Secondary Logo

Journal Logo


Treatment of Behavioral Emergencies: A Summary of the Expert Consensus Guidelines


Author Information
Journal of Psychiatric Practice®: January 2003 - Volume 9 - Issue 1 - p 16-38



The number of episodes of psychiatric care more than doubled between 1970 and 1994 while the number of inpatient beds was cut by half. 1 This shift toward treatment in the least restrictive setting, which was fueled by economic factors, has occurred in the context of increasing public concern about violence committed by individuals with severe mental illness in the community. Concern has also increased about the potential for physicians to abuse their so-called police powers, and this had led to a debate on the use of physical and chemical restraint or seclusion. 2 All of these factors have created an urgent need to establish coherent policy concerning the delivery of psychiatric emergency care that will help psychiatric emergency services balance the rights of patients with considerations of safety and good standards of care. However, the process of developing such policies is complicated by a number of problems. First, key constructs, such as agitation, have not been adequately operationalized, 3 so that the criteria defining a behavioral emergency are vague. Second, there are few data on which to base clinical policies, given the relative lack of research data in this area.

In a related development, payment for psychiatric hospital care is now often linked with dangerousness more than need for treatment. This has led to an increased concentration of aggressive patients in the hospital and emergency setting. 4 Mental health professionals are asked to make rapid decisions about interventions in situations in which the safety of patients and staff may be at risk. In an extensive review of the literature, Fisher concluded that restraint and seclusion “work” in the limited sense that they “can prevent injury and reduce agitation.” However, Fisher and others have also described deleterious effects on patients, who perceive such interventions as coercive and traumatic. 5–8

The perception that at least some use of restraint and seclusion is unnecessary was reinforced by the finding published by Way and Banks 9 in 1990 that there was wide variability in the use of restraint and seclusion across sites that was accounted for by institutional culture rather than by characteristics of individual patients. Relatively few data are available on the actual extent to which restraint and seclusion are used in emergency settings. Based on the results of a recent survey of approximately 50 psychiatric emergency services in the United States, it was estimated that 37.2% of patients presented involuntarily but that only 8.5% were restrained at any point during their time in the emergency setting. The mean duration of restraint reported in this survey was 3.3 ± 2.9 hours. 10

The controversy in this area was heightened by a number of reports of patient deaths while in restraint or seclusion. In 1994, the New York State Commission on Quality of Care reported 111 patient deaths over the 10-year period ending in 1993, a finding that led the Commission to undertake a statewide review of restraint and seclusion practices. 8 The controversy was further heightened by the publication in 1998 of a 5-part series in The Hartford Courant entitled “Deadly Restraint.”11The Hartford Courant series reported 142 deaths of patients in restraint or seclusion over a 10-year period. A study by The Harvard Center for Risk Analysis commissioned by the Courant estimated that between 50 and 100 such deaths occur each year. While the methodology employed to generate this estimate is open to question, the large number of deaths cited galvanized political momentum to regulate restraint use in healthcare settings more tightly. The New York State Commission and the National Association of State Mental Health Program Directors have both issued statements questioning the therapeutic value of restraint and seclusion and stressing their traumatic nature. 8,12 The National Alliance for the Mentally Ill (NAMI) has been collecting reports concerning adverse outcomes associated with the use of restraint and seclusion since the publication of The Hartford Courant series. 13

Such concerns led the Health Care Financing Administration (HCFA), now the Centers for Medicare and Medicaid Services (CMS), to introduce interim final rules for conditions of participation for facilities receiving Medicare and Medicaid payments. 14 These rules address patients’ rights in general and specifically discuss issues related to restraint and seclusion. The Joint Commission on Accreditation of Healthcare Organizations (JCAHO) has also produced regulations concerning the use of restraint and seclusion in psychiatric and medical settings. 15

Behavioral emergencies are not rare events. For example, it has been reported that there are approximately 135,000 psychiatric emergency visits per year in New York State alone. 16 Whether in the emergency room or in an inpatient psychiatric setting, immediate assessment and effective intervention can reduce the danger to patients and staff and more quickly speed patients to recovery.

Behavioral emergencies are complex situations. Routine care generally involves a cooperative patient and adequate time to perform an assessment and to reach agreement with the patient on a course that maximizes benefits and minimizes risk. By contrast, emergencies are dynamic situations; the diagnosis is often unknown or provisional at best; there is a sense of urgency, limited time for decision making, and a need both to intervene immediately despite limited data and to change course rapidly as new information becomes available, including responses to prior interventions. Any course of action or inaction may have serious adverse effects. It is therefore important for clinicians to remember that they must first do no harm. Even an objectively good response may leave the patient feeling traumatized and angered by the process.

Recent developments and innovations in pharmacology have combined with a rapidly changing regulatory environment to increase the level of complexity and difficulty already inherent in managing behavioral emergencies. The subspecialty of Emergency Psychiatry is emerging in the context of these demands. Unfortunately, high-quality, empirical data on the most effective and appropriate methods of managing behavioral emergencies are quite limited. As a result, there are no comprehensive evidence-based practice guidelines on the best treatment approaches for managing these situations. This has resulted in a need to create new and useful educational materials and programs to help train emergency physicians to meet current standards for behavioral emergency treatment. We therefore undertook a consensus survey of expert opinion on the management of behavioral emergencies. In this article, we summarize the Expert Consensus Guidelines on the Treatment of Behavioral Emergencies, which were published in May 2001. 17

In developing the survey and guidelines, we had a number of important goals in mind.

  1. To help clinicians address the many overlapping and complex factors involved in the management of behavioral emergencies, such as varying local and state practices governing the use of restraint and seclusion, the appropriate use of pharmacological agents, the selection and application of alternatives to physical restraint, and the use of physical restraint itself.
  2. To assist hospitals and clinics to establish policies for the management of behavioral emergencies as has been increasingly mandated by regulatory requirements.
  3. To assist hospitals and other clinical services to provide structured staff training in the management of behavioral emergencies and the documentation of adherence to pertinent policies, as required by regulatory agencies. Such educational resources are especially necessary because many emergency settings are very active venues for training and education. In a recent survey of Psychiatric Emergency Service administrators, more than 90% of respondents reported that medical residents rotated through their service and 74% reported that their services were involved in training medical students. 10
  4. To promote adoption and use of new knowledge concerning the treatment of acute behavioral dyscontrol. Up to now, the integration of newer drugs and formulations into standard practice has been slow. Such a lag in the application of new knowledge has a significant adverse impact on patients and their families.
  5. To address two important issues that have not previously been well addressed in educational protocols for behavioral emergencies: the patient’s perspective and a focus on specific diagnostic treatments. For the patient who has lost control or is in danger of losing control, how the episode is resolved can have enormous implications for the remaining course of illness. However, there has been little careful study on how patients experience such episodes and which crisis intervention approaches are preferred by patients. The lack of focus on specific diagnostic treatments has led to the prolific use of “blanket” regimens, nonspecific treatments meant to simplify decision-making by covering as broad a range of medical presentations as possible. Little attention has been paid to developing treatment regimens that would be more specific and appropriate for the underlying cause of the behavioral emergency and would, as a result, lead to a more rapid resolution of the problem underlying the behavioral emergency.


The contribution of expert consensus to practice guideline development continues to evolve throughout medicine, alongside the “gold standard” of meta-analysis of clinical trials and other experimental data. The sheer number of possible combinations and sequences of available treatments for many diseases makes it difficult to provide comparative recommendations based entirely on clinical trial data. 18,19 A method for describing expert opinion in a quantitative, reliable manner to help fill some of the gaps in evidence-based guidelines has been developed. This method has been applied to a variety of psychiatric disorders. 20–28

Creating the Surveys

We first created a skeleton algorithm based on a literature review. We sought to identify key decision points in the management of behavioral emergencies as well as a list of feasible options for intervention. We highlighted important clinical questions that had not yet been adequately addressed or definitively answered in the literature. 29 A written questionnaire was then developed with 61 questions with a total of 808 options. We asked about the types of assessments, how to select the most appropriate emergency interventions, when and how to use restraint and medication, and how to tailor selection of interventions to the most likely etiology of the behavioral dyscontrol. We also addressed lack of adequate response to initial intervention and safety and tolerability issues, such as management of behavioral emergencies in pregnant women, children, and individuals with complicating conditions.

The Rating Scale

For most of the options in the survey, we asked raters to evaluate appropriateness by means of a 9-point scale slightly modified from a format developed by the RAND Corporation for ascertaining expert consensus. 30 (In some questions, we asked raters to write in answers.) We explicitly asked the raters to consider what would be the best possible approach for the first few hours of intervention in order not to have a negative impact on the clinician’s ability to diagnose and then treat the disorder in continuing care. We asked the experts to draw on both their knowledge of the research literature (we did not provide a literature review) and their best clinical judgment in making their ratings, but not to consider financial cost. We presented the rating scale to the experts with the anchors shown in Figure 1.

Figure 1.
Figure 1.:
The rating scale

Figure 2 shows Survey Question 22 as an example of our question format.

Figure 2.
Figure 2.:
Sample survey question

Composition of the Expert Panel

We identified 52 leading American experts in psychiatric emergency medicine. The experts were identified from several sources: members of the American Association of Emergency Psychiatry who are board certified and have administrative responsibilities for a psychiatric emergency service as well as academic affiliations and individuals who have published research on emergency psychiatry or psychopharmacology. We offered a $500 honorarium. Panelists reported taking 2 or more hours to complete the survey.

We received responses from 50 of the 52 experts to whom the survey was sent. Of the respondents, 49 hold an M.D. degree and 1 a D.O. degree. 76% are male. The experts’ mean age was 47 years (S.D. 7.2, range 36–66), with a mean of 16 years in practice (S.D. 7.9, range 4–41) and a mean of 11 years in emergency psychiatric care (S.D. 5.5, range 1–25). 70% reported spending at least half their work time seeing patients. 59% practice in a general hospital, 18% in a psychiatric hospital, and 4% in a V.A. medical center. Of those practicing in a general hospital, 61% work in a separate psychiatric emergency service, 17% in a component of the medical emergency department, and 22% as consultants to the medical emergency department. The respondents reported the following percentages of patients by diagnostic group:



78% of the respondents’ departments sponsor clinical research. 24% of their psychiatric emergency services evaluate fewer than 250 patients each month, 41% 250–500 patients, and 35% more than 500 patients. The authors acknowledge that many panel members were drawn from urban academic medical centers, which may affect the applicability of their recommendations for rural settings. The respondents reported that a mean of 31% of patients were treated involuntarily (S.D. 29%, range 0%–100%).

Data Analysis for Options Scored on the Rating Scale

For each option, we first defined the presence or absence of consensus as a distribution unlikely to occur by chance by performing a chi-square “goodness-of-fit” test 31 (P < 0.05) of the distribution of scores across the 3 ranges of appropriateness (1–3, 4–6, 7–9). Next we calculated the mean and 95% confidence interval (C.I.). A categorical rating of first, second, or third line was designated based on the lowest category in which the C.I. fell, with boundaries of 6.5 or greater for first line, and 3.5 or greater for second line. Within first line, we designated an item as “treatment of choice” if at least 50% of the experts rated it as 9. Of the 739 options rated on the 9-point scale, consensus was reached on 617 options (83%) as defined by the chi-square test. When there was no first-line recommendation, we chose the highest-rated second-line option as the “preferred” treatment and indicated this in the guideline.

Displaying the Survey Results

To document the basis for the guideline recommendations and allow clinicians to examine the ratings for all options, the complete results of the survey are displayed graphically on pages 51–88 of the full guideline publication. 17 The type of graphic used is shown in Figure 3, which presents the results of Question 22 shown in Figure 2. The C.I.s for each treatment option are shown as horizontal bars and the numerical values are given in the table on the right. The following graphical conventions are used:


Figure 3.
Figure 3.:
Results of survey question 22

First-line treatments are those strategies that came out on top when the experts’ responses to the survey were statistically aggregated. These are options that the panel felt were usually appropriate as initial treatment for a given situation. Treatment of choice, when it appears, is an especially strong first-line recommendation (having been rated as “9” by at least half the experts). In choosing between several first-line recommendations, or deciding whether to use a first-line treatment at all, clinicians should consider the overall clinical situation, including the patient’s prior response to treatment, side effects, general medical problems, and patient preferences.

Second-line treatments are reasonable choices for patients who cannot tolerate or do not respond to the first-line choices. A second-line choice might also be used for initial treatment if the first-line options are deemed unsuitable for a particular patient (e.g., because of poor previous response, inconvenient dosing regimen, particularly annoying side effects, general medical contraindication, potential drug interaction, or if the experts did not agree on a first-line treatment). For some questions, second-line ratings dominated, especially when the experts did not reach any consensus on first-line options. In such cases, to differentiate within the pack, we labeled those items whose C.I.s overlap with the first-line category as “high second line.”

Third-line treatments are usually inappropriate or used only when preferred alternatives have not been effective.

From Survey Results to Guidelines

After the survey results were analyzed and ratings assigned, the next step was to turn these recommendations into user-friendly guidelines. We distinguished two levels, preferred options and alternate options, that generally correspond to first- and higher second-line ratings. Whenever the guideline includes more than one treatment in a rating level, we listed them in the order of their mean scores. As an example, the results of question 22 presented above were used to develop Guideline 2B: Interventions for an Imminently Violent Patient (Table 1). As initial strategies in this situation, the treatments of choice are verbal intervention, voluntary medication, and a show of force. As noted in the legend of the guideline table, bold italics indicate a treatment of choice rating, an especially strong opinion. Other first-line options are offering food, beverage, or other assistance and emergency medication. High second-line (alternate) options are the use of restraint or locked or unlocked seclusion.

Table 1
Table 1:
Guideline 2B: Interventions for an imminently violent patient


We have employed the expert consensus survey method in an attempt to describe an inherently complex, nonlinear process in which a variety of actors are potentially engaged in a number of conflicting parallel processes within a compressed time frame. Furthermore, the clinical problem we are addressing here differs from others for which these methods have been used. 20–28 Most of the treatment algorithms on which previous expert consensus guidelines have been based begin with a diagnosis, whereas this set of guidelines must deal with situations in which the diagnosis is unknown. Hence, many of these guidelines on behavioral emergencies are derived from the results of questions that involved forced decisions based on various assumptions about urgency, cooperation, amount of available information, diagnostic confidence, and individual risk factors.

What do the survey results tell us about the state of optimum practice in treating behavioral emergencies? In the following sections, we summarize the key recommendations from the guidelines and consider how the experts’ recommendations relate to the available research literature. A summary of key recommendations is presented in the Treatment Selection Algorithms in the Appendices at the end of this article (pp. 37–38).

Readers are referred to a supplement published in 2000 for more detailed discussions of the research literature on the acute care of agitated psychotic patients. 32 Note that literature in this area is relatively limited, because studies of agitation in emergency settings are difficult to justify ethically and are also difficult to conduct from a practical point of view. This was a major reason why this survey of expert opinion was undertaken.

Initial Assessment

The goal in a behavioral emergency is to facilitate the resumption of a more typical patient-physician relationship, with an emphasis on informed consent and long-term treatment outcome. Target symptoms associated with agitation interfere with assessment and treatment during a period when immediate intervention appears to be needed because of dangerous behavior or warning signs of such behavior. Since assessment clearly plays a key role in selecting the most appropriate intervention in a behavioral emergency, we asked the experts about the kinds of assessments they considered most appropriate.

A key step in the initial evaluation is to identify the medical etiology for the agitation, if one is present. This is especially important, because available data suggest that delirium, in particular, should be managed according to the underlying etiology, if it can be identified. If the psychiatric emergency service personnel are responsible for performing the initial medical evaluation, the experts considered it most important to obtain vital signs, a medical history, and perform a visual examination of the patient. They also considered a urine toxicology screen and a cognitive examination (e.g., a Mini-Mental State Examination) key assessments to perform. If the patient is a woman of childbearing age, the experts also recommended obtaining a pregnancy test, since this will have a bearing on subsequent treatment selection, especially if medication is needed. The experts gave somewhat less support to more complete forms of physical examinations, probably reflecting issues related to time constraints and availability of personnel. Obviously, the level of examination will depend on the specific signs and symptoms with which a patient presents. More complete evaluations will be indicated in some circumstances, and may also be indicated later in the patient’s treatment.

According to the CMS interim final rules, the distinction between what is considered a “chemical restraint” versus a treatment appears to hinge on whether medication is being given as part of a plan of care for the patient’s condition or merely to control the patient’s behavior. The CMS document specifies that “A drug used as a restraint is a medication used to control behavior or to restrict the patient’s freedom of movement and is not a standard treatment for the patient’s medical or psychiatric condition” (42CFR 482.13(e)). 14 A subsequent CMS bulletin 33 (for “guidance only”) appears to suggest that the distinction between a chemical restraint and treatment is the extent to which the patient has been assessed and medication prescribed as part of a plan of care. To create such a plan of care, the experts considered a brief assessment leading to the determination of a general category of presentation (e.g., intoxication, psychosis) adequate. A more comprehensive assessment leading to a specific diagnosis was also supported but may be impractical for various reasons.

Before intervening with medication, the experts considered it most important to determine if there is a causal medical etiology that should be managed first, to review the patient’s records if available, and to determine if substance abuse may be complicating the presentation. The experts considered it appropriate but less imperative to obtain a history of the patient’s previous medication response, if this information is available, and to determine the patient’s treatment preferences.

What Is Considered a Treatment Versus a Restraint

We asked the experts to rate a number of interventions in terms of whether they consider them a form of treatment. We defined a treatment to mean an intervention that follows from an assessment of the patient and a plan of care intended to improve the patient’s underlying condition. Nearly all of the experts strongly agreed that medication used to treat a specific psychiatric diagnosis would be considered a treatment rather than a chemical restraint. A majority of the experts also felt that medication used to treat symptoms, even in the absence of a clear diagnosis, would be considered a treatment. There was less agreement on how to view other interventions, such as unlocked or locked seclusion or physical restraint.

In a separate question, we asked the experts how strongly they agreed or disagreed with a number of statements about what can be considered a treatment. As described in the previous section, the CMS interim final rules specify that a medication must be prescribed as part of a plan of care to be considered a treatment. Three quarters of our panel rejected the notion that such a plan of care is necessary to consider medication a treatment. Instead they endorsed the idea that administering medication in a behavioral emergency is a form of treatment and comports with the standard of care. They thus appear to be more in agreement with the JCAHO guidelines on restraint and seclusion, 15 which reject the concept of chemical restraint, maintaining instead that if a medication is used to treat behavioral symptoms, then it can be considered a treatment.

Voluntary Versus Involuntary Treatment

We also asked the experts about what constitutes voluntary treatment. For the most part, the experts felt that any dose of oral medication to which a patient assents in an emergency situation can be considered voluntary. They rejected the idea that the situation is so coercive that any medication must be considered involuntary even if it the patient appears to accept it.

Defining a Behavioral Emergency

We asked the experts specifically what types of presentations they felt justify use of emergency intervention (involuntary medication or physical restraint). Their recommendations are summarized in Table 2. The experts would always consider it appropriate to initiate an emergency intervention when a patient is directly threatening or assaultive.

Table 2
Table 2:
Guideline 2A: Range of behavioral emergencies and appropriate responses

We also asked the experts what methods they use to document the need for an emergency intervention. Most of the experts (83%) use unstructured clinical observation and assessment; a good number (39%) also use structured checklists. Only four of the experts indicated that they use structured rating scales for this purpose.

Selecting Emergency Interventions

We attempted to determine the relative value of different initial strategies in dealing with a patient who appears imminently violent (Table 1). There was strong support for efforts to reduce tension and deescalate the crisis by approaching the patient in a calm and solicitous manner. Therefore, the experts recommended beginning with the least paternalistic or aggressive approaches—verbal intervention, offering food, beverage, or other assistance, or voluntary medication—before moving to more intrusive strategies. The experts believed these initial interventions are associated with the least risk of acute injury and negative long-term sequelae. Their next step would be a show of force. If those interventions were not successful, the experts would then consider use of emergency medication or physical restraints or seclusion. They did not recommend leaving the patient alone, which the experts considered to be associated with the greatest risk of injury and negative sequelae.

In terms of the goals of different interventions, the experts considered safety issues (e.g., control of aggressive behavior) somewhat more important in the short-term, whereas they placed more emphasis on collaboration between patient and physician and honoring the wishes of the patient in achieving the most favorable long-term outcome.

We asked the experts about their perceptions of consumer preferences. The experts believed that consumers consider oral medication most acceptable, followed by injectable medication or seclusion, but that they do not favor the use of physical restraints. Among the various classes of medications, the experts believed that benzodiazepines and atypical antipsychotics are most acceptable to consumers. These responses agree with the results of a survey of patient preferences in a psychiatric emergency service, which found that patients favored medication over restraint or seclusion by a 2:1 margin, that their first choice was generally benzodiazepines, and that almost one third of the respondents considered conventional antipsychotics a last resort. 34

Use of Restraint

When to use restraint.

As noted above, the experts consider restraint a last resort. The CMS interim rules 14 specify that use of restraint for “managing behavioral emergencies is allowed only when all less restrictive measures have failed and unanticipated severely aggressive or destructive behavior places the patient or others in imminent danger...” We asked the experts about situations in which they felt that the use of physical restraints was appropriate. Their recommendations are shown in Table 3. 47% of the experts reported that emergency interventions (parenteral medication, restraint, seclusion) were required for 1%–5% of patients seen in their services, and 32% said they were required for 6%–20% of patients. This means that, in this sample, more than 80% of patients are managed without the need for parenteral medication, restraint, or seclusion.

Table 3
Table 3:
Guideline 3A: When to use physical restraints

Personnel issues.

There are basically three different sets of personnel involved in the restraint process. First, someone makes the decision to initiate restraint. Then, a group of staff members physically places the patient in restraints. Finally, a face-to-face assessment is done to evaluate the need for restraint. The CMS interim rules state that hospitals should have a protocol “to specify who can initiate restraints or seclusion in an emergency prior to obtaining a physician’s or licensed independent practitioner’s order.” They further specify that “a physician or other licensed independent practitioner must see and evaluate the need for restraint or seclusion within 1 hour after the initiation of the intervention.” However, this regulation has caused some confusion, since the categories of providers who are licensed as independent practitioners vary from state to state. To try to clarify the situation, we asked the experts who they believe can most appropriately initiate restraint and who should perform the subsequent face-to-face evaluation. In both situations, they believed that attending psychiatrists or psychiatric residents, preferably with training and/or experience in emergency psychiatry, or nurses with psychiatric experience or advanced training are the most appropriate personnel both to initiate restraint and perform subsequent face-to-face evaluation. It should be noted that there was less support for psychologists and physicians in other specialties performing these functions, and that the experts generally did not consider it appropriate for social workers, licensed counselors, or unlicensed clinical staff to perform these functions, given the current state of training of these categories of providers. The experts were in agreement with the CMS rules that 1 hour is the most appropriate minimum interval between when a patient is put in restraints (or seclusion) and the initial face-to-face evaluation is performed. There have been some objections to this 1-hour rule, mainly because of logistical difficulties, from the American Medical Association and the American Psychiatric Association; it is therefore interesting that the experts confirmed that this is the appropriate standard of care.

We also asked the experts some questions about the mechanics of placing and maintaining a patient in restraints. The experts considered nursing staff and trained security officers the most appropriate personnel to participate in actually placing a patient in restraints, although they would also consider physicians sometimes appropriate. They do not consider untrained security personnel appropriate to perform this function. They would generally use leather restraints, but would also consider cloth or other soft restraints, with less support for the use of plastic and velcro restraints or restraint chairs.

Duration of episode.

The CMS interim rules specify that restraint orders are limited to 4 hours for adults. 14 The experts favored an interval of 2 hours (69% first line) but also supported 4 hours (57% first line). This may reflect the pattern of regulations already in place in the states where the panel members practice.

Intensity of monitoring.

The experts recommended continuous monitoring while a patient is in restraints (either in person or using a combination of audiovisual and direct personal observation). Many of the experts also considered in-person evaluation at 15 minute intervals reasonable, but they did not support longer intervals (30–60 minutes) between observations. The CMS interim final rules specify continuous audio and visual monitoring while in restraints. 14 The JCAHO regulations specify continuous in-person monitoring for individuals in restraints (with continuous audiovisual monitoring allowed after the first hour for patients in seclusion). 15

Use of medication while in restraints.

We also asked the experts about the appropriateness of using medication while a patient is in restraints. If the patient becomes calmer in restraints, the experts were divided as to whether to use no medication or to offer oral medication. They did not recommend parenteral medication in this situation. However, if a patient continues to be violent and agitated in restraints, the experts strongly supported the use of parenteral medication in combination with the restraints and would also consider using oral medication in this situation. They did not consider it appropriate to leave such a patient unmedicated in restraints. Overall, these recommendations appear to reflect the experts’ view that the goal in this situation is to use medication to minimize time in and/or complications of restraints.

Use of Medications

Factors influencing selection.

The factors the experts considered most important in selecting a specific medication for use in a behavioral emergency are shown in Table 4.

Table 4
Table 4:
Guideline 4A: Factors determining initial choice of medication


We asked the experts to compare the effectiveness for decreasing agitation and the level of sedation associated with four types of medications that are often used in the psychiatric emergency setting: droperidol, lorazepam, haloperidol, and atypical antipsychotics. The experts considered droperidol, lorazepam, and haloperidol the most effective agents for decreasing agitation, followed by the atypical antipsychotics. The experts considered lorazepam and droperidol most sedating, followed by haloperidol and the atypical antipsychotics.

There is very little evidence in the literature of differential effectiveness among the available agents that cannot be accounted for by dosage or pharmacokinetics. The largest number of studies have been done with haloperidol, 35,36 though a number of studies have looked at other antipsychotics, including thiothixene, 37 molindone, 38 and loxapine, 39,40 and have found comparable effectiveness with haloperidol. Although chlorpromazine is often mentioned for behavioral emergencies because of its sedative side effects, haloperidol has been found to be superior to chlorpromazine at usual doses. 41,42

Droperidol is a butyrophenone approved by the U.S. Food and Drug Administration (FDA) that is available only for parenteral administration and has been used primarily in anesthesia. There is strong anecdotal support for the use of droperidol as a calming agent in behavioral emergencies. 43 One of the few placebo-controlled studies of droperidol demonstrated its effectiveness for agitation. 44 However, only three studies comparing droperidol to other agents have been done, all of which have methodological problems. 45–47 These studies seem to suggest that droperidol is certainly faster and perhaps more potent but not necessarily more efficacious.

The atypical antipsychotics are associated with a much lower risk of extrapyramidal side effects (EPS) than high-potency conventional antipsychotics. Although they are recommended as the first-line agents for treatment of schizophrenia in most situations, 23 they have not up to now been as widely used as the conventional agents in emergency settings. This may be due in part to the slower titration schedules recommended for some of these agents and the fact that, until very recently, none of the atypical agents was available in an I.M. formulation.

Data on the use of atypical agents in psychiatric emergency settings or on their use to treat acute aggression or agitation are very limited. However, a number of studies in more chronic care settings have demonstrated that the atypical antipsychotics appear to be more effective than the conventional antipsychotics in treating aggression and agitation. 48–56

A recent study examined the relative efficacy, safety, and tolerability of oral risperidone (liquid concentrate) plus lorazepam versus I.M. haloperidol plus lorazepam. 57 Both treatment groups showed improvement over time, with no significant differences between the groups. One patient in the haloperidol group developed a dystonic reaction; there were no adverse reactions in the risperidone group.

I.M. ziprasidone was approved in the fall of 2001. Olanzapine is available in a wafer that dissolves to form a liquid in the oral cavity and an I.M. formulation is in late-stage development. These I.M. formulations were investigational at the time the survey was done and the guidelines were being developed. Published studies have appeared for acute I.M. forms of both olanzapine and ziprasidone but have focused mainly on the treatment of psychosis and safety issues, rather than agitation or behavioral emergencies. 58–60[Editor’s note: for a detailed discussion of recent studies with the atypical antipsychotics in the treatment of agitation, see the article by Currier on p. 59]

Studies concerning the use of benzodiazepines in psychiatric emergencies suggest that they are at least as effective as haloperidol alone. Most of the studies have been done with lorazepam, 47,61–65 but controlled data have also been published concerning midazolam, 66 clonazepam, 67 and flunitrazepam. 68 Studies comparing 5 mg of haloperidol with 2 mg of lorazepam found that the two agents were equal on some measures, 61,63,64 but that 2 mg of lorazepam was superior on measures of aggression 63 and clinical global improvement. 64 Flunitrazepam 1 mg was compared with haloperidol 5 mg and found to be superior using the Overt Aggression Scale as a measure of outcome. 68 Midazolam 5 mg was reported to be superior to haloperidol 10 mg in its effect on a measure of motor agitation. 66 These studies suggest that benzodiazepines used at the doses that are currently usual in emergency settings may be more effective than haloperidol. Battaglia et al. 61 found lorazepam used alone to be more sedating than haloperidol used alone.

Use of combination treatment.

The most common medication strategy in psychiatric emergency settings today is the use of haloperidol and lorazepam in combination (usually 5 mg haloperidol and 2 mg of lorazepam in the same injection). 69 Although this strategy is generally considered to be safe and effective, research evidence concerning this practice is very limited, with only two randomized, controlled studies comparing the use of the combination versus the component agents alone published to date. 61,62 These studies found that the combination was more effective early in treatment, but that differences in treatment tended to disappear within 2–4 hours, perhaps because additional doses were given in the interval. One study 57 has also been done that compared a combination of haloperidol and lorazepam with a combination of risperidone and lorazepam and found they were equally efficacious (see discussion in preceding section).

When asked about the advantages of using combination treatment, the experts considered the most important potential benefits to be greater efficacy for symptoms of arousal, faster onset of action, and reduced side effect liability. The authors noted that the limited literature is inconclusive as to whether combination treatment does indeed produce these benefits. However, the literature does appear to support the advantage of being able to use lower doses of each of the component medications, thus reducing the liability for side effects, especially from haloperidol 65; the experts also considered this a benefit of this strategy.


Time to onset is also an important characteristic. In managing the agitated and potentially violent patient, faster onset may reduce the chance of injuries and the need for, or time in, restraints. We therefore asked the experts to consider the speed of onset of a number of medications and formulations that are used in psychiatric emergency settings.

The experts’ recommendations agree with the literature concerning the rapidity of effect of I.M. formulations of lorazepam, midazolam, and haloperidol, while I.M. diazepam and chlordiazepoxide are absorbed slowly and erratically, so that they are not recommended for this use. 70 The authors noted that published pharmacokinetic data suggest that some oral preparations are absorbed more rapidly than some parenteral preparations. 70 The onset of I.M. haloperidol is usually reported to be 30–60 minutes and it has been found that the effect of haloperidol was still rising at 1 hour when the offset of the droperidol was already beginning. 46 In the same study, it was reported that subjects treated with droperidol spent significantly less total time in the emergency department than those treated with lorazepam (5.9 versus 8.6 hours). These rapid and profound effects are doubtless the reason this agent is commonly used in certain parts of the country. However, droperidol is not considered to be a part of the usual treatment of any psychiatric condition, which would seem to place it more in the class of a chemical restraint than a medication treatment. It should also be noted that droperidol was recently withdrawn from the European market 71 and has received a “black box” warning in the United States due to concerns about prolongation of the QTc interval. 72

Route of administration.

The experts considered speed of onset and reliability of delivery the two most important factors to consider in choosing a route of administration; they also considered patient preference quite important. When asked which route of administration they would prefer to use to treat a behavioral emergency, assuming the medication is available in both oral and I.M. formulations, the experts gave their highest ratings to oral liquid concentrates, orally dissolving formulations, and I.M. formulations. Oral tablets were not preferred, presumably because of slower onset and the risk of “cheeking.”

The experts’ recommendations are consistent with the results of another recent survey of approximately 50 directors of psychiatric emergency services, in which the majority advocated the use of oral medication whenever possible, with liquid formulations preferred to tablets because of their more rapid onset and because it is easier to verify compliance with liquid medication. 10 In that same survey, the medical directors estimated that only 1 in 10 patients in their emergency services require an injection. It has been reported elsewhere that most agitated patients will assent to oral medications. 73 As noted earlier, the experts felt that consumers’ first preference in an emergency situation is oral medication. The CMS rules 14 specify that “chemical restraint” be considered a last resort, suggesting that oral medication should be offered to the patient first, if possible.

When asked about factors that limit their willingness to use an I.M. formulation, the experts considered risk of side effects, mental or physical trauma to the patient, and the danger of compromising the patient-physician relationship most important.

When asked about their preferences among the oral atypical antipsychotics, the experts preferred risperidone and olanzapine, with quetiapine an alternate choice (note that ziprasidone had not yet been approved at the time of this survey and was therefore not included as an option) and would prefer to use a liquid formulation of the atypical antipsychotic.

Dose and frequency.

The experts’ recommendations concerning dosing levels and intervals between doses are summarized in Guideline 4H (Table 5). The experts recommended a minimum single dose of 1.0 mg and a maximum single dose of 10 mg for haloperidol; in a separate question the experts indicated that they considered a dose equivalent to 2.0–5.0 mg haloperidol most appropriate as initial treatment (either oral or parenteral) for a patient with a behavioral emergency. The experts recommended a minimum single dose of 0.5 mg of lorazepam and a maximum single dose of 2 mg.

Table 5
Table 5:
Guideline 4H: Dosing levels

Only three studies have compared different doses of medication for agitation, all of which looked at haloperidol. 35,36,41 Baldessarini et al. 74 combined the results of these studies and produced a dose-response curve. Their results suggest that a single dose of 7.5–10 mg of haloperidol might be expected to produce the most benefit possible with fewest side effects, and that higher doses, which are associated with an increased incidence of side effects, are not likely to produce much additional benefit. These findings are consistent with the experts’ recommendations.

The literature concerning the most appropriate initial doses of benzodiazepines for agitation is very limited. Most published studies concerning the use of lorazepam in agitation have used a dose of 2.0 mg. There is, however, some controversy in the literature as to the most appropriate dose of benzodiazepine with which to begin in a behavioral emergency. Bienek et al. 62 discussed the use of a higher initial dose of 3–4 mg, which would seem to agree with the results of Baldessarini’s meta-analysis, 74 which supported the use of 7.5 mg haloperidol as a starting dose.

Selecting Interventions Based on Etiology/Diagnosis

Agitation in patients who present in a psychiatric emergency setting may be associated with several different etiologies. Identifying the underlying cause of the patient’s agitation can help the clinician more accurately tailor the intervention to the presentation. We therefore asked the experts to recommend the most appropriate interventions for patients with agitation due to three general classes of suspected etiology: a general medical condition (e.g., delirium, HIV encephalopathy), substance intoxication (e.g., with cocaine, PCP), and a primary psychiatric disturbance (e.g., schizophrenia, mania). For each situation, we asked the experts what general strategies they would begin with during the first hour after presentation in 1) a very agitated, uncooperative patient whose behavior appears to require immediate intervention to prevent injury to self or others, and 2) a patient who is agitated but responsive to direction and does not appear to present an immediate danger to self or others (see Appendix, pp. 37–38).

General medical etiology.

If a patient is very confused and a general medical etiology is suspected, the experts recommended taking vital signs, gathering history from the family or other sources, talking to the patient if possible, performing a visual examination of the patient, requesting a consultation from the medical emergency department, and performing tests such as pulse oximetry, blood glucose and a toxicology screen. If the patient’s behavior appears to require immediate intervention to prevent danger to self or others, the experts would next consider intervening with physical restraints, administering parenteral medication or offering oral medication, and performing a focused or cursory physical examination. If the patient is responsive to direction and does not appear to pose any immediate danger to self or others, the experts considered performing a focused physical examination a first-line strategy, presumably because the patient is more likely to cooperate with such an examination. They did not recommend the use of parenteral medication or physical restraints for a cooperative patient.

If it is decided to use medication, either oral or parenteral, to treat agitation in a behavioral emergency that appears to have a general medical etiology, the majority of experts would begin with a conventional antipsychotic, a benzodiazepine, or a combination of the two. Among oral medications, 43% also considered risperidone a first-line option in this situation. If a parenteral medication is needed, 44% also considered droperidol first line.

Available data suggest that delirium should be managed according to the underlying etiology, if this can be identified. Delirium due to a general medical etiology has usually been treated with high-potency conventional antipsychotics. For example, Breitbart et al. found that conventional antipsychotics were superior to lorazepam in efficacy and side effects in a group of prospectively defined patients with AIDS delirium. 75 As reported in the Expert Consensus Guidelines for the Treatment of Agitation in Older Persons with Dementia, 22 a panel of experts on the treatment of dementia in older patients recommended conventional high-potency antipsychotics for delirium due to a general medical etiology (e.g., congestive heart failure, urinary tract or upper respiratory infections) in patients with dementia, with risperidone a high second-line choice.

Substance intoxication.

If it is strongly suspected that the patient’s agitation is associated with substance intoxication and the patient’s behavior appears to require immediate intervention to prevent danger to self or others, the experts recommended attempting to take vital signs, talking to the patient if possible, gathering history from the family or other sources, performing tests such as a toxicology screen, and a visual examination of the patient. High second-line interventions in this situation (presumably interventions the experts would recommend performing next) are offering oral medication or administering parenteral medication, performing a cursory physical examination, and testing for breath alcohol content.

Note that the use of restraint received higher ratings for an uncooperative and imminently violent patient whose symptoms appear to have a medical etiology (e.g., a patient with delirium) (restraint rated first line by 67% of the experts) than for a patient whose symptoms appears to be related to substance intoxication (restraint rated first line by 51% of the experts but third line by 43%). This difference may reflect a number of concerns, including worry that a delirious patient may wander, concern about the risk of vomiting and aspiration in an intoxicated patient, and a reluctance to use medication that might increase confusion in a delirious patient.

The experts’ recommendations were similar for a patient who is responsive to direction and does not appear to pose any immediate danger to self or others, except that they considered testing for breath alcohol content first line in this situation and they would be more inclined to perform a focused physical examination and to observe the patient and wait for the substance intoxication to resolve or else to offer oral medication. The experts did not recommend the use of parenteral medication or restraint for a cooperative patient who does not appear to pose a danger to self or others.

If it is decided to use oral or parenteral medication to treat agitation associated with substance intoxication, the experts gave the strongest support to the use of a benzodiazepine alone. For stimulant or hallucinogen intoxication, the next choice would be a benzodiazepine plus a high-potency conventional antipsychotic. A report in the literature suggests that individuals who abuse stimulants may be more prone to EPS, 76 which may be the reason that the experts preferred benzodiazepines in this situation (i.e., antipsychotics are not likely to have any special benefits for this population but may be more likely to cause EPS). Cocaine toxicity may also involve seizures, and the experts may have preferred benzodiazepines to antipsychotics for their protective effect in this situation. The preference for benzodiazepines in the treatment of hallucinogen intoxication may reflect the experts’ recognition that some hallucinogens are anticholinergic and their wish to avoid treating the patient with another drug with anticholinergic properties or that might require the use of an adjunctive anticholinergic agent for EPS.

The experts had no first-line recommendations for treatment of agitation due to alcohol intoxication, but did rate a benzodiazepine alone as high second line. The slight preference for benzodiazepines for patients intoxicated with alcohol may reflect the fact that a component of withdrawal may be contributing to the agitation for which the benzodiazepine might be specifically indicated. The American Psychiatric Association Guideline for the Treatment of Substance Use Disorders77 recommends benzodiazepines for alcohol withdrawal states. The CMS bulletin 33 referred to earlier in this article also mentions the use of benzodiazepines for behavioral disturbances associated with alcohol withdrawal as an appropriate use of medication for treatment rather than as a chemical restraint.

There was not much support for the use of any medication in patients intoxicated with opioids. This may reflect the belief that patients intoxicated with opioids are usually not agitated enough to risk adding a medication that might cause unwanted sedation or respiratory depression.

Primary psychiatric disturbance.

If the presentation or history suggest that the patient’s agitation is due to a primary psychiatric disturbance and the patient is uncooperative and appears to require immediate intervention to prevent danger to self or others, the experts recommended attempting to take vital signs, talking to the patient if possible, gathering history from the family or other sources, administering parenteral medication or offering oral medication, a visual examination of the patient, and performing tests such as a toxicology screen. High second-line interventions in this situation (presumably interventions the experts would recommend performing next) are intervening with physical restraints to ensure patient safety and performing a cursory physical examination.

The experts’ recommendations were similar for a patient who is responsive to direction and does not appear to pose any immediate danger to self or others, except that the experts did not recommend using parenteral medication or restraint in this situation.

The experts’ recommendations for medication to treat agitation that appears to be due to a primary psychiatric disturbance depended on the provisional diagnosis. We first describe their recommendations for oral medications and then review those for parenteral agents.

There are situations in which an immediate response is required but no data are available on which to base even a provisional diagnosis. If it is decided to use an oral medication in this situation, the experts considered a benzodiazepine alone first line and a benzodiazepine plus a high-potency conventional or atypical antipsychotic high second line. There was no first-line consensus on choice of parenteral medication when there are no data on which to base a more specific provisional diagnosis; high second-line options were a benzodiazepine alone or in combination with a high-potency conventional antipsychotic.

If it is decided to use an oral medication to treat a patient with a provisional diagnosis of schizophrenia or mania, the experts recommended a benzodiazepine plus a high-potency conventional or atypical antipsychotic. High second-line options for schizophrenia or mania were monotherapy with risperidone, a high-potency conventional antipsychotic, or olanzapine. Monotherapy with a benzodiazepine was also a high second-line option for a patient with a provisional diagnosis of mania.

There were no first-line recommendations for oral medication for a provisional diagnosis of psychotic depression or personality disorder. High second-line recommendations for psychotic depression were a benzodiazepine used either in combination with an atypical or conventional antipsychotic or alone, or risperidone alone; a benzodiazepine alone was rated high second line for personality disorder. A benzodiazepine alone was the first-line recommendation for a provisional diagnosis of posttraumatic stress disorder (PTSD).

If it is decided to use a parenteral medication to treat a patient with a provisional diagnosis of schizophrenia, the experts recommended a benzodiazepine plus a high-potency conventional antipsychotic as first line, with a high-potency conventional antipsychotic alone a high second-line option.

If it is decided to use a parenteral medication for a patient with a provisional diagnosis of mania, a benzodiazepine in combination with a high-potency conventional antipsychotic or used alone was first line, with a high-potency conventional antipsychotic alone high second line. For a provisional diagnosis of psychotic depression, a benzodiazepine plus a conventional antipsychotic was first line, with a benzodiazepine alone a high second-line option for parenteral treatment. There were no first-line recommendations for a provisional diagnosis of personality disorder; a benzodiazepine alone or in combination with a high-potency conventional antipsychotic was high second line. For a provisional diagnosis of PTSD, a benzodiazepine alone was the first-line recommendation, with a benzodiazepine combined with a high-potency conventional antipsychotic high second line.

Note that oral high-potency conventional antipsychotics used alone did not receive much support in most situations and that the experts gave equal or greater support to the atypical antipsychotics for patients with a primary psychiatric etiology. These results are consistent with the recommendations presented in the Expert Consensus Guidelines on schizophrenia 23 and mania, 25 in which atypical antipsychotics were generally preferred over conventionals for the treatment of schizophrenia and in which atypicals received equal or greater support for use in psychotic mania and were preferred for the treatment of nonpsychotic mania. See Ghaemi 78 for a review of recent findings concerning the role of atypical antipsychotics in the treatment of bipolar disorder.

Among parenteral medications, high-potency conventional antipsychotics used alone received somewhat more support, perhaps because of the lack of injectable atypical antipsychotics at the time of the survey. However, they were generally viewed as inferior to benzodiazepines alone.

In a survey of emergency psychiatrists, it was reported that, if a mood stabilizer is needed in this setting, 90% would use divalproex/valproate, while only 8% would choose lithium and only 2% another mood stabilizer. 10 We did not, therefore, ask about choice of mood stabilizer in this survey, but we did ask about dosing strategies for divalproex. The experts clearly favored divalproex dosing strategies that employ higher doses over usual titration. They would recommend either beginning with 20 mg/kg and continuing until blood levels are available or starting with a loading dose of 30 mg/kg for 2 days, followed by 20 mg/day beginning on day 3. Factors that would encourage the experts to use a loading dose strategy for divalproex include history of response to divalproex in the past, normal liver function tests, and a desire on the part of the patient and family to try to avert hospitalization. The experts considered the use of a loading dose appropriate for all types of manic episodes, probably reflecting the fact that lithium is not generally used in emergency settings and that loading doses of divalproex may help to stabilize the patient more quickly. 79

Next Steps If There Is an Inadequate Response

If a patient was initially treated with a single agent, either a benzodiazepine alone or an antipsychotic alone, and there has not been an adequate response after 45–60 minutes, the experts recommended either proceeding to a combination of a benzodiazepine and an antipsychotic or giving another dose of the initial agent alone. They would also consider giving a dose of the medication that was not yet tried.

We also asked the experts when they would recommend changing medication strategies—switching to a different agent or using a combination of agents if they had begun with a single agent.

If the patient was initially treated with a single agent, the experts recommended a change of strategy after 2 doses of the single agent have been totally ineffective (i.e., the patient is still extremely agitated and uncooperative) or 3–4 doses have been only partially effective (i.e., the patient is somewhat calmer but is still not able to converse with caregivers or take oral medication).

If the patient was initially treated with a combination of an antipsychotic and a benzodiazepine, the experts appeared to be willing to continue the same treatment strategy somewhat longer, probably reflecting the more limited options available at this point. In this situation, they recommended a change of strategy after 3 doses of the combination have been totally ineffective or 4 doses have been only partially effective.

Safety and Tolerability

In general, differences in the effectiveness of the various medications in the first few hours are hard to discern. In this situation, considerations of safety and tolerance become more important in selecting a particular medication. As we noted earlier, it is very important to first do no harm.


We asked the experts about the most appropriate medication strategies for a pregnant woman who is agitated, psychotic, and unresponsive to direction and for whom immediate medical intervention is judged necessary to prevent harm to the mother or fetus or to reduce the deleterious effects that the stress of agitation may have on the maternal-fetal system. In this situation the experts clearly preferred a conventional high-potency antipsychotic (rated first line by 76% of the experts), probably reflecting the much larger database concerning the use of this type of agent and the lack of teratogenicity reported for high-potency conventional antipsychotics. 80 There was also some support for the use of benzodiazepines alone (rated first line by 40% of the experts) and for droperidol (rated first line by 35% of the experts). The experts’ recommendations concerning choice of antipsychotics for a pregnant patient agree with the recommendation for treating psychotic depression in pregnant women in the recently published Expert Consensus Guidelines on the Treatment of Depression in Women 2001.27 It is interesting that, while the FDA rates conventional and atypical antipsychotics similarly in their Use-in-Pregnancy ratings 81 (category C: “risk cannot be ruled out”), the experts were less willing to endorse the use of atypical antipsychotics, presumably because of less experience with these agents in pregnant women. Note that, in selecting an atypical antipsychotic for a pregnant woman in this setting, the experts showed a slight preference for risperidone.


The experts had no first-line consensus on the most appropriate medication strategy for a child who is unmanageable and violent. A low-dose benzodiazepine or an antihistamine received high second-line ratings in this situation. The experts’ responses probably reflect the desire to be as conservative as possible in terms of safety and to minimize antipsychotic exposure in treating a child. If an antipsychotic is needed, the experts showed a slight preference for risperidone or olanzapine over a conventional antipsychotic and they would prefer to use lower doses of the antipsychotic.

It should be noted that, while the experts supported the use of a combination of an antipsychotic and a benzodiazepine in a number of other emergency situations, they would not generally recommend use of combination medication for children.

Complicating conditions and side effects.

The experts’ recommendations for choice of medication classes when complications are present were consistent with the general literature. The experts would avoid using high-potency conventional antipsychotics in patients with a history of EPS. They were reluctant to use benzodiazepines in patients with a history of substance abuse/dependence or drug-seeking behavior. However, the authors noted that a benzodiazepine rather than an antipsychotic is recommended for a patient with a significant blood alcohol level, which probably reflects the experts’ concern about withdrawal syndromes and the risk of seizures. As noted earlier, this recommendation is consistent with the examples provided in the CMS bulletin concerning the treatment of alcohol withdrawal. 33 Note that benzodiazepines may be initiated even while alcohol is still present in the patient’s system. Benzodiazepines were also preferred for patients with a history of seizures (e.g., because of substance or alcohol abuse). Although concerns have been raised on theoretical grounds about the risk of respiratory depression when benzodiazepines are used in combination with alcohol or other sedatives and about the possibility of behavioral disinhibition with benzodiazepines, these concerns were not reflected in the high ratings the experts generally gave benzodiazepines throughout the survey nor are they supported by the research. 82 The experts would use benzodiazepines with caution in patients with chronic obstructive pulmonary disease or in frail older patients. It should also be noted that the experts preferred atypical antipsychotics to conventional antipsychotics for frail older patients. The experts preferred benzodiazepines to antipsychotics for patients with cardiac arrhythmia or conduction defects, probably because of concern about adverse effects on cardiac function. Atypical antipsychotics were preferred for patients with mental retardation/developmental delay. This agrees with the recommendations in the Expert Consensus Guidelines on the Treatment of Psychiatric and Behavioral Problems in Mental Retardation, in which atypical antipsychotics were strongly preferred over conventional antipsychotics for the treatment of agitation, aggression, or self-injurious behavior in this population. 26

We also asked the experts which of the atypical antipsychotics they would use, if it is decided to use an atypical, when a variety of complicating conditions are present. Their recommendations were consistent with the literature and the side-effect profiles of the specific medications. As would be expected, the experts did not recommend olanzapine for patients with diabetes or concern about weight gain and they preferred quetiapine for patients with a history of EPS. Risperidone was preferred for delirious patients, probably because the other atypicals have anticholinergic properties that might increase confusion and sedation. The experts did not rate any of the atypical antipsychotics first line for patients with seizures, probably reflecting the lack of significant differences in the potential for seizures among the atypical antipsychotics other than clozapine and also the experts’ preference for using benzodiazepines rather than antipsychotics in this patient population (see above).


When does an emergency exist? In this survey, the authors have “piloted” an empirical approach to this question—that is, most of the experts would advocate an emergency intervention in a given set of circumstances. This case-based approach might be expanded to include a wider variety of scenarios using this methodology.

Structured approaches to assessment should also be examined. A number of the experts in our expert panel reported that they use structured instruments, although few use rating scales per se. The lack of a good operational definition of agitation, much less of an emergency, also constitutes a significant barrier to both social and scientific progress in this area.

Given the existence of an emergency, who decides what to do? The core problems in a behavioral emergency are the perceived need to do something immediately and the lack of agreement between the individual at the center of the emergency and those responsible for managing it. This guideline is an effort to reach an agreement among providers. But how do we deal with the problem of agreement between patients and providers? Strategies with a narrow focus on the technical issues that determine short-term outcome may do so at the cost of relationship issues that influence long-term outcome. No medication that is now available has a large enough immediate effect to outweigh the importance of facilitating collaboration between patient and provider over time. Although the attitudes and behaviors that foster autonomy and respect are difficult to incorporate into guidelines, there is evidence from our survey that providers would use information concerning patient preferences if it were available.

Since communication in emergencies is problematic, a number of communication strategies can be envisioned. Strategies that are popular in the consumer community are advance directives, wellness and recovery action plans, and other methods of care planning driven by the individual consumer. Another solution would be to develop a “guideline” based on a consumer perspective—a document similar to this one that would attempt to represent the consumer attitudes and beliefs that might operate in a behavioral emergency if they could be communicated to providers. There is a striking lack of information on this topic. Research is also needed concerning the influence of race, ethnicity, and culture.

What data are available suggest that consumers understand the need for emergency interventions but often feel frightened and abandoned in the midst of them. In this respect, recent regulations that stress continuing contact with the patient during the episode and debriefing afterward may bring improvement in this area. The inclusion of consumer perspectives in the training of providers should help to sensitize providers and peer counselors. Advocates and families can also give providers proxy data that might help avoid or shorten episodes.

Can we also narrow the gap between consumers and providers on pharmacological management issues? A survey of a representative “expert panel” of consumers with personal experience with restraint, seclusion, or emergency medications would be very instructive. Given the relatively modest differences between available agents, consumer preferences, as manifested either in individual advance directives or credible consumer surveys, could play the deciding role. The results of this expert survey and of an older survey of consumers 33 suggest some convergence of opinion concerning the preferential use of benzodiazepines. A new survey of consumers that includes the newer atypical antipsychotics is needed. [Editor’s note: see the article by Allen et al., p. 39, for the results of a recent consumer survey and workshops.]

Even as tremendous strides are made in the treatment of psychiatric illness, behavioral emergencies will continue to be a problem because of their tendency to occur outside the usual context of health care. This will remain a difficult and controversial area of practice because it involves limitations on patient autonomy and control, although it is hoped that better practices will contribute to improvement in this area. This guideline is dedicated to a new climate of increased respect and an effort to move from control to care.


These guidelines can be viewed as an expert consultation, to be weighed in conjunction with other information and in the context of each individual patient-physician relationship. The recommendations do not replace clinical judgment, which must be tailored to the particular needs of each clinical situation. We describe groups of patients and make suggestions intended to apply to the average patient in each group. However, individual patients will differ greatly in their treatment preferences and capacities, history of response to previous treatments, family history of treatment response, and tolerance for different side effects. Therefore, the experts’ first-line recommendations certainly will not be appropriate in all circumstances.

We remind readers of several other limitations of these guidelines:

  1. The guidelines are based on a synthesis of the opinions of a large group of experts. From question to question, some of the individual experts would differ with the consensus view.
  2. We have relied on expert opinion precisely because we are asking crucial questions that are not yet well-answered by the literature. One thing that the history of medicine teaches us is that expert opinion at any given time can be very wrong. Accumulating research will ultimately reveal better and clearer answers. Clinicians should therefore stay abreast of the literature for developments that would make at least some of our recommendations obsolete. We hope to revise the guidelines periodically based on new research information and on reassessment of expert opinion to keep them up-to-date.
  3. The guidelines are financially sponsored by the pharmaceutical industry, which could possibly introduce biases. Because of this, we have made every step in guideline development transparent, reported all results, and taken little or no editorial liberty.
  4. These guidelines are comprehensive but not exhaustive; because of the nature of our method, we omit some interesting topics on which we did not query the expert panel.

Despite the limitations, these guidelines represent a significant advance because of their specificity, ease of use, and the credibility that comes from achieving a very high response rate from a large sample of the leading experts in the field.


Advances in public health do not always require technological breakthroughs or long periods of waiting for new data. Immediate gains can be made by increasing the speed with which best practices are implemented. Guidelines offer a rapid means for communicating a distillate of expert opinion. When reaching a clinical decision point, practitioners and patients can use guidelines to generate a menu of reasonable choices and then select the option that is judged best for each individual. This process drives the next round of expert opinion and the next round of empirical studies.


1. Allen MH. Building level I psychiatric emergency services: Tools of the crisis sector. Psychiatr Clin North Am 1999; 22:713–34.
2. Fisher WA. Restraint and seclusion: A review of the literature. Am J Psychiatry 1994; 151:1584–91.
3. Laughren T. Regulatory issues on behavioral and psychological symptoms of dementia in the United States. International Psychogeriatrics 2000; 12(suppl 1):331–6.
4. Snyder W. Hospital downsizing and increased frequency of assaults on staff. Hosp Community Psychiatry 1994; 45: 378–80.
5. Currier GW, Allen MH. Emergency psychiatry: Physical and chemical restraint in the psychiatric emergency service. Psychiatr Serv 2000; 51:717–9.
6. Meyerson AT, Delaney B, Herbert JD, et al. Do aspects of standard emergency care have potentially traumatic sequelae? Emergency Psychiatry 1998; 45:44–8.
7. Binder RL, McCoy SM. A study of patients’ attitudes toward placement in seclusion. Hosp Community Psychiatry 1983; 34:1052–4.
8. Sundram CJ, Stack EW, Benjamin WP. Restraint and seclusion practices in New York State psychiatric facilities. Albany, NY: New York State Commission on Quality of Care for the Mentally Disabled, 1994.
9. Way BB, Banks SM. Use of seclusion and restraint in public psychiatric hospitals: Patient characteristics and facility effects. Hosp Community Psychiatry 1990; 41:75–81.
10. Allen MH, Currier GW. Use of restraints and pharmacotherapy in psychiatric emergency services. Gen Hosp Psychiatry (submitted).
11. Altimari D, Blint DF, Weiss EM, et al. Deadly restraint. Hartford: Hartford Courant; 1998. (Summary available at
12. Medical Directors Council of the National Association of State Mental Health Program Directors. Reducing the use of seclusion and restraint: Findings, strategies, and recommendations. Alexandria, VA: National Association of State Mental Health Program Directors, 1999.
13. National Alliance for the Mentally Ill. Cries of anguish: A summary of reports of restraints and seclusions abuse received since the October 1998 investigation by The Hartford Courant. (Available at
14. U.S. Department of Health and Human Services, Centers for Medicare and Medicaid Services. Hospital conditions of participation for patients’ rights. 42CFR 482.13. Baltimore, MD: 1999 (Available at
15. Joint Commission on Accreditation of Healthcare Organizations. Comprehensive accreditation manual for hospitals (CMAH) (
16. Dawes SS, Bloniarz PA, Mumpower JL, et al. Supporting psychiatric assessments in emergency rooms. Albany, NY: Center for Technology in Government; 1995.
17. Allen MH. Currier GW, Hughes DH, et al. The expert consensus guideline series: Treatment of behavioral emergencies. Postgrad Med Special Report May 2001:1–88..
18. Djulbegovic B, Hadley T. Evaluating the quality of clinical guidelines: Linking decisions to medical evidence. Oncology 1998; 12( 11A):310–4.
19. Shekelle PG, Kahan JP, Bernstein SJ, et al. The reproducibility of a method to identify the overuse and underuse of medical procedures. N Engl J Med 1998; 338:1888–95.
20. McEvoy JP, Weiden PJ, Smith TE, et al. The expert consensus guideline series: Treatment of schizophrenia. J Clin Psychiatry 1996; 57(suppl 12b):1–58.
21. March JS, Frances A, Carpenter D, et al. The expert consensus guideline series: Treatment of obsessive-compulsive disorder. J Clin Psychiatry 1997; 58(suppl 4):1–72.
22. Alexopoulos GS, Silver JM, Kahn DA, et al. The expert consensus guideline series: Treatment of agitation in older persons with dementia. Postgrad Med Special Report 1998; April:1–88.
23. McEvoy JP, Scheifler PL, Frances A. The expert consensus guideline series: Treatment of schizophrenia 1999. J Clin Psychiatry 1999; 60(suppl 11):1–80.
24. Foa EB, Davidson JRT, Frances A. The expert consensus guideline series: Treatment of posttraumatic stress disorder. J Clin Psychiatry 1999; 60(suppl 16):1–76.
25. Sachs GS, Printz DJ, Kahn DA, et al. The expert consensus guideline series: Medication treatment of bipolar disorder 2000. Postgrad Med Special Report 2000;April:1–104.
26. Rush AJ, Frances A. Expert consensus guideline series: Treatment of psychiatric and behavioral problems in mental retardation. AJMR 2000; 105:159–228.
27. Altshuler LL, Cohen LS, Moline ML, et al, eds. The expert consensus guideline series: Treatment of depression in women 2001. Postgrad Med Special Report 2001;March:1–116.
28. Alexopoulos GS, Katz IR, Reynolds CF. The expert consensus guideline series: Pharmacotherapy of depressive disorders in older patients. Postgrad Med Special Report 2001;October: 1–86.
29. Kahn DA, Docherty JP, Carpenter D, et al. Consensus methods in practice guideline development: A review and description of a new method. Psychopharmacol Bull 1997; 33: 631–9.
30. Brook RH, Chassin MR, Fink A, et al. A method for the detailed assessment of the appropriateness of medical technologies. Int J Tech Assess Health Care 1986; 2:53–63.
31. Jacoby HI, Brodie DA. Laparoscopic herniorrhaphy: Diagnostic and therapeutic technology assessment. JAMA 1996; 275:1075–82.
32. Allen MH, ed. Acute care of the agitated psychotic patient. J Clin Psychiatry 2000; 61(suppl 14):1–55.
33. Centers for Medicare and Medicaid Services. Hospital CoP for Patients’ Rights: Questions and Answers. US Dept of Health and Human Services. Baltimore, MD: October 10, 2001. (Available at
34. Sheline Y, Nelson T. Patient choice: Deciding between psychotropic medication and physical restraints in an emergency. Bull Am Acad Psychiatry Law 1993; 21:321–9.
35. Anderson WH, Kuehnle JC, Catanzano DM. Rapid treatment of acute psychosis. Am J Psychiatry 1976; 133:1076–8.
36. Neborsky R, Janowsky D, Munson E, et al. Rapid treatment of acute psychotic symptoms with high- and low-dose haloperidol. Arch Gen Psychiatry 1981; 38:195–9.
37. Stotsky BA. Relative efficacy of parenteral haloperidol and thiothixene for the emergency treatment of acutely excited and agitated patients. Dis Nerv Syst 1977; 38:967–73.
38. Binder R, Glick I, Rice M. A comparative study of parenteral molindone and haloperidol in the acutely psychotic patient. J Clin Psychiatry 1981; 42:203–6.
39. Paprocki J, Versiani M. A double-blind comparison between loxapine and haloperidol by parenteral route in acute schizophrenia. Curr Ther Res Clin Exp 1977; 21:80–100.
40. Tuason VB. A comparison of parenteral loxapine and haloperidol in hostile and aggressive acutely schizophrenic patients. J Clin Psychiatry 1986; 47:126–9.
41. Reschke RW. Parenteral haloperidol for rapid control of severe, disruptive symptoms of acute schizophrenia. Dis Nerv Sys 1974; 35:112–5.
42. Gerstenzang ML, Krulisky TV. Parenteral haloperidol in psychiatric emergencies: Double-blind comparison with chlorpromazine. Dis Nerv Syst 1977; 38:581–3.
43. Chambers RA., Druss BG. Droperidol: Efficacy and side effects in psychiatric emergencies. J Clin Psychiatry 1999; 60:664–7.
44. van Leeuwen AMH, Molders J, Sterkmans P, et al. Droperidol in acutely agitated patients: A double-blind placebo-controlled study. J Nerv Ment Dis 1977; 164:280–3.
45. Resnick M, Burton BT. Droperidol vs haloperidol in the initial management of acutely agitated patients. J Clin Psychiatry 1984; 45:298–9.
46. Thomas H, Schwartz E, Petrilli R. Droperidol versus haloperidol for chemical restraint of agitated and combative patients. Ann Emerg Med 1992; 21:407–13.
47. Richards JR, Derlet RW, Duncan DR. Chemical restraint for the agitated patient in the emergency department: Lorazepam versus droperidol. J Emerg Med 1998; 16:567–73.
48. Buckley P, Bartell J, Donenwirth K, et al. Violence and schizophrenia; Clozapine as a specific antiaggressive agent. Bull Am Acad Psychiatry Law 1995; 23:607–11.
49. Glazer WM, Dickson RA. Clozapine reduces violence and persistent aggression in schizophrenia. J Clin Psychiatry 1998; 59(suppl 3):8–14.
50. Czobor P, Volavka J, Meibach RC. Effect of risperidone on hostility in schizophrenia. J Clin Psychopharmacol 1995; 15: 243–9.
51. Chengappa KNR, Levine J, Ultich R, et al. Impact of risperidone on seclusion and restraint in a state psychiatric hospital. Presented at the 40th annual meeting of the New Clinical Drug Evaluation Unit, Boca Raton, FL, May 30–June 2, 2000.
52. Buckley PF, Ibrahim ZY, Singer B, et al. Violence and schizophrenia: Efficacy of risperidone. J Am Acad Psychiatry Law 1997; 25:173–81.
53. Beasley CM, Saylor ME, Kiesler GM, et al. The influence of pharmacotherapy on self-directed and externally directed aggression in schizophrenia (abstract). Schizophr Res 1998; 29:28.
54. Goldstein JM. “Seroquel” (quetiapine fumarate) reduces hostility and aggression in patients with acute schizophrenia. Presented at the 151st annual meeting of the American Psychiatric Association, Toronto, Ontario, Canada, May 30–June 5, 1998.
55. Hellewell JSE, Cameron-Hands D, Cantillon M. Seroquel: Evidence for efficacy in the treatment of hostility and aggression. Schizophr Res 1998; 29:154–5.
56. Lucey JV, Brook S, Daniel DG, et al. Intramuscular (IM) ziprasidone: A novel treatment for the short-term management of agitated psychotic patients. Presented at the 10th Biennial Winter Workshop on Schizophrenia; Davos, Switzerland, Feb 5–11, 2000.
57. Currier GW, Simpson GM. Risperidone liquid concentrate plus lorazepam versus intramuscular haloperidol plus lorazepam for treatment of psychotic agitation. J Clin Psychiatry 2001; 62: 153–7.
58. Jones B, Taylor CC, Meehan K. The efficacy of a rapid-acting intramuscular formulation of olanzapine for positive symptoms. J Clin Psychiatry 2001; 62(suppl 2):22–4.
59. Lesem MD, Zajecka JM, Swift RH, et al. Intramuscular ziprasidone, 2 mg versus 10 mg, in the short-term management of agitated psychotic patients. J Clin Psychiatry 2001; 62:12–8.
60. Brook S, Lucey JV, Gunn KP. Intramuscular ziprasidone compared with intramuscular haloperidol for the treatment of acute psychosis. J Clin Psychiatry 2000; 61:933–41.
61. Battaglia J, Moss S, Rush J, et al. Haloperidol, lorazepam, or both for psychotic agitation? A multicenter, prospective, double-blind, emergency department study. Am J Emerg Med 1997; 15:335–40.
62. Bienek SA, Ownby RL, Penalver A, et al. A double-blind study of lorazepam versus the combination of haloperidol and lorazepam in managing agitation. Pharmacotherapy 1998; 18:57–62.
63. Salzman C, Solomon D, Miyawaki E, et al. Parenteral lorazepam versus parenteral haloperidol for the control of psychotic disruptive behavior. J Clin Psychiatry 1991; 52:177–80.
64. Foster S, Kessel J, Berman ME, et al. Efficacy of lorazepam and haloperidol for rapid tranquilization in a psychiatric emergency room setting. Int Clin Psychopharmacol 1997; 12: 175–9.
65. Salzman C, Green A, Rodriguez-Villa F, et al. Benzodiazepines combined with neuroleptics for management of severe disruptive behavior. Psychosomatics 1986; 27(suppl 1):17–22.
66. Wyant M, Diamond B, O’Neal E, et al. The use of midazolam in acutely agitated psychiatric patients. Psychopharmacol Bull 1990; 26:126–9.
67. Chouinard G, Annable L, Turnier L, et al. A double-blind randomized clinical trial of rapid tranquilization with I.M. clonazepam and I.M. haloperidol in agitated psychotic patients with manic symptoms. Can J Psychiatry 1993; 38(suppl 4):S114–21.
68. Dorevitch A, Katz N, Zemishlany Z, et al. Intramuscular flunitrazepam versus intramuscular haloperidol in the emergency treatment of aggressive psychotic behavior. Am J Psychiatry 1999; 156:142–4.
69. Binder RL, McNiel DE. Emergency psychiatry: Contemporary practices in managing acutely violent patients in 20 psychiatric emergency rooms. Psychiatr Serv 1999; 50:1553–4.
70. American Hospital Formulary Service Drug Information 28:24:08. Bethesda, MD: American Society of Health System Pharmacists, 1998.
71. Medicines Control Agency. Important safety message: Discontinuation of Droleptan tablets, suspension and injection (droperidol). (Available from
72. Food and Drug Administration. Available at
73. Currier GW. Atypical antipsychotic medications in the psychiatric emergency service. J Clin Psychiatry 2000; 61(suppl 14):21–6.
74. Baldessarini RJ, Cohen BM, Teicher M. Significance of neuroleptic dose and plasma level in the pharmacological treatment of psychoses. Arch Gen Psychiatry 1988; 45:79–91.
75. Breitbart W, Marotta R, Platt MM, et al. A double-blind trial of haloperidol, chlorpromazine, and lorazepam in the treatment of delirium in hospitalized AIDS patients. Am J Psychiatry 1996; 153:231–7.
76. van Harten PN, van Trier JCAM, Horwitz EH, et al. Cocaine as a risk factor for neuroleptic-induced acute dystonia. J Clin Psychiatry 1998; 59:128–30.
77. American Psychiatric Association. Practice guideline for the treatment of patients with substance use disorders: Alcohol, cocaine, opioids. Am J Psychiatry 1995; 152(suppl 11):1–59.
78. Ghaemi SN. New treatments for bipolar disorder: The role of atypical neuroleptic agents. J Clin Psychiatry 2000; 61(suppl 14):33–42.
79. Hirschfeld RM, Allen MH, McEvoy JP, et al. Safety and tolerability of oral loading divalproex sodium in acutely manic bipolar patients. J Clin Psychiatry 1999; 60:815–8.
80. Altshuler LL, Cohen L, Szuba MP, et al. Pharmacologic management of psychiatric illness during pregnancy: Dilemmas and guidelines. Am J Psychiatry 1996; 153:592–606.
81. Physicians’ Desk Reference. Montvale, NJ: Thomson PDR; 2003: 3551.
82. Salzman C. Use of benzodiazepines to control disruptive behavior in inpatients. J Clin Psychiatry 1988; 49(suppl):13–5


Figure. Appendix A: Treatment of a Behavioral Emergency
Figure. Appendix A: Treatment of a Behavioral Emergency:
Figure. Appendix B: Summary of Preferred Medications by Etiology
Figure. Appendix B: Summary of Preferred Medications by Etiology

restraint; seclusion; culture change; training; individualized treatment

Copyright © 2003 Wolters Kluwer Health, Inc. All rights reserved.