- Email: [email protected]
Are enough explicit resuscitation directives obtained for critically ill patients?
COMMENTARY
COMMENTARY
Are enough explicit resuscitation directives obtained for critically ill patients? See page 1941
Decisions to withhold or withdraw life-sustaining therapy necessitate the balancing of two undesirable possibilities: failure to provide life-support to a patient who might recover to an acceptable level of function, and failure to withhold or withdraw life-supporting therapy to a patient who has very little likelihood of returning to an acceptable level of function. In this issue of The Lancet, Deborah Cook and colleagues present the results of an observational study, conducted in 15 intensive-care units (ICUs) in four countries, of practices related to explicit decisions (as stated in the progress notes or order sheets) to resuscitate or not resuscitate in the event of a cardiopulmonary arrest. Their findings contribute usefully to the current discussion about the role of patients’ autonomy in decisions about lifesustaining procedures. To find out the prevalence of explicit directives to or not to carry out cardiopulmonary resuscitation (CPR), to clarify the factors that predict whether there would be explicit directives, and to define how the explicit orders were obtained, the investigators collected data on 2916 patients undergoing a first admission to an ICU. For only 10·9% of patients were there explicit directives, half of which were do-not-resuscitate (DNR) orders. Predictors of explicit directives (either to resuscitate or not) included severe illness and admission to the ICU at night or on weekends. Predictors of DNR orders were advanced age (odds ratio 8·8 for patients >75 years) and inability of patients to participate in decisions. Perhaps the most striking finding, though, was the wide variability in practice between countries, between cities within one country (Canada), and between hospitals within the same city. The investigators correctly identify the strengths of their approach: a large cohort, well-defined estimates, and systematic, comprehensive data collection. They also acknowledge the limitations of the study. Because all of the Canadian centres (11 of the 15) were university-affiliated teaching hospitals in which ICU physicians coordinated care of the patients, these results cannot be extrapolated widely. However, a reasonable speculation is that a more broadly representative sample would have identified qualitatively similar patterns. The investigators carefully confine their conclusions within the limits of the data. They emphasise the importance of aligning directives with patients’ preferences and with cultural influences. However, if, in highly structured ICU environments, only 10·9% of all patients, only 18% of patients aged over 75, and only 30·2% of the most critically ill patients (Acute Physiology and Chronic Health Evaluation Score >30) received explicit directives, the inevitable question is: “What percentage of explicit directives would represent good clinical practice?” As the Cook and colleagues point out, lack of an explicit directive regarding CPR represents an implicit decision to resuscitate. An alternative question would be whether more than 5·5% of patients should have received DNR orders. Of 1920
course, to that 5·5% must be added an unknown proportion of patients who were not transferred to ICUs because they or their families had decided with their physicians that aggressive care was inappropriate. The answer for surgical patients would probably be different from that for medical patients. In this study, 237 (21%) of 1130 medical patients received specific directives (DNR for 111 of the 237), compared with only 30 (2·6%) of 1149 non-trauma surgical patients (DNR for 21 of the 30). These findings suggest that there are fundamental differences in the management of critically ill medical patients and that of critically ill surgical patients that may require distinct approaches to defining an appropriate target. But should not an explicit directive be drawn up for all patients, irrespective of whether in this study the total number of explicit and implicit directives to offer resuscitation was defensible? The Study to Understand Prognoses and Preferences for Outcomes and Risks of Treatments (SUPPORT) project reported that only 29% of seriously ill patients and only 48% of patients who did not want CPR had discussed their preferences with their physician.1 Investigators from this project stated that the process for making decisions about DNR orders should be improved to accurately reflect patients’ preferences and probable outcomes,2 yet these same investigators, in another report,3 noted that a majority of patients who had not discussed their preferences with a physician preferred not to discuss them. The process of discussion is necessarily time-consuming and complex. For patients who are not moribund with clearly fatal diseases, discussions of whether to withhold or withdraw treatment require consideration of a nearly limitless variety of contingencies. Should CPR be done if a cardiopulmonary arrest accompanies a massive neurological complication—or if respiratory depression occurs as a consequence of postoperative narcotic analgesia, or if a massive myocardial infarction occurs, or if ventricular fibrillation accompanies less severe myocardial ischaemia? Perhaps many apparent failures to establish explicit directives simply represent the enormous effort required of both physicians and patients to discuss and understand the diverse circumstances under which cardiac arrest might occur. Such complexity would inevitably be compounded by the facts that many patients do not want to discuss end-of-life care with their physicians3 and that health-care providers vary greatly in their opinions on appropriateness of care in specific clinical circumstances.4 Routinely exploring such complex, value-laden, and emotional issues entails the implicit risk of forcing explicit directives that do not actually serve patients’ needs. Another cautionary note is in order. In Cook and colleagues’ study, ICU junior doctors (residents) provided the most specific directives (45·6% of all specific directives); ICU physicians provided far fewer (17%), and most of these were DNR directives. Physicians who had admitted patients to the hospital provided 32·7% of
THE LANCET • Vol 358 • December 8, 2001
For personal use. Only reproduce with permission from The Lancet Publishing Group.
COMMENTARY
decisions. Patients admitted to ICUs at night and on weekends were more likely to receive specific directives. Although the details of individual decisions obviously could not be included in the analysis, one is left with the discomforting thought that directives were most likely to be arrived at by junior physicians who have little previous knowledge of the patient or their social environment and who are under the pressure that accompanies clinical care of a large group of critically ill patients during off-hours. In summary, Cook and colleagues have contributed a valuable, well-conducted study to the existing publications on compliance with patients’ preferences in decisions on whether to offer life-sustaining therapy. The heuristic value of this report will no doubt result in further studies that clarify appropriate strategies for patients, patients’ families, and physicians as they confront serious illness that could result in death. Donald S Prough Department of Anesthesiology, University of Texas Medical Branch, Galveston, TX 77555, USA (e-mail: [email protected]) 1
2
3
4
Phillips RS, Wenger NS, Teno J, et al. Choices of seriously ill patients about cardiopulmonary resuscitation: correlates and outcomes. SUPPORT Investigators. Study to Understand Prognoses and Preferences for Outcomes and Risks of Treatments. Am J Med 1996; 100: 128–37. Hakim RB, Teno JM, Harrell FE Jr, et al. Factors associated with donot-resuscitate orders: patients’ preferences, prognoses, and physicians’ judgments. SUPPORT Investigators. Study to Understand Prognoses and Preferences for Outcomes and Risks of Treatment. Ann Intern Med 1996; 125: 284–93. Hofmann JC, Wenger NS, Davis RB, et al. Patient preferences for communication with physicians about end-of-life decisions. SUPPORT Investigators. Study to Understand Prognoses and Preference for Outcomes and Risks of Treatment. Ann Intern Med 1997; 127: 1–12. Cook DJ, Guyatt GH, Jaeschke R, et al. Determinants in Canadian health care workers of the decision to withdraw life support from the critically ill. Canadian Critical Care Trials Group. JAMA 1995; 273: 703–08.
Present, rather than, advance directives The concept of advance directives has caught the attention of the general public over the past decade, though such directives (the commonest being “don’t put me in a nursing home when I am old”) are difficult to incorporate into everyday practice. Much impetus for advance directives was provided by the Patient Self-Determination Act (PSDA) in the USA. The PSDA arose following a high-profile case over continued feeding of a young woman with persistent vegetative state.1 This and other similar cases sensitised the general public to some of the difficulties associated with care decisions in the face of severe disability and reduced decision-making capacity. Although wishful thinking on the part of some commentators would see the PSDA as an enabling process promoting dialogue at all levels of care,1 in reality the act has been monitored almost exclusively in terms of end-of-life care and advance directives. A recent randomised trial of advance-care planning discussions among surgical patients seen at pre-operative anaesthesia clinic2 is typical of this trend. On the credit side, Denise Grimaldo and colleagues acknowledged that patients ought to discuss with their close families the kind of treatment that they would wish for should they become too ill to speak for themselves, and the trial was intended to examine the effect of a 5–10 min discussion at the clinic aimed at fostering such communication. The discussions did increase dialogue between patients and their proxies. On the debit side, there was an assumption that living wills and the appointment of a durable power of attorney are a
THE LANCET • Vol 358 • December 8, 2001
“good thing”; the number of patients completing advance directives was another of the main outcome measures. Although these considerations have worth in their own right, the disadvantage is that reports of this kind may feed into a modern paranoia about overtreatment at the end of life, analogous to that of mediaeval fears of being buried alive:3 in the words of the philanthropist and promoter of open societies, George Soros, being in “a state of limbo between living and dying”.4 This focus has emphasised protection against the worst of medical paternalism at the expense of a sense of the complexity of medical decision making, trust in an ethical caring relationship with the health profession, and the dangers of rigid definitions of proxy decision-making. It may also blind the general population to ageist currents in society, whereby a greater hazard in later life is likely to be undertreatment rather than overtreatment.5 This trend for undertreatment also extends to severe disability—in one study of those with a diagnosis of persistent vegetative state, 43% were misdiagnosed and many opportunities for communication were missed.6 Concerns have been expressed by a small but growing number of publications about the skewed emphasis and often reductionist nature of the pressure for advance directives. One of the most articulate critics writes of the danger of losing sight of the fact that autonomy does not exist in a vacuum but is developed, enunciated, and ultimately exercised in the embrace of others.7 Acknowledgment of this fact would mean eschewing simplistic solutions to impaired capacity, such as power of attorney for healthcare. The difficulties of surrogate decision-making are manifold—not only can the role of surrogate be intensely stressful but it also is discharged in a background of societal underestimation of personhood in dementia.8 Moreover, surrogates’ predictions resemble more closely the surrogate’s own treatment wishes than they do those of the individual they are acting for.9 Other solutions are possible in the face of impaired decision-making capacity. In Ireland, for example, the legal precedent stipulates a consensual approach with the family, using both substituted judgment and best interest, and demanding a second medical opinion if agreement cannot be found.8 The generally low level of implementation of advance directives is usually seen in published work as an unwelcome outcome, an interpretation shared by Grimaldo and colleagues. Is it not time to look at another explanation? Is it not possible that patients and their physicians may realise that the complexities of the future cannot be easily contained or expressed adequately within the narrow confines of an advance directive, and that future resolution of care issues may be better handled in a relationship of trust and beneficence? In essence, it may be a validation of the “present directive”, a real-time adaptive mode of relating with a physician that recognises the complexities, change, and growth that occurs during the course of a severe illness. Just as the practice of medicine is an iterative process, so too communication with patient and caregivers evolves and grows with time, progression of the disease, and adaptation of all the players to changed circumstances. An emphasis on present directives might also help to redirect the emphasis of research from areas of uncertain benefit to patients and their carers to those that affect present choices in a more relevant and direct way. Are present dangers feared less than horrible imaginings? For example, are patients aware of the full range of potential risks due to surgery, such as acquired cognitive impairment? According to some estimates, cognitive deterioration will affect one in ten elderly people 1921
For personal use. Only reproduce with permission from The Lancet Publishing Group.
COMMENTARY
Are enough explicit resuscitation directives obtained for critically ill patients? See page 1941
Decisions to withhold or withdraw life-sustaining therapy necessitate the balancing of two undesirable possibilities: failure to provide life-support to a patient who might recover to an acceptable level of function, and failure to withhold or withdraw life-supporting therapy to a patient who has very little likelihood of returning to an acceptable level of function. In this issue of The Lancet, Deborah Cook and colleagues present the results of an observational study, conducted in 15 intensive-care units (ICUs) in four countries, of practices related to explicit decisions (as stated in the progress notes or order sheets) to resuscitate or not resuscitate in the event of a cardiopulmonary arrest. Their findings contribute usefully to the current discussion about the role of patients’ autonomy in decisions about lifesustaining procedures. To find out the prevalence of explicit directives to or not to carry out cardiopulmonary resuscitation (CPR), to clarify the factors that predict whether there would be explicit directives, and to define how the explicit orders were obtained, the investigators collected data on 2916 patients undergoing a first admission to an ICU. For only 10·9% of patients were there explicit directives, half of which were do-not-resuscitate (DNR) orders. Predictors of explicit directives (either to resuscitate or not) included severe illness and admission to the ICU at night or on weekends. Predictors of DNR orders were advanced age (odds ratio 8·8 for patients >75 years) and inability of patients to participate in decisions. Perhaps the most striking finding, though, was the wide variability in practice between countries, between cities within one country (Canada), and between hospitals within the same city. The investigators correctly identify the strengths of their approach: a large cohort, well-defined estimates, and systematic, comprehensive data collection. They also acknowledge the limitations of the study. Because all of the Canadian centres (11 of the 15) were university-affiliated teaching hospitals in which ICU physicians coordinated care of the patients, these results cannot be extrapolated widely. However, a reasonable speculation is that a more broadly representative sample would have identified qualitatively similar patterns. The investigators carefully confine their conclusions within the limits of the data. They emphasise the importance of aligning directives with patients’ preferences and with cultural influences. However, if, in highly structured ICU environments, only 10·9% of all patients, only 18% of patients aged over 75, and only 30·2% of the most critically ill patients (Acute Physiology and Chronic Health Evaluation Score >30) received explicit directives, the inevitable question is: “What percentage of explicit directives would represent good clinical practice?” As the Cook and colleagues point out, lack of an explicit directive regarding CPR represents an implicit decision to resuscitate. An alternative question would be whether more than 5·5% of patients should have received DNR orders. Of 1920
course, to that 5·5% must be added an unknown proportion of patients who were not transferred to ICUs because they or their families had decided with their physicians that aggressive care was inappropriate. The answer for surgical patients would probably be different from that for medical patients. In this study, 237 (21%) of 1130 medical patients received specific directives (DNR for 111 of the 237), compared with only 30 (2·6%) of 1149 non-trauma surgical patients (DNR for 21 of the 30). These findings suggest that there are fundamental differences in the management of critically ill medical patients and that of critically ill surgical patients that may require distinct approaches to defining an appropriate target. But should not an explicit directive be drawn up for all patients, irrespective of whether in this study the total number of explicit and implicit directives to offer resuscitation was defensible? The Study to Understand Prognoses and Preferences for Outcomes and Risks of Treatments (SUPPORT) project reported that only 29% of seriously ill patients and only 48% of patients who did not want CPR had discussed their preferences with their physician.1 Investigators from this project stated that the process for making decisions about DNR orders should be improved to accurately reflect patients’ preferences and probable outcomes,2 yet these same investigators, in another report,3 noted that a majority of patients who had not discussed their preferences with a physician preferred not to discuss them. The process of discussion is necessarily time-consuming and complex. For patients who are not moribund with clearly fatal diseases, discussions of whether to withhold or withdraw treatment require consideration of a nearly limitless variety of contingencies. Should CPR be done if a cardiopulmonary arrest accompanies a massive neurological complication—or if respiratory depression occurs as a consequence of postoperative narcotic analgesia, or if a massive myocardial infarction occurs, or if ventricular fibrillation accompanies less severe myocardial ischaemia? Perhaps many apparent failures to establish explicit directives simply represent the enormous effort required of both physicians and patients to discuss and understand the diverse circumstances under which cardiac arrest might occur. Such complexity would inevitably be compounded by the facts that many patients do not want to discuss end-of-life care with their physicians3 and that health-care providers vary greatly in their opinions on appropriateness of care in specific clinical circumstances.4 Routinely exploring such complex, value-laden, and emotional issues entails the implicit risk of forcing explicit directives that do not actually serve patients’ needs. Another cautionary note is in order. In Cook and colleagues’ study, ICU junior doctors (residents) provided the most specific directives (45·6% of all specific directives); ICU physicians provided far fewer (17%), and most of these were DNR directives. Physicians who had admitted patients to the hospital provided 32·7% of
THE LANCET • Vol 358 • December 8, 2001
For personal use. Only reproduce with permission from The Lancet Publishing Group.
COMMENTARY
decisions. Patients admitted to ICUs at night and on weekends were more likely to receive specific directives. Although the details of individual decisions obviously could not be included in the analysis, one is left with the discomforting thought that directives were most likely to be arrived at by junior physicians who have little previous knowledge of the patient or their social environment and who are under the pressure that accompanies clinical care of a large group of critically ill patients during off-hours. In summary, Cook and colleagues have contributed a valuable, well-conducted study to the existing publications on compliance with patients’ preferences in decisions on whether to offer life-sustaining therapy. The heuristic value of this report will no doubt result in further studies that clarify appropriate strategies for patients, patients’ families, and physicians as they confront serious illness that could result in death. Donald S Prough Department of Anesthesiology, University of Texas Medical Branch, Galveston, TX 77555, USA (e-mail: [email protected]) 1
2
3
4
Phillips RS, Wenger NS, Teno J, et al. Choices of seriously ill patients about cardiopulmonary resuscitation: correlates and outcomes. SUPPORT Investigators. Study to Understand Prognoses and Preferences for Outcomes and Risks of Treatments. Am J Med 1996; 100: 128–37. Hakim RB, Teno JM, Harrell FE Jr, et al. Factors associated with donot-resuscitate orders: patients’ preferences, prognoses, and physicians’ judgments. SUPPORT Investigators. Study to Understand Prognoses and Preferences for Outcomes and Risks of Treatment. Ann Intern Med 1996; 125: 284–93. Hofmann JC, Wenger NS, Davis RB, et al. Patient preferences for communication with physicians about end-of-life decisions. SUPPORT Investigators. Study to Understand Prognoses and Preference for Outcomes and Risks of Treatment. Ann Intern Med 1997; 127: 1–12. Cook DJ, Guyatt GH, Jaeschke R, et al. Determinants in Canadian health care workers of the decision to withdraw life support from the critically ill. Canadian Critical Care Trials Group. JAMA 1995; 273: 703–08.
Present, rather than, advance directives The concept of advance directives has caught the attention of the general public over the past decade, though such directives (the commonest being “don’t put me in a nursing home when I am old”) are difficult to incorporate into everyday practice. Much impetus for advance directives was provided by the Patient Self-Determination Act (PSDA) in the USA. The PSDA arose following a high-profile case over continued feeding of a young woman with persistent vegetative state.1 This and other similar cases sensitised the general public to some of the difficulties associated with care decisions in the face of severe disability and reduced decision-making capacity. Although wishful thinking on the part of some commentators would see the PSDA as an enabling process promoting dialogue at all levels of care,1 in reality the act has been monitored almost exclusively in terms of end-of-life care and advance directives. A recent randomised trial of advance-care planning discussions among surgical patients seen at pre-operative anaesthesia clinic2 is typical of this trend. On the credit side, Denise Grimaldo and colleagues acknowledged that patients ought to discuss with their close families the kind of treatment that they would wish for should they become too ill to speak for themselves, and the trial was intended to examine the effect of a 5–10 min discussion at the clinic aimed at fostering such communication. The discussions did increase dialogue between patients and their proxies. On the debit side, there was an assumption that living wills and the appointment of a durable power of attorney are a
THE LANCET • Vol 358 • December 8, 2001
“good thing”; the number of patients completing advance directives was another of the main outcome measures. Although these considerations have worth in their own right, the disadvantage is that reports of this kind may feed into a modern paranoia about overtreatment at the end of life, analogous to that of mediaeval fears of being buried alive:3 in the words of the philanthropist and promoter of open societies, George Soros, being in “a state of limbo between living and dying”.4 This focus has emphasised protection against the worst of medical paternalism at the expense of a sense of the complexity of medical decision making, trust in an ethical caring relationship with the health profession, and the dangers of rigid definitions of proxy decision-making. It may also blind the general population to ageist currents in society, whereby a greater hazard in later life is likely to be undertreatment rather than overtreatment.5 This trend for undertreatment also extends to severe disability—in one study of those with a diagnosis of persistent vegetative state, 43% were misdiagnosed and many opportunities for communication were missed.6 Concerns have been expressed by a small but growing number of publications about the skewed emphasis and often reductionist nature of the pressure for advance directives. One of the most articulate critics writes of the danger of losing sight of the fact that autonomy does not exist in a vacuum but is developed, enunciated, and ultimately exercised in the embrace of others.7 Acknowledgment of this fact would mean eschewing simplistic solutions to impaired capacity, such as power of attorney for healthcare. The difficulties of surrogate decision-making are manifold—not only can the role of surrogate be intensely stressful but it also is discharged in a background of societal underestimation of personhood in dementia.8 Moreover, surrogates’ predictions resemble more closely the surrogate’s own treatment wishes than they do those of the individual they are acting for.9 Other solutions are possible in the face of impaired decision-making capacity. In Ireland, for example, the legal precedent stipulates a consensual approach with the family, using both substituted judgment and best interest, and demanding a second medical opinion if agreement cannot be found.8 The generally low level of implementation of advance directives is usually seen in published work as an unwelcome outcome, an interpretation shared by Grimaldo and colleagues. Is it not time to look at another explanation? Is it not possible that patients and their physicians may realise that the complexities of the future cannot be easily contained or expressed adequately within the narrow confines of an advance directive, and that future resolution of care issues may be better handled in a relationship of trust and beneficence? In essence, it may be a validation of the “present directive”, a real-time adaptive mode of relating with a physician that recognises the complexities, change, and growth that occurs during the course of a severe illness. Just as the practice of medicine is an iterative process, so too communication with patient and caregivers evolves and grows with time, progression of the disease, and adaptation of all the players to changed circumstances. An emphasis on present directives might also help to redirect the emphasis of research from areas of uncertain benefit to patients and their carers to those that affect present choices in a more relevant and direct way. Are present dangers feared less than horrible imaginings? For example, are patients aware of the full range of potential risks due to surgery, such as acquired cognitive impairment? According to some estimates, cognitive deterioration will affect one in ten elderly people 1921
For personal use. Only reproduce with permission from The Lancet Publishing Group.