Prophylactic haloperidol: too early to lose hope

Delirium is a very prevalent syndrome and an important independent predictor of negative outcome in patients in the intensive-care unit (ICU), but there is no proven pharmacological intervention to prevent or treat this disorder. Therefore, the study by Valerie J Page and colleagues, which provided an assessment of the use of haloperidol in the ICU setting, was needed. However, the results of this trial should not be classed as defi nitive because the study has some limitations. First, haloperidol was used as both a prophylaxis and treatment, and the number of patients with delirium at enrolment was not stated. We would expect that the dose needed to treat a patient with delirium would be higher than the prophylactic dose. Also, the criteria for stopping the drug should be reconsidered, since although patients might be 2 days free of delirium, they still could be at risk of developing delirium. Second, regarding the statistical analysis, the primary outcome was the number of delirium-free and comafree days in the first 14 days after randomisation, but patients who died before day 14 were recorded as having zero days free of delirium and coma, restricting the analysis to survivors, and correcting the incidence of the main outcome (delirium) according to the competing event (death). When analysing time-to-event data and competing outcomes, a technique known as cumulative incidence analysis can be used to assess the actual incidence of delirium. This type of analysis would have provided more accurate results in Page and colleagues’ study, and should be used in future trials that intend to address delirium incidence. Last, the study authors conclude that their results do not support the use of haloperidol in patients needing mechanical ventilation, irrespective of whether patients screen positive for delirium or are in a coma. They believe that haloperidol should be reserved for short-term management of acute agitation; however, we think it is too soon to come to this conclusion.