Article Text

Download PDFPDF
The role of the principle of double effect in ethics education at US medical schools and its potential impact on pain management at the end of life
  1. Robert Macauley1,2
  1. 1Department of Clinical Ethics, Fletcher Allen Health Care, Burlington, Vermont, USA
  2. 2University of Vermont, Department of Pediatrics, Vermont, USA
  1. Correspondence to Dr Robert Macauley, Department of Clinical Ethics, Fletcher Allen Health Care, 111 Colchester Avenue, Smith 266, Burlington, Vermont 05401, USA; robert.macauley{at}


Background Because opioids can suppress respiratory drive, the principle of double effect (PDE) has been used to justify their use for terminally ill patients. Recent studies, however, suggest that the risk of respiratory depression in typical end-of-life (EOL) situations may be overstated and that heightened concern for this rare occurrence can lead to inadequate treatment of pain. The purpose of this study is to examine the role of the PDE in medical school ethics education, with specific reference to its potential impact on pain management at EOL.

Method After obtaining institutional review board approval, an electronic survey was sent to ethics educators at every allopathic medical school in the USA.

Results One-third of ethics educators felt that opioids were ‘likely’ to cause significant respiratory depression that could hasten death. Educators' opinions of opioid effects did not influence their view of the relevance of the PDE, with approximately 70% deeming it relevant to EOL care. Only 15% of ethics educators believed that associating the PDE with opioid use might discourage clinicians from optimally treating pain, out of concern for respiratory depression.

Conclusion This study demonstrates that a significant minority of ethics educators believe, contrary to current evidence, that opioids are ‘likely’ to cause significant respiratory depression that could hasten death in terminally ill patients. Yet, many of those who do not feel this is likely still rely on the PDE to justify this possibility, potentially (and unknowingly) contributing to clinical misperceptions and underutilisation of opioids at EOL.

Statistics from


Physicians must first do no harm (primum non nocere), but sometimes it may be necessary to risk harm in order to achieve a greater good. In order to determine whether such an action with both good and bad effects is justified, philosophers have often relied on the principle of double effect (PDE), first formulated by Thomas Aquinas in relationship to killing in self-defence.1 The PDE has four basic components:

  1. The act itself must be, at worst, morally neutral.

  2. The bad effect cannot be the means to the good effect.

  3. The good effect must outweigh the bad effect (principle of proportionality).

  4. The agent must only intend the good effect, although the bad effect may be foreseen.2

One of the most common applications of the PDE is to the use of opioid administration at the end of life (EOL). Because of the effect of opioids on respiratory drive, ‘the level of analgesia necessary to relieve the patient's pain … may also have the effect of shortening the patient's life.’3 The PDE is used to justify potentially hastened death because (1) opioid administration is not morally bad; (2) opioids relieve pain directly (rather than doing so through the death of the patient); (3) relief of extreme pain outweighs the shortened life of a patient who is dying; and (4) the agent intends to minimise suffering, rather than hasten death.

The application of the PDE to opioid administration at EOL has been criticised, however, for a variety of reasons including: an oversimplified concept of human intention, the fact that respiratory depression is (at worst) only a potential side effect of opioids and the assumption that death is always a ‘bad’ effect. Some have gone even further, arguing that appealing to the PDE to justify the use of opioids at EOL may not only be unnecessary, but could also be harmful if this practice reinforces the misperception that opioids inevitably hasten death, thus leading to inadequate pain management. The term ‘opiophobia’ has been used to refer to physicians' reluctance to treat pain aggressively, out of concern for addiction or (in this case) hastened death.4

To explore the connection between the PDE and opiophobia, we identified medical school ethics education as a potential context where the PDE might be used to justify opioid use at EOL. We hypothesised a direct correlation between an ethics educator believing that opioids hasten death and appealing to the PDE in education. We also hypothesised that clinicians (ie, physicians and nurses) would be less likely than non-clinicians to link opioid treatment to hastened death, given their bedside experience with dying patients.


We devised a 12-question electronic survey exploring the role of the PDE in medical school ethics education in the USA. Individuals involved in ethics education at all 127 allopathic medical schools in the 50 states of the USA were identified, through personal knowledge and internet searching. (Three individuals were identified for one school, and two contacts for 12 other schools.) The study was approved by the University of Vermont Institutional Review Board.

Invitations to participate in the study were emailed to 141 individuals, with a request to forward the survey to other ethics educators. A follow-up invitation was sent 2 weeks later. Eight emails were returned as undeliverable. Respondents were asked to provide the name of their institution (to permit measurement of both inter- and intrainstitutional variability) but were assured that, in order to ensure anonymity of respondents, institutional affiliations would not be reported.

Data were analysed by χ2 analysis using SPSS V.18, with a 95% CI.


Seventy-two responses were received, representing 59 medical schools. This represents 46% of all allopathic medical schools in the 50 states in the USA, with multiple respondents from five schools (three schools had two respondents each, one had three respondents and one had ten respondents). Thirty-nine per cent of respondents were course directors for their medical school's ethics curriculum. Fifty-one per cent of respondents were clinicians (allopathic physician, nurse or chiropractic physician), with the remainder holding the following degrees: PhD (44%), MA/MS/MPH (17%) and JD (15%).

Thirteen respondents did not complete the entire survey, but the responses they did provide were included in data analysis.

Prevalence of PDE in medical school ethics education

The majority of respondents were ‘very familiar’ with the PDE (57%), with only 11% reporting that they were ‘not at all familiar’ with it. There was, however, substantial inter-institutional variability regarding the teaching of the PDE in the medical school curriculum. Eighty-seven per cent of respondents reported that the PDE was taught either by name or ‘implicitly, in response to relevant situations’, while 8% said it was not referred to in their curriculum (and 5% were not sure). If the PDE was part of the curriculum at the respondent's school, the most common clinical example was the administration of opioids to terminally ill patients (72%), followed by palliative sedation (64%) (table 1).

Table 1

Role of PDE in medical school education in the USA

There was also significant intrainstitutional variability at the school with the most respondents. Of these ten respondents, six reported the PDE was taught in relation to palliative sedation, five to the use of opioids at EOL, three to termination of intrauterine pregnancy and two to self-defence. Two respondents stated the PDE was not taught at all in the school's curriculum.

Relationship of PDE to use of opioids in EOL care

Respondents generally felt that the PDE was both relevant (71%) and useful (68%) in ‘justifying optimal pain and other symptom management at the end of life’. The vast majority of respondents felt that both patients and staff would benefit from an explanation of how the PDE justifies optimal pain and other symptom management at EOL. Only 7% felt that the PDE should not be used to justify pain and symptom management in this situation.

Some respondents noted that, even if they did not personally believe the PDE was necessary to justify opioid use at EOL, nevertheless it was useful to others who had a different world view. When asked who would benefit from discussion of the PDE, one respondent commented: ‘Individuals who believe that it is wrong to kill an innocent person intentionally. The PDE is irrelevant to many of my students because they believe it is ok to kill patients to relieve intractable pain’ (table 2).

Table 2

Relevance and utility of PDE in relationship to symptom management at EOL

A significant minority of respondents believed that, in typical EOL situations, opioids were ‘likely’ to cause significant respiratory depression that could hasten death (34%). Of these respondents, 71% deemed the PDE relevant to EOL care. Of respondents who felt that opioids were not likely to suppress respiration, a similar percentage (69%) affirmed the relevance of the PDE. The only statistically significant difference noted involved the 5% of respondents who had previously stated that the PDE was ‘not at all’ relevant or useful, as these respondents were significantly less likely to believe that opioids depress respiration in typical EOL situations (p=0.04).

Respondents were then asked, if they did not feel the PDE was relevant to opioid administration, what there reason(s) were for reaching this conclusion. Reasons cited included overstated risk of respiratory depression (31%), that opioids were no different from other medical treatments having significant side effects (23%), that death was not always a bad effect (19%) and that intention was too subjective to say that a bad effect was merely ‘foreseen’ (14%). Emphasising this complex view of intention, one respondent commented: “Employing the PDE is teaching people to lie to themselves and others. People are responsible for the foreseeable consequences of their actions.”

Interestingly, while 71% of respondents initially deemed the PDE relevant to EOL care, when subsequently prompted with potential reasons why this might not be so, only 48% reaffirmed its relevance. Those who did so were more likely to believe that opioids hastened death in typical EOL situations (57% vs 44%, p=0.02).

Regardless of respondents' views regarding the role of opioids in respiratory depression, few were concerned about negative ramifications of justifying their use according to the PDE. Even though two-thirds of respondents thought that opioids were unlikely to hasten death in EOL situations, only 15% felt that associating the PDE with opioid administration at EOL ‘will discourage clinicians from optimally treating pain, out of excessive concern for respiratory depression’ (table 3).

Table 3

Opioids and respiratory depression

Clinicians versus non-clinicians

There was no statistically significant difference between clinicians (ie, physicians and nurses) and non-clinicians with regard to familiarity with the PDE (p=0.2), relevance and usefulness of the PDE (p>0.7), or the risk of discouraging clinicians from adequately treating pain (p>0.7). Non-clinicians were slightly more likely to believe that opioids were likely to cause significant respiratory depression in typical EOL situations (39% to 30%), but this did not reach statistical significance (p=0.07).


Criticisms of the PDE

While the PDE has historically been used to justify the use of medications which may depress respiratory drive at EOL,5 6 the applicability of the PDE to such situations has recently been questioned. Some have argued that the PDE cannot practically be applied because it relies on an analysis of intention that is ‘difficult to validate externally, and inconsistent with other analyses of human intention’.7 Such critics claim that human intention is complex and multi-faceted, and thus it is unrealistic to claim that a foreseen (and, to some, not unwelcome) side effect can be wholly unintended. For example, if a clinician's primary intention is to relieve pain, but at the same time he would be relieved if the patient died peacefully, does this fulfil the criteria of the PDE?

Others have argued that the PDE is unnecessary because death is not necessarily a ‘bad effect’. Such critics point to the Judeo-Christian origins of the PDE in its original formulation, calling into question its applicability in a multi-cultural society. Instead of viewing life as a gift from God and death as something to be prevented (or at least delayed) at all costs, death is viewed as a part of life, an inevitable conclusion that should reflect the patient's values and preferences.7 Current debates surrounding Physician Assisted Death reveal that, for many, there are fates worse than death (including unrelieved suffering, whether physical or emotional).8

Even those who defend the PDE seem to accept this criticism, while still advocating for the relevance of the PDE for those individuals who do believe that death is a bad effect (and thus must never be intentionally hastened). Sulmasy and Pellegrino argue that for individuals who are ‘fearful of unwittingly participating in euthanasia if a patient's death is hastened, … the rule of double effect provides moral reassurance and thus encourages optimal care of the dying’.9

Still others have asserted that the PDE is not relevant to such situations because even if opioids may hasten death, they do not necessarily. Thus, instead of having two consistent effects (ie, the good effect of pain relief and the bad effect of hastened death), opioids have a clear beneficial effect with several associated risks, which could be said of most medical procedures (for which the PDE is not appealed to).10 A more appropriate approach would be to consider the relative benefits and burdens of opioids in EOL situations. Angell describes over-reliance on the PDE in this way: ‘I can't think of any other area in medicine in which such an extravagant concern for side effects so drastically limits treatment.’11

Yet even if opioids can suppress respiratory drive, this does not mean that they hasten death in typical EOL situations. Quite the contrary, recent studies have shown that appropriately dosed opioids may actually prolong—rather than shorten—a patient's life. In a large study of hospice patients, for instance, opioid use was directly (rather than inversely) correlated with duration of life for terminally ill patients.12 In light of these emerging data, Fohr concludes: ‘In the case of medication to relieve pain in the dying patient, the PDE should be rejected not on ethical grounds, but for a lack of medical reality.’13 Yet, despite recent editorials (bearing titles like ‘Morphine kills the pain, not the patient’14), one-third of ethics educators in this study felt that opioids were likely to hasten death in typical EOL situations.

The arguments considered up to this point centre on whether the PDE is practical, relevant and necessary. Others have taken the criticism of the PDE further, arguing that its intimate connection to the use of opioids at EOL could be harmful. Clearly, the only reason why the PDE might be relevant to the use of opioids at EOL is if such medications hasten death. Concern for this side effect could act as a deterrent to optimal symptom management, though, as illustrated in a recent study, which found that 41% of physicians and nurses agreed that ‘clinicians give inadequate pain medication most often out of fear of hastening a patient's death’.15

Hypothesis 1: perception of opioids and relevance of PDE

Reasoning that physicians-in-training would learn of the PDE through ethics education, we sought to determine whether the risk of hastened death due to opioids might be overstated in this context. Of concern, one-third of educators in this study believed that opioids were likely to hasten death, contrary to current evidence.

We hypothesised a direct correlation between an ethics educator believing that opioids hasten death and appealing to the PDE in education. This hypothesis was not supported by the data. Respondents' opinions of the relevance of the PDE were not influenced by their estimate of opioid effects, with approximately 70% asserting that it was relevant to EOL care.

In a later question, however, after several of the aforementioned arguments for the inapplicability of the PDE to pain at EOL had been noted, only 48% of respondents affirmed its relevance, yielding a statistically significant correlation with views on opioid effects at EOL (p=0.02). One respondent's free-text response may explain this discrepancy:In a prior question, I said that PDE is very relevant but I did so on theoretical, not empirical grounds. In this question I checked the ‘overstated’ option because to my knowledge empirical studies have not demonstrated any diminution in survival from intensive palliation, and indeed there is some evidence to suggest the contrary.

It is not surprising that ethics educators who feel opioids are likely to hasten death find the PDE relevant in justifying the use of such medications, and thus teach medical students about applications of this principle. Yet, this study reveals that those educators who believe opioids rarely hasten death are just as likely to teach students about the relevance of the PDE to EOL care. The PDE is not necessary for some healthcare professionals to adequately treat pain (eg, those who do not view hastened death as necessarily a ‘bad effect’), the reasoning goes, but it is necessary for those concerned with unwittingly participating in euthanasia, even if the risk is very small. The intention, if you will, of the ethics educators is to provide physicians-in-training with philosophical justification for optimal symptomatic management of dying patients, even if many of these educators who have a more accurate understanding of opioid side effects—when pressed on the practicalities of EOL care—admit that the PDE is not so relevant, after all.

Continued appeal to the PDE may have negative consequences, though. It is possible that persistent identification of the PDE with opioids could unintentionally reinforce the misperception that opioids are likely to hasten death (a misperception, it should be repeated, that a third of ethics educators fall prey to). If the vast majority of medical schools provide students with a philosophical response to the exception to the rule (ie, the rare instances when opioids could hasten death), could students graduate with the belief that the exception is the rule?

Clearly, ethics educators do not see this as a risk, as only 15% were worried that tying the PDE to opioids could impede optimal pain management. Yet, given the risks of opiophobia (and its potential contribution to inadequate pain management at EOL), we believe that greater attention should be paid to the manner in which the PDE is taught to ensure that its (at most) rare applicability to EOL care is fully appreciated by physicians-in-training. For students (and healthcare professionals) who believe death is always a bad effect and who are concerned about the potential for opioids to hasten death in extreme circumstances, the PDE can empower them to provide optimal pain management. Misapplied, however, the PDE could have the opposite effect, acting as an impediment to adequate symptom control.

Hypothesis 2: clinicians versus non-clinicians

We also hypothesised that clinicians would be less likely to believe that opioids hasten death in typical EOL situations, given their bedside experience with dying patients. While slightly fewer clinicians held this belief (compared with non-clinicians), the difference only approached statistical significance. It is possible that some clinician/ethicists might, over time, come to focus more on ethics than on clinical medicine, and thus may subscribe to exaggerated risks of opioids that have since been called into question. Alternatively, these clinician/ethicists may feel that even a small risk of hastened death in certain EOL situations requires moral justification for aggressive pain management, at least for those students who are deeply concerned about their own complicity (as noted by Sulmasy and Pellegrino above).

Of course, it is also possible that this study was not sufficiently powered to detect a statistically significant difference between clinicians and non-clinicians, resulting in a type 2 error.


Other limitations of the study include the fact that slightly less than half of US medical schools were represented, with approximately one-sixth of responses coming from one medical school. Most of the schools had only one respondent, thus raising the possibility that the respondent was not aware of all the ways the PDE is taught at that institution (a suspicion heightened by the inconsistent reports from the 10 respondents at one institution). An individual response also might not adequately represent a diversity of opinion among faculty at that institution. The discrepancy between the acknowledged relevance of the PDE in theoretical (71%) and practical (48%) situations highlights the need for more nuanced exploration of these distinctions.


Special thanks to Casey Johnson, MD, for his assistance in conceptualising and instituting the study; to Bob Orr, MD, for his helpful comments on the manuscript; to Alan Howard, PhD, for his assistance in data analysis; and to Gordon Meyer for his technical and logistical support.



  • Competing interests None.

  • Ethics approval Ethics approval was provided by University of Vermont Institutional Review Board.

  • Provenance and peer review Not commissioned; externally peer reviewed.

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

Other content recommended for you