Text version
Rapid Tranquilisation in the Emergency Setting
Managing acute agitation in the emergency room is one of the most challenging scenarios in psychiatry, and there is so much disagreement about the optimal approach. In emergency settings, affect often runs very high around patients who are yelling, striking out, and generally being disruptive. Prior recommendations have suggested everything from IV Haldol to ketamine to droperidol and Versed to IM ziprasidone to even oral risperidone as first-line strategies.
Each hospital I’ve worked at has favored a different approach. Frustratingly, psychiatrists are often not leading the development of agitation algorithms and sometimes are not even involved. Studies of agitation management frequently offer conflicting recommendations, and most don’t take into account factors that seem important to psychiatrists, such as understanding the driver of the agitation.
In that context, pooled analyses of multiple studies could be a useful way of understanding the overall landscape of the literature. Today’s study, recently published in Lancet Psychiatry, is a network meta-analysis examining the effectiveness and safety of approaches for rapid tranquilization in emergency settings.
Download PDF and other files
Reading Network Rankings With Caution
Network meta-analyses are becoming increasingly visible in psychiatry. There have been some really important examples recently from the same group who published today’s study, comparing the safety and efficacy of antipsychotics or antidepressants and then stratifying agents according to their relative propensity to cause various side effects.
The key strength of a network meta-analysis is the ability to compare all available interventions for a given indication simultaneously and to rank them, providing evidence when direct head-to-head trials are not available. On the flip side, it’s important to keep in mind that especially where the networks are poorly connected (that is, where treatments are linked by only a single small trial) estimates may be imprecise and rank ordering may be unreliable. We should keep that in mind as we dig through today’s study.
Head-to-Head Rapid Sedation Trials
Today’s network meta-analysis included randomized controlled trials comparing intramuscular or intravenous treatments used for agitation management in emergency settings. The authors specifically looked at trials in which the agents were targeted to achieve sedation within 15 to 30 minutes, and in which sedation was evaluated close to 20 minutes and not more than 30 minutes after the medication was administered.
Trials that focused on one specific etiology of agitation, such as delirium, were excluded, as were studies conducted in children or older adults. Placebo-controlled trials were also excluded, so the analysis looked only at head-to-head trials involving at least two different medications or combinations.
The study ultimately included 13 trials involving about 2700 patients. 58% of included individuals were men, and mean age was 36 years.
Download PDF and other files
Haloperidol Monotherapy Fared Poorly
The major finding of this network meta-analysis was that the odds of achieving sedation within 30 minutes for moderate agitation were higher with several approaches compared to haloperidol alone:
- Antipsychotic–benzodiazepine combinations
- Benzodiazepines alone
- Other antipsychotics alone
For severe agitation, antipsychotic–benzodiazepine combinations were also associated with higher odds of achieving sedation than haloperidol. In other words, Haldol monotherapy did not do well in this study.
On the safety side, Haldol monotherapy was associated with a higher risk of extrapyramidal side effects (EPS), and benzodiazepines (either alone or in combination)nwere associated with a higher risk of hypotension.
Confidence Was Rated As Very Low
So based on these results, should we stop using Haldol for agitation? It’s important first to understand these findings in a broader context and to think about some major limitations of this particular study.
Most importantly, confidence in today’s findings was rated as very low. This is related in part to the lack of connections between agents used in the individual studies and to the heterogeneity of the studies. The odds ratios in a lot of the comparisons here have huge ranges, which speaks to the imprecision. Based on this alone, I’d be very wary of buying fully into the headline claims that accompany this article.
Download PDF and other files
Setting and Route Confound the Comparison
Another significant limitation has to do with setting and route of administration. Haldol was primarily studied in psychiatric emergency settings, whereas antipsychotic–benzodiazepine combinations were mostly studied in general ED settings. These settings may have very different patient populations, and one could imagine that patients in the general ED may be more likely to have other medical conditions that could contribute to sedation.
Furthermore, Haldol was mostly given intramuscularly in these studies, whereas other medications were often given IV. This difference would obviously have a huge impact on the major finding of time to sedation. It’s also worth noting that IM haloperidol is associated with high rates of EPS, whereas IV Haldol, for reasons we don’t really understand, is not.
A final point is that agents were essentially separated into Haldol and everything else, with the everything-else group lumped together and looked at as a single entity. This introduces the possibility that one agent in that group accounted for most of the better performance, and it makes it really hard to compare individual agents with Haldol.
All of this is to say that I’m taking these findings with a grain of salt. I don’t know that we can draw really meaningful conclusions from these data or that they teach us anything new.
Sedation Speed May Be the Wrong Endpoint for Psychiatry
But let’s assume for a second that these findings are real and that Haldol really does perform worse than other agents or combinations. Would I stop using it?
Here’s a really important thing to remember about agitation studies: most use a similar endpoint: time to sedation or time to tranquilization. Today’s study specifically looked at agents targeted to achieve sedation within 15 to 30 minutes.
This makes complete sense if you think about it from the standpoint of the emergency medicine physician. If you have an agitated patient in your emergency room yelling and striking out, your primary goal is to get that patient sedated so you can properly examine and assess them for acute medical issues. This is why emergency medicine docs often favor options that provide immediate sedation, and why medications like ketamine and midazolam that were once used primarily in anesthesia settings increasingly find their way into official algorithms for agitation in EM settings. The faster a medication knocks someone out, the better it performs in these studies.
As psychiatrists working in the ED, though, our goals are often different from and sometimes at odds with those of emergency medicine physicians. Of course we also want the person calmer so we can conduct an interview, but we don’t want them so sedated that they can’t participate. In fact, few things frustrate ED psychiatrists more than being asked to see a patient who is completely snowed and unable to answer any questions.
Download PDF and other files
Treat the Driver of the Agitation
Additionally, our primary goal is to treat the underlying driver of the agitation, which may be anxiety or psychosis or withdrawal or something else entirely, depending on the presentation. In that way, we tend to take less of a one-size-fits-all and more of a targeted approach to managing agitation.
If someone’s agitation is being driven by an exacerbation of a primary thought disorder, why would I want to use a medication like midazolam or ketamine that will do nothing for their psychosis and might even make it worse? If I’m being thoughtful, I’d like to give them something that will start to treat the psychosis while at the same time relieving fear and helping to calm them down. I’m still going to strongly consider Haldol in that case, even if I know it might take a little longer to act.
I certainly might consider adding a little lorazepam, so that Haldol can potentiate the sedative effect of the benzodiazepine. In the case of primary psychosis, that seems low risk. In fact, for the manic patient, the benzo is likely to add to the therapeutic effect.
On the other hand, if I think the agitation is driven by delirium, I’m going to think twice about adding that benzo to the Haldol. That’s not to say I would never do it; I’m definitely not an absolutist that benzos are bad for delirium, and sometimes a little benzo does go a long way toward calming when given in combination with Haldol and used judiciously. But in that case, I’m probably going to try Haldol by itself first, and I would certainly never use ketamine or midazolam for a delirious patient who’s agitated.
What about olanzapine? I think it’s a reasonable alternative to Haldol in many cases. We just have to keep in mind that it can’t be combined with a benzodiazepine directly, because of that somewhat controversial warning about cardiopulmonary suppression, and that in delirious patients it carries a high anticholinergic burden, especially with repeated dosing.
Dropping the Anticholinergic Third Agent
One last thing to highlight. When I trained in the mid-2000s, there was a third component to the traditional ED agitation cocktail. We used a 5-2-1 or a 5-2-50: the 5 being Haldol, the 2 being lorazepam, and the 1 being Cogentin, while the 50 was Benadryl, one of those two as the third agent. The idea was that an anticholinergic agent would further mitigate the risk of dystonic events with Haldol.
More recently, our emergency department has shifted away from adding that anticholinergic component, and I actually think this is a good change. I’ve been pleasantly surprised to find that it’s not really needed, and removing it has not seemed to lead to any higher rates of EPS. I imagine that’s partly because the presence of the benzo is probably somewhat protective on its own.
In fact, using two agents instead of three has probably led to less prolonged sedation or snowing of patients, which allows for faster reassessments following medication. So if you’re still using that third component, you might try shifting away from it and seeing if you notice a difference.
Download PDF and other files
Abstract
Comparative effectiveness and safety of pharmacological treatments for rapid tranquilisation in emergency settings: a systematic review and individual participant data network meta-analysis
Spyridon Siafis; Frederik Philipona; Nobuyuki Nomura, et al
Background
Several medications are available for rapid tranquilisation in psychomotor agitation, but choosing among them varies across local practices and this variation is compounded by inconsistent guidelines. We performed a systematic review with individual participant data network meta-analysis to inform evidence-based recommendations.
Methods
In this systematic review and individual participant data meta-analysis, we searched multiple databases from database inception to Nov 14, 2025, for randomised trials comparing intramuscular or intravenous treatments for rapid tranquilisation (primary outcome defined as sedation within 15-30 min) in patients with psychomotor agitation in general or psychiatric emergency settings. Anonymised individual participant data were collected and harmonised into a common dataset. Risk of bias was assessed using the RoB 2 tool. We performed Bayesian one-stage random-effects individual participant data network meta-regressions, accounting for between-study heterogeneity, drug classes, prognostic factors, and subgroup effects based on agitation severity. We combined individual participant data and aggregate data to evaluate side-effects. Confidence in the evidence was assessed using the Confidence In Network Meta Analysis (CINeMA) framework. People with lived experience were involved in the design and interpretation of the findings. The protocol was registered with PROSPERO (CRD42023402365).
Findings
We included 18 trials across eight regions (3411 participants; 1988 [58·3%] men, 1423 [41·7%] women; mean age 36·0 [SD 11·7] years), of which 13 trials (2705 participants) provided individual participant data for antipsychotics, benzodiazepines, and their combination. In moderate agitation, odds of achieving sedation relative to haloperidol monotherapy were higher with antipsychotic-benzodiazepine combinations (odds ratio [OR] 12·93, 95% credible interval [95% CrI] 3·00-50·91; relative risk [RR] 1·58), benzodiazepines (5·52, 1·37-21·02; 1·49), and other antipsychotics (4·54, 1·35-14·45; 1·45). In severe agitation, antipsychotic-benzodiazepine combinations were more effective than haloperidol (4·86, 1·28-17·54; 1·73), whereas results were uncertain for benzodiazepines (2·09, 0·58-6·99; 1·38) and other antipsychotics (1·70, 0·62-4·59; 1·28). Confidence in these estimates was very low, mainly due to imprecision and heterogeneity. Haloperidol monotherapy was associated with higher risk of extrapyramidal side-effects and benzodiazepines, alone or in combination, with hypotension.
Interpretation
Antipsychotic-benzodiazepine combinations might be among the most effective options for rapid tranquilisation in patients with psychomotor agitation but carry a risk of hypotension, whereas haloperidol monotherapy appeared among the least effective and is associated with extrapyramidal side-effects. These findings should be contextualised to the specific setting and patient characteristics, including the underlying agitation aetiology. Large trials are needed to provide more precise recommendations.
Funding
German Ministry of Research, Technology and Space and Swiss National Science Foundation.
Reference
Siafis, S.; Philipona, F.; Nomura, N.; Milosavljević, F.; Chalkou, K.; Schneider-Thoma, J. et al. (2026). Comparative effectiveness and safety of pharmacological treatments for rapid tranquilisation in emergency settings: a systematic review and individual participant data network meta-analysis. Lancet Psychiatry; 13(7):567-580.
