end-of-life decision making—had a significant impact on cluster membership (see Table 1).

Discussion

As expected, participants expressed a wide range of positions regarding the acceptability of refusing care on the basis of perceived futility. These responses can be.

described by, and are consistent with, what physicians and medical ethicists would see as Beauchamp and Childress’s four principles of medical ethics.

Even if the principles of Beauchamp and Childress may or may not be “the best moral framework,” they were useful for our study. First, as noted by Holm in his critique [25], their work is influential around the world, even if their application of their principles is more appropriate in the American context. Second, we did not use the principles to guide judgments but to study them. Third, the scenarios described realistic situations that could be encountered in real life. We chose the four factors because they seemed most relevant to real decision making, not because they were Beauchamp and Childress’s principles. Fourth, we found that the Beauchamp and Childress framework was useful in interpreting the results of studies conducted in countries as diverse as India, Turkey, Kuwait, Togo, and France [35]. Fifth, the study was a judgment analysis, not an ethical analysis; we looked at judgments of the acceptability of the physician’s decision, not of its ethical correctness.

The most frequent position (35% of the participants) – Mainly Depends on Painfulness—was consistent with the principle of non-maleficence. This position was in keeping with a judgment that refusing to administer an extremely painful treatment would always be justified. If, however, the treatment would be less painful, the acceptability of refusing it would depend, in part, on the cost of the treatment and the chances of obtaining a positive effect. Nonetheless, the acceptability of refusing a less painful treatment that was very costly would still be moderately high.

Participants expressing the next most frequent position (23%)—Almost Always Acceptable—made judgments consistent with supporting the refusal of wasteful and non-beneficial treatment although only moderately so if the treatment would not be very costly, would not be extremely painful, and would offer some hope of benefit. Since the public insurance system in France would have to pay for the wasteful treatment, this position was consistent with the principle of social justice. We do not know, of course, how many of these participants actually had the interest of other citizens in mind.

The position called Depends on Chances and Painfulness (19%) was most consistent with a balance between the principles primarily of beneficence and secondarily of non-maleficence. If there was some chance that the treatment would have a positive effect, it would be unjust to refuse it. Refusal would be tolerable only if the treatment would be both extremely painful and very expensive.

Finally, two extreme positions were expressed by small minorities of participants. For some (7%), futile treatment should always be refused. This position was consistent either with a willingness to follow the physician’s decision in all cases or, for those aware that the futile treatment would waste limited resources, with the principle of social justice. For the others (10%), in contrast, a physician should not refuse to administer a treatment desired by a patient, regardless of its cost and expected lack of benefit. This position is consistent with a focus on the principle of patient autonomy. It also mirrors the insistence by the lay people studied by Neville and colleagues [15] and Kadooka and colleagues [16] that patients and families, not physicians, should be the primary decision makers.

These results are consistent with French culture, which, in general, is characterized by an emphasis both on individual autonomy and on social (or distributive) justice, i.e., on egalitarianism [34]. It was not surprising, therefore, to find that some participants emphasized autonomy and others social justice. Yet, as expected, considerations of beneficence and non-maleficence also played a role.

Surprisingly, none of the few demographic variables considered were associated with the positions expressed. Even religious belief or involvement—found in other studies to motivate people to want to continue any possible life-sustaining treatments [14, 35]—failed to produce any notable differences, possibly because treatments in the scenarios were not linked to prolonging life.

The study has, of course, several limitations. First, it was performed on a modest-sized convenience sample of people in the south of France; generalization of the findings must, therefore, be done with care. The aim was, however, to identify the different positions on medical futility that lay people might take, not to determine exact percentages. Second, the applicability of our findings to countries in which, unlike in France, people tend to place less emphasis on individualism and social justice (as described by Hofstede [34]) may seem reduced. Yet, as Ng and colleagues point out [18], the issues remain the same. It is likely, therefore, that—as in the case of physician-assisted suicide [35]—the same, or similar, positions will be found even if they vary in importance. Third, the number of relevant factors and their levels was limited by the need to construct scenarios according to a within-subject factor design. Nonetheless, the chosen factors are important in decision making, and the participants demonstrated in their responses that they saw important differences in the factor levels. The scenario method is much better at determining participants’ views than questionnaires listing possible factors because it presents patients with real-life combinations of factors [30]. Fourth, for this reason, the study could not assess the importance. shown by others [15], of people’s trust in physicians and the medical system nor of some people’s belief, often religious, in the overwhelming value of life and the need to prolong it at all costs [14, 36]. In future studies, it would be useful to ask participants, after they finish judging the scenarios, if they can explain on what bases they made their ratings. In addition, administering personality tests and measures of collectivism-individualism and gathering additional demographic information—including the participants’ economic situations and political affiliations—might help explain their responses, although some participants might find these questions too burdensome or intrusive. Fifth, since not all medical ethicists agree that the principles of Beauchamp and Childress are the most applicable (as noted above), future studies could—through interviews and vignettes, the use of other analytical tools, and consideration of other ethical approaches (such as Reis-Dennis’s insistence on “the absolute right to be respected as a human being with inherent dignity” [37])—assess the applicability and usefulness of other moral theories of lay persons’ judgments of futility. Sixth, the scenario portrayed a cancer patient at the end of life. Future studies might look at physicians’ assertions of futility in the case of other patients, with both terminal and non-terminal diseases, to ascertain to what extent people’s positions depend on the nature of the disease and on the ultimate prognosis.

Implications

The issue of medical futility can be very difficult for physicians to discuss with patients and their surrogates, and, for this reason, it is important for physicians to delve deeply into it [38]. To accomplish this, clinicians need to be aware of the very different positions on futility taken by lay people [39, 40]. Bailoor and colleagues [17] already showed, in a nationwide survey in the United States, that explaining to the public how decisions are made increases their acceptance of determinations of medical futility. The current study provides a description of different lay attitudes to futility, which may be useful for those dealing with patients, their family members, or surrogates when facing critical illness. Even though their views may be altered, at least to some degree, by the dire state in which the patients find themselves, a description of lay people’s positions in terms of the principles of medical ethics may offer clinicians a conceptual tool for exploring the moral issues of futility with patients and families that may help to improve communication and shared decision making, which are essential [4, 40]. The clinicians can explore, and appreciate, which of these principles are most important to the patient, to the family members, to surrogates, and to the others involved in the patient’s life. This has the potential to offer a common moral language to clinicians and patients as they try to deal with a tragic situation.