Journal of General Internal Medicine

, Volume 29, Issue 2, pp 335–340

Directive Counsel and Morally Controversial Medical Decision-Making: Findings from Two National Surveys of Primary Care Physicians

Authors

  • Michael S. Putman
    • Pritzker School of MedicineThe University of Chicago
    • The Department of MedicineThe University of Chicago
  • John D. Yoon
    • The Department of MedicineThe University of Chicago
    • MacLean Center for Clinical Medical EthicsThe University of Chicago
    • Program on Medicine and ReligionThe University of Chicago
  • Kenneth A. Rasinski
    • Program on Medicine and ReligionThe University of Chicago
    • The Department of MedicineThe University of Chicago
    • MacLean Center for Clinical Medical EthicsThe University of Chicago
    • Program on Medicine and ReligionThe University of Chicago
Original Research

DOI: 10.1007/s11606-013-2653-4

Cite this article as:
Putman, M.S., Yoon, J.D., Rasinski, K.A. et al. J GEN INTERN MED (2014) 29: 335. doi:10.1007/s11606-013-2653-4

ABSTRACT

BACKGROUND

Because of the potential to unduly influence patients’ decisions, some ethicists counsel physicians to be nondirective when negotiating morally controversial medical decisions.

OBJECTIVE

To determine whether primary care providers (PCPs) are less likely to endorse directive counsel for morally controversial medical decisions than for typical ones and to identify predictors of endorsing directive counsel in such situations.

DESIGN AND PARTICIPANTS

Surveys were mailed to two separate national samples of practicing primary care physicians. Survey 1 was conducted from 2009 to 2010 on 1,504 PCPs; Survey 2 was conducted from 2010 to 2011 on 1,058 PCPs.

MAIN MEASURES

Survey 1: After randomization, half of the PCPs were asked if physicians should encourage patients to make the decision that the physician believes is best (directive counsel) with respect to “typical” medical decisions and half were asked the same question with respect to “morally controversial” medical decisions. Survey 2: After reading a vignette in which a patient asked for palliative sedation to unconsciousness, PCPs were asked whether it would be appropriate for the patient’s physician to encourage the patient to make the decision the physician believes is best.

KEY RESULTS

Of 1,427 eligible physicians, 896 responded to Survey 1 (63 %). Physicians asked about morally controversial decisions were half as likely (35 % vs. 65 % for typical decisions, p < 0.001) to endorse directive counsel. Of 986 eligible physicians, 600 responded to Survey 2 (61 %). Two in five physicians (41 %) endorsed directive counsel after reading a vignette describing a patient requesting palliative sedation to unconsciousness; these physicians tended to be male and more religious.

CONCLUSIONS

PCPs are less likely to endorse directive counsel when negotiating morally controversial medical decisions. Male physicians and those who are more religious are more likely to endorse directive counsel in these situations.

KEY WORDS

medical ethicsend of life caremedical decision-makingprimary caredirective counselpalliative care

INTRODUCTION

Debates over the proper role of directive counsel become especially contentious when patients and physicians negotiate morally controversial medical decisions. Some ethicists oppose the use of directive counsel,13 arguing that physicians’ moral values in these situations fall outside of “expert clinical judgment”1 and introduce the possibility of physicians “covertly or openly imposing [their] moral beliefs on patients”.4 Yet others have argued that a patient may best be supported by an open dialogue that explores the values of both the patient and physician, with the physician offering his or her own recommendations.5 In these “enhanced autonomy” models,2,68 physicians might candidly and respectfully encourage patients to make the decision the physician feels is best. Such models, it is argued, avoid reducing physicians to “technicians” who merely “present the ‘facts’ unadorned by values”.9

This debate has its broadest clinical ramifications in the field of primary care, since primary care physicians (PCPs) handle more than one-third of all patient encounters,10 often act as “gatekeepers to specialist referrals,11 and advise patients on end-of-life decisions.12 Despite a burgeoning body of research on medical decision-making in general,1316 very little is known about how primary care physicians negotiate morally controversial medical decisions. In a recent national survey, Yoon et al. found that the majority of obstetrician gynecologists oppose giving directive counsel in the case of morally controversial decisions.17 Yet obstetrician gynecologists more commonly face the morally charged issues in reproductive health, and they may have adopted nondirectiveness to avoid conflicts over these issues, which have been associated with higher rates of emotional stress and burnout.18 In the context of genetic counseling, a study by Geller et al. indicated that PCPs were more likely than obstetrician-gynecologists, pediatricians, or psychiatrists to offer an opinion about prenatal testing or abortion.19

Using two large, nationally representative samples of practicing primary care physicians, this study investigates the degree to which PCPs endorse directive counsel in typical or morally controversial situations. The study then uses an experimental vignette to see whether physicians’ in-general opinions about directive counsel are reflected in their opinions about directive counsel for a patient requesting palliative sedation to unconsciousness. Because previous studies have indicated a relationship between religious variables and physician responses in morally controversial situations,17,1921 religious characteristics and other demographic covariates were included in the analysis.

METHODS

Survey 1

Between September 2009 and June 2010, we mailed a self-administered confidential questionnaire to a stratified random sample of 1,504 US generalist physicians under the age of 65 with a primary board specialty of internal medicine, family medicine/practice, or general practice and with no secondary subspecialty. The sample was generated from the American Medical Association Physician Masterfile, a database intended to include all practicing US physicians. Hospitalists in our sample were identified by a survey item asking, “Are you a hospitalist? (yes/no)” and were then subsequently excluded. To increase the study’s overall power to examine differences between religious groups, we included modest oversamples of physicians with typical Arabic, South Asian, and Jewish surnames.2225 Physicians received up to three separate mailings of the questionnaire. The first mailing included a $20 bill and the third offered an additional $30 for participation.

Respondents were randomized to receive one of two versions of the same item. Half the sample received the following: “Please indicate to what extent you agree or disagree with the following statements: When dealing with typical medical decisions, a physician should”: “a) avoid influencing the patient’s decision one way or another”; “b) encourage the patient to make the decision that the physician believes is best”. Response categories for both statements were Agree strongly, Agree somewhat, Disagree strongly, and Disagree somewhat. Those who agreed with statement b) were classified as endorsing directive counsel. The other half of the sample received the same item, except that “typical medical decisions” was replaced with “morally controversial medical decisions”.

Survey 2

Between July 2010 and February 2011, we mailed a self-administered confidential questionnaire to a second stratified random sample of 1,058 practicing US generalist physicians under the age of 65 with a primary board specialty of internal medicine, family medicine/practice, or general practice and with no secondary subspecialty. We followed similar procedures as noted above in Survey 1 to obtain a true primary care sample and to increase the power of our sample to include minority religious perspectives. All data from both surveys were double keyed, cross-compared, and corrected against the original questionnaires. Both studies were also approved by the University of Chicago institutional review board.

In Survey 2, we created a clinical vignette about palliative sedation to unconsciousness (PSU) to evaluate the extent to which physicians’ in-general opinions from Survey 1 translated into opinions regarding directive counsel in a specific morally controversial situation. Palliative sedation to unconsciousness, sometimes referred to as “terminal sedation”, has previously been associated with both moral controversy26,27 and physicians’ religious characteristics.2830 The vignette read:

KD is a 62-year-old woman dying at home from metastatic lung cancer. Her pain has been treated with high-dose long-acting narcotics. Her dyspnea has been treated with a combination of oxygen, narcotics, and intermittent nebulizer treatments. KD tells her physician that her pain and dyspnea are well controlled, but she is distressed at the constant thought of her impending death. She says, ‘I know I am going to die; I just cannot tolerate lying here thinking about it day after day’. KD asks her physician to sedate her to unconsciousness until she dies.

Physicians were then asked to respond to the following: “Please indicate to what extent you agree or disagree with the following statements: When talking to KD about this decision, the physician should… 1) Try to avoid influencing the patient’s decision one way or another”; “2) encourage the patient to make the decision that the physician believes is best”. Response categories for 1) and 2) were Agree strongly, Agree somewhat, Disagree strongly, and Disagree somewhat. Other items from this vignette have been published elsewhere.31

Predictors

For both studies, respondent religious characteristics were primary predictors. Religious affiliation was categorized as: None, Hindu, Jewish, Muslim, Roman Catholic/Eastern Orthodox, evangelical Protestant, non-evangelical Protestant, and Other religion. We also assessed attendance at religious services (categorized as Never, Once a month or less, and Twice a month or more) and self-rated importance of religion (categorized as Not very important, Fairly important, Very important, and Most important). Finally, we assessed theological pluralism, a measure of the extent to which a respondent believes that no religious tradition is uniquely and comprehensively true,17 by asking respondents to indicate agreement or disagreement with the following three statements: 1) “Different religions have different versions of the truth, and each may be equally right in its own way”, 2) “There is one religion that is uniquely and comprehensively true” and 3) “There is no one, true, right religion”. After reverse coding the second item, responses were summed to create a scale (Study 1 Cronbach alpha = 0.80, Study 2 Cronbach alpha = 0.74), which was divided into three categories (Low, Moderate, High). Additional predictors included gender, age, race/ethnicity, and immigration history (US or foreign born).

Statistical Analysis

The methods of both surveys have been published elsewhere,31,32 and full methodology reports are available online.33,34 Case weights were incorporated to account for oversampling and differences in response rates between groups. Respondents who left questions blank were omitted from analyses of those items. First, we generated population estimates for responses to each survey measure. Next, we used the Pearson χ2 test and multivariable logistic regression to evaluate differences in responses to the two criterion measures by each of the predictors. Because of multicollinearity between measures of religiosity, results for each religious characteristic are adjusted for demographic characteristics, but not for other religious characteristics. All analyses take into account survey design and were performed on Stata/SE 12.0 statistical software (Stata Corp., College Station, TX).

RESULTS

Respondents

Respondent characteristics for both surveys are displayed in Table 1.
Table 1

Respondent Characteristics from Two National Surveys* of Practicing Primary Care Physicians

Demographic and religious characteristics

Survey 1: no. (%)

Survey 2: no. (%)

Age

 24–37 years

174 (23)

130 (26)

 38–45 years

173 (23)

117 (23)

 46–54 years

205 (27)

121 (24)

 55–65 years

202 (27)

133 (27)

Gender

 Female

284 (38)

220 (44)

 Male

470 (62)

281 (56)

Race/ethnicity

 White, non-Hispanic

537 (72)

357 (72)

 Black, non-Hispanic

41 (6)

21 (4)

 Asian

112 (15)

78 (16)

 Hispanic/Latino

36 (5)

21 (4)

 Other

18 (2)

20 (4)

Immigration history

 Born in the United States

560 (75)

363 (74)

 Immigrated to United States

185 (25)

130 (26)

Religious affiliation

 None

80 (11)

58 (12)

 Hindu

33 (5)

23 (5)

 Jewish

85 (12)

39 (8)

 Muslim

42 (6)

27 (6)

 Roman Catholic/Eastern Orthodox

176 (24)

135 (28)

 Protestant, evangelical

84 (12)

46 (9)

 Protestant, nonevangelical

194 (27)

130 (27)

 Other

36 (5)

30 (6)

Attendance at religious services

 Never

96 (13)

67 (14)

 Once a month or less

340 (47)

230 (47)

 Twice a month or more

294 (40)

195 (40)

Importance of religion

 Not important/not applicable

179 (24)

134 (27)

 Fairly important

239 (32)

141 (29)

 Very important

210 (29)

142 (29)

 Most important

108 (15)

75 (15)

Theological pluralism§

 High

312 (44)

226 (46)

 Moderate

170 (24)

126 (26)

 Low

235 (33)

136 (28)

*Survey 1: Data from a national survey of 1,504 practicing primary care physicians, 2009–2010; Survey 2: Data from a national survey of 1,058 practicing primary care physicians, 2010–2011

Because of rounding error, numbers may not add up to 100 %. Because of partial nonresponse, numbers do not add up to 1,504 or 1,058

There were no significant differences in any variables between Survey 1 and Survey 2 (p < 0.05)

§Theological pluralism measures the extent to which a person believes that no religious tradition is uniquely and comprehensively true

  1. Survey #1:

    Because of retirement, leaving the country, or bad addresses, 5 % (77/1,504) of PCPs were declared out of scope. Among eligible physicians, the response rate was 63 % (896/1,427). PCP response rates differed by medical school graduation (56 % [218/387] for foreign vs. 65 % for U.S [678/1,039]), and for practice type (teaching and office-based 75 % [103/138] vs. all other types 62 % [793/1,288]). Response rates did not differ significantly by age, gender, region, or board certification, or typical vs. morally controversial decision scenarios.

     
  2. Survey #2:

    Because of retirement, leaving the country, or bad addresses, 7 % (72/1,058) of PCPs were declared out of scope. Among eligible physicians, the response rate was 61 % (600/986). PCP response rates differed by medical school graduation: 53 % (172/326) for foreign vs. 65 % for U.S (428/660), p < 0.001; and by gender: 57 % (344/599) for male respondents and 66 % (256/387) for female respondents, p = 0.006. Response rates did not differ significantly by age, region, or board certification. PCPs in Survey 1 and Survey 2 did not differ significantly with respect to their demographic or religious characteristics.

     

Survey 1

Experiment

As hypothesized, physicians who were asked about “morally controversial” clinical decisions were half as likely (35 % vs. 65 % for typical decisions, p < 0.001) to endorse directive counsel. Similarly, they were more likely to agree that they should avoid influencing the patient one way or another (65 % vs. 35 % for typical decisions, p < 0.001).

The following data from Survey 1 are not shown in tables. With respect to morally controversial situations, endorsing directive counsel was independently associated with being male (40 % vs. 28 % female, p = 0.04; adjusted OR 1.8, CI 1.1–3.2), and inversely associated with being non-evangelical Protestant (20 % vs. 48 % no affiliation, p = 0.004; OR 0.3, CI 0.1–0.8). With respect to typical medical decisions, endorsing directive counsel was independently associated with being male (67 % vs. 60 % female, p = 0.16; OR 2.2 CI, 1.3–3.8), Asian (81 % vs. 59 % white, p = 0.002; OR 4.3, CI 1.4–13.3), Roman Catholic/Eastern Orthodox (71 % vs. 47 % no affiliation, p = 0.02; OR 3.5, CI 1.5–7.9), with attending religious services once a month or less (69 % vs. 50 % never, p = 0.02; OR 2.5 CI 1.2–5.1) or twice a month or more (65 % vs. 50 %, p = 0.10; OR 2.3 CI 1.1–4.7), and with considering religion to be very important (70 % vs. 54 %, p = 0.04; OR 2.3 CI 1.2–4.4). Physicians who had immigrated to the US were more likely to endorse directive counsel for typical decisions (74 % vs. 60 % US born, p = 0.02), but this difference did not persist in adjusted analyses.

Survey 2

When presented with a clinical vignette in which a patient asks about palliative sedation to unconsciousness, 41 % of physicians endorsed directive counsel, and 57 % of physicians endorsed avoiding influencing the patient’s decision one way or another.

Table 2 displays the prevalence and adjusted odds of endorsing directive counsel in this case, stratified by physicians’ demographic and religious characteristics. In bivariate analysis, physicians were more likely to endorse directive counsel if they were Asian, attended religious services, or had low theological pluralism. In multivariable analysis, these differences persisted for Asian ethnicity (56 % vs. 38 % for white, OR 3.6 CI 1.5–8.4), attendance at religious services once a month or less (46 % vs. 21 % for never, OR 3.1 CI 1.5–6.4), attendance at religious services twice a month or more (42 % vs. 21 % for never, OR 3.2 CI 1.6–6.7), and low theological pluralism (53 % vs. 38 % high, OR 2.1 CI 1.3–3.4). While not significant in bivariate analysis, physicians were more likely in adjusted analyses to endorse directive counsel if they were Roman Catholic/Eastern Orthodox (43 % vs. 30 % for no affiliation, OR 2.5 CI 1.2–5.0), evangelical Protestant (49 % vs. 30 % for no affiliation, OR 2.3 CI 1.1–4.7), non-evangelical Protestant (39 % vs. 30 % for no affiliation, OR 2.3 CI 1.1–4.7), and if they considered religion to be most important (48 % vs. 31 % not important, OR 2.7 CI 1.4–5.2).
Table 2

US Primary Care Physicians Endorsing Directive Counsel to Patients in a Clinical Vignette About Palliative Sedation to Unconsciousness, Stratified by Physicians’ Demographic and Religious Characteristics, 2010–2011*

“A physician should encourage the patient to make the decision the physician believes is best” (agree strongly/agree somewhat)

Demographic and religious

Bivariate

Multivariable

Characteristics

No. %

P(χ2)

OR [95 % CI]

Age

 24–37 years

53 (43)

0.49

Referent

 38–45 years

40 (37)

 

1.1 [0.5–2.1]

 46–54 years

44 (38)

 

1.3 [0.7–2.4]

 55–65 years

59 (46)

 

1.9 [1.0–3.7]

Gender

 Female

80 (38)

0.22

Referent

 Male

116 (44)

 

1.3 [0.8–2.0]

Race/ethnicity

 White, non-Hispanic

132 (38)

p = 0.024

Referent

 Black, non-Hispanic

4 (23)

 

0.4 [0.1–1.3]

 Asian

42 (56)

 

3.6 [1.6–8.4]

 Hispanic/Latino

7 (38)

 

1.0 [0.3–3.0]

 Other

10 (51)

 

2.4 [0.6–8.9]

Immigration history

 Born in the United States

135 (39)

p = 0.051

Referent

 Immigrated to United States at any age

61 (50)

 

1.1 [0.6–2.1]

Religious affiliation

 None

15 (30)

p = 0.43

Referent

 Hindu

11 (56)

 

1.4 [0.4–4.7]

 Jewish

12 (37)

 

1.7 [0.6–4.3]

 Muslim

16 (51)

 

1.5 [0.4–5.6]

 Roman Catholic/Eastern Orthodox

53 (43)

 

2.5 [1.2–5.0]

 Protestant, evangelical

21 (49)

 

3.5 [1.5–8.3]

 Protestant, non-evangelical

49 (39)

 

2.3 [1.1–4.7]

 Other

12 (38)

 

1.2 [0.4–3.5]

Attendance at religious services

 Never

15 (21)

p = 0.004

Referent

 Once a month or less

96 (46)

 

3.1 [1.5–6.4]

 Twice a month or more

79 (42)

 

3.2 [1.6–6.7]

Importance of religion

 Not important/not applicable

40 (31)

p = 0.075

Referent

 Fairly important

60 (45)

 

1.7 [1.0–3.0]

 Very important

56 (42)

 

1.6 [1.0–2.8]

 Most important

34 (48)

 

2.7 [1.4–5.2]

Theological pluralism‡§

 High

79 (38)

p = 0.01

Referent

 Moderate

41 (35)

 

0.8 [0.5–1.3]

 Low

70 (53)

 

2.1 [1.3–3.4]

*Table 2 displays survey-design adjusted prevalence and odds ratios of those who agree strongly or somewhat to the following statement regarding the clinical vignette described in the Methods section: “When talking to KD about this decision, the physician should encourage the patient to make the decision that the physician believes is best”. Data from a 2010–2011 national survey among a stratified, random sample of 1,058 US Primary Care Physicians. Percentages do not sum to 100 % because the Disagree strongly/somewhat data was omitted from this table for clarity

p < 0.05

Religious characteristics were modeled independently from each other but with all demographics characteristics included

§ Theological pluralism measures the extent to which a person believes that no religious tradition is uniquely and comprehensively true

DISCUSSION

In two large national surveys, two out of three PCPs agreed that with respect to typical medical decisions, physicians should encourage their patient to make the decision the physician believes is best. With respect to morally controversial decisions, however, two out of three PCPs opposed giving directive counsel and three in five opposed giving directive counsel to a patient who is considering palliative sedation to unconsciousness. In the latter case, physicians were more likely to endorse directive counsel if they attended religious services at all, considered religion to be fairly or most important, or were less theologically pluralistic.

These findings indicate that a majority of practicing PCPs embrace the “independent choice” or “fact-provider” models of the doctor–patient relationship for morally controversial situations.1,6,7 When negotiating morally controversial decisions, physicians must navigate complex moral questions for which no medical or societal consensus may exist.1,3 Concern about bias has led to a general trend toward trying to remain morally neutral in such situations.35 While the present study suggests that PCPs have similarly adopted nondirective counsel as the norm, roughly one in three still endorse directive counsel for morally controversial decisions, both in theory and in response to a vignette patient asking for palliative sedation to unconsciousness.

Regarding religious physicians, those who identified as Christian, attended religious services regularly, considered religion to be most important, or were less theologically pluralistic were more likely to endorse directive counsel in the clinical vignette about terminal sedation. The basis for this difference between more and less religious physicians may best be understood with respect to theological pluralism. Physicians who believe that one religious tradition is uniquely and comprehensively true may have more confidence that they can, on the basis of that tradition, discern the right decision regarding a morally controversial practice. In this study, more than half of such physicians endorsed giving directive counsel to the patient seeking palliative sedation to unconsciousness.

This study has a number of limitations. Though Survey 2 moved from an in-general question about directive counsel to giving directive counsel in a specific clinical scenario, physicians’ opinions might differ substantially with respect to other clinical scenarios. The vignette neither specified whether the patient was imminently dying nor addressed whether fluids or nutrition would be withheld; different assumptions about these issues may have influenced physicians’ responses. Although our response rates for both surveys (63 % and 61 %, respectively) were high, non-respondents may have differed from our sample in ways we cannot measure.36 Due to the cross-sectional design of these surveys, the study cannot support inferences about causation regarding the association between physician religious predictors and their opinions about directive counsel. Priming effects, recall biases, and social desirability biases may limit the survey’s ability to reflect the actual opinions of respondents. Finally, physicians’ self reports may not reflect what they actually do.

Notwithstanding these limitations, this study confirms that PCPs are much less likely to endorse directive counsel for morally controversial situations than they are for typical medical decisions, and that most endorse a nondirective approach to the former. A minority of physicians, who tend to be more religious, still endorse directive counsel with respect to morally controversial clinical decisions. Both patients and PCPs should be aware of these findings when negotiating clinical decisions, and further study is needed to understand how these physician-level differences impact the decisions patients ultimately make regarding morally controversial clinical practices.

Acknowledgements

The authors gratefully acknowledge the feedback of Michael Combs, Robert Stern, and Helen Shin on an earlier version of this manuscript.

This study was supported by grants from the John Templeton Foundation and the National Center for Complementary and Alternative Medicine (1 K23 AT002749, to Dr. Curlin).

Conflict of Interest

The authors declare that they do not have a conflict of interest.

Copyright information

© Society of General Internal Medicine 2013