Influence of Institutional Culture and Policies on Do-Not-Resuscitate Decision Making at the End of Life

Importance  Controversy exists regarding whether the decision to pursue a do-not-resuscitate (DNR) order should be grounded in an ethic of patient autonomy or in the obligation to act in the patient’s best interest (beneficence).

Objective  To explore how physicians’ approaches to DNR decision making at the end of life are shaped by institutional cultures and policies surrounding patient autonomy.

Design, Setting, and Participants  We performed semistructured in-depth qualitative interviews of 58 internal medicine physicians from 4 academic medical centers (3 in the United States and 1 in the United Kingdom) by years of experience and medical subspecialty from March 7, 2013, through January 8, 2014. Hospitals were selected based on expected differences in hospital culture and variations in hospital policies regarding prioritization of autonomy vs best interest.

Main Outcomes and Measures  This study identified the key influences of institutional culture and policies on physicians’ attitudes toward patient autonomy in DNR decision making at the end of life.

Results  A hospital’s prioritization of autonomy vs best interest as reflected in institutional culture and policy appeared to influence the way that physician trainees conceptualized patient autonomy. This finding may have influenced the degree of choice and recommendations physician trainees were willing to offer regarding DNR decision making. Trainees at hospitals where policies and culture prioritized autonomy-focused approaches appeared to have an unreflective deference to autonomy and felt compelled to offer the choice of resuscitation neutrally in all situations regardless of whether they believed resuscitation to be clinically appropriate. In contrast, trainees at hospitals where policies and culture prioritized best-interest–focused approaches appeared to be more comfortable recommending against resuscitation in situations where survival was unlikely. Experienced physicians at all sites similarly did not exclusively allow their actions to be defined by policies and institutional culture and were willing to make recommendations against resuscitation if they believed it would be futile.

Conclusions and Relevance  Institutional cultures and policies might influence how physician trainees develop their professional attitudes toward autonomy and their willingness to make recommendations regarding the decision to implement a DNR order. A singular focus on autonomy might inadvertently undermine patient care by depriving patients and surrogates of the professional guidance needed to make critical end of life decisions.

Although beneficence, the obligation to act in the patient’s best interest, traditionally embodied the primary obligation of health care professionals, this approach has given way in recent decades to the primacy of patient autonomy.^1-4 Some worry that the pendulum has swung too far by replacing concerns of paternalism, in which a person’s preferences are intentionally overridden, with simplistic understandings of autonomy that “abandon patients to their autonomy.”^{5(p 446)} This attitude is problematic in situations in which patients are overwhelmed by, do not want to, or are not able to choose from the menu of different options. These situations may result in patients choosing treatments that are neither in their best interest nor consistent with their goals and values.

Some health care professionals believe that to respect autonomy, they must present patients with all possible options and refrain from expressing recommendations.⁶ This belief stands in contrast to the guidance of most professional organizations, which state that futile therapies should not be offered and recommendations consistent with the patient’s goals and values should be addressed in a resuscitation status discussion to honor patient autonomy.^7-9 Futility is typically defined as an inability to regain consciousness or survive outside the intensive care unit (ICU) setting.³ Navigating the balance between beneficence and autonomy can be especially fraught at the end of life and in particular regarding resuscitation decisions.

Studies^10,11 have found that survival to discharge after cardiopulmonary resuscitation (CPR) is highly unlikely in certain conditions, such as sepsis or metastatic cancer, and is near zero when irreversible dying is imminent. Offering CPR when not clinically indicated can cause harm.^12,13 On the other hand, leaving such decisions completely in the hands of physicians may lead to biased decision making partially based on unconscious factors (eg, patient race, age, and socioeconomic status).¹⁴ Some have argued that do-not-resuscitate (DNR) decisions should be tailored to 3 distinct patient populations: (1) those whom CPR should be considered a plausible option, (2) those whom CPR should be recommended against, or (3) those whom CPR should not be offered (ie, those imminently dying or who have no chance of surviving to discharge).¹⁵ This framework embraces the notion that providing clear recommendations is a part of a physician’s duty to act with beneficence and thereby empowers rather than hinders the patient’s ability to make an informed, autonomous decision.

The concept of informed assent operationalizes this best-interest framework, in which the health care professional does not insist that the decision ultimately be made by the patient or family member and explicitly recommends against CPR when it is clearly not indicated.¹⁶ National and local policies differ between and within the United States and the United Kingdom. Whereas US institutions tend to prioritize autonomy and individual choice, UK institutions promote best-interest decision making guided primarily by the physician, who has the unilateral authority to withhold CPR when deemed appropriate.^17-19

Studies^20-23 have found that institutional cultures surrounding advanced care planning contribute to variation in care provided by hospitals in end-of-life ICU care. Others^22,24 have found considerable variation in use of DNR orders and withdrawal of life-sustaining therapies, even after accounting for variations in patient characteristics. The aim of this qualitative study was to explore how institutional cultures and policies surrounding patient autonomy influence physicians’ clinical approaches to DNR decision making at the end of life.

Semistructured in-depth interviews performed from March 7, 2013, through January 8, 2014, investigated physicians’ views regarding the influence of institutional culture and policies on physicians’ attitudes toward choice and autonomy in the DNR decision-making process. This qualitative study was exploratory in nature, intended to deepen conceptual understanding of underlying phenomena that drive physician attitudes and behavior. The theories developed emerged from initial interviews and analysis and were refined and validated in subsequent interviews.

We purposively sampled 3 large academic medical centers with well-established residency and fellowship teaching programs in urban cities in the United States (hospitals A, B, and C) and 1 in the United Kingdom (hospital D) based on expected differences in hospital culture and variations in hospital policies (Table 1). The policies and culture of hospitals A and B reflect prioritization of patient autonomy, whereas the policies and culture of hospitals C and D prioritized best-interest decision making. The initial intention of the first 2 hospitals selected (hospitals B and D) was to explore differences between the United States and the United Kingdom and to contrast autonomy vs best-interest–focused policies. Subsequently, 2 additional hospitals were added in the United States to gain further insight into this evolving theoretical framework. These hospitals were chosen based on variations on end-of-life care.³¹ In addition, the prior work experience of one of the authors (E.D.) in hospitals A and B, the observation of differing policies and cultures, and known differences in norms regarding the end of life, such as the Death with Dignity movement in the Pacific Northwest, informed the selection of hospitals.³²

During a 9-month period, one investigator (E.D.) recruited and interviewed 58 internal medicine physicians (including internal medicine subspecialists) in person, with the exception of 2 interviews that were conducted via Skype. Participants were eligible if they were full-time physicians at the selected institutions who were routinely involved in DNR conversations with patients at the end of life. They were purposively sampled by stage of training, years of experience, and medical subspecialty to provide a wide range of perspectives and contribute to the understanding of emerging patterns and themes. Physicians were excluded if they had not been practicing in the United States or the United Kingdom for most of their clinical career. Recruitment occurred through group e-mail advertisements, individual solicitations, and personal referrals.

Written informed consent was obtained from all interviewees, and interview data were anonymized during transcription. The study was approved by the Johns Hopkins University Institutional Review Board and the UK National Health Service National Research Ethics Service.

An interview guide used across all sites provided thematic continuity (the eAppendix in the Supplement lists the sample interview guide questions). However, the interview format was open-ended, thereby encouraging participants to explore those aspects they considered most relevant. Interviews lasted 45 to 120 minutes and were audiotaped and transcribed verbatim. Data collection concluded when we reached theoretical saturation, a point where no new themes arose from the interviews.

Our qualitative approach is grounded in a framework that acknowledges that multiple perspectives are intrinsic to the research process and the particular importance of the perspective that the researchers bring to the fieldwork and analysis.^33-35 Throughout the analyses, the researchers drew on their own clinical experiences in a reflexive manner, understanding how these experiences would inform and potentially bias their interpretation of the interview data. Data were analyzed and theories developed as more interviews were conducted and coded. Developed hypotheses and themes became the subject of questions in subsequent interviews to further confirm the trustworthiness of the data. Disconfirming cases were recognized and analyzed in light of their effect on the emerging theory.

Two independent readers (E.D., a physician and sociologist, and A.C., a medical student with experience in bioethics) identified initial key themes and words that occurred through a subset of the interviews and developed a codebook through an iterative process. They subsequently coded 20% of the interviews after meeting to discuss emerging themes and patterns. There was rarely disagreement among coders. One researcher (E.D.) then analyzed and coded the remaining interviews using the codebook and added additional themes and adapting categories as needed.

A total of 13 to 16 physicians participated at each site, with approximately equal numbers of attending physicians, fellows, and residents (or UK equivalent) at each site interviewed. Years of experience ranged from 1 to 45 years (Table 2).

At each hospital, physicians’ ethical attitudes toward DNR decision making reflected its hospital’s policy position regarding its prioritization of autonomy vs best interest. In general, physicians at the 2 hospitals (A and B) whose policies prioritized autonomy believed that patient autonomy was more important than other ethical principles (Table 3). In contrast, at hospitals whose policies emphasized best-interest decision making (hospitals C and D), physicians were more likely to believe that they had the ability to make clinical recommendations that respected both autonomy and the patient’s best interest or prioritized best interest as a whole. Most physicians at hospital C were comfortable using the informed assent approach when appropriate.

Notably, hospital C’s DNR policy begins with a determination of futility before the section on eliciting patient preferences. In this section, the policy lists that likelihood of successful resuscitation is extremely low with preexisting conditions, such as metastatic cancer and sepsis, and emphasizes that “there is consensus that CPR would not be effective or even if successful would not be beneficial to the patient. There are some patients in whom CPR should not be attempted.”^{37(p 1)} This language emphasizes the importance of the physician’s clinical judgment in determining DNR status.

Experienced physicians at all sites generally were comfortable engaging in best-interest decision making and, when clinically appropriate, not offering or making explicit recommendations against offering resuscitation (Table 3). Respondents stated that they would discuss the situation with the patient or surrogate from a goal of care perspective and recommend against resuscitation or focus on comfort care without explicitly mentioning resuscitation. Many of them expressed the belief that failure to give recommendations should be seen as a derogation of responsibility. As one attending physician at hospital B stated, “I would recommend the following and the patient decides. I think that’s autonomy. I think to say, ‘You could be resuscitated or not, what would you like to do?’ seems to me to be an abdication of responsibility to the patient in the name of autonomy. As the patient reminds me often, ‘You’re the doctor, how do I know?’”

Experienced physicians at hospitals that prioritized autonomy recognized the primacy of autonomy at their institution but did not believe autonomy meant they should offer futile treatments that would not serve the best interest of their patients. These physicians remained committed to what they believed was ethically the right thing to do by finding ways to balance autonomy and the law and making unilateral clinical decisions when they thought they were appropriate. This statement from a senior physician at hospital A demonstrated this balance: “I have to do it because I cannot break the law. You do your best and try to minimize the amount of ethical or moral compromise by limiting the attempted resuscitation as much as possible … When it comes to a procedure that I know far better than the patient or the family, to question whether it is appropriate for me not to exercise my judgment, I think would be wrong. It is just the opposite. I would call it being a responsible doctor and exercising good judgment … My primary responsibility is to the patient, and if a family is asking me to do something that clearly is harmful to the patient, I will explain to them patiently why I don’t think [it’s right.] I am not going to let somebody suffer because of it.”

Several respondents acknowledged a great degree of variation among physicians in terms of skill, attitudes, and practices, although many also expressed that their own views were relatively mainstream.

Although hospital policies primarily dictated whether patient preferences should be honored and none of the hospitals required physicians to offer the choice of resuscitation if it were unlikely to work, trainees at hospitals where policies reflected autonomy-focused approaches often interpreted policies to mean that they were obligated to offer a choice of resuscitation in all cases regardless of the patient’s likelihood of survival and that they should not provide clinical recommendations (Table 3).

Trainees were less comfortable recommending that resuscitation be withheld even if the chance of successful resuscitation were negligible. One resident at hospital A thought it was inappropriate to “inject your own personal opinions of life and death.” To another resident, respecting autonomy meant providing information without biasing the patient with a recommendation and that “it’s my responsibility … to do everything as they say” (hospital B). Less experienced physicians at autonomy-focused hospitals expressed their responsibility for decision making up to a certain point but appeared to be more hesitant to take on that responsibility.

In contrast, trainees at hospitals with policies that encouraged a best-interest decision-making approach (hospitals C and D) were more comfortable expressing clinical judgment against resuscitation when appropriate. Many of the trainees at hospital C were willing to use informed assent and strongly recommend against resuscitation or did not offer resuscitation if it was unlikely to work. One resident said, “I can think of a few times per rotation where informed assent was done. I did it at least a few times in the ICU, where if things head towards CPR, then I’d say, ‘It wouldn’t be something that we think would be appropriate for you and here is why.’”

A hospital’s culture and policies’ prioritization of autonomy vs best interest appears to influence the way in which physician trainees conceptualize patient autonomy. This finding may subsequently influence the degree of choice and recommendations physician trainees are willing to offer regarding DNR decision making. There is a complex interplay between institutional cultures and policies; the ability to parse out the exact interaction is beyond the scope of this article. Policies and cultures complement and drive each other.

Although experienced physicians in this study acknowledged institutional culture and policies, it appears they did not exclusively allow their actions to be defined by policies and were willing to make recommendations against resuscitation if they believed this procedure would be futile. Although variation existed among respondents, in general, experienced physicians at all sites understood and acted on their ethical obligations to recommend and offer treatments responsibly.

Although hospital policies, even at the most autonomy-focused institutions, do not require physicians to offer futile resuscitation or to withhold recommendations, we found that less experienced physicians at autonomy-focused hospitals tended to overinterpret hospital policies and embrace a reductionist, limited understanding of autonomy. Ethics and communications at the end of life are generally present in the medical school and residency didactic curricula, but lack of time and space on the wards for reflection might promote an excessive, unreflexive deference to patient autonomy that risks placing unwanted and unreasonable responsibility for medical decision making onto the patient or surrogate.^6,38,39

Unlike more senior physicians, trainees have not yet developed the experience or professional confidence to advocate for their patient’s best interest by making explicit recommendations regardless of their institution’s policy or cultural constraints. Trainees at autonomy-focused hospitals appeared compelled to remain neutral when offering the choice of resuscitation without providing a specific recommendation regardless of whether they believed it to be clinically appropriate. Their attitudes might reflect perceived constraints because of policies and expectations, which focus on patient autonomy. This finding suggests that the ethical principles embraced by experienced physicians were not transmitted to learners at autonomy-focused institutions.

The culture and policies at hospitals C and D actively encouraged best-interest decision making. Although some of the physicians at hospital C had never heard of the term informed assent, the cultural norm at that institution—similar to the hospital in the United Kingdom—was not only that it was appropriate to discourage ineffective CPR but also that it would be ethically suspect to offer CPR in futile situations. These trainees appeared to have the time and space within their workflow and culture to develop similar attitudes as experienced physicians. Even though interns and residents come from medical schools across the country, they seemed to adopt similar attitudes as their peers in residency training.

Studies^40,41 have found that, in practice, physicians and residents do not provide adequate information during conversations regarding resuscitation. Most omit critical information on prognosis, goals of care, or explanations of the processes involved in various interventions. Although medical residents in these interviews generally had a great respect for autonomy, omission of these critical elements may deprive patients of the very autonomy that physicians strive to respect.^38,39

Interventions to improve end-of-life communication skills have focused on formal curricula and increasing opportunities for house staff to practice these skills.⁴² Several studies^43-46 have explored the effects of a hidden curriculum on the development of medical trainees and its effect on their ethical beliefs and communication strategies. This hidden curriculum captures the inadvertent transmission of attitudes and behaviors to trainees through everyday interactions during their training that are in conflict with the normative ideals and official policies of the medical profession.⁴⁷ Because much of residency training involves apprenticeship-based peer learning, the influence of the cultural milieu cannot be overemphasized.

It is often perceived that the United States and the United Kingdom are on opposite ends of the autonomy and best-interest spectrum, with the greater focus in the United States on individual choice and patient rights and the culture in the United Kingdom reflecting a greater acceptance of more paternalistic practices.⁴⁸ This study suggests that there may be important differences in approaches not only between the United States and the United Kingdom but within the United States itself. Trainees at the US hospital that encouraged best-interest decision making had attitudes and beliefs about DNR decisions that appeared more similar to UK physicians than physicians at the other US sites. In contrast to trainees at the autonomy-focused hospitals, they seemed more willing to not offer resuscitation if they did not think it was indicated.

This study has several limitations. Interviews with physicians do not reflect actual practices but instead reflect physicians’ opinions, attitudes, and perceptions of clinical practice. Social desirability bias may have influenced answers. There are many internal and external factors that influence physician attitudes and communication practices regarding end-of-life decision making. For example, it is possible that the backgrounds of patients and the cultural milieu in which these hospitals are located influence aggressiveness of care, attitudes toward death, and degree of deference to authority. Furthermore, institutional interest and leadership in palliative care also influences hospital culture surrounding end-of-life issues. These issues emerged in our interviews and will be explored in future articles. Although the difference between more and less experienced physicians’ willingness to offer clinical recommendation could reflect more traditional paternalistic attitudes among older respondents, our study suggests a different explanation because trainees at the best-interest–focused hospital had similar attitudes to experienced physicians in those hospitals.

Inadequacies exist in physician communication at the end of life, especially among medical trainees.⁴⁹ Prior studies^20,21 have found that advanced care norms affect end-of-life ICU use, but, to our knowledge, this is the first study to explore the influence of a hospital’s policy or culture on physicians’ understanding of autonomy and willingness to recommend resuscitation decisions. This study hypothesizes that culture and policies oriented toward autonomy may have unintended consequences of constraining a physician’s willingness to make recommendations, especially that of trainees.

It is ironic that the very policies that seek to empower patients might inadvertently deprive them of their autonomy through an overemphasis on a reductionist checklist approach to autonomy. Perhaps policies more oriented toward best-interest decision making might allow physicians the leeway to shift their focus from a discourse of choice to one of care and compassion.

Accepted for Publication: December 24, 2014.

Corresponding Author: Elizabeth Dzeng, MD, MPH, MPhil, MS, Program in Palliative Care, Johns Hopkins School of Medicine, 600 N Wolfe St, Blalock 359, Baltimore, MD 21287 (dzeng@stanfordalumni.org).

Published Online: April 6, 2015. doi:10.1001/jamainternmed.2015.0295.

Author Contributions: Dr Dzeng had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis

Study concept and design: Dzeng, Roland, Smith, Kelly, Barclay, Levine.

Acquisition, analysis, or interpretation of data: Dzeng, Colaianni, Chander, Smith, Levine.

Drafting of the manuscript: Dzeng, Smith, Kelly, Levine.

Critical revision of the manuscript for important intellectual content: All authors.

Obtained funding: Dzeng, Levine.

Administrative, technical, or material support: Dzeng, Smith.

Study supervision: All authors.

Conflict of Interest Disclosures: None reported.

Funding/Support: This study was funded by training grant T32 HP10025-20 from the Health Resources and Service Administration, the Gates Cambridge Scholarship, and the Ho-Chiang Palliative Care Research Fellowship at the Johns Hopkins School of Medicine (Dr Dzeng).

Role of the Funder/Sponsor: The funding sources had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and the decision to submit the manuscript for publication.

Additional Contributions: J. Randall Curtis, MD, MPH (Division of Pulmonary and Critical Care Medicine, University of Washington), provided institutional support and participant recruitment and commented on early drafts of the manuscript. Eric Larson, MD, MPH (Department of Medicine, University of Washington and Group Health Research Institute), provided insightful comments during the manuscript revision process. Jimmy Hoard, BFA (Palliative Care Center of Excellence, University of Washington), and Jonathan Fuld, MB ChB, PhD (Department of Respiratory Medicine, Addenbrooke’s Hospital), recruited participants. Sydney Dy, MD (Department of Health Policy, Johns Hopkins University), provided mentorship and helpful suggestions. Lucy Burkitt-Gray, BA (Hons) (University of Cambridge), assisted with data management. Zoe Fritz, MB ChB (Department of Acute Medicine, Addenbrooke’s Hospital), participated in discussions at early stages of the project. Lucy Burkitt-Gray was compensated for her contributions; no others listed here were otherwise compensated. We thank the physicians who generously gave their time to participate in the interviews.

Correction: This article was corrected on May 6, 2015, to fix an error in the text and Table 1.