Pharmacologic management of delirium was evaluated primarily by the use of antipsychotic medications with no identifiable, consistent differences in efficacy and tolerability between first- and second-generation antipsychotics. One study identified a difference in tolerability between antipsychotics and benzodiazepines, suggesting that benzodiazepines be avoided in the delirious population.16
Similarly, pharmacologic treatments for the prevention of incident delirium in hospitalized patients have resulted in controversial outcomes. We found inconsistent results in the development of delirium, with one study suggesting that the duration and severity of delirium symptoms appear to be reduced in patients given haloperidol22
. Two trials with donepezil20,21
, and one with a cholinergic precursor23
, have consistently shown no impact on the development of delirium symptoms when administered to patients undergoing orthopedic surgical procedures.
The studies included in our review enrolled both patients with and without baseline cognitive impairment. Although baseline cognitive function has been recognized as a risk factor for the development of delirium3
, no study stratified outcomes based on patient’s baseline cognitive status. Studies also inconsistently reported differences in hospital length of stay. Whereas one study identified a statistically significant difference in hospital length of stay when given haloperidol for the prevention of delirium22
, another study identified a difference in hospital length of stay only between those who experienced delirium and those who did not, though this difference was not statistically different26
Our systematic review identified several similar reports as those recognized in two recently published reviews13,14
. Seitz and colleagues used similar search criteria and included 14 studies in their analysis13
. Their results evaluated single-agent, non-comparison trials in addition to comparison trials, whereas our search included only comparison studies. The authors identified similar methodological limitations to the available literature as described below, most notably the lack of a placebo comparator. However, the authors recognized that patients exposed to antipsychotics during an acute delirious event did show signs of improvement in nearly all studies. Lacasse and colleagues also identified comparison studies of delirium management in acute settings14
, though both reviews excluded studies aimed at delirium prevention. As in our review, these reports have described an improvement in delirium symptoms following the use of antipsychotic medications, with no particular agent proving superior to others that have been studied.
Similarly, our review supports the recommendations of delirium management as previously published through the American Psychiatric Association (APA)29
and the Society of Critical Care Medicine (SCCM)30
. The APA recommends low-dose haloperidol as a first-line agent in the symptomatic management of delirium episodes, with few comparisons of newer second-generation antipsychotic medications included in their evaluation29
. Our review supports the use of haloperidol as the first-line agent in delirium management based on similar results achieved when first-generation and second-generation antipsychotics were compared in head-to-head trials. SCCM guidelines suggest dose titration of haloperidol with the possible need of continuous infusion to control delirium symptoms30
; however, our review identified no studies that evaluated the use of continuous infusions of haloperidol, and in fact utilized mean daily doses of no more than 6.5 mg, much lower than the range suggested in the critical care guidelines. Our search results suggest that benzodiazepines should not be considered in delirium in patients without a history of psychiatric illness or alcohol withdrawal due to poor outcomes and limited use in the literature included in this review.
Multiple limitations of this review exist that limit the interpretation of this collected data set. First, several of the available studies lacked or failed to report standard methods of randomization processes and blinding techniques, potentially contaminating their results. Notably, studies that had higher JADAD scores were more likely to identify a difference in delirium incidence, severity, duration, or hospital length of stay. Secondly, the included studies also compared a variety of different pharmacologic interventions in different clinical settings, making universal conclusions regarding delirium management inconclusive. To date, no study has compared the pharmacologic management of delirium symptoms with placebo in hospitalized patients. Similarly, the included studies lacked uniformity in reporting delirium severity, duration and length of hospital stay. The studies included in the review were also limited by the utilization of small sample sizes, thereby exposing the risk of type II errors. No differences in treatment effect could be compared between populations with or without baseline cognitive impairment, those with hypoactive or hyperactive delirium, or those with emergent or elective surgical procedures. With the use of a heterogeneous set of population characteristics, interventions, and evaluation scales in evaluating delirium severity and change over time, it is impossible to pool results to improve the strength of recommendation for clinical practice.
Delirium is commonly encountered in hospitalized older patients especially in medical, surgical and critical care units. A lack of clinical data with strong evidence limits the ability to provide evidence-based recommendations for pharmacologic interventions in this population. In patients at risk for delirium or with identifiable signs of mental status changes, screening with the Confusion Assessment Method (CAM/CAM-ICU) is recommended to confirm a diagnosis of delirium29,30
. When appropriate, non-pharmacologic interventions should be performed to minimize the burden of delirium on hospital complications. Haloperidol has been the most widely used and studied agent in hospitalized delirium with suggested benefits in reducing symptom severity and duration of pre-existing delirium episodes. The second-generation antipsychotics may be an alternative in patients who are not candidates for or who do not tolerate first-generation antipsychotics, though the class has shown no benefit over the first-generation antipsychotics on either efficacy or safety parameters in delirium trials. Because of the risks of cardiovascular and musculoskeletal toxicities, the use of antipsychotics in patients with delirium should be continuously evaluated to minimize the potential for toxicity or over-sedation.
The current literature lacks well-designed studies comparing interventions in a controlled environment. Future studies should include appropriate randomization and blinding techniques with adequate sample sizes to ensure accurate and reproducible outcome measurement. Several specific research questions should be addressed, including the most appropriate dose, duration, and time to initiate agents for delirium treatment. Similarly, appropriate prevention interventions should be evaluated with adequately designed studies specifically targeting pain management, sleep cycle restoration, and antipsychotic use as potential targets to reduce delirium episodes.