Do Antipsychotics Help with ICU Delirium?

Abstract & Commentary

By Andrew M. Luks, MD, Pulmonary and Critical Care Medicine, University of Washington, Seattle, is Associate Editor for Critical Care Alert.

Synopsis: This small, randomized, placebo-controlled trial demonstrated that use of haloperidol or ziprasidone in mechanically ventilated medical and surgical ICU patients was not associated with adverse outcomes, but did not improve the number of days alive without delirium or coma.

Source: Girard TD, et al. Feasibility, efficacy, and safety of antipsychotics for intensive care unit delirium. Crit Care Med 2010;38:428-437.

Although delirium is a common complication of critical illness associated with considerable morbidity and mortality and haloperidol is widely used for treatment of the problem, there are no controlled studies examining whether this or other medications are effective therapeutic modalities in ICU patients. Girard and colleagues conducted a randomized, controlled trial to determine if antipsychotic agents were safe and effective at treating delirium in mechanically ventilated patients with the larger goal of assessing whether a larger, multicenter trial is warranted to address these issues.

They enrolled mechanically ventilated adults (> 18 years) on sedative and/or analgesic medications in medical or surgical ICUs at six tertiary medical centers. Patients were excluded if they were pregnant, had been on mechanical ventilation > 60 hours, had no plan for gastric access within 48 hours, took antipsychotics as an outpatient, or had underlying neurologic diseases (ongoing seizures, recent stroke, dementia, or anoxic brain injury), or other problems such as severe heart failure, predisposition to tachyarrhythmia, or severe electrolyte abnormalities. Enrolled patients were then randomized to one of three treatments (haloperidol 5 mg, ziprasidone 40 mg, or placebo) every 6 hours for up to 14 days, after which time all study drugs were discontinued. Short-term use of IV formulations was permitted if gastric access had not been established but study drugs were otherwise administered orally or via nasogastric tube.

Patients were evaluated twice daily for the presence of delirium and their level of sedation using the Confusion Assessment Method for the ICU (CAM-ICU) and Richmond Agitation-Sedation Scale (RASS), respectively. Dosing frequency was reduced or the study drug was discontinued according to a pre-specified protocol if patients were CAM-ICU negative on consecutive assessments or had evidence of over-sedation. Study drug could be restarted or the dosing frequency increased if patients developed recurrent delirium following such changes. The managing ICU teams determined all other treatments, including the approach to sedation, and could use open-label antipsychotic medications for severe agitation or delirium. The primary endpoint was the number of days alive without delirium or coma during the 21-day study period, while secondary endpoints included the duration of coma or delirium, ventilator-free days, time to ICU and hospital discharge, and all-cause 21-day survival.

Of 3297 patients screened over a 2-year period, a total of 101 (35 haloperidol, 30 ziprasidone, 36 placebo) were enrolled in the study and included in the final analysis, with adequate matching between the three groups. Over the 21-day study period, patients treated with haloperidol spent a similar number of days in delirium or coma (14.0; interquartile range, 6.0-18.0), as patients in the ziprasidone (15; interquartile rate, 9.1-18.0) and placebo (12.5; interquartile range, 1.2-17.2) groups. Daily sedation goals were also similar in the three groups throughout the entire study period. There were no statistically significant differences in secondary outcomes or the use of additional open-label antipsychotics, while the numbers of patients developing akathisia or extrapyramidal symptoms were similar between the two groups. No patients developed neuroleptic malignant syndrome or ventricular arrhythmias following initiation of the study drugs.


On the surface, this study appears to call into question a long-standing, but poorly supported practice, the use of haloperidol and other antipsychotics for the management of delirium in the ICU, as there were no differences in the number of delirium-free days between the various treatment groups. It is important to remember, however, that the primary goal of this study was to demonstrate the feasibility and safety of conducting a larger multicenter trial on this question rather than show a clear difference between these treatment options for delirium itself. In fact, with only 36 or fewer patients in each treatment group, the likelihood of a Type II error for the primary hypothesis is reasonably high.

Beyond the issue of the study size, there are some other important methodological issues that warrant attention when considering these results. First, although the study design included a mechanism for decreasing the frequency of the study drug, there was no protocolized mechanism for dose increases. Second, all other decisions regarding management of the study patients were made at the discretion of the primary treating teams. To the extent that inter-physician and inter-institutional differences affect choice of sedative medications, and that the various sedative medications (e.g., dexmedetomidine vs midazolam) differentially affect the likelihood of developing delirium, the lack of protocolized sedation practices across study sites may have affected the results.

If the authors are able to proceed with a larger trial building on these results, these issues will need to be incorporated into study design. If this study is completed and shows similar results to those seen in this pilot study, intensivists will be left with the challenging question of how best to treat a vexing problem when non-pharmacologic measures are not sufficient. Recent work has suggested that dexmedetomidine may be associated with less delirium than benzodiazepines,1,2 but the evidence on that question is still inconclusive and no other good pharmacologic options appear to be on the horizon.


  1. Riker RR, et al. Dexmedetomidine vs midazolam for sedation of critically ill patients. JAMA 2009;301:489-499.
  2. Pandharipande PP, et al. Effect of sedation with dexmedetomidine vs lorazepam on acute brain dysfunction in mechanically ventilated patients. JAMA 2007;298:2644-2653.