Evaluation of Discontinuation of Atypical Antipsychotics Prescribed for ICU Delirium

Introduction

The American Psychiatric Association’s Diagnostic and Statistical Manual of Mental Disorders (Fourth Revision) defines delirium as a disturbance of consciousness and cognition that develops over a short period and fluctuates over time. ¹ This change in mental status is common during critical illness, with a prevalence rate between 20% and 80%. Intensive care unit (ICU) delirium is associated with poor patient outcomes, including prolonged hospital stay, increased health care cost, and increased mortality. ² At the time of this study, the 2002 Society of Critical Care Medicine guidelines recommended that patients be screened for delirium, and haloperidol is suggested as the drug of choice for its treatment. ³ Since the publication of this guideline, newer “atypical” antipsychotics have been studied in the treatment of ICU delirium. ^{4 –6} These studies have shown efficacy equal to that of haloperidol and have since gained popularity in use as alternative agents for this specific indication.

As atypical antipsychotics become more broadly used in the treatment of ICU delirium, there is a concern about the potential side effects that accompany long-term use of such agents. These include an increased risk of weight gain, hyperlipidemia, new onset diabetes mellitus, QTc interval prolongation, extrapyramidal side effects, and increased mortality in elderly patients with dementia-related psychosis. ^{7 –9} The financial burden of managing these side effects as well as that of an unnecessary prescription warrants prompt discontinuation of these agents within a reasonable time frame. At our institution, authors observed hesitation to discontinue therapy by some prescribers who credited the improvement in mental status to antipsychotic therapy. Conversely, if mental status had not improved there was concern that patients needed more time on therapy. Therefore, this study was designed to determine whether atypical antipsychotics prescribed for ICU delirium were discontinued upon discharge from the ICU or upon hospital discharge. Furthermore, the economic impact of potentially inappropriate continuation of the prescribed agents, as unnecessary inpatient costs and outpatient annual costs were assessed.

Methods

A retrospective review of electronic and paper medical records was performed of patients over 18 years of age who were newly prescribed a scheduled or as needed atypical antipsychotic while in the medical intensive care unit (MICU). Since these agents were not initiated for other indications in this ICU, all patients were assumed to be receiving treatment for delirium even in the absence of diagnostic documentation. This study was performed in an academic medical center with data collection performed from January 1, 2010, to October 1, 2010. The research protocol was approved by the University of Arizona Institutional Review Board and the University Medical Center Site Review Authority. All data were collected by 1 individual with the exception of physician documentation of delirium. Patients were excluded if they were treated with an antipsychotic within 30 days prior to MICU admission. Repeat admissions during this time frame could be included if they met inclusion criteria.

Data collection included demographic information (age, sex, and race), hospital and ICU length of stay, atypical antipsychotic regimen, Confusion Assessment Method for the ICU (CAM-ICU) scores, discharge status, and whether or not medical staff documented delirium in the medical record. Initial CAM-ICU scores were defined as measurements within 24 hours of the first dose of atypical antipsychotic and final CAM-ICU scores were those taken 24 hours within discontinuation of the agent or prior to transfer, whichever occurred first. Of the patient’s who survived, data collection included whether or not the atypical antipsychotic was continued or discontinued upon transfer from the ICU to general wards and upon hospital discharge. The cost of atypical antipsychotics on the general ward was based on wholesale acquisition cost and the number of doses administered. Cost data for the outpatient setting were determined using the antipsychotic regimen prescribed at discharge and average wholesale price. Full compliance with the medication prescription was assumed for both analyses.

Both descriptive and inferential testing (for differences in proportions) were utilized. Demographic comparisons between the patients discharged or not discharged on antipsychotic medications were compared using a 2 sample t test or Fisher exact test. Potential independent variables (eg, age, sex, race, CAM-ICU score, hospital and ICU length of stay, antipsychotic medication, and discharge site) associated with the discharge of a patient on an antipsychotic medication were evaluated using logistic regression analysis. Significance was defined as P < .05 for all comparisons. Data are reported as mean ± standard deviation.

Results

Of the 111 adult patients who received an atypical antipsychotic while in the MICU, 31 (27.9%) were treated with the agent within 30 days prior to MICU admission and thus excluded. The mean age of the patients was 59 ± 17.2 years. The average length of hospital stay was 22.4 ± 15 days and the average ICU length of stay was 15.6 ± 11.9 days.

Half of the patients had an initial CAM-ICU score recorded (n = 40) of which 26 were reported as positive (65%). Final CAM-ICU scores were recorded for 41 patients of which 12 were positive (29.3%). Physician documentation of delirium in the medical record was available for 54% of patients while in the ICU.

Fifty-nine patients survived to hospital discharge (73.75%). Of these patients, 28 were transferred from the ICU on the atypical agent (47.5%) and 20 were discharged on an atypical antipsychotic (33.9%). If the atypical antipsychotic was not discontinued prior to ICU transfer, 71.4% (20 of 28) of patients received a prescription for 1 of these agents at hospital discharge. There were no significant differences in age, sex, or race between patients discharged or not discharged on an antipsychotic medication. The only factors associated with antipsychotic continuation at hospital discharge were increased length of ICU stay (odds ratio [OR] = 1.073, 95% confidence interval [CI] = 1.021-1.128), length of hospital stay (OR = 1.041, 95% CI = 1.003-1.080), and final CAM-ICU score (OR = 0.068, 95% CI = 0.011-0.428).

In terms of discharge location, 22 (37.3%) patients were discharged home, 18 (31.0%) to skilled nursing facilities, 11 (19.0%) to rehabilitation facilities, 3 (5.3%) to hospice, 3 (5.3%) to inpatient psychiatry, and 1 (1.7%) unknown. Interestingly, no patient discharged to inpatient psychiatry was continued on the agent.

The total cost of continuing atypical antipsychotics on the general ward during the evaluation period was $888. Based on the prescription given at discharge and assuming full compliance as outpatients, the added pharmacy costs for these patients for 1 year totaled $45 107.

Discussion

In recent years, numerous studies have suggested that atypical antipsychotics may be useful in the treatment of ICU delirium. At our institution, 47.5% of patients who were pharmacologically treated for ICU delirium were continued on these agents when transferred to a general medical ward from the ICU. A third of patients initiated on an atypical antipsychotic were discharged from the hospital on the agent. The true impact of prolonged therapy for ICU delirium is unknown, but this certainly raises concern of an increased risk of potentially serious side effects as well as an added financial burden for the patient.

Relatively few studies have evaluated atypical antipsychotics for the treatment of ICU delirium. A randomized study comparing olanzapine and haloperidol in treating ICU delirium in 73 patients found that the resolution of delirium symptoms were similar in both treatment groups. ⁴ More extrapyramidal side effects occurred in the haloperidol group compared to the olanzapine group; however, significance was not reported. Patients were treated for a total of 5 days. Recently, 2 separate trials investigating the effects of quetiapine and ziprasidone on ICU delirium were published. The Modifying the Incidence of Delirium (MIND) trial was a randomized, placebo controlled trial comparing haloperidol, ziprasidone, and placebo for up to 14 days for treatment of delirium in 101 mechanically ventilated medical and surgical ICU patients. ⁵ The primary end point, number of days patients were alive without delirium or coma, was not different between the 3 groups. No differences were found in secondary outcomes such as ventilator-free days, hospital length of stay, mortality, or extrapyramidal side effects. Another prospective randomized trial compared quetiapine to placebo in 36 ICU patients with delirium. ⁶ Therapy was discontinued with resolution of symptoms, greater than or equal to 10 days of therapy or discharge from the ICU. Quetiapine was associated with less time spent in delirium and less time spent agitated versus the placebo; however, duration of mechanical ventilation, length of ICU stay, length of hospitalization, and hospital mortality were not different.

All of these trials had small sample sizes and numerous exclusion criteria, which may limit their generalizability. Furthermore, despite patients spending less time in delirium when quetiapine was used to treat ICU delirium when compared to placebo, these results did not translate into clinically significant end points.

While new trials are underway to assess the effectiveness of atypical antipsychotics in the treatment of delirium in critical and noncritical acute illness, it is important to note that all trials to date had clearly defined time lines in which therapy would be discontinued. ¹⁰ There are no data supporting the long-term use of these agents for acute delirium, even in the setting of suboptimal improvement in mental status.

The reported acute adverse effects of atypical antipsychotics and haloperidol noted in the previously discussed trials include QTC prolongation, hyperglycemia, extrapyramidal symptoms, and sedation. Additional risks related to prolonged usage of atypical antipsychotics include weight gain, hyperlipidemia, new onset diabetes mellitus, hyperprolactinemia, tardive dyskinesia, orthostatic hypotension, and an increased risk of death when used in the elderly individuals for dementia-related psychosis. ^{7 –9} Furthermore, a recent case series found an increased relative risk of hip fracture and pneumonia in the elderly individuals who had began either typical or atypical antipsychotics for treatment of dementia. ¹¹ The highest risk of hip fracture was noted within the first week of initiation of atypical antipsychotics but was still significantly increased at 12 weeks. The incidence of pneumonia was increased at all time points for both typical and atypical antipsychotics. Our patients who were continued on these agents may have experienced side effects related to long-term use, but such outcomes were not investigated in our study.

ICU delirium is often unrecognized by clinicians but may be improving with the validation of assessment tools such as the CAM-ICU. ¹² In our study, the documentation of CAM-ICU score related to the initiation of drug was poor. It is unclear whether these assessments are not being consistently performed or whether they are simply documented incorrectly. Similarly, diagnostic documentation of delirium by the medical staff was poor. Multidisciplinary education on the identification, documentation, and treatment of this significant comorbidity should be among the primary goals for all critical care teams.

The primary limitations of our study are the small sample size of only medical ICU patients and its performance at a single institution. Documentation of CAM-ICU scoring, delirium as a diagnosis, improvement of symptoms, and reasons for continuation was suboptimal. Extent of long-term use and side effects could not be obtained for those discharged on therapy.

Systems should be implemented at ICU discharge as well as hospital discharge to ensure that the utilization of all inpatient medications are carefully evaluated for continued use. This should include regular assessment of the CAM-ICU, involvement of patient and caregivers, medication reconciliation with previously prescribed outpatient regimens, and systems alerts for medications specifically prescribed for an inpatient indication. At our institution, the computerized physician order entry for atypical antipsychotics was modified to indicate use specific for ICU delirium. In addition, a multidisciplinary approach to educate prescribers on delirium recognition, prevention, and treatment has been implemented. Delirium assessment using the CAM-ICU is part of an ICU checklist that is reviewed regularly for each patient on rounds and discussed with the medical team. Although this appears to be prompting earlier discontinuation of therapy for resolved delirium and improved documentation for the rationale for continued use (eg, discharge to hospice), a formal assessment of the effectiveness of these interventions is currently underway.

Conclusion

At one university teaching hospital, patients prescribed an atypical antipsychotic in the medical ICU for delirium were continued on the agent after ICU discharge and many received prescriptions for use after hospital discharge. Systems should be implemented to prevent extended use of these agents without careful evaluation of potential risks and added health care costs.