Accredited Psychiatry & Medicine

Subjective Data and Suicide Assessment in the Light of Recent Legal Developments
Part II: Clinical Uses of Legal Standards in the Interpretation of Subjective Data

Harold Bursztajn,* Thomas G. Gutheil,** Robert M. Hamm,*** and Archie Brodsky****

*Clinical Instructor in Psychiatry, **Associate Professor, ****Research Associate, Harvard Medical School, Program in Psychiatry and the Law, Massachusetts Mental Health Center, 74 Fenwood Road, Boston, MA 02115.
***Senior Research Associate, Center for Research on Judgment and Policy, Box 344, University of Colorado, Boulder, CO 80309.
The authors acknowledge their indebtedness to Ms. Audrey Bleakley for her assistance in the preparation of this manuscript.
Reprint requests to Thomas Gutheil, Program in Psychiatry and the Law, Massachusetts Health Center, 74 Fenwood Rd., Boston, MA 02115, USA.

Suicide assessment is a problematic area for the psychiatrist, given the tension between the uncertainty inherent in subjective assessment and the risk of a fatal outcome. In addition to the dread of a tragic outcome, there is also an ever-present fear of malpractice liability, along with discomfort at the prospect of having to defend clinical judgments based on data that may not lend themselves to the usual notion of scientific objectivity. These medicolegal concerns reflect deeper intellectual misgivings about the reliability of data that are neither quantified nor based on absolutely verifiable observations. On the one hand, together with more objective data, clinicians routinely take into consideration their perceptions of patients' feelings — and their own — in the assessment of suicide risk. On the other, they can never be quite sure that such data meet the standards of precision set by "hard science," let alone the law.

In Part I of this paper we demonstrated that psychiatrists' fears of indiscriminately rendered malpractice judgments for suicidal outcomes are largely groundless. For the careful, documented use of subjective data in arriving at a reasoned clinical assessment (particularly in the context of the attempt to obtain informed consent) satisfies all three major standards of due care typically invoked by the courts in determining negligence. Thus the clinician can follow the guidelines of clinical experience and theory without being dogged by the fantasy of judicial retribution for a lack of omniscience or omnipotence. Here we will take the argument a step further and show that the three standards of due care as well as the informed consent requirement can be turned to clinical advantage. That is, they can provide additional guidelines for disciplining clinical judgment so as to make wise use of subjective data. In this way awareness of the reasoning behind the law can make good clinical practice even better.

Let us begin by acknowledging the affective and cognitive biases of clinical judgment that make it impossible for the psychiatrist to be totally objective. Through an examination of a classic case in the suicide literature, we shall note some of the common shortcuts of human judgment — some of them useful, some amounting to distortions — that influence the clinician's interpretations and decisions. As we shall show, however, these shortcuts are inescapable, whether the data under consideration are objective or subjective. Therefore, rather than engage in futile attempts to eliminate subjectivity and diagnostic or prognostic uncertainty, the clinician should seek to be conscious of the judgment strategies employed. It is here that attention to the three standards for the judicial determination of negligence can provide a check against undisciplined subjectivity. Similarly, the interchange with the patient with regard to informed consent serves as a convenient framework for articulating and critically examining one's decision making strategies.

Clinical Uses of the Three Standards of Negligence

In Part I we reviewed the standards, derived from three traditions of negligence law, by which the courts decide whether a psychiatrist has provided due care in the treatment of a suicidal patient. These are the community standard (professional custom), the Learned Hand rule of maximizing benefits relative to costs, and the standard of a "reasonable and prudent" practitioner. We here propose that adherence to these standards is consistent with a sound clinical assessment; to realize this potential, each principle can be translated into questions one can ask oneself to monitor the adequacy of one's assessment.

Community Standards

The oldest tradition of negligence law, that of "community" or professional standards, holds that one has exercised due care if one has done what others in the relevant professional community (whether defined as local or national) would do in similar circumstances [1]. This legal standard suggests guidelines for the practitioner, which can be phrased as questions such as these: "Would anyone else take into account these same subjective factors?" "Would other clinicians feel as I do toward the patient (e.g., become depressed) — that is, do my reactions tell me something about the way the patient affects others (especially those closest to him or her) rather than simply about myself?" "Would one of my colleagues or, particularly, supervisors, remind me of other considerations that I am overlooking?" This sort of question, when asked either hypothetically or in direct consultation with colleagues, affords a degree of protection from either neglect of, or uncritical reliance on, subjective data.

Maximization of Benefits Relative to Costs

As a corrective to the consensual emphasis of the community standards criterion, the Learned Hand rule establishes a mathematical standard for judging what is best for the patient [2]. Briefly stated, the rule holds that "negligent behavior is the failure to invest resources up to a level that [is commensurate with] the anticipated saving in damages" [3]. Although the Learned Hand rule allows for subjective as well as objective data, it frames subjective judgment in a structure of cost-benefit calculation. The questions it invites one to ask oneself are questions about probabilities, values, short-term and long-term gains and losses. Such questions might include: "Is this a situation where it is relatively safe to rely on subjective data?" "Am I overlooking objective data such as the statistical probability that someone with this patient's diagnosis will attempt suicide?" "What if I'm right about the consequences?" "What if I'm wrong?" "What will be the impact on the patient's immediate safety? On the patient's ultimate well-being? On the therapeutic alliance?" "Am I sure enough of a high probability of small gains to risk a very low probability of a large loss?"[*]

Last but far from least, one should ask oneself, "Have I involved the patient as much as possible in the consideration of costs and benefits through the process of attempting to obtain informed consent?" As will be shown in a later section of this paper, the informed consent procedure is explicitly concerned with the costs and benefits of alternative courses of action. By engaging in a dialogue with the patient over informed consent, one avails oneself of an invaluable device for focusing attention on possibilities that might otherwise be overlooked.

The "Reasonable and Prudent Practitioner"

The third tradition in negligence law, as applied to medical malpractice, holds that one has exercised due care if one has done what a reasonable and prudent practitioner of the same discipline would do in similar circumstances [5]. This legal tradition interprets the other two in a manner consistent with both modern probabilistic science [6] and good clinical sense. It spurs the clinician to a more discriminating use of the "rules of thumb" that are legitimized by professional consensus. At the same time, it tempers the clinical use of formal decision making procedures with an appreciation of the uncertainty that enters into all judgmental processes, scientific and otherwise.

In daily practice this standard of due care prompts the clinician to ask questions such as: "What is my implicit philosophy of science, the set of standards by which I judge my clinical reasoning to be 'scientific'?" "Am I focusing on a single cause of the patient's illness to the exclusion of other possible causes (and of chance)?" "In assessing observed improvements or deteriorations in the patient's condition, am I keeping in mind that causal factors (and the ways they interact) can change?" "Am I falling into the old trap of looking at the patient in isolation from the effects of the therapeutic alliance and the therapeutic milieu?" "Am I being as sensitive as I can to the effects that my own feelings (e.g., countertransference hate) may have on the patient?" What a reasonable and prudent psychiatrist would do, then, includes making wise use of subjective as well as objective data in the assessment of suicide risk.

Documentation of Due Care as a Clinical Tool

Table 1 shows how each of the three standards of due care for the suicidal patient is best met and how it in turn becomes an ingredient of good clinical practice. In all three cases, careful documentation of how one's clinical reasoning satisfies the legal standard is an essential element of malpractice prevention [7]. Beyond that, such documentation has clinical value in that it protects against the vicissitudes of the clinician's own hindsight. By reviewing the documentation of one's previous emotional reactions to the patient, one can put in perspective one's present countertransference feelings. For instance, when a borderline patient experiencing a regression evokes countertransference hate [8], the therapist can balance this feeling by recalling a more empathic response which the patient evoked during calmer periods of treatment. Such a review can help prevent counter-therapeutic acting out on the part of the therapist and thereby allow the clinical work to proceed. At the same time, it provides a measure of the extent to which the present crisis has shaken the therapeutic alliance and put the patient at greater risk for suicide. Finally, it reminds the therapist of what buttressed the alliance in previous crises. The therapist can then engage the patient in this reflective process as well. In this way, the therapist can support the patient's use of ego functions to put in perspective feelings that would otherwise be overwhelming.

TABLE 1
Clinical Implications of the Three Standards of Negligence In Suicide Assessment


Legal Standard	Method of Meeting Standard	Clinical Usefulness

Community standards	Seek consultation with colleagues	Safeguards patient from idiosyncratic countertransference reaction by therapist.
	Document deviations from customary practices	Enables therapist to generalize from own reactions to patient to reactions of those in patient's support system. Helps therapist avoid overlooking diagnostic or therapeutic options which have worked in the past.

Maximization of benefits relative to costs	Obtain informed consent by outlining possible gains and losses as well as odds.	Builds therapeutic alliance through informed consent procedure.
	Document weighing of risks against benefits.	Tests patient's competence. Sensitizes therapist and patient to therapeutic benefits or reduced restriction and monitoring. Helps patient exercise ego functions of anticipation and planning. Provides corrective to hindsight reconstructions of subjective reasoning.

Reasonable and prudent practitioner	Consider multiple and changing causes, impact of therapeutic alliance on patient, and subjective data such as patient's and therapist's feelings.	Counteracts heuristic focus on single cause.
	Document how clinical reasoning is consistent with contemporary scientific practice.	Counteracts heuristic focus on patient in isolation from therapeutic alliance. Sensitizes therapist to possible impact of changing circumstances (e.g., milieu, vacation). Sensitizes therapist to possible impact of therapist's own feelings (e.g., countertransference hate). Provides a check against repression of clinical uncertainty.

The therapist, meanwhile, can exercise similar ego functions so as to make good clinical use of necessary legal precautions that would otherwise seem extraneous and irritating. The right-hand column of Table 1 lists some important ways in which attention to the requirements of the law can make for increased awareness and more discriminating use of the affective and cognitive biases of clinical judgment. These biases, their useful functions, and remedies for their misuse are illustrated in greater detail in the following case discussion.

A Case Study of Clinical Judgment: The Use of Simplifying Strategies

Havens [9] uses a well-documented case of a suicidal outcome to study the process of decision making in the care of a suicidal patient. Although the case is not one of negligence by any of the three standards discussed here, these standards might have proved useful (both in and out of the context of the informed consent discussion) as clinical guidelines for the therapists involved. Havens' analysis is, of course, an exercise in hindsight. Hindsight, like other information-processing heuristics discussed below, can be either instructive or misleading. It is dangerous, as Havens is well aware, when it is employed without an appreciation of the prospective unpredictability of events whose causation may in retrospect seem clear. Whether or not this particular suicide was preventable - and here Havens' conclusions are controversial - its analysis is a useful exercise. It points to problem areas in decision making and to preventive strategies that the treating clinician should be aware of. Our concern here is especially with the relationship between the use made (or not made) of subjective data and the clinical decisions that were reached.

The patient was a 55-year-old mother of two who had undergone a hysterectomy two years prior to her psychiatric admission. Subsequently she became convinced that she had breast cancer because her breasts were of unequal size. She also showed somatic signs and symptoms which included poor appetite, weight loss, and insomnia. Her internist, finding no significant abnormalities on physical examination and laboratory workup, referred her for psychiatric evaluation.

The psychiatrist initially noted that the patient looked depressed, felt depressed, and made him, the psychiatrist, feel depressed. This "depressive triad" (a time-tested subjective indicator), together with her vegetative symptoms, convinced the psychiatrist that the patient had a major affective disorder — depressive type. Further history revealed that the patient had an obsessive premorbid character clinically associated with later-life depression. She was the sort of woman who always kept things in order and had few friends outside of her involvement in several charitable organizations.

In the three years prior to admission the patient had undergone a series of losses. These included the symbolic loss of function signified by the removal of her uterus and the more concrete loss of close family relationships. Her two children had gone away to college. An aunt who had been living with the family had moved out. The patient and her husband, ostensibly to save money, had stopped celebrating the anniversary of their marriage (which was described as "amiable"). Still another major loss loomed on her horizon: her daughter's impending marriage, twice postponed at her insistence. Her reasons for opposing the marriage had always been unclear, her only explanation being that her prospective son-in-law was "too handsome" for her daughter. It seemed relevant as well that her daughter, a nurse, had cared for her during her surgical and medical difficulties.

Hospitalized for her depression, the patient initially improved as she developed a trusting relationship with a resident in psychiatry. Now more energetic and less delusional, she was able to leave the hospital to attend her daughter's wedding, which she reported that she "enjoyed." However, in addition to her daughter's departure on her honeymoon, the patient suffered three other significant losses over a two-week period leading to her death. First, the physician who had been treating her was promoted to chief resident and terminated with her. In addition, a nurse whom the patient had found especially supportive was kept at home by illness, and the medical student who had been working with her ended her psychiatric rotation.

Thus enumerated, her losses appear substantial. Why, then, did her therapists overlook them? For one thing, because it was part of her nature to minimize their impact. As Havens puts it, she "had an extraordinary gift of making the special ordinary, of turning the remarkable into the commonplace and unobtrusive. ..." Her composure, her imperturbability, her stoicism constituted a misleading self-presentation which the clinicians may have overidentified with, i.e., empathized uncritically [10]. Thinking, as it were, only with their hearts, they appear to have joined with her in keeping her conflicts buried.

A week after the wedding, the patient requested and received permission to go home for the weekend. Once there, telling her husband and son that she was going to visit her mother, she drove the family car to a high cliff by the sea and leapt to her death. At her funeral her husband remarked, "Maybe this was all for the better."

The therapists underestimated this patient's suicide potential when they granted her a weekend pass. Did they make what in retrospect seems a misjudgment by relying too heavily on subjective data? On the contrary, they seem not to have given much consideration to their countertransference reactions, either as clues to how the patient might be feeling or as warnings of ill-considered therapeutic responses of their own. "The anger that we suspect was carried secretly in her may have been conveyed to us and been just as secretly present in our clinical decisions," Havens concludes. In retrospect, he attributes the clinicians' apparent incomprehension of their patient's life-and-death needs to unconscious rejection motivated by dislike: "... we may have hurried our patient toward a false death out of disgust for something we felt in her, all the time largely unaware of what we were doing. I wonder if what disgusted us was not that possessive love of the daughter, with all its power to cling, deaden and destroy" [9].

Patterns of clinical response such as this can be understood in terms of both affective and cognitive processes. In the first place, the therapist's affective experience of the patient can affect clinical judgment for better or for worse [11-13]. Moreover, the clinician, having a limited amount of time and energy to process information, is compelled to find a satisfactory (rather than ideal) way of doing so [14]. The clinician thereby risks errors arising from the inevitable - and often helpful - use of cognitive short-cuts [15, 16]. The risk of error is usually assumed to be especially serious in the case of subjective data because of the leeway for interpretation that these data allow.

Affective Factors: Countertransference

The affective processes involved in clinical judgment are by their very essence subjective, as is the material they operate on. The therapist is seeking to interpret the patient's feelings. In so doing, the therapist's own feelings are engaged, thereby providing another layer of subjective data for interpretation. Psychodynamic theory directs the therapist to sift through his or her feelings for clues to what the patient is feeling. But as Strupp and Wallach make clear, countertransference (here understood to include all of the feelings aroused in the therapist by the encounter with the patient [17]) is a double-edged sword, with the power both to reveal and to obscure:

In its positive aspects, the therapist's own feelings and emotional responses equip him to function as a superbly sensitive clinical instrument which far surpasses any "objective" techniques for assessing the personality "state" of another person, his mood, feelings, attitudes, etc. On the negative side, this sensitivity renders him subject to distortions, faulty assessments, and personal idiosyncratic judgments which he might project onto the patient. [18]

In attempting to infer the patient's state of mind from one's own empathic response, one cannot simply assume a congruence between the two [10]. One must try to sort out the transference from the countertransference, the patient's feelings from the therapist's. In the case of the hostile feelings that surround a suicidal patient, the therapist must make a judgment about whether the hate felt by the therapist can properly be attributed to the patient's hate, either for himself or for the therapist (an indicator of suicide risk), or to the therapist's countertransference hate (which presents the risk of rejecting or exerting too much control over the patient) [8]. Thus, the critical use of subjective data is a fundamental requirement of psychodynamic therapy. Our interpretation of the case under consideration is that the clinicians perceived the patient's condition as they did in part because, in letting their subjective experience go unexamined, they did not look critically at potentially useful data.

Cognitive Factors: Heuristics

Resolving the affect-laden questions that arise in the interpretation of subjective data — e.g., "Is this hate coming from me or the patient?" — entails a process of simplification. One knows that the hate one feels can never come only from the patient or only from oneself. Yet one must answer such questions. One must adopt a simplified understanding of the situation in order to act.

Cognitive psychologists use the term heuristics to describe the simplifying strategies by which people make sense of a world that would otherwise be too complex and chaotic to grasp. In the words of Tversky and Kahneman:

. . . people rely on a limited number of heuristic principles which reduce the complex tasks of assessing probabilities and predicting values to simpler judgmental operations. In general, these heuristics are quite useful, but sometimes they lead to severe and systematic errors. [19]

The cognitive strategies are not in themselves errors. Rather, they are means of coping with ambiguous situations, emotional or otherwise. Uncertainty and the need for interpretation arise with objective as well as subjective data. One can abstract different implications, draw different conclusions, from what one sees and hears as well as from what one feels.

The tendency to rely on simplifying strategies is greater in more ambiguous situations. A person's ability to use them critically, on the other hand, is a function not only of knowledge and training, but also of personality factors such as tolerance for ambiguity [20]. Several such heuristics, as summarized in Table 2, are discussed here as they apply to the case reported by Havens. The purpose of this discussion lies in the fact that clinicians can only improve their use of simplifying strategies by becoming aware of how they use them.

TABLE 2
Heuristics Influencing the Interpretation of Subjective Data In Suicide Assessment (partial list)


Heuristic	Description	Example

Availability	Ease with which factors that might bear upon a decision are called to mind affects judgment.	Imminence of patient's suicidal feelings is repressed by therapists and is therefore not available to their judgment.
Recency	The last items in a sequential presentation assume undue importance.	Therapists relax vigilance because patient appears to have improved.
Ignoring base-rate	Concrete information (i.e., vivid or based on experience/incidents) dominates abstract information (e.g., summaries, statistics).	Therapists disregard statistical risks of suicide on the strength of observations of particular patient.
Anchoring and adjustment	Prediction is made by anchoring on a cue or value and then adjusting to allow for the circumstances of the present case.	Therapists adjust statistically derived estimate of suicide risk (e.g., one based on diagnostic categories) to take into account special characteristics of particular patient.
Selective perception: confirmation bias	People seek information consistent with their own views/hypotheses and downplay/ disregard conflicting evidence.	Patient's health, as evidenced by capacity to form relationships, obscures the increased risk of suicide in the face of loss.
Single cause	Investigator discounts the possibility of other causal explanations once one has been found.	Therapists believe they have prevented suicide simply by helping patient negotiate one major precipitating event or issue.
Attribution of causality: Internal	Explanations of behavior are biased toward personal qualities or dispositions as opposed to situational forces.	Therapists place too much faith in intact personality projected by patient at a time of situational crisis such as object loss.
Locus of control: external	People attribute the outcomes they experience primarily to chance and to the actions of others rather than to their own.	Therapists underestimate their own power to precipitate a suicide inadvertently by withdrawing themselves from the patient.
Hindsight	Plausible explanations are constructed to show that what in fact happened could easily have been predicted to happen.	Patient's suicide is characterized as inevitable in post-mortem review.

Note: Portions of this table are adapted from R.M. Hogart, Judgment and Choice: The Psychology of Decision. Chichester: John Wiley, 1980, pp. 166-170

The close link between cognitive heuristics and the affective responses they serve to interpret is especially apparent in the case of availability [21], a heuristic that equates the importance of a factor in a decision with the ease with which it is called to mind. Availability can be a function of immediate environmental cues or of recall from memory, which operates selectively in line with one's present emotions. One tends to remember events experienced in a mood similar to one's mood during recall [22]. Thus, the memories that become available to a therapist during a session with a patient are essential to empathic understanding, in that they give clues to the moods transmitted to the therapist from the patient, and in turn to the patient's memories [23]. Availability is a cognitive strategy which, if used uncritically, supports the defense of repression. What is repressed is not available and therefore, by this heuristic, is overlooked. In the case at hand one of the most important predictors of suicide, the rising pressure of the patient's suicidal feelings [24], was not available to the therapists, either through the patient's words and behavior or through the clinicians' empathy.

The patterns of judgment observed in the laboratories of experimental psychologists are also those that clinical wisdom teaches the therapist to guard against. For example, the therapists who relaxed their vigilance on the basis that, in Havens' words, "the patient appeared better" (without considering whether the vulnerability she had previously shown might reappear under stress) exhibited the recency bias [25], i.e., gave undue weight to the most recent information.

Havens notes as well that the clinicians were not alerted to the increased risk of suicide posed by the patient's diagnosis of involutional melancholia (or agitated depression). The clinicians apparently disregarded the high rate of suicide in patients with this diagnosis; instead, they responded more to the normal, healthy appearance this particular patient projected in the face of loss. To the extent that they were persuaded by their observations and intuitions about her that the statistical evidence was irrelevant to her case, they were ignoring the base-rate [26], i.e., ignoring the prevalence of suicide intrinsic to this diagnostic category. A more sophisticated and more accurate variant of this heuristic is anchoring and adjustment [19], whereby both kinds of information are taken into account. Here one would anchor on a prior or general (perhaps statistical) estimate and adjust for the particular case, e.g., "This woman is special. She is different from the average manic-depressive patient and therefore is less suicidal."

At the center of the therapists' assessment of their patient's condition lay a set of hopeful assumptions. The clinicians assumed that the period just before, rather than after, her daughter's marriage was of greatest risk for the patient. Once this crisis was past, they became less vigilant. They were also too easily reassured by the patient's having "shown the ability most gratifying to psychiatrists, the capacity to form a relationship" [9]. In attaching such importance to these two positive signs, they disregarded the fact that the patient had lost (along with — in her view — her daughter) three of the therapeutic relationships she had formed in the hospital.

That the therapists gave less weight to this loss of object ties than to other, more encouraging developments can be understood in terms of several heuristics. One is a form of selective perception known as the confirmation bias, whereby people seek information that confirms — and reject information that refutes — their own hypotheses [27]. There is also the single cause heuristic, by which the investigator, having found one cause (in this case the daughter's wedding), tends to discount the importance of any others [28]. By placing too much faith in the patient's seemingly intact personality, as registered by her bearing, poise, and self-control, the therapists showed an explanatory bias toward internal causation (i.e., personal qualities or dispositions) while discounting external causation (i.e., situational forces like losses) [29, 30]. Finally, in underestimating the effect on the patient of the loss of her relationships with clinical personnel, the therapists in effect underestimated the causal role of their own presence in her life [31]. that is, they assumed that the locus of control [32] of her state of mind and behavior lay almost entirely outside themselves. In Havens' words, ". . . we are the great placebos of our pharmacopeia, and the power of the placebo can be measured by the results of its withdrawal" [9].

Most importantly, the retrospective character of the case illustrates the seductive appeal of hindsight [33], or after-the-fact explanation. Havens' review is a critical exercise in hindsight, undertaken with the knowledge that an uncertain situation could have been understood before the fact only in terms of probabilities. When used uncritically, on the other hand, hindsight expresses the belief that what did happen could easily have been predicted to happen. Clinical hindsight is exemplified by Light's description of the typical presentation by the therapist at a suicide review, in which "much evidence points to the inevitability of suicide, much more so than presentations made by the same therapist about the same patient at conferences while he was living" [34]. Legal hindsight is the belief — rejected by the courts [35] but still at the root of some malpractice claims — that a tragic outcome is prima facie evidence of negligence. There is also a subtler misuse of hindsight that both clinicians and the law should avoid, one that holds an error in judgment resulting from the use of simplifying strategies such as those discussed here to be an unacceptable deviation from scientific objectivity.

Is an Escape to Objectivity Possible?

We have seen that the information-processing strategies that people habitually use to come to terms with subjective data are prone to error, and that judgment may also be biased or made unreliable by the decision maker's emotional responses. Can we, then, bypass subjective data in the assessment of suicide risk? Clearly not, for as Kohut [36] emphasizes, introspection and empathy are essential tools needed to observe and understand any psychological fact. Even if we could undertake a suicide assessment by using only objective data, we could not escape the uncertainties inherent in the use of simplifying strategies. For the strange fact is that simplifying strategies of inconsistent reliability are used in the assessment of objective as well as subjective data - for example, in drug monitoring [37]. In fact, some of the classic experimental studies of cognitive heuristics and the biases they introduce have involved seemingly cut-and-dried statistical relationships such as gambles with numerical payoffs [38].

Consider a type of objective data that figures in the clinician's estimate of suicide risk — statistical generalizations. For example, before any individual factors are taken into consideration, a manic-depressive patient may be assumed, on the basis of statistical frequency, to have a 15 percent risk of death by suicide. Even to make the diagnosis, however, the clinician must make a subjective "similarity" judgment, which itself involves the use of cognitive heuristics [39]. In other words, the use of statistical categories for risk assessment only transfers the exercise of subjectivity from the assessment of risk to the assessment of what categories the patient fits into. To take one illustration, in making the necessary similarity judgments it makes a difference whether one asks, "How similar is this patient to those whose characteristics make up the patient profile for this category?" (thus emphasizing the common features) or "How different is this patient from those whose characteristics make up the profile?" (thus emphasizing the discrepant features) [40]. Not only does the use of objective as well as subjective data bring in cognitive simplifying strategies; it is also accompanied by affective responses that demand critical interpretation. Judgments of similarity may be influenced, for example, by the wish to master an unfamiliar feeling that is thought to be at the heart of the deja vu phenomenon [41].

Given that subjective data cannot be excluded from consideration, one might still seek to exclude subjective judgment by feeding the data into an objective decision-making procedure such as formal decision analysis [37]. This wish for objectivity "once removed" can be heard in Kaplan et al.'s recommendation of computer-generated assessments of suicide risk as a model for rational assessment by the clinician:

Since the computer's management decisions were based on a rigid and obviously nonintuitive decision rule, the extent of agreement between individual raters and the computers is an estimate of how systematic the human decisions were . . . [i.e.,] of the actual level of rationality. [24]

Critical studies of decision analysis have shown, however, that subjective judgment (involving the same simplifying strategies used in informal decision making, such as judgments of similarity) makes itself felt at every stage of the process - namely, the structuring of the decision tree, estimation of probabilities, and assessment of values [6, 42, 43]. Moreover, the manner in which decision analysis is presented to the patient may either increase or reduce the empathy available, heuristically and therapeutically, as part of the therapeutic alliance [44]. As with any clinical tool, the empathic use of decision analysis requires subjective judgment.

To exercise subjectivity - and the simplifying strategies associated with it - from clinical reasoning would be to destroy much of what makes human judgment possible and clinical judgment therapeutic [36]. Whether the data under examination are subjective or objective, whether the judgments to be made are clinical or actuarial, the characteristic patterns of thought and feeling by which people make sense of the world inevitably come into play. The clinician must learn to use them wisely, both to minimize the errors and to maximize the discoveries to which they can lead.

How the Informed Consent Requirement Promotes Critical Use of Subjective Data

Given the right tools, the clinician can make use of heuristics in such a way that they reveal rather than deceive. But what are the right tools? What is the best framework for focusing the simplifying strategies of clinical judgment so that they yield a sound assessment? As table 3 shows, clinical reasoning consistent with the three judicial standards of due care itself provides a framework that might lead to more critical use of heuristics in cases such as that reported by Havens [9] and further analyzed above. Informed consent [7] also assumes strategic importance in improving clinical decision making. The process of attempting to obtain informed consent can serve not only to meet a legal requirement, but also to bring to consciousness, organize, and deploy the therapist's perceptions and judgments.

TABLE 3
Use of Legal Standards as Correctives to Clinical Heuristics


Heuristic	Example	Applicable Legal Standard(s)* and Example of Clinical Use

Availability	Imminence of patient's suicidal feelings is repressed by therapists and is therefore not available to their judgment.	#1: Ask if other clinicians would make the same judgment.
Recency	Therapists relax vigilance because patient appears to have improved.	#1: Ask if others who have seen the patient at a different time and would make the same judgment.
Ignoring base-rate	Therapists disregard statistical risk of suicide on the strength of observations of particular patient.	#2: Construct probability estimates as precisely as possible according to data in literature. Think of the consequences of Type I and Type II errors.
Anchoring and adjustment	Therapists adjust statistically derived estimate of suicide risk (e.g., one based on diagnostic categories) to take into account special characteristics of particular patient.	#2: Use this heuristic to full advantage by bringing to bear relevant objective and subjective data.
Single cause	Therapists believe they have prevented suicide simply by helping patient negotiate one major precipitating event or issue.	#3: Consider multiple causes and changing causal factors in patient's life situation and therapeutic milieu.
Selective perception: confirmation bias	Patient's health, as evidenced by capacity to form relationships, obscures increased risk of suicide in the face of loss.	#1: Seek other opinions. #3: Consider multiple causes.
Attribution of causality: internal	Therapists place too much faith in intact personality projected by patient at a time of situational crisis such as object loss.	#3: Consider multiple causes and changing causal factors in patient's life situation and therapeutic milieu.
Locus of control: external	Therapists underestimate their own power to precipitate a suicide inadvertently by withdrawing themselves from the patient.	#3: Take into account the effects of therapists observing presence on the patient when attempting to foresee the patient's state in the therapist's absence.
Hindsight	Patient's suicide is characterized as inevitable in post-mortem review.	#1,2: Recall original basis of erroneous judgments in terms of standards of profession and cost-benefit analysis. #3: Recall that all things involve some uncertainty before the fact.

* #1 - community standards
#2 - benefit/cost maximization
#3 - reasonable and prudent practitioner

The attempt to obtain informed consent is usually treated as if it were separate from the clinical determination itself. One decides what is best for the patient, and then one seeks the patient's consent. Actually, the full therapeutic value of informed consent can be realized only when this procedure is seen as contributing to the thorough clinical assessment achieved through continuous monitoring [45]. Continuous monitoring makes possible the involvement and interchange necessary for feedback and correction (both of the therapist's impressions of the patient and of the patient's impressions of the therapist). One monitors not only the patient, but oneself — one's counter-transference, one's empathy, one's effects on the interaction of character and life situation that constitutes the patient's present state of being. One's interchange with the patient over informed consent then becomes both a tool for monitoring and an ingredient of the therapeutic alliance being monitored.

As indicated in table 4, informed consent not only protects the patient's right to choose and offers the therapist legal protection as a defense against negligence (as discussed in Part I of this paper), but also has three major therapeutic benefits. First, the informed consent process in itself can have therapeutic value. Whatever transpires between patient and therapist, including the gathering of both objective and subjective data, has an impact, for good or for ill, on the therapeutic alliance and with it the patient's well being. It has been observed, for example, that engaging in a dialogue with the patient over a procedure such as petitioning for commitment can change the therapeutic alliance even as it illuminates it [46]. The effect can be a positive one if sensitive attention is given to the manner in which the subjective data needed for informed consent are obtained. Questions must be asked empathically [9, 36, 45], rather than in a way that only elicits repeated false reassurances. Options likewise must be presented with care, in language such as the following: "Some people, when they are depressed and suffering, feel that life isn't worth living. Right now it may feel even to you that there is no hope of ending your suffering other than by ending your life. Of course, this depression will lift, so in the meantime, how can we best work together to decrease the risk of suicide?" This approach can have the therapeutic effect of distancing the notion of suicide, making the suicide ego-alien, even as it begins to establish a "situational" alliance of shared observation [47]. The informed consent procedure, when carried out with this kind of involvement and mutuality, gives the patient responsibility without conveying a message of rejection, lack of caring, or therapeutic passivity [48].

TABLE 4
Clinical Uses of the Informed Consent Procedure In Suicide Assessment


Benefits for Patient	Benefits for Therapist	Benefits for Therapeutic Alliance

Safeguards right to informed choice of treatment options Gives responsibility while demonstrating therapeutic involvement Facilitates conscious reinterpretation of data and reconsideration of costs and benefits of options	Provides legal protection as a defense against negligence (when documented) Yields access to relevant diagnostic information: Patient's affects, cognitions, and competence to make informed choice Therapist's affects and cognitions (countertransference) State of therapeutic alliance Facilitates conscious reinterpretation of data and reconsideration of costs and benefits of options	Builds alliance by providing a context for mutual exchange and shared effort

Second, the alliance thus created (or strengthened) can be one of the best sources of information about the risk of suicide. Whether or not the patient can follow the language of risks and benefits is a test of the patient's competence to give informed consent, which in turn is data for the assessment of risk. (One would petition for commitment when a patient is both suicidal and incompetent.) The process yields other diagnostic information as well. By observing how the patient deals with concepts such as costs, benefits, gambles, and consequences, one can identify not only the affective indicators of suicidal risk — helplessness, hopelessness, worthlessness, and aloneness [49, 50] — but other indicators as well, such as diminished future orientation [51], a sense of time passing slowly [52], excessive risktaking [53], and cognitive rigidity [54].

One should keep in mind that the patient has cognitions as well as affects. The exchange over informed consent opens a window to the patient's ways of viewing the world (e.g., the cognitive biases that Beck [54] has shown to be associated with depression), which in turn both reflect and structure the patient's experience of living. Likewise, the clinician has affects as well as cognitions. The subjective data to which the clinician attends in monitoring the therapeutic alliance and the informed consent process include the clinician's responses as well as the patient's. Havens lists several signs (in addition to the depressive triad cited earlier [9]) of such affective responses on the part of the clinical observer, beginning with the feeling of absence of communication that is known as Bleuler's sign of schizophrenia:

Awareness of praecox gefühl, the eerie feeling schizophrenic people convey, is old. There has been backstairs talk for years of what we call in Boston Hendrick's sign: what young female hysterics convey to the doctor's penis. But the number of distinct observer effects are few. Obsessional patients make us yawn. I named another Sullivan's sign from the following anecdote: Sullivan is rumored to have said, "I can detect when I am in the presence of a strongly homosexual person by a tightening of the anal sphincter." ... Finally, I like to call Havens' sign what the praecox gefühl does to many: Like a horror movie the young schizophrenic's account of derealization or of deja vu or of strange bodily feelings produces pilo-erection; one can feel the small hairs on one's neck go up. [55]

These data must, however, be interpreted critically. No one emotion has reliable clinical significance except in the context of other feelings and other kinds of data. In the case of the patient's responses one must not necessarily believe what one hears, but must ask oneself if it makes sense, or if there is some contradiction between statement and affect, between statement and life situation. One must consider whether there are personality or situational factors that make withholding of the truth a likely possibility. In the case of the clinician's attitude toward the patient, which can be one of the most sensitive indicators of the seriousness of suicidal risk [56], one must likewise be critical.

To take one example, people tend to be attracted to others who are like themselves [57]. One must therefore consider whether one's feeling of dislike for a patient signifies that the patient is distant, rejecting, and painful or that the patient simply is unlike oneself. If the former, is this affect representative of the aversion the patient creates in relationships with others? [58] If so, then this should appropriately be interpreted as decreasing the likelihood that there is anyone who cares for the patient - a subjective datum that indicates the patient to be at high suicide risk [59]. If, on the other hand, one finds the patient likable, is it really a matter of likability (to others as well as oneself) - suggesting that the patient may have "anchors to life" - or is it more a matter of likeness (so that outside of a "narcissistic alliance" [47] with the therapist, the patient is actually all alone)? Or is it yet another situation: the patient simply trying to please? [31]

Questions such as these cannot be resolved definitively, but they are valuable heuristics for thinking critically about the subjective data on whose interpretation the assessment of suicide risk turns. It is here, in making conscious and rational use of the simplifying strategies of judgment, that the third clinical advantage of the informed consent process comes into play. Informed consent not only gives access to data (i.e., the patient's and the clinician's reasoning and emotions) which, properly interpreted, make the assessment less prone to distortion; it also gives the clinician and the patient an opportunity to think out loud about how data translate into costs and benefits of therapeutic options. In the informed consent discussion, where one attempts to justify a particular course of action to the patient, one can give conscious attention to one's reasoning strategies and test whether the patient can similarly attend to his or her own heuristics (which, when suicide is at issue, are at least as important as the clinician's). When the patient is capable of joining in such an examination, the therapeutic alliance becomes a mutual effort to achieve rationality. The latter would appear to offer the closest possible approximation to freedom from cognitive and affective bias while incorporating both subjective and objective data in therapeutic decisions.

Applying these principles specifically to Havens' case of the woman who committed suicide while home on a weekend pass from the hospital, we can consider (admittedly from hindsight) how the use of informed consent might have helped prevent this tragic outcome. As it was, the woman's therapists found it all too easy to approve the pass, both because they were giving the patient what she wanted and because (by Havens' account) they may have had unexamined reasons for wanting her out of their way. The informed consent procedure might have served here as a valuable safeguard in several ways. In the first place, the patient's ability to consider whether there might be some risk in her going home would itself have been a significant clinical datum. By being asked this question, the patient also would have been afforded the opportunity to realize and communicate the despair she actually felt.

In addition, examination of this issue might have focused the clinicians' attention on their own subjective reactions, both as impediments to good judgment on their part and as possible evidence of the patient's unlikability and consequent isolation. By putting a critical check on their unconscious rejection of the patient, the clinicians might have been able to act toward her in a more therapeutic manner. Finally, by demonstrating that they were concerned enough about her well being to question the wisdom of her leaving the hospital, they might have strengthened the therapeutic alliance and thereby dissuaded her from ending her life.

Conclusion

The dilemmas of suicide assessment demonstrate forcefully and poignantly that no information is completely objective or completely subjective. The two kinds of data are inextricably linked, and both must be approached with respect for their inherent uncertainties. For these reasons, the characterization of the clinician's use of subjective data as unscientific and therefore negligent is, by today's standards, both scientific and legal, unfounded.

Neither the ultimate unpredictability of suicide nor the range of affective and cognitive influences on psychiatric decision making can be done away with. Nonetheless, the psychiatrist is responsible for making wise use of both objective and subjective data in order to give all relevant clinical factors their proper weight in a risk-benefit analysis. The creation of a therapeutic alliance by involving the patient in informed choices, far from being a clinically irrelevant legal requirement, is a precondition for a sound and thorough assessment of this sort. Similarly, the awareness that one's clinical reasoning must be supportable by the three standards of due care in negligence law can guide the clinician in finding the right questions to ask.

These recommendations for making clinical use of legal requirements are in keeping with the spirit of the legal tradition, i.e., its effort to temper the distortions of intuitive judgment, where heuristics fueled by passion can become biases. Legal principle, like those gleaned from long clinical experience, introduce an element of deliberateness enabling heuristics to be used to simplify complex, highly charged diagnostic and prognostic judgments in the service of better clinical care.

* In posing these questions the clinician should be aware of the distinction between Type I statistical error (rejecting a hypothesis which is true) and Type II error (accepting a hypothesis which is false). To fail to diagnose and act on a serious suicidal risk is a Type I error. To treat a patient who does not in fact present such as risk as if he or she did is a Type II error. Medical and legal concerns, together with ordinary superstition ("Better safe than sorry!"), predispose us to seek to avoid any Type I errors even at the cost of numerous Type II errors. Nonetheless, although the gravity of the risk of Type I error should be kept in mind, the choice between risking a Type I or Type II error is, as the questions above suggest, too complex to be resolved by any one maxim [4].

Reference Notes

Epstein, R. A. The limits of medical malpractice. New England Journal of Medicine, 1978, 298, 1311-1312.
Helling v. Carey (1974), 83 Wash. 2d 514, 519 P. 2d 981.
Schwartz, W. B., & Komesar, N. K. Doctors, damages and deterrence: an economic view of medical malpractice. New England Journal of Medicine, 1978, 298, 1289.
Scheff, T. J. Decisions in medicine. In Being mentally ill: A sociological theory. Chicago: Aldine, 1966.
Fiederlein v. City of New York Health and Hospitals Corporation (1981), 80 App. Div. 2d 821, 437 N.Y. 2d 321.
Bursztajn, H., Feinbloom, R. I, Hamm, R. M., & Brodsky, A. Medical choices, medical chances: How patients, families, and physicians can cope with uncertainty. New York: Delacorte/Seymour Lawrence, 1981.
Gutheil, T. G., & Appelbaum, P. S. Clinical handbook of psychiatry and the law. New York: McGraw-Hill, 1982.
Maltsberger, J. T., & Buie, D. H., Countertransference hate in the treatment of suicidal patients. Archives of General Psychiatry, 1974, 30, 625-633.
Havens, L. L. The anatomy of a suicide. New England Journal of Medicine, 1965, 272, 401-406.
Buie, D. H. Empathy: its nature and limitations. J Amer Psychoanal Assn, 1981, 29, 281-307.
Winnicott, D. W. Countertransference. Brit J Med Psychol, 1960, 33, 17-21.
Kernberg, O. F. Countertransference. J Amer Psychoanal Assn, 1965, 13, 38-56.
Beres, D., & Arlow, J. A. Fantasy and identification in empathy. Psychoanalysis Quarterly, 1974, 43, 26-50.
Simon, H. A. Rational choice and the structure of the environment. Psychology Review, 1956, 63, 129-138.
Hogarth, R. M. Judgment and choice: The psychology of decision. Chichester: John Wiley, 1980.
Kahneman, D., Slovic, P., & Tversky, A. (Eds.). Judgment under uncertainty: Heuristics and biases. Cambridge: Cambridge University Press, 1982.
Winnicott, D. W. Hate in the countertransference. In Through paediatrics to psycho-analysis. New York: Basic Books, 1975.
Strupp, H. H., & Wallach, M. S., A further study of psychiatrists' responses in quasi-therapy situations. Behavioral Science, 1965, 10, 113-134.
Tversky, A., &. Kahneman, D. Judgment under uncertainty: heuristics and biases. In D. Kahneman, P. Slovic & A. Tversky (Eds.), Judgment under uncertainty: heuristics and biases. Cambridge: Cambridge University Press, 1982.
Budner, S. Intolerance of ambiguity as a personality variable. J Pers, 1962, 30, 29-50.
Tversky, A., & Kahneman, D. Availability: a heuristic for judging frequency and probability. Cognitive Psychology, 1973. 5, 207-232.
Bower, G. H. Mood and memory. American Psychologist, 1981. 36, 129-148.
Day, M. Counter-transference in everyday practice. In Issues in psychotherapy. Vol. 1: Counter-transference. Boston: Boston Institute for Psychotherapies, 1981.
Kaplan, R. D., Kottler, D. B., & Frances, A. J. Reliability and rationality in the prediction of suicide. Hospital Community Psychiatry, 1982, 33, 212-215.
Slovic, P., & Lichtenstein, S. Comparison of Bayesian and regression approaches to the study of information processing in judgment. Organic Behavior and Human Performance, 1971, 6, 649-744.
Tversky, A., & Kahneman, D. Causal schemas in judgments under uncertainty. In M. Fishbein (Ed.), Progress in social psychology. Hillsdale, NJ: Erlbaum, 1980.
Wason, P. C. On the failure to eliminate hypotheses in a conceptual task. Quarterly Journal of Experimental Psychology, 1960, 12, 129-140.
Shaklee, H., & Fischhoff, B. Strategies of information search in causal analysis. Decision Research Report 79-1, 1979.
Shweder, R. A. Likeness and likelihood in everyday thought: magical thinking and everyday judgments about personality. In P. N. Johnson-Laird & P. C. Wason (Eds.), Thinking: readings in cognitive science. Cambridge: Cambridge University Press, 1977.
Nisbett, R. E., & Ross, L. Human inference: Strategies and short-comings of social judgement. Englewood Cliffs, NJ: Prentice-Hall, 1980.
Ferenczi, S. Confusion of tongues between the adult and the child. International Journal of Psychoanalysis, 1949, 30, 225-230.
Rotter, J. Generalized expectancies for internal versus external control of reinforcement. Psychological Monographs, 1966, 80, Whole no. 609.
Fischhoff, B. Hindsight ≠ foresight: the effect of outcome knowledge on judgment under uncertainty. Journal of Experimental Psychology: Human Perception and Performance, 1975, 1, 288-299.
Light, D. Becoming psychiatrists: the professional transformation of self. New York: Norton, 1980.
Staloch v. Holm (Minn. 1907), 111 N.W. 264.
Kohut, H. Introspection, empathy, and psychoanalysis: an examination of the relationship between mode of observation and theory. Journal of the American Psychoanalytical Association, 1959, 7, 459-483.
Hogarth, R. M. Judgment, drug monitoring and decision aids. In W. H. W. Inman (Ed.), Monitoring for drug safety. Lancaster, England: MTP Press, 1980.
Tversky, A., & Kahneman, D. The framing of decisions and the psychology of choice. Science, 1981, 211, 453-458.
Glass, A. L., Holyoak, K. J., & Santa, J. L. Cognition. Reading, MA: Addison-Wesley, 1979.
Tversky, A., & Gati, I., Studies of similarity. In E. Rosen & B. B. Lloyd (Eds.), Cognition and categorization. Hillsdale, NJ: Erlbaum, 1978.
Arlow, J. A. The structure of the deja vu experience. J Amer Psychoanal Assn, 1959, 7, 611 -631.
Tribe, L. H. Trial by mathematics: precision and ritual in the legal process. Harvard Law Review, 1971,84, 1329-1393.
Tribe, L. H. Policy science: analysis or ideology. Phil Public Affairs, 1972, 2, 66-110.
Bursztajn, H., & Hamm, R. M. The clinical utility of utility assessment. Medical Decision Making, 1982, 2, 161-165.
Shein, H. M., & Stone, A. A. Psychotherapy designed to detect and treat suicidal potential. American Journal of Psychiatry, 1969, 125, 1247-1251.
Eisenberg, G. C, Barnes, B. M., & Gutheil, T. G. Involuntary commitment and the treatment process: a clinical perspective. Bulletin of the American Academy of Psychology and the Law, 1980, 8, 44-55.
Gutheil, T. G., & Havens, L. L. The therapeutic alliance: contemporary meanings and confusions. International Review of Psycho-Analysis, 1979, 6, 467-481.
Olin, H. S. Psychotherapy of the chronically suicidal patient. American Journal of Psychotherapy, 1976, 30, 570-575.
Leonard, C. V. Self-ratings of alienation in suicidal patients. Journal of Clinical Psychology, 1973, 29, 423-428.
Carter, C. A. Some conditions predictive of suicide at termination of psychotherapy. Psychotherapy: Theory, Research and Practice, 1971,5, 156-157.
Yufit. R. I., & Benzies, B. Assessing suicidal potential by time perspective. Life-Threatening Behavior, 1973, 3, 270-282.
Neuringer, C., & Levenson, M. Time perception in suicidal individuals. Omega: Journal of Death & Dying, 1972, 3, 181-186.
Adams, R. L., Giffen, M. B., & Garfield, F. Risk taking among suicide attempters. Journal of Abnormal Psychology, 1973, 82, 262-267.
Beck, A. T. Cognitive therapy and emotional disorders. New York: International Universities Press, 1976.
Havens, L. L. Approaches to the mind. Boston: Little, Brown, 1973.
Dressier, D. M., Prusoff, B., Mark, H., & Shapiro, D. Clinician attitudes toward the suicide attempter. Journal of Nervous and Mental Disorders, 1975, 160, 146-155.
Berscheid, E., & Walster, E. Interpersonal attraction. Reading, MA: Addison-Wesley, 1969.
Moras, K., & Strupp, H. H. Pretherapy interpersonal relations, patients' alliance and outcome in brief therapy. Archives of General Psychiatry, 1982, 39, 405-409.
Motto, J. A., Heilbron, D. C, Juster, R. P., & Bostrom, A. G. Suicide risk assessment: development of a clinical instrument. In Proceedings, Fourteenth Annual Meeting, American Association of Suicidology, Albuquerque, New Mexico, 1981.

Subjective Data and Suicide Assessment in the Light of Recent Legal Developments Part II: Clinical Uses of Legal Standards in the Interpretation of Subjective Data