Paper Five: Competency to Decide on Treatment and Research-Commissioned Papers
Author(s):
Elyn R. Saks
Introduction
The issue of competency to decide on treatment and research has come center stage in the arena of bioethics.With the advent of the doctrine of informed consent came attention to both the quality of the doctors informing and the quality of the patients understanding. Since World War II, moreover, concern about the ethics of doing medical research on human subjects, particularly vulnerable subjects, has led commentators to focus on competency to consent. Finally, since the 1970s, many courts, first on federal and now on state law grounds, have granted competent psychiatric patients the right to refuse treatment absent an emergency. Indeed, probably the vast majority of treatment competency hearings today arise out of efforts to medicate psychiatric patients.1 Courts and commentators have struggled with the test for competency in this setting. Unfortunately, courts typically cite language that often obscures more than it helps, seldom engaging in illuminating analysis. Commentators have done better. The premier work on competency to make treatment and research decisions has come out of the MacArthur network on law and mental health.2 The MacArthur researchers, in particular, Paul Appelbaum and Thomas Grisso, have operationalized three research instruments to measure four capacities involved in treatment competency and have studied the instruments in their application to patients and control groups. They have also produced a treatment capacity measure that can be administered in the clinic.3 Their work has been impressive indeed; they have given us a set of generally well-designed instruments that have achieved high reliability, can be administered with relative ease, and have been studied in interesting and informative ways.
The three research instruments are:
- the Understanding Treatment Disclosures (UTD) instrument, which measures understanding 4
- the Perceptions of Disorder (POD) instrument, which measures ones appreciation of disclosures about illness and treatment as they apply to ones own situation5
- the Thinking Rationally About Treatment (TRAT) instrument, which measures ones reasoning skills as one decides about a hypothetical treatment dilemma based on ones own condition.6
A subset of the latter test is a question which measures ones ability to express a choice.
Appelbaum and Grisso argue that the three instruments measure capacities relevant to different standards of legal competency found in the case law and statutes.7 They distinguish between capacity and competency - capacity refers to abilities relevant to performing a task, while competency is a legal judgment that one has sufficient abilities to perform the task - and say that their instruments measure only capacities.8 The term impaired means that the subject scored two standard deviations below the mean of those studied.9
The researchers have recently designed a treatment capacity instrument, to be used for actual evaluations rather than research purposes (the MacArthur Competence Assessment
Tool - Treatment, or MacCAT - T).10 The MacCAT - T incorporates many of the questions found in the research instruments, while being more economical to administer and tailoring the questions to the persons particular situation. The investigators do not suggest that MacCAT-T scores translate directly into competency or incompetency findings (although those who score average or better on all the tests are said to be likely to be competent).11 Clinical judgment is required to make these findings.
This paper seeks to evaluate the MacArthur instruments with a view to understanding their implications for assessing the competency to participate in research of vulnerable psychiatric
patients. This assessment takes place against the background of some of the important normative issues that arise whenever one selects a competency instrument. The paper begins by describing the three MacArthur instruments in greater detail. It then sketches a framework for thinking about the normative questions raised by any competency instrument, discussing the implicit judgments the MacArthur investigators smuggle in, despite protestations that they have not addressed these issues and absent an effort to justify the choices. The paper then evaluates the MacArthur instruments in light of the normative framework, focusing on both the classification schema itself and specific features of each instrument. The evaluation leads into a discussion of the issues involved in importing these treatment instruments into the research context - as well as additional normative issues raised in this context. The paper concludes by addressing thedilemma of what IRBs should require of their investigators in this regard: first, should researchers be required to evaluate competency when recruiting subjects for research? And second, should they be required to use particular instruments to do so? A call for further research will end this section.
My conclusions? First, adopting any standard or instrument to assess capacity/competency will involve normative considerations. The MacArthur investigators have arguably omitted an essential step in developing their instruments. Second, the implicit judgments that the MacArthur investigators do make may strike the wrong balance between autonomy and paternalism. Both of these conclusions are equally applicable to the treatment and research contexts, and are important for the National Bioethics Advisory Commission to consider in assessing what to require of IRBs in regard to competency evaluations in the research context.
Back to Top
The Nature of the MacArthur Treatment Instruments
Before evaluating the MacArthur instruments, let us consider in more detail what they do, and what studies of their application have revealed about different groups of people. The UTD measures the subjects understanding of treatment disclosures about the illness he suffers from and its treatment.12 Form disclosures were devised for schizophrenia, depression, and ischemic heart disease (angina). Each disclosure, using language understandable at the junior high level, consists of five simple paragraphs briefly describing the illness and its treatment.
The first paragraph speaks of the illness itself, as well as of two common symptoms of the illness (schizophrenia is a mental disorder. People with schizophrenia often have unpleasant experiences, called symptoms. For example, they . . . may hear voices talking about what they are doing, even when there are no other people around).13 The second paragraph talks about treatment, how it is administered, and what is required of the patient for it to be effective (fortunately, schizophrenia can be treated with medicine. . . . But if patients stop taking this medicine, their symptoms may come back14). The third speaks of the potential benefits of the treatment (the medicines used to treat schizophrenia help many patients to think more clearly. They often stop the frightening voices that some patients with schizophrenia hear15). The fourth paragraph speaks of the potential side effects of the treatment (the medicine might make patients restless or cause their muscles to tighten up16). The fifth paragraph speaks of alternatives, benefits of the alternatives, and potential problems with the alternatives (there is also psychotherapy [to help treat schizophrenia]. . . . This talking therapy may help patients better understand themselves and their feelings. But psychotherapy alone does not usually help
with schizophrenia by itself. . . . [it] is most helpful when the patient is also taking medicine17).
The UTD is administered in three forms. First, the patient is read the entire disclosure and asked to paraphrase what has been said (with questions prompting him if need be). Second, the patient is read each element of the disclosure in turn, and after each element is read, is asked to paraphrase what has been said. Third, the same element disclosure format is followed, except the patient is asked, after each element, whether a statement read is the same as or different from what has been said.
Patients receive points depending on how much they have remembered and (presumably) understood. For example, if two symptoms of schizophrenia have been disclosed, a patient
will receive a full score on that issue if he repeats or paraphrases those two symptoms. He will also receive a maximum score if he includes those two but adds others that were not disclosed
to him. He will receive no credit if he remembers none of the symptoms or, interestingly, if he brings up other symptoms - even if they are bona fide symptoms of schizophrenia - that he did not hear in the disclosure (and he does not additionally mention disclosed items).
The POD measures peoples appreciation of their illness and its treatment: it requires them to apply general information to their own situation.18 There are two subtests, the Non - Acknowledgment of Disorder (NOD) subtest and the Non - Acknowledgment of Treatment Potential (NOT) subtest. The first measures the patients failure to acknowledge his diagnosis, the severity of his condition, or the symptoms he has been demonstrating. Objective measures of these three are provided by the diagnosis given in the patients medical chart, the severity of his symptoms as measured by the Brief Psychiatric Rating Scale, and the symptoms recently reported in his medical chart.
The NOT measures patients failure to acknowledge the potential value of treatment for their illnesses even when successful treatment is likely. It focuses on the extent to which patients believe
- any treatment might be of benefit to them,
- medication specifically might benefit them,
- the course of improvement is likely to be lessened absent treatment.
If they fail to acknowledge the potential benefits of treatment, they are provided a hypothetical premise that logically nullifies their reasoning (e.g. imagine that a doctor tells you that there is a medication that has been shown in research to help 90 percent of people with your problem, even people who had not gotten better with any other medication19). Nonacknowledgment is scored only if the patient fails to acknowledge the potential benefits of treatment under the hypothetical condition. The NOT does not assess whether patients would agree to the medication - just whether they believe it might be of possible benefit.
There are three additional elements of the POD that contribute to no subscale; they have been included for exploratory reasons only. These items assess patients acknowledgment of
potential side effects of medication generally, their perceptions of the beneficence of the hospital staff, and their perceptions of their own need for hospitalization.
The third instrument, the TRAT, measures patients ability to think rationally about treatment.20 The instrument gives a vignette including information about a disorder, various treatment alternatives, and their probable risks and benefits. It then asks the subject to recommend one of the treatments to a friend with the relevant illness, and to describe the reasons for the selection. The patients reasoning is then scored for various cognitive activities that are considered important to making a decision. A second set of procedures examines more formal cognitive functions with relevance for decision making. The cognitive functions identified are seeking information (tendency to seek information beyond what is provided); consequential thinking (consideration of consequences of treatment alternative when deciding whether to accept or reject it); comparative thinking (simultaneous processing of information about two treatment alternatives, such that they are considered in relation to each other); complex thinking (attention to the full range of treatment alternatives); and generating consequences (generation of potential real-life consequences of the liabilities described in the informed consent disclosure, such as effects of a side effect of medication on ability to perform job functions).
In the second part of the TRAT, three additional cognitive functions are measured independent of the vignette: weighting consequences (tendency for consistent application of preferences);
transitive thinking (assessment of relative quantitative relationships between several alternatives based on paired comparisons); and probabilistic thinking (ability to distinguish correctly the relative values of percentage probabilities).
All of the abilities measured on the two parts of the TRAT were derived from discussions in the literature of essential reasoning abilities. Presumably they are measured in a different way
given the constraints of the vignette procedure.
The way the vignette abilities are scored is by presenting the vignette to the patient, asking her if she needs further information, and asking her to choose one of the alternatives and to
give her reasons for doing so. She is then asked for further reasons, as well as for her least preferred choice and her reasons for that. Scoring occurs by seeing how many of the kinds of cognitive operations identified earlier occur. For instance, did the patient compare risks and benefits of the alternatives with each other?
The way the three further abilities are scored is by presenting the patient with a series of questions that tap into those abilities. For instance, to test probabilistic thinking the patient is
told that some event has a 90 percent probability of occurring and is then asked if he thinks it likely to occur.
Finally, the TRAT has a question that measures the patients ability to express a choice. A full score is received if the patient unambiguously chooses an option, and partial credit is received if the patient chooses two or no alternatives, but chooses one alternative during a repeat inquiry.
Early indications are that two of the cognitive operations - weighting consequences and seeking information - are frequent outliers, and when factor analyses are performed with these subscales removed, they produce two very consistent factors: consequential, comparative, and complex thinking on the one hand, transitive and probabilistic thinking together with generating consequences on the other.
Formulating these instruments to measure capacities relevant to competency has been an impressive achievement. The instrument designers have also done a large-scale, multicenter
study evaluating the reliability and validity of the instruments.21 The instruments have done well on such indicators. In addition, the study has enabled the researchers to draw certain important conclusions about different populations of patients. The research design was to study populations of patients diagnosed with schizophrenia, depression, and ischemic heart disease, and to match the patients on various measures (e.g., age, race, gender, socioeconomic status) to a sample of community nonpatients.
The results are intriguing. The most important is that a significant proportion of patients and nonpatients, in all categories, scored in the nonimpaired range, although the schizophrenic patients did the least well. Impaired, as noted, was defined as two standard deviations below the mean for the aggregate of everyone studied, patients and nonpatients alike. Given this definition, approximately 25 percent of the schizophrenic patients scored in the impaired range on each of the three principal instruments, and approximately 50 percent scored in that range if the scores on the different instruments were aggregated. But this means, of course, that approximately 50 percent scored reasonably well.
The researchers noted that their study probably understated the impairment in the group of patients studied, because probably the most disturbed patients were not deemed suitable for participation by their treaters. One could, however, look at the matter another way. The study looked only at recently hospitalized patients who were likely to be in the throes of the most acute phase of their illness. One would want a study of these patients much later in their hospital stay as well - the study did look at them a little later - in order to judge the capacities of hospitalized schizophrenics. And a study evaluating schizophrenics in different settings, such as day hospitals, community mental health centers, and group homes, would probably have found a much higher percentage of patients scoring in the nonimpaired range. These patients, of course, also have to make treatment decisions. Moreover, insofar as schizophrenia is a chronic illness, studying schizophrenics decisionmaking abilities should include them; they may be displaying some symptoms even when not hospitalized. In short, many if not most patients - even
those with the most severe psychiatric disorder - are quite capable of making their own decisions. Beliefs otherwise may reflect bald prejudice.
The second important finding of the study was that the three different instruments seemed to be picking out different patients. While the UTD and the TRAT somewhat tracked each other, the POD was not correlated with either of the former two at all.
The final instrument that these researchers have designed is the MacCAT-T.22 The MacCAT-T incorporates many of the questions found in the research instruments, while being more economical to administer and tailoring the questions to the patients particular situation. On the normative matters, the researchers have pointed out that the research instruments pick out different groups of people as impaired, and therefore ought all to be used in the MacCAT-T. This is a policy choice which we will consider below.
Quantitatively, the researchers consider two alternatives: using a fixed level of performance as a basis for a finding of competency - say understanding, appreciating, and reasoning
about 75 percent of the information provided; or varying the level based on the net balance of expected benefits and risks of the patients choice compared to the alternatives (i.e., requiring
more capacity for more problematic decisions). The researchers seem to prefer the latter, although they do not clearly take a position in this work. I consider the issue of varying the level of competency required depending on the decision below.
Finally, qualitatively, the researchers note that some items of information to be understood, appreciated, and reasoned about may be much more important than others. Decisions about the weight to be given to each item are of great importance in formulating a competency standard.
The MacCAT, then, is an instrument, unlike the three research instruments, which can be used to aid clinical evaluators in a determination of legal competency. It does not definitively answer what level of impairment should equal incompetence. And its authors recommend that the instrument be used in conjunction with a clinical evaluation that takes into account such things as contextual variables. Nevertheless, it, like the research instruments, is an important contribution to the literature on competency.
Back to Top
Normative Judgments Required When Designing Competency and Capacity Instruments
While the MacArthur investigators claim that they have left open the normative questions in designing their instruments23 - a claim I question below - the adoption of any standard or instrument for capacity and/or competency requires careful normative analysis.24 The most critical normative issue facing any designer of a capacity and/or competency instrument is how to strike the balance between autonomy and paternalism. It may be true that bioethicists have moved beyond this simple dichotomy in many areas. But in the arena of assessing competency, this conflict remains the most critical. Competency standards are the mechanism by which we draw the line between those who will be permitted to exercise their autonomy and those who will be treated paternalistically.
Of course, it is at best incomplete to say that we strike the balance in our society between autonomy and paternalism by holding that competent patients, and competent patients alone, are given the right to exercise their autonomy. For this simply pushes back the question: how we define competency is itself the pivotal determiner of how that balance is struck - requiring much in the way of competency favors paternalistic interests and vice versa.
Perhaps more important, the definitional question replicates within itself autonomy/paternalism concerns. The MacArthur researchers recognize that the issue of when to decide for a patient and when to allow him to decide himself raises an important autonomy dilemma. We are concerned about allowing people choice - say of treatment or participation in research - when it is appropriate to do so. But the researchers do not seem to recognize that the very definition of competency requires us to decide how much latitude we will give the decision maker to exercise in selecting a method of decision making. Are intuitive methods adequate? Must decision makers compare all alternatives? A most important subset of this issue is how much scope we will give decision makers to select the version of the truth they will embrace.
What does this all mean? In adopting a competency standard, we must be mindful of several important problems. First, we want to protect the vulnerable who are unable to decide for themselves (our paternalism interest); this translates into a careful inquiry into what abilities are essential for making decisions. Second, an important purpose of our doctrines regarding competency - which must therefore be reflected in the standard we adopt - is to protect the unconventional (our autonomy interest). Third, we must be mindful of the discovery of psychoanalysts, psychiatrists, and psychologists that irrationality in decision making is really quite pervasive: people misunderstand statistics, overvalue vivid memories, form somewhat distorted beliefs about their doctors as a result of transference, and so forth.25 This requires us to question whether we want to find many
more people incompetent than our current practices do - to encompass all of these people who are mildly irrational. If we do not make this normative choice, it requires us to be very careful not to find incompetent those who are mentally ill26 but really suffering from no more irrationality in the relevant regard than many people. To do otherwise is to mistake the floridness of many of the patients symptoms for sufficient decisional impairment and therefore to stigmatize the mentally ill unnecessarily (our nondiscrimination
interest).
These, it seems to me, are some of the normative parameters in terms of which we must measure a competency standard and its implementing instrument. Concretely, what this means
is that we must justify which abilities we require for competency, as well as the level of these abilities which we require. Are these abilities, with this level of performance, really necessary,
and if so why? Or are they nice, but inessential, much as speaking a foreign language with a good accent is inessential to basic communication? Even if we think them arguably important, do we trench too much on patient autonomy by requiring them? And if their absence or impairment is widespread, do we risk discrimination by applying them only to the mentally ill? In short, deciding on the abilities and levels are thoroughly normative endeavors.27 The choices will be manifest in the kinds of abilities chosen; the skill level in the tests of those abilities chosen (e.g., the reading level in the UTD); and the level of performance required in order to be deemed competent.28
Can the MacArthur researchers avoid criticism of their instruments by claiming that they have not made normative choices - an argument they have made in the Law, Psychology, and Public Policy issue devoted to their research?29 In my view, they cannot. It is true that they avowedly leave some choices open: how deficient must one be to be incompetent (vs. impaired)? But they have also made certain other choices - and these choices must be justified. For instance, they do set a level at which impairment is found. In addition, in their discussion of the MacCAT-T (admittedly at a time after they invoked the shield of supposed normative-lessness), they suggest that a particular score should, whatever else we decide, definitely lead to a competency finding. Further, they suggest that we should probably adopt a variable competency standard depending on a cost-benefit judgment about the patients choice.30 These are not pure, value-free statistical questions, but rather normative choices.
Perhaps most important, the MacArthur researchers point out that the three main research instruments seem to be picking out different populations of patients, so that a treatment capacity instrument (their MacCAT-T) should aggregate the three measures. This judgment presupposes that all the skills measured by the three instruments are important to competency - something which is not at all obvious and needs to be justified.31 Indeed, even by proposing the research instruments the MacArthur researchers make normative choices.32 True, the instruments measure only capacities, and not competency. But why would one measure these particular capacities unless one thought them relevant (critical?) to competency? One could also measure foreign language skills, but no one would bother to measure these skills in a capacity instrument administered in ones own language that was designed to be significant for measuring treatment competency. Simply by selecting the abilities measured, and measuring them down from and up to certain levels, the MacArthur researchers are making normative choices. Once again, these choices must be justified. (In addition, as a practical matter many future competency
administerers may mistake the nature of certain of the instruments - thinking that impairment simply translates into incompetency or imagining that the standard given for
clearly competent on the MacCAT-T should divide the competent from the incompetent. It would be interesting to try to judge how often these instruments are being used in current practice, and whether such mistakes are being made.) The absence of normative justification is the most important flaw in the MacArthur instruments. (It is difficult to imagine leaving a harder task to the evaluator on the streets, so to speak.) Below, I evaluate the instruments by suggesting that they may strike the balance in the wrong place without myself undertaking the sustained normative evaluation that would fully establish the point. I intend to do so in future work. Until that time, one might see my critique and proposals as the kind of critique and proposals one would make if one very much valued autonomy. Still, undertaking the appropriate normative justification is the biggest challenge facing future scholars.
Back to Top
Evaluating the MacArthur Instruments
Problems with the MacArthur Scheme
Although it is not clear until one reads through the questions in the instruments themselves, the UTD is what I have elsewhere called a pure understanding standard, while the POD is what
I have elsewhere called a naive understanding and belief standard.33 By a pure understanding standard, I mean that what is assessed is hether the patient comprehends what the caregiver tells him without necessarily believing it,34 while by an understanding and belief standard I mean that what is assessed is the adequacy of the beliefs the patient forms (such as standard is naive to the extent it rests on the notion that truth is easy to discern). The UTD looks, in particular, at pure understanding of general information about an illness and its treatment, while the POD looks at naive understanding and belief concerning the illness and the treatment that the patient applies to himself.
Why do I argue that the UTD measures pure understanding? First, the questions in the UTD ask the patient not to tell the evaluator what is the case, what she believes, or even what the evaluator believes; they ask, instead, only what the evaluator says. This suggests that the UTD is not measuring beliefs, even only fairly obvious beliefs such as what someone else purports to believe. Rather, it is measuring simply the ability to say back what has been said. Second, the questions in the third part of the UTD ask whether a statement is the same as or different from what has just been said. Once again, not belief but comprehension is being measured. Indeed, if the patient says that a proposed statement is true, rather than the same, he is corrected,because that is not what is being tested.
Third, the researchers distinguish between the UTD and the POD on the basis that the former identifies what the patient knows while the latter identifies what he believes. In at least one common philosophical tradition, knowing implies believing, and so is not something less than believing as suggested by the researchers here. Still, what the researchers seem to want to get at is clear - the difference between grasping some information and truly believing it. This is precisely the difference between abilities measured by the pure understanding and those measured by the understanding and belief standards.
If I am correct in my assessment of what these MacArthur instruments are measuring, the next question becomes whether the precise distinction they draw makes sense, and the answer
is that the distinction is problematic. The problem with the MacArthur schema is that it incorporates two distinctions - pure understanding vs. understanding and belief on the one hand, and general information vs. specific information on the other - in a way that does not cover the entire field. While it does cover pure understanding of general information and understanding and belief of specific information, it omits pure understanding of specific information and understanding and belief of general information. Put differently, only two of the four cells generated by these two
distinctions is covered.
The problem is that pure understanding is relevant as to both general information and information as applied to the patients situation; and adequate beliefs are important both as to the general information and the information as applied to the patients particular situation. If this is so, the critical division should be between a standard requiring only comprehension and one requiring belief, and not between a standard involving understanding of general information and belief about information as applied to ones own case.
In my view, the best solution, then, is to preserve the distinction between pure understanding and understanding and belief standards, but discard the distinction between general information and information as applied to oneself; the latter distinction is a red herring. Otherwise, the instruments do not cover the entire field.
Evaluating Whether the General Capacities Picked Out by the MacArthur Instruments are Necessary to Competent Decision making
In my view, the capacities the MacArthur instruments treat as essential are necessary, but perhaps not sufficient, to adequate decision making; I discuss below whether the level of abilities
the MacArthur instruments pick out is plausible.35 I understand the essential abilities to be: pure comprehension of relevant information; the ability to assess evidence and form appropriate beliefs about that information; the ability to reason with that information; and the ability to evidence a choice. All of these abilities can be normatively justified as necessary for competent decision making.
Why is pure understanding required? Comprehension of relevant information is a sine qua non for a persons assessing that information as it bears on how the patient should decide. To see that a competency standard should require pure understanding, consider the following thought experiment. John is a captive faced with two contraptions between which he understands that he must decide, on pain of death. One of the contraptions will torture him and the other will grant his every wish. John cannot tell from looking at the contraptions what they will do, and he cannot understand his captors explanation of them because it is in a foreign language.
It seems plausible to say that John is incompetent to decide between the two contraptions - with one reservation. We may want to reserve the term incompetent for people who are
not simply ignorant. Although well-known philosophers have justified paternalism in the face of ignorance (recall, e.g., John Stuart Mills broken bridge example), the law may prefer to reserve the term incompetent for those who lack abilities, perhaps as a function of their mental illness, and not simply knowledge. There may be practical reasons for doing this - e.g., we often do not know the truth, so we want assurances that the person is under some disability and so likelier to be self-deceived. These practical concerns may not speak to what we would do if we could always know the truth; perhaps in an ideal universe, all ignorance would amount to incompetency. Whatever we decide in the real world, surely most people would want, in our example above, to be disabled from deciding for themselves, and to have benign and knowledgeable others decide for them.36
Indeed, we do not need fanciful thought experiments to recognize the importance of pure understanding: imagine being asked to make any important decision the implications
of which are described in a foreign language. One is simply not in a position to decide in that case. Pure understanding, then, is a prerequisite for competency.
Yet Pure Understanding, while necessary, is not sufficient. The ability to assess evidence and form appropriate beliefs is also necessary; MacArthurs inclusion of this ability in one of its capacity instruments makes eminent sense. Consider our example again. If the captive comprehends his captors information, but does not credit it at all - say because he believes the captor is delusional - he is not even going to advert to the information in making his decision. Because making a decision in ones best interests requires assessing how those interests are likely to be affected, the patient must be able to form adequate beliefs in order to be a competent decision maker.
We can think of the matter yet another way. Decisions are based on desires37 and beliefs; one desires x, and believes that y is the way to get x, and thus one decides to y. Believing y is the way to get x, in turn, requires other subsidiary beliefs. A deficiency in ones beliefs may therefore severely affect ones decisionmaking capacity. (How deficient is too deficient is, of course, another question - one which I shall address below.) One forms beliefs as a result of assessments of the evidence, so that the skill
tapped here is the ability to assess evidence; and this skill is clearly needed in some degree or another for competency.
But pure understanding and the ability to assess evidence are also arguably not enough; one needs also to be able to reason with some degree of intactness. Reasoning allows one to put together the information one has purely understood and, having assessed, has formed beliefs about. Consider, at its simplest, the practical syllogism recited above. If one knows that one desires x and one believes that y is the way to get x (and say that not doing y will guarantee not getting x), and if one then concludes to not y on the basis of deficient reasoning,38 one has clearly not made a competent choice to not y. Some level of reasoning ability is required. Thus the MacArthur instruments inclusion of a measure of this makes sense.
Should evidencing a choice also be considered a necessary skill for making a competent choice? There are two possibilities here. If making a competent choice means (in part) expressing
a choice, then the answer is obviously yes; so much is a tautology. If, on the other hand, making a competent choice can include going through intact decisionmaking processes without
necessarily telling anyone what one has chosen, then the answer is obviously no. Take a person who is paralyzed and unable to communicate. He may very well decide, after careful
consideration, that he would like some procedure done. Suppose that by any (other) measure we could formulate he would be deemed competent. Does his inability to say what he wants
make him incompetent? If not, making a choice and showing ones choice are two different things.
Two views are possible, then - evidencing a choice is prima facie evidence of ones competency that triggers an inquiry or is actually itself necessary for ones competency - and it is not
important which view we choose, since in either case the ability must be assessed. Thus, the subtest in the MacArthur instruments measuring evidencing a choice seems justified.
While all the abilities the MacArthur instruments target are arguably necessary for competency, are there others they have omitted? For instance, practical syllogisms also refer to the persons desires; so perhaps one must be able to identify ones desires. Indeed, perhaps one must make a choice that is true to ones desires. More robustly (and relatedly), perhaps one
must be oneself so that one can be true to oneself and ones values. Other noncognitive abilities should also arguably be required - for instance, that one not be under the sway of internal compulsion.39
The MacArthur instruments omit any mention of these abilities and, to the extent we think them important, the instruments are lacking. I myself have doubts at least about our ability to identify such deficits reliably - but it is clearly an open question whether we should want to try to test for them.
If the MacArthur instruments plausibly pick out abilities that any competent decision maker must have, at least formulated at a high level of abstraction, do they set the level of abilities required paying sufficient regard for the various other values involved in designing competency instruments? We now turn to that question, as well as to other questions about the specific instruments.
Assessing the Individual Instruments and the Balance Struck Among the Values Implicated by Competency Instruments
Assessment of the UTD
The UTD is a well-crafted instrument, with two qualifications. First, it seems that the UTD may handle, in the wrong way, information relevant to the patients decision which he acquired apart from the disclosure. The patient receives no points if she mentions even real symptoms that were not part of the current disclosure. Second, the UTD may both overmeasure and undermeasure pure understanding.
The UTDs treatment of extra-disclosure information makes sense up to a point. We arguably want the patient to listen to and understand what he has been told. If he cannot do this, he
cannot assimilate (and eventually assess) information which we believe to be relevant to his decision.
On the other hand, proper respect for unconventionality might counsel us to allow the patient completely to diverge from what he has just been told, provided he recites true information. Patients may get just as good information - or better information for their situation - from other sources. They should arguably be entitled to choose what information is important to them, and they clearly purely understand that information.
Indeed, the patient may have better information relevant to his decision than that which the researcher has given him.What may be most salient about schizophrenia to him, for instance,
may not be the voices mentioned in the disclosure, but the disorganization of his thinking process. So that is what he recites as a symptom of schizophrenia.
On this view, then, patients do seem to have the requisite ability stated at a high enough level of generality. Indeed, requiring them to understand the particular information disclosed may not be necessary for them to be able to make decisions; it may not disserve our interest in protecting the vulnerable to fail to so require. It takes choice away from patients to require them to purely understand just certain information, thus disserving our autonomy interest.
In fact, it may be the case that the UTD requires a specificity of information that is not necessary if a patient is to be capable of decision making.40 It is arguable that all a patient needs to understand to make a treatment decision is that, as a result of mental illnesses, people feel distress and suffer certain disabilities; that there are treatments such as medication for the illness; and that the medication can help people with the distress and/or disability but might have some unpleasant consequences of its own.41 Patients must have a ballpark sense of the seriousness of the condition described, as well as the risks and benefits of the treatment; but perhaps ballpark generalities are enough.
Even if we wish for patients to understand only what they have just heard, patients may not understand well enough that they are to repeat just that. They may have heard there are
symptoms of schizophrenia, including x, y, and z, and appended in their mind some others they know only too well. They may think that what is important to understand is that there are unpleasant happenings associated with schizophrenia, and not the particular ones the researcher has mentioned.
So, at the very least, subjects should be instructed that it is necessary to mention all of the items on the list and only those items. Those who nevertheless give extra-disclosure information
should be warned that this is not the information wanted. Subjects may still recite extra-disclosure information - they may have source amnesia, as it were - while understanding both in - disclosure and extra-disclosure information perfectly well; but we may want to be sure patients understand the disclosed information, and so we should arguably not be overly concerned with false positives.
There remains another more serious problem with the UTD: it may overmeasure and undermeasure pure understanding.42 To do well on the UTD, one must be able to attend, process, record, retain, and recite back the information provided. Being able to attend and process information are clearly important for (pure) understanding. But being able to record, retain, and recite back are not necessarily so. Thus the UTD may measure abilities not essential to pure understanding.
Conversely, a person who is able to recite back - which is all the UTD arguably asks a person to do - may not have (purely) understood anything he has heard. A parrot can recite phrases
back with no understanding at all. One suggested modification of the UTD? - in the section asking whether a statement is the same as one made in the disclosure, use sufficient numbers of
statements that are equivalent in meaning but use completely different words.43 (Of course there may be people who understood the original information but not the equivalent statements; but no measure can be perfect.)
The UTD, then, is an impressive instrument: it does a good job spelling out the items of information, for the three relevant diagnoses, that patients ought to understand; it explains this information with a simple enough vocabulary that most people should be able to understand it; and it tests understanding of these items of information in several different ways, thus allowing patients full scope to demonstrate what they have learned in the best way they can. On the other hand, the UTD has some problems. It may set the level required for being unimpaired too high by requiring disclosed rather than extra-disclosure information or information at a higher level of generality. In addition, it may not require enough in the way of understanding inasmuch as it tests essentially recall and repeating ability rather than understanding. Finally, difficulties remembering may not disable someone from making an adequate decision, provided he remembers enough at the moment of decision to process all the relevant information. Despite these difficulties, the UTD has many virtues.
Assessment of the TRAT
Requiring highly developed reasoning ability - e.g., knowledge of all the rules of logic, absence of any strong emotions influencing a decision, etc. - is problematic for a number of reasons, and we shall have to assess whether the TRAT does so.
First, it is unclear that pure or pristine reasoning plays an essential role in all effective decision making; intuitive, idiosyncratic processes actually may improve decision making in some cases. (Consider cases in which people dream of solutions to difficult mathematical problems.) Thus, fully intact reasoning may not be necessary for adequate decision making (recall our example of the good accent). Concerns about protecting the vulnerable, then, may not be implicated.Moreover, what qualities of reasoning are good may be open to dispute, so that to
require some particular form of reasoning may be to discriminate against unconventionality.
Perhaps most important, even generally effective decision makers who clearly have the ability to form accurate beliefs misuse statistics, misunderstand probabilities, and accord undue weight to vivid examples. They also may be affected profoundly by irrational and unconscious factors. Thus, unless we are to declare most people incompetent, to so declare only the mentally ill who exhibit these deficiencies amounts to prejudicial discrimination.
Requiring some reasoning ability, on the other hand, seems to make sense, as I argued above: if one does not know how to put together the information one has understood, so to speak, one will not be able to reach the right conclusion. The question is how much reasoning ability should be required. Does the TRAT set the right level?
In one way, the TRAT does seem to require the presence of fairly basic abilities. For instance, in testing the understanding of probabilities, it requires only the understanding of a grossly obvious inference. One problem with the TRAT, on the other hand, may be that it does not justify requiring all of these abilities it tests for - or rather, giving better scores the more abilities there are in evidence. A particular decision, for instance, may involve only two alternatives; what, then, is the relevance of (say) transitive thinking, or of complex versus comparative thinking in that case? More important, a patient might not engage in many of these cognitive functions because, for her, one consideration is decisive; she may (say) so disvalue a risk of one of the alternatives that thinking consequentially is all she needs to do to choose between two alternatives - particularly since she may trust her doctor not to propose otherwise horrendous alternatives.
On the other hand, it may be that evidencing all the TRAT abilities puts one in a better position to decide by giving one more options. If one asks for further information, for example, of if one considers unstated consequences of the choices, one may be in a better position to decide. The question is whether this added strength of ones decisionmaking process is really necessary to competent decision making.
Another problem with the TRAT may be its requirement that one evidence all of these other functions; they may be occurring at an implicit level. For instance, a patient who says I want x
and not y because I am terribly frightened of the significant seizure risk carried by y - my dad died in a car accident when I was three as a result of a seizure - will often have gauged that x does not carry such a seizure risk (or anything equally abhorrent to the patient). If she has not done so, her decision making is arguably deficient. But she may well have done so and may simply not say the words and I have compared x to y and x does not have any such abhorrent consequences to me.
If this is right, there is at least a problem with the TRAT, i.e., that it is difficult to test for the different functions. Perhaps instead of simply asking for reasons, the patient, once having given a reason, should be asked directly if he compared y to x and, if so, what in the comparison led to his choice? On the other hand, this approach may be too leading. I leave to psychometricians
the task of arriving at effective, nonleading ways of eliciting this information.44
In short, the TRAT does an impressive job in identifying reasoning abilities, arguably necessary for competency, and testing for them in a reasonable way. It nevertheless suffers from two problems. First, it may sometimes require abilities that do not really add, in a given case, to the patients decisionmaking process. Second, it may underestimate how often those cognitive processes are actually occurring, while the patient is simply not saying the magic words. Nevertheless, the TRAT remains an impressive instrument for the reasons given above.45
Assessment of the POD
The POD, as will be recalled, measures a patients appreciation of her or his illness as well as of the potential benefits of treatment. The ability it taps is the ability to assess evidence; and so the quality of the patients beliefs is at issue. While an ability to assess evidence is important, the POD requires too high a level of that ability - and too high a level in a context when we sometimes cannot even decide what a high level is. In my view, the POD is seriously flawed and needs to be radically revised.
Deciding what beliefs a patient must have to be deemed competent is fraught with danger. Since decisions take effect in the world, we want the patient to come to accurate beliefs about the world. The problem is that it is - more often than we like to think - an open question what is true; very few beliefs are completely indisputable. This means that requiring particular beliefs may not further our interest in protecting the vulnerable; if the beliefs we require are wrong, we are not putting the patient in a better position to decide. We should not say the patient is incompetent - indeed, if most of us are wrong, the patient may in fact be super competent.
Perhaps, more important, freedom includes freedom to decide what is true no less than what is good. If we require particular beliefs, we prevent the patient from pursuing the truth according to her or his own lights. There may be some limits on what patients can believe. But too stringent limits severely curtail patients freedom to be unconventional in their pursuit of truth. And once again, patients may turn out to be right.
In addition, we must be mindful of the fact that many people have distorted beliefs and make decisions on the basis of those beliefs. Unless we want to call many, if not most, people incompetent, we are discriminating against the mentally ill if we disable them on the basis of many of their distortions. There is a range in the kinds of beliefs we can require or proscribe. The POD is at the far end of the range: patients must, essentially, believe what their doctors believe about their illness and treatment. This cannot be a correct standard. Given the many contested things in our society, an individual doctor cannot be held out as the final authority on truth.What would become of second opinions if doctors were?
Let us look at the submeasures of the POD in a little more detail. The NOD (Non-Acknowledgment of Disorder) measures appreciation of ones illness. One receives a full score if one accepts the diagnosis ones doctor has given one; judges ones illness as severe as a particular measure of symptom severity does; and accepts the symptoms reported in ones chart.
The first two are problematic. Ones doctor may be wrong about ones diagnosis. To put this in a more technical way, the NOD is limited by the reliability and validity of psychiatric diagnosis. In fact, doctors disagree quite often about a particular patients diagnosis. They often disagree about the category of illness (e.g., psychotic disorder vs. mood disorder vs. personality
disorder). They sometimes disagree about whether a patient even has a significant illness.
A single individual cannot be made the final authority on truth.46 Indeed, the patient may be quite willing to believe an earlier doctors diagnosis or even that he is seriously ill; but if he disagrees with this particular diagnosis, he is counted impaired on this measure of the NOD.47
The second measure of the NOD asks whether the patient rates his symptoms as severe as the Brief Psychiatric Rating Scale does. This measure also makes little sense. First, the patient may be the most important judge of how severely ill he feels. Second, even if the question is how severe his illness is on some common metric, how shall we expect patients to know that metric? This is a comparative judgment which patients are not in a good position to make. A result that diverges from the scale is not a profound distortion of reality. Finally, as with the first measure of the NOD, this measure is limited to the extent that the severity ratings of the Brief Psychiatric Rating Scale are not highly reliable and valid.
The third measure of the NOD is more acceptable. It asks whether patients acknowledge the presence of symptoms mentioned in their chart. Many of these symptoms will be grossly demonstrable. If a patient denies that he has just been frenetically pacing, or hasnt slept in days, he is severely distorting reality. Some symptoms, of course, involve more interpretation.
Is the patient agitated? Maybe not for him. Other symptoms essentially duplicate the illness question; patients cannot deny they are ill and admit to hallucinations and delusions, so-called, although they can admit they are seeing, hearing, and believing things that others do not. Framing the question in terms of whether patients are experiencing delusions or hallucinations should therefore be off limits.48
The NOT (Non-Acknowledgment of Treatment Potential), which measures acknowledgment of treatment potential, is also problematic. The NOT requires one to accept a good prognosis with treatment if it exists; a good prognosis in particular with medication if it exists; and a worse prognosis without treatment. As will be recalled, if the patient has reasonable grounds to disagree with the doctors judgment, a hypothesis nullifying his premise is presented and he is asked again his beliefs. (Imagine that a doctor tells you there is a medication that has been shown in research to help 90 percent of people with your problem, even people who had not gotten better with any other medication.49)
The NOT is problematic once again because the doctors may simply be wrong about ones particular likelihood of benefiting from treatment and declining without it. For instance, some patients may become demoralized and depressed at the need to take medication and essentially stop trying, just as some may regress in hospitals and never want to leave. It may be indisputable how patients on average do with and without a particular treatment. But averages dont speak to this particular patient, and he may be right that he will be in the 10 percent that do not respond to a particular treatment.50
Because no one can predict the future with complete confidence, it is problematic to require patients to form beliefs about a particular outcome they will experience in the future. Asking patients to understand what happens generally may make sense; asking them to believe that the general rule will apply to them does not have.
To look at this in another way, in essence the NOT measures optimism and pessimism. Many people are unduly optimistic or pessimistic about many things. Thus, in addition to requiring patients to believe things that may be false, and taking away their right to decide about truth as they will, the NOT requires patients (generally mentally ill people) to manifest a trait - optimism or pessimism as the case may be - that many people do not.51
I have noted that there is a range in the kinds of beliefs we can require. At the far end is the naive view incorporated in the POD that we can require the patient to believe essentially what the doctor believes. This view is problematic, it seems to me, for all the reasons given above. At perhaps the other end of the range is the view that patients can believe anything but beliefs impossible in the nature of things. But quite apart from the fact that some philosophers will tell you that virtually anything, technically, is possible,52 this view probably allows beliefs that are such a serious distortion of reality that we should not allow them. They are most probably wrong and are not widespread, so that ruling them out does not unduly infringe the right to be unconventional.
Within these extremes other standards are possible. Perhaps patients must believe what most doctors would believe about them. Or, so as to accommodate people who do not buy into
the medical model,53 perhaps they must believe what most people would believe. Or should it be most rational or most reasonable people? On the other hand, is this not just requiring patients to conform to norms? Perhaps we should try - if this is even possible - to characterize the standard in a way that does not refer essentially to majorities.
To my own way of thinking, we want a standard somewhere above a standard that proscribes only impossible beliefs and below a standard that makes doctors the final authorities on truth - somewhere, in other words, in the middle: patients cannot hold beliefs that obviously distort reality/are based on little evidence/are indisputably false/are patently delusional - we can choose the precise locution we wish. Perhaps referring to what most people would believe is the way to get at this; there may be others. The reader will recognize that I have evaluated the POD according to a kind of patent distortion of reality standard; I undertake to justify such a standard elsewhere.54
I want to conclude, however, by suggesting an idea that may be somewhat radical: that, according to such a standard, denial of mental illness often does not disqualify one from competency. 55 I hasten to add that I have no doubts whatsoever myself about the reality of mental illness and the severe suffering it causes. I subscribe completely to the medical model.56 On the other hand, denying that one is mentally ill may have characteristics that should lead us to say that it should generally not go toward a finding of incompetency.
Let us consider why. First, a person denying he is mentally ill may simply not be willing to admit to something that is stigmatizing and carries negative consequences in our society. He may be frankly lying about what he thinks, or may be conflicted about what to think for these completely understandable reasons. In either case, a person trying to avoid the negative consequences of a mental illness diagnosis may be thought to be acting quite rationally in our society - and far from incompetently.
Second, even if the person cannot admit, even to himself, that he is mentally ill, he may be acting on the basis of a common, understandable, and often quite adaptive defense. Denial of difficult things is quite common. And so one might through denial be attempting to avoid the narcissistic injury of having a mental illness. People identify with - and accept ownership of - the contents of their mind; it is hard to believe that such an intimate part of oneself is diseased.
Denial is also understandable in a way that other primitive defenses may not be. People can perhaps understand another persons saying that she just cant fully believe that her daughter
has died or that she herself has cancer; they do not in the same way understand a persons saying he just cant help believing that someone put a transistor in his brain.
In addition, denial can be very adaptive. There is evidence that seriously physically ill people live longer if they deny the seriousness of their illness.57 And there are other less striking examples of the same kind of beneficial effects from denial. A person denying he is mentally ill might draw on resources he would be too discouraged to use if he admitted his illness.
Third, mental illness diagnoses are simply less certain than many, if not most, physical illness diagnoses.Unlike physical illnesses, where there can often be definitive physical findings that unequivocally establish the diagnosis, there are no physical tests for any nonorganic mental illness.58
Thus, even when there is considerable consensus among physicians about a particular persons diagnosis, it is possible, though perhaps not likely, that the consensus is mistaken. Consider seizures prior to the advent of EEGs; or lumps in the breast prior to the advent of biopsies. Physicians could have reasonable bases for diagnosing epilepsy or cancer; but they could always be wrong. In the case of the seizures, an EEG could establish, with some reliability, that there were no electrical charges firing and that hysterical seizures were the more likely diagnosis. Prior to EEGs, that judgment was much less certain.
In the same way, a person presenting with the symptoms of a psychotic disorder might not actually have a psychotic disorder; she might instead have what used to be called a hysterical psychosis or what might be called today a factitious disorder with psychological symptoms. Alternatively, she could have a less serious diagnosis, such as obsessive-compulsive disorder with
bizarre obsessional thoughts that she does not quite credit and that therefore should not be called delusions. Or perhaps she is frankly malingering for some reason we have not discerned. Even such serious disorders as schizophrenia are thought by many to be a collection of different conditions; and sometimes what looks like schizophrenia may not be an illness at all - or at least the illness doctors thought it was.
Thus, while psychiatrists may have reasons for preferring to say a person has a psychotic disorder, they can always be wrong. In the same way, physical illnesses, without clear physical findings, cannot be as certainly diagnosed as those with clear physical findings. A diagnosis of Gulf War syndrome, chronic fatigue syndrome, Epstein Barr virus, soft tissue damage, irritable bowel syndrome, even Crohns disease, cannot be made as certainly as, say, a diagnosis of cancer.59
The point I am making is epistemological, not ontological. Again, I have no doubts about the reality of mental illness. In the same way, to say that we cannot definitively prove someone has soft tissue damage is not to deny that there is such an illness as soft tissue damage or that it can cause considerable pain and disability. The two issues are different.
Now the reader may object that I have established only that someones belief that she is not mentally ill is not impossible, not that it does not patently distort reality. Some diagnoses of mental illness are simply pretty obviously true - the person has recurrent episodes, responds well to treatment, etc. But denial of mental illness may still often not patently distort reality because, coupled with the fact that the belief is not impossible, why the patient believes as she does given her evidence may be completely understandable. People often have a hard time accepting a mental illness diagnosis because they have felt the way they are feeling for a long time - the illness has come on so gradually or feels so appropriate to their current surroundings, that it feels a part of them.Why are they suddenly mentally ill? In addition, people often accept far more responsibility for their thoughts, feelings, and actions than they arguably should,
so a patient may feel she has chosen to be the way she is. Her thoughts, feelings, and behaviors are not an illness, but a choice. (Compare the reasons people give for their behaviors that are
clearly a response to posthypnotic suggestion.) In short, when a belief is possible, and when the belief does not wildly depart from the evidence the patient has, we should not deem the patient patently delusional.
Fourth, many members of society are skeptical about mental illness - or at least about whether particular behavior patterns or symptom constellations amount to an illness.60 They may think that mental illness is a failure of will or consists of problems in living or is motivated by a desire to be cared for. They may attribute symptoms to stress and believe that the best response is coping with or avoiding the stress.
Many of these beliefs are not what we would call enlightened. Some amount to frank prejudice, or are at least based on ignorance. But my point is that if these beliefs are not all that
uncommon, then a particular patients believing them does not represent a gross departure from ordinary ways of thinking. Once again, we dont want to single out only the mentally ill
who hold certain beliefs when these beliefs are relatively widespread in society at large. In addition, that many people think this way is further reason to hold that the patients belief in this
nonimpossible belief is reasonable given his evidence and understandings of the world.
Fifth, I have suggested that it does represent a patent distortion of reality to deny that one is suffering from grossly demonstrable symptoms. But the patient who can admit that she or he is agitated, pacing, scared, or whatever, has every reason to accept treatment doctors say will help those symptoms abate. It is not clear that we need to make the patient admit to the illness; it is almost forcing a humiliation on the patient to do so. In the same way, a patient who admits to abdominal pain and all the symptoms of Crohns disease, and understands that her or his doctor can recommend treatments that will help, need not admit to the disease to be in an adequate position to decide.
Perhaps, however, we should require more; patients need to accept not only that they are pacing, say, but that they have some condition, even if it is not the condition their doctors say they have. Or should we require patients to admit they have some condition that looks like schizophrenia and that most doctors would so diagnose and that is thought antecedently to be as likely to benefit from treatment as any other similar presentation. These claims are fairly indisputable in many cases - we dont need a physiological test to establish them. Thus, while a patient may not trust the individual doctor telling him about his diagnosis, he can and should trust the DSM-IV. In addition, these claims may be a necessary added ingredient before some will consent to treatment. I think it is a close call whether we require these additional beliefs or whether simply admitting to ones symptoms and ones doctors belief in potential benefit of treatment is enough. An intermediate position would be to require patients to admit, simply, that somethings wrong.
All of this said, there are at least two tacks one could take: first, one could simply exclude most denial as a basis for an incompetency finding.61 Alternatively, one could probe the denial further to see if the patients reasoning is such that one can understand - see as somewhat reasonable - his denial. Perhaps he is not speaking honestly. Perhaps he is narcissistically wounded but, in his heart of hearts, knows the truth. Perhaps he thinks of his behaviors as his choice. Perhaps he holds widely held views about mental illness that lead him to think he is not really ill. In short, one would probe to see whether a given case of denial should be thought to amount to a patent distortion of reality. The implicit assumption is that if a belief is not impossible, then one must consider how plausible it is - and whether it is an understandable or common belief - to see whether the belief patently distorts reality.
One final point about denial: allowing denial to be a basis for an incompetency finding - and thus forced treatment - is in fact fraught with danger. Not only would it permit us to force treatment on an obsessive-compulsive person who denies that he is ill - and who among us is free of maladaptive personality traits? - but it would also allow us to characterize political dissidents as ill, and then to use their understandable denial that they are ill as a basis for their involuntary treatment, despite the fact that their denial is to be expected.
In addition to this central problem of treating denial in the wrong way, the POD may also, at times, understate the presence of incompetency by focusing too exclusively on disavowal of
what ones doctor believes and not enough on the degree of distortion which the belief represents. Take the patient who admits he has the diagnosis his doctor says he has and agrees with his prognosis with and without treatment. This person would receive a full score on the POD. But suppose he also believes that he has the diagnosis his doctor says he has because aliens are manipulating his neurotransmitters from afar; and that taking the medication will enrage the aliens and cause them to destroy the earth - even though he thinks it will cure his illness. Again, this person would receive a full score on the POD. But is he really competent to refuse treatment? Do we not want to look for patently false beliefs and not just disagreement with what ones doctor says?
The POD, then, is a fundamentally flawed instrument. It naively requires patients to believe what their doctors believe, even though doctors - certainly individually but maybe even collectively - can be wrong. It may also, at times, not only overstate but also understate the presence of incompetency by overlooking patently false beliefs that affect the patients decision but do not involve denial. The POD should therefore be changed in one of two ways.
One way - pending normative argument for the appropriate level of impairment in assessing evidence - would be to design an instrument with different levels of the adequacy of ones belief. For instance, one could look at whether the patient denies what his doctor says; what most doctors would say; what most reasonable people would say; what is patently true; and what must be true. Future scholars would then establish which level of each belief to require. Another way is to tentatively adopt a particular level that one thinks is plausible. I myself think a level somewhere in the middle - a patently false belief standard, however characterized, looks likely to be correct. I intend to try to justify this level normatively in future work. In addition to either of these recommendations, one would want a method to probe the patients reasons for her beliefs, so as to ferret out misconceptions at whatever level they may occur.
Assessment of the MacCAT-T
The MacCAT-T, as discussed earlier, is a streamlined version of the three research instruments described above. Therefore, many of the observations regarding the instruments described above also apply to the MacCAT-T. The MacCAT-Ts appreciation component does try to acknowledge the difference between nonagreement that is nondelusional and has some reasonable explanation and nonagreement that is based on a delusional premise or some other belief that seriously distorts reality and does not have a reasonable basis in the patients cultural or religious background. But this effort is not entirely successful, inasmuch as the range of reasonable explanations given is quite narrow and, once again, only culturally or religiously sanctioned beliefs - conventional beliefs in that sense - are permitted to ground reasonable disagreements. Indeed, the MacCAT-T scores as 0 a patients belief that his symptoms are related to circumstances other than a psychiatric disorder, such as stress or overwork. But the patient may be right or - given widespread beliefs in our society about psychological distress - is at least holding a nonpatently false belief.
The MacArthur Instruments Authors Response to Similar Critiques of Their View
A 1996 symposium issue of Psychology, Public Policy, and Law, devoted to the MacArthur Treatment Capacity research instruments, contains a number of articles critiquing the POD on the basis of my work; Christopher Slobogin62 does so in the most sustained way, but Susan Stefan63 and Trudi Kirk and Donald Bersoff64 critique the instrument on similar grounds as well.
The authors response to the basic critique is severalfold.65 First, they note that the critics all seem to want some measure of appreciation of illness and treatment to be included in a competency instrument, even if they object to the precise measure used. Second, they suggest that they may well not be all that far apart from the critics in the measure they want: they acknowledge that mere nonacknowledgment of ones disorder or of the realistic consequences of treatment is not enough to constitute incapacity, but the acknowledgment must be related to delusional thinking or other medical or psychological conditions that are responsible for a serious distortion of reality. They add that they accept the concept of a patently false belief, provided it is not restricted to delusions but may also include nondelusional reasons for denying the existence of ones disorder, such as parietal lobe damage and intolerable anxiety related to recognition of the disorder. Third, they acknowledge that their instrument does not formulate a criterion for patently false beliefs, and suggest that it was difficult for them to operationalize this concept; they invite others to try. Finally, they note that the MacCAT-T requires clinicians to make a judgment about patients reasons for denial of their symptoms in
order to rate their appreciation. The requirement represents an effort to include the patently false belief component in the capacity standard; the authors thought it possible to do so only by relying on clinical judgment, at the cost of sacrificing some psychometric reliability.
The authors response is helpful - and points to some recognition of the concerns that animate my argument - but is not completely satisfactory. Their first point is well taken: everyone weighing in on this issue does, and everyone should, care about beliefs the decision maker forms; they are simply extremely important to competency.
But I take issue with their second point - that their view is at all close to mine. They say they want to pick out only beliefs that seriously distort reality - which is important. But while I agree that there may be a variety of reasons for serious distortions of reality, such as anxiety or dissociation (although if the distortions are serious, dont they necessarily amount to delusions?), part of my point in discussing denial of illness has been that denial, at least of mental illness, is often not a sufficient distortion of reality to help justify a finding of incompetency.
The third point the authors make is that, although they approve of some notion like a patently false belief to help measure competency, they found it very difficult to operationalize such a notion, and therefore did not include it in their research instruments. They invite others to try to formulate a workable concept. It would be tempting to say that this notion would be fairly easy to apply; but given the authors misunderstanding of what the terms mean in the case of denial of illness and treatment efficacy, that claim is rather hard to make. In any case, in future work I intend to undertake to operationalize this concept in collaboration with others skilled in test design.
Finally, the authors note that the MacCAT-T attempts to introduce the notion of a patently false belief by requiring examiners to assess the reasons for patients denial. This is a step in the right direction, and I am pleased that the authors see a need to introduce such a concept. Given this approach, the most obviously well-reasoned bases for disagreement with ones doctor would not result in a finding of incapacity, as they currently do according to the POD. On the other hand, the reasons that the researchers would allow to justify disavowals are much more limited than makes sense; their exclusion for cultural and religious bases of distorted beliefs is also too narrow and they count beliefs that I would find acceptable as the most serious disavowals.
In short, the MacArthur researchers have not adequately answered some of the same types of criticisms leveled elsewhere, although they make a good faith effort to do so.
Concluding Evaluative Comments About the MacArthur Instruments
The MacArthur instruments identify the abilities necessary for competency, thus enabling us to protect the vulnerable; but they set the level of those abilities too high. The central failing of the instruments is insufficient attention to the need to protect unconventional behavior and to the pervasiveness of minor decisional impairment throughout the population as a whole.
The instrument designers recognize that competency issues involve normative judgments that balance freedom of choice and protection of the vulnerable. But they fail to recognize that freedom of choice includes not only freedom to choose treatment (or no treatment) but also freedom to choose what skills to use in deciding about treatment. Most important, patients, within broad limits, should be given their choice of what to believe, no less than what to decide.66 The patients decisionmaking process, as measured by capacity and/or competency instruments, implicates the same normative issues as the patients choice.
Back to Top
Importing the MacArthur Instruments into the Research Context
The MacArthur instruments were designed for measuring treatment capacity, but may well be of use in the context of consent to participate in research. To think about this issue we must ask, first, whether any abilities in addition to those the instruments assess are needed in the research context. Second, even over the range of abilities the instruments cover, certain adaptations may be necessary to tailor the instruments to the research context. Third, we must face additional normative issues.67
The answer to the first question is, arguably, yes. The patient asked to participate in research is serving the interests not only of himself but also of the researcher.68 This means that the researcher may have something of a conflict of interest, and so perhaps cannot be counted on to be entirely benign. (I in no way think that researchers are selfish or malevolent; but unconscious motives for wanting to do their research may make them less reliable guardians of patients interests.) It also means that, given the enormous transference patients bring to doctor/patient interactions, the patient may not be in a good position to protect himself - to make the best judgment for himself in the absence of a doctor whose job it is to make the best judgment for him.
Indeed, patients may have all sorts of unconscious reasons to consent to research when a doctor asks them to do so. I have mentioned transference - clearly the most powerful. Involved in this may be a desire to please the doctor, a desire not to be the object of her or his animus, a belief that the doctor offers protection from all harm, or a belief that the doctor must have only the patients interests at heart. In addition, patients may believe that they will not get other therapeutic treatment if they are unwilling to participate, will get the best treatment only if they participate, or will be able to survive financially only if they are treated through a research protocol. Finally, the doctor may put some pressure on the patient to consent, and many people have a hard time saying no.
I recognize that patients suffer transference distortions in the treatment context as well. But again, the doctors and patients interests are largely the same in that context. Moreover, patients are likelier, in the research context, to want to decline participation, but feel they cannot. All of this at least suggests that we may want some measure of ability to protect oneself against pressure - to negotiate to protect ones own interests, or something of the sort - in a capacity/competency measure for consent to participation in research.
Second, the MacArthur instruments will need to be adapted in major ways to accommodate the research context. Under the UTD, they will need to include the most important information patients need to be able to (purely) understand. Most notably, the patients will need to understand that nontherapeutic research will not help them, how likely therapeutic research is to help them, and that the doctors have something of an interest in doing the research. Also, patients must understand their right to not participate in the research without incurring any penalties. In addition, patients may well want to understand the potential gains of the research in terms of increased knowledge - of what kinds and how much. Of course, patients will also need the usual disclosures describing the procedure itself and its potential risks. The UTD offers a good model for level of language comprehension to require and for different ways to test for understanding; but, once again, the actual content of the information will need to be adapted to the research context.
In the same way, the POD will have to be reformulated to ask about the patients appreciation (belief formation) concerning these important things. As in the context of treatment, we arguably want patients to avoid believing only what is patently false. In nonobvious cases, a patient (who can resist pressure) who understands that a certain protocol is not designed to help him and probably will not help him may nevertheless optimistically hope it will; the point is, maybe the protocol will help him, even if there is little reason to expect so in advance.
On the other hand, if there is no chance the protocol will help, the patient must at least recognize that that is what his doctors think, and should perhaps be required to believe this himself. It all depends on whether this is one of those things on which doctors could be wrong.
Of course, I am again proposing a kind of patently false belief standard, and one might reject that view. Another level of belief, required in this context,must then be proposed. Alternatively, the POD in the research context could include a number of different levels at which ideas must be appreciated.
The POD must not only include additional items for the patient to appreciate, and at the right level, but it must also omit items which seem irrelevant to the inquiry. For instance, denial of mental illness seems irrelevant to the case of a patient who consents to participate in research.69
Third, there are additional normative issues in the research context. These include whether the additional important value, in this context of advancing science, justifies a different balance in our assessment of competency, and whether it makes normative sense to require more ability here because consent is of more questionable value to the patient in the research context than in the treatment context.
I address only the second issue here. The question raises a broader question of whether we should vary our competency standard depending on the quality of the decision being made. Everyone agrees that if a task is more difficult, one must have additional abilities, or a higher level of the relevant abilities, or both, to perform it. The proposal here is different: that if a patient is making a good decision, we should require only a low level of competency, while if he is making a bad decision, we should require a high level. Here, the decision to participate in research is not so beneficial to the patient as the decision to consent to conventional treatment.70
The reason we might want to vary the level of competency is clear: if patients are about to choose something that will not help them and may harm them, we want to be very sure that they know what they are doing.
To my mind, however, there is a very serious conceptual problem with varying the level of competency: doing so is only a distant cousin to declaring people who make good choices competent and people who make bad choices incompetent - a practice roundly criticized. In essence, one sets up each individual evaluator as judge of the quality of the patients decision, and allows him to substitute his judgment for the patients. If the evaluator, for example, disvalues limitations on occupational functioning to a high degree, he may disable the patient, by varying the level of competency, from making a choice that risks such limitations in his occupational functioning. But what if the patient himself does not much care about this?
One response to this charge - although not entirely satisfactory - is that whatever level of competency we set we are balancing well-being against autonomy, so that striking the balance differently makes perfect sense when well-being is likely to be affected more seriously.
But there is a difference between saying that one must have certain abilities as a general matter in order to take responsibility for ones own choices, without scrutiny of particular choices, and saying that one must have more abilities when we judge ones particular choices bad. Arguably, competency doctrine does and should set the balance once, as it were, so as to avoid second guessing patients decisions and setting someone else up as a judge of what is a good choice; isnt that up to the patient?
If we reject varying the level of competency based on how good a decision we think the patient is making, there may nevertheless be good reasons for requiring a higher level of competency in general in the research context. For instance, we may think that, as a risk-of-error matter, evaluators are likelier to have an interest in finding competency so that their patients will be able to consent to research that will help the researchers. We should err on the side of requiring a higher level of competency.
In addition, we may be concerned that the researcher cannot be counted on to be entirely well-motivated in recommending the particular research protocol; once again, the researcher doesnt completely have the patients interests at heart. This matters here, not because we think the patient is making a bad choice, but because we expect him to play a larger role in evaluating
the decision, and he may need more capacity for this evaluation.
What about the argument that research subjects do not lose much by not being permitted to participate in research, and may at times suffer injuries by participating, so we should set a
high bar before we allow a vulnerable subject to participate? This is a slight variation on the sliding scale competency approach we have just discussed. Consenting to participate in research is not making an obviously good choice for oneself - and so we want one to know well what one is doing. In addition, one might argue that denying the person the choice to participate does not much harm him. Indeed, we might even think that many research subjects do not feel harmed when deprived of the right to make such choices, because they are likelier to be acquiescing in what their doctor wants than independently choosing to participate.
This suggestion has some merit, but does not seem to me to carry the day. Patients can desperately want to participate in therapeutic research when nothing else is helping them, and
indeed it may be the best choice for them to do so. They can derive great utility from the thought of helping others. And they can feel terribly demeaned when their choice is not respected.71 Thus, it is not obvious that we should raise the bar for competency to participate in research.
Given all of these competing concerns, I am not sure how I come down on the controversy around sliding scale competency measures, although I am tempted by a single level of competency. At the very least, I believe that we should increase the level that we require of patients only in the most exigent circumstances: when their choice exposes them to a serious risk of very substantial, perhaps irreversible harm. This policy would avoid letting individual evaluators make normative judgments about their evaluees choices.72 Even then, I might call the impairment justifying overriding the choice in exigent circumstances something other than incompetency (perhaps impairment?), so as not to open the door to more manipulation of the concept. In addition, we may have reasons for thinking that competency should be understood in realist terms as either there or not there, and not variable according to context.73
In sum, how we eventually come down on this issue will require careful normative consideration, which I have not undertaken here. And this question must be answered before we definitively adopt a competency standard or measure.
Back to Top
What recommendations should the National Bioethics Advisory Commission make to institutional review boards regarding competency evaluations?
Two questions arise with respect to NBAC recommendations to IRBs: first, should researchers be required to evaluate competency when recruiting subjects for research? And second, should they be required to use particular instruments to do so? It seems clear that researchers should inquire into competency whenever there is reason to suspect it is not present. This is particularly so in the research context where consent is not obviously in the patients interests. Perhaps there should be a presumption that hospitalized psychotic patients should be evaluated, although requiring evaluation of anyone with a chronic psychotic illness may underestimate competency in this population. However we draw the line, if there is reason to doubt a potential subjects competency, it should clearly be evaluated.
Then the question becomes how to evaluate it.What should the NBAC recommend, if anything, for IRBs to require of investigators who wish to perform research on vulnerable populations? One possibility is that IRBs might wish to require use of the MacArthur instruments with suitable adaptations. MacArthur seems the most promising instrument available, but if the criticisms presented here and elsewhere are sound, it is also flawed.
If the MacArthur instruments are required, IRBs must realize that it is never sound for investigators simply to use them without more. At the least, they must be adapted to the research context. Perhaps they should also be modified in the light of criticisms I and others have made.
Perhaps most important, because the MacArthur researchers have omitted the crucial normative judgments necessary before the instruments can be used, some effort must be made to justify the normative choices the evaluator wishes to apply. For there is a real danger that an investigator faced with a requirement to use the MacArthur instruments may simply adopt its definition of impairment as the cutoff point for incompetence, or decide that the line the Mac- Arthur researchers say indicates clear competence should also be the line below which incompetence falls. The MacArthur researchers did not justify these normative choices, so that using these cutoff points, without normative analysis, is improper.
If all of this is so, IRBs could require, at the least, that investigators study the MacArthur instruments, as well as critiques of the instruments as given here and by other scholars. Alternatively, they could require some sustained analysis, submitted to the IRB, of why the investigators are making the normative choices they are making - as well as of their responses to other criticisms of the instruments. To my way of thinking, the second provides much better assurance that the required analysis is being performed, and that investigators are not simply
adopting the MacArthur researchers suggestions.
On the other hand, it may simply be too costly to require such analysis, and researchers on the street, so to speak, may not be very skilled at this kind of analysis. It is, arguably, not within doctors and other scientists expertise to engage in this kind of philosophical inquiry. Thus we should encourage scholars to engage in the hard analysis of the normative questions. On the
other hand, investigators can hardly be expected to wait for this kind of analysis, especially if some consensus is required - research must go on.
There are ultimately only three possibilities:74 require investigators to use MacArthur with suitable adaptations after at least considering problems with the instruments and choices that must be made in order to use them, and perhaps justify those choices in some sustained way; require investigators to use some other instrument - say an instrument of their own devising; or allow, instead of the use of instruments, clinical investigations of competency.
Which option IRBs should use depends on how problematic one finds the MacArthur instruments to be; how confident one is of the ability of investigators to moderate the problems; and whether one thinks more open-ended evaluation involves the same kinds of problematic normative judgments, but simply sweeps them under the table, possibly achieving less accuracy in the end. How to answer the question of what IRBs should require is difficult given the state of the art of capacity/competency instruments today. Further research is needed both on the substantive question of what ones instrument should look like and on the procedural question of what to do in the face of the problems identified in existing instruments.
Back to Top
Conclusion
The MacArthur instruments are an extremely important contribution to the literature on treatment capacity. I have suggested two essential problems. First, the instruments omit necessary
normative evaluation and smuggle in certain judgments without justifying them. Second, in some cases, they arguably strike the balance between autonomy and paternalism in the wrong
place. Future research must focus on remedying these problems. In particular, to the extent that the normative inquiry leads us to select a patently false belief standard to measure appreciation, effort must be made to operationalize such a standard. Notwithstanding the need for future research, the MacArthur instruments are an impressive achievement and will
no doubt be a focal point for the debate on capacity/competency for many years to come.
Back to Top
Footnotes