Accredited Psychiatry & Medicine | Harold J. Bursztajn, MD, Forensic Psychiatrist | Parens Patriae Considerations in the Commitment Process

Parens Patriae Considerations in the Commitment Process

Harold Bursztajn, M.D.*
Thomas G. Gutheil, M.D.*
Robert M. Hamm, Ph.D.**
Archie Brodsky, B.A.*
Mark J. Mills, J.D., M.D.***

*Program in Psychiatry and the Law, Massachusetts Mental Health Center, 74 Fenwood Road, Boston, MA 02115.
**Department of Management Sciences, University of Iowa, Iowa City, IA 52242.
***Department of Psychiatry, School of Medicine, University of California at Los Angeles, Los Angeles, CA.

From the Program in Psychiatry and the Law, Department of Psychiatry, Massachusetts Mental Health Center, 74 Fenwood Rd., Boston, MA 02115. Address requests for reprints to Dr. Bursztajn. Supported in part by Grant #5T01-MH-16460-03 from the National Institutes of Mental Health.
Thanks to Mr. Norm House for his help with the data analysis, and to Ms. Audrey Bleakley and Ms. Ingrid Young for their help in the preparation of this manuscript.

Abstract: Seventy-one commitment-seeking decisions made by 36 clinicians in a state-funded mental health center were studied to determine the extent to which clinicians attend to legally mandated "dangerousness" criteria in seeking commitment. A previous finding that clinicians rely largely on the dangerousness criteria was replicated. In addition, clinicians were found to be sensitive to clinical indicators of the patient's need for treatment, a question which is central to the parens patriae approach to involuntary hospitalization. Further, patients who were judged to be more seriously ill or more dangerous were more likely to retract their requests for discharge. This finding suggests that the patient's request for release and the psychiatrist's petition for commitment together constitute an interactive, transactional process in which the clinician's and the patient's views of the patient's need for hospitalization influence each other.

Introduction

Previous empirical studies of the psychiatrist's decision to petition for civil commitment have shown that psychiatrists are appropriately attentive to the "dangerousness" criteria established by law for involuntary commitment. [1,2] These criteria are based on a "police powers" or public safety rationale, by which potentially violent individuals who are mentally ill may be preventively detained in the hospital. [3] Some psychiatrists, however, would prefer a modification of the older parens patriae model, by which need for treatment is a primary factor in the decision to hospitalize a patient involuntarily. [4] This preference is reflected in the American Psychiatric Association's model state statute on civil commitment. [5,6,7] Moreover, a preliminary empirical study of judges' commitment decisions indicates that judges, in interpreting the statutory requirements for commitment, are not oblivious to the clinically and ethically salient consideration of need for treatment. [8] Presumably, the same clinically-based viewpoint should be expected to figure, along with the statutory requirements, in the clinician's thinking as well.

The present study includes a replication of a prior empirical investigation of the decision to seek commitment by Appelbaum and Hamm, [1] with an added focus on the parens patriae aspects of commitment as seen from both the clinician's and patient's perspectives. The earlier study was conducted at the Massachusetts Mental Health Center in Boston in 1979 and 1980; the current study is based on new data collected during the period from 1980 to 1983 at the same institution. In Massachusetts the statutory requirements for commitment are met when the patient exhibits any of the following characteristics in conjunction with serious mental illness: dangerousness to self, dangerousness to others, or inability to care for self. [9] Beyond the ten-day emergency commitment which can be ordered solely by a psychiatrist, commitment becomes a two-step process involving both the clinician and a judge: the psychiatrist does or does not petition the court; if petitioned, the judge does or does not commit. For example, voluntarily hospitalized patients must give three days' written notice (a "three-day paper") signifying an intention to leave the hospital against medical advice. The patient's psychiatrist then has three days to decide whether or not to petition the court for a commitment lasting up to six months.

The legally mandated three-day wait between the patient's application for release and the psychiatrist's decision to seek or not to seek commitment offers an ideal "natural laboratory" for studying how parens patriae works in practice. As in previous studies, [1,2] correlations between the commitment decision and clinicians' descriptive ratings of patient characteristics (clinico-legal, psychosocial, interpersonal) give an indication of the ground on which the decision to seek commitment is made. For this study, however, the questionnaire has been expanded to include more clinical dimensions (both diagnostic and predictive), so that the influence of "need for treatment" considerations on the clinician's decision can be more thoroughly explored.

The "three-day paper" situation also shed light on the patient's reaction to the clinician's intervention. Parens patriae has been called the "thank you" theory of civil commitment, the assumption being that a patient who is temporarily incompetent to make life-sustaining choices will, upon recovery, be grateful that the clinician stepped in temporarily to override the patient's destructive or self-destructive impulses. [4] A patient who has filed a three-day paper can also respond positively to the clinician (whether for petitioning for commitment or for other therapeutic measures or expressions of concern) by retracting the formal request for release before the three days expire; that is, retraction of the three-day paper communicates that the patient now agrees with the clinician that the patient's best interests are served by continued hospitalization. But just which patients change their minds in this way? Correlations between the patient's decision to retract and the clinician's decision to petition, as well as between the decision to retract and the descriptive ratings of patient characteristics, should tell us something about the other side of the parens patriae model — elucidating what determines the patient's (perhaps grateful, perhaps compliant) change of heart in response to the clinician's attempts to act in the patient's best interest.

Method

Questionnaires were given to clinicians faced with formal requests for discharge filed by voluntary patients at Massachusetts Mental Health Center between November 1980 and August 1983. To avoid biasing the data base with repeated filings by the same patients, only the first such request by a given patient in a given hospitalization was used in the study. In each instance the questionnaire was given to the patient's treating clinician, in most cases a resident in psychiatry. In a few cases, however, the questionnaire was given to a clinician of another discipline (e.g., psychology intern) who was more directly involved with the patient's day-to-day care. Respondents were instructed to wait to fill out the questionnaire until the commitment decision had been made, which was typically at the end of the three-day period. If the patient retracted the request for discharge, the clinician was to complete the questionnaire immediately. Although the clinician's intention is not measured until the decision is made, during most of the three-day waiting period clinicians usually have a good idea of what they will do and they communicate it in their negotiations with patients.

In addition to some demographic questions about the patient, the questionnaire asked respondents whether or not they had decided to petition for commitment and whether or not the patient had retracted the request for discharge. In cases where the patient retracted, the respondents were asked what decision they would have made had a decision been required. The respondent was then asked to rate the patient from 1 to 7 on each of 23 statutory, clinical, psychological, interpersonal, and miscellaneous variables having potential relevance to the decision to seek commitment. The clinical variables used in this study included: the patient's Need for Treatment, Psychotic Thought, Psychotic Behavior, and Wellness versus Illness. A further measure on the instrument, a rating of each variable's impact on the commitment-seeking decision, will be reported elsewhere. A total of 74 questionnaires were returned by 36 respondents, with each participating clinician returning from 1 to 5 questionnaires. Of the questionnaires returned, 73 indicated whether the clinician sought (or would have sought) commitment. Of these, 71 indicated whether the patient retracted the request for release.

In the analysis of the data the clinician was considered to have made the decision to seek commitment if (a) the patient did not retract the three-day paper and the clinician initiated the commitment process, or (b) the patient retracted the three-day paper and the clinician would have sought commitment had the patient not retracted. Similarly, the clinician was considered to have made the decision not to seek commitment if (a) the patient did not retract the three-day paper and the clinician allowed the patient to be discharged, or (b) the patient retracted the three-day paper but the clinician would have allowed the patient to be discharged had the patient not retracted. Table 1 shows the number of patients in each of these categories.

Table 1

The Relation Between the Clinicians' Commitment Seeking Decision and the Patients' Retractions

	Patients who retracted	Patients who did not retract	Total
Patients for whom clinician sought or would have sought commitment	40	5	45
Patients for whom clinicians did not seek or would not have sought commitment	16	10	26
Total	56	15	71

Correlations were obtained between the decision to seek commitment and the clinician's ratings of the patient on each of the 23 characteristics. Similarly, correlations were obtained between the ratings of patient characteristics and the patient's decision to retract, both for the entire sample and separately for the groups of patients for whom the clinicians did and did not choose to seek commitment. In addition, with respect to the clinician's commitment decision, multiple linear regression [10] was performed to identify those patient characteristics whose unique contribution to the decision variance was significant. This method shows the proportion of the variance that was attributable to a given characteristic in the context of other relevant characteristics, in this case the three legally mandated criteria for civil commitment. In this way the independent contribution to the decision made by each of the other 20 patient characteristics was isolated from that which resulted from simple covariance with the three "judicial" criteria.

Results

Relationship between clinician's decision to seek commitment and patient's decision to retract three-day paper

Table 1 shows the number of patients who retracted their three-day papers and the number of clinicians who sought commitment for their patient or would have done so had the patient not retracted. Among the 71 completely filled out questionnaires, only 15 of the patients (21.1%) stood by their requests to leave. For five of these (33.3%) the clinician sought commitment. Fifty-six patients (78.9%) retracted their three-day papers, and the clinician responsible would have sought commitment for 40 of them (71.49%) had the patient not retracted. The patient's retraction decision and the clinician's decision to seek commitment were significantly related: if the clinician was likely to petition to commit the patient at the end of the three-day period, the patient was likely to retract the request to be released before the end of the period (chi-squared = 5.85, df = 1, p < .02).

Relationship between clinician's decision to seek commitment and various patient characteristics

Table 2 shows the relationship between the clinician's decision whether to seek to commit the patient and each of the 23 patient characteristics. The characteristics are grouped into major categories — the three judicial dangerousness criteria (Dangerousness to Others, Danger to Self, and Unable to Care for Self), diagnostic and predictive clinical judgments of the patient's psychosocial situation, and judgments of the patient's effect on others.

TABLE 2

Relation Between Clinician's Decision to Seek Commitment and the Patient Characteristics


		Partition of Variance (b)

Patient Characteristic	Correlation (a)	Unique (extrajudicial)	Common	Unique (judicial)

Judicial Variables
Danger to Others	.40***
Danger to Self	.35***
Able to Care for Self	−.45***
Extrajudical Variables
Clinical Judgments (Diagnostic)
Acute	−.03	.001	.000	.389***
Needs Treatment	.34**	.048*	.068	.321***
Ill	.52***	.058**	.208	.181***
Psychotic Thought	.49***	.033†	.203	.185***
Psychotic Behavior	.61***	.110***	.265	.125***
Clinical Judgments (Predictive)
Reliable Outpatient	−.29**	.005	.081	.309***
Reliable with Meds	−.16†	.006	.021	.372***
Psychosocial Judgments
Place to Live	−.35***	.029†	.093	.296***
Support on Outside	−.15	.009	.013	.376***
Able to Work	−.22*	.000	.050	.339***
Poor	.04	.000	.002	.387***
Family Favors Commitment	.18†	.003	.029	.384***
Interpersonal Factor Judgments
Likeable	.07	.015	−.010	.399***
Discouraging	−.03	.012	−.011	.401***
Frightening	.38***	.016	.128	.261***
Seductive	.14	.018	.002	.387***
Depressing	.13	.001	.016	.372***
Disruptive on Ward	.12	.000	.013	.376***
Miscellaneous Judgments
Well Known to Doctor	.04	.012	−.010	.400***
Of Academic Interest	.16	.032†	−.006	.395***

a. Correlation between clinicians' decisions whether to seek commitment and clinicians' ratings of each patient characteristic.
b. Proportion of variance of clinicians' decisions that is explained uniquely by each extrajudicial variable in turn, uniquely by the judicial variables, or in common by both.
N = 73 except for "Reliable with Medications," N = 70, and "Family Favors Commitment," N = 68.
† = p < .10; * = p < .05; ** = p < .01; *** = p < .001; 1-tailed.

Column 1 shows the correlations between the patient characteristics and the clinician's decision to seek commitment. All of the patient characteristics that were correlated with the decision at the p < .10 level or less had the direction of relationship that would be expected. This was also found when correlations were computed separately for patients who did and did not retract their three-day papers (data not shown). For example, the clinician was more likely to seek commitment when the patient was more dangerous to others (r = .40) or to self (r = .35) and when the patient needed treatment (r = .34). The judicial variables and the diagnostic clinical judgments in particular had strong relations to the clinician's decision.

Column 2 shows the unique additional contribution that each extrajudicial variable makes to the explanation of the clinician's decision variance, over and above what is accounted for by the three legally mandated judicial variables. (Each result is derived from regressing the decision variable onto four patient characteristic variables: the three judicial variables plus the extrajudicial variable in question.) Only the diagnostic clinical variables (specifically Needs Treatment, Ill, and Psychotic Behavior) make significant (p < .05) contributions. A number of patient characteristics (such as Frightening) are significantly correlated with the decision to seek commitment, yet make no unique contribution beyond that of the three judicial criteria. The contribution of Frightening can be accounted for most parsimoniously by saying that the decision to seek commitment is related to the judicial (dangerousness) variables, which are in turn correlated with Frightening. The other variables whose unique contributions are insignificant, although their correlations with the clinician's decision are significant, can be explained similarly.

The complementary question is also of interest: what unique contribution do the three judicial patient characteristics make to the clinician's decision to seek commitment, beyond what would be explained by each extrajudicial patient characteristic? This is important because it allows direct comparison of the extents to which the clinicians depend on the dangerousness and parens patriae factors. The proportion of variance uniquely due to the three judicial variables is shown in Column 4 of Table 2, and the common variance, i.e., the proportion explained by both the judicial and the extrajudicial variables, is shown in Column 3. In each of the four-variable regression equations, the unique contribution of each of the extrajudicial variables was less than the contribution of the three judicial variables. The mean proportion of variance uniquely due to the three judicial variables was .333, compared with the mean of .020 uniquely due to the extrajudicial variable. The clinical extrajudicial variables (parens patriae) not only have the highest unique contributions but also share the largest amounts of common variance with the judicial variables (dangerousness). The unique contribution of the Psychotic Behavior variable is nearly as large as for the dangerousness criteria, which is all the more notable given that it is a comparison of one variable against three.

Relationship between patient's decision to retract and various patient characteristics

Correlations between (a) whether the patient retracted the three-day paper before the expiration of the three days and (b) each of the patient characteristics are given in Table 3, for all cases and separately for the cases in which the clinicians would have sought commitment or would not have sought commitment.

Table 3

Relation of Patient Retraction to Patient Characteristics


	Correlations^©			Regression
Patient Characteristic	All patients N = 71	Patients for whom clinicians would seek commitment N = 26	Patients for whom clinicians wouldn't seek commitment N = 45	Interaction term^©©

Judicial Variables
Danger to Others	−.02	−.05	−.26*	−.165
Danger to Self	.18†	.09	.06	−.067
Able to Care for Self	−.20*	−.07	−.07	.084
Extrajudical Variables
Clinical Judgments (Diagnostic)
Acute	.16†	.31†	.10	−.525
Needs Treatment	.21*	.23	−.14	−.381
Ill	.17†	.17	−.21†	−1.122
Psychotic Thought	.21*	.19	−.06	−.254
Psychotic Behavior	.28**	.12	.12	−.038
Clinical Judgments (Predictive)
Reliable Outpatient	−.11	−.16	.12	.477
Reliable with Meds	−.09	−.05	−.06	.000
Psychosocial Judgments
Place to Live	.06	.06	.31*	.173
Support on Outside	.07	−.09	.31*	.577
Able to Work	.05	−.08	.36**	.621†
Poor	−.04	−.07	−.06	.050
Family Favors Commitment	.02	−.06	−.03	.136
Interpersonal Factor Judgments
Likeable	.01	−.00	−.01	.000
Discouraging	−.09	−.28†	.03	.693†
Frightening	−.02	−.14	−.20	.108
Seductive	.09	.01	.05	.038
Depressing	−.03	−.31†	.08	.665†
Disruptive on Ward	−.03	.11	−.25*	−.415
Miscellaneous Judgments
Well Known to Doctor	−.20*	−.31†	−.18	.375
Of Academic Interest	−.07	−.14	−.12	.058

t = p < .10; * = p < .05; ** = p < .01, 1-tailed.
# Correlations between whether patient retracted 3-day paper and clinicians' direct ratings of each patient characteristic, for all clinicians and separately for the cases where clinicians would seek commitment and the cases where they would not.
## Decision X Characteristic interaction term in regression of patient's retraction onto clinician's decision and each individual patient characteristic.

Looking at all cases together (Column 1 of Table 3), five of the patient characteristics were significantly related (p < .05) to the decision to retract the three-day papers: Able to Care for Self (the less able to care for self, the more likely to retract), Needs Treatment (the more needed, the more likely to retract), and finally Well-Known to clinician (the less well-known, the more likely to retract). In addition, Danger to Self, Acute (versus Chronic), and Ill (versus Well) were marginally related to the patient's decision to retract (at p < .10): the more suicidal, the more acute, and the more ill the patient, the more likely he or she would retract. Thus the overall impression is that patients are prudent, or at least cooperate with the clinician's prudence.

This "prudent patient" pattern holds especially for the more seriously dangerous or seriously ill patients, and is less true or even reversed when the patients are not so ill or dangerous. The second, third, and fourth columns of Table 3 present data that reveal this pattern, as explained in the Appendix. For example, with the Needs Treatment variable, if the clinician would seek commitment, then the more the clinician thought the patient needed treatment, the more likely the patient was to retract the request to be released (r = .23). But if the clinician would not seek commitment, then the more the patient was judged to need treatment, the less likely he or she was to retract the three-day paper (r = −.14). Although this pattern may be weak for each particular variable, it occurs for the vast majority of the characteristics (see Appendix). It suggests that the patient's tendency to retract the application for release does not depend solely on the patient's degree of need or dangerousness; it is rather an interactive function of the patient's need or dangerousness and the clinician's response. That is, when the clinician would seek to commit the patient, the more dangerous or mentally ill the patient was, the more likely the patient was to retract the request to be released from the hospital. However, when the clinician would not seek to commit the patient, i.e., when the patient was not judged to be too dangerous or too ill to be released, then the relationship between dangerousness or illness and retraction was weakened and often reversed: the less dangerous or mentally ill patients were more likely to reconsider their requests to leave the psychiatric hospital. Thus what the clinician does or would do influences what the patient does.

In summary, the clinicians sought or would have sought commitment for two-thirds of the patients, and more than three-fourths of the patients retracted their three-day papers. Although the clinician's commitment-seeking decision was significantly correlated with a number of the patient characteristics, only the characteristics related to the patient's need for treatment made a significant (p < .05) contribution to the decision variance over and above what the three judicial variables explained. As for the patients' decision making, those patients who were more dangerous and more in need of treatment were more likely to retract the three-day paper. An unanticipated pattern in the results suggests that this last relation is lessened or reversed for those patients whose clinician intended to let them go: the healthier of these patients were more likely to retract.

Discussion

First, this study replicated earlier findings [1,2] that the statutory requirements for civil commitment figure prominently in the psychiatrist's decision to seek commitment, while interpersonal factors (e.g., seductiveness) have little impact on the decision. Next, the present study investigated clinical variables pertaining to the patient's need for treatment more extensively than the earlier studies and found that clinicians make the decision to petition on the basis of parens patriae as well as dangerousness considerations. That is, they take into account the nature of the illness and the need for treatment in conjunction with (i.e., not merely in redundancy with, as in the case of some psychosocial and interpersonal variables) the legally mandated criteria. These findings are consistent with the conclusions drawn from our preliminary investigation of judges' decisions to order commitment, where evidence of parens patriae thinking was less to be expected. [8]

Finally, on the question of patients' retractions of the requests for release, our data support the generalization that the patients who most need continued hospital care are the ones who retract their petitions. Based on our clinical experience, this finding may be interpreted in several ways.

First, retraction may be a gesture of compliance on the part of patients with paranoid tendencies who would rather let the clinician take responsibility for the decision than retain that responsibility themselves and face the intervention of a judge. On the other hand, there may be a basic wisdom that asserts itself even in the mentally ill, enabling them to choose what is best for themselves. Alternatively, this "wisdom" may be due to clinicians' use of the three-day interval to bring to bear various means of persuasion (including confrontation with the possibility of court-ordered commitment) on patients who, in the clinician's view, need to remain in the hospital. [11]

Sometimes it may be the patient who persuades the clinician that release is justified, or a mutual agreement may be arrived at. It is also possible that clinicians revise post hoc their rating of patient characteristics or their hypothetical commitment decision to be more consistent with the patient's retraction decision. (A clinician may, for example, identify with a patient's wish to be healthy.) We will have less need to discriminate among these explanations if we conceive of rationality as an interpersonal rather than a purely individual attribute; we can then accept that the three-day paper is one event in an ongoing clinical process, a patient-clinician-judge transaction [8] that can lead to shared understanding.

One patient characteristic that correlates significantly with the patient's decision to retract requires a more complex interpretation. It is not intuitively obvious why patients who are less well-known to their clinicians are more likely to retract their three-day paper. The answer may lie in the distinction between a transferential and a situational misalliance. [12] Any filing of a three-day paper may be considered at least a temporary failure in the therapeutic alliance. Nonetheless, the better the patient and clinician know each other, the more serious a problem in the alliance the filing indicates. Patients who do not know their primary treating clinician very well frequently file for release out of a desire to gain attention, to decrease the anxiety generated by the clinician's unfamiliarity, or to cope with the uncertainty of illness by doing something to gain control, however illusory. [13,14] The filing of the paper often serves to mobilize the attention and/or explanation and thus generate the familiarity and trust needed to resolve these problems and strengthen the alliance. But when a patient who is already familiar with (and to) the clinician requests release, then the threat to the alliance is more than a matter of uncertainty or lack of information, and the alliance is not likely to be reestablished (as signaled by the patient's retraction) during the three-day period.

The fact that this study and its two predecessors [1,2] were part of the institution's utilization review procedure allowed the participation of a large number of clinicians from the institution (all those considering whether to initiate legal commitment proceedings for patients seeking to leave the psychiatric hospital against medical advice), but limited the amount of effort that could be demanded from each of them. The methodology that was developed for this situation has been helpfully criticized on a number of counts. We now respond to these criticisms in the hope of deepening the context in which the limitations and significance of the findings may be understood.

First, the fact that a different clinician made the decision about each patient distinguishes this method from the method cognitive psychologists commonly use to describe individuals' judgment policies, [15] which requires the individual to judge a large number of cases. For an infrequently occurring decision, such as the clinician's response to the three-day paper, these cases would have to be hypothetical. The "institutional review" mandate for the study was not presumed to be powerful enough to motivate all the clinicians to make such a set of judgments. Therefore we do not claim to describe the judgment policy used by every individual clinician in the institution. We do, however, claim that the analysis of clinicians' responses to these patients in the aggregate can serve as a "paramorphic" description [16] of the decision behavior of the institution as a whole.

A second issue is that because some of the clinicians filled out more than one questionnaire, they are overrepresented in the results. Although some critics view this as a fault, it is appropriate methodology for our goals because in order to describe the decision policy of the institution as a whole, it is necessary that clinicians who represent the institution more frequently than others in a particular type of decision be included in the study in proportion to the relative frequency with which they make the decisions. The critics go on to say that the inclusion of several cases by the same clinician artificially inflates the N of the analysis from 36 (the number of clinicians) to 71 (the number of patients). However, the appropriate N in a study is determined by the kind of conclusion one wishes to draw. If just one clinician had made all 71 judgments, it would still be appropriate to base the statistics on all 71 cases. [15] The judgment policy description, however, would apply only to that one clinician. If our goal had been a general description of all clinicians, we could only base our conclusions on an N of 36. However, since we are describing the behavior of the institution, using a representative sample, it is appropriate to base statistical conclusions on an N of 71, the number of cases for which complete observations were available.

A third criticism is that it is difficult to know the dynamics of the decision-making process because the questionnaires were filled out only after the decisions were made. While our methods revealed important effects, it would pay in a future study to use methods that focus more closely on the process, as by having the clinician rate the patient and state the most likely decision at each of a number of points during the three-day period. Such methods would demand more of participating clinicians.

Fourth, because all judgments of patient characteristics were made after the decision, by the clinician who made the decision, it is possible that the patient ratings are influenced by the clinician's decision or by the fact that the patient retracted the three-day paper. If the clinician rated the patient in a way that justified the decision or that reduced cognitive dissonance, this would influence the relation that our analysis revealed between the decision and the patient characteristics. Although the method we used does not control for this possibility, analysis of the self-reported use of the various cues in making a decision in the first study [1] showed no evidence that the clinicians engage in such self-justification with respect to the implied requirement to rely solely on the three judicial dangerousness criteria: rather, clinicians reported relying on the extrajudicial factors to a larger extent than they actually did. In future studies, obtaining ratings of patient characteristics at a number of points in the three-day period, before the clinician makes a decision and before the patient might retract the three-day paper, would limit any influence of the clinician's decision on his or her ratings of patient characteristics. Another possible method would be to have other clinicians who are familiar with the patient rate the patient characteristics.

A related issue is the possibility that the clinician's statement of whether he or she would have sought commitment may be determined by the fact of the patient's retraction. (This would be a "Catch 22" situation for the patient: in order to prove to the clinician that he or she is sane enough to be allowed to leave the hospital, the patient must retract the request to leave, and thus not leave the hospital.) If the clinician's decisions are indeed influenced by this factor, the temporally fine-grained observation method described above would control for its effects in the analysis of the data.

To summarize, use of the present study's methodology has revealed important relationships between patient characteristics and both the clinicians' decisions and the patients' retractions. In response to criticism of the methodology, we have discussed the basis for the methodology and suggested methodological variants for future studies that, while demanding more from the respondents, would clarify some questions about the doctor-patient transaction that have been raised by our findings.

Appendix

While the more seriously ill and more dangerous patients were more likely to retract their three-day papers, there is evidence in our data that the strength and even the direction of this relationship depended on the clinician's intentions, as evidenced by whether the clinician sought commitment (or would have, had the patient not retracted the three-day paper). The nature of this unanticipated interaction pattern appears to be: For those patients where the clinician was not going to accede, the less ill the patient the harder the fight; for those patients where the clinician was ready to accede, the more seriously ill the patient the harder the fight.

To demonstrate how often this interaction pattern occurs, we must first distinguish whether each patient characteristic would be expected to influence the clinician to seek commitment. A priori, twelve patient characteristics would lead the clinician to seek commitment: Danger to Others, Danger to Self, Acute (as opposed to Chronic), Needs Treatment, Ill, Psychotic Thought, Psychotic Behavior, Family Favors Commitment, Frightening, Disruptive on Ward, Depressing, and Discouraging. Six patient characteristics would favor the patient's release: Able to Care for Self, Reliable Outpatient, Reliable with Medication, Has Place to Live, Has Support on the Outside, and Able to Work. The remaining five patient characteristics are plausibly neutral with respect to the decision to seek commitment: Poor, Likeable, Seductive, Well-known to Clinician, and Academically Interesting.

When a patient characteristic would lead the clinician to seek to commit the patient (e.g., Danger to Others), there is evidence for the hypothesized interaction pattern when either:

the correlation between the patient characteristic and the patient's decision to retract the three-day paper is more positive (or less negative) when the clinician would seek commitment than when the clinician would allow the patient to leave the hospital, or
the interaction term (in the regression of the patient's retraction onto the patient characteristic, the clinician's decision, and their interaction) has a negative regression coefficient.

For the purpose of measuring how often this interaction pattern occurs, a conservative approach will be adopted: a commitment-favoring patient characteristic will be considered to manifest the interaction pattern only when it has both of these signs.

Conversely, when the patient characteristic would lead the clinician to release the patient (e.g., Able to Care for Self), there is evidence for this interaction pattern when either:

the correlation between the patient characteristic and the patient's decision to retract is less positive (or more negative) when the clinician would seek commitment than when he or she would not, or
the interaction term has a positive regression coefficient.

A release-favoring patient characteristic is considered to manifest this interaction pattern only when it has both of these signs.

Eleven variables meet the criteria for this interaction pattern: Danger to Others, Danger to Self, Acute, Needs Treatment, Ill, Psychotic Thought, Reliable Outpatient, Has Place to Live, Has Support on the Outside, Able to Work, and Disruptive. For only three variables - Depressing, Discouraging, and Family Favors Commitment - is there evidence for the opposite interaction pattern. This opposite pattern is quite strong for the first two, which are interpersonal factor judgments - when the clinician would seek commitment, the more discouraging or depressing the patient, the less likely the patient was to retract; however, when the clinician would not seek commitment, the more discouraging or depressing the patient, the more likely the patient was to retract. The preponderance of the paradoxical pattern, which occurred with 11 patient characteristics while the reverse pattern occurred with only three, approaches significance at the p < .10 level (Chi-squared = 3.50, df = 1, two-tailed).

References

Appelbaum PS, Hamm RM: Decision to seek commitment: Psychiatric decision making in a legal context. Arch Gen Psychiatry 39:447-451, 1982.
Schwartz HI, Appelbaum PS, Kaplan RD: Clinical judgments in the decision to commit: Psychiatric decision and the law. Arch Gen Psychiatry 41: 811-815, 1984.
Gutheil TG, Appelbaum, PS: Clinical Handbook of Psychiatry and the Law. New York: McGraw-Hill, 1982.
Stone AA: Mental health and Law: A System in Transition. New York: Jason Aronson, 1976.
Stromberg CD, Stone AA: A model state law on civil commitment of the mentally ill. Harvard Journal on Legislation 20:275-396, 1983.
Appelbaum PS: Special section on APA's model commitment law: An introduction. Hospital and Community Psychiatry 36:966-968, 1985.
Stone AA: A response to comments on APA's model commitment law. Hospital and Community Psychiatry 36:984-989, 1985.
Bursztajn H, Gutheil TG, Mills MJ, Hamm RM, Brodsky A: Process analysis of judges' commitment decisions: A preliminary empirical study. American Journal of Psychiatry 143:170-174, 1986.
Massachusetts General Law, Chap. 123, Para. 1, 11.
Cohen J, Cohen P: Applied Multiple Regression: Correlation Analysis for the Behavioral Sciences. Hillside, N.J.: Erlbaum, 1975.
Frank JD: Persuasion and Healing. Baltimore: Johns Hopkins Press, 1961.
Gutheil TG, Havens LL: The therapeutic alliance: Contemporary meanings and confusions. Int. Rev. Psycho-Anal 6:467-481, 1979.
Bursztajn H, Feinbloom RI, Hamm RM, Brodsky A: Medical Choices, Medical Chances: How Patients, Families, and Physicians Can Cope with Uncertainty. New York: Delacorte/Lawrence, 1981.
Langer E: The illusion of control. J. Pers. Soc. Psychol 32:311-328, 1985.
Hammond KR, McClelland GH, Mumpower J; Human Judgment and Decision Making: Theories, Methods, and Procedures. New York: Praeger, 1980.
Hoffman PJ: The paramorphic representation of clinical judgment. Psychological Bulletin 57:116-131, 1960.