“Patient autonomy has achieved paradigmatic status in both the ethics and the law of medicine,” wrote Carl Schneider in 1998 (p7).1 More recent articles testify to its continued pre-eminence.2^–4 Typically, this paradigm requires physicians to provide scientific and medical expertise, while the patients decide which clinical strategies are most consistent with their values.1 5 6

Although the principle of autonomy is widely affirmed at the theoretical level, its application in concrete situations can be problematic. For example, how does a physician respect autonomy when patients do not want to make their own decisions?1 5 7 What about those with shifting preferences?1 How much persuasion is appropriate?1 In these situations and others, the proper implementation of patient autonomy is less than clear. Additionally, there has been concern that elevating the role of autonomy will lead physicians to adopt a laissez-faire attitude towards patients, lessening physicians’ sense of responsibility.1

Despite a decade of autonomy having “paradigmatic status”, little is known about how doctors apply the principle in clinical settings. We hypothesised that if physicians believe patient autonomy to be the paramount concern, they will give patient’s expressed wishes the highest weight possible, while giving competing concerns less weight. To test this, we surveyed physicians and asked them how much weight they give to patient preferences. For comparison we selected three commitments that sometimes compete with patient autonomy and asked how much weight physicians give to these.

We also explored whether physicians’ religious commitments affect the role given to patient autonomy. Earlier we found that religious physicians are less likely to believe they are obligated to provide information about or refer patients for medical interventions they find objectionable8—a trend at odds with prevalent interpretations of the autonomy principle. The data from this study provide a view into how decisions are made in clinical settings, particularly the clinical role of autonomy 10 years after Schneider’s statement of its triumph in ethical literature.

METHODS

In 2007, we mailed a confidential, self-administered questionnaire to a stratified random sample of primary care physicians drawn from the American Medical Association Physician Masterfile—a database intended to include all physicians in the USA. From the universe of practising internal medicine, general practice and family medicine physicians aged 60 or younger, we first selected 500 physicians at random. These constitute the principle sample. Another aim of this survey was to explore physicians’ religious characteristics, and in order to increase Muslim, Hindu and Buddhist representation, we utilised validated surname lists9 10 to select an additional 250 physicians with typical South Asian surnames and 250 physicians with typical Arabic surnames. Demographic characteristics included sex, race, age, region and immigration history. Physicians received up to three separate mailings of the questionnaire. The first included a US$5 Starbucks gift card, and the third offered $30 for participation. The study was approved by the University of Chicago institutional review board.

Questionnaire

The primary criterion variables for this analysis were physicians’ responses to the following questions: “When making an ethically complex medical decision, how much weight should physicians give to each of the following considerations? 1) the patient’s expressed wishes and values, 2) the physician’s own judgment about what is in the patient’s best interest, 3) standards and recommendations from professional medical bodies, and 4) moral guidelines from religious traditions”. Response options were: little to no weight, some weight, a lot of weight and the highest possible weight.

We considered whether physicians gave patients’ expressed wishes and values the highest possible weight, or a higher relative weight than the other three considerations. Relative weight was obtained by looking at each respondent’s answers, and noting which factor(s) received the most weight. Secondary criterion variables were how much weight respondents assigned to guidelines from religious traditions, along with whether or not physicians agreed with the following statement: “Physicians should not let their religious beliefs keep them from providing patients legal medical options.”

Primary predictors were measures of physicians’ religious characteristics. Religious affiliation was categorised as no religion, Hindu, Muslim, Catholic/Orthodox, evangelical Protestant, non-evangelical Protestant and other religion (includes Buddhist (n = 5), Jewish (n = 16), and other (n = 14)). We measured intrinsic religious motivation—the extent to which individuals embrace their religion as the “master motive” that guides and gives meaning to their life11—by asking seven questions derived from the Hoge Intrinsic Religious Motivation Scale:12

(1) I try hard to carry my religious beliefs over into all my other dealings in life.

(2) My whole approach to life is based on my religion.

(3) My faith involves all of my life.

(4) I seek God’s guidance when making every important decision.

(5) My faith sometimes sets limits on my actions.

(6) Nothing is as important to me as serving God as best I know how.

(7) In my life I experience the presence of the Divine.

These items have a Cronbach α of 0.94 in our sample. Responses, which ranged from 1 (strongly agree) to 4 (strongly disagree), were averaged, and respondents were classified as having high, medium or low intrinsic religious motivation. Organisational or participatory religiosity was measured as physicians’ frequency of attendance at religious services and was categorised as never, once a month or less, or twice a month or more. A substantial minority of physicians consider themselves spiritual but not religious. To identify this group, we asked physicians “To what extent do you consider yourself a spiritual person?” and “To what extent do you consider yourself a religious person?” Responses were dichotomised (very/moderately versus slightly/not at all) and respondents were categorised as religious, spiritual not religious, or neither spiritual nor religious.

Statistical analysis

In our analysis we used weights to correct for sampling and response bias, so that we could make statements about all US physicians, not just those in our sample. These weights take into account the sampling strategy, differences in response rates among the three samples, and different response rates along lines of region, gender, and specialty. After correcting for these biases, we generated overall population estimates for agreement with each of the criterion measures. We then used the χ² test to examine the associations between each predictor and each criterion measure. Finally, we used multivariate logistic regression to test whether bivariate associations remained significant after adjusting for other relevant covariates. All analyses were conducted with Stata SE V.10 statistical software. Respondents who left questions blank were omitted from our analysis of those questions.

RESULTS

Survey response

Approximately 12% (121) of the questionnaires were returned as undeliverable. The response rate among eligible physicians was 51% (446/879). Response rates varied by sample: 55% (246 respondents/450 eligible) of the primary sample responded, 49% (104/212) of those with South Asian surnames responded and 44% (96/217) of those with Arabic surnames responded. There was no significant variation in response by gender, region or specialty. Respondent characteristics are shown in table 1.

Relevant criteria for ethically complex medical decisions

When making ethically complex medical decisions, 55% of physicians gave the patient’s expressed wishes and values “the highest possible weight”. However, many doctors gave more than one consideration “the highest possible weight”. Thus, patient wishes were given the highest relative weight by only 40% of doctors (table 2). Incidentally, when patient wishes tied with other concerns as chief priority, they most often tied with professional guidelines (12%, n = 76), best interest (9%, n = 32) or both (16%, n = 76), or the physicians ranked all four concerns equally (8%, n = 32). Thirteen per cent of physicians ranked the patient’s wishes behind some other concern.

While most physicians take patients’ wishes and values very seriously, most also consider their own judgment about the patient’s best interest to merit either “the highest possible weight” (15%) or “a lot of weight” (59%). Most also gave professional standards either “the highest possible weight” (18%) or “a lot of weight” (59%).

A minority of physicians believed moral guidelines from religious traditions should receive “the highest possible weight” (5%) or “a lot of weight” (16%). Not surprisingly, then, most (78%) agreed that physicians should not let their religious beliefs keep them from providing legal medical options to patients (table 2).

As seen in table 3, religious doctors tended to give less weight (absolute or relative) to patient wishes. In particular, doctors with high intrinsic religious motivation were less likely than those with low to give patient’s expressed wishes the highest possible weight (47% vs 67%, OR 0.5, CI 0.3 to 0.8). They were also less likely to give a patient’s expressed wishes the highest relative weight (31% vs 56%, OR 0.4, CI 0.2 to 0.7). Similarly, doctors who described themselves as religious were less likely than the “neither religious nor spiritual” to give patient preferences the highest relative weight (33% vs 52%, OR 0.5, CI 0.3 to 0.9). After correction for all religious and demographic characteristics, older physicians (aged 47–60 years) were found less likely than younger ones (aged 26–29 years) to give patient preferences the highest relative weight (26% vs 49%, OR 0.4, CI 0.2 to 0.97).

Non-religious doctors tended to think that a physician’s religion should not limit the treatment options made available to patients. Specifically, the belief that moral guidelines from religious traditions should receive little to no weight was unpopular among doctors with high intrinsic religiosity (10% agree) but had considerable support among doctors with low intrinsic religiosity (56% agree, OR 0.1, CI 0.04 to 0.2). Similarly, the belief that doctors should not let religious beliefs keep them from providing legal medical options had some support among doctors with high intrinsic religiosity (63% agreed) but was approved by the great majority of doctors with low intrinsic religiosity (91% agree, OR 0.2, CI 0.1 to 0.5) (table 4).

DISCUSSION

When making complex medical decisions, doctors place considerable emphasis on patients’ expressed wishes and values. Overall, concern for patient autonomy received more weight than any other criteria, with most doctors (93%) believing that patients’ expressed wishes and values deserve at least “a lot of weight” and over half of doctors (55%) giving those expressed wishes and values “the highest possible weight”.

Despite this strong emphasis on respect for autonomy, many doctors also gave much weight to other criteria, with the result that only 40% of physicians named autonomy as the dominant concern. This raises the question of whether patient autonomy has the degree of importance often advocated in theoretical ethics literature. If physicians truly considered respect for autonomy to be “the preeminent value … the top of the moral mountain” (Joffe et al, 2003, p103),3 if they truly believed their purpose was “not only to relieve suffering but to enhance patients’ autonomy” (Godolphin, 2003, p692)2 and if they truly held that “the competent adult patient … retains final decisional authority” (Whitney et al, 2003, p54),4 would not more physicians give patient’s expressed wishes and values the highest relative weight?

The data suggest that, while valuing patient preferences, physicians tend to weigh them in the balance against other factors, which are often equally valued. This approach resembles the model proposed by Jonsen, Siegler and Winslade, who note, “although it is frequently said that the principle of autonomy holds priority in American bioethics … all principles and all the facts of a case must be viewed together in order to make a balanced judgment” (p4).13 Physicians appear to share Schneider’s concern about “simplistic and extravagant versions of the autonomy paradigm” and prefer “a less absolutist, better modulated, and more proportional version of autonomy”, choosing not to “promote the autonomy paradigm by every conceivable means, on every conceivable front, and at every conceivable occasion” (p33).1

This picture suggests that writers’ concerns about doctors overemphasising the autonomy principle to the detriment of patients have not materialised in mainstream clinical practice. For example, Schneider warned that “doctors can make the autonomy paradigm a welcome and acceptable way of passing on burdensome problems to patients” (p5).1 He also noted that patient preferences can be poorly thought out, or subject to change—precluding a straightforward application of the autonomy principle. Alternatively, Appelbaum and Roth, commenting on empirical data, worried about physicians being “too ready to concede patients’ “right to refuse” rather than to recognize the clinical problems that lay at the bottom of the refusal (e.g., poor or inconsistent communication) and to take steps to remedy them” (p1301).14 The physicians in our study do not obviously fall victim to these concerns, but only because they have curtailed the role of autonomy. Whether this represents a forward-looking and balanced implementation of the autonomy principle or a continuation of paternalistic models is open for discussion.

Our finding that younger physicians give more weight to patient preferences parallels Schneider’s report that younger patients are more inclined to make their own treatment decisions.1 This trend of older doctors downplaying autonomy could represent an echo of paternalistic models that were once prominent in medical education.6 If so, it suggests that views acquired during medical training continue to influence practice patterns decades later, even when the predominant view has shifted. Alternatively, it could represent a stable difference between young and old, with veteran physicians holding firmer opinions shaped by years of experience. Regardless of its origins, the consequence of this observation is that physicians must be aware that their tendencies may not match the expectations of patients, particularly when patients belong to another generation.

Most physicians were interested in limiting the role of religion in medicine. This concern is given voice in a recent statement by the American College of Obstetricians and Gynecologists ethics committee, which supported limits on conscientious refusals that “constitute an imposition of religious or moral beliefs on patients …” (p1203).15 Others, however, have challenged the notion that doctors ought to keep their personal values separate from their public and professional lives. For example, Pellegrino has equated “value neutrality” with requiring physicians “to sacrifice moral integrity to the requirements of their social role as that role is interpreted by secular bioethicists” (p78).16 Of relevance to this debate, we found that a majority of religious physicians agree that religious beliefs should not keep doctors from providing legal medical options. This suggests that most religious physicians are open, on some level, to accommodating societal and professional expectations that are in tension with their religious commitments.

Our study has several limitations. First, we only surveyed primary care physicians, yet doctors in different areas of medicine face different kinds of clinical and ethical decisions. Also, because we were probing basic tendencies our question was quite general, and different doctors may have envisaged very different clinical scenarios when answering the questions. Future studies using vignettes would help to eliminate this type of variability. Most clinical decisions require physicians to weigh more than four competing commitments, but we suspect that a similar pattern holds: patient autonomy receives high priority but has active competition from other concerns. Limiting our survey to four commitments helped us to address our primary hypothesis (whether autonomy is paramount) but leaves many questions about physician decision-making unexplored. While our analysis found many correlations, the cross-sectional design does not permit inferences about the causes of the associations. Additionally religious and other characteristics may have systematically affected physicians’ willingness to respond to the survey. Finally, we recognise that self-reports are imperfect measures of physicians’ beliefs and practices.

CONCLUSIONS

Concern for patient autonomy appears to influence physicians’ decisions more than other criteria, but it may not hold the degree of preeminence frequently advocated in the bioethics literature. While autonomy is regarded highly, doctors often give equal weight to other considerations, such as their perception of what is in the patient’s best interest and guidelines from professional bodies. The result is a decision-making process that resembles Schneider’s recommendation: patient autonomy acting not as a single beguiling flower, but as the centrepiece for a whole bouquet of concepts.1