Barriers to communication regarding end-of-life care: perspectives of care providers

Journal of Critical Care (2005) 20, 214 – 223

Barriers to communication regarding end-of-life care: perspectives of care providersB Anjali H. Anselm MDa, Valerie Palda MD, MScb,c,*, Cameron B. Guest MD, MEdd,e, Richard F. McLean MDf, Mary L.S. Vachon RN, PhDg,h, Merrijoy Kelner PhDi, Jenny Lam-McCulloch MScj a

Faculty of Medicine, University of Ottawa, Ottawa, Ontario, Canada K1H 8M5 Department of Medicine, University of Toronto, Toronto, Ontario, Canada M58 1W8 c Department of Health Policy Management and Evaluation, University of Toronto, Toronto, Ontario, Canada M5S 1A8 d Department of Critical Care Medicine, Sunnybrook and Women’s College Health Sciences Centre, Toronto, Ontario, Canada M4N 3M5 e Department of Anaesthesia, Sunnybrook and Women’s College Health Sciences Centre, Toronto, Ontario, Canada M4N 3M5 f Department of Anaesthesia, McMaster University, Hamilton, Ontario, Canada L8N 3Z5 g Department of Psychiatry, University of Toronto, Toronto, Ontario, Canada M5T 1R8 h Department of Health Policy Management and Evaluation, University of Toronto, Toronto, Ontario, Canada M5S 1A8 i Institute for Human Development, Life Course and Aging, University of Toronto, Toronto, Ontario, Canada M5T 3J1 j Department of Anaesthesia, University of Toronto, Toronto, Ontario, Canada M4N 3MS b

Received 29 June 2004; revised 19 March 2005; accepted 31 May 2005

Keywords: Resuscitation decisions; Do-not-resuscitate orders; Terminal care; Health surveys; Communication barriers; Attitude of health personnel

Abstract Objective: Communication regarding end-of-life care is frequently perceived as suboptimal, despite the intent of both health care providers and patients. We interviewed health care providers to determine their perspective regarding these barriers to communication. Materials and Methods: Eleven focus groups with a total of 10 attending physicians, 24 residents, and 33 nurses were convened to explore barriers to end-of-life discussions on the Internal Medicine service at a 600-bed tertiary care hospital in Toronto, Canada. An interview schedule was designed to elicit information regarding the process of end-of-life discussions, barriers to these discussions, and possible interventions for limiting such barriers. Transcripts were qualitatively analyzed by 6 raters who independently identified bthemes.Q Themes were refined using the Delphi technique and classified under broader bcategories.Q Results: Four main categories of barriers emerged, relating to (1) patients, (2) the health care system, (3) health care providers, and (4) the nature of this dialogue. Attending physicians and residents most frequently identified patient-related factors as barriers to discussions, followed by system, dialogue, and

B The work should be attributed to the Departments of Medicine and Anaesthesia, University of Toronto, and Department of Anaesthesia, McMaster University. We would like to acknowledge the Sunnybrook Trust for Medical Research Grant 724179409. We would like to state bno reprints.Q T Corresponding author. Cardinal Carter Wing, St. Michael’s Hospital, Toronto, Ontario, Canada M5B 1W8. E-mail address: [email protected] (V. Palda).

0883-9441/$ – see front matter D 2005 Elsevier Inc. All rights reserved. doi:10.1016/j.jcrc.2005.05.012

Barriers to communication regarding end-of-life care

215

provider barriers (43%, 39%, 10%, and 8%, respectively, for attending physicians; 40%, 34%, 13%, and 13%, respectively, for residents). Nurses similarly identified patient-related and system barriers most frequently, but provider barriers were discussed more often than dialogue barriers (46%, 28%, 22%, and 4%, respectively). Conclusions: Attending physicians, residents, and nurses perceive the recipients of their care, and the system within which they provide this care, to be the major source of barriers to communication regarding end-of-life care. This finding may impact on the effectiveness of quality-improvement initiatives in end-of-life care. D 2005 Elsevier Inc. All rights reserved.

1. Introduction End-of-life care is assuming increasing importance with the aging of the Canadian population. This has particular significance when we consider that approximately 50% of hospital deaths occur in the intensive care unit [1]. Resuscitative measures in hospitalized patients are associated with significant morbidity, mortality, and cost [2-6]. The incidence of withholding and withdrawing life support has shown a significant increase over the last 2 decades [7], and there is marked variability in the practice of end-of-life care between centers [8]. The guiding principle in any approach to life support requires that patient autonomy and well-being be primary determinants of care [9]. However, physicians may not possess the necessary skills to engage in end-of-life conversations [10-12]. They overestimate the value of therapies such as cardiopulmonary-resuscitation and mechanical ventilation [13,14], leaving patients misinformed about the nature and value of resuscitation [15,16]. This can lead to interventions that would otherwise be rejected had patients been adequately informed [16-18]. Sometimes patients are not even offered the opportunity to discuss these issues [19,20], with the result that when asked, physicians [21] and family members [22] are often unaware of the patient’s wishes regarding aggressive therapy. The Study to Understand Prognoses and Preferences for Outcomes and Risks of Treatments demonstrated that simply providing physicians with patient information pertinent to end-of-life issues was insufficient to enhance the discussion and decision-making process [23]. Goold et al [24] suggest that there are other barriers to engaging in discussions, including family tensions and denial, prognostic uncertainty, discomfort with death, misinformation from media, and organizational factors such Table 1

as unclear policies or economic pressures. They go on to recommend educational interventions as a possible method of facilitating end-of-life discussions. To design and implement interventions that would adequately educate health care providers on initiating endof-life discussions, it is important to first more clearly delineate the barriers that actually exist in that context and environment. The purpose of the present study was to identify what health care providers perceive as barriers to end-of-life discussions in a Canadian health care setting.

2. Materials and methods After obtaining approval from the research ethics board, all attending physicians, residents, and nurses caring for patients on the Internal Medicine service at a 600-bed tertiary care hospital in Toronto, Canada, were approached about participation during the 8-month study period. Internal Medicine patients represent a large proportion of those eventually requiring resuscitative therapy. Attending staff were chosen for their longer experience, confidence in assessment of prognosis, and involvement in major treatment decisions. Residents were interviewed because they have more frequent, direct contact with patients and families, and are usually the first to engage in end-of-life conversations. Nurses spend the most time with the patients and families and will often discuss personal and spiritual concerns with patients, especially regarding endof-life decisions. A total of 11 separate focus groups were organized for the residents (grouped by level of training), attending physicians, and nurses, all of whom signed written informed consent. They were segregated into homogeneous groups in terms of status, experience, and training, to facilitate

Composition of focus groups

Health care provider

No. of focus groups

No. of subjects per group

Total No. of subjects interviewed

Sex distribution

Age range

Staff Residents Nurses Summary

2 3 6 11

5 3-12 4-7 3-12

10 24 33 67

M, F M, F F M, F

35-50 20-30 30-50 20-50

216

A.H. Anselm et al.

Box 1: Recipient barriers 1. Exclusion by family of patients or their wishes Families sometimes prefer to shield patients from these discussions and/or disregard patients’ previously expressed wishes. bI’ve had a couple of instances where the patient himself/herself was very calm and could appreciate the discussion and could carry on a reasonable conversation but the family didn’t want this discussion with the patient. Quite often, we tell them that that’s inappropriate because where they can, the patient is still in charge of his or her own decision making. On occasion the family is the biggest barrier.Q

2. Difficulty in designating a decision maker or reaching consensus Families have difficulty either determining who the decision maker is, or what the family’s consensus is regarding the desired level of intervention. b[When] it is really unclear as to who the designated family member is that can be really bothersome because one family member will make a decision and then another family member will disagree. Obviously the family members themselves have not communicated with each other when that happens. . .You feel sometimes as though you’re caught in the middle. . .and the danger there is you tend to favour the decision that you think is more appropriate.Q

3. Family tensions Coping mechanisms of individuals increase family tension and make it difficult to establish communication. These include feelings of intense guilt, relieving stress through confrontation, and distancing themselves from the discussions. bThe family wanted us to do everything despite realizing that it was futile and that this patient was going to suffer, and so we felt that there was some inner guilt in the family members. They just wouldn’t let the patient pass on and they wouldn’t let us use the right to make the decision to not resuscitate.Q

4. Differences in culture or values Certain cultures, religions, or other sources of deeply held values (which may conflict with those of providers) inhibit discussion. bUnfortunately, our concepts of patient autonomy and about decisions about treatment are very Anglo-Saxon based ideologies where it is a little more open in terms of dialogue among family members. In other cultures it just doesn’t work that way. The idea is you’re the doctor. The presumption is that the doctor makes all the decisions, not the family. I think that western Anglo-Saxon cultures are more used to making decisions for themselves. This ideology is not as ingrained in other cultures. Often patients say to you, dWell, you’re the doctor, so you decide,T or dWhatever you say, doctor.T Some people from other cultures expect you to make that decision for them.Q

5. Variable capacity to understand and appreciate discussions Patient or family incapacity to understand or appreciate these discussions limits communication. bQuite often the family is confused and although you have an idea about how you want to manage the patient and what would be appropriate actions, the family doesn’t necessarily understand you.Q

6. Appropriate timing A poorly timed discussion may raise anxiety in or alienate patients who are relatively well, young, insufficiently informed about their condition, afraid of death, unprepared for death, or who have not achieved closure in personal relationships. bIt’s difficult when people are very sick and are approached at a moment of high stress. This is not the right time to broach the topic for the first time. Everybody is confused and nobody knows what to say and what to think.Q

Barriers to communication regarding end-of-life care

217

7. Temporal lability of appropriateness of resuscitation The appropriateness and desirability of resuscitation might be different at different times for either the patient or the provider. bTherein lies the difficulty with those people who are bfineQ—the durability of a DNR decision. If you ask a person about DNR and a DNR is signed, once out of the hospital and some time in the near or distant future, how definite was that decision made 5 months, 10 months, or 10 years ago? Q

comfortable, open, and frank discussions. To allow participants ample opportunity to contribute, the size of the groups was kept to 12 or fewer individuals. Each focus group was led by 1 or 2 interviewers with experience in qualitative research in medical settings (MLSV, MK). An additional researcher was present to record nonverbal cues (JLM). The sessions were approximately 1 hour in duration, and subjects were paid a small honorarium in accordance with the University of Toronto’s interpretation of Canadian federal guidelines. An interview schedule was designed to elicit information regarding (1) who was responsible for initiating end-of-life discussions, and the role of various team members in this process; (2) perceived institutional, patient, and family barriers to discussions; (3) personal difficulties in initiating and participating in such discussions; (4) views on what should happen during such decision making; (5) personal and institutional problems encountered; (6) how hospital management could help facilitate the resolution of these difficulties; and (7) suggestions on interventional strategies for educating providers on approaching end-of-life discussions. The interview schedule is reproduced in Appendix A. Audiotapes of interviews were transcribed verbatim, with the exception of identifying information, which was deleted. Six analysts (VP, CG, RM, MLSV, MK, JLM) who have experience with qualitative research and training in diverse disciplines applicable to this field (anesthesia, critical care, internal medicine, oncology, nursing, and sociology) independently reviewed the transcriptions. They identified word clusters that corresponded to discrete ideas related to barriers. A list of themes, which describe these ideas, was developed by each analyst. The analysts then met to establish consensus regarding the meaning of ambiguous comments and to harmonize the various lists of themes. To minimize redundancy in the aggregate list of themes, an iterative process (Delphi method) of 4 cycles was used. After consensus on the list of themes was achieved, the 6 analysts again independently read the interviews to determine the number of times, that is, frequency with which, each theme was addressed during each interview. To ensure adequate reproducibility of the thematic structure developed, reliability analysis (intraclass correlation coefficient and Cronbach’s a) was performed to determine the reproducibility of analysts’ coding of the frequency with which each theme was discussed in the interviews. For this analysis, focus groups were treated as

subjects, analysts were treated as raters, and themes were treated as test items.

3. Results 3.1. Focus group composition Characteristics of the participants are outlined in Table 1. All 67 subjects participated except for one resident from the PGY2/3 group who made no comments.

3.2. Categories and themes identified as barriers Fifty-seven themes were initially identified as barriers to end-of-life discussions. Through a process of 4 iterations, these were successively distilled to 51, then 31, and finally 20 themes. The analysts classified these 20 themes into 4 categories: recipient barriers, system barriers, provider barriers, and dialogue barriers (Box 1 -4). Health care brecipientsQ were seen to be the source of many barriers to communication regarding end-of-life care. Comments classified as recipient barriers referred to issues specific to the patient or family and are detailed in Box 1 with sample quotes. bSystemQ barriers refer to institutional, professional, and societal factors perceived to limit communication regarding end-of-life care and are detailed in Box 2. The interviewees also identified factors specific to themselves and their personal issues as sometimes being the source of barriers to the discussions. We classified these as bproviderQ barriers (Box 3). Finally, the fourth category, bdialogueQ barriers (Box 4), focused on the terminology used in end-of-life discussions. Here, providers felt that several difficulties involved in initiating these discussions arose because of the intrinsic nature of the communication itself.

3.3. Ranking of categories and themes Themes were ranked by frequency of identification as a barrier by each of the 3 groups of subjects. That is, of the total number of times particular barriers were addressed, the percentage of times an individual barrier was discussed was calculated, and then this percentage ranked. Attending physicians, residents, and nurses all identified recipient factors most frequently as being barriers to end-of-life discussions (43%, 40%, and 46%, respectively). The second most frequently discussed category of barriers was system

218

A.H. Anselm et al.

Box 2. System barriers 8. Suboptimal coordination of information exchange Providers in teaching hospitals do not communicate optimally with each other or with other institutions regarding endof-life discussions, even if these have occurred previously. The system for sharing information is inefficient. The resultant uncertainty regarding optimal management can delay initiation of communication. bThere could be better communication among health care personnel. . .These are things that get people into trouble, not discussing with family ahead of time, not communicating among medical personnel, somebody else making a decision. It’s difficult for families because they see multiple doctors.Q

9. Impersonality of large teaching hospitals Teaching hospitals are large impersonal institutions; care is typically short-term, with minimal involvement of community providers. bIt’s not easy. Decisions for us are different than those made by long-term care physicians. . .Our usually short-term relationship with patients can pose a barrier. . .My willingness is reflected by my not really knowing the patient on a long term basis.Q

10. Providers unskilled in discussions as a result of specialization in certain areas In teaching hospitals, care is specialty based; certain specialties are unskilled at conducting (or recognizing the need for) these discussions. bCertain specialties are better at discussing issues like that. Oncologists tend to be better, infectious disease specialists for HIV tend to be better, but other disciplines seem to forget that their patients can die. Cardiologists are very poor communicators of this discussion. I think also, people who are nephrologists tend to forget that their patients can die from terminal conditions.Q

11. Scheduling difficulties Busy work schedules of providers and the physical environment of hospitals make it difficult to arrange for private discussions. bWe are very busy and by definition if you are going to discuss this you have to be prepared to do it very slowly and patiently and wait for questions, answer questions. That’s the biggest barrier for me. The absence of time that this sort of thing merits.Q

12. Lack of external support External factors work against providers to create barriers to discussion: fear of legal action, lack of effective policy documents, and lack of institutional resources including education programs, better staffing, or 24-hour support for ethical decision making by resource people. bWe don’t have good legal or institutional backup when we get into situations like this. . .It stems back to our training for codes. Our training starts with the ACLS course and there we’re thrown in with nurses and other health care attendants and we all sit around learning algorithms. There’s no integration of that whole process so you leave the course saying to yourself that you’ll run the code for 20 minutes and consult the algorithms but what you really need is to have policy to back us up.Q

13. Risk of abandonment for bDNRQ patients bDNRQ labels the patient and leads to abandonment or less aggressive care by others. bOne of the problems that I’ve come across is that when you do put a DNR on a patient’s chart they frequently do not get the same care that they should get up until the point where they have to be resuscitated. It does brand them. . .That’s one barrier that I have to the idea of DNR.Q

Barriers to communication regarding end-of-life care

219

Box 3. Provider barriers 14. Inadequate expertise in prognosticating and leading discussions A lack of expertise due to inadequate training or inexperience makes providers feel uncomfortable about leading these discussions. bNo one teaches us how to do these things. There’s no course on this and quite frankly our role models for this are few. . .a lot of this is learned at the bedside. I think there is a role to be had for educating physicians in an approach.Q

15. Discomfort with emotion involved Identification with the patient, and/or other emotions, makes these discussions difficult. bSome doctors have difficulty. . .we had 2 physicians recently who, no matter how hard we tried, they never would talk with the patients and family about this. . .They themselves had difficulty dealing with it. . .They couldn’t come to grips with it.Q

16. Role ambiguity Providers’ roles and responsibilities in this domain are not well outlined; they fear reprimand due to overstepping the boundaries of their position. bI think overall we need the development of clear definitions of roles. . .What’s the role of the physician, the role of the nurse?Q

17. Prognostic uncertainty Providers prefer not to discuss end-of-life care until they are certain that the patient’s prognosis is dismal. bOften you don’t know with 100% certainty that there’s no hope. . .It’s awkward, but I guess you can say that the chance is unlikely or less likely. However, people often want you to be more specific and that’s hard because again, you just don’t know.Q

barriers (39%, 34%, and 28%, for attending physicians, residents, and nurses, respectively). Although attending physicians and residents identified provider barriers least frequently (8% and 13%, respectively), nurses discussed these barriers third most frequently (22%) (Fig. 1). Within these categories of themes, attending physicians most frequently identified scheduling difficulties, family tensions, cultural issues, and poor timing as barriers (13%, 12%, 10%, and 10%, respectively) (Fig. 2A). Residents discussed the inadequacy of external support, poor timing, and lack of expertise most frequently (19%, 16%, and 6%, respectively) (Fig. 2B). Nurses identified poor timing, inadequacy of external support and cultural issues most frequently (13%, 13%, and 13%, respectively) (Fig. 2C).

3.4. Reliability of thematic structure and frequency analysis Intraclass correlation coefficients and Cronbach’s a were calculated for each data set to determine the internal consistency and reliability of the analysts’ coding of frequency of themes. The intraclass correlation coefficient was 0.79 (95% CI 0.61-0.91) for the resident focus group

themes, 0.88 (95% CI 0.77-0.94) for themes from the nursing focus groups, and 0.91 (95% CI 0.83-0.96) for attending physician focus groups. Cronbach’s a was 0.83 for residents, 0.90 for nurses, and 0.92 for attending physicians.

4. Discussion With an aging population, Canadian health care providers have the responsibility of providing adequate end-oflife care to patients and their families. Central to this is upholding the principle that patient autonomy and wellbeing determine the goals of care. The Canadian Medical Association’s position is that although health care professionals are not obligated to provide futile care, bit is the patient’s right to accept or refuse treatmentQ [25]. The ability to exercise this autonomy depends on a number of factors including patient competence and family involvement. The Canadian Medical Association’s statement on principles for developing cardiopulmonary-resuscitation policy suggests using open discussion, sensitivity, and the exchange of information to determine treatment based on the patient’s

220

A.H. Anselm et al.

Box 4. Dialogue barriers 18. Nature of bDNRQ that may be perceived as nonsensical or defeatist Discussing or ordering bDNRQ is either nonsensical (because it specifies things not to do, which is unique), or inappropriate, because it is seen as being defeatist. bSo many of our doctor’s orders are written in the negative. For instance, bDo Not Resuscitate, No CPR, No This, No That.Q If things were written out in a more positive manner, it wouldn’t be construed by the persons reading the chart as a bnegativeQ thing, and in many cases as a Do Not Treat situation. We need to change the way we chart things and then you won’t get that sort of confusion or misunderstanding.Q

19. Societal values surrounding death Society does not generally recognize or appreciate death as a natural and acceptable part of life. This is reflected in expectations of unrealistically high survival rates from CPR, due to media portrayal. bAnother barrier is the perception of the general public and the perception of the families in terms of the success of a resuscitation effort. . .and how it’s altered by the media and television shows like ER and Chicago Hope. . .That’s the ER ward that I want to go to because all of those patients live through their resuscitation!Q

20. Lack of trust in providers commitment or competence Recipients of care lacked trust in providers. They questioned providers’ commitment or competence by charging that issues such as resource allocation were interfering with acting in their interest, or by simply not believing their diagnosis and/or prognosis. bOften if we aggressively seek DNR status on patients who may not be that ill or may not be that old, it undermines the trust that they have in you especially when they believe that your job is to provide them with excellent care. This is particularly important in the current status of the health care system with all the cutbacks and bed closures. They may misinterpret your intentions as dkilling themT to make way for someone who dneedsT their bed.Q

best interests [26]. Effective discussions between health care providers, patients, and their families may be the key to optimal end-of-life care [10], but providers are still reluctant to engage in these, even in the presence of interventions designed to facilitate the process [23]. Ineffective communication may unnecessarily complicate end-of-life care as attested to by a focus group study of nurses in intensive care units [27].

Fig. 1

This study has identified factors that health care providers perceive to be barriers precluding effective communication regarding end-of-life care. The 20 themes identified here cover a broad range of issues highlighting the breadth and scope of contributing variables that may hinder these discussions. This confirms the findings of Cook et al [28] who showed that factors ranging from patient demographics and nature of illness to time of admission

Distribution of categories identified as barriers to end-of-life discussions.

Barriers to communication regarding end-of-life care

Fig. 2 Ranking of ten themes most frequently identified as barriers to end-of-life discussions.

and geo-cultural differences affect when and how the issues of life support are broached. In this study, the themes most frequently identified by providers were those belonging to recipient and system categories of barriers. It is perhaps most interesting to note that although attending physicians, residents, and nurses deal with these issues in different capacities, and from different perspectives, they are in agreement with respect to

221 what they most commonly identify as the source of the problem, that is, the patients and families receiving their care, and the system within which they provide this care. A study by Eliasson et al [29] in the general medicine ward of a tertiary care hospital interviewed attending physicians regarding why they did not discuss resuscitation with patients in cases where they felt resuscitation would be inappropriate. Limits in the extent of the physician-patient relationship appeared to be the most frequent impediment to writing a do-not-resuscitate (DNR) order. In a poorly developed provider-recipient relationship (which may stem from pressures within the health care system), it can be difficult for providers to understand the nuances of recipient factors such as family tensions, appropriate timing, readiness for such discussions, and cultural differences. Time pressure was also identified by Eliasson et al [29] as an impediment to discussing DNR orders. This supports our finding that the most frequently cited theme by attending physicians was bscheduling.Q However, Eliasson et al [29] also found that attending physicians opted not to discuss end-of-life care because they believed the patient’s primary care provider should discuss DNR issues. Although our study did not explicitly identify this barrier, it could be viewed as a system barrier related to poor coordination between health care providers, a theme that was identified in our study. A focus group study conducted in the family medicine department of a tertiary care center asked physicians to identify barriers to code-status discussions [30]. Factors identified here also lend support to the findings of our study. They included recipient issues such as appropriate timing and family tensions; system factors such as scheduling difficulties; dialogue factors such as patients’ trust regarding the motivation of physicians; and provider barriers such as expertise in broaching the topic, or personal emotions related to mortality. Similar findings were again cited in a questionnaire-based study of internal medicine physicians [31]. A potential limitation of our study was our choice of a focus group design. Focus groups are limited to participants who volunteer. Each group is a unique mix of individuals whose social dynamics can be variable and unpredictable. Moderators have less control over the direction of the discussion than would be possible in a one-on-one interview. As such, analysis and interpretation of these qualitative data can be challenging. However, the purpose of focus groups is to promote self-disclosure among participants who share something in common. They are especially useful when trying to understand differences in perspectives between groups of people, and when looking for a range of ideas, feelings, or factors that influence opinions, behavior, or motivations [32]. These were certainly the objectives of this study. We believe the limitations inherent in focus groups were addressed in our study with well-developed questions, appropriately recruited participants, skilled moderation, and systematic analysis.

222 The objective of this study was to identify barriers to communication regarding end-of-life care, some of which might be amenable to educational interventions. The Medical Education for Care Near the End of Life National Consensus Conference [33] made several recommendations for improving education in this area for critical care providers. Principal among these was respect for the diversity of patients’ opinions and values. It also suggested that providers communicate information regarding the patient’s illness and prognosis in a coherent and coordinated manner, with proactive listening (clarifying with questions what they do, and do not, understand). Similar strategies are endorsed by Norton and Talerico [34] following a focus group study asking nurses, physicians, and family members what provider behaviors facilitated end-of-life decision making. Need for such educational strategies is supported by a study of nursing staff and assistants at nursing homes, who felt that difficulties with communication were a major impediment to adequate end-of-life care [35]. Incorporating the above concepts into educational programs may help providers better understand who patients and their families are, what tensions exist among them, when it is best to talk to them, and how cultural differences may complicate communication. This approach would address some of the key recipient factors perceived to be barriers in our study. The conference also advocated education during multidisciplinary team rounds that provide an opportunity for collaborative learning, with a focus on individual patients. It suggested problem-based learning, videotapes, and roleplaying exercises with feedback for teaching trainees about effective discussions. Similar suggestions have been made by others [30,31,36]. These authors note that efforts to teach an ideal model of care will only be successful with institutional support including policy makers, subspecialty organizations, and curriculum planners. This approach would address the issue of inadequate expertise and also provide the bexternal supportQ that is felt to be lacking by health care providers in our study. Institution-level engagement in optimizing end-of-life communication might also help address system barriers such as scheduling difficulties and poor coordination between institutions and providers.

A.H. Anselm et al. has delineated provider perspectives on several categories of barriers to communication regarding end-of-life care. Further investigation of patients’ perspectives on the various barriers will be important to gain a complete picture. Understanding how patients perceive such issues is crucial to our understanding of barriers, and how they might be amenable to education. Johnston and Pfeifer [38] support the need for physician education, whereas Clover et al [39] encourage nursing research regarding patients’ perceptions in this area. These are important goals in helping providers and patients attempt to engage in a difficult dialogue on life and death.

Appendix A. Focus group questions 1.

2.

3. 4. 5.

6. 7.

Who initiates the discussion around end-of-life decision making on your service? a. What is your role? b. What is the role of other members of the team (housestaff vs attending staff)? What do you see as the barriers to discussions regarding end-of-life decision making? a. At Sunnybrook Health Science Centre? b. In your interactions with patients? c. In your interactions with families? What do you find difficult about initiating/participating in these discussions? What are your own views regarding what should happen in end-of-life decision making? What problems have you had in making your ideas work? a. Personal difficulties b. Institutional difficulties How could the hospital/environment be helpful in facilitating the resolution of these difficulties? We are considering an educational intervention regarding this issue. What are your views on such an intervention and what would you like to see implemented?

References 5. Conclusions The views expressed here by attending physicians, residents, and nurses may reflect real and important barriers to communication between health care providers and patients. On the other hand, they may simply represent a bdefensiveQ attitude whereby providers attribute much of the bblameQ to factors extrinsic to themselves. In a recently published study by Wenrich et al [37], focus groups composed of patients, family members, physicians, and other health care providers identified only physician-related barriers to communication with dying patients. Our study

[1] Heyland DK, Lavery JV, Tranmer JE, Shortt SE, Taylor SJ. Dying in Canada: is it an institutionalized, technologically supported experience? J Palliat Care(16 Suppl.):S10 - 6. [2] Jacobs P, Noseworthy TW. National estimates of intensive care utilization and costs: Canada and the United States. Crit Care Med 1990;18:1282 - 6. [3] Lakshmipathi C, Pinsky MR, Donahoe MP, Grenvik A. Long-term outcome of critically ill elderly patients requiring intensive care. JAMA 1993;269:3119 - 23. [4] Ridley S, Plenderleith L. Survival after intensive care. Anaesthesia 1994;49:933 - 5. [5] Ballew KA, Philbrick JT, Caven DE, Schorling JB. Predictors of survival following in-hospital cardiopulmonary resuscitation. Arch Intern Med 1994;154:2426 - 32.

Barriers to communication regarding end-of-life care [6] Bialecki L, Woodward RS. Predicting death after CPR: experience at a nonteaching community hospital with a full-time critical care staff. Chest 1995;108:1009 - 17. [7] Prendergast TJ, Luce JM. Increasing incidence of withholding and withdrawal of life support from the critically ill. Am J Respir Crit Care Med 1997;155(1):15 - 20. [8] Prendergast TJ, Claessens MT, Luce JM. A national survey of end-oflife care for critically ill patients. Am J Respir Crit Care Med 1998;158:1163 - 7. [9] American Thoracic Society. Withholding and withdrawing lifesustaining therapy. Am Rev Respir Dis 1991;144:726 - 31. [10] Gilligan T, Raffin TA. Whose death is it anyway? Ann Intern Med 1996;125:137 - 40. [11] Tulsky JA, Chesney MA, Lo B. How do medical residents discuss resuscitation with patients? J Gen Intern Med 1995;10:436 - 42. [12] Hansot E. A letter from a patient’s daughter [letter] Ann Intern Med 1996;125:149 - 51. [13] Ghusn HF, Teasdale TA, Skelly JR. Limiting treatment in nursing homes: knowledge and attitudes of nursing home directors. J Am Geriatr Soc 1995;43:1131 - 4. [14] Miller DL, Gorbein MJ, Simbarth LA, Jahnigen DW. Factors influencing physicians in recommending in-hospital CPR. Arch Intern Med 1993;153:1999 - 2003. [15] Godkin MD, Toth EL. Cardiopulmonary resuscitation and older adults expectations. Gerontologist 1994;34:797 - 802. [16] Miller DL, Jahnigen DW, Gorbein MJ, Simbarth L. Cardiopulmonary resuscitation: how useful? Attitudes and knowledge of an elderly populations. Arch Intern Med 1992;152:578 - 82. [17] Hofmann JC, Wenger NS, Davis RB, Teno J, Connors Jr AF, Desbiens N, et al. Patient preferences for communication with physicians about end-of-life decisions. Ann Intern Med 1997;127:1 - 12. [18] Murphy DJ, Burrow D, Santilli S, Kemp AW, Tenner S, Kreling B, et al. The influence of the probability of survival on patients preferences regarding cardiopulmonary resuscitation. N Engl J Med 1994;330:545 - 9. [19] Higget CL, Forsyth DR, Connor GD. Improving the documentation and appropriateness of CPR decisions. J R Soc Med 1995;88:136 - 40. [20] Owen C, Tennant C, Levi J, Jones M. Resuscitation: patient and staff attitudes in the context of cancer. Gen Hosp Psychiatry 1992;14:327 - 33. [21] Teno JM, Hakim RB, Knaus WA, Wenger NS, Phillips RS, Wu AW, et al. Preferences for cardiopulmonary resuscitation: physicianpatient agreement and hospital resource use. J Gen Intern Med 1995;10:179 - 86. [22] Gerety MB, Chiodo LK, Kanten DN, Tuley MR, Cornell JE. Medical treatment preferences of nursing home residents: relationship to function and concordance with surrogate decision makers. J Am Geriatr Soc 1993;41:953 - 60. [23] The SUPPORT Principal Investigators. A controlled trial to improve care for seriously ill hospitalized patients. The Study to Understand

223

[24] [25] [26] [27]

[28]

[29]

[30]

[31]

[32] [33]

[34]

[35]

[36]

[37]

[38]

[39]

Prognoses and Preferences for Outcomes and Risks of Treatment (SUPPORT). JAMA 1995;274:1591 - 8. Goold SD, Williams B, Arnold RM. Conflicts regarding decisions to limit treatment. A differential diagnosis. JAMA 2000;283:909 - 14. Canadian Medical Association. Resuscitation of the terminally ill. CMAJ 1987;136:424A. Canadian Medical Association RM. Joint statement on resuscitative interventions. CMAJ 1994;151:1176A - C. Kirchoff KT, Spuhler V, Walker L, Hutton A, Cole BV, Clemmer T. Intensive care nurses’ experience with end-of-life care. Am J Crit Care 2000;9:36 - 42. Cook DJ, et al, For the Canadian Critical Care Trials Group. Cardiopulmonary resuscitation directives on admission to intensivecare unit: an international observational study. Lancet 2001;358: 1941 - 5. Eliasson AH, Parker JM, Shorr AF, Babb KA, Harris R, Aaronson BA, et al. Impediments to writing do-not-resuscitate orders. Arch Intern Med 1999;159:2213 - 8. Calam B, Far S, Andrew R. Discussions of bcode statusQ on a family practice teaching ward: what barriers do family physicians face? CMAJ 2000;163:1255 - 9. Morrison RS, Morrison EW, Glickman DF. Physician reluctance to discuss advance directives. An empiric investigation of potential barriers. Arch Intern Med 1994;154:2311 - 8. Krueger RA, Casey MA. Focus groups: a practical guide for applied research. 3rd ed. Thousand Oaks (Calif)7 Sage Publications; 2000. Danis M, Federman D, Fins JJ, Fox E, Kastenbaum B, Lanken PN, et al. Incorporating palliative care into critical care education: principles, challenges, and opportunities. Crit Care Med 1999;27:2005 - 13. Norton SA, Talerico KA. Facilitating end-of-life decision-making: strategies for communicating and assessing. J Gerontol Nurs 2000; 26:6 - 13. Ersek M, Kraybill BM, Hansberry J. Assessing the educational needs and concerns of nursing home staff regarding end-of-life care. J Gerontol Nurs 2000;26:16 - 26. Tulsky JA, Chesney MA, Lo B. See one, do one, teach one? House staff experience discussing do-no-resuscitate orders. Arch Intern Med 1996;156:1285 - 9. Wenrich MD, Curtis JR, Shannon SE, Carline JD, Ambrozy DM, Ramsey PG. Communicating with dying patients within the spectrum of medical care from terminal diagnosis to death. Arch Intern Med 2001;161:868 - 74. Johnston SC, Pfeifer MP. Patient and physician roles in end-of-life decision making. End-of-life study group. J Gen Intern Med 1998; 13:43 - 5. Clover A, Browne J, McErlain P, Vandenberg B. Patient approaches to clinical conversations in the palliative care setting. J Adv Nurs 2004;48:333 - 41.