Objectives: Due to inherent dangers and barriers to research in emergency settings, few data are available to guide clinicians about how best to manage behavioral emergencies. Key constructs such as agitation are poorly defined. This lack of empirical data led us to undertake a survey of expert opinion, results of which were published in the 2001 Expert Consensus Guidelines on the Treatment of Behavioral Emergencies. Several second-generation (atypical) antipsychotics (SGAs) are now available in new formulations for treating behavioral emergencies (e.g., intramuscular [i.m.] olanzapine and ziprasidone; rapidly dissolving tablets of olanzapine and risperidone). Critical questions face the field. The SGAs are significantly different from the FGAs and from each other and have not been studied in unselected patients as were the FGAs. Can the SGAs can be thought of as a class, do all antipsychotics have similar anti-agitation effects in different conditions, and, if equally effective, what limits might their safety profiles impose? Should antipsychotics be used more specifically to treat psychotic conditions, while benzodiazepines (BNZs) alone are used nonspecifically? Few data are available concerning combinations of SGAs and BNZs, and findings concerning the traditional combination of haloperidol plus a BNZ may not be relevant to combinations with SGAs. The culture is also evolving with more emphasis on patient involvement in treatment decisions. An international consensus has been developing that calming rather than sedation is the appropriate endpoint of behavioral emergency interventions. We undertook a new survey of expert opinion to update recommendations from the earlier survey.
Method: A written survey of 61 questions (1,020 options) was mailed to 50 experts in the field, 48 (96%) of whom completed it. The survey sought to define level of agitation at which emergency interventions are appropriate, scope of assessment depending on urgency and patients' ability to cooperate, guiding principles for selecting interventions, and appropriate physical and medication strategies at different levels of diagnostic confidence for a variety of provisional diagnoses and complicating conditions. A modified version of the RAND Corporation's 9-point scale for rating appropriateness of medical decisions was used to score most options. Consensus was defined as a non-random distribution of scores by chi-square "goodness-of-fit" test. We assigned a categorical rank (first line/preferred, second line/alternate, third line/usually inappropriate) to each option based on the 95% confidence interval around the mean. Ratings were used to develop guidelines for preferred strategies in key clinical situations. This study received financial support from multiple sponsors, with the panel kept blind to sponsorship to reduce possible bias. Medication ratings were based on responses of only those respondents with direct experience with each drug. In reporting practice patterns, the panel was asked to respond based on actual data rather than estimates.
Results: The expert panel reached consensus on 78% of the options rated on the 9-point scale. The responses suggest that physicians can make provisional diagnoses with some confidence and that pharmacological and nonpharmacological interventions are selected differentially based on diagnosis and other salient demographic and medical features. BNZs are recommended when no data are available, when there is no specific treatment (e.g., personality disorder), or when they may have specific benefits (e.g., intoxication). No single SGA emerges as a nonspecific replacement for haloperidol; instead, different SGAs are preferred in various circumstances consistent with current evidence. To the degree that haloperidol is recommended, it is almost always in combination with a BNZ; haloperidol alone is preferred only in the medically compromised. In contrast, the SGAs are more often recommended for use alone, and the panel would avoid combining BNZs with some SGAs. Oral risperidone alone or combined with a BNZ receives strong support in a variety of situations. Oral olanzapine was rated very similarly to risperidone, with slightly higher ratings than risperidone in situations where it has been studied (e.g., schizophrenia, mania) and slightly lower ratings where it has not been studied or safety may be a concern; there was less support for combining oral olanzapine with a BNZ. For oral treatment of agitation related to schizophrenia or mania, olanzapine alone, risperidone alone or combined with a BNZ, and haloperidol plus a BNZ are first line, with strong support also for combining divalproex with the antipsychotic for presumed mania. Oral ziprasidone and quetiapine generally received similar second-line ratings in most situations. If a parenteral agent is needed, i.m. olanzapine alone received somewhat more support than i.m. ziprasidone alone; however, there was more support for i.m. ziprasidone alone or combined with a BNZ than for i.m. olanzapine plus a BNZ, probably reflecting safety concerns. For example, for a provisional diagnosis of schizophrenia, first-line parenteral options are i.m. olanzapine or ziprasidone alone or i.m. haloperidol or ziprasidone combined with a BNZ. Neither of the new parenteral formulations received as much support as traditional agents (i.m. BNZs, i.m. haloperidol) when no data are available or the diagnosis involves medical comorbidity or intoxication. When initial intervention with risperidone, ziprasidone, or haloperidol is unsuccessful, the panel recommended adding a BZD to the antipsychotic. However, when initial treatment with olanzapine or quetiapine is unsuccessful, increasing the dosage is recommended. Perphenazine was consistently rated second line and droperidol and chlorpromazine received third-line ratings throughout.
Conclusions: Within the limits of expert opinion and with the expectation that future research data will take precedence, these guidelines suggest that the SGAs are now preferred for agitation in the setting of primary psychiatric illnesses but that BNZs are preferred in other situations.