[Skip to Content]
Sign In
Individual Sign In
Create an Account
Institutional Sign In
OpenAthens Shibboleth
[Skip to Content Landing]
Policy Perspectives
August 12, 1998

The Practice of Euthanasia and Physician-Assisted Suicide in the United StatesAdherence to Proposed Safeguards and Effects on Physicians

Author Affiliations

From the Center for Outcomes and Policy Research, Division of Cancer Epidemiology and Control, Dana-Farber Cancer Institute, Boston, Mass (Dr Emanuel and Ms Daniels); Center for Research Methodology and Biometrics, Cancer Research Center, American Medical Center, Denver, Colo (Dr Fairclough); and the Center for Survey Research, University of Massachusetts, Boston (Dr Clarridge).

JAMA. 1998;280(6):507-513. doi:10.1001/jama.280.6.507

Context.— Despite intense debates about legalization, there are few data examining the details of actual euthanasia and physician-assisted suicide (PAS) cases in the United States.

Objective.— To determine whether the practices of euthanasia and PAS are consistent with proposed safeguards and the effect on physicians of having performed euthanasia or PAS.

Design.— Structured in-depth telephone interviews.

Setting and Participants.— Randomly selected oncologists in the United States.

Outcome Measures.— Adherence to primary and secondary safeguards for the practice of euthanasia and PAS; regret, comfort, and fear of prosecution from performing euthanasia or PAS.

Results.— A total of 355 oncologists (72.6% response rate) were interviewed on euthanasia and PAS. On 2 screening questions, 56 oncologists (15.8%) reported participating in euthanasia or PAS; 53 oncologists (94.6% response rate) participated in in-depth interviews. Thirty-eight of 53 oncologists described clearly defined cases of euthanasia or PAS. Twenty-three patients (60.5%) both initiated and repeated their request for euthanasia or PAS, but 6 patients (15.8%) did not participate in the decision for euthanasia or PAS. Thirty-seven patients (97.4%) were experiencing unremitting pain or such poor physical functioning they could not perform self-care. Physicians sought consultation in 15 cases (39.5%). Overall, oncologists adhered to all 3 main safeguards in 13 cases (34.2%): (1) having the patient initiate and repeat the request for euthanasia or PAS, (2) ensuring the patient was experiencing extreme physical pain or suffering, and (3) consulting with a colleague. Those who adhered to the safeguards had known their patients longer and tended to be more religious. In 28 cases (73.7%), the family supported the decision. In all cases of pain, patients were receiving narcotic analgesia. Fifteen patients (39.5%) were enrolled in a hospice. While 19 oncologists (52.6%) received comfort from having helped a patient with euthanasia or PAS, 9 (23.7%) regretted having performed euthanasia or PAS, and 15 (39.5%) feared prosecution.

Conclusions.— Intractable pain or poor physical functioning seem to be nearly absolute requirements for physicians to perform euthanasia or PAS. Only one third of cases are performed consistently with proposed safeguards. For some patients, end-of-life care that includes opioid analgesia and hospice care does not obviate their desire for euthanasia or PAS. While the majority of physicians seem comforted by their actions, some experience adverse consequences from having performed euthanasia or PAS.

ONCE AGAIN, euthanasia and physician-assisted suicide (PAS) have become controversial social issues,15 prompting substantial empirical research. There have been numerous surveys of physicians in Australia, Britain, Canada, the Netherlands, the United States, and other countries.1,619 These surveys have tended to focus on attitudes toward euthanasia and PAS and how frequently these interventions have been performed. Few studies have elucidated the actual details of the practices of euthanasia and PAS. The largest of such studies have been conducted in the Netherlands.9,11 We have identified only 2 studies in the United States examining the details of actual euthanasia or PAS cases.16,18

To provide additional empirical data on the actual practices of euthanasia and PAS in the United States, we conducted in-depth interviews with US oncologists who, on screening questions, acknowledged performing these interventions. Using a method modeled on the Remmelink study in the Netherlands,9,11 the in-depth interviews focused on whether practice is consistent with proposed safeguards and the effects on the physicians themselves of performing these interventions.

We identified 4 proposed primary safeguards that have usually been required in the Netherlands, incorporated into US proposals on PAS, and endorsed by advocates of euthanasia and PAS1,2025; they include the following: (1) the patient must be terminally ill; (2) the patient must be competent and initiate and repeatedly request euthanasia or PAS; (3) the patient must be experiencing severe pain and/or suffering; and (4) the patient must be evaluated by another physician, who may be a psychiatrist. Several secondary safeguards have been suggested but usually not as requirements, as follows: (1) optimal pain and palliative care services have been provided to the patient; (2) the physician and patient have a long-standing relationship; and (3) the patient's family has been informed and supports the decision.

We also examined the effects on physicians of performing euthanasia or PAS. It has been reported that among Dutch "physicians who had practiced euthanasia [many] mentioned that they would be most reluctant to do so again."9 And in a television program reporting a euthanasia case, the Dutch physician who performed euthanasia noted that:

To kill someone is something very far reaching and that is something that nags at your conscience. . . . I wonder what it would be like not to have these cases in my practice. Perhaps I would be a much more cheerful person.

We focused on oncologists because the Dutch data demonstrate that 80% of patients receiving euthanasia have cancer.9,11 In addition, proposals to legalize euthanasia or PAS in the United States typically require the patients to have 6 months or less to live, making oncology patients the largest group of patients for whom euthanasia and PAS would be an option if legalized.2325 Therefore, in the United States, oncologists are most likely to administer the interventions.

Participants, Eligibility Requirements, and Response Rates

Participant selection has previously been described.26 In brief, no participants were paid to participate. We selected 2 samples: (1) all adult medical, gynecological, and surgical oncologists from Connecticut, Maine, Massachusetts, New Hampshire, Rhode Island, and Vermont and (2) from all 44 non-New England states and the District of Columbia, we randomly selected 10% of medical oncologists listed in the 1993 American Society of Clinical Oncology Directory.26 Of the 498 oncologists selected, 3 had died, 5 were retired, and 1 was hospitalized. Of the 489 eligible oncologists, 115 refused to participate, 7 could not be traced, and 12 could not be interviewed before the study ended. We completed 355 interviews for a response rate of 72.6%.

All 355 oncologists who participated in the initial interview were contacted 6 to 12 months later for a follow-up interview. In both the initial and follow-up interviews, oncologists were asked 2 screening questions: "Have you ever actually injected drugs to intentionally end a patient's life?" and "Have you ever actually prescribed drugs to a patient knowing the patient intended to use them to end his or her life?" Oncologists who answered yes to either of these questions were given an in-depth interview. Except for being conducted on the telephone, the in-depth interviews were modeled on the methods of the Remmelink report, asking oncologists about their most recent case of euthanasia or PAS.9,11

Survey Instrument Development, Pretesting, and Administration

Survey development occurred in 6 steps: (1) literature search, (2) focus groups, (3) instrument creation, (4) cognitive pretesting, (5) behavioral pretesting, and (6) reliability assessment. The extensive pretesting ensured content and construct validity, test-retest reliability, and comprehensibility. Because the terms euthanasia and PAS can be ambiguous and emotionally charged, they were replaced by the carefully worded and pretested phrases described above.1,26 Specially trained interviewers from the Center for Survey Research, University of Massachusetts, Boston, conducted all interviews on the telephone.

Data Confidentiality

Oncologists were informed that their responses would be kept confidential and that personal identifiers linked to the survey would be destroyed prior to any publication. All files with participants' names, telephone numbers, and survey identification numbers have been destroyed.

Data Analysis

As previously reported, there were no differences between New England and national oncologists on 98% of the survey questions, so responses were combined for analysis.26 The characteristics of oncologists who participated or never reported participating in euthanasia or PAS were compared using analysis of variance for age; the Cochran Mantel-Haenszel χ2 test for ordered categorical variables, level of religious belief, and number of requests for euthanasia or PAS in the last year; and the χ2 test of independence for unordered categorical variables, sex, religious affiliation, race, and ever having been asked to perform euthanasia or PAS. Candidate-independent variables that were used in multivariate logistic regression analyses of the use of primary safeguards, regret, fear of prosecution, and other outcomes included oncologist age, sex, race, religion, level of religious beliefs, length of physician-patient relationship, consistency with primary safeguards, whether the patient was experiencing pain or receiving opioid analgesia, and whether the patient's family supported a decision for euthanasia or PAS.

Human Subjects Approval

The institutional review boards at the Dana-Farber Cancer Institute and the University of Massachusetts approved the study.


Of the 355 participating oncologists, 56 (15.8%) reported ever participating in either euthanasia or PAS on the screening questions (Figure 1).26 Of these 56 oncologists, only 3 refused to participate in the in-depth interviews (94.6% response rate). Three indicated that the event occurred so long ago that they could not recall the details of the case to answer the in-depth questions. During the interviews, 7 oncologists' answers indicated that they did not perform euthanasia or PAS but either terminated life-sustaining treatments or increased narcotics for pain relief with the unintended result of hastening a patient's death. Also, 5 cases were situations in which patients overdosed on medications they previously possessed without asking their physician for information about ending their lives or requesting a prescription to end their lives. Overall, the in-depth interview revealed that for 12 (21.4%) of 56 oncologists who indicated that they had "intentionally ended a patient's life," the actual cases were not consistent with established definitions of euthanasia or PAS.1 Therefore, all subsequent analyses are based on the in-depth interviews of the 38 oncologists (10.7% of 355 interviewed oncologists) in which the description of cases is clearly consistent with accepted definitions of euthanasia or PAS. There were no significant differences regarding age, sex, race, religion, or geography between oncologists who reported performing euthanasia and PAS and those who did not (Table 1). However, oncologists who performed euthanasia or PAS were significantly less likely to report themselves as religious and more likely to receive requests for these interventions.26

Image description not available.
Study overview. PAS indicates physician-assisted suicide.
Table 1.—Characteristics of Oncologists Who Have Performed Euthanasia or Physician-Assisted Suicide (PAS)
Table 1.—Characteristics of Oncologists Who Have Performed Euthanasia or Physician-Assisted Suicide (PAS)
Image description not available.
The Practice of Euthanasia and PAS

Of the 38 oncologists, 28 (73.7%) reported that they "prescribed drugs . . . knowing the patient intended to use" them for suicide (PAS), 2 (5.3%) claimed they "injected drugs . . . to intentionally end a patient's life" (euthanasia), and 8 (21.1%) reported doing both. However, the details provided in the in-depth interviews indicated that in 7 cases (18.4%), the intervention that was identified by the oncologist as PAS was more accurately characterized as euthanasia. Misclassified cases usually occurred because oncologists did not actually inject medications to end the patient's life, but wrote an order for a nurse to inject intravenous medications to end the patient's life. Therefore, of the 38 cases, 17 (44.7%) appeared to be euthanasia, 20 (52.6%) were PAS, and 1 (2.6%) case remained ambiguous.

Many oncologists indicated that they were more comfortable playing a "passive" rather than "active" role in ending patients' lives, believing that writing a prescription and letting the patient decide constitutes a passive role, as indicated in the following responses: "I philosophically find it difficult to take an active role. I'm willing to take a passive role. I think injecting drugs is active. Allowing [a patient the] means to end [his or her] own life is passive. I am comfortable with that." "[The] patient requested narcotics and asked if she took more than the prescribed amount would she die. I told her yes . . . I will help control and treat pain, but I'm not going to actually take any action [to end patients' lives]."

Among the patients who received euthanasia or PAS, 60.5% were women. The dominant diagnoses were lung cancer (8 [21.1%]), breast cancer (5 [13.2%]), prostate cancer (4 [10.5%]), leukemia or lymphoma (4 [10.5%]), and human immunodeficiency virus infection (3 [7.9%]). In all cases, patients were deemed to be terminally ill, ie, having 6 months or less to live.

In 3 (15%) of the 20 PAS cases, patients explicitly asked for information about how to commit suicide; usually the patients possessed medications and asked the physicians about how much to take to end their lives. In 4 (20%) of the 20 PAS cases, the patients never used the medication for suicide. In an additional 15% of the PAS cases, the attempted suicide failed. By physicians' estimates, 50% of patients had their life shortened by 1 month or less.

In about 75% of cases, pain (29), dependency (28), or emotional burden on the patient's family (28) were reported by oncologists as the dominant motivating factors. Frequently (in 16 cases [42.1%]), all 3 factors were reported simultaneously. In 3 cases (7.9%), concerns about financial burdens played a key role in motivating euthanasia or PAS. In 1 case (2.6%), the family wanted to "get it over with."

In the vast majority of cases (78.9%), narcotics were used to end the patients' lives. In 7 cases (18.4%), barbiturates were used; in 2 cases (5.3%), benzodiazepines were used. Muscle relaxants and potassium chloride were each used in 1 case (2.6%). In 5 cases (13.2%), multiple medications were used.

Proposed Primary Safeguards

Table 2 presents the proportion of these 38 cases in which the practices were consistent with proposed primary safeguards. (Since all patients were terminally ill, this safeguard was not included.) In 30 cases (78.9%), patients initiated the request for euthanasia or PAS; in 23 cases (60.5%), patients both initiated and repeated their requests. Thus, in 7 cases, patients did not repeat their requests. In 5 cases (13.2%), the family and in 1 case (2.6%), the physician initiated the request for euthanasia or PAS. In 6 cases (15.8%), the patient neither initiated nor discussed the decision about euthanasia or PAS and in 2 cases (5.3%), the patient was unconscious at the time of euthanasia. In all but 1 case (97.4%), the patients were experiencing substantial physical symptoms. Oncologists often stated that pain and poor quality of life were necessary to justify committing euthanasia or PAS, as indicated in the following responses: "The pain, the quality of life, the duration of life are all factors [in the decision]." "The pain lasted hours on end."

Table 2.—Proposed Primary Safeguards for Performing Euthanasia or Physician-Assisted Suicide (PAS)
Table 2.—Proposed Primary Safeguards for Performing Euthanasia or Physician-Assisted Suicide (PAS)
Image description not available.

In 15 cases (39.5%), the oncologist discussed the request for euthanasia or PAS with another physician, although the patient was not necessarily examined by the other physician. In 2 cases (5.3%), the patient received a psychiatric evaluation. In 2 cases (5.3%), the patients refused oncologists' referrals for a psychiatric evaluation.

Overall, in 13 cases of euthanasia and PAS (34.2%), practices were consistent with all the proposed primary safeguards. In 23 cases (60.5%), patients initiated and repeated their request for euthanasia or PAS and also had severe physical symptoms. Multivariate logistic analysis revealed that oncologists' actions were significantly more likely to be consistent with primary safeguards the longer they had a relationship with the patient (P=.008). There was a trend that did not reach statistical significance, indicating that more religious oncologists were more likely to adhere to the primary safeguards (P=.06).

Proposed Secondary Safeguards

Table 3 presents the proportion of cases in which the practices were consistent with proposed secondary safeguards. All patients experiencing pain were being treated with opioid analgesia; in only 3 cases (7.9%) did this adequately control the pain. In addition, 15 patients (39.5%) were receiving hospice care at the time of euthanasia or PAS. The retrospective nature of the study precluded assessment of the quality of these palliative care interventions. In over 80% of cases, physicians had known the patients for whom they provided euthanasia or PAS for more than 6 months. For many physicians, having a long-term relationship with the patient was critical to their willingness to perform euthanasia or PAS: "There must be an on-going relationship with the patient and family." "I think you have to know the patient well." In 28 cases (73.7%), the family initiated the request or supported the decision for euthanasia or PAS. In 1 case (2.6%), the family was opposed to the decision; in an additional 10 cases (26.3%), the decision was not discussed with the family.

Table 3.—Proposed Secondary Safeguards for Performing Euthanasia or Physician-Assisted Suicide (PAS)
Table 3.—Proposed Secondary Safeguards for Performing Euthanasia or Physician-Assisted Suicide (PAS)
Image description not available.
Effect on Physicians

A majority of oncologists (54.0%) found comfort in knowing they "helped a patient end his or her life the way the [patient] wished" (Table 4). A quarter of oncologists (9) regretted performing euthanasia or PAS. While some of these oncologists (5) also feared prosecution, it is clear from the interviews that in all cases the regret resulted from concerns other than prosecution. One oncologist regretted his action because "a daughter had not said good-bye" to her mother. Another thought "the patient might have benefited from living to the end." Others expressed worries about "playing God a little bit too much" and a sense that ending a patient's life made physicians feel "conflicted, at odds with myself [and my role]." In none of these cases, did the form of death—euthanasia or PAS—predict the reason for regret. In none of the cases in which the physicians expressed regret was the patient unconscious.

Table 4.—Effect on Oncologists of Performing Euthanasia or Physician-Assisted Suicide (PAS)
Table 4.—Effect on Oncologists of Performing Euthanasia or Physician-Assisted Suicide (PAS)
Image description not available.

A third of oncologists (12) felt that the "emotional burden associated with [their euthanasia and PAS] decision . . . affected the way they practice medicine." Importantly, many (5) felt it made them listen to their patients with more sympathy: "It makes me more open to listen to subsequent patients." I am "much more aware of the patient's emotional and physical needs apart from their medical needs." Conversely, the others (6) said the emotional burden was adverse. For some it made them avoid situations that might create a request for euthanasia or PAS. One physician felt so "burned out" he moved from the city in which he was practicing to a small town. Another said, "I have the luxury of being in academic medicine which has allowed me to be exposed to these difficult situations relatively infrequently."

Nearly 40% (15) of oncologists who had performed euthanasia or PAS feared prosecution. There were no significant differences in age, sex, geography, religion, or level of religious belief, between those oncologists who feared prosecution and those who did not. A similar proportion of oncologists who feared prosecution had performed euthanasia (39.1%) as those who did not fear prosecution. This fear neither prevented the oncologists from performing these interventions nor from answering questions about their case with an interviewer; none reported legal proceedings.


This study provides additional empirical data on the actual performance of euthanasia and PAS by US physicians that challenge the claims advanced by both proponents and opponents of euthanasia and PAS. Seven implications seem important.

First, in as many as a third of responses, physicians may have been confused or may have misclassified or misrepresented their actions, suggesting that data on the performance of euthanasia and PAS not collected through in-depth interviews are unreliable. More than 12 (20%) of 56 oncologists, who in screening questions reported that they intentionally ended a patient's life, revealed through the in-depth interviews actions such as withholding life-sustaining treatments and providing narcotics for pain relief that shortened a patient's life, which are neither euthanasia nor PAS. Some of the confusion and misclassification could be a result of poor survey questions. However, our questions were carefully crafted and extensively pretested. The wording of our questions is similar to that used in most other careful surveys,1318 and are more precise than the questions used in some other surveys.1 More likely, physicians may be uncertain and confused as to whether increasing narcotics for pain control, withholding blood transfusions, or giving prescriptions for medications that could be used to overdose constitute "intentionally ending a patient's life." Regardless, these data suggest that existing surveys of US physicians and physicians from other countries may significantly overstate, by more than 20%, the actual practice of euthanasia and PAS.

Second, in at least 7 cases, physicians identified an action ending a patient's life as PAS, but in-depth interviews revealed it to be euthanasia. In many of the cases, a physician wrote an order leaving a hospice or hospital nurse or even a family member to perform the actual injecting of the life-ending medication. As survey data and comments during our in-depth interviews indicate, many more physicians find PAS more acceptable than euthanasia because it seems to connote a passive rather than an active life-ending role.1,1318,26 Thus, identifying their action as PAS rather than as euthanasia diminishes the importance of the physician's role, thereby decreasing physicians' psychological burden. Such data suggest existing studies may significantly understate the proportion of cases that are euthanasia.

While laws permitting PAS but prohibiting euthanasia may have appeal by providing a barrier to the "slippery slope," there may be serious practical problems.2325 Despite significant education through public debate, media attention, and many articles in both medical journals and the lay press, physicians and others may not accurately distinguish PAS from euthanasia. These data warn that if PAS is legalized, physicians may unintentionally and unknowingly commit many cases of euthanasia believing they are performing a legal act of PAS, thus creating potential enforcement problems. This is just a caution because with legalization it is conceivable that additional instructional efforts may successfully educate physicians about the differences between PAS and euthanasia.

A third important implication of these data is the inconsistent adherence to well-accepted safeguards. Extending the results of Back et al16 and Meier et al,18 we found that being severely ill—either having pain despite narcotic medications or extremely poor physical functioning—is nearly an absolute requirement for oncologists to perform euthanasia or PAS. Conversely, oncologists' practices were not as consistent with the proposed safeguards ensuring that the requests were initiated by the patient, voluntary, and well considered. Most worrisome is that in 15.3% of cases, the patients were not involved in the decision but families wanted the patients' lives ended. This lack of involvement even occurred in cases where the patients were conscious and could have participated in the decision. Meier et al18 also reported that in some cases family members made the request for euthanasia and that in other cases patients were confused more than half the time. Therefore, both the US and Dutch data suggest that despite safeguards, in 15% to 20% of cases, physicians are willing to provide euthanasia even without patient involvement in the decision when there are severe physical symptoms.911 The illegal status of euthanasia and PAS may itself prevent adherence to the primary safeguards and legalization may improve adherence to these safeguards. In the Netherlands, minimizing legal barriers has increased reporting of euthanasia cases, although it did not measurably affect euthanasia performed on patients with compromised mental capacity who could neither contemporaneously request nor consent to euthanasia.11 Again, these data warn that safeguards meant to ensure patient autonomy could be violated in a significant minority of cases making nonvoluntary or involuntary euthanasia a concomitant of permitting voluntary euthanasia or PAS.27

Like Back et al16 and Meier et al,18 we found a lack of consultation. Because euthanasia and PAS were illegal when the interviews were conducted, it is understandable that US physicians may be unwilling to consult a colleague about a request. Nevertheless, we found that 40% of physicians did discuss their cases with colleagues. More important, there was very little use of psychiatric evaluations despite the fact that a physician can refer a patient for psychiatric evaluation without revealing that a request for euthanasia or PAS prompted the evaluation. Contrary to available data, physicians may believe that it is "natural" or "normal" for terminally ill patients to be depressed and not take measures for evaluation and treatment of depression.2830 These data are worrisome because depression is one of the leading predictors of interest in euthanasia or PAS.16,26,31,32 While we do not know how many of these patients were depressed, in at least 1 case reported here, the physician felt the patient was depressed. The data suggest that laws like Oregon's, which permit PAS but do not mandate a psychiatric evaluation, may not lead to adequate detection and care of mental illness among patients requesting PAS.33 We also found that in 2 cases, patients refused to receive psychiatric care despite a physician's recommendation. Laws permitting euthanasia or PAS must consider how to respond to patient refusal of such evaluations.2325

A fourth implication of this study is that, contrary to claims by some opponents of euthanasia and PAS, use of palliative care is not necessarily an effective response to patients' request for euthanasia and PAS. Of the patients who received these interventions, all patients experiencing pain received opioid analgesia and almost 40% were enrolled in hospice programs. Because of the retrospective nature of our interviews, we could not evaluate the quality of the pain care or hospice. Nevertheless, at least use of these palliative measures did not necessarily obviate desire for euthanasia or PAS.

A fifth implication is that there will be failures of PAS. We found that in 15% of PAS attempts the patient failed to die. Such failures of PAS may occur because physicians do not know what medications to prescribe to end patients' lives17 or because patients do not take the life-ending medications appropriately. Indeed, one problem may be the reported use of morphine for PAS, which is reported in the Netherlands not to be the optimal agent. If PAS is legalized, educational efforts on the appropriate drugs to use may be necessary. In addition, how to handle these failed PAS attempts is problematic. In the Netherlands, the ethos is that physicians should be within close proximity and be willing to perform euthanasia if the PAS attempt fails.11,20 This approach is unlikely to work in the United States. First, it appears that if any life-ending intervention will be legalized it will permit PAS only.2,2325 Thus, euthanasia to remedy a failed PAS case is likely to remain illegal. Further, our data and other surveys suggest that many physicians who are willing to perform PAS are often unwilling to perform euthanasia.1,1318,26 Physicians may find being near and willing to perform euthanasia if the PAS attempt fails too active. Legalizing PAS but not euthanasia could create serious difficulties in addressing the failed PAS attempt.

Sixth, like Back et al,16 we found that some patients did not use a prescription for PAS, suggesting that the prescription did serve as "psychological insurance"34; having the control afforded by the prescription without ending one's life was sufficient for some patients. Our data suggest that this occurs in 20% of PAS cases, whereas Back et al16 reported 39% of patients did not use a prescription for PAS. The differences may just be random variation because of the small numbers of interviews and cases of PAS, or it may be because of the different survey methods (written vs in-depth interviews) used and the higher response rates in this study.

Finally, like Meier et al,18 we found that the majority of oncologists would perform euthanasia and PAS again in a similar case and received comfort from having helped a patient. Nevertheless, a significant minority of oncologists experience substantial problems from their actions. In almost a quarter of cases, physicians regretted performing euthanasia or PAS after the incident and in almost 1 in 6 cases, physicians experienced emotional distress. Importantly, distress and regret did not seem linked with either fear of prosecution or performing euthanasia. Given that euthanasia and PAS are irreversible acts, such a high frequency of regret should be a cause of concern. In many cases, physicians who regretted the decision consulted colleagues, and in other cases the regret resulted from effects on the family that might not have been anticipated. In at least a few cases, the emotional distress seemed significant. One oncologist went so far as to move to another city and others changed their practices to avoid caring for patients who might request euthanasia or PAS.


This study has several limitations. The number of physicians interviewed who performed euthanasia or PAS is small, meaning that proportions of responses have wide confidence intervals, the conclusions must be provisional and tentative, and it is difficult to identify predictors of actions. However, because of the low rate at which physicians perform euthanasia and PAS and the costs of conducting in-depth interviews, studies on these practices will inherently involve relatively small numbers of cases. Indeed, the other 2 studies of these practices in the United States reported data from written responses of 52 and 81 physicians without follow-up in-depth interviews16,18 and the well-regarded Dutch studies report interviews of 215 physicians who actually performed euthanasia or PAS.9,11

A second limitation is that these data may not generalize to nononcologists. Some physicians treating patients with the acquired immunodeficiency syndrome (AIDS) report performing euthanasia or PAS at higher rates.19 And many of the patients who have been assisted by Dr Jack Kevorkian have had neurological diseases, such as amyotrophic lateral sclerosis. However, cancer causes more than 10 times as many deaths in the United States as AIDS and more than 20 times as many deaths as neurological diseases. More important, the Dutch data suggest that 80% of euthanasia or PAS patients will have cancer, while less than 5% will be patients with neurological diseases. Consistent with this, nononcology specialists report performing euthanasia and PAS much less frequently than oncologists.16,18

A third limitation is that despite reassurances of confidentiality, physicians might be inhibited from discussing these illegal actions. The high response rates in the initial interviews (72.6%) and follow-up interviews (94.6%) make this seem unlikely. Finally, while the safeguards we identified are accepted by ethicists and those actively engaged in the debate about euthanasia and PAS,1,2325 we are applying the standards after the fact. These safeguards are not legally mandated and what precisely is required may be unclear to practicing physicians. If euthanasia and/or PAS were legalized, the practice environment might change and physicians' awareness of and adherence to safeguards could be substantially different. Thus, these data can only warn about areas that might be problematic if interventions are legalized.


This study provides additional data on the practice of euthanasia and PAS in the United States. It suggests that (1) even with well-crafted questions, existing data may overstate the practices of euthanasia and PAS; (2) physicians frequently confuse euthanasia with PAS; (3) practices are frequently inconsistent with proposed safeguards, although having severe physical symptoms seems a strong requirement for physicians to perform euthanasia or PAS; (4) for some patients, receiving opioid analgesia and/or hospice care does not obviate the desire for euthanasia or PAS; (5) there are some unsuccessful attempts at PAS; and (6) a significant minority of physicians experience regret and emotional distress at having performed euthanasia or PAS.

Emanuel EJ. Euthanasia: historical, ethical, and empiric perspectives.  Arch Intern Med.1994;154:1890-1901.
 Vacco et al v Quill et al,  117 S Ct 2293 (1997).
Quill TE. Death and Dignity: Making Choices and Taking Control . New York, NY: WW Norton & Co Inc; 1993.
Brody H. Assisted suicide: a challenge for family physicians.  J Fam Pract.1993;37:123-125.
Eddy DM. A conversation with my mother.  JAMA.1994;272:179-181.
Stevens CA, Hassan R. Management of death, dying and euthanasia: attitudes and practices of medical practitioners in South Australia.  J Med Ethics.1994;20:41-46.
Ward BJ, Tate PA. Attitudes among NHS doctors to requests for euthanasia.  BMJ.1994;308:1332-1334.
Kinsella TD, Verhoef MJ. Alberta euthanasia survey, 1: physicians' opinions about the morality and legalization of active euthanasia.  CMAJ.1993;148:1921-1933.
van der Maas PJ, van Delden JJM, Pijnenborg L, Looman CWN. Euthanasia and other medical decisions concerning the end of life.  Lancet.1991;338:669-674.
Pijnenborg L, van Delden JJM, Kardaun JWPF, Glerum JJ, van der Maas PJ. Nationwide study of decision concerning the end of life in general practice in the Netherlands.  BMJ.1994;309:1209-1212.
van der Maas PJ, van der Wal G, Haverkate I.  et al.  Euthanasia, physician-assisted suicide, and other medical practices involving the end of life in the Netherlands, 1990-1995.  N Engl J Med.1996;335:1699-1705.
Folker AP, Holtug N, Jensen B, Kappel K, Nielsen JK, Norup M. Experiences and attitudes towards end of life decisions amongst Danish physicians.  Bioethics.1996;10:233-249.
Fried TR, Stein MD, O'Sullivan PS, Brock DW, Novack DH. The limits of patient autonomy: physician attitudes and practices regarding life-sustaining treatments and euthanasia.  Arch Intern Med.1993;153:722-728.
Shapiro RS, Derse AR, Gottlieb M, Schiedermayer D, Olson M. Willingness to perform euthanasia: a survey of physician attitudes.  Arch Intern Med.1994;154:575-584.
Cohen JS, Fihn SD, Boyko EJ, Jonsen AR, Wood RW. Attitudes toward assisted suicide and euthanasia among physicians in Washington State.  N Engl J Med.1994;331:89-94.
Back AL, Wallace JI, Starks HE, Pearlman RA. Physician-assisted suicide and euthanasia in Washington State.  JAMA.1996;275:919-925.
Lee MA, Nelson HD, Tilden VP, Ganzini L, Schmidt TA, Tolle SW. Legalizing assisted suicide: view of physicians in Oregon.  N Engl J Med.1996;334:310-315.
Meier DE, Emmons CA, Wallenstein S, Quill T, Morrison RS, Cassel CK. A national survey of physician-assisted suicide and euthanasia in the United States.  N Engl J Med.1998;338:1193-1201.
Slome LR, Mitchell TF, Charlebois E, Benevedes JM, Abrams DI. Physician-assisted suicide and patients with human immunodeficiency virus disease.  N Engl J Med.1997;336:417-421.
de Wachter MAM. Active euthanasia in the Netherlands.  JAMA.1989;262:3316-3319.
Gevers JKM. Legislation on euthanasia: recent developments in the Netherlands.  J Med Ethics.1992;18:138-141.
Ryan CJ, Kaye M. Euthanasia in Australia: the Northern Territory Rights of the Terminally Ill Act.  N Engl J Med.1996;334:326-328.
Quill TE, Cassel CK, Meier DE. Care of the hopelessly ill: proposed clinical criteria for physician-assisted suicide.  N Engl J Med.1992;327:1380-1384.
Miller FG, Quill TE, Brody H, Fletcher JC, Gostin LO, Meier DE. Regulating physician-assisted death.  N Engl J Med.1994;331:119-123.
Baron CH, Bergstresser C, Brock DW.  et al.  Statute: a model state act to authorize and regulate physician-assisted suicide.  Harv J Legislation.1996;33:1-34.
Emanuel EJ, Fairclough DL, Daniels ER, Clarridge BR. Euthanasia and physician-assisted suicide: attitudes and experiences among oncology patients, oncologists, and the general public.  Lancet.1996;347:1805-1810.
Callahan D, White M. The legalization of physician-assisted suicide: creating a regulatory Potemkin village.  Univ Richmond Law Rev.1996;30:1-83.
Derogatis LR, Morrow GR, Fetting J.  et al.  The prevalence of psychiatric disorders among cancer patients.  JAMA.1983;249:751-757.
Portenoy R, Thaler HT, Kornblith AB.  et al.  Symptom prevalence, characteristics and distress in a cancer population.  Qual Life Res.1994;3:183-189.
Allebeck P, Bolund C, Ringback C. Increased suicide rate in cancer patients: a cohort study based on the Swedish Cancer-Environment Register.  J Clin Epidemiol.1989;42:611-616.
Breitbart W, Rosenfeld BD, Passik SD. Interest in physician-assisted suicide among ambulatory HIV-infected patients.  Am J Psychiatry.1996;153:238-242.
Emanuel E. Whose right to die?  The Atlantic.1997;279(3):73-79.
Emanuel EJ, Daniels E. Oregon's physician-assisted suicide law: provisions and problems.  Arch Intern Med.1996;156:825-829.
Brock DW. Voluntary active euthanasia.  Hastings Cent Rep.1992;22:10-22.