Dying for Advice: Code Status Discussions between Resident Physicians and Patients with Advanced Cancer

Abstract

Background:

Patients with treatment-resistant advanced cancer rarely benefit from cardiopulmonary resuscitation (CPR) but infrequently discuss end-of-life care with physicians until hospitalized. Admitting resident physicians may conduct initial code status discussions, but may elicit patients' preferences without providing necessary guidance.

Objective:

We surveyed residents' approach to code status discussions to identify barriers hindering informed decision making.

Methods:

We developed an online case-based survey and enrolled subjects via e-mail requests to program directors. U.S. internal medicine residents (n=358; response rate 22.0%) from 19 programs participated. We measured respondents' likelihood of discussing prognosis and benefit of CPR, likelihood of offering code status recommendations, preference for discussing code status independent of attending physicians, and satisfaction with end-of-life discussions.

Results:

Nearly all residents felt CPR would be unhelpful. Yet fewer than half (46.7%) were likely to discuss an estimate of prognosis and the value of CPR with the patient. Only 30% were likely to offer a recommendation on CPR. A majority (69%) of residents who were unwilling to offer a recommendation stated that deference to patient autonomy prevented them from providing guidance. Residents preferred to discuss code status independent of attendings, primarily due to a sense of responsibility. Ultimately, only a minority was satisfied with end-of-life discussions.

Conclusion:

U.S. internal medicine resident physicians are unlikely to discuss prognosis or offer recommendations on CPR in treatment-refractory cancer principally because of a conflict with their concept of patient autonomy. Given the futility associated with CPR in this setting, these data define an unmet need in training and practice.

Introduction

Cardiopulmonary resuscitation (CPR) offers extremely little chance of survival to patients with advanced treatment-refractory cancers^1,2—as little as a one-day median postresuscitation survival.³ Such patients routinely overestimate their life expectancy.⁴ Patients typically rely on physicians to initiate end-of-life conversations,⁵ and a large majority of seriously ill hospitalized patients prefer to make end-of-life decisions with physician advice.⁶ Thus, physicians must provide information and guidance on end-of-life decision making. Discussing prognosis may actually improve coping with illness without decreasing hopefulness or survival.⁷ End-of-life planning also reduces the aggressiveness of care and increases hospice usage.⁸

Despite recommendations,⁹ a minority of outpatients with advanced cancer has a documented code status.¹⁰ Therefore, the first code status discussion may occur upon hospitalization for complications. Limited data address whether resident physicians, frequently responsible for such discussions at teaching hospitals, advise seriously ill patients on end-of-life decision making.^11,12 To address this question we surveyed U.S. internal medicine residents to assess willingness to share prognostic information and recommendations on code status, given a hypothetical patient hospitalized for complications of treatment-resistant advanced cancer.

Methods

The Tufts Medical Center institutional review board reviewed this project. All participants gave informed consent. We developed an online survey using Qualtrics (Provo, UT). Table 2 details the hypothetical case and survey questions. In Part 1 of the survey we asked respondents about their likelihood of sharing prognostic information and providing a recommendation on code status in the hypothetical case. Next we asked whether respondents felt that attendings should be primarily responsible for conducting code status discussions. In Part 2 we asked general questions about code status discussions, including frequency of conducting them and satisfaction with both one's own discussions and with attendings' discussions.

The target population was the approximately 25,000 internal medicine residents at all 387 U.S. programs accredited by the Accrediation Council for Graduate Medical Education (ACGME).¹³ Using contact information obtained from ACGME and program websites, we e-mailed every program director with a request to participate by forwarding the online survey to their residents. We followed up with nonresponders and called programs with unlisted or nonfunctional addresses. We analyzed data with Stata statistical software version 12.1 (StataCorp; College Station, TX) and tested hypothesized correlations between variables. Because our dependent variables were ordered categorical variables and dichotomous variables, we used appropriate linear regression models, the ordered logit and logit models, respectively.¹⁴

Results

Of the 387 programs we contacted, 22 program directors responded: 19 agreed to participate and 3 declined (see Fig. 1). The participating programs represented 1627 residents and varied widely in size (median size 96, mean 85.6, standard deviation 49) and geographic region. A total of 375 respondents began the survey and 358 completed it. We present all recorded responses, including 17 incomplete surveys. Based on completed surveys, the response rate was 22.0%. Table 1 displays demographic information.

FIG. 1.

Study enrollment.

Table 1.

Respondent Demographics

	n	%
Q7. What is your level of training?
PGY 1	135	38.2%
PGY 2	100	28.3%
PGY 3	105	29.8%
PGY 4	13	4.0%
Total	358
Q8. What is your gender?
Male	188	53.3%
Female	165	46.7%
Total	358
Q9. What is your career interest?
Prelim/Transitional	25	7.0%
Primary care	90	25.1%
Q9a. Specialty	240	67.9%
Allergy/Immunology	7	2.0%
Cardiology	52	14.4%
Endocrinology	10	2.8%
Gastroenterology	29	8.2%
Geriatrics	4	1.1%
Hematology/Oncology	42	11.9%
Infectious Disease	17	4.8%
Nephrology	19	5.4%
Palliative Care	48	13.6%
Pulmonology/Critical Care	6	1.7%
Rheumatology	6	1.7%

Table 2 presents survey responses and distributions. Following the hypothetical case presentation, less than half (46.7%) of respondents reported they were likely or very likely to share prognostic information. Over two-thirds (69.6%) reported they were neither likely or very likely to offer a recommendation on code status.

Table 2.

Survey and Responses

(This is the only case you will encounter.)

AB is a patient with stage IV lung adenocarcinoma with brain metastases who was diagnosed 10 months ago and has had disease progression despite first- and second-line therapy.

AB presents to the ED one night with dyspnea and is admitted to your service. Lungs are diffusely rhonchorous to auscultation with decreased breath sounds in the right lung base. Chest CT shows right upper lobe opacity consistent with known primary tumor and a new right-sided pleural effusion.

While doing the admission you note that AB is slightly tachypneic but speaking in complete sentences and maintaining oxygen saturation at 94% on two liters via nasal cannula. You decide to ask AB about code status given the possibility of respiratory failure. AB has not discussed code status with a physician before.

Please answer the following questions about your interaction with AB.

Part I.
Q1. While discussing code status, I would tell the patient (a) my estimate of his/her prognosis and (b) my estimate of how successful CPR would be in the event of cardiopulmonary arrest.
	Very Unlikely	Unlikely	Somewhat unlikely	Undecided	Somewhat likely	Likely	Very likely	Total
%	5.6%	11.7%	10.7%	3.5%	21.9%	26.7%	20.0%
n	21	44	40	13	82	100	75	375
Q2. I would make a recommendation about code status.
	Very Unlikely	Unlikely	Somewhat unlikely	Undecided	Somewhat likely	Likely	Very likely	Total
%	14.1%	14.9%	12.3%	8.8%	19.5%	19.7%	10.7%
n	53	56	46	33	73	74	40	375
Q2a. (If at all likely to make a recommendation in Q2)
Would you recommend …
	Full code	DNR/DNI
%	6.5%	93.5%
n	12	172
Q2b. (If at all unlikely to make a recommendation in Q2)
Tell us more about why you would not want to make a recommendation.
				n	%
I don't know enough about the patient's disease to make a recommendation.				24	15.5%
The patient or his/her family might not want me to.				26	16.8%
I don't know how likely it is for CPR to successfully resuscitate the patient.				14	9.0%
The attending or primary oncologist might not want me to make a recommendation.				41	26.5%
CPR offers a reasonable chance of successful resuscitation.				2	1.3%
The patient should make the decision about code status without my influence.				107	69.0%
Other				13	8.4%
Q3. Unless it's an emergency, the attending physician or primary oncologist—not I—should discuss code status with AB.
	Strongly disagree	Disagree	Neither agree nor disagree	Agree	Strongly agree	Total
%	26.4%	40.4%	12.6%	14.3%	6.3%
n	96	147	46	52	23	375
Q3a. (If agrees at all with Q3)
Tell us more about why you agree.
				n	%
A more experienced physician will provide more accurate information about prognosis of the disease and the value of CPR.				25	33.3%
The primary oncologist will have known the patient for longer so can better estimate prognosis.				50	66.7%
The attending or primary oncologist would not want me to discuss code status with the patient.				6	8.0%
The patient will more likely respect the opinion of an attending or primary oncologist.				25	33.3%
Other				9	12.0%
Q3b. (If disagrees at all with Q3)
Tell us more about why you disagree.
				n	%
Discussing code status is part of my responsibility as a physician.				227	93.4%
I have enough training and knowledge to discuss code status with the patient.				160	65.8%
The attending or primary oncologist is unlikely to discuss code status with the patient.				42	17.3%
I probably would disagree with the attending's or primary oncologist's estimate of prognosis.				6	2.5%
The attending or primary oncologist is unlikely to be honest with the patient about prognosis.				12	4.9%
Other				16	6.6%
Part II.
Q4. Do you routinely ask patients about code status on admission?
				n	%
I almost always ask, regardless of diagnosis, prognosis, or reason for admission.				178	49.2%
I ask only for patients who I think have at least a small chance of needing intubation or cardiac resuscitation.				93	26.3%
I ask only for patients who I think have a substantial chance of needing intubation or cardiac resuscitation, i.e., a critically ill patient.				49	13.6%
I usually don't ask unless the patient volunteers it or I see it documented previously.				7	2.0%
My electronic medical record or admission order set requires documentation of code status.				33	8.8%
Total				360
Q5. How satisfied are you with the discussions of code status that YOU have had with your patients with advanced cancer?
	Very dissatisfied	Dissatisfied	Somewhat dissatisfied	Neutral	Somewhat satisfied	Satisfied	Very satisfied	Total
%	0.6%	4.8%	10.3%	18.9%	30.9%	29.3%	5.3%
n	2	17	37	68	111	105	19	359
Q6. How satisfied are you with the discussions of code status that ATTENDINGS or PRIMARY ONCOLOGISTS have had with your patients with advanced cancer?
	Very dissatisfied	Dissatisfied	Somewhat dissatisfied	Neutral	Somewhat satisfied	Satisfied	Very satisfied	Total
%	7.5%	10.6%	12.0%	18.1%	20.3%	22.8%	8.6%
n	27	38	43	65	73	82	31	359

We explored respondents' reasoning in follow-up questions, allowing for multiple answers and free-text answers. Of the 155 respondents who declined to recommend a code status, many (69%) thought the patient should make the decision without the resident's influence. One-quarter (26.5%) felt the attending might not want the resident to make a recommendation. Free-text answers included endorsements of patient autonomy:

“Code status is supposed to be a decision a patient makes for him/herself.”

Others were hesitant about giving advice without being asked or making a recommendation during an initial meeting:

“If asked I would offer my opinion but not unprompted.”

“I may not have that rapport to make such a recommendation [during the first meeting] but would consider it if I already had a [prior] relationship.”

Importantly, very few (1.3%) felt that CPR offered a reasonable chance of resuscitation. Likewise, the 187 respondents who reported they would recommend a code status overwhelmingly recommended “do not resuscitate/do not intubate” (DNR/DNI).

A majority of respondents (66.8%) felt they should discuss code status themselves while a minority (20.6%) preferred to defer to an attending. The 75 respondents who preferred an attending discussion most commonly believed that the primary oncologist would have known the patient for longer and would be better able to estimate prognosis. Free-text answers included statements of support for the primary oncologist's preexisting relationship with the patient:

“The primary oncologist will know the patient and have a trusting relationship, which is the ideal setup for a difficult yet crucial conversation.”

Of the 243 respondents who preferred to discuss code status, the vast majority (93.4%) felt that this discussion was part of their responsibility as a physician, and a majority (65.8%) felt that they had enough training and knowledge to do so.

Overall, only 34.2% were satisfied or very satisfied with their discussions. But a very small minority (5.4%) was dissatisfied or very dissatisfied, while the dissatisfied proportion was significantly higher when it came to attendings' discussions (18.1%, p<0.01).

We tested for hypothesized correlations between responses, as shown in Table 3. Unsurprisingly, more-senior residents were more likely to share prognostic information and offer recommendations. Interestingly, those planning on pursuing hematology/oncology or palliative care training were more likely to share prognostic information, but were not more likely to provide recommendations. This is an important test of response bias, since these respondents are presumably above-average in interest in end-of-life discussions.

Table 3.

Correlations between Resident Characteristics and Discussion Practices

Dependent variable	Independent variable	p value	Direction of correlation
Likelihood of sharing prognosis and giving an estimate of the value of CPR	Frequency of discussing code status with inpatients	>0.050
	Training year	0.002	Positive
	Interest in Hematology/Oncology or Palliative Care	0.026	Positive
Likelihood of making a recommendation on code status	Likelihood of sharing prognosis and giving an estimate of the value of CPR	0.001	Positive
	Frequency of discussing code status with inpatients	>0.050
	Satisfaction with one's own discussions of code status	0.355
	Training year	0.002	Positive
	Interest in Hematology/Oncology or Palliative Care	0.213
Preference to discuss code status oneself	Likelihood of sharing prognosis and giving an estimate of the value of CPR	0.551
	Likelihood of making a recommendation on code status	0.145
	Satisfaction with one's own discussions of code status	<0.001	Positive
	Satisfaction with attendings' discussions of code status	0.840
	Training year	0.235
	Female gender	0.052
	Interest in Hematology/Oncology or Palliative Care	0.890
Satisfaction with one's own discussions of code status	Training year	0.046	Negative
Satisfaction with attendings' discussions of code status	Training year	<0.001	Negative

Discussion

By discussing end-of-life planning with patients with limited life expectancy, physicians uphold several bioethical principles: respect for patient autonomy through documentation of wishes, beneficence by decreasing end-of-life distress, and nonmaleficence in avoiding futile care. By focusing on documentation of wishes alone, residents may not fulfill the professional responsibility of physicians to guide their patients through difficult medical decisions.¹⁵ Incomplete discussions can cause, at minimum, improper documentation of patients' wishes and, at most, psychological harm, damage to the physician-patient relationship, and the potential for unwanted and futile attempts at resuscitation.¹⁶

In this survey, U.S. internal medicine residents knew the limitations of CPR in patients with advanced treatment-refractory cancer, yet were frequently reluctant to provide the information on prognosis and treatment outcome necessary to make informed decisions on code status. Less than one-third of respondents reported that they were likely to provide a recommendation on code status, largely attributable to an emphasis on the concept of patient autonomy. Nonetheless, residents did prefer to retain responsibility for code status discussions, primarily due to a sense of physician responsibility, as seen previously.¹⁷ Whether justified or not, they did not feel underprepared to conduct such discussions, contrasting with prior reports.¹⁸

Only one third of residents were substantially satisfied with the code status discussions that they or their attendings conduct in this common scenario. This discontent may result from tension between residents' strong sense of the futility of CPR in advanced cancer and their inability to communicate this opinion because of a priority given to patient autonomy instead of informed decision making. Residents' dissatisfaction may also communicate to medical students.¹⁹ Given an anticipated rise in cancer incidence²⁰ and deaths²¹ as well as the associated increase in end-of-life care costs,²² end-of-life communication is of increasing importance,²³ and these findings suggest an unmet need in training and practice.

There are several strengths of this study. This is the first nationwide survey that illuminates residents' self-assessed attitudes, skills, and practice in treatment-refractory cancer, the major emerging cause of worldwide mortality. The survey sample included participants from across the United States and from programs of varying size, allowing good external validity. Although most of the program directors we contacted did not respond, the response rate among residents we did reach allowed for generalizability and was comparable to other resident surveys.^19,24

Our study is limited by problems of self-report and response bias. Response bias is of particular concern with a low response rate, although our within-sample test of response bias was negative. Both self-report and response bias may cause us to overestimate the actual likelihood of providing prognostic information or recommendations. However, this bolsters our conclusion that residents provide insufficient guidance. Our findings also cannot be generalized to residents outside the United States, who may receive different training and experience different attitudes on end-of-life decision making.

Ideally, attending physicians should discuss advance planning early in the disease course, but in practice, residents are often the first to document code status while admitting patients who have developed complications. Previous authors have argued that prognostication and making recommendations should be taught as medical competencies but acknowledge that attendings often fail to model such skills.²⁵ Indeed, in a study of attending hospitalists discussing code status, none shared an estimate of the benefit of CPR or provided a recommendation about code status.²⁶ Furthermore, 41 of our respondents felt that attendings would discourage them from offering recommendations. Physicians, residents, and attendings alike should be expected to provide guidance on the utility of medical interventions, including CPR. If unsure of what to recommend, they should seek guidance from expert resources, which may in turn encourage oncologists to discuss end-of-life planning early.

Footnotes

Acknowledgments

We wish to thank the Tufts Medical Center residents who provided feedback on a pilot survey and Dr. Michael Fisch for comments on this manuscript.

Author Disclosure Statement

No competing financial interests exist.

References

Landry

, Parker

, Phillips

: Outcome of cardiopulmonary resuscitation in the intensive care setting. Arch Intern Med, 1992; 152:2305–2308.

Ewer

, Kish

, Martin

, et al.: Characteristics of cardiac arrest in cancer patients as a predictor of survival after cardiopulmonary resuscitation. Cancer, 2001; 92:1905–1912.

, Hong

, Naing

, et al.: Outcome analyses after the first admission to an intensive care unit in patients with advanced cancer referred to a phase I clinical trials program. J Clin Oncol, 2011; 29:3547–3552.

Weeks

, Catalano

, Cronin

, et al.: Patients' expectations about effects of chemotherapy for advanced cancer. N Engl J Med, 2012; 367:1616–1625.

Hanson

, Danis

, Garrett

: What is wrong with end-of-life care? Opinions of bereaved family members. J Am Geriatr Soc, 1997; 45:1339–1344.

Heyland

, Tranmer

, O'Callaghan

, et al.: The seriously ill hospitalized patient: Preferred role in end-of-life decision making?. J Crit Care, 2003; 18:3–10.

Mack

, Smith

: Reasons why physicians do not have discussions about poor prognosis, why it matters, and what can be improved. J Clin Oncol, 2012; 30:2715–2717.

Mack

, Cronin

, Keating

, et al.: Associations between end-of-life discussion characteristics and care received near death: A prospective cohort study. J Clin Oncol, 2012; 30:4387–4395.

National Comprehensive Cancer Network: Practice guidelines in oncology: Palliative care. National Comprehensive Cancer Network. Fort Washington, PA, 2014. www.nccn.org/professionals/physician_gls/pdf/palliative.pdf. (Last accessed March 5, 2014.)

10.

Temel

, Greer

, Admane

, et al.: Code status documentation in the outpatient electronic medical records of patients with metastatic cancer. J Gen Intern Med, 2009; 25:150–153.

11.

Tulsky

, Chesney

, Lo

. How do medical residents discuss resuscitation with patients?. J Gen Intern Med, 1995; 10:436–442.

12.

Deep

, Griffith

, Wilson

. Communication and decision making about life-sustaining treatment: Examining the experiences of resident physicians and seriously-ill hospitalized patients. J Gen Intern Med, 2008; 23:1877–1882.

13.

American Board of Internal Medicine: Number of programs and residents. American Board of Internal Medicine. Philadelphia, PA, 2014. www.abim.org/about/examInfo/data-res/chart-01.aspxj. (Last accessed April 2014 .)

14.

Ananth

, Kleinbaum

: Regression models for ordinal responses: A review of methods and applications. Int J Epidemiol, 1997; 26:1323–1333.

15.

Quill

, Brody

: Physician recommendations and patient autonomy: Finding a balance between physician power and patient choice. Ann Intern Med, 1996; 125:763–769.

16.

Tulsky

, Chesney

, Lo

: See one, do one, teach one? House staff experience discussing do-not-resuscitate orders. Arch Intern Med, 1996; 156:1285–1289.

17.

Siddiqui

, Holley

: Residents' practices and perceptions about do not resuscitate orders and pronouncing death: An opportunity for clinical training. Am J Hosp Palliat Care, 2011; 28:94–97.

18.

Schwartz

, Goulet

, Gorski

, et al.: Medical residents' perceptions of end-of-life care training in a large urban teaching hospital. J Palliat Med, 2003; 6:37–44.

19.

Billings

, Engelberg

, Curtis

, et al.: Determinants of medical students' perceived preparation to perform end-of-life care, quality of end-of-life care education, and attitudes toward end-of-life care. J Palliat Med, 2010; 13:319–326.

20.

Smith

, Smith

, Hurria

, et al.: Future of cancer incidence in the United States: Burdens upon an aging, changing nation. J Clin Oncol, 2009; 27:2758–2765.

21.

Ferlay

, Soerjomataram

, Ervik

, et al.: GLOBOCAN 2012 v1.0, Cancer incidence and mortality worldwide. International Agency for Research on Cancer, Lyon, France, 2012. globocan.iarc.fr. (Last accessed April 21, 2014.)

22.

Chastek

, Harley

, Kallich

, et al.: Health care costs for patients with cancer at the end of life. J Oncol Pract, 2012; 8:75s–80s.

23.

Zhang

, Wright

, Huskamp

, et al.: Health care costs in the last week of life: Associations with end-of-life conversations. Arch Intern Med, 2009; 169:480–488.

24.

Drolet

, Spalluto

, Fischer

: Residents' perspectives on ACGME regulation of supervision and duty hours: A national survey. N Engl J Med, 2010; 363:e34.

25.

Lamas

, Rosenbaum

: Freedom from the tyranny of choice: Teaching the end-of-life conversation. N Engl J Med, 2012; 366:1655–1656.

26.

Anderson

, Chase

, Pantilat

, et al.: Code status discussions between attending hospitalist physicians and medical patients at hospital admission. J Gen Intern Med, 2011; 26:359–366.

Dying for Advice: Code Status Discussions between Resident Physicians and Patients with Advanced Cancer—A National Survey

Abstract

Abstract

Background:

Objective:

Methods:

Results:

Conclusion:

Introduction

Methods

Results

Discussion

Footnotes

Acknowledgments

Author Disclosure Statement

References