Views of United States Physicians and Members of the American Medical Association House of Delegates on Physician-assisted Suicide

Abstract

OBJECTIVE

To ascertain the views of physicians and physician leaders toward the legalization of physician-assisted suicide.

DESIGN

Confidential mail questionnaire.

PARTICIPANTS

A nationwide random sample of physicians of all ages and specialties, and all members of the American Medical Association (AMA) House of Delegates as of April 1996.

MEASUREMENTS

Demographic and practice characteristics and attitude toward legalization of physician-assisted suicide.

MAIN RESULTS

Usable questionnaires were returned by 658 of 930 eligible physicians in the nationwide random sample (71%) and 315 of 390 eligible physicians in the House of Delegates (81%). In the nationwide random sample, 44.5% favored legalization (16.4% definitely and 28.1% probably), 33.9% opposed legalization (20.4% definitely and 13.5% probably), and 22% were unsure. Opposition to legalization was strongly associated with self-defined politically conservative beliefs, religious affiliation, and the importance of religion to the respondent (P < .001). Among members of the AMA House of Delegates, 23.5% favored legalization (7.3% definitely and 16.2% probably), 61.6% opposed legalization (43.5% definitely and 18.1% probably), and 15% were unsure; their views differed significantly from those of the nationwide random sample (P < .001). Given the choice, a majority of both groups would prefer no law at all, with physician-assisted suicide being neither legal nor illegal.

CONCLUSIONS

Members of the AMA House of Delegates strongly oppose physician-assisted suicide, but rank-and-file physicians show no consensus either for or against its legalization. Although the debate is sometimes adversarial, most physicians in the United States are uncertain or endorse moderate views on assisted suicide.

Physician-assisted suicide has been a topic of active debate for more than a decade. Advocates achieved a breakthrough in Oregon, where the practice is now legal and eligible patients receive open assistance from physicians in ending their lives. Opponents have scored key victories of their own, winning a decision by the Supreme Court that the terminally ill have no constitutional right to physician-assisted suicide and defeating initiatives in Washington, California, and Michigan.

The leadership of the American Medical Association (AMA) has been outspoken and influential on this topic. The AMA Council on Ethical and Judicial Affairs has stated that participation in physician-assisted suicide is “fundamentally incompatible with the physician's role as healer.”¹ The Council's opinion is vigorously supported by the AMA's trustees and House of Delegates and has been cited by the Supreme Court.² The AMA's strong position might suggest that it speaks for a united profession, but there has never been a nationwide study of the opinions of physicians of all specialties toward physician-assisted suicide. Three excellent state-specific surveys^3–5 showed a majority of physicians in those states favored the legalization of physician-assisted suicide under some circumstances, but it is not known if these states are representative of the country as a whole. In contrast, a survey of oncologists found that a majority do not find physician-assisted suicide acceptable, even for patients with unremitting pain.⁶

We conducted a nationwide survey of attitudes toward the legalization of physician-assisted suicide among physicians in the United States. We included physicians of all specialties to ensure an inclusive sample and because almost all physicians help patients and their families make medical decisions of lasting importance. We also surveyed all members of the AMA House of Delegates. Our objectives were to measure the attitudes of physician in the United States toward physician-assisted suicide, to compare their views with the views of members of the AMA House of Delegates, and to evaluate the relationship between personal characteristics and attitudes toward physician-assisted suicide.

METHODS

Questionnaire Development

The first draft of the questionnaire drew on questions used in other studies.^3–5,7 Authors and leaders of organizations that have been active in the debate over physician-assisted suicide were invited to participate in questionnaire development. These consultants included proponents and opponents of physician-assisted suicide and individuals with no identified position. As the questionnaire evolved, pilot versions were sent to 3 physician populations. The final questionnaire included 6 questions about physician-assisted suicide, 3 questions about the involvement of the AMA in the policy debate, and 13 questions about respondents' demographic and practice characteristics. All of the attitudinal questions were closed-ended (i.e., multiple choice). A disclaimer—“This confidential survey is not affiliated with the American Medical Association”—was placed prominently on the questionnaire. The Stanford University Institutional Review Board approved this study and the consent information provided to the subjects. Copies of the questionnaire are available on request.

Previous studies employed a variety of definitions of physician-assisted suicide. For example, Meier et al.⁷ defined physician-assisted suicide as “the practice of providing a competent patient with a prescription for medication for the patient to use with the primary intention of ending his or her own life.” We chose this definition in response to concerns by some members of our consultant panel that a more restrictive definition—for example, one that required that the patient be terminally ill and experiencing unrelenting suffering—might be difficult to define and ineffective in preventing abuses.

Selection of the Samples

The nationwide physician sample, designated “U.S. physicians,” was drawn from the AMA Masterfile, which is the most comprehensive list of physicians in the United States. It includes both members and nonmembers of the AMA, and licensed allopathic and osteopathic physicians of all ages and all specialties, whether in practice or in training, and whether in clinical, administrative, or other positions. One thousand names were randomly selected from the list, geographically stratified by state. Selection was made proportional to the number of physicians in each state. Every active physician in the United States was eligible to be part of the first sample.

The second group, members of the AMA House of Delegates, was obtained from the AMA Official Call⁸ and included every person who was a delegate as of April 1996 and who had not retired by the time the survey was conducted. A comparison of the 2 lists showed no overlap.

Questionnaire Distribution and Collection

Both groups of physicians were sent the questionnaire, a cover letter with consent information, and a postage-paid return envelope on February 18, 1997. Subjects who did not respond to the first mailing were sent a second questionnaire a month later. Persistent nonrespondents received a third questionnaire and then a fourth. As an incentive to complete the questionnaire, a check for $10 was enclosed with the fourth questionnaire. The cutoff date for responses was July 17, 1997. Double data entry with 100% verification was used.

The primary dependent variable for most of the analyses was the respondent's attitude toward the legalization of physician-assisted suicide, which was measured on a 5-point Likert scale. Differences in the dependent variable were first considered for respondents who received the cash incentive and then for the independent variables. Nominal independent variables included gender, specialty, AMA membership, geographical region of practice, ethnicity, involvement in direct patient care, completion of training, and religious affiliation. Age was a continuous variable. The importance of religion was measured on a 3-point scale, with 1 indicating that religion was “not at all important” and 3 indicating that it was “very important” to the respondent. Political conservatism was also measured on a 3-point scale (“liberal,”“moderate,” and “conservative”). Frequency of caring for terminally ill patients was measured on a 4-point scale (“never” to “regularly”).

Critical comparisons of attitudes toward the legalization of physician-assisted suicide were made separately for the U.S. physicians and the AMA House of Delegates. The Wilcoxon-Mann-Whitney test (for comparisons of 2 groups) and the Kruskal-Wallis 1-way analysis of variance (for comparisons of 3 or more groups) were used for these analyses.

In adjusted analyses, the 5-point Likert responses for attitude toward the legalization of physician-assisted suicide were dichotomized for the regressions; respondents who believed that physician-assisted suicide should definitely or probably be illegal were combined as “opposed” and all other respondents combined as “not opposed.” Logistic regression was then used to examine the impact of physician characteristics on opposition to physician-assisted suicide (subjects who had missing values on any variable were dropped from these analyses). For all analyses, significance levels were 2-tailed. The analyses were done with SPSS for Windows version 10.0.5 (SPSS Inc., Chicago, Ill).

Response Rate

U.S. Physicians Sample.

Of the 1,000 randomly selected physicians, 70 were ineligible for the survey: 37 questionnaires were returned because of incorrect addresses, 1 physician was out of the country, and 32 questionnaires were returned by physicians who had retired and were therefore ineligible as defined at the start of the study. Of the 930 eligible physicians, 658 (71%) returned completed questionnaires. Respondents were compared with nonrespondents in terms of gender, specialty, and geographic region for both the U.S. physicians sample and the House of Delegates sample. No statistically significant differences were found. There was also no significant relationship between receipt of the cash incentive and attitude toward the legalization of assisted suicide (P = .262). The demographic and practice characteristics of the respondents are shown in Table 1.

Table 1.

Respondent Characteristics^*

	U.S. Physicians	AMA House of Delegates
No. of respondents	658	315
Mean age, y±SD	44±11	59±9
Gender
Male	519 (79)	284 (91)
Female	138 (21)	28 (9)
Specialty
Anesthesiology	31 (5)	8 (3)
Family and general practice	101 (15)	42 (13)
Internal medicine	133 (20)	72 (23)
Obstetrics and gynecology	47 (7)	9 (3)
Pathology	21 (3)	12 (4)
Pediatrics	79 (12)	11 (4)
Psychiatry	43 (7)	9 (3)
Radiology	35 (5)	20 (6)
Surgery	113 (17)	110 (35)
Other	52 (8)	21 (7)
Still in training	101 (15)	9 (3)
Provide direct care	625 (94)	293 (83)
Care for terminally ill patients
Never	104 (16)	61 (20)
Rarely	200 (31)	70 (23)
Sometimes	204 (31)	89 (29)
Regularly	146 (22)	89 (29)
Region
Midwest	137 (21)	84 (27)
Northeast	157 (24)	61 (20)
South	216 (33)	106 (34)
West	139 (21)	59 (19)
Ethnicity
Asian or Pacific Islander	83 (13)	6 (2)
African American	20 (3)	2 (1)
Hispanic	29 (5)	4 (1)
Native American, American Indian or Alaskan	2 (0.3)	1 (0.3)
White	482 (76)	288 (95)
Other	20 (3)	3 (1)
Political self-identification
Conservative	213 (35)	132 (43)
Moderate	303 (49)	143 (46)
Liberal	102 (17)	33 (11)
Religious affiliation
Jewish	107 (18)	42 (14)
Catholic	159 (26)	70 (23)
Protestant	245 (40)	151 (50)
Other	95 (16)	39 (13)
None	4 (1)	1 (0.3)
Importance of religion to respondent
Not at all important	115 (19)	43 (14)
Moderately important	263 (43)	143 (47)
Very important	239 (39)	118 (39)

^*Values expressed number (percent), except where otherwise indicated. The percentages represent those physicians who answered a particular question and may not sum to 100 because of rounding. AMA indicates American Medical Association.

American Medical Association House of Delegates Sample.

There were 430 positions in the House of Delegates in April 1996. At the time of our survey, 4 of those positions were vacant, 2 physicians who were members in April 1996 had since died, 27 had retired, and no valid address could be obtained for an additional 7. Of the 390 eligible delegates, we received completed questionnaires from 315 (81%).

RESULTS

Attitudes of U.S. Physicians

Among the nationwide random sample of physicians, 45% believed that physician-assisted suicide should definitely or probably be legal, 34% believed that it should definitely or probably be illegal, and 22% were uncertain (Table 2). This nationwide sample was studied to ascertain the attitudes of subgroups and the relationship between personal characteristics and opinions about the legalization of physician-assisted suicide. Only political orientation and religion showed strong and consistent relationships to attitudes. Physicians who identified themselves as politically conservative were far more likely to oppose physician-assisted suicide than those who were politically liberal (47% vs 19%, P < .001). This finding was robust after adjustment for possible confounding by all other independent variables (Table 3).

Table 2.

Attitudes Toward Legalization of Physician-Assisted Suicide^*

	U.S. Physicians†
	AMA Members (N = 277)	AMA Nonmembers§ (N = 367)	Overall (N = 658)	AMA House of Delegates†† (N = 315)
Should definitely be illegal	19.5	21.0	20.4	43.5
Should probably be illegal	13.7	13.1	13.5	18.1
Unsure	24.5	19.1	21.6	14.9
Should probably be legal	26.4	29.4	28.1	16.2
Should definitely be legal	15.9	17.4	16.4	7.3

^*Values expressed as percent of those responding and may not total 100 because of rounding. Responses given were to the question, “The following questions concern assisted suicide, i.e., the practice of providing a competent patient with a prescription for medication for the patient to use with the primary intention of ending his or her own life. How do you feel about the legal status of physician-assisted suicide?”

^†Fourteen respondents were not sure of their AMA membership status.

^‡Views of U.S. Physicians sample and American Medical Association (AMA) House of Delegates were significantly different (Wilcoxon-Mann-Whitney P < .001).

^§Views of AMA members and nonmembers in the U.S. Physicians sample were not significantly different (Wilcoxon-Mann-Whitney P = .625).

Table 3.

Factors Associated with Opposition to Physician-Assisted Suicide (N = 541^*

	Odds Ratio (Confidence Interval)	P Value
Age, y
20–29	1.0 (reference)
30–39	1.3 (0.4 to 3.6)	.665
40–49	2.1 (0.7 to 6.8)	.202
50–59	2.3 (0.7 to 7.8)	.170
60+	2.2 (0.6 to 8.0)	.220
Gender
Male	1.0 (reference)
Female	1.1 (0.6 to 1.9)	.763
Specialty
Internal medicine	1.0 (reference)
Anesthesiology	0.5 (0.2 to 1.5)	.219
Family and general practice	0.8 (0.4 to 1.6)	.494
Surgery	0.6 (0.3 to 1.1)	.104
Obstetrics/gynecology	0.5 (0.2 to 1.3)	.167
Pathology	0.4 (0.1 to 2.0)	.262
Pediatrics	0.8 (0.3 to 1.7)	.505
Psychiatry	0.7 (0.3 to 2.1)	.575
Radiology	0.3 (0.1 to 1.0)	.044
Other	0.6 (0.3 to 1.5)	.291
Training status
No longer in training	1.0 (reference)
Still in training	2.0 (0.9 to 4.1)	.070
Provide direct care
Yes	1.0 (reference)
No	1.4 (0.4 to 4.9)	.638
Care for terminally ill patients
Never	1.0 (reference)
Rarely	0.8 (0.4 to 1.7)	.623
Sometimes	1.6 (0.8 to 3.5)	.200
Regularly	1.3 (0.6 to 3.0)	.467
Region
Midwest	1.0 (reference)
Northeast	1.2 (0.6 to 2.2)	.575
South	0.8 (0.4 to 1.4)	.399
West	1.0 (0.5 to 1.9)	.972
Ethnicity
White	1.0 (reference)
Asian or Pacific Islander	0.8 (0.4 to 1.6)	.498
African American	0.7 (0.2 to 2.4)	.551
Hispanic	0.2 (0.1 to 0.7)	.010
Other	0.6 (0.2 to 2.0)	.385
Political self-identification
Conservative	1.0 (reference)
Moderate	0.5 (0.3 to 0.8)	.002
Liberal	0.3 (0.2 to 0.6)	.001
Religious affiliation
Jewish	1.0 (reference)
Catholic	3.1 (1.5 to 6.2)	.002
Protestant	1.7 (0.9 to 3.5)	.129
Other†	2.7 (1.2 to 6.3)	.017
Importance of religion
Not at all important	1.0 (reference)
Moderately important	1.0 (0.5 to 2.0)	.938
Very important	4.1 (2.2 to 7.8)	<.001

^*The dependent variable is opposition to the legalization of physician-assisted suicide. Only members of the U.S. physicians sample were used in this regression. Ordered multinomial logistic regression using all 5 points of the Likert scale resulted in similar overall probabilities.

^†There were many different “other” religious affiliations, including, Buddhist, Latter-Day Saint (Mormon), Humanism, and “just a Christian.”

Both religious affiliation and the intensity of religious belief were significant predictors of attitude. Physician-assisted suicide was opposed by 45% of Catholic respondents, 32% of Protestant respondents, and 16% of Jewish respondents (P < .001). Intensity of religious belief was also a powerful predictor of attitude.

Respondents identified the role of religion in their lives as “not at all important,”“moderately important,” and “very important.” There was much greater opposition to the legalization of physician-assisted suicide among physicians for whom religion was very important (55%) than those for whom it was moderately important (23%) or not at all important (18%). Logistic regression showed no statistically significant difference between the latter 2 groups and confirmed a strong difference between those for whom religion was very important and those for whom it was not very important (P < .001).

In simple bivariate analysis, physicians of different ethnic groups did not have statistically significantly different views toward the legalization of physician-assisted suicide. However, logistic regression demonstrated that if other variables were held constant, Hispanic ethnicity was strongly predictive of support for the legalization of physician-assisted suicide. In bivariate analysis, physicians who were involved in direct patient care were less likely to support physician-assisted suicide than those who were not (43% vs 62%, P = .015), females were less likely to support physician-assisted suicide than males (36% vs 47%, P = .037), and physicians of different specialties showed varying levels of support for physician-assisted suicide (P < .001). However, none of these effects was seen in the logistic regression.

There was no consistent statistically significant pattern of opposition or support with age, time spent with the terminally ill, or any of the other independent variables. No statistically significant relationship was found between geographic region and attitude toward physician-assisted suicide. Physicians from the 3 states that have been previously studied—Oregon, Washington, and Michigan—held opinions toward the legalization of physician-assisted suicide that were not significantly different from physicians in the rest of the country (n = 47 for these states, P = .516).

American Medical Association Leaders and Members

Among respondents in the AMA House of Delegates, legalization was favored by 24% of the delegates and opposed by 62%; 15% were uncertain. The House of Delegates members were older (mean, 14 years) and, on average, were more likely to be male, white, and politically conservative than members of the U.S. physicians sample (Table 1). However, logistic regression demonstrated that even when all other personal characteristics were held constant, being a delegate was an independent and strongly significant predictor of opposition toward physician-assisted suicide (odds ratio, 3.0; 95% confidence interval, 2.0 to 4.5; P < .001). In contrast, AMA members within the U.S. physicians sample were not significantly more opposed to physician-assisted suicide than were nonmembers (Table 2).

The Role of Law

As in a previous study by Bachman et al.,⁴ respondents were asked specifically if they would prefer no law at all, leaving decisions about assisted suicide under the purview of the doctor-patient relationship or the medical profession. Fifty-five percent of the U.S. physicians favored a “no law” option, as did 58% of members of the AMA House of Delegates (Table 4). The most common preference in both groups was to leave this decision to the doctor-patient relationship rather than having the medical profession provide regulations or guidelines.

Table 4.

Should There Be a Law?^*

There should be …	U.S. Physicians†	AMA House of Delegates
A law allowing physician-assisted suicide	22	9
A law prohibiting physician-assisted suicide	17	28
No law, leave it to physician-patient relationship	31	38
No law, medical profession should provide guidelines	24	20
No answer or unsure	7	5
Total respondents	651	313

^*Responses were given to the question: “Some physicians feel that physician-assisted suicide should be legal; others feel it should be prohibited. Some physicians prefer no law at all, preferring instead to leave end-of-life decisions to the doctor-patient relationship or to regulations or guidelines to be provided by the medical profession. Which one of these options would you favor most?”

^†Values expressed as percent of those responding and may not sum to 100 because of rounding. AMA indicates American Medical Association.

DISCUSSION

Physician attitudes toward deliberately hastening death, whether by active or passive means, vary with the specifics of the situation. We probed the lower limits of support for physician-assisted suicide, stipulating only that the patient be competent. Previous surveys usually indicated that assisted suicide would be restricted to the terminally ill, often with detailed safeguards against abuse^4,5 sometimes the word “suicide” was not used at all.³ In contrast, this study's questionnaire consistently used the phrase “physician-assisted suicide” and made no mention of limiting this practice to the terminally ill or of the restrictions and safeguards that have been proposed^9–11 or implemented.^12,13 One limitation of our method is that some responses undoubtedly reflect the opinions of physicians toward physician-assisted suicide in general rather than toward our particular definition of assisted suicide. Another and more fundamental limitation is that no survey can tell us whether physician-assisted suicide is ever a valid moral choice or whether its legalization would be wise public policy.

In our study, 45% of the nationwide physician sample believed physician-assisted suicide should either definitely (16%) or probably (28%) be legal, and 34% felt it should either definitely (20%) or probably (14%) be illegal. These results can be compared to those of previous surveys of physicians in Oregon,⁵ Washington,³ and Michigan,⁴ which found that between 53% and 60% of physicians favored physician-assisted suicide and between 33% and 38% of physicians were opposed. Our data do not suggest strong regional differences, so it is more likely that our finding of lower levels of support for legalization stems from differences in the wording of our questionnaire, our choice of topics, reduced physician support over time, or the absence of safeguards. It is worth noting that our survey might have yielded different results if we had included questions about ethics, patient rights, or exceptional cases. For example, among respondents in a study of Oregon physicians by Lee and colleagues, 60% felt that physician-assisted suicide should be legal in some cases, 66% felt that it would be ethical in some cases, and 73% felt that a terminally ill patient has a right to commit suicide (this question did not specify physician assistance in the suicide).⁵

Our study confirmed earlier work^4,14 showing that the intensity of a physician's religious beliefs is an excellent predictor of attitude toward physician-assisted suicide. We did not, however, replicate the findings of Bachman and colleagues⁴ who found that physicians who spend more time working with the terminally ill were less likely to support physician-assisted suicide. We also did not replicate the findings of Cohen et al.³ that women physicians were more supportive of physician-assisted suicide than men, and psychiatrists more supportive than other specialty groups. In previous studies, physicians who were nonwhite¹⁵ or African American¹⁶ were found to be less likely than white physicians to favor physician-assisted suicide. Our U.S. physicians sample contained only 20 African-American physicians, so it is not surprising that our study had insufficient statistical power to distinguish their views from those of their white colleagues. Our finding that Hispanic ethnicity is an independent predictor of support for physician-assisted suicide should be viewed as a preliminary result since it is based on the views of only 29 Hispanic respondents; further research into the views of Hispanic physicians would be of value.

Most physicians, like most members of other professional groups, prefer a minimum of legal intervention in their practices. In the survey of Michigan physicians by Bachman and coworkers, 37% favored “no law” over other choices.⁴ We used the same wording for this particular question and found agreement between our 2 study groups on this topic, with 55% support for “no law” among the U.S. physicians sample and 58% among members of the AMA House of Delegates. The greater preference for “no law” among our physician samples might be explained by changes in opinion over time, our study's broader definition of physician-assisted suicide, or other differences in the questionnaires.

It is commonplace, and perhaps too easy, to say that views of the public, or physicians, toward physician-assisted suicide are polarized. Authors and editorial writers often have strong views for or against physician-assisted suicide, and within any group of practicing physicians, there are some who vigorously favor or oppose the practice. Most physicians in our study, however, declined to unequivocally endorse either legal alternative. Only 37% of the nationwide physician sample believed that physician-assisted suicide should “definitely” be legal or illegal, 42% indicated that the practice should “probably” be either legal or illegal, and another 22% were unsure. Many respondents added the handwritten comment, “It depends on the case.”

The views of members of the AMA House of Delegates are strikingly different from those of the nationwide physician sample; 61% of the delegates opposed legalization versus 34% of the nationwide sample. This difference is significant at the P < .001 level. One limitation of this study is that because the instrument used a closed-ended response format, it cannot tell us about the subjective dimensions of the respondents' opinions and the personal and moral beliefs that underlie their views. Perhaps the rank-and-file physician focuses primarily on his or her individual patients, while the physician leader gives more weight to the harm that legalizing physician-assisted suicide might cause to the profession and to the nation; perhaps other factors are at work. Further research to explore the differences in attitude between physicians and physician leaders would be valuable.

The AMA leadership has emphasized that it believes physician-assisted suicide to be morally wrong and poor public policy; our results suggest that this view is probably not shared by most practicing physicians. This discrepancy between physicians and physician leaders raises important questions. What are the implications of this difference for the political work of the AMA and future AMA policy regarding physician-assisted suicide? How might disagreement in moral values between grassroots physicians and the AMA leadership best be addressed? These questions offer an opportunity for thoughtful discussion and further research.