Recently the MINDS-USA trial evaluated the use of haloperidol or ziprasidone for delirium in critical illness.  Before jumping into the results of this study, it will help to establish a couple of foundational principles.

Foundational principles

Delirium is a syndrome with multiple etiologies and no single treatment.

Delirium isn’t a specific disease state.  Rather, it is a syndrome – an associated group of signs and symptoms which often occur together.  This may be caused by dozens of disorders (e.g. stroke, hypoglycemia, hypernatremia, sleep deprivation, medications).

As a syndrome, delirium has no single pathophysiology.  Different forms of delirium are caused by different imbalances.  For example, patients with delirium due to alcohol withdrawal have deficient GABA signaling and excessive glutamine activity.  Alternatively, patients with delirium due to anti-cholinergic medication have deficient acetylcholine activity.  Most patients have combinations of numerous imbalances.

The ideal treatment of delirium is to reverse whatever is causing it.  One patient with delirium due to hypoglycemia needs IV dextrose.  Another patient with delirium due to alcohol withdrawal may respond nicely to phenobarbital.  Unfortunately, in most cases there is no clear cause of delirium.  Critically ill patients often have multifactorial delirium due to a host of causes which are not easily reversed (e.g. sepsis, sleep deprivation).

In summary, delirium is a syndrome which is widely heterogeneous, resulting from a variety of causes, involving different pathophysiologies, and mandating variable treatments.   It should come as little surprise that there doesn’t seem to be any silver-bullet treatment for delirium.

Positive symptoms are easier to treat than negative symptoms

Symptoms of schizophrenia are generally divided into positive symptoms and negative symptoms (above).  Positive symptoms generally respond well to antipsychotic therapy (with similar efficacy among either typical or atypical antipsychotics).  Unfortunately, negative symptoms are much harder to treat.  In fact, most antipsychotic medications haven’t demonstrated significant improvement in negative symptoms.¹

This shouldn’t be too surprising.  Most antipsychotic medications have sedative effects.  It’s somewhat paradoxical to expect a sedative agent to resolve negative symptoms (e.g. apathy, lack of energy).

Delirium may involve more diffuse brain dysfunction compared to schizophrenia (e.g. patients with delirium can exhibit the symptoms of schizophrenia plus additional symptoms, such as inattention).  Given that delirium is a more complex and variable disorder than schizophrenia, it ought to be harder to treat.  If antipsychotics don’t work for negative symptoms in schizophrenia, then it would be a reasonable guess that they don’t work for negative symptoms in delirium either.

MINDS-USA trial

This is a multicenter RCT evaluating haloperidol or ziprasidone for delirium among patients with either respiratory failure or shock.²  Patients were identified using CAM-ICU screening, a sensitive tool to detect cases of delirium which may not be clinically obvious.  This resulted in a population of 566 patients, of whom 89% had hypoactive delirium.  There was no effect in any of the outcomes (e.g., mortality, days with delirium, ICU length of stay).

No. And this is why crit care studies have to be carefully constructed— we keep studying blunt tools in a diverse group of patients who arent even narrowed down to a single disease state or phenotype. Some helped, some harmed, and unable to tell the diff

— Sage Whitmore (@spwhitmore) October 23, 2018

Ineffectiveness of antipsychotics in this study comes as no surprise, for two reasons:

1. 89% of these patients had hypoactive delirium, which is marked by somnolence, inattention, and impaired cognition. Haloperidol and ziprasidone have sedative effects, which threaten to make somnolent patients even more somnolent.  Ineffectiveness of antipsychotics for hypoactive delirium is predicted by the ineffectiveness of antipsychotics for negative symptoms in schizophrenia.
2. Open-label antipsychotics were allowed in all arms of the trial, with 21% of patients receiving them.  The use of antipsychotics across all arms of the trial will minimize observed differences between groups.

What is surprising is the safety of haloperidol and ziprasidone in this population.  I would have expected some signs of harm (e.g. excessive somnolence, torsade de pointes, or extrapyramidal effects).  However, there was no real evidence of toxicity.  Two patients treated with haloperidol developed torsade de pointes, but this occurred >4 days after stopping the haloperidol.  Similar numbers of patients in each group stopped the medication due to concern for extrapyramidal symptoms.  Ziprasidone did cause increase QTc duration, without causing any clinical arrhythmias.

Is the antipsychotic glass half empty or half full?  Antipsychotics didn’t cause benefit, but they didn’t cause harm either.   Absence of harm may be reassuring in situations where we have little choice but to use these medications (e.g. dangerous agitation).

Is there a role for antipsychotics in delirium? 

MINDS-ICU provides us with a cautionary tale of how not to use antipsychotics:  screen for delirium using CAM-ICU and then “treat” delirium with antipsychotics.   However, some roles for antipsychotics likely remain:

1. Insomnia:  Quetiapine may be used for insomnia in patients at high risk of delirium.³
2. Acutely uncontrolled agitation:  Intravenous haloperidol have traditionally been used for the management of delirious patients with profound agitation.  Intravenous olanzapine may be used similarly.
3. Difficult-to-sedate patient:  There may also be a role for atypical antipsychotics among patients with less severe symptoms.  Quetiapine or olanzapine may be used primarily for their sedative properties among difficult-to-sedate patients (e.g., patients who have been intubated for several days).  Especially at lower doses, the sedative antihistamine effects of these drugs may outweigh their anti-dopaminergic activity.  These agents can therefore function as “non-deleriogenic sedatives.”⁴  Both quetiapine and olanzapine have established efficacy in anxiety disorders, so there is some evidentiary support for their use as sedatives.⁵

Yes, I know:  none of these indications is supported by Level 1 evidence.  None are ideal.  In a perfect world we would use no neuroactive medications, but rather control agitation with the calming properties of the human voice.  However, in reality, situations will arise where an antipsychotic agent is preferable to alternative therapies or lack of any treatment:

1. Insomnia:  Most sedative/hypnotic agents used for sleep will worsen delirium (e.g. benzodiazepines, zolpidem, antihistamines).   Dexmedetomidine could arguably the ideal agent for insomnia, but logistically this can be challenging.
2. Acutely uncontrolled agitation:  Uncontrolled agitation is dangerous to the patient and staff.   Most other agents used to treat agitation (e.g. opioids, benzodiazepines) will ultimately worsen the underlying delirium.
3. Difficult-to-sedate patient:  Propofol or dexmedetomidine are generally preferred sedatives, but situations will arise when these drugs are undesirable (especially hemodynamically unstable patients).  Prolonged use of propofol may lead to hypertriglyceridemia, whereas prolonged use of dexmedetomidine may lead to tolerance and dependence.  For hemodynamically unstable patients or patients intubated for extended periods, sedating atypical antipsychotics can be useful.

Overall, antipsychotics don’t reverse the underlying pathophysiologic causes of delirium.  However, antipsychotics can be useful to alleviate symptoms of agitation and insomnia, allowing the patient to be managed safely while delirium abates.

Parting shot – to make matters even more confusing, “antipsychotics” have widely variable mechanisms of action

Most classes of neuroactive medications have a consistent mechanism of action (e.g. benzodiazepines, barbiturates, SSRIs).  Uniform pharmacodynamics within a class of drugs allows knowledge about one drug to be applied to other drugs within the same class.  For example, if lorazepam works for seizure and your department runs out of lorazepam, then it’s a fair assumption that another benzodiazepine would work as well.

Although antipsychotics are lumped together, they have enormously variable mechanisms of action:⁶

The medications studied in MINDS-USA may be roughly described as follows:

• Haloperidol is classic “typical antipsychotic” which works mostly as a dopamine (D2-receptor) antagonist.
• Ziprasidone is a relatively pure “atypical antipsychotic” which works largely as a 5-HT2A antagonist.

Atypical antipsychotics which are commonly used in critical care include quetiapine and olanzapine.  These drugs have different mechanisms of action:

• Quetiapine’s primary effect seems to blocking the H1 receptor, particularly at lower doses.  This anti-histamine effect may cause it to act largely as a sedative (a “non-deliriogenic sedative”).
• Olanzapine inhibits the 5HT2A inhibitor more than the D2 dopamine receptor, with some antihistamine activity as well.  Olanzapine may be roughly conceptualized as a 2:1:1 mixture of ziprasidone, haloperidol, and quetiapine.

The inconvenient truth is that evidence obtained from haloperidol or ziprasidone doesn’t necessarily apply to other antipsychotics.  Thus, each of these medications needs to be studied individually, ideally in subsets of patients (e.g. does quetiapine work for patients with agitated delirium?).

Ultimately, we are left with the fact that we still don’t know much about antipsychotics for delirium.  Until more is known, the following general principles are reasonable (this applies for most drugs in the ICU):

• Don’t assume that the drug will make the patient better.
• Use the lowest possible dose of the drug required to alleviate problematic symptoms (which, for a patient with hypoactive delirium will be zero milligrams).
• Assess responsiveness to the medication, if it’s not working try something else.
• Use a lot of fancy terminology to make it seem like you know what you’re doing.

• MINDS-USA investigated the use of haloperidol or ziprasidone among a group of delirious ICU patients, 89% of whom had hypoactive delirium.
• Haloperidol and ziprasidone were both ineffective. However, they were both also surprisingly safe (with comparable side-effect rates compared to placebo).
• This study suggests that haloperidol and ziprasidone are ineffective for hypoactive delirium. This makes sense given the clinical features of hypoactive delirium (somnolence) and the variable neurobiological underpinnings of this disorder (making a single therapy unlikely to work for all patients).
• All patients in the study were allowed to have open-label PRN antipsychotics for breakthrough agitation. Therefore, this study cannot reveal anything about this utilization of antipsychotics.
• The receptor targets of atypical antipsychotics are widely variable. Thus, results obtained in this study are not necessarily generalizable to all antipsychotic medications.

Addendum 11/24:

MINDS-USA first author Dr. Timothy Girard responded to this blog in a thoughtful twitter thread.  I've reproduced it here, followed by some responses to it:

Thanks for @emcrit for pointing their followers to our recent @NEJM publication on antipsychotics for @ICU_delirium: https://t.co/r4htItyP9R

— Timothy Girard (@timothygirard) November 24, 2018

I appreciate their thoughtful and thorough commentary, especially the description of when to consider using an antipsychotic. But a number of points should be clarified. https://t.co/0a5vDZk6PQ

— Timothy Girard (@timothygirard) November 24, 2018

“89% of these patients had hypoactive delirium” – True at randomization but the percent with hypoactive delirium fluctuated during the trial; 209 (37%) experienced 1 or more days of hyperactive (agitated) delirium.

— Timothy Girard (@timothygirard) November 24, 2018

“Open-label antipsychotics were allowed in all arms of the trial, with 21% of patients receiving them.” Half of the 21% received <5 mg of open-label haloperidol total; 75% received <10 mg total.

— Timothy Girard (@timothygirard) November 24, 2018

“The use of antipsychotics across all arms of the trial will minimize observed differences between groups.” 79% of participants received 0 mg open-label antipsychotic. The other 21% received cumulative doses that were 1/10 or less what they received as double-blind study drug.

— Timothy Girard (@timothygirard) November 24, 2018

“Intravenous haloperidol has traditionally been used for the management of delirious patients with profound agitation.” Not unreasonable, but one (small) RCT compared haloperidol with dexmedetomidine for agitation and found the latter was better. https://t.co/emBKOGFGQa

— Timothy Girard (@timothygirard) November 24, 2018

“Evidence obtained from haloperidol or ziprasidone doesn’t necessarily apply to other antipsychotics.” This point–often used to justify the use of other antipsychotics–doesn't change the fact that we have no high-quality evidence supporting the use of any antipsychotic.

— Timothy Girard (@timothygirard) November 24, 2018

“Before publication of this study was anyone using antipsychotics for hypoactive delirium?” Yes, there's plenty of published data showing this to be the case. This is just one of many examples where what doctors believe they do differs from what they actually do.

— Timothy Girard (@timothygirard) November 24, 2018

In fact, multiple studies have found that 25% of antipsychotic-naive patients who were started on an antipsychotic for @ICU_delirium end up being discharged from the hospital on the drug. https://t.co/3gvo5ips5x

— Timothy Girard (@timothygirard) November 24, 2018

First, thanks for commenting on the blog.  It's always extremely illuminating when the authors of a study add commentary.

I agree that antipsychotics are over-utilized for patients with hypoactive delirium.  You study should shine light on this problem and thereby lead to practice improvement.

I agree that few patients received PRN haloperidol.  The problem is that this minority of patients who received haloperidol, patients with severe hyperactive delirium and dangerous agitation, were the very subgroup patients who likely benefitted most from haloperidol.   Since all such patients were treated with haloperidol, it's impossible to know the consequences of not administering haloperidol in this scenario.

In order to determine if there is a role for haloperidol in severe agitated delirium in the ICU, a RCT is needed which would randomized patients to haloperidol versus another treatment (without the use of rescue haloperidol for the non-haloperidol group).  This would be difficult to execute because there aren't a lot of great alternatives to haloperidol for immediate control of dangerous agitated delirium (e.g. benzodiazepines or opioids are occasionally used, but these likely aggravate delirium).  I could imagine the following study design possibly being performed outside the United States (it would be extremely hard to get it through a US IRB):

• Antipsychotic group:  Haloperidol IV for breakthrough agitation, scheduled haloperidol for maintenance therapy
• Non-antipsychotic group:  IV ketamine/propofol for breakthrough agitation, dexmedetomidine infusion for maintenance therapy

Overall, I agree that we have precious little high-quality evidence regarding the treatment of agitated delirium in the ICU.  Thanks for your study, like any successful study it illuminates some points while leading to many unanswered questions.

-Josh Farkas (@PulmCrit)

Image credit: Neurons

References

• Author
• Recent Posts

Social Me

Josh Farkas

Josh is the creator of PulmCrit.org. He is an associate professor of Pulmonary and Critical Care Medicine at the University of Vermont.

Social Me

Latest posts by Josh Farkas (see all)

• PulmCrit Hot Take: Steroid for severe pneumonia (CAPE COD trial) - March 21, 2023
• PulmCrit Blogitorial – SIESTA syndrome:  Sedation Induced EEG Suppression with Transient Agitation - December 19, 2022
• PulmCrit Hot Take – Acetazolamide plus furosemide for decongestion of heart failure (ADVOR trial) - August 27, 2022