Elizabeth Knauft, MD, MS; Elizabeth L. Nielsen, MPH; Ruth A. Engelberg, PhD; Donald L. Patrick, PhD, MSPH; and J. Randall Curtis, MD, MPH, FCCP

Barriers and Facilitators to End-of-Life Care Communication for Patients with COPD* Elizabeth Knauft, MD, MS; Elizabeth L. Nielsen, MPH; Ruth A. Engelberg, PhD; Donald L. Patrick, PhD, MSPH; and J. Randall Curtis, MD, MPH, FCCP Objective: Patients with COPD frequently do not discuss end-of-life care with physicians; therefore, we sought to identify the barriers and facilitators to this communication as a first step to overcoming barriers and capitalizing on facilitators. Design: Fifteen barriers and 11 facilitators to patient-physician communication about end-of-life care were generated from focus groups of patients with COPD. We subsequently conducted a cross-sectional study of 115 patients with oxygen-dependent COPD and their physicians to identify the common barriers and facilitators and examine the association of these barriers and facilitators with communication about end-of-life care. Participants and setting: Patients with oxygen-dependent COPD were recruited from clinics at a university, county, and Veterans Affairs teaching hospital, and an oxygen delivery company. We also recruited the physician identified by each patient as the doctor primarily responsible for their lung disease. Measurements and results: Patients were interviewed by trained research interviewers. Physician data collection was completed by mail survey. Participation rates were 40% for patients and 86% for physicians. Only 32% of patients reported having a discussion about end-of-life care with their physician. Two of 15 barriers and 8 of 11 facilitators were endorsed by > 50% of patients. The most commonly endorsed barriers were I d rather concentrate on staying alive, and I m not sure which doctor will be taking care of me. Two barriers were significantly associated with lack of communication, as follows: I don t know what kind of care I want, and I m not sure which doctor will be taking care of me. The greater the number of barriers endorsed by patients, the less likely they were to have discussed end-of-life care with physicians (p < 0.01), suggesting the validity of these barriers. Conversely, the more facilitators, the more likely patients were to report having had end-of-life discussions with their physicians (p < 0.001). Conclusion: Although patients endorsed many barriers and facilitators, few barriers were endorsed by most patients. Barriers and facilitators associated with communication are targets for interventions to improve end-of-life care, but such interventions will likely need to address the specific barriers relevant to individual patient-physician pairs. (CHEST 2005; 127:2188 2196) Key words: barriers; COPD; death; dying; end-of-life care; palliative care; physician-patient communication Abbreviation: VA Veterans Affairs COPD is a chronic progressive illness that affects 6% of the general population and is the fourthleading cause of death in the United States. 1 Patients hospitalized with an acute exacerbation of COPD *From Division of Pulmonary and Critical Care Medicine (Dr. Knauft, Engelberg, and Curtis, and Ms. Nielsen), Department of Medicine, School of Medicine; and Program in Social and Behavioral Sciences (Dr. Patrick), Department of Health Services, School of Public Health, University of Washington, Seattle, WA. This project was supported by a Career Investigator Award from the American Lung Association and the American Lung Association of Washington (Dr. Curtis). Manuscript received May 20, 2004; revision accepted December 15, 2004. Correspondence to: J. Randall Curtis, MD, MPH, Division of Pulmonary and Critical Care Medicine, Harborview Medical Center, Box 359762, 325 Ninth Ave, Seattle, WA 98104-2499; e-mail: jrc@u.washington.edu have a median survival of 2 years, and 50% of patients are readmitted to the hospital within 6 months. 2 COPD patients are more likely to die with aggressive, technological care directed at preservation of life and are less likely to receive home nursing For editorial comment see page 1886 and palliative care services than patients with lung cancer, despite having similar preferences favoring palliation. 3 Ensuring that patients receive the care they desire requires that patients and their physicians discuss patient treatment preferences. However, only a small proportion of patients with moderate-to-severe COPD have actually discussed endof-life issues with their physicians. 4,5 In addition, 2188 Ethics in Cardiopulmonary Medicine

patients with COPD perceive many shortcomings in the education they receive about many aspects of their care, including their disease, treatment, prognosis, and advance-care planning. 6 Despite this need for improved patient-physician communication about end-of-life care and improved access to palliative care for patients with COPD, there are relatively few data to guide physicians in how to accomplish these goals. For example, the updated Global Initiative on COPD guidelines, 7 assembled by an expert panel under the auspices of the National Heart Lung and Blood Institute and the World Health Organization, do not include recommendations on providing quality end-of-life care for patients with COPD in part because of the relative scarcity of data on how to improve this care. American Thoracic Society/ European Respiratory Society COPD guidelines 8 contain a section on end-of-life care that highlights the limited data available regarding patients with COPD and calls for research on the barriers to communication about end-of-life care. Identification of specific ways to overcome barriers to this communication thereby promoting high-quality, end-of-life care would be beneficial, especially since end-of-life care may be worse for patients with COPD compared with those with other diseases such as lung cancer. 3,9 Because patient-clinician communication about end-of-life care is important for patients with COPD and yet does not commonly occur, the goal of this study was to identify the common and important barriers and facilitators of this communication. Once identified, these barriers and facilitators could become the targets for interventions to improve communication about end-of-life care and ultimately be used to help physicians find ways to improve the quality of this care. Materials and Methods Participant Recruitment and Enrollment We report results of a cross-sectional analysis of a prospective study of patients with oxygen-dependent COPD recruited from ambulatory clinics at a university, county, and Veterans Affairs (VA) teaching hospital, and an oxygen delivery company in Seattle, WA. The methods have been described previously in an article 10 describing the quality of communication about end-oflife care and an article 11 describing the role of depression and quality of life on treatment preferences for life-sustaining treatments. Patients were enrolled between July 1999 and June 2002. Patients were eligible if they were 18 years old, spoke English, had a physician diagnosis of COPD, and had been prescribed long-term oxygen therapy. Patients were excluded if their doctor did not confirm a clinical diagnosis of COPD, if the doctor anticipated that the patient would only need oxygen temporarily, or if mental status precluded participation. We used two enrollment methods because of differing requirements from different Institutional Review Boards. At the county and university hospitals, the physician or nurse asked the patient whether he or she was willing to talk with a study staff member. At the VA medical center and oxygen delivery company, a letter was mailed to all patients receiving oxygen, asking them to call a toll-free voice mail if they did not want to participate; if they did not leave a message declining participation, they received a call from a study staff member. Every participant was also asked to identify the doctor they believed to be primarily responsible for caring for their lung disease. The Human Subjects Committee at the University of Washington and the Institutional Review Board at all institutions involved approved all study procedures, and all patients signed consent forms. Interviews and Questionnaires Patient data collection was conducted with in-person interviews by a trained research interviewer. Demographic information was gathered including age, gender, and race. Patients were asked their amount of time receiving oxygen therapy and current comorbid illnesses. FEV 1 was obtained using a portable spirometer according to American Thoracic Society standards. 12 Alternatively, if patients refused spirometry and consented to medical record review, then the most recent FEV 1 measure was obtained from the medical record. Physician data collection was completed with a mail survey. Development of the Barriers and Facilitators Questionnaire The barriers and facilitators questionnaire for patients was adapted from a version developed through focus groups of patients with AIDS 13 and previously validated in a cohort of patients with advanced AIDS. 14 To adapt this questionnaire to patients with COPD, we conducted three additional focus groups comprising 15 patients with COPD. During these focus groups, we asked patients to identify the barriers and facilitators for patient-physician communication about end-of-life care. Focus groups were audiotaped and transcribed verbatim. Investigators reviewed the transcripts to identify barriers and facilitators using content analysis. 15 In addition, focus group participants were asked to examine the barriers and facilitators identified in the prior study 13 and state whether these applied to them. On the basis of these focus groups, 1 new barrier was added, increasing the number of barriers from 14 to 15, and 2 new facilitators were added, increasing the number of facilitators from 9 to 11 (see Results section for new questions). To determine whether each barrier and facilitator applied to each patient, we read each item and asked, Does this apply to you? (questionnaire available from the investigators). The barriers and facilitators questionnaire for physicians was developed from focus groups of physicians in general medicine, pulmonary medicine, and infectious diseases working in university and community settings. 13 The physician questions describe patient-specific barriers, and physicians completed a separate questionnaire for each patient. The physician questions used the same format as the patient questionnaire. Data Analysis Demographic information, end-of-life care discussions, barriers, and facilitators were analyzed using descriptive statistics. We performed analyses to identify whether each barrier and facilitator was associated with patient and physician demographic characteristics, using 2 analysis with dichotomous variables (eg, race and sex) and Mann-Whitney U tests for continuous variables (eg, age and total months receiving oxygen). We did not conduct multivariate analyses for associations with patient and physician www.chestjournal.org CHEST / 127 / 6/ JUNE, 2005 2189

characteristics because there were no barriers or facilitators that were significantly associated with more than one patient characteristic or physician characteristic. Although barriers and facilitators were patient specific, we used the number of patients as the denominator for all analyses. We used a Spearman correlation coefficient to examine associations between the number of barriers or facilitators endorsed by patients with the numbers endorsed by their physicians and Mann-Whitney U tests to examine differences between the numbers of barriers or facilitators endorsed by patients who had treatment conversations with their physicians and those who did not. A two-tailed p value 0.05 was considered statistically significant. Results Of the 295 eligible patients contacted, 118 patients were initially enrolled in the study for 40% participation. Of the 118 enrolled patients, 3 patients were unable to complete interviews due to fatigue, leaving a sample size of 115 patients. At the two study sites where a clinician familiar with the patient introduced the study, 78 eligible patients were identified, and 50 patients were enrolled for a participation rate of 64%. At the two study sites where patients were sent a mailing describing participation, 217 eligible patients were identified, and 65 patients enrolled for a participation rate of 30%. Questionnaires were sent to physicians for 102 patients; we were unable to locate or identify physicians for the remaining 13 patients. Questionnaires were returned from physicians for 89 patients, giving a response rate from physicians of 86%. There were 56 individual physicians returning the patient-based survey. Forty-one physicians had one patient, 6 physicians had two patients, 4 physicians had three patients, 2 physicians had four patients, 2 physicians had five patients, and 1 physician had six patients. Demographic characteristics of the patients and physicians are shown in Table 1. Patients had a mean age of 67.5 and a mean FEV 1 of 29.0% of predicted. Approximately three fourths of the patients were male, and 84% of the patients were white. More than 80% of the physicians were trained in internal medicine, and almost 60% had subspecialty training in pulmonary medicine. Occurrence of Communication About End-of-Life Care We asked patients several questions about their communication with their physicians. Descriptive statistics for some of these items are reported in Table 2. Only 32% of patients reported having had a discussion with their physicians about care they would want if they were too sick to speak for themselves (95% confidence interval, 24 to 42%). Nonetheless, almost three fourths of patients reported that they thought their doctor probably or Table 1 Participant Characteristics* Characteristics Patients (n 115) Physicians (n 56) Age, yr 67.5 9.3 43.4 9.0 Months using oxygen 35.0 24.6 FEV 1 % predicted 29.0 14.3 Years since medical school graduation 16.4 10.2 Male gender 72 (83) 80 (45) White race 84 (96) 82 (46) Married or with partner 46 (53) Education 8th grade 4 (5) 9th to12th grade/general 39 (45) equivalency diploma 12th grade/some college 34 (39) College degree 12 (14) College 9 (10) Health insurance Private, prepaid 21 (24) Government program 94 (108) None 1 (1) Income $18,000 46 (53) $18,000 to $24,000 9 (10) $24,000 to $36,000 12 (14) $36,000 to $48,000 7 (8) $48,000 16 (18) Type of practice Solo or small group private 5 (3) Group practice 14 (8) Health maintenance organization 11 (6) VA clinic 43 (24) University hospital clinic 14 (8) County hospital clinic 11 (6) Specialty Family practice 8 (4) Internal medicine 82 (46) Other 9 (5) Subspecialty training Pulmonary 59 (33) Critical care 50 (28) COPD patients cared for 200 44 (25) *Data are presented as mean SD or No. (%) unless otherwise indicated. Respondents were able to endorse more than one subspecialty. definitely knows the kinds of treatments they would want if they were too sick to speak for themselves. Barriers and Facilitators to Communication About End-of-Life Care The three most common barriers endorsed by patients were as follows: (1) I would rather concentrate on staying alive than talk about death (75%); (2) I m not sure which doctor will be taking care of me if I get very sick (64%); and (3) I don t know what kind of care I would want if I get very sick (37%). The four most common patient-endorsed facilitators are as follows: (1) I have had family or friends who 2190 Ethics in Cardiopulmonary Medicine

Table 2 Patient Responses on Health-Care Communication Items* Health-Care Communication Items Asked of All Patients Do you think that your doctor knows the kinds of treatments you want if you get too sick to speak for yourself? Have you ever discussed with your doctor in a face-to-face discussion, the kind of treatments you want if you get too sick to speak for yourself? Definitely no, 9 (10) Yes, 32 (37) Probably no, 14 (16) No, 65 (75) Probably yes, 30 (35) Don t know, 2 (2) Definitely yes, 40 (46) Do not know, 6 (7) *Data are presented as % (No.). have died (88%); (2) I trust my doctor (87%); (3) My doctor is very good at taking care of lung/respiratory disease (83%); and (4) My doctor cares about me as a person (83%). Of the 15 patient barriers, 2 barriers were endorsed by 50% of patients, and an additional 9 barriers were endorsed by 10 to 40% of patients. Only four patient barriers were endorsed by 10% of patients. Of the 11 facilitators, 7 were endorsed by 50% of patients, and all but 1 facilitator was endorsed by 10% of patients. The only facilitator endorsed by 10% of the patients was My doctor often asks me about end-of-life care. Table 3 shows the descriptive statistics for these individual patient barriers and facilitators (Table 4). We also examined the frequency with which physicians endorsed barriers and facilitators. The three most common physician-endorsed barriers were as follows: (1) There is too little time during our appointments to discuss everything we should (64%); (2) I worry that discussing end-of-life care will take away his/her hope (23%); and (3) The patient is not ready to talk about the care she/he wants if she/he gets sick (21%). Only the first of these barriers was endorsed by 50% of physicians. Of the six physician facilitators, the three most commonly endorsed were as follows: (1) The patient and I have good relationship (85%); (2) I have cared for many patients with lung disease (80%); and (3) The patient has been very sick in past (71%). Table 5 shows the descriptive statistics for these individual physician barriers and facilitators (Table 6). Patient and Physician Predictors of the Presence of Barriers and Facilitators We examined whether each barrier or facilitator was more common among some patient or physician subgroups. For the barriers, male patients were more likely to endorse Doctors look down on Table 3 Patient-Endorsed Barriers* Barriers Applies Does Not Apply Do Not Know Not Answered 1. I would rather concentrate on staying alive than talk about death. 75 (86) 20 (23) 0 (0) 5 (6) 2. I m not sure which doctor will be taking care of me if I get very 64 (74) 29 (33) 3 (3) 4 (5) sick. 3. I don t know what kind of care I want if I get very sick. 37 (43) 57 (65) 2 (2) 4 (5) 4. I don t like to talk about getting very sick. 35 (40) 60 (69) 1 (1) 4 (5) 5. My ideas about the kind of medical care I want change at 31 (36) 62 (71) 3 (3) 4 (5) different times. 6. I have not been very sick. 31 (36) 64 (73) 1 (1) 4 (5) 7. I have a living will, and that means I don t need to talk with my 26 (30) 69 (79) 1 (1) 4 (5) doctor about the care I want if I m too sick to speak for myself. 8. Doctors look down on people who developed lung/respiratory 21 (24) 70 (80) 4 (5) 5 (6) disease because of smoking. 9. I don t want to make plans for the future. 20 (23) 74 (85) 2 (2) 4 (5) 10. I m not ready to talk about the care I want if I get very sick. 15 (17) 77 (88) 4 (5) 4 (5) 11. I feel that talking about death can bring death closer. 14 (16) 80 (92) 2 (2) 4 (5) 12. My doctor never seems to have the time to talk about issues like 9 (10) 77 (88) 4 (4) 11 (13) end-of-life care. 13. My doctor doesn t like to talk about me getting very sick. 8 (9) 64 (73) 21 (24) 8 (9) 14. I have concerns about bringing up assisted suicide. 7 (8) 87 (100) 0 (0) 6 (7) 15. I worry that talking about getting sick is too depressing for my doctor. 2 (2) 92 (106) 2 (2) 4 (5) *Data are presented as % (No.); n 115; some totals are 100% due to rounding. These barriers were added in response to the focus groups of patients with COPD. www.chestjournal.org CHEST / 127 / 6/ JUNE, 2005 2191

Table 4 Patient-Endorsed Facilitators* Facilitators Applies Does Not Apply Do Not Know Not Answered 1. I have had family or friends who have died. 88 (101) 8 (9) 0 (0) 4 (5) 2. I trust my doctor. 87 (100) 5 (6) 3 (3) 5 (6) 3. My doctor is very good at taking care of lung/respiratory 83 (95) 4 (5) 7 (8) 6 (7) disease. 4. My doctor cares about me as a person. 83 (95) 5 (6) 7 (8) 5 (6) 5. I feel sure that my doctor will be there for me if I get very 77 (89) 12 (14) 4 (5) 6 (7) sick. 6. I have been very sick. 64 (73) 32 (37) 0 (0) 4 (5) 7. I worry that I could be a burden on my friends and family if I 61 (70) 31 (36) 2 (2) 6 (7) got very sick. 8. I worry about the quality of my life in the future. 49 (56) 46 (53) 1 (1) 4 (5) 9. Someone other than my doctor has talked with me about the 35 (40) 59 (68) 0 (0) 6 (7) care I would want if I got too sick to speak for myself (like a nurse, home health worker, chaplain or clergy, social worker). 10. My doctor is very good at talking about end-of-life care. 23 (27) 42 (48) 23 (26) 12 (14) 11. My doctor often asks me about end-of-life care. 7 (8) 86 (99) 1 (1) 6 (7) *Data are presented as % (No.); n 115; some totals are 100% due to rounding. These facilitators were added in response to the focus groups of patients with COPD. smokers (p 0.035), and nonwhite patients were more likely to endorse I feel that talking about death can bring death closer (p 0.033). For the facilitators, patients with a lower health status were more likely to endorse I worry about my quality of life in the future (p 0.034). Among physicians, female physicians were more likely to endorse I work in a system that expects me to have end-of-life discussions with my patients (p 0.001). Nonwhite physicians were more likely to endorse a number of barriers as follows: (1) This patient doesn t know what kind of care he wants if he gets sick (p 0.040); (2) This patient is not ready to talk about the care he wants if he gets sick (p 0.022); (3) This patient s ideas about the kind of care he wants change over time (p 0.012); and (4) I worry that discussing end-of-life care will take away his hope (p 0.031). Summary scores for barriers or facilitators, comprised of the total number of items endorsed, were calculated for both patients and physicians. Patients endorsed a median of 4 barriers, with 3% (n 3) of patients endorsing no barriers, and no patients endorsing all 15 barriers. Nine was the maximum number of barriers endorsed by 3% (n 3) of patients. For facilitators, patients endorsed a median of 7 facilitators, with a minimum endorsement of 2 facilitators (2%; n 2), and a maximum endorsement of all 11 facilitators (3%; n 3). Physicianendorsed barriers were patient specific and, therefore, we used the number of patients as the denominator for physician-identified barriers. Physicians endorsed a median of one barrier per patient, with 16% (n 18) of patients having their physicians endorsing no barriers, and 2% (n 2) of patients having their physicians endorsing six barriers. No patient had a physician endorsed all seven barriers. For facilitators, the physicians endorsed a median of Table 5 Physician-Endorsed Barriers* Barriers Applies Does Not Apply Do Not Know Not Answered 1. There is too little time during our appointments to 64 (57) 34 (30) 1 (1) 1 (1) discuss everything we should. 2. I worry that discussing end-of-life care will take 23 (20) 75 (67) 1 (1) 1 (1) away her/his hope. 3. The patient is not ready to talk about the care she/ 21 (19) 76 (68) 2 (2) 0 (0) he wants if she/he gets sick. 4. She/he has not been very sick yet. 20 (18) 79 (70) 1 (1) 0 (0) 5. The patient doesn t know what kind of care she/he 15 (13) 83 (74) 2 (2) 0 (0) wants if she/he gets sick. 6. The patient s ideas about the kind of care she/he 11 (10) 85 (76) 2 (2) 1 (1) wants change over time. 7. My role as a doctor is to make him/her feel better. 10 (9) 88 (78) 1 (1) 1 (1) *Data are presented as % (No.); n 89; some totals are 100% due to rounding. 2192 Ethics in Cardiopulmonary Medicine

Table 6 Physician-Endorsed Facilitators* Facilitators Applies Does Not Apply Do Not Know Not Answered 1. The patient and I have a good relationship. 85 (76) 12 (11) 2 (2) 0 (0) 2. I have cared for many patients with lung disease. 80 (71) 18 (16) 2 (2) 0 (0) 3. The patient has been very sick in the past. 71 (63) 27 (24) 2 (2) 0 (0) 4. We have known each other a long time. 60 (53) 38 (34) 2 (2) 0 (0) 5. I work in a system that expects me to have end-oflife 39 (35) 58 (52) 2 (2) 0 (0) discussions with my patients. 6. The patient has had family or friends who have died. 29 (26) 66 (59) 3 (3) 1 (1) *Data are presented as % (No.); n 89; some totals are 100% due to rounding. four facilitators per patient, with 1% (n 1) of patients having physicians endorsing no facilitators, and 8% (n 9) of patients having physicians endorsing all six physician facilitators. The associations between patient and physician counts of barriers and facilitators were assessed using Spearman correlation coefficients. This analysis showed that there was no correlation between the number of barriers or facilitators endorsed by each patient and the number of barriers or facilitators endorsed by that patient s physician (barriers: Spearman r 0.15 [p 0.16]; facilitators: Spearman r 0.10 [p 0.37]). Barriers and Facilitators That Predict Having Discussions About End-of-Life Care Those patients who reported having had a discussion about the treatments she/he would want if she/he became too sick to speak for him/herself reported a median of three barriers compared with those patients who reported not having such a discussion who reported a median of four barriers (p 0.01). Similarly, those patients who reported having had this type of discussion reported a median of eight facilitators, whereas those who reported not having had such a discussion reported a median of only six facilitators (p 0.001). Therefore, patients who reported having had a discussion about end-oflife care reported fewer barriers and more facilitators than patients who did not report having had such a discussion. In addition to examining the association between the total numbers of barriers and facilitators and communication about end-of-life care, we also examined whether individual barriers and facilitators were associated with having had end-of-life discussions, as reported by patients. For patient-endorsed barriers, two barriers were associated with not having had end-of-life discussions, as follows: (1) I don t know what kind of care I want if I get very sick (p 0.002); and (2) I m not sure which physician will be taking care of me if I get very sick (p 0.004). There were also two patient facilitators associated with having had end-of-life discussions, as follows: (1) My physician is good at talking about end-of-life issues (p 0.001); and (2) Others have talked to me about end-of-life care (p 0.006). There were no physician-identified barriers associated with not having had end-of-life discussions and one physician-identified facilitator associated with physician reports of having had end-of-life discussions, as follows: The patient and I have a good relationship (p 0.002). Discussion Occurrence of Communication About End-of-Life Care This study suggests that only one third (32%) of the patients with oxygen-dependent COPD discuss with their physicians the type of care they would want if they were too sick to speak for themselves. Even the lower limit of the 95% confidence interval around this proportion (24%) is higher than the 19% found by Heffner and colleagues 4,5 in a study of patients with COPD enrolled in a pulmonary rehabilitation program. The increase likely represents temporal changes, with an increase in emphasis on advance-care planning and discussion about end-oflife care, but could also reflect differences in patient population or regional differences. Despite this relative increase in the proportion reporting discussions with their physicians about end-of-life care, it is still a minority of patients who report such a discussion has occurred. When these discussions do occur, patients rate the quality of the discussion relatively high. 10 We previously reported that if patients report having had a discussion about end-of-life care, on average they rate the quality of the discussion at a score of 8.9 of 10. The low occurrence of these discussions but relatively high-quality rating if they do occur suggests that a major impediment to improving communication about end-of-life care is to find ways to overcome the barriers to this communication. However, it is important to note that although patient satisfaction with physician communication about end-of-life care may be relatively www.chestjournal.org CHEST / 127 / 6/ JUNE, 2005 2193

high, other studies 16 22 have shown that patient understanding of advance-care planning is poor, physician understanding of patient preferences is poor, and there are significant shortcomings in the quality of this communication. Therefore, improving communication about end-of-life care will likely require interventions that can both overcome the barriers to this communication and improve the quality of the communication. One potential method to increase the communication about end-of-life care may be to incorporate nonphysician clinicians into these discussions. Prior studies 23,24 have suggested a potential role for such clinicians. It is interesting that discussions with such clinicians were reported to be a facilitator to patientphysician communication about end-of-life care in our prior qualitative work, 13 and in the current study patients who endorsed speaking with other clinicians about end-of-life care were more likely to have discussed end-of-life care with their physician. These findings further support the potential role for nonphysician clinicians in facilitating discussions about end-of-life care. Intervention studies are needed to confirm the role of such clinicians in improving communication between patients and physicians. One of the systems features important to patients and their families about end-of-life care is continuity of care. 23,25,26 Two thirds of patients reported that a barrier to discussions about end-of-life care was that patients were not sure which doctor will be taking care of them, and this was the second most commonly endorsed barrier. This finding is likely influenced by the fact that the majority of patients in this study were cared for in an academic setting, where continuity may be more difficult to achieve than in community-based practice. This may not be as important a barrier in nonacademic settings. Barriers and Facilitators to Communication About End-of-Life Care Patients endorsed large numbers of different barriers and facilitators. However, only two barriers were endorsed by more than half of the patients and, therefore, these two barriers represent potential targets for intervention with a majority of patients. The first barrier, I d rather concentrate on staying alive than talk about death, suggests that physicians must be skilled in talking about this difficult subject with patients who would rather not talk about it. It may be helpful in these situations to acknowledge that this is a difficult topic for discussion but an important one for patients and physicians to address. The second barrier, I m not sure which physician will be taking care of me if I get very sick, suggests that discussions of continuity may be an important aspect to encouraging these discussions. The fact that 11 of 15 barriers were endorsed by at least 10% of patients, whereas only 2 barriers were endorsed by 50%, suggests that interventions to overcome these barriers may need to target specific barriers that are relevant to individual patient-physician dyads. However, 7 of 11 facilitators were endorsed by 50% of the patients. Therefore, these facilitators may represent topics that physicians could use to encourage communication about end-of-life care with a majority of patients. For example, if patients have a family member or friend who has died, this may represent an opportunity for physicians to begin and encourage a discussion about the patient s preferences for end-of-life care. Several patient barriers are associated with a lack of communication about end-of-life care and, therefore, these specific barriers may represent particularly important targets for interventions to improve end-of-life care. However, the fact that only one of these two barriers was endorsed by more than half of the patients also suggests that future interventions will need to target the specific barriers and facilitators that are relevant to individual patients and their physicians. The lack of an association between the number of barriers (or facilitators) identified by patients and the number of barriers (or facilitators) identified by physicians also suggests that interventions may be more successful if they target both patients and physicians. Those patients who reported having had a discussion about end-of-life care report significantly fewer barriers and more facilitators than those who have not had such a discussion. This association suggests that the barriers and facilitators identified may be influential on the occurrence of communication. However, because of the cross-sectional study design, we cannot attribute a causal relationship between occurrence of communication and the barriers or facilitators. Future studies will be needed to determine whether an intervention targeting these barriers and facilitators can increase the occurrence of communication. Finally, we found that several patient and physician characteristics were associated with some barriers to discussing end-of-life care. Although we do not believe these associations should be used to profile certain patients or physicians, nonetheless, exploration of the role of race, ethnicity, and gender on the barriers and facilitators to discussing end-of-life care may allow better understanding of the best way to use these barriers and facilitators to improve quality of care. Prior research 14,27,28 has shown that patient race and ethnicity are associated with patient attitudes toward communication about end-of-life care. Of note, the demographic variable associated with 2194 Ethics in Cardiopulmonary Medicine

the most barriers was physician race. There is little research on the role of physician race or ethnicity on discussions about end-of-life care. Future research is needed to confirm and explore this relationship to ensure that interventions to improve communication about end-of-life care are culturally appropriate not only for patients but also for physicians. Limitations and Conclusions There were several potential limitations to our study. First, the majority of patients and physicians were in academic institutions and were in one city. These features may limit generalizability to other practice settings and other regions. Second, a large number of statistical comparisons were done and, therefore, one or more of the associations may have occurred by chance. We did not formally control for the number of statistical comparisons because the goal of this study was to generate hypotheses about the barriers and facilitators that could be used in intervention studies to improve the quality of communication. Furthermore, all associations occurred in the direction hypothesized. Nonetheless, it is important to acknowledge that some of the statistical associations may have occurred by chance alone. Third, some physicians had more than one patient enrolled in the study, and, therefore, these physicians were represented more than once. However, most physicians had only one patient (71%), and the highest number of patients any physician had in the study was six, which was the case for only one physician. Our sample size would not allow us to control for this clustering using statistical methods such as hierarchical linear models. Given the low number of patients per physician, it seems unlikely that this clustering had a significant effect on analyses. Nonetheless, this remains a limitation of this study. In addition, our sample size does not allow us to assess for differences across the different practice settings or recruitment strategies. Finally, the overall response rate was approximately 40% and was different with two different recruitment methods. The different response rate suggests that having a clinician known to the patient introduce the study may be a more effective method of recruiting patients than mailings for this type of study. Patients unwilling to participate in a study concerning end-of-life care may differ in their attitudes toward this care from those willing to participate. Although these results may not be generalizable to patients unwilling to participate in such research, there is no ethical alternative to disclosing the topic of research and obtaining full consent for participation. Future studies should find ways to improve participation rates in studies concerning end-of-life care. We currently have a limited understanding of the perspective of patients with severe COPD with regard to the problems and issues they face in confronting advance-care planning. There is evidence 3,9,29 that end-of-life care for patients with severe COPD may be worse than for patients with other diseases. Despite the fact that the majority of patients with severe COPD report that they wish to discuss their preferences for end-of-life care with their physicians, only a minority actually do so. 4,5 We have identified specific barriers and facilitators to this communication, and this description of barriers and facilitators is the first step in overcoming barriers and taking advantage of facilitators. Our findings may be useful in future interventions to improve communication about end-of-life care for these patients. References 1 Mannino DM. COPD: epidemiology, prevalence, morbidity and mortality, and disease heterogeneity. Chest 2002; 121(suppl):121S 126S 2 Connors AF, Dawson NV, Thomas C, et al. Outcomes following acute exacerbation of severe chronic obstructive lung disease. Am J Respir Crit Care Med 1996; 959 967 3 Claessens MT, Lynn J, Zhong Z, et al. Dying with lung cancer or chronic obstructive pulmonary disease: insights from SUP- PORT. J Am Geriatr Soc 2000; 48:S146 S153 4 Heffner JE, Fahy B, Hilling L, et al. Outcomes of advance directive education of pulmonary rehabilitation patients. Am J Respir Crit Care Med 1997; 155:1055 1559 5 Heffner JE, Fahy B, Hilling L, et al. Attitudes regarding advance directives among patients in pulmonary rehabilitation. Am J Respir Crit Care Med 1996; 154:1735 1740 6 Curtis JR, Wenrich MD, Carline JD, et al. Patients perspectives on physicians skills at end-of-life care: differences between patients with COPD, cancer, and AIDS. Chest 2002; 122:356 362 7 Pauwels RA, Buist AS, Calverley CR, et al. Global strategy for the diagnosis, management, and prevention of chronic obstructive pulmonary disease, NHLBI/WHO Workshop Summary. Am J Respir Crit Care Med 2001; 163:1256 1276 8 American Thoracic Society. Standards for the diagnosis and management of patients with COPD. Available at: http:// www.thoracic.org/copd; accessed March 9, 2005 9 Gore JM, Brophy CJ, Greenstone MA. How do we care for patients with end stage chronic obstructive pulmonary disease (COPD)? A comparison of palliative care and quality of life in COPD and lung cancer. Thorax 2000; 55:1000 1006 10 Curtis JR, Engelberg RA, Nielsen EL, et al. Patient-physician communication about end-of-life care for patients with severe COPD. Eur Respir J 2004; 24:200 205 11 Stapleton R, Nielsen EL, Engelberg RA, et al. Association of depression and life-sustaining treatment preferences in patients with COPD. Chest 2005; 127:328 334 12 American Thoracic Society. Standards for the diagnosis and care of patients with chronic obstructive pulmonary disease. Am J Respir Crit Care Med 1995; 152:S77 S120 13 Curtis JR, Patrick DL. Barriers to communication about end-of-life care in AIDS patients. J Gen Intern Med 1997; 12:736 741 14 Curtis JR, Patrick DL, Caldwell E, et al. Why don t patients www.chestjournal.org CHEST / 127 / 6/ JUNE, 2005 2195

with AIDS and their clinicians talk about end-of-life care? Barriers to communication for patients with AIDS and their primary care clinicians. Arch Intern Med 2000; 160:1597 1601 15 Denzin NK, Lincoln YS. The handbook of qualitative research. Thousand Oaks, CA: Sage Publications, 1994; 576 16 Upadya A, Muralidharan V, Thorevska N, et al. Patient, physician, and family member understanding of living wills. Am J Respir Crit Care Med 2002; 166:1430 1435 17 Sulmasy DP, Terry PB, Weisman CS, et al. The accuracy of substituted judgment in patients with terminal diagnoses. Ann Intern Med 1998; 128:621 629 18 Uhlmann RF, Pearlman RA, Cain KC. Physicians and spouses predictions of elderly patients resuscitation preferences. J Gerontol 1988; 43:M115 M121 19 Seckler AB, Meier DE, Mulvihill M, et al. Substituted judgment: how accurate are proxy predictions? Ann Intern Med 1991; 115:92 98 20 The SUPPORT Principal Investigators. A controlled trial to improve care for seriously ill hospitalized patients: the study to understand prognoses and preferences for outcomes and risks of treatments (SUPPORT). JAMA 1995; 274:1591 1598 21 Tulsky JA, Fischer GS, Rose MR, et al. Opening the black box: how do physicians communicate about advance directives? Ann Intern Med 1998; 129:441 449 22 Tulsky JA, Chesney MA, Lo B. How do medical residents discuss resuscitation with patients? J Gen Intern Med 1995; 10:436 442 23 Carline JD, Curtis JR, Wenrich MD, et al. Physicians interactions with health care teams and systems in the care of dying patients: perspectives of dying patients, family members, and health care providers. J Pain Symptom Manage 2003; 25:19 28 24 Rabow MW, Petersen J, Schanche K, et al. The comprehensive care team: a description of a controlled trial of care at the beginning of the end of life. J Palliat Med 2003; 6:489 499 25 Steinhauser KE, Christakis NA, Clipp EC, et al. Factors considered important at the end of life by patients, family, physicians, and other care providers. JAMA 2000; 284:2476 2482 26 Steinhauser KE, Clipp EC, McNeilly M, et al. In search of a good death: observations of patients, families, and providers. Ann Intern Med 2000; 132:825 832 27 Blackhall LJ, Murphy ST, Frank G, et al. Ethnicity and attitudes toward patient autonomy. JAMA 1995; 274:820 825 28 Blackhall LJ, Frank G, Murphy ST, et al. Ethnicity and attitudes towards life sustaining technology. Soc Sci Med 1999; 48:1779 1789 29 Edmonds P, Karlsen S, Khan S, et al. A comparison of the palliative care needs of patients dying from chronic respiratory diseases and lung cancer. Palliat Med 2001; 15:287 295 2196 Ethics in Cardiopulmonary Medicine