Antipsychotic prescribing patterns during and after critical illness: a prospective cohort study

BACKGROUND Antipsychotics are used to treat delirium in the intensive care unit (ICU) despite unproven efficacy. We hypothesized that atypical antipsychotic treatment in the ICU is a risk factor for antipsychotic prescription at discharge, a practice that might increase risk since long-term use is associated with increased mortality. METHODS After excluding patients on antipsychotics prior to admission, we examined antipsychotic use in a prospective cohort of ICU patients with acute respiratory failure and/or shock. We collected data on medication use from medical records and assessed patients for delirium using the Confusion Assessment Method for the ICU. Using multivariable logistic regression, we analyzed whether age, delirium duration, atypical antipsychotic use, and discharge disposition (each selected a priori) were independent risk factors for discharge on an antipsychotic. We also examined admission Acute Physiology and Chronic Health Evaluation (APACHE) II score, haloperidol use, and days of benzodiazepine use in post hoc analyses. RESULTS After excluding 18 patients due to prior antipsychotic use and three who withdrew, we included 500 patients. Among 208 (42%) treated with an antipsychotic, median (interquartile range) age was 59 (49-69) years and APACHE II score was 26 (22-32), characteristics that were similar among antipsychotic nonusers. Antipsychotic users were more likely than nonusers to have had delirium (93% vs. 61%, p < 0.001). Of the 208 antipsychotic users, 172 survived to hospital discharge, and 42 (24%) of these were prescribed an antipsychotic at discharge. Treatment with an atypical antipsychotic was the only independent risk factor for antipsychotic prescription at discharge (odds ratio 17.6, 95% confidence interval 4.9 to 63.3; p < 0.001). Neither age, delirium duration, nor discharge disposition were risk factors (p = 0.11, 0.38, and 0.12, respectively) in the primary regression model, and post hoc analyses found APACHE II (p = 0.07), haloperidol use (p = 0.16), and days of benzodiazepine use (p = 0.31) were also not risk factors for discharge on an antipsychotic. CONCLUSIONS In this study, antipsychotics were used to treat nearly half of all antipsychotic-naïve ICU patients and were prescribed at discharge to 24% of antipsychotic-treated patients. Treatment with an atypical antipsychotic greatly increased the odds of discharge with an antipsychotic prescription, a practice that should be examined carefully during medication reconciliation since these drugs carry "black box warnings" regarding long-term use.