EDITORIALS
3. Rusnak RA, Stair TO, Hansen K, et aI: Litigation against the emergency physician: Common features in cases of missed myocardial infarction. Ann Emerg Med 1989;18:1029q034. 4. Goldman L, Cook El:, Brand DA, et al: A computer protocol to predict myocardial infarction in emergency department patients with chest pain. N Engl J Med 1988;318:797-803. 5. Zarling EJ, Sexton H, Milnor P: Failure to diagnose acute myocardial infarction in emergency room patients with chest pain. JAMA 1983;250:1177-I181. 6. Kannel WB, Abbott RD: Incidence and prognosis of unrecognized myocardial infarction. An update on the Framingham Study. N EngI ] Med 1984;311:1144-1147.
absence of chest pain in patients with suspected acute myocardial infarction. Am J Emerg Med 1989;7:372-377. 9. McGuiness JB, Begg TB, Semple T: First electrocardiogram in recent myocardial infarction. Br Med J 1976;2:449-451. i0. Pozen MW, D'Agostino RB, Selker HP, et al: A predictive instrument to improve corona13r-care-unit admission practices in acute ischemic heart disease: A prospective multicenter clinical trial. N EngI J Med 1984;310:1273-I278. 11. Pozeu MW, Stechmiller JK, Voight GC: Prognostic efficacy of early clinical categorization of myocardial infarction patients. Circ~ilatJon 1977;56:816-819.
7. Klaeboe G, Otterstad JE, Winsnes T, et ah Predictive value of prodromal symptoms in myocardiK1 infarction. Acta ivied Scand 1987;222:27-30.
12. Lee TH, Juarez G, Cook ElZ, et aI: Ruling out acute myocardial infar-ction: A prospective multicenter validation of a i2-hour strategy for patients at low risk. N Engl J ivied 1990;324:1239-1246.
8. Fesmire FM, Wears RL: The utility of the presence or
13. Lee TH, Goldman L: Serum enzyme assays in the
diagnosis of acute myocardial infarction: Recommendations on acute quantitative analysis. Ann Intern Med 1986;105:221-233. 14. Eisenberg JM, Horowitz LN, Busch RR, et ah Diagnosis of acute myocardial infarction in the emergency room: A prospective assessment of cIinical decision making arid the usefuiness of immediate cardiac enzyme determination. J Commun Health I979;4:190-198. 15. Giblet WB, Lewis LM, Erb RE, et aI: Early detection of acute myocardial infarction in patients presenting with chest pain and nondiagnostic ECGs: Serial CK-MB sampling in the emergency department. Ann Emerg Med 1990;19:1389q366. 16. Green GB, Hansen KN, Fleetwood D, et aI; The usefulness of a rapid creatine phosphokinase (CPK) MB assay in the evaluation of emergency department patients with possible myocardial infarction (abstract). Ann Emerg Med 1990;19:1227.
I n f o r m e d C o n s e n t to R e s e a r c h on Emergency Care In the article, "Informed Consent for Biomedical Research in Acute Care Medicine," in this issue of Annals, Abramson, Iserson, McKay, and Cohen agree on the statement of the central ethical problem in research on emergency medical treatment: the conflict between the autonomy and welfare of subjects and the need to conduct research. This conflict is present in all medical research, but it is especially difficult to resolve for emergency research because subjects are likely to be unconscious or otherwise incapable of exercising autonomy through informed consent and because of the urgency of beginning treatment. The physicians (Abramson and Iserson) and the government official (McKay) present resolutions to the conflict that rely on the provisions for waiver of consent in the Department of Health and Human Services (DHHS) regulations for institutional review board (IRB) review of research. The attorney (Cohen) presents a legal challenge to the regulatory approach of his fellow panelists: courts will be responsive to arguments that the DHHS regulations do not sufficiently protect a subject's statutory, common law, or constitutional rights, and this places investigators at r i s k of liability even though the research was closely and thoroughly scrutinized by an IRB. 138/1267
Iserson and McKay argue that the impossibility of prospective consent requires that the risk:benefit ratio for subjects be strictly reviewed by IRBs and that these bodies adopt procedures to m o n i t o r the research. When subjects can give prospective consent, the IRB members can approve research on the assumption that individuals can make their own assessments of whether the risk:benefit ratio of being a subject compares favorably with that of receiving care outside the research protocol. When this assumption cannot be met, it sounds reasonable that an IRB should make its own more careful assessment. However, neither author provides more than an exhortation for this. They do not state the standards for stricter review and explain how they differ from those for nonstrict review. McKay identifies some questions that an IRB should ask, but they are questions an IRB should ask about any clinical research. This absence of standards is not limited to Iserson and McKay; one of the m o s t v e x i n g l a c u n a e in the DHHS regulations is the lack of any account of how an IRB should judge that "risks to subjects are reasonable in relation to anticipated benefits and the importance of the knowledge." Quantitative standards for weighing risk:benefit ratios are theoretAnnals of Emergency Medicine
ically dubious and practically unworkable. However, it is possible to present a qualitative comparison of the risk/benefit of being a subject in a trial and the risk/benefit of not being in the trial. Abramson's criteria for using deferred c o n s e n t to research, as well as my own criteria for research without prospective consent (discussed by Abramson), suggest a framework for doing this. For an IRB to make this comparison, protocols should contain a section in which the risks and benefits of being a subject in the trial are listed next to the risks and benefits of not being in the trial, and the differences explained. Protocols contain explanations of the possible benefits of the research, and a long list of the risks of the research treatments and related interventions. Seldom is a t t e n t i o n given to the comparative risks and benefits for the patients if they are not enrolled in the protocol. See related article, p 1251. Most IRB members will not have enough expertise in the area of the research to do the comparison for themselves. Even if they could, it is the burden of the investigator to justify the research to the IRB. Of course, the investigator can appear before the IRB and answer questions, 20:11 November 1991
EDITORIALS
but this information is so central to IRB review that it should be in the protocol and not treated as an addition or clarification. Iserson and McKay also recommend that the IRB review be augm e n t e d by surrogate lay panels. There is an ambiguity in the function of such panels that is revealed in the difference b e t w e e n calling t h e m "surrogate panels" and "lay oversight boards." The former suggests that the members would be asked whether they would wish to participate in the research if they were in the circumstances of the potential subjects. A strong affirmative vote on this would be used to support the claim that the actual subjects would consent if they were capable of consent. The latter suggests a m u c h broader charge s o m e t h i n g similar to IRB review from a strictly lay point of view. If the concern is that IRBs have a professional bias and research should be evaluated from the perspective of the lay public who are the subjects and beneficiaries of research, then the composition of IRBs can be altered and the additional layer of bureaucracy avoided. The deeper question about "surrogate panels" is whether their "approval" of a research protocol adequately replaces the absence of prospective consent. Cohen's remarks on h o w courts regard i n d i v i d u a l rights to autonomy shed light on this issue. He says that if true informed consent cannot be obtained, there must be a decision as close as possible to what the individual would have wanted. The fact that a panel of individuals who were not subjects said they would consent if they were does not support the claim that a particular individual would have consented. The right to autonomy is the right to be different, the right to not do what a majority, or nearly everyone else, would do. In case after case, courts have held that a proxy may refuse life-saving treatment for an incompetent patient, provided that the proxy has evidence that the patient would have done so. The evidence can be from a living will or from clearly expressed wishes of the pa20:11 November 1991
tient. It is the wishes of the individual patient that are relevant, not the wishes of persons who might be such a patient. A court might be persuaded to accept a less demanding standard for initiating research than for with&awing life-saving treatment on the grounds that a patient's rights and welfare are less at risk in research on emergency treatment, w h i c h provides life-saving care, than in decisions to withdraw life-saving care. Until the courts reach such a result, researchers will have to proceed under a cloud of legal uncertainty. Abramson and McKay each present an additional argument for the acc e p t a b i l i t y of doing research on emergency treatment without prospective consent from a competent patient. Abramson defends the use of deferred consent and McKay the use of presumed consent. For both authors the standard that must be met is that the treatments that are randomly assigned are of equal efficacy. In an emergency situation, either treatment could be chosen for a patient under the emergency exemption to informed consent, and hence patient well-being and rights are not compromised by random assignment. There is a problem with making the supporting claim as a positive assertion of knowledge. If it means what it seems to say, that the treatments are known to be equally efficacious, then there is no reason to do the research. Abramson and McKay would certainly deny that this is what they mean. McKay hints at, and Abramson is specific about, defending randomization on the basis of a negative claim of uncertainty of clinical knowledge. One of the criterion for deferred consent in the Brain Resuscitation Clinical Trials (BRCT) is that "the investigator does not (and cannot) know which treatment is better." One way to support this is by appeal to the general uncertainty in medicine. Whenever a treatment choice is made, the results are not certain. No treatment is 100% effective. Abramson says that this shows that the provision of care in the patient's best interest and the conduct of scientific research are not as disparate as might Annals of Emergency Medicine
be believed. The problem with this way of putting it is that it justifies too much. If it is the lack of certain knowledge of the outcome of a treatment that justifies it for a random clinical trial, then any pair of treatments could be used in a trial. My a l t e r n a t i v e criteria for the Brain Resuscitation Clinical Trials would be an attempt to be more precise about what must be known to justify randomization. In place of the BRCT criterion in which the investigator does not (and cannot) know which therapy is better, I suggested the following two criteria: the probability of death with experimental or control therapy is not greater than the probability of death with usual therapy, and the probability of normal outcome is greater with experimental or control therapy than with usual therapy. These criteria are limited to the circumstance of the BRCT and may not be applicable to other research on emergency care. The relevant feature of these trials is that they were an attempt to find a treatment for an outcome of global brain ischemia for which there was no treatment. Usual therapy was simply palliative, and the investigators had evidence for believing that the new therapy would be beneficial. Control therapy in these trials differed from usual therapy because the patient was in the trial and would have interventions that would not be done outside the research context; this difference may be negligible depending on the particular features of the trial and the u s u a l care.
The motivation behind my way of stating the criteria for research on emergency care is to recognize the uncertainty of medicine and at the same time make a positive statement about what sort of evidence for beliefs about treatment efficacy is required to justify initiating a clinical trial of emergency care. The focus is on the crucial question for research w i t h o u t prospective consent: the comparison of risks and benefits of being in the trial with the risks and benefits of not being in the trial. A b r a m s o n says that m y points were modified to require that there 1268/139
EDITORIALS
be reason to believe that the experimental therapy is beneficial and little reason to believe that it is harmful. This is an acceptable modification only when applied to a situation in which there is no beneficial treatment. If there is a beneficial treatment, even though it m a y be limited in degree and success rate, it is imperative for the rights and well-being of the incompetent patient that there be some evidence that being in the trial be at least as beneficial as not being in t h e trial and that being in the trial does not expose the patient to significant additional risk. It will not be possible to reliably quantify the differences, but it should be possible to present the evidence in a way
140/1269
that will allow a reasoned judgment. C o h e n m a k e s two p o i n t s a b o u t randomization that buttress my views: first, it is rare that all things are exactly equal; second, the implication for a jury is likely to be that the t r e a t i n g p h y s i c i a n really just wanted to do the experiment, especially if the physician chooses that response for a n u m b e r of patients. The first point pushes toward recognizing the importance of an explicit comparison of the risks and benefits of being in a trial versus not being in a trial. The second works toward recognizing the investigator perspective on treatment efficacy - that it is a m a t t e r of generalized results for a large number of patients as opposed
Annals of Emergency Medicine
to a clinical judgment of what is best for a particular patient. Investigators m a y bemoan the fact that they cannot obtain final answers on w h a t ethics, federal regulation, and legal rights of patients require of them. Law and ethics stand to justice the same w a y m e d i c i n e stands to health: in the easy cases a definitive a n s w e r can be given; in the hard cases we have to learn to live with uncertainty.
Bruce L Miller, PhD Department of Philosophy Michigan State University East Lansing
20:11 November 1991
3. Rusnak RA, Stair TO, Hansen K, et aI: Litigation against the emergency physician: Common features in cases of missed myocardial infarction. Ann Emerg Med 1989;18:1029q034. 4. Goldman L, Cook El:, Brand DA, et al: A computer protocol to predict myocardial infarction in emergency department patients with chest pain. N Engl J Med 1988;318:797-803. 5. Zarling EJ, Sexton H, Milnor P: Failure to diagnose acute myocardial infarction in emergency room patients with chest pain. JAMA 1983;250:1177-I181. 6. Kannel WB, Abbott RD: Incidence and prognosis of unrecognized myocardial infarction. An update on the Framingham Study. N EngI ] Med 1984;311:1144-1147.
absence of chest pain in patients with suspected acute myocardial infarction. Am J Emerg Med 1989;7:372-377. 9. McGuiness JB, Begg TB, Semple T: First electrocardiogram in recent myocardial infarction. Br Med J 1976;2:449-451. i0. Pozen MW, D'Agostino RB, Selker HP, et al: A predictive instrument to improve corona13r-care-unit admission practices in acute ischemic heart disease: A prospective multicenter clinical trial. N EngI J Med 1984;310:1273-I278. 11. Pozeu MW, Stechmiller JK, Voight GC: Prognostic efficacy of early clinical categorization of myocardial infarction patients. Circ~ilatJon 1977;56:816-819.
7. Klaeboe G, Otterstad JE, Winsnes T, et ah Predictive value of prodromal symptoms in myocardiK1 infarction. Acta ivied Scand 1987;222:27-30.
12. Lee TH, Juarez G, Cook ElZ, et aI: Ruling out acute myocardial infar-ction: A prospective multicenter validation of a i2-hour strategy for patients at low risk. N Engl J ivied 1990;324:1239-1246.
8. Fesmire FM, Wears RL: The utility of the presence or
13. Lee TH, Goldman L: Serum enzyme assays in the
diagnosis of acute myocardial infarction: Recommendations on acute quantitative analysis. Ann Intern Med 1986;105:221-233. 14. Eisenberg JM, Horowitz LN, Busch RR, et ah Diagnosis of acute myocardial infarction in the emergency room: A prospective assessment of cIinical decision making arid the usefuiness of immediate cardiac enzyme determination. J Commun Health I979;4:190-198. 15. Giblet WB, Lewis LM, Erb RE, et aI: Early detection of acute myocardial infarction in patients presenting with chest pain and nondiagnostic ECGs: Serial CK-MB sampling in the emergency department. Ann Emerg Med 1990;19:1389q366. 16. Green GB, Hansen KN, Fleetwood D, et aI; The usefulness of a rapid creatine phosphokinase (CPK) MB assay in the evaluation of emergency department patients with possible myocardial infarction (abstract). Ann Emerg Med 1990;19:1227.
I n f o r m e d C o n s e n t to R e s e a r c h on Emergency Care In the article, "Informed Consent for Biomedical Research in Acute Care Medicine," in this issue of Annals, Abramson, Iserson, McKay, and Cohen agree on the statement of the central ethical problem in research on emergency medical treatment: the conflict between the autonomy and welfare of subjects and the need to conduct research. This conflict is present in all medical research, but it is especially difficult to resolve for emergency research because subjects are likely to be unconscious or otherwise incapable of exercising autonomy through informed consent and because of the urgency of beginning treatment. The physicians (Abramson and Iserson) and the government official (McKay) present resolutions to the conflict that rely on the provisions for waiver of consent in the Department of Health and Human Services (DHHS) regulations for institutional review board (IRB) review of research. The attorney (Cohen) presents a legal challenge to the regulatory approach of his fellow panelists: courts will be responsive to arguments that the DHHS regulations do not sufficiently protect a subject's statutory, common law, or constitutional rights, and this places investigators at r i s k of liability even though the research was closely and thoroughly scrutinized by an IRB. 138/1267
Iserson and McKay argue that the impossibility of prospective consent requires that the risk:benefit ratio for subjects be strictly reviewed by IRBs and that these bodies adopt procedures to m o n i t o r the research. When subjects can give prospective consent, the IRB members can approve research on the assumption that individuals can make their own assessments of whether the risk:benefit ratio of being a subject compares favorably with that of receiving care outside the research protocol. When this assumption cannot be met, it sounds reasonable that an IRB should make its own more careful assessment. However, neither author provides more than an exhortation for this. They do not state the standards for stricter review and explain how they differ from those for nonstrict review. McKay identifies some questions that an IRB should ask, but they are questions an IRB should ask about any clinical research. This absence of standards is not limited to Iserson and McKay; one of the m o s t v e x i n g l a c u n a e in the DHHS regulations is the lack of any account of how an IRB should judge that "risks to subjects are reasonable in relation to anticipated benefits and the importance of the knowledge." Quantitative standards for weighing risk:benefit ratios are theoretAnnals of Emergency Medicine
ically dubious and practically unworkable. However, it is possible to present a qualitative comparison of the risk/benefit of being a subject in a trial and the risk/benefit of not being in the trial. Abramson's criteria for using deferred c o n s e n t to research, as well as my own criteria for research without prospective consent (discussed by Abramson), suggest a framework for doing this. For an IRB to make this comparison, protocols should contain a section in which the risks and benefits of being a subject in the trial are listed next to the risks and benefits of not being in the trial, and the differences explained. Protocols contain explanations of the possible benefits of the research, and a long list of the risks of the research treatments and related interventions. Seldom is a t t e n t i o n given to the comparative risks and benefits for the patients if they are not enrolled in the protocol. See related article, p 1251. Most IRB members will not have enough expertise in the area of the research to do the comparison for themselves. Even if they could, it is the burden of the investigator to justify the research to the IRB. Of course, the investigator can appear before the IRB and answer questions, 20:11 November 1991
EDITORIALS
but this information is so central to IRB review that it should be in the protocol and not treated as an addition or clarification. Iserson and McKay also recommend that the IRB review be augm e n t e d by surrogate lay panels. There is an ambiguity in the function of such panels that is revealed in the difference b e t w e e n calling t h e m "surrogate panels" and "lay oversight boards." The former suggests that the members would be asked whether they would wish to participate in the research if they were in the circumstances of the potential subjects. A strong affirmative vote on this would be used to support the claim that the actual subjects would consent if they were capable of consent. The latter suggests a m u c h broader charge s o m e t h i n g similar to IRB review from a strictly lay point of view. If the concern is that IRBs have a professional bias and research should be evaluated from the perspective of the lay public who are the subjects and beneficiaries of research, then the composition of IRBs can be altered and the additional layer of bureaucracy avoided. The deeper question about "surrogate panels" is whether their "approval" of a research protocol adequately replaces the absence of prospective consent. Cohen's remarks on h o w courts regard i n d i v i d u a l rights to autonomy shed light on this issue. He says that if true informed consent cannot be obtained, there must be a decision as close as possible to what the individual would have wanted. The fact that a panel of individuals who were not subjects said they would consent if they were does not support the claim that a particular individual would have consented. The right to autonomy is the right to be different, the right to not do what a majority, or nearly everyone else, would do. In case after case, courts have held that a proxy may refuse life-saving treatment for an incompetent patient, provided that the proxy has evidence that the patient would have done so. The evidence can be from a living will or from clearly expressed wishes of the pa20:11 November 1991
tient. It is the wishes of the individual patient that are relevant, not the wishes of persons who might be such a patient. A court might be persuaded to accept a less demanding standard for initiating research than for with&awing life-saving treatment on the grounds that a patient's rights and welfare are less at risk in research on emergency treatment, w h i c h provides life-saving care, than in decisions to withdraw life-saving care. Until the courts reach such a result, researchers will have to proceed under a cloud of legal uncertainty. Abramson and McKay each present an additional argument for the acc e p t a b i l i t y of doing research on emergency treatment without prospective consent from a competent patient. Abramson defends the use of deferred consent and McKay the use of presumed consent. For both authors the standard that must be met is that the treatments that are randomly assigned are of equal efficacy. In an emergency situation, either treatment could be chosen for a patient under the emergency exemption to informed consent, and hence patient well-being and rights are not compromised by random assignment. There is a problem with making the supporting claim as a positive assertion of knowledge. If it means what it seems to say, that the treatments are known to be equally efficacious, then there is no reason to do the research. Abramson and McKay would certainly deny that this is what they mean. McKay hints at, and Abramson is specific about, defending randomization on the basis of a negative claim of uncertainty of clinical knowledge. One of the criterion for deferred consent in the Brain Resuscitation Clinical Trials (BRCT) is that "the investigator does not (and cannot) know which treatment is better." One way to support this is by appeal to the general uncertainty in medicine. Whenever a treatment choice is made, the results are not certain. No treatment is 100% effective. Abramson says that this shows that the provision of care in the patient's best interest and the conduct of scientific research are not as disparate as might Annals of Emergency Medicine
be believed. The problem with this way of putting it is that it justifies too much. If it is the lack of certain knowledge of the outcome of a treatment that justifies it for a random clinical trial, then any pair of treatments could be used in a trial. My a l t e r n a t i v e criteria for the Brain Resuscitation Clinical Trials would be an attempt to be more precise about what must be known to justify randomization. In place of the BRCT criterion in which the investigator does not (and cannot) know which therapy is better, I suggested the following two criteria: the probability of death with experimental or control therapy is not greater than the probability of death with usual therapy, and the probability of normal outcome is greater with experimental or control therapy than with usual therapy. These criteria are limited to the circumstance of the BRCT and may not be applicable to other research on emergency care. The relevant feature of these trials is that they were an attempt to find a treatment for an outcome of global brain ischemia for which there was no treatment. Usual therapy was simply palliative, and the investigators had evidence for believing that the new therapy would be beneficial. Control therapy in these trials differed from usual therapy because the patient was in the trial and would have interventions that would not be done outside the research context; this difference may be negligible depending on the particular features of the trial and the u s u a l care.
The motivation behind my way of stating the criteria for research on emergency care is to recognize the uncertainty of medicine and at the same time make a positive statement about what sort of evidence for beliefs about treatment efficacy is required to justify initiating a clinical trial of emergency care. The focus is on the crucial question for research w i t h o u t prospective consent: the comparison of risks and benefits of being in the trial with the risks and benefits of not being in the trial. A b r a m s o n says that m y points were modified to require that there 1268/139
EDITORIALS
be reason to believe that the experimental therapy is beneficial and little reason to believe that it is harmful. This is an acceptable modification only when applied to a situation in which there is no beneficial treatment. If there is a beneficial treatment, even though it m a y be limited in degree and success rate, it is imperative for the rights and well-being of the incompetent patient that there be some evidence that being in the trial be at least as beneficial as not being in t h e trial and that being in the trial does not expose the patient to significant additional risk. It will not be possible to reliably quantify the differences, but it should be possible to present the evidence in a way
140/1269
that will allow a reasoned judgment. C o h e n m a k e s two p o i n t s a b o u t randomization that buttress my views: first, it is rare that all things are exactly equal; second, the implication for a jury is likely to be that the t r e a t i n g p h y s i c i a n really just wanted to do the experiment, especially if the physician chooses that response for a n u m b e r of patients. The first point pushes toward recognizing the importance of an explicit comparison of the risks and benefits of being in a trial versus not being in a trial. The second works toward recognizing the investigator perspective on treatment efficacy - that it is a m a t t e r of generalized results for a large number of patients as opposed
Annals of Emergency Medicine
to a clinical judgment of what is best for a particular patient. Investigators m a y bemoan the fact that they cannot obtain final answers on w h a t ethics, federal regulation, and legal rights of patients require of them. Law and ethics stand to justice the same w a y m e d i c i n e stands to health: in the easy cases a definitive a n s w e r can be given; in the hard cases we have to learn to live with uncertainty.
Bruce L Miller, PhD Department of Philosophy Michigan State University East Lansing
20:11 November 1991
