Application of Involuntary Admission Criteria in Psychiatric Emergency Rooms

Abstract

The authors developed Three Ratings of Involuntary Admissibility (TRIAD) to reflect the way clinicians in psychiatric emergency rooms apply legal criteria for the involuntary commitment of patients. Using TRIAD to rate patients on danger to self, danger to others, and grave disability, psychiatric social workers observing patients and clinicians correctly predicted 82 percent of 89 dispositions in two metropolitan county hospitals. These results indicate that clinicians agree on the meaning of the legal criteria and are able to apply them consistently.

THE CIVIL commitment of mentally ill persons poses a major dilemma for mental health professionals throughout the United States. Given the ambiguities in commitment statutes and the difficulty inherent in predicting patients’ behavior, clinicians must make decisions that may, on the one hand, violate individuals’ rights, or on the other, result in the neglect of community safety or of individuals who need care. Although it is generally agreed that commitment is necessary in some cases, there is widespread concern that the process is irrational, arbitrary, and discriminatory.¹ Furthermore, the process has been seriously questioned in many cases that have come to the attention of the courts.²

Most efforts to prevent improper use of commitment have focused on procedural safeguards to ensure “due process” for patients, which implies the existence of standards that are thoroughly and consistently applied in all cases.³ The courts and legislatures have left the substantive interpretation of commitment criteria to professional discretion.⁴ They have assumed that despite the lack of evidence that clinicians can accurately predict patients’ behavior, professional standards exist that can be consistently applied. In view of this assumption, it is surprising to find that of the several studies examining clinical reasons for admission decisions, none has attempted to describe the clinical application of legal or statutory criteria.⁵

According to Schwitzgebel, most states specify two or three criteria for involuntary commitment: Danger to self and/or others or likelihood of serious harm to self or others is usually combined with a criterion similar to California’s “grave disability” standard.⁶ Although state statutes vary in the degree of restrictiveness implied by their wording, “the trend has been to narrow the population of those who may be committed.”⁷ Because the California statute first implemented in 1969 was a harbinger of this trend, information about application of this standard by clinicians may be relevant to other states with similar laws.

CIVIL COMMITMENT IN CALIFORNIA

The Lanterman-Petris-Short Act (LPS) specifies criteria for civil commitment in California, but the law provides little definition of these standards. The commitment process begins with a 72-hour emergency detention of the patient for observation and treatment. Although a variety of mental health and law enforcement officials are authorized to begin this process, the critical decision about hospitalization is made by personnel in the psychiatric emergency rooms of general hospitals. No data have been compiled to indicate how these clinicians apply legal commitment criteria.

According to LPS, involuntary hospitalization requires (1) that the person be dangerous to himself or herself, dangerous to others, or gravely disabled and (2) that the person’s condition be due to a mental disorder. Thus, the law requires that two separate assessments be made, but it gives almost no statutory guidance. In failing to specify the meaning of these criteria, the legislature clearly intended that determinations be guided by clinicians’ judgment.

The criteria that have been the focus of greatest concern are those of dangerousness and grave disability. In a statewide evaluation of involuntary treatment procedures, Schwitzgebel and Swenson reported to the California legislature that there is need for clarification of the criteria to be used in the detention of patients under the three LPS standards…. Consistently applied interpretations have been lacking. Facility staff members frequently seem to want information or suggestions about the detention or commitment criteria. Ambiguity of interpretation allows an unnecessary and unintended abuse of liberties. The preparation of regulations or guidelines describing involuntary detention criteria might, with suitable inservice training, reduce considerably the present diversity in the application of LPS standards.⁸

PSYCHIATRIC ADMISSIONS

Uncertainty among clinicians about how the involuntary commitment law should be interpreted and corresponding inconsistency in its application do not belie the assumption of a body of relevant clinical opinion, nor do they reveal in what particulars and to what extent its application is inconsistent. A number of studies over two decades are widely cited to suggest that admission decisions are significantly and inappropriately influenced by social characteristics of the patient or the emergency-room setting.⁹ However, many of the findings are contradictory, and most studies report a role for social factors that is secondary to that of the “severity of illness”—usually a global concept that includes violent or suicidal activity or inability to care for oneself.¹⁰

Moreover, most previous studies of determinants of admission decisions are seriously flawed. The conclusions of these studies are valid only to the extent that all significant variables that influenced the decision were included in the analysis. Applebaum and Hamm studied decisions to apply for commitment of already-hospitalized voluntary patients.¹¹ They found that psychiatrists, using self-report measures, attributed the greatest influence to the patient’s status with regard to one or more of the legal criteria (danger to self, danger to others, and ability to care for self) and to nonlegal variables closely related to ability to care for and protect oneself in the community. However, studies purportedly demonstrating the influence of sociodemographic and environmental factors on the admission decision have not specifically considered the influence of legal commitment criteria as they are clinically construed.¹² Indeed, only two studies included as an independent variable a clinical assessment of the state of the patient with regard to one legal criterion for commitment (danger to self).¹³ No study has used an independent rating by an observer other than the evaluating clinician of the patient’s status on any legal criteria.

In short, it is too soon to conclude that mental health professionals need administrative guidelines to achieve substantial agreement and consistency in applying involuntary admission criteria. Further effort is warranted to establish (1) the extent to which there is already agreement among clinicians as to the meaning of the criteria and (2) the extent to which there is consistency in the application. Note that the question being addressed here is not the predictive validity of emergency psychiatric assessments, but rather the question of whether clinicians respond to similar cases with similar judgments. To this end, the authors report the preliminary results of efforts to develop a tool to describe the application of legal commitment criteria in psychiatric emergency rooms.

METHOD

In an attempt to reflect the way clinicians in psychiatric emergency rooms interpret and apply the legal criteria of danger to self, danger to others, and grave disability, the authors developed a prototype index entitled “Three Ratings of Involuntary Admissibility” (TRIAD). The authors developed the instrument through an iterative process that included a literature review, observation of actual cases, and debriefing of clinicians after dispositions were made. This process resulted in the identification and ranking of patterns of behavior and circumstance more and less likely to lead to a determination that a patient is involuntarily admissible by LPS standards.

The authors theorized that through professional training and experience, clinicians are sensitized to patterns of behavior and circumstance that they believe to be associated with a patient’s danger to self, danger to others, or grave disability and that the clinicians internalize scales by which they weigh or rank these patterns. Thus, clinicians will react to certain patterns as unambiguously dangerous or as not dangerous, and they will consistently respond to these patterns with decisions that a person is admissible or not admissible according to involuntary commitment criteria.

This hypothesis is supported by the findings of Meyerson et al. that clinicians’ experience affected admission decisions in the middle range of mental illness, but at the extremes of mental illness, experienced and inexperienced clinicians admitted patients at the same rate.¹⁴ The authors predicted, therefore, that admission decisions will be highly consistent in cases involving unambiguous patterns that clinicians associate with dangerousness and disability. However, clinicians will experience other patterns as more ambiguous, and this ambiguity will lead to a greater variation in the outcome of decision making.

Expecting that many patients would present complaints or behavior related to more than one of the legal criteria, the authors hypothesized further that an ambiguous presentation on any single criterion would be more likely to lead to a decision that the person was admissible if it was accompanied by at least a low-level presentation on a different criterion. For example, a person who presented some moderate threat to the safety of others would more likely be judged admissible if he or she also seemed to present a moderate or mild potential for self-harm. Thus, the authors expected that a total score across all three criteria on TRIAD would also predict the clinician’s judgment.

Observations.

Using TRIAD, the researchers observed evaluation interviews in the Psychiatric Emergency Services (PES) of San Francisco General Hospital and Highland General Hospital, Oakland. These are the major emergency evaluation units for the two largest San Francisco Bay Area counties. Eighty-nine patients were chosen on the basis of their availability at a time when an observer was free to follow a new case. Observers—psychiatric social workers experienced in assessment of acutely disturbed patients—followed patients and the assigned clinicians as long as the patients remained in the PES, usually for a period of several hours. TRIAD was scored by the observer when a disposition decision had been reached. The score was based on the items checked off during the evaluation period that still applied at the time of disposition. The clinician handling the case was not involved in the scoring process.

Description of TRIAD.

TRIAD is an easily scored instrument consisting of three scales. The three scales, organized as checklists, consist of a total of 88 numbered items that can be combined to yield 156 patterns of behavior and circumstance relevant to the clinical prediction of violence and suicide and the assessment of grave disability. For each of the scales, a number of patterns are assigned to the highest score, a number are assigned to the next highest score, and so on.

No pattern combines more than nine items, and most involve two, three, or four items. For example, “threatened to harm another” is one item that, by itself, scores at a moderate level (2) on the danger to others scale. However, such a threat may yield the highest score (4) if it occurs in combination with three other particular items. The first additional item has to do with provocation or lack thereof. The other items involve indication that a patient has a concrete plan or weapon; is in a volatile, unpredictable, or enraged state; and has a history of assault. According to the authors’ hypothesis, if such a presenting picture is accompanied by a current diagnosis of mental disorder, the evaluating clinician will determine that the patient is clearly admissible by LPS standards. To prevent hospitalization, the clinician may attempt to bring about some change in the patient through crisis intervention or medication in the emergency room, but if such efforts fail, admission will follow. If the efforts succeed, the danger to others score will be lower. TRIAD is scored at the time of disposition by finding the standard pattern represented by the items checked during the evaluation that are applicable at disposition and yield the highest score.

RESULTS

Interrater Reliability.

Three pairs of observers, using TRIAD, rated ten cases each and achieved interrater reliability coefficients (Pearson’s r) of .94 on the danger to self score; .89 on the danger to others score; .77 on the grave disability score; and .89 on the total admissibility score. The results demonstrate that it is possible to use this instrument reliably to rate actual cases in the psychiatric emergency room.

Characteristics of the Sample.

Table 1 summarizes some of the demographic and diagnostic characteristics of the sample of 89 patients observed in the two hospital emergency rooms. Using data from most of the sample, the authors were able to describe a typical patient in the sample. This typical patient was a white (62 percent) male (69 percent), age 26–44 (66 percent), born in the United States (87 percent), and fluent in English (94 percent). The patient had never been married (46.6 percent), had ten to 12 years of education (63 percent), and was out of the job market as a result of disability (70.4 percent) for which he was receiving Supplemental Security Income (SSI) (54.4 percent). He was more likely to receive a diagnosis of psychosis (66 percent) than a nonpsychotic diagnosis (34 percent).

Table 1.

Characteristics of Patients Seen in Psychiatric Emergency Services (N = 89)

Characteristics	Percentagea
Sex
Male	69.3
Female	30.7
Age
14–25	11.5
26–44	65.8
45–87	22.4
Ethnicity
White	61.7
Black	24.7
Spanish surname	8.6
Other	4.9
Birthplace
United States	86.5
Foreign	13.5
Marital Status
Never married	46.6
Married	25.9
Divorced or separated	25.8
Widowed	1.7
Education
Less than 10 years	16.7
10–12 years	62.5
13–17 years	20.8
Employment
Full- or part-time	5.6b
Unemployed	8.5
Disabled	70.4
Other	15.4
Income Source
None	10.5
Ramily or friends	12.3
Employment	10.5b
Disability	54.4
Other	12.5
Living Arrangements
No address	29.3
Alone	22.7
With others	33.4
Sheltered care	8.0
Other	6.6
Number of Previous Visits to This Psychiatric Emergency Service
None	52.9
1–5	24.2
6 or more	18.5
Uncertain	4.4
Number of Prior Hospitalizations
None	14.3
Multiple (number unknown)	49.2
1–5	25.4
7–13	11.2
Previous Diagnosisc
Major affective disorder ± schizophrenia	27.9
Schizophrenia ± substance abuse	44.2
Substance abuse only	11.6
Organic psychotic disorder	4.7
Acute or atypical psychotic disorder	4.6
Adjustment reaction or anxiety disorder	7.0
Current Diagnosis (DSM-IIlAxis I) Psychoticd
Schizophrenia or schizoaffective disorder	32.8
Major affective disorder	18.8
Organic psychotic disorder	5.9
Other psychotic disorder	8.3
Current Diagnosis (DSM-IIUAxis I) Nonpsychoticd
Adjustment or anxiety disorder	13.1
Acute organic or substance use disorder	6.0
Other nonpsychotic disorder	8.1
No major disorder or diagnosis deferred	7.0

^aPercentages in each category may not total 100 due to rounding.

^bAlthough 10.5 percent of respondents stated that their source of income was from employment, only 5.6 percent were employed at the time of evaluation.

^cprimary diagnosis in the past, by category.

^dThe total of psychotic and nonpsychotic diagnoses equals 100 percent.

Clinicians and Setting.

Forty-nine percent of the cases were evaluated by psychiatrists, 25 percent by nurses, 15 percent by social workers, 11 percent by other professionals, paraprofessionals, or unlicensed professionals in training. The clinical experience of evaluators ranged from two to 23 years, and the emergency psychiatric experience of the clinicians ranged from less than six months to 13 years. Of the patients observed, 90.2 percent were examined by clinicians with two years of PES experience or more.

The number of patients evaluated in the emergency service on any day of the observations ranged from 14 to 32, and in most cases was 20 to 26. If admitted to a ward following the emergency evaluation, the patient was most likely to remain for 7 to 9 days (30 percent) or for 15 to 17 days (25 percent). Average occupancy rates for the inpatient wards at the two hospitals during the study period were 94 percent and 91 percent.

Severity of Presenting Problem.

Table 2 shows how scale scores axe combined to yield different TRIAD severity levels, given a range of 0–4 on the danger to self and danger to others scales and a range of 0–3 on the grave disability scale.¹⁵ The distribution across severity levels of the 89 observed patients was as follows: 69.7 percent at the highest severity level, 4; 2.2 percent at level 3; 10.1 percent at level 2; and 18.0 percent at level 1, the lowest level.

Table 2.

Disposition of Cases by Severity Level (N = 89)

	Disposition
Severity Level on TRIAD Scalesa	Released	Retained Voluntarily	Retained Involuntarily	Total
Level 1 (DSS, DOS. GDS = 0 or 1; total score = 3 or less)b	13 (81%)	0	3 (19%)	16 (100%)
Level 2 (DSS, DOS, GD = 2; total score = 2)	7 (78%)	0	2 (22%)	9 (100%)
Level 3 (DSS, DOS, GDS = 2; total score = 3)	1 (50%)	0	1 (50%)	2 (100%)
Level 4 (DSS, DOS, GD = 3 or 4; total score = 4 or more)	10 (16%)	4 (7%)	48 (77%)	62 (100%)

^aTRIAD = Three Ratings of Involuntary Admissibility.

^bDSS = danger to self scale; DOS = danger to others scale; GDS = grave disability scale.

Disposition.

Disposition was consistent with TRIAD severity scores in 82 percent of the cases (gamma = .82; see Table 2), and agreement was roughly equivalent for both hospitals.¹⁶ After the initial evaluation, 58 patients (65 percent) were retained.

As expected, the most and least severe presentations were most predictive of disposition (84 percent and 81 percent correct predictions, respectively). The high scorers who were retained and the low scorers who were released are the true positives and true negatives. False positives and false negatives are identified in Table 2. False negatives are patients who scored low on TRIAD (level 1) but were retained by the clinicians (19 percent); false positives are the high scorers (level 4) who were released by the clinician (16 percent).

Severity levels 2 and 3 represent the hypothesized ambiguous range on TRIAD. However, severity level 2 also turned out to be quite discriminating, with 78 percent of the patients being released. At severity level 2, the picture presented by patients was ambiguous, but clinicians were inclined to let the patients go. The least predictive score configuration represents the situation in which the patient presents only a moderate degree of concern on any one criterion but raises one other issue at a low level (severity level 3). With only two cases at this level, the dispositions of 50 percent released and 50 percent admitted is far from conclusive. However, the difference in retention rates between severity levels 2 and 3 suggests that the index is capable of representing the decision-making process at a fine level. Future observations will be necessary to test the authors’ hypothesis that severity level 3 represents the most ambiguous situations that provide wider latitude for clinical discretion.

Most (69.7 percent) of the 89 patients scored at the highest level of severity. Table 3 describes the disposition of patients at level 4 according to whether their high score resulted from danger to self (8 percent), danger to others (35 percent), grave disability (38 percent), or a combination (2 percent). Thirteen percent (n = 12) scored at the highest level on two scales.

Table 3.

Disposition of Cases at Severity Level 4 by TRIAD Scale and Total Scores (n = 62)

	Disposition
TRIAD Scale and Total Scoresa	Released	Retained Voluntarily	Retained Involuntarily	Total
Danger to Self = 3 or 4	5 (71%)	0	2 (29%)	7 (100%)
Danger to Others = 3 or 4	4 (13%)	2 (6%)	25 (81%)	31 (100%)
Grave Disability = 3	1 (3%)	2 (6%)	31 (91%)	34 (100%)
Total = 4 or more but no scale score = 3 or 4	0	0	2 (100%)	2 (100%)

^aTRIAD = Three Ratings of Involuntary Admissibility. Twelve cases scored at the highest level on two scales.

Of the patients whose scores on danger to others and grave disability placed them in the highest severity level, 87 percent and 97 percent, respectively, were retained. Of those who attained the highest severity level by reason of a high danger to self score, 71.4 percent were released, contrary to the authors’ expectation, and 28.6 percent were retained.

Diagnosis.

Disposition may legitimately be influenced, by legal and clinical considerations in addition to dangerousness and grave disability. The presence or absence of a mental disorder and the severity of the disorder are major criteria. To the extent that the presence or absence of a diagnosis of psychosis captures these concerns, the authors are able to report the influence of legal and clinical concerns—in addition to dangerousness and grave disability—on disposition.

To facilitate analysis, DSM-III Axis-I diagnoses were categorized as psychotic and nonpsychotic.¹⁷ Although the presence of psychosis was moderately related to severity of presentation on TRIAD (gamma = .53), it was strongly related to disposition (gamma = .79), although not as strongly as TRIAD severity (gamma = .82). Thus, it appears that degree of dangerousness and disability, on the one hand, and presence or absence of psychosis, on the other, make partially independent contributions to the explanation of disposition. It is not surprising that the relationship between disposition and TRIAD severity was stronger for nonpsychotic patients (gamma = .89) than for psychotic patients (gamma = .74). Presence or absence of psychosis is helpful in explaining dispositions that differ from those predicted by the TRIAD score. However, it is important to point out that only three psychotic patients were held with low TRIAD scores, and in only one of these cases was there no manifestation of dangerousness or grave disability as measured by TRIAD.

Discrepant Cases.

It appears that the best explanation for the discrepancy between TRIAD scores and disposition in the false positive cases is the clinician’s judgment in each case that admission was not clinically indicated—that is, severity of mental disorder (insofar as it is reflected by diagnosis), the availability of treatment alternatives, and the judgment that patients would not benefit from hospital care appear to have been the critical factors. However, the authors’ process recordings and clinicians’ comments on their evaluations led the authors to believe that the clinicians did not doubt that these patients’ presentations sufficiently reflected the legal criteria to justify holding them, if clinically indicated. At least two discrepant dispositions appear, however, to have been determined by nonclinical considerations—in one case a false positive and the other a false negative. A second false negative appears to be explained by clinical considerations unrelated to the legal criteria, whereas the third reflected a deficiency in TRIAD that the authors’ have corrected.

COMMENT

The results of this preliminary study strongly suggest that clinicians in urban psychiatric emergency rooms in the San Francisco Bay Area have a common understanding of danger to self, danger to others, and grave disability; that these constructs can be reliably applied in actual cases; and that at least in observed cases most involuntary admissions are predictable from the severity of the patient’s status with respect to these criteria. Furthermore, it appears that these shared constructs can be operationalized.

The study provides a test of TRIAD as an instrument that describes clinicians’ judgments about whether a patient meets legal standards for involuntary admission. In this instance, the concurrent measure was disposition. By this criterion, the construct validity of the scales of danger to others and grave disability was supported. Also supported was the validity of the total TRIAD score as a measure of involuntary admissibility. However, the validity of the danger to self scale has yet to be demonstrated.

Although dispositon proved a useful concurrent measure of the construct validity of TRIAD, it is obviously limited by the fact that variables beyond the clinician’s assessment of dangerousness and disability appropriately influence this decision. The authors are currently proceeding with other ways to test the validity of TRIAD as a measure of clinicians’ constructs of danger to self and others and grave disability. In a study of 250 additional cases, the authors are obtaining clinicians’ independent global ratings of danger to self, danger to others, and grave disability at the time of disposition as well as researchers’ TRIAD ratings. To measure the external validity (generalizability) of TRIAD, the authors have expanded the study to include five emergency rooms.

If these procedures establish that TRIAD reflects the way clinicians interpret the legal criteria, the discussion of emergency involuntary commitment criteria and procedures should be greatly facilitated. TRIAD could provide a most useful description of the state of patients considered involuntarily admissible, as well as assurance that it is possible to apply the legal criteria consistently and equitably.