Antipsychotic Medication for Prevention and Treatment of Delirium in Hospitalized Adults: A Systematic Review and Meta-Analysis.

OBJECTIVES

To evaluate the effectiveness of antipsychotic medications in preventing and treating delirium.

DESIGN

Systematic review and meta-analysis.

SETTING

PubMed, EMBASE, CINAHL, and ClinicalTrials.gov databases were searched from January 1, 1988, to November 26, 2013.

PARTICIPANTS

Adult surgical and medical inpatients.

INTERVENTION: Antipsychotic administration for delirium prevention or treatment in randomized controlled trials or cohort studies.

MEASUREMENTS: Two authors independently reviewed all citations, extracted relevant data, and assessed studies for potential bias. Heterogeneity was considered as chi-square P < .1 or I(2) > 50%. Using a random-effects model (I(2) > 50%) or a fixed-effects model (I(2) < 50%), odds ratios (ORs) were calculated for dichotomous outcomes (delirium incidence and mortality), and mean or standardized mean difference for continuous outcomes (delirium duration, severity, hospital and intensive care unit (ICU) length of stay (LOS)). Sensitivity analyses included postoperative prevention studies only, exclusion of studies with high risk of bias, and typical versus atypical antipsychotics.

RESULTS

Screening of 10,877 eligible records identified 19 studies. In seven studies comparing antipsychotics with placebo or no treatment for delirium prevention after surgery, there was no significant effect on delirium incidence (OR = 0.56, 95% confidence interval (CI) = 0.23-1.34, I(2) = 93%). Using data reported from all 19 studies, antipsychotic use was not associated with change in delirium duration, severity, or hospital or ICU LOS, with high heterogeneity among studies. No association with mortality was detected (OR = 0.90, 95% CI = 0.62-1.29, I(2) = 0%).

CONCLUSION

Current evidence does not support the use of antipsychotics for prevention or treatment of delirium. Additional methodologically rigorous studies using standardized outcome measures are needed.