- Email: [email protected]
Do-Not-Resuscitate Decisions in the Medical ICU
ethics in cardiopulmonary medicine Do-Not-Resuscitate Decisions in the Medical ICU* Comparing Physician and Nurse Opinions Ltc Am H. Eliasson, MC, USA; Robin S. Howard, MA; Col Kenneth G. Torrington, MC, USA; Ltc Thomas A. Dillard, MC, USA; and Col Yancy Y Phillips, MC, USA
Study objective: To detennine how soon after admission to a medical ICU physicians and nurses decide that attempts at resuscitation are inappropriate and how frequently physicians and nurses disagree about do-not-resuscitate (DNR) decisions. Design: Prospective, opinion survey of care providers. Setting: Ten-bed adult medical ICU in a university-afftliated tertiary care referral hospital. Patients: Consecutive adult medical ICU admissions. Interventions: Over 10 months, physicians and nurses were surveyed independently every day regarding their opinions about DNR issues on each patient in the ICU. Measurements: ICU day when DNR order was deemed appropriate by either physicians or nurses. Results: Of 368 consecutive admissions, 84 (23%) patients were designated DNR during their ICU stay. In 6 of these 84 cases (7%), the responsible nurse did not agree that DNR orders were appropriate. In the remaining 78 patients designated DNR, the median time for physicians to recommend DNR (median, 1 day; range, 0 to 22 days) was not significantly different from the median time for nurses (median, 1 day; range, 0 to 13 days); (p=0.45). For the 284 patients not designated DNR, physicians and nurses both believed DNR was appropriate in 14 cases (5%), but a DNR order was not written five times (2%) because there was not time to do so and nine times (3%) because patient or family did not concur. Physicians and nurses disagreed about a DNR recommendation in 33 of the 284 patients not designated DNR (12%). Physicians were more likely to believe that DNR was appropriate than were nurses (p<0.0005), with physicians alone recommending DNR 29 times (10%) and nurses alone favoring DNR in four cases (1 %). Conclusions: At our institution, recognition of DNR appropriateness by nurses and physicians occurs over a similar time frame. However, physicians are more likely to recommend DNR in cases of disagreement between nurses and physicians. (CHEST 1997; 111:1106-11) Key words: do-not-resuscitate orders; ethics; nurse opinions; physician opinions Abbreviations: APACHE Il=acute physiology and chronic health evaluation (version 2) ; CI=confidence intervals; DNR= do not resuscitate
*From the D epartments of Me dicine and Clinical Investigation, Walter Reed Army Medical Cente r, Washington, DC, and the Department of Medicine, Uniformed Services University of the Health Sciences, Bethesda, Md. The opinions or assertions contained herein are the private views of the authors and are not to be construed as official or as reflecting the views of the Department of the Army or the D epartment of Defense. Manuscript received June 11, 1996; revision accepted October 1. Reprint requests: Ltc Am H. Eliasson, MC, USA , 12515 Davan Dr, Silver Spring, MD, 20904 1106
the medical ICU at Walter Reed Army Medical I nCenter, a vocal contingent of nurses asserted that
the physician team often lost sight of realistic goals for some critically ill patients. They believed that patients and their families, and the entire medical team, were frequently subjected to an unnecessary ordeal that often included invasive studies, monitoring, and ventilatory support even though no realistic chance for recovery existed. This debate has been Ethics in Cardiopulmonary Medicine
framed in published literature, but without supporting data. 1 If the nurses' impression could be verified, there would be obvious implications for both medical costs and patient comfort. Warranted changes in practice could have a powerful influence on the morale of the medical staff, especially the nursing staff, and could improve quality of care while cutting cost. 2 The physician staff believed that all important discussions regarding end-of-life decisions did include the nurses and that their opinions about do-not-resuscitate (DNR) issues were routinely solicited. The physicians believed that delays in writing DNR orders stemmed from a variety of sources, including uncertainty about prognosis, difficulty reaching surrogate decision makers for incompetent patients, and the inherent lag time which patients and families require to adjust to new and desperate circumstances. To address these unsettling impressions, we performed a quality improvement study to examine three questions in our ICU: (1) When do physicians and nurses arrive at the DNR decision? (2) How often do physicians and nurses disagree about DNR decisions? (3) How often do patients or their surrogates disagree with the recommendations of the medical staff?
MATERIALS AND METHODS Walter Reed Army Medical Center is a tertiary care, university-affiliated teaching hospital. All patients in the medical ICU are cared for by one team of physicians comprised of a staff attending physician certified in pulmonary and critical care medicine, a fellow in pulmonary and critical care medicine, two internal medicine residents, and three interns from a variety of backgrounds. One-on-one nursing care is delivered by a registered nurse or a licensed practical nurse who is supervised directly by a registered nurse. All nurses have graduated from a credentialed intensive care nursing course. For a 10-month period, demographic data were gathered on all patients on admission to the medical ICU. In addition, during the first half of the study, each patient's mental and general neurologic status was evaluated using the Glasgow Coma Scale. During the second half of the study, acute physiology and chronic health evaluation (APACHE II) scores were calculated on every admission as an index of severity of illness. 3 For each day of every adult medical ICU patient's stay, the medical team (attending physician and fellow) and each nurse assigned to bedside care were independently surveyed by the principal investigator or his or her assistant for their opinions about DNR issues involving patients under their direct care. When the principal investigator attended in the medical ICU, his assistant performed all opinion surveys. Every interview was conducted using the same, simple, objective, verbally administered questionnaire. Demographic and clinical characteristics of subjects with DNR orders and patients without DNR orders were compared using the two-sample t test, Wilcoxon rank sum test, and Fisher's Exact Test (two-tailed). The time to DNR decision for each care
provider was calculated as the difference between the date of ICU admission and the date that either the physician team or nurse decided that a DNR order would be appropriate for the patient. The difference in the time to DNR decision between the teams was analyzed using the Wilcoxon signed rank test. Disagreement regarding the DNR decision was examined using McNemar's test for matched proportions. Data are expressed as means (::'::SDs), medians (with ranges), or proportions with 95% confidence intervals (95% CI).
RESULTS Over a 10-month period, 368 consecutive admissions were studied. Age ranged from 16 to 102 years, with a mean (±:SD) of 60.8 (±:15.5) years. There were 216 men (59%) and 152 women (41 %). Characteristics of patients for whom a D NR order was written and those patients not designated DNR are presented in Table 1. The mean Glasgow Coma Scale was 13.4 (±:3.5) points in the aggregate group. The mean APACHE II score was 14.1 (±:8.8) for the patients in aggregate. Length of medical ICU stay ranged from 1 to 226 days: 22% of patients stayed 1 day, 56% stayed from 2 to 5 days, and 22% stayed longer than 5 days. Patients designated DNR stayed in the medical ICU for a median of 2 days (range, 0 to 34 days) after the DNR order was written. Eighty-eight percent of DNR patients stayed 7 days or less after the DNR order was written. Over 10 months, there were 368 consecutive admissions to the medical ICU, and DNR orders were written for 84 (23%) of these patients. There were six cases (7%) in whom the nurses did not agree that DNR orders were appropriate (Fig 1). In the remaining 78 patients designated DNR, the median time for physicians to recommend DNR was 1 day (range, 1 to 22 days), which was not significantly different from the median time for nurses (median, 1 day; range, 1 to 13 days; p=0.45). Nurses suggested D NR on the same day as physicians in 46
Table !-Characteristics of Patients
DNR Order Written (n=84) Age, yr, mean±SD 66.2±14.3 Male, no. (%) 48 (57) APACHE II score, mean±SD 1 22.6±8.5 Glasgow Coma Scale, mean±SD§ 10.8±4.7 5 (1-39) Length of stay, d, median (range) Mortality in medical ICU, No. (%) 32 (38)
NoDNR Order (n=284) 59.2±15.5* 166 (59)1 12.3±7.8* 14.3±2.3* 2 (1-226)* 14 (5)*
*p<0.005. 1 p=0.81. 1n=31 and n=145, respectively. §n=43 and n=122, respectively. CHEST I 111 I 4 I APRIL, 1997
1107
Patients Not Made DNA {n=284)
Patients Made ON R {n=84) 93%
r;;]MD & RN Agree- YES il •MD & RN Agree- NO
:ll
• MD-YES & RN-NO • MD-NO & RN-YES FIGURE 1. Pie graph depicting the frequency of agreement between physicians and nurses about the approp1iateness of DNR orders, both for patients who were and those who were not designated DNR. Disagree ments by patients (or surrogates ) are also depicted.
patients (50%), preceded the physicians' opinion in 15 cases (19%), and suggested DNR after the physicians for 17 patients (22%) . Of the 284 patients not designated DNR, physicians and nurses agreed DNR was appropriate in 14 cases (5%), but a DNR order was not written, five times (2%) because there was not enough time and nine times (3%) because the patient or family did not concur. Physicians and nurses disagreed with each other about a DNR recommendation in 33 cases (12%). Physicians alone recommended DNR in 29 cases (10%), while nurses alone favored DNR in four cases (1 %) . In these circumstances, physicians were significantly more likely to recommend DNR compared with nurses (p<0.0005). Reasons cited by nurses and physicians in cases of disagreement are presented in Table 2. For the 84 patients designated DNR, the physicians suggested DNR on the same date as the DNR order was written for 73 patients (87%) . T en patients (12%) were made DNR 1 to 5 days after the physician team first suggested DNR. In one case, the patient was designated DNR 11 days after the physicians' recommendation. Twenty-four of the 84 patients (29%) with DNR orders specified their own DNR status on ICU admission. DNR opinions of nurses and physicians were correlated with the patients' medical ICU survival. In 127 patients for whom physicians believed that a DNR order was appropriate, 40 patients died in the ICU (31%; 95% CI, 24 to 40%). In 96 cases in which 1108
nurses recommended DNR, 37 patients died (39%; 95% CI, 29 to 49%). In patients for whom physicians decided against recommending DNR (241 cases), only six died (3%; 95% CI, 1 to 5%). When nurses recommended against DNR orders (272 cases), there were nine deaths (3%; 95% CI, 2 to 6%). Comparison of physicians and nurses did not show statistically significant differences for any of these parameters. If physicians and nurses agreed that DNR was appropriate (92 cases), 37 patients died (40%; 95% CI, 30 to 51%) and when physicians and
Table 2-Reasons Cited for Decisions When Physicians and Nurses Disagreed Patients designated DNR (n=6): MD-Yes DNR
RN-No DNR
Terminal illness (3) Patient requests (2) Poor prognosis (l )
Condition reversible (3 ) Condition stable (l ) Patient has good faculties (l ) Patient indecision (l )
Patients not designated DNR (n=29): MD-Yes DNR
RN-No DNR
Terminal illness (ll ) Poor prognosis (ll ) Poor qual ity of life (5) Patient requests (2) Patients not designated DNR (n=4): RN-Yes DNR
Poor prognosis (2) Poor quality of life (l ) Terminal illness (l )
Condition reversible (12) Condition stable (9) Patient indecision (5) Ill-defined condition (3) MD-NoDNR
Condition reversible ( 2) Patient indecision (2)
Ethics in Cardiopulmonary Medicine
nurses agreed not to implement DNR (237 cases), six patients died (3%; 95% CI, 1 to 5%).
DISCUSSION
The European literature contains major work comparing the opinions of physicians and nurses. Abizanda et al4 from Spain, surveyed 183 physicians and 198 nurses to determine their attitudes about not resuscitating patients, limiting care in certain cases, and accepting responsibility for decisions when disagreements occur. On a broad range of ethical issues posed by the study's questionnaire, attitudes of nurses and physicians were in close agreement. Likewise, an English study by Davies et al5 reported results of a mailed survey of 249 nurses, 300 geriatricians, and 300 consultant physicians. There was concurrence on which factors are important to consider in deciding DNR issues. 5 All three groups of professionals believed that the patient's prognosis and wishes were most important, while age alone was unimportant. In the United States, Gillick et al6 compared physician and nurse preferences for their own care using 12 possible interventions in four hypothetical scenarios. They reported that both physicians and nurses were unlikely to wish aggressive treatment if they had become terminally ill, demented, or persistently vegetative. Nurses had a significantly higher refusal rate for aggressive care than physicians. Factors that predicted refusal patterns were age and being a nurse. An individual's values obviously influence behaviors and attitudes and physicians must be aware that nursing and patient values may differ vastly from their own. Bedell and Delbanco 7 demonstrated this in a 1981 study of patients who had suffered cardiac arrest. Only 19% had discussed resuscitation with a physician before the event. Among the 24 competent survivors of resuscitation (from the entire series of 154 resuscitated patients), only a very weak correlation between physician and patient preferences existed. Johnston et al 8 reported for the End of Life Study Group in 1995 that patients believed discussion about advance directives should occur at a younger age, earlier in the natural history of disease, and earlier in the patient-physician relationship than did the physicians. We believe that physicians could benefit from consulting with their nursing colleagues in order to explore a patient's preferences regarding DNR issues. Asch 9 recently explored the attitudes and practices of critical care nurses with regard to euthanasia and assisted suicide, noting that "critical care nurses frequently care for patients who wish to die, and these nurses are often in a position to hasten their
deaths." Asch's dramatic report included findings that 20% of the nurses in the study had engaged in these practices and that an additional 40% had wanted to engage in euthanasia but did not for a variety of reasons, most commonly for fear of getting caught. The willingness of so many nurses to perform euthanasia or assisted suicide was explained by a chorus of themes, which included "concern about the overuse of life-sustaining technology, aprofound sense of responsibility for the patient's welfare, a desire to relieve suffering, and a desire to overcome the perceived unresponsiveness of physicians toward that suffering." These reported attitudes are consistent with findings of the SUPPORT study in which physicians' behavior was not affected by nurses' reports of the preferences of terminally ill patients. 10 Understandably, this lack of physician responsiveness is at the heart of frustration experienced by critical care nurses who are tasked to remain at the bedside and journey with their patients through circumstances that can be indescribably difficult. The characteristics of our adult medical ICU patients are not unusual. Compared with other ICU s, our rate of 23% of patients being designated DNR is on the high end of the spectrum.l 1 The fact that DNR patients are older and sicker (higher APACHE II scores) than patients without a DNR order concurs with common-sense expectations and previous reports. 11 - 13 Parker et al 14 have previously observed at our institution that one factor that discriminates between patients who have a D NR order and those who do not is tl1eir level of mental competence as measured by the Glasgow Coma Scale. 14 As the patient's mental capacity diminishes, physicians and surrogate decisions makers more readily comprehend the grim prognosis, and the odds of the patient being designated DNR increase.l 5 · 16 Our finding of markedly lower Glasgow Coma Scale scores in D NR patients confirms this observation. In contrast to previously published concerns that DNR orders often lead to a diminished level of patient care,l1- 19 our DNR patients often remained in the ICU for days after DNR orders were written. Our patients were not transferred to medical wards for comfort care unless the intensity of required nursing activities could be satisfactorily provided. This finding is similar to the experience of Smedira et al, 16 who reported a longer median duration of intensive care among patients from whom life support was withheld or withdrawn compared to other patients who survived \vith aggressive care or those who died in the ICU despite aggressive resuscitative efforts. Our study showed very close agreement in the timing of D NR decisions between the physicians and CHEST I 111 I 4 I APRIL, 1997
1109
nurses in our institution's medical ICU. How do we account for the strongly held opinion of the few vocal nurses whose allegations stimulated this study? We believe that our nurses' incorrect impression that they were frequently ready to designate DNR earlier than physicians probably stemmed from their role of providing one-on-one bedside care and witnessing fluctuating clinical courses with patients and their families. Physicians must understand the emotional stresses encountered by nurses and the impact the nursing role has on the formulation of their opinions about the care plan. 20 When opinions are passionately held, even brief disagreements among healthcare providers can generate divisiveness. If differences involve issues as important as end-of-life decisions, the educational and liaison role that nurses often assume for their patients is disrupted. This role, previously described as "culture brokering,"2l highlights the powerful influence the nurse may have on acceptance by the patient (or the surrogate decision maker) of the medical recommendation for a DNR order. For communication with the patient concerning DNR recommendations to go smoothly, it is extremely important that all members of the multidisciplinary care team agree about the patient's progress, prognosis, perceived quality of life, and expressed desires. This approach has been formally outlined by Luce and Fink22 as a result of their experiences at San Francisco General Hospital. Designating a patient DNR is not equivalent to deciding to withdraw intensive care. To explore this issue more fully, it would have been useful to ask physicians and nurses their opinions about withdrawing care. This line of questioning might have shown a difference between physicians and nurses that was not detected by inquiries about DNR. Further investigation will be required to clarify this issue. Patients or families in our medical ICU infrequently disagreed with the medical recommendation for a DNR order. While this observation is in line with previously reported data by Smedira et al, 16 our report is especially noteworthy in view of the lack of monetary constraints or incentives affecting patients or surrogate decision makers in the military healthcare system. The high degree of compliance in our patient population suggests willingness of either patient or family to accept recommendations of the health-care team, that is "to obey orders." Additionally, the lack of staff concern about malpractice issues may grant physicians the freedom to press their opinions about DNR decisions more vigorously than in other settings where that threat is more palpable. The high level of compliance of our patients may also suggest that military physicians and nurses are sensitive to the patients' needs and desires. 1110
The manner in which physicians discuss end-oflife decisions with patients will influence the patient's or family's likelihood of accepting a DNR order. It was not possible to control for this variable in our observational study. The styles of communication and decision making differ among our ICU staff physicians. While a minority of our doctors ask patients or surrogate decision makers to choose among several possible treatment options, most physicians present medical recommendations firmly and with conviction. This style of communication is more persuasive and convincing to patients and families and would likely reduce disagreements between physicians and nurses as well. A recommendation for implementation of a DNR order by nurses and physicians was associated with a much higher ICU mortality rate (40%) than agreement by care providers that a DNR order was not indicated (3%). Clearly this stems from the fact that DNR orders are appropriate in hopelessly ill patients and those in the final phase of a terminal illness. It is worthwhile commenting that fully 60% of patients designated DNR did survive the ICU stay at our hospital, signifYing an aggressive level of continued care after implementation of a DNR order. A potential weakness of our study is the bias induced by the study itself. Any time something is evaluated, as in our daily questionnaires in the workplace, that system is changed by the intervention. Our daily practice of soliciting opinions of the physician and the nursing teams, even though confidential, potentially improved communication or at least suggested that each team member's opinions were valued. However, we saw no way to avoid this potential bias. Another source of bias was the DNR order itself. If a nurse had not yet decided that a DNR order should be implemented, the physician's written order would influence the nurse's assessment of appropriateness. In this circumstance, the two groups would not arrive at the same conclusion independently. Of note, however, the impression among the nurses that they would advocate D NR earlier than physicians is not supported. The fact that this study was instituted and given serious consideration at our hospital provides evidence of an evolution away from the paternalistic, physician-driven model of health care to a multidisciplinary team approach. We believe that critical decisions about life support issues are best addressed with input from both nurses and physicians to enhance communication with the patient and family. The favorable statistical findings of our study point to the success of our current system of bedside rounds, in which opinions of the multidisciplinary healthcare team are usually solicited. However, the fact Ethics in Cardiopulmonary Medicine
that some nurses felt disenfranchised from the decision-making process is an indicator that improvements can yet be made in our system. On the basis of our findings, and in view of the national debate about end-of-life decisions, we strongly encourage an active and open dialogue between health-care providers. We also encourage an ongoing conversation about DNR decisions with patients and their family members, as early in the course of illness as possible.
REFERENCES 1 Slater AL, Fassnacht-Hanrahan K, Slater H, et a!. From hopeful to hopeless ...when do we write "Do not resuscitate"? Focus Crit Care 1991; 18:476-79 2 Berwick DM. Eleven worthy aims for clinical leadership of health system reform. JAMA 1994; 272:797-802 3 Knaus WA, Draper EA, Wagner DP, eta!. APACHE II: a severity of disease classification system. Crit Care Med 1985; 13:818-29 4 Abizanda R, Almendros CL, Balerdi PB. Ethical aspects of intensive medicine: results of an opinion survey. Med Clin (Bare) 1994; 102:521-26 5 Davies KN, King D, Silas JH. Professional attitudes to cardiopulmonary resuscitation in departments of geriatric and general medicine. J R Coli Physicians Lond 1993; 2:127-30 6 Gillick MR, Hesse K, Mazzapica N. Medical technology at the end of life-what would physicians and nurses want for themselves? Arch Intern Med 1993; 153:2542-47 7 Bedell SE, Delbanco TL. Choices about cardiopulmonary resuscitation in the hospital-when do physicians talk with patients? N Eng! J Med 1984; 310:1089-93 8 Johnston SC, Pfeifer MP, McNutt R. The discussion about advance directives-patient and physician opinions regarding when and how it should be conducted. Arch Intern Med 1995; 155:1025-30
9 Asch DA. The role of critical care nurses in euthanasia and assisted suicide. N Engl J Med 1996; 334:1374-79 10 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; 274:1591-98 11 Jayes RL, Zimmerman JE, Wagner DP, et al. Do-notresuscitate orders in intensive care units-current practices and recent changes. JAMA 1993; 270:2213-17 12 Clark CD, Lucas K, Stephens L. Ethical dilemmas and decisions concerning the do-not-resuscitate patient undergoing anesthesia. J Am Assoc Nurse Anesthetists 1994; 62: 253-56 13 Johnson RF Jr, Baranowski-Birkmeier T, O'Donnell JB. Advance directives in the medical intensive care unit of a community teaching hospital. Chest 1995; 107:752-56 14 Parker JM, Landry FJ, Phillips YY. Use of do-not-resuscitate orders in an intensive care setting. Chest 1993; 104:1592-96 15 Webster GC, Mazer CD, Potvin CA, eta!. Evaluation of a do not resuscitate policy in intensive care. Can J Anaesth 1991; 38:553-63 16 Smedira NG, Evans BH, Grais LS, et a!. Withholding and withdrawal oflife support from the critically ill. N Eng! J Med 1990; 322:309-15 17 Webster MA. Some decisions can't be easy. RN 1992; 55:19-22 18 Fowler MDM. When did 'do not resuscitate' mean 'do not care'? Heart Lung 1989; 18:424-25 19 Sulmasy DP, Geller G, Faden R, et al. The quality of mercy-caring for patients with 'do not resuscitate' orders. JAMA 1992; 267:682-86 20 Scanlon C. Euthanasia and nursing practice-right question, wrong answer. N Eng! J Med 1996; 334:1401-02 21 Jezewski MA. Do-not-resuscitate status: conflict and culture brokering in critical care units. Heart Lung 1994; 23:458-65 22 Luce JM, Fink C. Communicating with families about withholding and withdrawal of life support. Chest 1992; 101: 1185-86
CHEST /111 /4/ APRIL, 1997
1111
Study objective: To detennine how soon after admission to a medical ICU physicians and nurses decide that attempts at resuscitation are inappropriate and how frequently physicians and nurses disagree about do-not-resuscitate (DNR) decisions. Design: Prospective, opinion survey of care providers. Setting: Ten-bed adult medical ICU in a university-afftliated tertiary care referral hospital. Patients: Consecutive adult medical ICU admissions. Interventions: Over 10 months, physicians and nurses were surveyed independently every day regarding their opinions about DNR issues on each patient in the ICU. Measurements: ICU day when DNR order was deemed appropriate by either physicians or nurses. Results: Of 368 consecutive admissions, 84 (23%) patients were designated DNR during their ICU stay. In 6 of these 84 cases (7%), the responsible nurse did not agree that DNR orders were appropriate. In the remaining 78 patients designated DNR, the median time for physicians to recommend DNR (median, 1 day; range, 0 to 22 days) was not significantly different from the median time for nurses (median, 1 day; range, 0 to 13 days); (p=0.45). For the 284 patients not designated DNR, physicians and nurses both believed DNR was appropriate in 14 cases (5%), but a DNR order was not written five times (2%) because there was not time to do so and nine times (3%) because patient or family did not concur. Physicians and nurses disagreed about a DNR recommendation in 33 of the 284 patients not designated DNR (12%). Physicians were more likely to believe that DNR was appropriate than were nurses (p<0.0005), with physicians alone recommending DNR 29 times (10%) and nurses alone favoring DNR in four cases (1 %). Conclusions: At our institution, recognition of DNR appropriateness by nurses and physicians occurs over a similar time frame. However, physicians are more likely to recommend DNR in cases of disagreement between nurses and physicians. (CHEST 1997; 111:1106-11) Key words: do-not-resuscitate orders; ethics; nurse opinions; physician opinions Abbreviations: APACHE Il=acute physiology and chronic health evaluation (version 2) ; CI=confidence intervals; DNR= do not resuscitate
*From the D epartments of Me dicine and Clinical Investigation, Walter Reed Army Medical Cente r, Washington, DC, and the Department of Medicine, Uniformed Services University of the Health Sciences, Bethesda, Md. The opinions or assertions contained herein are the private views of the authors and are not to be construed as official or as reflecting the views of the Department of the Army or the D epartment of Defense. Manuscript received June 11, 1996; revision accepted October 1. Reprint requests: Ltc Am H. Eliasson, MC, USA , 12515 Davan Dr, Silver Spring, MD, 20904 1106
the medical ICU at Walter Reed Army Medical I nCenter, a vocal contingent of nurses asserted that
the physician team often lost sight of realistic goals for some critically ill patients. They believed that patients and their families, and the entire medical team, were frequently subjected to an unnecessary ordeal that often included invasive studies, monitoring, and ventilatory support even though no realistic chance for recovery existed. This debate has been Ethics in Cardiopulmonary Medicine
framed in published literature, but without supporting data. 1 If the nurses' impression could be verified, there would be obvious implications for both medical costs and patient comfort. Warranted changes in practice could have a powerful influence on the morale of the medical staff, especially the nursing staff, and could improve quality of care while cutting cost. 2 The physician staff believed that all important discussions regarding end-of-life decisions did include the nurses and that their opinions about do-not-resuscitate (DNR) issues were routinely solicited. The physicians believed that delays in writing DNR orders stemmed from a variety of sources, including uncertainty about prognosis, difficulty reaching surrogate decision makers for incompetent patients, and the inherent lag time which patients and families require to adjust to new and desperate circumstances. To address these unsettling impressions, we performed a quality improvement study to examine three questions in our ICU: (1) When do physicians and nurses arrive at the DNR decision? (2) How often do physicians and nurses disagree about DNR decisions? (3) How often do patients or their surrogates disagree with the recommendations of the medical staff?
MATERIALS AND METHODS Walter Reed Army Medical Center is a tertiary care, university-affiliated teaching hospital. All patients in the medical ICU are cared for by one team of physicians comprised of a staff attending physician certified in pulmonary and critical care medicine, a fellow in pulmonary and critical care medicine, two internal medicine residents, and three interns from a variety of backgrounds. One-on-one nursing care is delivered by a registered nurse or a licensed practical nurse who is supervised directly by a registered nurse. All nurses have graduated from a credentialed intensive care nursing course. For a 10-month period, demographic data were gathered on all patients on admission to the medical ICU. In addition, during the first half of the study, each patient's mental and general neurologic status was evaluated using the Glasgow Coma Scale. During the second half of the study, acute physiology and chronic health evaluation (APACHE II) scores were calculated on every admission as an index of severity of illness. 3 For each day of every adult medical ICU patient's stay, the medical team (attending physician and fellow) and each nurse assigned to bedside care were independently surveyed by the principal investigator or his or her assistant for their opinions about DNR issues involving patients under their direct care. When the principal investigator attended in the medical ICU, his assistant performed all opinion surveys. Every interview was conducted using the same, simple, objective, verbally administered questionnaire. Demographic and clinical characteristics of subjects with DNR orders and patients without DNR orders were compared using the two-sample t test, Wilcoxon rank sum test, and Fisher's Exact Test (two-tailed). The time to DNR decision for each care
provider was calculated as the difference between the date of ICU admission and the date that either the physician team or nurse decided that a DNR order would be appropriate for the patient. The difference in the time to DNR decision between the teams was analyzed using the Wilcoxon signed rank test. Disagreement regarding the DNR decision was examined using McNemar's test for matched proportions. Data are expressed as means (::'::SDs), medians (with ranges), or proportions with 95% confidence intervals (95% CI).
RESULTS Over a 10-month period, 368 consecutive admissions were studied. Age ranged from 16 to 102 years, with a mean (±:SD) of 60.8 (±:15.5) years. There were 216 men (59%) and 152 women (41 %). Characteristics of patients for whom a D NR order was written and those patients not designated DNR are presented in Table 1. The mean Glasgow Coma Scale was 13.4 (±:3.5) points in the aggregate group. The mean APACHE II score was 14.1 (±:8.8) for the patients in aggregate. Length of medical ICU stay ranged from 1 to 226 days: 22% of patients stayed 1 day, 56% stayed from 2 to 5 days, and 22% stayed longer than 5 days. Patients designated DNR stayed in the medical ICU for a median of 2 days (range, 0 to 34 days) after the DNR order was written. Eighty-eight percent of DNR patients stayed 7 days or less after the DNR order was written. Over 10 months, there were 368 consecutive admissions to the medical ICU, and DNR orders were written for 84 (23%) of these patients. There were six cases (7%) in whom the nurses did not agree that DNR orders were appropriate (Fig 1). In the remaining 78 patients designated DNR, the median time for physicians to recommend DNR was 1 day (range, 1 to 22 days), which was not significantly different from the median time for nurses (median, 1 day; range, 1 to 13 days; p=0.45). Nurses suggested D NR on the same day as physicians in 46
Table !-Characteristics of Patients
DNR Order Written (n=84) Age, yr, mean±SD 66.2±14.3 Male, no. (%) 48 (57) APACHE II score, mean±SD 1 22.6±8.5 Glasgow Coma Scale, mean±SD§ 10.8±4.7 5 (1-39) Length of stay, d, median (range) Mortality in medical ICU, No. (%) 32 (38)
NoDNR Order (n=284) 59.2±15.5* 166 (59)1 12.3±7.8* 14.3±2.3* 2 (1-226)* 14 (5)*
*p<0.005. 1 p=0.81. 1n=31 and n=145, respectively. §n=43 and n=122, respectively. CHEST I 111 I 4 I APRIL, 1997
1107
Patients Not Made DNA {n=284)
Patients Made ON R {n=84) 93%
r;;]MD & RN Agree- YES il •MD & RN Agree- NO
:ll
• MD-YES & RN-NO • MD-NO & RN-YES FIGURE 1. Pie graph depicting the frequency of agreement between physicians and nurses about the approp1iateness of DNR orders, both for patients who were and those who were not designated DNR. Disagree ments by patients (or surrogates ) are also depicted.
patients (50%), preceded the physicians' opinion in 15 cases (19%), and suggested DNR after the physicians for 17 patients (22%) . Of the 284 patients not designated DNR, physicians and nurses agreed DNR was appropriate in 14 cases (5%), but a DNR order was not written, five times (2%) because there was not enough time and nine times (3%) because the patient or family did not concur. Physicians and nurses disagreed with each other about a DNR recommendation in 33 cases (12%). Physicians alone recommended DNR in 29 cases (10%), while nurses alone favored DNR in four cases (1 %) . In these circumstances, physicians were significantly more likely to recommend DNR compared with nurses (p<0.0005). Reasons cited by nurses and physicians in cases of disagreement are presented in Table 2. For the 84 patients designated DNR, the physicians suggested DNR on the same date as the DNR order was written for 73 patients (87%) . T en patients (12%) were made DNR 1 to 5 days after the physician team first suggested DNR. In one case, the patient was designated DNR 11 days after the physicians' recommendation. Twenty-four of the 84 patients (29%) with DNR orders specified their own DNR status on ICU admission. DNR opinions of nurses and physicians were correlated with the patients' medical ICU survival. In 127 patients for whom physicians believed that a DNR order was appropriate, 40 patients died in the ICU (31%; 95% CI, 24 to 40%). In 96 cases in which 1108
nurses recommended DNR, 37 patients died (39%; 95% CI, 29 to 49%). In patients for whom physicians decided against recommending DNR (241 cases), only six died (3%; 95% CI, 1 to 5%). When nurses recommended against DNR orders (272 cases), there were nine deaths (3%; 95% CI, 2 to 6%). Comparison of physicians and nurses did not show statistically significant differences for any of these parameters. If physicians and nurses agreed that DNR was appropriate (92 cases), 37 patients died (40%; 95% CI, 30 to 51%) and when physicians and
Table 2-Reasons Cited for Decisions When Physicians and Nurses Disagreed Patients designated DNR (n=6): MD-Yes DNR
RN-No DNR
Terminal illness (3) Patient requests (2) Poor prognosis (l )
Condition reversible (3 ) Condition stable (l ) Patient has good faculties (l ) Patient indecision (l )
Patients not designated DNR (n=29): MD-Yes DNR
RN-No DNR
Terminal illness (ll ) Poor prognosis (ll ) Poor qual ity of life (5) Patient requests (2) Patients not designated DNR (n=4): RN-Yes DNR
Poor prognosis (2) Poor quality of life (l ) Terminal illness (l )
Condition reversible (12) Condition stable (9) Patient indecision (5) Ill-defined condition (3) MD-NoDNR
Condition reversible ( 2) Patient indecision (2)
Ethics in Cardiopulmonary Medicine
nurses agreed not to implement DNR (237 cases), six patients died (3%; 95% CI, 1 to 5%).
DISCUSSION
The European literature contains major work comparing the opinions of physicians and nurses. Abizanda et al4 from Spain, surveyed 183 physicians and 198 nurses to determine their attitudes about not resuscitating patients, limiting care in certain cases, and accepting responsibility for decisions when disagreements occur. On a broad range of ethical issues posed by the study's questionnaire, attitudes of nurses and physicians were in close agreement. Likewise, an English study by Davies et al5 reported results of a mailed survey of 249 nurses, 300 geriatricians, and 300 consultant physicians. There was concurrence on which factors are important to consider in deciding DNR issues. 5 All three groups of professionals believed that the patient's prognosis and wishes were most important, while age alone was unimportant. In the United States, Gillick et al6 compared physician and nurse preferences for their own care using 12 possible interventions in four hypothetical scenarios. They reported that both physicians and nurses were unlikely to wish aggressive treatment if they had become terminally ill, demented, or persistently vegetative. Nurses had a significantly higher refusal rate for aggressive care than physicians. Factors that predicted refusal patterns were age and being a nurse. An individual's values obviously influence behaviors and attitudes and physicians must be aware that nursing and patient values may differ vastly from their own. Bedell and Delbanco 7 demonstrated this in a 1981 study of patients who had suffered cardiac arrest. Only 19% had discussed resuscitation with a physician before the event. Among the 24 competent survivors of resuscitation (from the entire series of 154 resuscitated patients), only a very weak correlation between physician and patient preferences existed. Johnston et al 8 reported for the End of Life Study Group in 1995 that patients believed discussion about advance directives should occur at a younger age, earlier in the natural history of disease, and earlier in the patient-physician relationship than did the physicians. We believe that physicians could benefit from consulting with their nursing colleagues in order to explore a patient's preferences regarding DNR issues. Asch 9 recently explored the attitudes and practices of critical care nurses with regard to euthanasia and assisted suicide, noting that "critical care nurses frequently care for patients who wish to die, and these nurses are often in a position to hasten their
deaths." Asch's dramatic report included findings that 20% of the nurses in the study had engaged in these practices and that an additional 40% had wanted to engage in euthanasia but did not for a variety of reasons, most commonly for fear of getting caught. The willingness of so many nurses to perform euthanasia or assisted suicide was explained by a chorus of themes, which included "concern about the overuse of life-sustaining technology, aprofound sense of responsibility for the patient's welfare, a desire to relieve suffering, and a desire to overcome the perceived unresponsiveness of physicians toward that suffering." These reported attitudes are consistent with findings of the SUPPORT study in which physicians' behavior was not affected by nurses' reports of the preferences of terminally ill patients. 10 Understandably, this lack of physician responsiveness is at the heart of frustration experienced by critical care nurses who are tasked to remain at the bedside and journey with their patients through circumstances that can be indescribably difficult. The characteristics of our adult medical ICU patients are not unusual. Compared with other ICU s, our rate of 23% of patients being designated DNR is on the high end of the spectrum.l 1 The fact that DNR patients are older and sicker (higher APACHE II scores) than patients without a DNR order concurs with common-sense expectations and previous reports. 11 - 13 Parker et al 14 have previously observed at our institution that one factor that discriminates between patients who have a D NR order and those who do not is tl1eir level of mental competence as measured by the Glasgow Coma Scale. 14 As the patient's mental capacity diminishes, physicians and surrogate decisions makers more readily comprehend the grim prognosis, and the odds of the patient being designated DNR increase.l 5 · 16 Our finding of markedly lower Glasgow Coma Scale scores in D NR patients confirms this observation. In contrast to previously published concerns that DNR orders often lead to a diminished level of patient care,l1- 19 our DNR patients often remained in the ICU for days after DNR orders were written. Our patients were not transferred to medical wards for comfort care unless the intensity of required nursing activities could be satisfactorily provided. This finding is similar to the experience of Smedira et al, 16 who reported a longer median duration of intensive care among patients from whom life support was withheld or withdrawn compared to other patients who survived \vith aggressive care or those who died in the ICU despite aggressive resuscitative efforts. Our study showed very close agreement in the timing of D NR decisions between the physicians and CHEST I 111 I 4 I APRIL, 1997
1109
nurses in our institution's medical ICU. How do we account for the strongly held opinion of the few vocal nurses whose allegations stimulated this study? We believe that our nurses' incorrect impression that they were frequently ready to designate DNR earlier than physicians probably stemmed from their role of providing one-on-one bedside care and witnessing fluctuating clinical courses with patients and their families. Physicians must understand the emotional stresses encountered by nurses and the impact the nursing role has on the formulation of their opinions about the care plan. 20 When opinions are passionately held, even brief disagreements among healthcare providers can generate divisiveness. If differences involve issues as important as end-of-life decisions, the educational and liaison role that nurses often assume for their patients is disrupted. This role, previously described as "culture brokering,"2l highlights the powerful influence the nurse may have on acceptance by the patient (or the surrogate decision maker) of the medical recommendation for a DNR order. For communication with the patient concerning DNR recommendations to go smoothly, it is extremely important that all members of the multidisciplinary care team agree about the patient's progress, prognosis, perceived quality of life, and expressed desires. This approach has been formally outlined by Luce and Fink22 as a result of their experiences at San Francisco General Hospital. Designating a patient DNR is not equivalent to deciding to withdraw intensive care. To explore this issue more fully, it would have been useful to ask physicians and nurses their opinions about withdrawing care. This line of questioning might have shown a difference between physicians and nurses that was not detected by inquiries about DNR. Further investigation will be required to clarify this issue. Patients or families in our medical ICU infrequently disagreed with the medical recommendation for a DNR order. While this observation is in line with previously reported data by Smedira et al, 16 our report is especially noteworthy in view of the lack of monetary constraints or incentives affecting patients or surrogate decision makers in the military healthcare system. The high degree of compliance in our patient population suggests willingness of either patient or family to accept recommendations of the health-care team, that is "to obey orders." Additionally, the lack of staff concern about malpractice issues may grant physicians the freedom to press their opinions about DNR decisions more vigorously than in other settings where that threat is more palpable. The high level of compliance of our patients may also suggest that military physicians and nurses are sensitive to the patients' needs and desires. 1110
The manner in which physicians discuss end-oflife decisions with patients will influence the patient's or family's likelihood of accepting a DNR order. It was not possible to control for this variable in our observational study. The styles of communication and decision making differ among our ICU staff physicians. While a minority of our doctors ask patients or surrogate decision makers to choose among several possible treatment options, most physicians present medical recommendations firmly and with conviction. This style of communication is more persuasive and convincing to patients and families and would likely reduce disagreements between physicians and nurses as well. A recommendation for implementation of a DNR order by nurses and physicians was associated with a much higher ICU mortality rate (40%) than agreement by care providers that a DNR order was not indicated (3%). Clearly this stems from the fact that DNR orders are appropriate in hopelessly ill patients and those in the final phase of a terminal illness. It is worthwhile commenting that fully 60% of patients designated DNR did survive the ICU stay at our hospital, signifYing an aggressive level of continued care after implementation of a DNR order. A potential weakness of our study is the bias induced by the study itself. Any time something is evaluated, as in our daily questionnaires in the workplace, that system is changed by the intervention. Our daily practice of soliciting opinions of the physician and the nursing teams, even though confidential, potentially improved communication or at least suggested that each team member's opinions were valued. However, we saw no way to avoid this potential bias. Another source of bias was the DNR order itself. If a nurse had not yet decided that a DNR order should be implemented, the physician's written order would influence the nurse's assessment of appropriateness. In this circumstance, the two groups would not arrive at the same conclusion independently. Of note, however, the impression among the nurses that they would advocate D NR earlier than physicians is not supported. The fact that this study was instituted and given serious consideration at our hospital provides evidence of an evolution away from the paternalistic, physician-driven model of health care to a multidisciplinary team approach. We believe that critical decisions about life support issues are best addressed with input from both nurses and physicians to enhance communication with the patient and family. The favorable statistical findings of our study point to the success of our current system of bedside rounds, in which opinions of the multidisciplinary healthcare team are usually solicited. However, the fact Ethics in Cardiopulmonary Medicine
that some nurses felt disenfranchised from the decision-making process is an indicator that improvements can yet be made in our system. On the basis of our findings, and in view of the national debate about end-of-life decisions, we strongly encourage an active and open dialogue between health-care providers. We also encourage an ongoing conversation about DNR decisions with patients and their family members, as early in the course of illness as possible.
REFERENCES 1 Slater AL, Fassnacht-Hanrahan K, Slater H, et a!. From hopeful to hopeless ...when do we write "Do not resuscitate"? Focus Crit Care 1991; 18:476-79 2 Berwick DM. Eleven worthy aims for clinical leadership of health system reform. JAMA 1994; 272:797-802 3 Knaus WA, Draper EA, Wagner DP, eta!. APACHE II: a severity of disease classification system. Crit Care Med 1985; 13:818-29 4 Abizanda R, Almendros CL, Balerdi PB. Ethical aspects of intensive medicine: results of an opinion survey. Med Clin (Bare) 1994; 102:521-26 5 Davies KN, King D, Silas JH. Professional attitudes to cardiopulmonary resuscitation in departments of geriatric and general medicine. J R Coli Physicians Lond 1993; 2:127-30 6 Gillick MR, Hesse K, Mazzapica N. Medical technology at the end of life-what would physicians and nurses want for themselves? Arch Intern Med 1993; 153:2542-47 7 Bedell SE, Delbanco TL. Choices about cardiopulmonary resuscitation in the hospital-when do physicians talk with patients? N Eng! J Med 1984; 310:1089-93 8 Johnston SC, Pfeifer MP, McNutt R. The discussion about advance directives-patient and physician opinions regarding when and how it should be conducted. Arch Intern Med 1995; 155:1025-30
9 Asch DA. The role of critical care nurses in euthanasia and assisted suicide. N Engl J Med 1996; 334:1374-79 10 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; 274:1591-98 11 Jayes RL, Zimmerman JE, Wagner DP, et al. Do-notresuscitate orders in intensive care units-current practices and recent changes. JAMA 1993; 270:2213-17 12 Clark CD, Lucas K, Stephens L. Ethical dilemmas and decisions concerning the do-not-resuscitate patient undergoing anesthesia. J Am Assoc Nurse Anesthetists 1994; 62: 253-56 13 Johnson RF Jr, Baranowski-Birkmeier T, O'Donnell JB. Advance directives in the medical intensive care unit of a community teaching hospital. Chest 1995; 107:752-56 14 Parker JM, Landry FJ, Phillips YY. Use of do-not-resuscitate orders in an intensive care setting. Chest 1993; 104:1592-96 15 Webster GC, Mazer CD, Potvin CA, eta!. Evaluation of a do not resuscitate policy in intensive care. Can J Anaesth 1991; 38:553-63 16 Smedira NG, Evans BH, Grais LS, et a!. Withholding and withdrawal oflife support from the critically ill. N Eng! J Med 1990; 322:309-15 17 Webster MA. Some decisions can't be easy. RN 1992; 55:19-22 18 Fowler MDM. When did 'do not resuscitate' mean 'do not care'? Heart Lung 1989; 18:424-25 19 Sulmasy DP, Geller G, Faden R, et al. The quality of mercy-caring for patients with 'do not resuscitate' orders. JAMA 1992; 267:682-86 20 Scanlon C. Euthanasia and nursing practice-right question, wrong answer. N Eng! J Med 1996; 334:1401-02 21 Jezewski MA. Do-not-resuscitate status: conflict and culture brokering in critical care units. Heart Lung 1994; 23:458-65 22 Luce JM, Fink C. Communicating with families about withholding and withdrawal of life support. Chest 1992; 101: 1185-86
CHEST /111 /4/ APRIL, 1997
1111