Pharmacologic Agents for the Prevention and Treatment of Delirium in Patients Undergoing Cardiac Surgery: Systematic Review and Metaanalysis* Jing Lan Mu, MMed; Anna Lee, PhD, MPH; Gavin M. Joynt, MBBCh, FCICM
Objectives: Postcardiac surgery delirium is associated with increased risks of morbidity, cognitive decline, poor health-related quality of life and mortality, and higher healthcare costs. We performed a systematic review of randomized controlled trials to examine the effect of pharmacologic agents for the prevention and the treatment of delirium after cardiac surgery. Data Sources: Electronic search on PubMed, Medline, Embase, Cochrane Central Register of Controlled Trials, ISI Web of Science, and CINAHL up to December 2013. Study Selection: Randomized controlled trials of pharmacologic agents used for the prevention and the treatment of delirium after emergency or elective cardiac surgery in adults. *See also p. 256. All authors: Department of Anaesthesia and Intensive Care, The Chinese University of Hong Kong, Shatin, Hong Kong. Drs. Lee and Joynt conceived the study. Dr. Lee coordinated the review, screened retrieved articles, appraised the quality of articles, extracted the data from articles, wrote to authors, checked data, performed statistical analysis in STATA, wrote initial draft of article, and is guarantor for the review. Dr. Mu undertook electronic database searches, organized retrieval of articles, screened retrieval articles against inclusion criteria, appraised the quality of articles, extracted the data from articles, entered data and performed metaanalyses in Review Manager, and performed previous work that was the foundation of the present study. Dr. Joynt helped resolved disagreements during the data extraction process and made critical revisions of the article for important intellectual content. All authors interpreted the data, helped revised the article, read and approved the final article. The work was performed at the Department of Anaesthesia and Intensive Care, The Chinese University of Hong Kong, Prince of Wales Hospital, Shatin, NT, Hong Kong. Supported, in part, by a grant from the Research Grants Council of the HKSAR, China (Project reference: CUHK469113). Part of this work was presented at the Australian and New Zealand Intensive Care Society/Australian College of Critical Care Nurses Annual Scientific Meeting, Hobart, Australia, October 17–19, 2013. Drs. Lee, Mu, and Joynt received support for article research from the Research Grants Council of the HKSAR, China (Project reference: CUHK469113). Their institutions received grant support from the Research Grants Council of the HKSAR, China (Project reference: CUHK469113). Dr. Mu’s institution received support for travel from The Chinese University of Hong Kong. For information regarding this article, E-mail: [email protected] Copyright © 2014 by the Society of Critical Care Medicine and Lippincott Williams & Wilkins DOI: 10.1097/CCM.0000000000000673
194
www.ccmjournal.org
Data Extraction: We extracted data on patient population, pharmacologic agents, delirium characteristics, rescue treatment, length of stays in the ICU and hospital, and mortality. For each trial, we assessed the risk of bias domains and rated the quality of evidence using the Grading of Recommendations Assessment, Development and Evaluation approach. Data Synthesis: Of the 13 studies (10 prevention and three treatment) involving 5,848 patients, one multicentered randomized controlled trial on prophylactic dexamethasone made up 77% of the total sample size. The use of pharmacologic agents (dexamethasone, rivastigmine, risperidone, ketamine, dexmedetomidine, propofol, and clonidine) reduced the risk of delirium (relative risk, 0.57; 95% CI, 0.40–0.80) with quality of evidence rated as moderate. There was high quality of evidence for no increased risk of mortality (relative risk, 0.89; 95% CI, 0.57–1.38) associated with the use of prophylactic pharmacologic agents. Metaanalysis of treatment trials was not undertaken because of high heterogeneity. In two small trials (total number of patients = 133), haloperidol did not appear to be effective in treating delirium. Conclusions: Moderate to high-quality evidence supports the use of pharmacologic agents for the prevention of delirium, but results are based largely on one randomized controlled trial. The evidence for treating postcardiac surgery delirium with pharmacologic agents is inconclusive. (Crit Care Med 2015; 43:194–204) Key Words: cardiac surgical procedures; delirium/drug therapy; metaanalysis; intensive care; postoperative complications; review; systematic
D
elirium is a syndrome characterized by the acute onset of fluctuating levels of attention, awareness, and cognition (1). Most ICU healthcare professionals recognize delirium not only as a common and potentially serious problem but also that it is frequently underdiagnosed (2). Patients undergoing cardiac surgery have unique characteristics and present special challenges. Dedicated cardiac surgical ICUs are common in the United States, Europe, and Australasia. Patients undergoing cardiac surgery are exposed to specific risk factors for delirium, such as cardiopulmonary January 2015 • Volume 43 • Number 1
Review Articles
bypass and circulatory arrest, and appear at high risk of delirium. The reported incidence varies from 12% to 52% (3, 4). Typically, delirium after cardiac surgery develops on the first to second day after admission to the ICU (3, 4) and lasts for 1 to 3 days (4–6) with hypoactive delirium (88%) being the most common subtype (7). ICU delirium is reversible but associated with poor patient outcomes, such as a prolonged stay in the ICU and in hospital, higher risk of mortality, slower recovery from cognitive impairment, decreased quality of life, and higher healthcare costs (3, 8, 9). Early management of delirium may reduce poor outcomes and may minimize costly treatments for long-term physical and mental disorders. The most recent 2013 American College of Critical Care medicine guidelines for the management of pain, agitation, and delirium in adult patients in ICU provided no recommendation for using pharmacologic agents for the prevention of delirium (10). For the treatment of delirium, the guidelines (10) found no evidence to support the use of haloperidol, lowquality evidence to support the use of atypical antipsychotics and recommended against the use of rivastigmine to reduce the duration of delirium. There was no recommendation for the use of dexmedetomidine for the prevention of delirium, but weak evidence to support its use for sedation, as an alternative to benzodiazepines, to reduce the prevalence of delirium (10). There are emerging, but not definitive, data evaluating the effectiveness of pharmacologic agents for the management of delirium in the cardiac surgical population (11–23). Our hypothesis was that pharmacologic agents are effective for the prevention and the treatment of delirium after cardiac surgery. Therefore, we performed a systematic review and metaanalysis of randomized controlled trials (RCTs) to address the hypothesis.
MATERIALS AND METHODS Literature Search We searched databases, such as Medline (1946 to December 16, 2013), Embase (1974 to December 16, 2013), CINAHL (1937 to December 16, 2013), Cochrane Central Register of Controlled Trials (1991 to December 16, 2013), and ISI Web of Science (1956 to December 16, 2013) using the following MESH headings “delirium,” “surgery,” “cardiac; surgery,” “heart”; “coronary artery bypass,” “cardiac valve annuloplasty,” “heart transplantation,” “haloperidol,” “olanzapine,” “quetiapine,” “ziprasidone,” “risperidone,” “cholinesterase inhibitors,” “rivastigmine,” “donepezil,” “galantamine,” “benzodiazepines,” “lorazepam,” “oxazepam,” “flunitrazepam,” “dexmedetomidine,” “clonidine,” “mivazerol,” “ketamine,” “propofol,” and “dexamethasone”. We also used text words “mitral valve surgery,” “atrial valve surgery,” “open-heart surgery,” “aneurysm repair,” “atypical antipsychotics,” “preventing,” “prevention,” “treatment,” and “control”. We limited the search to RCTs using the terms “randomized controlled trial.” The reference lists of reviews and published Cochrane reviews (24–27), as well as identified studies, were checked for further RCTs. Critical Care Medicine
Inclusion Criteria We included all RCTs of pharmacologic agents for the prevention and the treatment of delirium in adult patients undergoing emergency or elective cardiac surgery. Comparisons of pharmacologic agents with one another or with placebo were included. There was no language restriction. Studies without full text or with comparisons of pharmacologic agents for sedation or pain were excluded. Outcomes Definition The primary outcome was delirium reported as incidence, severity, or duration. Specific tools used to screen or diagnose delirium events for each trial were noted: Diagnostic and Statistical Manual for Mental Disorders (DSM), Confusion Assessment Method for the ICU (CAM-ICU), Confusion Assessment Method (CAM), Intensive Care Delirium Screening Checklist (ICDSC), Delirium Detection Score (DDS), and Nursing Delirium Screening Scale (28). The secondary outcomes included risk of mortality at hospital discharge and adverse effects associated with the use of pharmacologic agents, need for rescue pharmacologic agent for treating delirium, and duration of stays in ICU and hospital. Study Selection and Data Extraction Two reviewers (J.L.M. and A.L.) independently scanned the titles and abstracts, read the full-text articles, extracted the data using a standardized form, and resolved disagreements by discussion and consultation with a third reviewer (G.M.J.). One author (J.L.M.) entered the data into Review Manager 5.2 (Copenhagen; The Nordic Cochrane Collaboration, 2012) and another (A.L.) verified the consistency and accuracy of the data entry. For each trial, we extracted data on patient population, pharmacologic agents, delirium characteristics, rescue treatment (as defined by the study authors), length of stays in the ICU and in hospital, and mortality. Assessment of Risk of Bias Domains and Quality of Evidence We evaluated sources of bias (29): selection bias (random sequence generation and allocation concealment), performance bias (blinding of participants and personnel), detection bias (blinding of outcome assessment), attrition bias (incomplete outcome data), reporting bias (selective reporting), and other biases (baseline imbalance). For each domain, we made a judgment of low risk, high risk, or unclear risk of bias. A high-quality trial was defined when all domains were considered to be at low risk of bias. Conversely, a low-quality trial was defined when one or more of the domains were rated as high risk of bias. J.L.M. and A.L. independently assessed the risk of bias and any disagreements were settled by consensus after discussion with G.M.J. The overall quality of evidence was assessed using the Grading of Recommendations Assessment, Development and Evaluation approach that considers study design, risk of bias in individual trials and potential for publication bias, precision of pooled estimates, consistency of results across studies, www.ccmjournal.org
195
Mu et al
suitability of the individual study populations, interventions, and outcome assessments in directly addressing the clinical question in the systematic review and magnitude of effect (30). There are four levels of evidence (31): 1. high: very confident that the true effect lies close to that of the estimated effect; 2. moderate: the true effect is likely to be close to the estimated effect, but there is a possibility that it is substantially different; 3. low: the true effect may be substantially different from the estimated effect; and 4. very low: the true effect is likely to be substantially different from the estimated effect. Statistical Analysis and Data Synthesis Review Manager 5.2 (Copenhagen; The Nordic Cochrane Collaboration, 2012) and STATA 13.1 software (StataCorp, College Station, TX) were used for data analysis. Relative risk (RR) or mean difference (MD) with 95% CI was reported. Before data synthesis, we estimated sd from se, CI, interquartile range, and range using methods previously described (32, 33). Missing data were obtained by contacting trial authors. We used a DerSimonian and Laird random-effects model to combine the data (33). We assessed the heterogeneity as low, moderate, and high using I2 values of 25%, 50%, and 75% (34). If there was evidence of high heterogeneity, we attempted to explain the reason for it by performing subgroup analyses on the effects of individual pharmacologic agents used for preventing or treating delirium. We conducted a sensitivity analysis on low risk of bias trials to estimate the robustness of results. Contour-enhanced funnel plot was used to detect publication bias for the primary outcome in a funnel plot (35) when at least 10 studies were included in a metaanalysis. A network plot (36) was drawn to give a visual representation of the evidence base but formal network metaanalysis (synthesis of direct and indirect evidence of more than two alternative interventions for the same condition) was not undertaken because there were too few studies.
RESULTS The search on electronic databases yielded 358 citations (Fig. 1). After reviewing the title, abstract, and/or full-text articles, 61 potentially eligible studies were found but 50 were excluded because of nonrandomized study design (n = 28), inclusion of noncardiac surgery in adults (n = 11), and interventions (e.g., bispectral index–guided anesthesia [n = 1], cyclooxygenase-2 inhibitors [n = 1], nitrous oxide [n = 1], and social support [n = 1]) and outcomes (e.g., pain control [n = 3], sedation [n = 1], ICU costs [n = 1], renal function [n = 1], and fast-track cardiac anesthesia [n = 1]) not meeting the inclusion criteria. Two trials (11, 23) were found using other sources. Of the 13 RCTs involving 5,848 patients, 10 were prevention trials (n = 5,643) and three were treatment trials (n = 205) (Table 1). 196
www.ccmjournal.org
Figure 1. Flow chart of systematic search for randomized controlled trials (RCTs).
All trials, except one (16), were two-arm parallel trials (Table 1). The largest trial (12) involving 4,494 patients examined the effect of intraoperative high-dose dexamethasone on the prevalence of major adverse events, of which delirium was one of many outcomes assessed. This trial (12) made up 77% of the total sample size of the 13 RCTs. The pharmacologic agents examined in the 13 RCTs included clonidine (n = 1), dexamethasone (n = 2), dexmedetomidine (n = 3), ketamine (n = 1), haloperidol (n = 2), propofol (n = 1), risperidone (n = 2), and rivastigmine (n = 1) (Table 1). Eleven trials used validated and reliable delirium assessment tools, such as CAM (13, 19) or CAM-ICU (11, 18, 21, 23), DSM (14, 16, 17, 20), and ICDSC (14, 15). The risk of bias domains for each trial is shown in Table 2; four trials (12, 14, 18, 21) were assessed to be at low risk of bias overall and another four trials (16, 19, 22, 23) were at high risk of bias overall. A summary of results and quality of evidence for each outcome pooled is shown in Table 3. Prevention of Delirium Risk of Delirium. The pair-wise comparisons between intervention (dexamethasone, rivastigmine, risperidone, ketamine, dexmedetomidine, propofol, and clonidine) and control groups are shown in a network plot in Figure 2. In the overall analysis (Fig. 3), pharmacologic agents reduced the risk of delirium (RR, 0.57; 95% CI, 0.40–0.80) with a moderate degree of heterogeneity between the 10 studies (I2 = 53%), perhaps because of different control risk ranging from 12% (12) to 50% (16). The corresponding risk difference estimate was –0.15 (95% CI, –0.23 to –0.08). A sensitivity analysis on low-risk bias trials (12, 14, 18, 21) showed a reduction in the risk of delirium associated with pharmacologic agents (RR, 0.57; 95% CI, 0.37–0.86, I2 = 56%). The contour-enhanced funnel plot in Figure 4 showed asymmetry, suggesting the presence of publication bias. January 2015 • Volume 43 • Number 1
Review Articles
Table 1.
Characteristics of Included Studies
Reference (Country)
Strategy
Atalan et al (11) Treatment (Turkey)
Participants
Intervention
Control
Delirium Assessment Tool
Haloperidol 5 mg IM every hour until adequate sedation (max 20 mg/d)
Morphine 5 mg IM every hour until adequate sedation (max 20 mg/d)
CAM-ICU
Dieleman Prevention 4,494 (≥ 18 yr) et al (12) urgent or elective (Netherlands) cardiac surgery + CPB
Dexamethasone 1 mg/kg (max 100 mg) IV after induction of anesthesia before CPB
Placebo
Treatment with neuroleptic drug (not specified)
Gamberini et al (13) (Switzerland)
Rivastigmine 1.5 mg orally 3 times/d starting evening before surgery to POD6
Placebo
CAM
53 (≥ 18 yr) cardiac surgery ± CPB with hyperactive delirium
Prevention 120 (≥ 65 yr) elective cardiac surgery + CPB
Hakim et al (14) Prevention 101 (≥ 65 yr) cardiac Risperidone 0.5 mg orally Placebo (Egypt) 2 times/d starting 4 hr after surgery + CPB extubation with subsyndromal delirium
DSM, ICDSC
Hudetz et al (15) (United States)
ICDSC
Prevention 58 men (55–84 yr) Ketamine 0.5 mg/kg IV elective CABG or bolus during induction of valve replacement/ anesthesia before surgery repair + CPB
Placebo
Maldonado et al Prevention 118 (18–90 yr) (16) (United elective valve States) surgery + CPB
Dexmedetomidine 0.4 μg/kg Propofol 25–50 μg/kg/min DSM (control group A) or loading IV dose, midazolam 0.5–2 mg/hr 0.2–0.7 μg/kg (control group B) IV maintenance IV dose at infusion at sternal sternal closure and infusion closure and weaned off continued after extubation before extubation for up to 24 hr
Mardani et al (17) (Iran)
Prevention 110 (≤ 80 yr) elective CABG
Dexamethasone 8 mg IV before surgery, 8 mg IV 3 times/d for POD1–3
Prakanrattana et al (18) (Thailand)
Prevention 126 (≥ 40 yr) elective cardiac surgery + CPB
Risperidone 1 mg sublingually Placebo when awake in ICU
Placebo
DSM
CAM-ICU
Royse et al (19) Prevention 180 (≥ 18 yr) (Australia) primary CABG + CPB
Propofol target concentration 1.5–3 μg/mL and no volatile anesthesia
Sevoflurane induction and desflurane maintenance anesthesia
CAM
Rubino et al (20) (Italy)
Prevention 30 aortic dissection + CPB
Clonidine 0.5 mg/kg IV bolus at weaning off mechanical ventilation, 1–2 mg/kg/hr infusion over weaning phase
Placebo
DSM
Shehabi et al (21) (Australia)
Prevention 360 (≥ 60 yr) CABG ± valve replacement/ repair + CPB
CAM-ICU Dexmedetomidine Morphine (10–70 μg/ 0.1–0.7 μg/kg/hr with kg/hr) with open-label open-label propofol on propofol on arrival to ICU arrival to ICU and titrated to and titrated to a target a target sedation level sedation level
Tagarakis et al Treatment (22) (Greece)
80 elderly on-pump cardiac surgery with delirium
Haloperidol 5 mg IV
Ondansetron 8 mg IV
Resolution of delirium by 4-point scale
Yapici et al (23) Treatment (Turkey)
72 elective CABG ± valve replacement
Dexmedetomidne Midazolam CAM-ICU 0.3–0.7 μg/kg/hr infusion 0.05–0.2 mg/kg/hr until 1–2 hr after extubation infusion until 1–2 hr after extubation
CPB = cardiopulmonary bypass, CAM-ICU = Confusion Assessment Method for the ICU, DSM = Diagnostic and Statistical Manual for Mental Disorders, ICDSC = Intensive Care Delirium Screening checklist, CABG = coronary artery bypass graft.
Critical Care Medicine
www.ccmjournal.org
197
Mu et al
Table 2.
Risk of Bias Assessment of Included Studies Random Sequence Generation
Allocation Concealment
Blinding of Participants and Personnel
Blinding of Outcome Assessment
Incomplete Outcome Data
Selective Reporting
Other Bias
Unclear
Unclear
Low
Unclear
Low
Low
Low
Dieleman et al (12)
Low
Low
Low
Low
Low
Low
Low
Gamberini et al (13)
Low
Low
Low
Low
Low
Unclear
Low
Hakim et al (14)
Low
Low
Low
Low
Low
Low
Low
Hudetz et al (15)
Unclear
Low
Unclear
Low
Unclear
Low
Low
Maldonado et al (16)
Unclear
Unclear
High
High
Low
Low
Low
Mardani et al (17)
Low
Unclear
Unclear
Unclear
Low
Low
Low
Prakanrattana et al (18)
Low
Low
Low
Low
Low
Low
Low
Royse et al (19)
Low
Low
High
Low
Low
Low
Low
Rubino et al (20)
Unclear
Unclear
Low
Unclear
Low
Low
Low
Shehabi et al (21)
Low
Low
Low
Low
Low
Low
Low
Tagarakis et al (22)
High
Unclear
Unclear
Unclear
Unclear
Low
Low
Unclear
Unclear
Unclear
Unclear
High
Low
Low
Reference
Atalan et al (11)
Yapici et al (23)
The terms “low,” “unclear,” and “high” risk of bias follow the criteria defined by the Cochrane Risk of Bias Methods (29).
There was a subgroup difference effect across the seven different pharmacologic agents (p = 0.04): clonidine (20) (RR, 1.20; 95% CI, 0.47–3.09), dexamethasone (12, 17) (RR, 0.63; 95% CI, 0.32–1.26; I2 = 53%), dexmedetomidine (16, 21) (RR, 0.26; 95% CI, 0.07–1.00; I2 = 50%), ketamine (15) (RR, 0.11; 95% CI, 0.02–0.82), propofol (19) (RR, 0.60; 95% CI, 0.25–1.45), risperidone (14, 18) (RR, 0.38; 95% CI, 0.22–0.66; I2 = 0%), and rivastigmine (13) (RR, 1.08; 95% CI, 0.62–1.87). Severity of Delirium. The Delirium Detection Score was used to measure the severity of delirium in one trial (20). Clonidine was associated with less severity of delirium than placebo group (mean DDS, 0.6 ± 0.7 vs 1.8 ± 0.8, respectively; p < 0.001) (20). The Mini Mental State Examination (MMSE) measures cognitive symptoms only and not severity of delirium per se (37). Prophylactic dexamethasone was associated with improvements in cognition function because the MMSE scores were higher on postoperative days 1 (p = 0.04), 2 (p = 0.04) but not 3 (p = 0.36) than placebo (17). In contrast, the median MMSE scores during the first 6 days after surgery were similar between prophylactic rivastigmine and placebo groups (p > 0.60) after adjusting for type of operations in a separate study (13). Duration of Delirium. Four trials (13, 14, 16, 21) measured the duration of delirium. In the study of Maldonado et al (16), only one patient in the dexmedetomidine group had delirium, and it was not possible to estimate the MD in delirium duration against the combined propofol and midazolam control groups. There was no difference in the duration of delirium associated with rivastigmine (13) (MD, –0.50 d; 95% CI, –1.34 to 0.34) or with risperidone (14) (MD, 0.00 d; 95% CI, –1.15 to 1.15) when compared with placebo. However, dexmedetomidine reduced the duration of delirium (MD, –3.00 d; 95% CI, 198
www.ccmjournal.org
–6.93 to 0.93) when compared with morphine (21). Overall, the intervention group had similar duration of delirium to the control group (MD, –0.41 d; 95% CI, –1.14 to 0.33; I2 = 9%) when pooled (13, 14, 21). Need for Rescue Pharmacologic Agents. There was no reduction in the proportion of patients needing rescue haloperidol associated with prophylactic pharmacologic agents (RR, 0.79; 95% CI, 0.34–1.88; I2 = 0%) (Fig. 5). These results were the same as the sensitivity analysis on low risk of bias trials (14, 21). Similarly, there was no reduction in the proportion of patients needing rescue lorazepam between intervention and control groups (RR, 0.64; 95% CI, 0.25 to 1.67; I2 = 30%) (13, 16). Length of Stay in ICU and Hospital. Pharmacologic agents reduced the length of stay in ICU (MD, –0.30 d; 95% CI, –0.57 to –0.04), but there was high heterogeneity (I2 = 85%) between the 10 trials (n = 5,572) (Fig. 6). There was no subgroup difference across the different pharmacologic agents (p = 0.71). Sensitivity analysis on low risk of bias trials (12, 14, 18, 21) showed no effect of intervention on ICU length of stay (MD, –0.25 d; 95% CI, –0.78 to 0.28) with a high level of heterogeneity (I2 = 93%). For length of hospital stay, there were nine trials involving 5,547 patients. There was no difference between groups (MD, –0.22 d; 95% CI, –0.85 to 0.40) and there was high level of heterogeneity between studies (I2 = 83%) (Fig. 7). There was a subgroup difference across the six types of pharmacologic agents (p < 0.00001); dexamethasone (12, 17) (MD, –0.95 d; 95% CI, –1.18 to –0.71; I2 = 0%), dexmedetomidine (16, 21) (MD, –0.68 d; 95% CI, –1.78 to 0.43; I2 = 49%), ketamine (15) (MD, 1.00 d; 95% CI, –0.82 to 2.82), propofol (19) (MD, 1.00 d; 95% CI, 0.45–1.55), risperidone (14, 18) (MD, 0.03 d; 95% January 2015 • Volume 43 • Number 1
Review Articles
Table 3. Summary of Findings for the Effect of Prophylactic Pharmacologic Agents on Delirium After Cardiac Surgery Estimated Absolute Effects Outcomes
Delirium event
No. of Patients (Studies)
5,572 (10 studies)
Relative Effect (95% CI)
Assumed Risk or Value With Control
RR, 0.57 Study population (0.40–0.80) 145 per 1,000 100 per 1,000a (low risk)
43 fewer per 1,000 (from 20 to 60 fewer)b
a
129 fewer per 1,000 (from 60 to 180 fewer)b
500 per 1,000 (high-risk) 125 (3 studies)
Need for rescue haloperidol
400 (2 studies)
Length of stay in intensive care unit (d)
5,572 (10 studies)
Length of stay in hospital (d)
5,547 (9 studies)
Hospital mortality
5,300 (6 studies)
RR, 0.89 (0.57–1.38)
Quality of the Evidence (GRADE)
⊕⊕⊕⊖ Moderate because of publication bias
215 fewer per 1,000 (from 100 to 300 fewer)b Mean duration was 0.41 d lower (1.14 lower to 0.33 higher)
⊕⊕⊕⊖
12 fewer per 1,000 (from 37 fewer to 49 more)b
⊕⊕⊕⊖
Mean duration ranged from 1 to 4 d
Mean duration was 0.30 d lower (0.57–0.04 lower)
⊕⊕⊖⊖
Mean duration ranged from 6 to 14 d
Mean duration was 0.22 d lower (0.85 lower to 0.40 higher)
⊕⊕⊖⊖
Two fewer per 1,000 (from 7 fewer to 6 more)b
⊕⊕⊕⊕
Mean duration ranged from 3 to 5 d RR, 0.79 (0.34–1.88)
62 fewer per 1,000 (from 29 to 87 fewer)b
a
300 per 1,000 (moderate risk)
Duration of delirium (d)
Risk Difference or Reduction in Value With Intervention
56 per 1,000
16 per 1,000
Moderate because of imprecision Moderate because of imprecision Low because of risk of bias and inconsistency Low because of risk of bias and inconsistency High
GRADE = Grading of Recommendations Assessment, Development and Evaluation, RR = relative risk. a Estimated risk from PREdiction of DELIRium in ICu patients risk prediction model (38). b Absolute risk difference calculated by assumed risk – (assumed risk × relative risk).
CI, –0.66 to 0.71; I2 = 0%), and rivastigmine (13) (MD, 0.00 d; 95% CI, –2.57 to 2.57). Sensitivity analysis on low risk of bias trials (12, 14, 18, 21) showed no effect of intervention on hospital length of stay (MD, –0.33 d; 95% CI, –1.05 to 0.39) and a high level of heterogeneity (I2 = 77%). Hospital Mortality. In the study of Maldonado et al (16), two patients died in the propofol control group. Overall, there was no difference in the risk of hospital mortality between intervention and control groups (RR, 0.89; 95% CI, 0.57–1.38) (Fig. 8). Sensitivity analysis on low risk of bias trials (12, 14, 21) showed that there was no increased risk of hospital mortality associated with pharmacologic agents (RR, 0.90; 95% CI, 0.57–1.42), and this effect was consistent between trials (I2 = 0%).
Figure 2. Network plot of randomized controlled trials.
Critical Care Medicine
Treatment of Delirium Because the three treatment trials (11, 22, 23) in 205 patients reported different clinical outcomes, meta-analysis was not www.ccmjournal.org
199
Mu et al
Figure 3. Prophylactic effect on risk of delirium.
Figure 4. Contour enhanced funnel plot.
undertaken. There was a discrepancy in the number of patients remaining delirious reported in the text and result table in one dexmedomidine treatment trial (23). Attempts to contact the authors for clarification failed. The proportion of patients remaining delirious despite haloperidol treatment was similar to the ondansetron control group (15% vs 17.5%; RR, 0.86; 95% CI, 0.32–2.33) (22). In another trial, there was no difference in the mean (sd) duration of hyperactive delirium between haloperidol and morphine groups (34 ± 17 vs 32 ± 17 hr; p = 0.61) (11). The haloperidol group was more likely to require rescue lorazapam than the morphine group (30.7% vs 3.7%; RR, 8.31; 95% CI, 1.12–61.87) (11). Notably, the risk of reintubation (for unspecified reasons) appeared higher in the haloperidol group than in the morphine group (23.1% vs 3.7%; p = 0.05) (11). However, the risk of hospital mortality was similar between the haloperidol and the morphine groups (7.7% vs 3.7%; RR, 2.08; 95% CI, 0.20–21.55) (11). 200
www.ccmjournal.org
Adverse Effects Four trials (13, 14, 21, 23) described the adverse side effects associated with pharmacologic agents. When compared with placebo, rivastigmine had similar risks for nausea (RR, 1.29; 95% CI, 0.96–1.74), vomiting (RR, 1.16; 95% CI, 0.77–1.77), and diarrhea (RR, 1.21; 95% CI, 0.43–3.38) (13). Patients taking risperidone had a similar risk of extrapyramidal side effect with those in the placebo group (3.9% vs 2.0%; RR, 1.96; 95% CI, 0.18–20.94) (14). The risk of postoperative nausea and vomiting was similar between dexmedetomidine and morphine groups (RR, 0.83; 95% CI, 0.49–1.42) (21). When compared with morphine, dexmedetomidine was associated with a lower risk of systolic hypotension (23.0% vs 38.1%; RR, 0.62; 95% CI, 0.43–0.88) but a higher risk of bradycardia (16.5% vs 6.1%; RR, 2.74; 95% CI, 1.32–5.68) (21). In another study (23), dexmedetomidine was associated with lower mean heart rate at 12 hours (MD, –14 bpm; 95% CI, –20 to –8) and at 24 hours (MD, –18 bpm; 95% CI, –21 to –14) but not at 6 hours (MD, –2 bpm; 95% CI, –9 to 6) after the start of infusion when compared with midazolam.
DISCUSSION This systematic review and meta-analysis of pharmacologic agents found moderate quality of evidence from 10 RCTs (n = 5,572) to support the prophylactic use of pharmacologic agents. The PREdiction of DELIRium in ICu patients risk score (38) defines a risk of 30% (20–40%) as a moderate risk of delirium and 50% (40–60%) as a high risk. The estimated effect of pharmacologic prophylaxis in patients at moderate (30%) and high (50%) risk of delirium would be a corresponding reduction in delirium events of 13 and 22 per 100 patients (Table 3). The majority of RCTs that examined the effect of prophylactic pharmacologic agents on delirium were conducted with low to moderate risk of bias. It should be noted, however, that most of the evidence was derived from a high-quality large multicentered January 2015 • Volume 43 • Number 1
Review Articles
Figure 5. Rescue haloperidol.
Figure 6. Prophylactic effect on ICU length of stay.
Figure 7. Prophylactic effect on hospital length of stay.
trial that showed a significant effect of high-dose intraoperative dexamethasone on the risk of delirium (12). The result of the subgroup analysis on different types of pharmacologic agents on the risk of delirium suggests that Critical Care Medicine
there is probably a credible subgroup effect using the recent criteria outlined by Sun et al (39). In considering whether the type of pharmacologic agent used to prevent delirium is clinically important, additional consideration should be given to www.ccmjournal.org
201
Mu et al
Figure 8. Prophylactic effect on mortality.
the following issues. Although the risperidone subgroup analysis showed that it was effective in preventing delirium, a limitation is that it can only be used after extubation in oral (14) or sublingual form (18). The effect of ketamine may be overestimated in the trial of Hudetz et al (15) because the total fentanyl dose per body weight was not taken into account in the analysis (40). The post hoc analysis by Burkhart et al (41) of the trial of Gamberini et al of rivastigmine (13) showed that when fentanyl was adjusted by body weight, there was a significant association between intraoperative fentanyl use and postoperative delirium (adjusted odds ratio, 4.9; 95% CI, 1.7–13.8 per 10 μg/ kg increase). Dexmedetomidine is a selective α2-receptor agonist used widely in the early postoperative period for sedation and analgesia (42). A recent retrospective cohort study of 1,134 patients undergoing cardiac surgery showed that dexmedetomidine given after cardiopulmonary bypass and continued for less than 24 hours after surgery was associated with lower odds of delirium (adjusted odds ratio, 0.53; 95% CI, 0.37–0.75) (42). In a previous metaanalysis examining factors associated with delirium after cardiac surgery, the authors concluded that “sedation with dexmedetomidine and fast-track weaning protocols may be effective in reducing the risk of delirium”(43). Although our results suggest that dexmedetomidine may reduce the risk of delirium, caution is warranted because the relatively wide confidence interval suggests that the estimate was imprecise. To date, a direct comparison between dexamethasone and dexmedetomidine for the prevention of postcardiac surgery delirium has not been undertaken. We found imprecise estimates for the role of prophylactic pharmacologic agents in reducing the duration of delirium and in the need for rescue haloperidol and lorazepam. Although there was a significant reduction in ICU length of stay (0.30 d; 95% CI, 0.04–0.57) associated with prophylactic pharmacologic agents, the clinical importance of the finding may depend on individual ICU settings because there was a wide variation in the control group length of stay of 1 to 4 days. The different pharmacologic agents in the subgroup analysis did not explain the large heterogeneity in the ICU length of stay data. There 202
www.ccmjournal.org
was low quality of evidence to support a lack of association between all prophylactic pharmacologic agents and hospital length of stay. However, the subgroup analysis found a significant difference in the effect between individual pharmacologic agents. Rates of reported transient minor adverse effects and the risk of hospital mortality were low. However, none of the RCTs included in this systematic review followed up patients beyond hospital discharge or 30-day survival. Lengthening the duration of follow-up reporting in future studies may be indicated because a recent retrospective cohort study reported that dexmedetomidine was associated with a reduction in mortality 1 year after surgery (adjusted odds ratio, 0.49; 95% CI, 0.31– 0.70) (42). The three small RCTs (11, 22, 23) that examined the role of pharmacologic agents for treating established postcardiac surgery delirium were conducted with moderate to high risk of bias. The data reported in studies examining the effectiveness of haloperidol for treating delirium were inconclusive because both trials were underpowered (11, 22). These results support the 2013 American College of Critical Care Medicine guidelines that found no evidence to support the use of haloperidol for delirium treatment (10). The results of this systematic review may be limited because of the methodology used for the diagnosis of delirium. The DSM method is the gold standard used by psychiatrists and was used in four of the 13 RCTs included in this systematic review. The CAM-ICU and ICDSC methods were used in four and two trials, respectively. A metaanalysis of the diagnostic accuracy of CAM-ICU and ICDSC showed that these two screening tools had similar negative likelihood ratios (0.26 vs 0.28), but that CAM-ICU had a higher positive likelihood ratio than ICDSC (16.3 vs 3.06) (28). Thus, there is a possibility that not all delirium events were identified in the RCTs using these screening instruments. Unfortunately, such detection bias is almost unavoidable in delirium studies but is likely to be minimal in this systematic review because the CAM-ICU has excellent interrater reliability (κ = 0.89) (44). Finally, despite our efforts to conduct a thorough search for eligible RCTs, publication bias was detected. January 2015 • Volume 43 • Number 1
Review Articles
Thus, the summary estimates associated with the prophylactic use of pharmacologic agents may be an overestimate.
CONCLUSIONS Moderate to high-quality evidence supports the use of pharmacologic agents for the prevention of delirium, but the conclusion is based largely on one dexamethasone trial (12). Individual pharmacologic agents differ in their effect on the risk of delirium, but the observation of the apparent effectiveness of prophylactic dexmedetomidine requires caution because the results are imprecise. No robust comparison between dexamethasone and dexmedetomidine has been performed. Finally, the evidence for treating postcardiac surgery delirium with pharmacologic agents is inconclusive.
ACKNOWLEDGMENTS We thank Ms Lok Kuen Wong, medical librarian, for her help with developing the search strategies used in the electronic databases and Professor Wilson Tam for statistical review of this article.
REFERENCES
1. American Psychiatric Association: Diagnostic and Statistical Manual of Mental Disorders. Fifth Edition. Arlington, VA, American Psychiatric Association, 2013 2. Patel RP, Gambrell M, Speroff T, et al: Delirium and sedation in the intensive care unit: Survey of behaviors and attitudes of 1384 healthcare professionals. Crit Care Med 2009; 37:825–832 3. Katznelson R, Djaiani GN, Borger MA, et al: Preoperative use of statins is associated with reduced early delirium rates after cardiac surgery. Anesthesiology 2009; 110:67–73 4. Rudolph JL, Jones RN, Levkoff SE, et al: Derivation and validation of a preoperative prediction rule for delirium after cardiac surgery. Circulation 2009; 119:229–236 5. Ely EW, Gautam S, Margolin R, et al: The impact of delirium in the intensive care unit on hospital length of stay. Intensive Care Med 2001; 27:1892–1900 6. Lin SM, Liu CY, Wang CH, et al: The impact of delirium on the survival of mechanically ventilated patients. Crit Care Med 2004; 32:2254–2259 7. McPherson JA, Wagner CE, Boehm LM, et al: Delirium in the cardiovascular ICU: Exploring modifiable risk factors. Crit Care Med 2013; 41:405–413 8. Saczynski JS, Marcantonio ER, Quach L, et al: Cognitive trajectories after postoperative delirium. N Engl J Med 2012; 367:30–39 9. Koster S, Hensens AG, Schuurmans MJ, et al: Consequences of delirium after cardiac operations. Ann Thorac Surg 2012; 93:705–711 10. Barr J, Fraser GL, Puntillo K, et al; American College of Critical Care Medicine: Clinical practice guidelines for the management of pain, agitation, and delirium in adult patients in the intensive care unit. Crit Care Med 2013; 41:263–306 11. Atalan N, Efe Sevim M, Akgün S, et al: Morphine is a reasonable alternative to haloperidol in the treatment of postoperative hyperactive-type delirium after cardiac surgery. J Cardiothorac Vasc Anesth 2013; 27:933–938 12. Dieleman JM, Nierich AP, Rosseel PM, et al; Dexamethasone for Cardiac Surgery (DECS) Study Group: Intraoperative high-dose dexamethasone for cardiac surgery: A randomized controlled trial. JAMA 2012; 308:1761–1767 13. Gamberini M, Bolliger D, Lurati Buse GA, et al: Rivastigmine for the prevention of postoperative delirium in elderly patients undergoing elective cardiac surgery–a randomized controlled trial. Crit Care Med 2009; 37:1762–1768
Critical Care Medicine
14. Hakim SM, Othman AI, Naoum DO: Early treatment with risperidone for subsyndromal delirium after on-pump cardiac surgery in the elderly: A randomized trial. Anesthesiology 2012; 116:987–997 15. Hudetz JA, Patterson KM, Iqbal Z, et al: Ketamine attenuates delirium after cardiac surgery with cardiopulmonary bypass. J Cardiothorac Vasc Anesth 2009; 23:651–657 16. Maldonado JR, Wysong A, van der Starre PJ, et al: Dexmedetomidine and the reduction of postoperative delirium after cardiac surgery. Psychosomatics 2009; 50:206–217 17. Mardani D, Bigdelian H: Prophylaxis of dexamethasone protects patients from further post-operative delirium after cardiac surgery: A randomized trial. J Res Med Sci 2013; 18:137–143 18. Prakanrattana U, Prapaitrakool S: Efficacy of risperidone for prevention of postoperative delirium in cardiac surgery. Anaesth Intensive Care 2007; 35:714–719 19. Royse CF, Andrews DT, Newman SN, et al: The influence of propofol or desflurane on postoperative cognitive dysfunction in patients undergoing coronary artery bypass surgery. Anaesthesia 2011; 66: 455–464 20. Rubino AS, Onorati F, Caroleo S, et al: Impact of clonidine administration on delirium and related respiratory weaning after surgical correction of acute type-A aortic dissection: Results of a pilot study. Interact Cardiovasc Thorac Surg 2010; 10:58–62 21. Shehabi Y, Grant P, Wolfenden H, et al: Prevalence of delirium with dexmedetomidine compared with morphine based therapy after cardiac surgery: A randomized controlled trial (DEXmedetomidine COmpared to Morphine-DEXCOM Study). Anesthesiology 2009; 111:1075–1084 22. Tagarakis GI, Voucharas C, Tsolaki F, et al: Ondasetron versus haloperidol for the treatment of postcardiotomy delirium: A prospective, randomized, double-blinded study. J Cardiothorac Surg 2012; 7:25 23. Yapici N, Coruh T, Kehlibar T, et al: Dexmedetomidine in cardiac surgery patients who fail extubation and present with a delirium state. Heart Surg Forum 2011; 14:E93–E98 24. Lonergan E, Luxenberg J, Areosa SA: Benzodiazepines for delirium. Cochrane Database Syst Rev 2009;CD006379 25. Lonergan E, Britton AM, Luxenberg J, et al: Antipsychotics for delirium. Cochrane Database Syst Rev 2007;CD005594 26. Overshott R, Karim S, Burns A: Cholinesterase inhibitors for delirium. Cochrane Database Syst Rev 2008;CD005317 27. Siddiqi N, Stockdale R, Britton AM, et al: Interventions for preventing delirium in hospitalised patients. Cochrane Database Syst Rev 2007;CD005563 28. Neto AS, Nassar AP Jr, Cardoso SO, et al: Delirium screening in critically ill patients: A systematic review and meta-analysis. Crit Care Med 2012; 40:1946–1951 29. Higgins JP, Altman DG, Gøtzsche PC, et al; Cochrane Bias Methods Group; Cochrane Statistical Methods Group: The Cochrane Collaboration’s tool for assessing risk of bias in randomised trials. BMJ 2011; 343:d5928 30. Guyatt G, Oxman AD, Akl EA, et al: GRADE guidelines: 1. IntroductionGRADE evidence profiles and summary of findings tables. J Clin Epidemiol 2011; 64:383–394 31. Balshem H, Helfand M, Schünemann HJ, et al: GRADE guidelines: 3. Rating the quality of evidence. J Clin Epidemiol 2011; 64: 401–406 32. Walter SD, Yao X: Effect sizes can be calculated for studies reporting ranges for outcome variables in systematic reviews. J Clin Epidemiol 2007; 60:849–852 33. Higgins JPT, Green S: Cochrane Handbook for Systematic Reviews of Interventions Version 5.1.0 [updated March 2011]. The Cochrane Collaboration, 2011. Available at: http://www.cochrane-handbook. org. Accessed November 12, 2012 34. Higgins JP, Thompson SG, Deeks JJ, et al: Measuring inconsistency in meta-analyses. BMJ 2003; 327:557–560 35. Peters JL, Sutton AJ, Jones DR, et al: Contour-enhanced meta-analysis funnel plots help distinguish publication bias from other causes of asymmetry. J Clin Epidemiol 2008; 61:991–996 36. Chaimani A, Higgins JP, Mavridis D, et al: Graphical tools for network meta-analysis in STATA. PLoS One 2013; 8:e76654 www.ccmjournal.org
203
Mu et al 37. Grover S, Kate N: Assessment scales for delirium: A review. World J Psychiatry 2012; 2:58–70 38. van den Boogaard M, Pickkers P, Slooter AJ, et al: Development and validation of PRE-DELIRIC (PREdiction of DELIRium in ICu patients) delirium prediction model for intensive care patients: Observational multicentre study. BMJ 2012; 344:e420 39. Sun X, Ioannidis JP, Agoritsas T, et al: How to use a subgroup analysis: Users’ guide to the medical literature. JAMA 2014; 311:405–411 40. Tse L, Schwarz SK, Bowering JB, et al: Pharmacological risk factors for delirium after cardiac surgery: A review. Curr Neuropharmacol 2012; 10:181–196
204
www.ccmjournal.org
41. Burkhart CS, Dell-Kuster S, Gamberini M, et al: Modifiable and nonmodifiable risk factors for postoperative delirium after cardiac surgery with cardiopulmonary bypass. J Cardiothorac Vasc Anesth 2010; 24:555–559 42. Ji F, Li Z, Nguyen H, et al: Perioperative dexmedetomidine improves outcomes of cardiac surgery. Circulation 2013; 127:1576–1584 43. Lin Y, Chen J, Wang Z: Meta-analysis of factors which influence delirium following cardiac surgery. J Card Surg 2012; 27:481–492 44. Luetz A, Heymann A, Radtke FM, et al: Different assessment tools for intensive care unit delirium: Which score to use? Crit Care Med 2010; 38:409–418
January 2015 • Volume 43 • Number 1
Objectives: Postcardiac surgery delirium is associated with increased risks of morbidity, cognitive decline, poor health-related quality of life and mortality, and higher healthcare costs. We performed a systematic review of randomized controlled trials to examine the effect of pharmacologic agents for the prevention and the treatment of delirium after cardiac surgery. Data Sources: Electronic search on PubMed, Medline, Embase, Cochrane Central Register of Controlled Trials, ISI Web of Science, and CINAHL up to December 2013. Study Selection: Randomized controlled trials of pharmacologic agents used for the prevention and the treatment of delirium after emergency or elective cardiac surgery in adults. *See also p. 256. All authors: Department of Anaesthesia and Intensive Care, The Chinese University of Hong Kong, Shatin, Hong Kong. Drs. Lee and Joynt conceived the study. Dr. Lee coordinated the review, screened retrieved articles, appraised the quality of articles, extracted the data from articles, wrote to authors, checked data, performed statistical analysis in STATA, wrote initial draft of article, and is guarantor for the review. Dr. Mu undertook electronic database searches, organized retrieval of articles, screened retrieval articles against inclusion criteria, appraised the quality of articles, extracted the data from articles, entered data and performed metaanalyses in Review Manager, and performed previous work that was the foundation of the present study. Dr. Joynt helped resolved disagreements during the data extraction process and made critical revisions of the article for important intellectual content. All authors interpreted the data, helped revised the article, read and approved the final article. The work was performed at the Department of Anaesthesia and Intensive Care, The Chinese University of Hong Kong, Prince of Wales Hospital, Shatin, NT, Hong Kong. Supported, in part, by a grant from the Research Grants Council of the HKSAR, China (Project reference: CUHK469113). Part of this work was presented at the Australian and New Zealand Intensive Care Society/Australian College of Critical Care Nurses Annual Scientific Meeting, Hobart, Australia, October 17–19, 2013. Drs. Lee, Mu, and Joynt received support for article research from the Research Grants Council of the HKSAR, China (Project reference: CUHK469113). Their institutions received grant support from the Research Grants Council of the HKSAR, China (Project reference: CUHK469113). Dr. Mu’s institution received support for travel from The Chinese University of Hong Kong. For information regarding this article, E-mail: [email protected] Copyright © 2014 by the Society of Critical Care Medicine and Lippincott Williams & Wilkins DOI: 10.1097/CCM.0000000000000673
194
www.ccmjournal.org
Data Extraction: We extracted data on patient population, pharmacologic agents, delirium characteristics, rescue treatment, length of stays in the ICU and hospital, and mortality. For each trial, we assessed the risk of bias domains and rated the quality of evidence using the Grading of Recommendations Assessment, Development and Evaluation approach. Data Synthesis: Of the 13 studies (10 prevention and three treatment) involving 5,848 patients, one multicentered randomized controlled trial on prophylactic dexamethasone made up 77% of the total sample size. The use of pharmacologic agents (dexamethasone, rivastigmine, risperidone, ketamine, dexmedetomidine, propofol, and clonidine) reduced the risk of delirium (relative risk, 0.57; 95% CI, 0.40–0.80) with quality of evidence rated as moderate. There was high quality of evidence for no increased risk of mortality (relative risk, 0.89; 95% CI, 0.57–1.38) associated with the use of prophylactic pharmacologic agents. Metaanalysis of treatment trials was not undertaken because of high heterogeneity. In two small trials (total number of patients = 133), haloperidol did not appear to be effective in treating delirium. Conclusions: Moderate to high-quality evidence supports the use of pharmacologic agents for the prevention of delirium, but results are based largely on one randomized controlled trial. The evidence for treating postcardiac surgery delirium with pharmacologic agents is inconclusive. (Crit Care Med 2015; 43:194–204) Key Words: cardiac surgical procedures; delirium/drug therapy; metaanalysis; intensive care; postoperative complications; review; systematic
D
elirium is a syndrome characterized by the acute onset of fluctuating levels of attention, awareness, and cognition (1). Most ICU healthcare professionals recognize delirium not only as a common and potentially serious problem but also that it is frequently underdiagnosed (2). Patients undergoing cardiac surgery have unique characteristics and present special challenges. Dedicated cardiac surgical ICUs are common in the United States, Europe, and Australasia. Patients undergoing cardiac surgery are exposed to specific risk factors for delirium, such as cardiopulmonary January 2015 • Volume 43 • Number 1
Review Articles
bypass and circulatory arrest, and appear at high risk of delirium. The reported incidence varies from 12% to 52% (3, 4). Typically, delirium after cardiac surgery develops on the first to second day after admission to the ICU (3, 4) and lasts for 1 to 3 days (4–6) with hypoactive delirium (88%) being the most common subtype (7). ICU delirium is reversible but associated with poor patient outcomes, such as a prolonged stay in the ICU and in hospital, higher risk of mortality, slower recovery from cognitive impairment, decreased quality of life, and higher healthcare costs (3, 8, 9). Early management of delirium may reduce poor outcomes and may minimize costly treatments for long-term physical and mental disorders. The most recent 2013 American College of Critical Care medicine guidelines for the management of pain, agitation, and delirium in adult patients in ICU provided no recommendation for using pharmacologic agents for the prevention of delirium (10). For the treatment of delirium, the guidelines (10) found no evidence to support the use of haloperidol, lowquality evidence to support the use of atypical antipsychotics and recommended against the use of rivastigmine to reduce the duration of delirium. There was no recommendation for the use of dexmedetomidine for the prevention of delirium, but weak evidence to support its use for sedation, as an alternative to benzodiazepines, to reduce the prevalence of delirium (10). There are emerging, but not definitive, data evaluating the effectiveness of pharmacologic agents for the management of delirium in the cardiac surgical population (11–23). Our hypothesis was that pharmacologic agents are effective for the prevention and the treatment of delirium after cardiac surgery. Therefore, we performed a systematic review and metaanalysis of randomized controlled trials (RCTs) to address the hypothesis.
MATERIALS AND METHODS Literature Search We searched databases, such as Medline (1946 to December 16, 2013), Embase (1974 to December 16, 2013), CINAHL (1937 to December 16, 2013), Cochrane Central Register of Controlled Trials (1991 to December 16, 2013), and ISI Web of Science (1956 to December 16, 2013) using the following MESH headings “delirium,” “surgery,” “cardiac; surgery,” “heart”; “coronary artery bypass,” “cardiac valve annuloplasty,” “heart transplantation,” “haloperidol,” “olanzapine,” “quetiapine,” “ziprasidone,” “risperidone,” “cholinesterase inhibitors,” “rivastigmine,” “donepezil,” “galantamine,” “benzodiazepines,” “lorazepam,” “oxazepam,” “flunitrazepam,” “dexmedetomidine,” “clonidine,” “mivazerol,” “ketamine,” “propofol,” and “dexamethasone”. We also used text words “mitral valve surgery,” “atrial valve surgery,” “open-heart surgery,” “aneurysm repair,” “atypical antipsychotics,” “preventing,” “prevention,” “treatment,” and “control”. We limited the search to RCTs using the terms “randomized controlled trial.” The reference lists of reviews and published Cochrane reviews (24–27), as well as identified studies, were checked for further RCTs. Critical Care Medicine
Inclusion Criteria We included all RCTs of pharmacologic agents for the prevention and the treatment of delirium in adult patients undergoing emergency or elective cardiac surgery. Comparisons of pharmacologic agents with one another or with placebo were included. There was no language restriction. Studies without full text or with comparisons of pharmacologic agents for sedation or pain were excluded. Outcomes Definition The primary outcome was delirium reported as incidence, severity, or duration. Specific tools used to screen or diagnose delirium events for each trial were noted: Diagnostic and Statistical Manual for Mental Disorders (DSM), Confusion Assessment Method for the ICU (CAM-ICU), Confusion Assessment Method (CAM), Intensive Care Delirium Screening Checklist (ICDSC), Delirium Detection Score (DDS), and Nursing Delirium Screening Scale (28). The secondary outcomes included risk of mortality at hospital discharge and adverse effects associated with the use of pharmacologic agents, need for rescue pharmacologic agent for treating delirium, and duration of stays in ICU and hospital. Study Selection and Data Extraction Two reviewers (J.L.M. and A.L.) independently scanned the titles and abstracts, read the full-text articles, extracted the data using a standardized form, and resolved disagreements by discussion and consultation with a third reviewer (G.M.J.). One author (J.L.M.) entered the data into Review Manager 5.2 (Copenhagen; The Nordic Cochrane Collaboration, 2012) and another (A.L.) verified the consistency and accuracy of the data entry. For each trial, we extracted data on patient population, pharmacologic agents, delirium characteristics, rescue treatment (as defined by the study authors), length of stays in the ICU and in hospital, and mortality. Assessment of Risk of Bias Domains and Quality of Evidence We evaluated sources of bias (29): selection bias (random sequence generation and allocation concealment), performance bias (blinding of participants and personnel), detection bias (blinding of outcome assessment), attrition bias (incomplete outcome data), reporting bias (selective reporting), and other biases (baseline imbalance). For each domain, we made a judgment of low risk, high risk, or unclear risk of bias. A high-quality trial was defined when all domains were considered to be at low risk of bias. Conversely, a low-quality trial was defined when one or more of the domains were rated as high risk of bias. J.L.M. and A.L. independently assessed the risk of bias and any disagreements were settled by consensus after discussion with G.M.J. The overall quality of evidence was assessed using the Grading of Recommendations Assessment, Development and Evaluation approach that considers study design, risk of bias in individual trials and potential for publication bias, precision of pooled estimates, consistency of results across studies, www.ccmjournal.org
195
Mu et al
suitability of the individual study populations, interventions, and outcome assessments in directly addressing the clinical question in the systematic review and magnitude of effect (30). There are four levels of evidence (31): 1. high: very confident that the true effect lies close to that of the estimated effect; 2. moderate: the true effect is likely to be close to the estimated effect, but there is a possibility that it is substantially different; 3. low: the true effect may be substantially different from the estimated effect; and 4. very low: the true effect is likely to be substantially different from the estimated effect. Statistical Analysis and Data Synthesis Review Manager 5.2 (Copenhagen; The Nordic Cochrane Collaboration, 2012) and STATA 13.1 software (StataCorp, College Station, TX) were used for data analysis. Relative risk (RR) or mean difference (MD) with 95% CI was reported. Before data synthesis, we estimated sd from se, CI, interquartile range, and range using methods previously described (32, 33). Missing data were obtained by contacting trial authors. We used a DerSimonian and Laird random-effects model to combine the data (33). We assessed the heterogeneity as low, moderate, and high using I2 values of 25%, 50%, and 75% (34). If there was evidence of high heterogeneity, we attempted to explain the reason for it by performing subgroup analyses on the effects of individual pharmacologic agents used for preventing or treating delirium. We conducted a sensitivity analysis on low risk of bias trials to estimate the robustness of results. Contour-enhanced funnel plot was used to detect publication bias for the primary outcome in a funnel plot (35) when at least 10 studies were included in a metaanalysis. A network plot (36) was drawn to give a visual representation of the evidence base but formal network metaanalysis (synthesis of direct and indirect evidence of more than two alternative interventions for the same condition) was not undertaken because there were too few studies.
RESULTS The search on electronic databases yielded 358 citations (Fig. 1). After reviewing the title, abstract, and/or full-text articles, 61 potentially eligible studies were found but 50 were excluded because of nonrandomized study design (n = 28), inclusion of noncardiac surgery in adults (n = 11), and interventions (e.g., bispectral index–guided anesthesia [n = 1], cyclooxygenase-2 inhibitors [n = 1], nitrous oxide [n = 1], and social support [n = 1]) and outcomes (e.g., pain control [n = 3], sedation [n = 1], ICU costs [n = 1], renal function [n = 1], and fast-track cardiac anesthesia [n = 1]) not meeting the inclusion criteria. Two trials (11, 23) were found using other sources. Of the 13 RCTs involving 5,848 patients, 10 were prevention trials (n = 5,643) and three were treatment trials (n = 205) (Table 1). 196
www.ccmjournal.org
Figure 1. Flow chart of systematic search for randomized controlled trials (RCTs).
All trials, except one (16), were two-arm parallel trials (Table 1). The largest trial (12) involving 4,494 patients examined the effect of intraoperative high-dose dexamethasone on the prevalence of major adverse events, of which delirium was one of many outcomes assessed. This trial (12) made up 77% of the total sample size of the 13 RCTs. The pharmacologic agents examined in the 13 RCTs included clonidine (n = 1), dexamethasone (n = 2), dexmedetomidine (n = 3), ketamine (n = 1), haloperidol (n = 2), propofol (n = 1), risperidone (n = 2), and rivastigmine (n = 1) (Table 1). Eleven trials used validated and reliable delirium assessment tools, such as CAM (13, 19) or CAM-ICU (11, 18, 21, 23), DSM (14, 16, 17, 20), and ICDSC (14, 15). The risk of bias domains for each trial is shown in Table 2; four trials (12, 14, 18, 21) were assessed to be at low risk of bias overall and another four trials (16, 19, 22, 23) were at high risk of bias overall. A summary of results and quality of evidence for each outcome pooled is shown in Table 3. Prevention of Delirium Risk of Delirium. The pair-wise comparisons between intervention (dexamethasone, rivastigmine, risperidone, ketamine, dexmedetomidine, propofol, and clonidine) and control groups are shown in a network plot in Figure 2. In the overall analysis (Fig. 3), pharmacologic agents reduced the risk of delirium (RR, 0.57; 95% CI, 0.40–0.80) with a moderate degree of heterogeneity between the 10 studies (I2 = 53%), perhaps because of different control risk ranging from 12% (12) to 50% (16). The corresponding risk difference estimate was –0.15 (95% CI, –0.23 to –0.08). A sensitivity analysis on low-risk bias trials (12, 14, 18, 21) showed a reduction in the risk of delirium associated with pharmacologic agents (RR, 0.57; 95% CI, 0.37–0.86, I2 = 56%). The contour-enhanced funnel plot in Figure 4 showed asymmetry, suggesting the presence of publication bias. January 2015 • Volume 43 • Number 1
Review Articles
Table 1.
Characteristics of Included Studies
Reference (Country)
Strategy
Atalan et al (11) Treatment (Turkey)
Participants
Intervention
Control
Delirium Assessment Tool
Haloperidol 5 mg IM every hour until adequate sedation (max 20 mg/d)
Morphine 5 mg IM every hour until adequate sedation (max 20 mg/d)
CAM-ICU
Dieleman Prevention 4,494 (≥ 18 yr) et al (12) urgent or elective (Netherlands) cardiac surgery + CPB
Dexamethasone 1 mg/kg (max 100 mg) IV after induction of anesthesia before CPB
Placebo
Treatment with neuroleptic drug (not specified)
Gamberini et al (13) (Switzerland)
Rivastigmine 1.5 mg orally 3 times/d starting evening before surgery to POD6
Placebo
CAM
53 (≥ 18 yr) cardiac surgery ± CPB with hyperactive delirium
Prevention 120 (≥ 65 yr) elective cardiac surgery + CPB
Hakim et al (14) Prevention 101 (≥ 65 yr) cardiac Risperidone 0.5 mg orally Placebo (Egypt) 2 times/d starting 4 hr after surgery + CPB extubation with subsyndromal delirium
DSM, ICDSC
Hudetz et al (15) (United States)
ICDSC
Prevention 58 men (55–84 yr) Ketamine 0.5 mg/kg IV elective CABG or bolus during induction of valve replacement/ anesthesia before surgery repair + CPB
Placebo
Maldonado et al Prevention 118 (18–90 yr) (16) (United elective valve States) surgery + CPB
Dexmedetomidine 0.4 μg/kg Propofol 25–50 μg/kg/min DSM (control group A) or loading IV dose, midazolam 0.5–2 mg/hr 0.2–0.7 μg/kg (control group B) IV maintenance IV dose at infusion at sternal sternal closure and infusion closure and weaned off continued after extubation before extubation for up to 24 hr
Mardani et al (17) (Iran)
Prevention 110 (≤ 80 yr) elective CABG
Dexamethasone 8 mg IV before surgery, 8 mg IV 3 times/d for POD1–3
Prakanrattana et al (18) (Thailand)
Prevention 126 (≥ 40 yr) elective cardiac surgery + CPB
Risperidone 1 mg sublingually Placebo when awake in ICU
Placebo
DSM
CAM-ICU
Royse et al (19) Prevention 180 (≥ 18 yr) (Australia) primary CABG + CPB
Propofol target concentration 1.5–3 μg/mL and no volatile anesthesia
Sevoflurane induction and desflurane maintenance anesthesia
CAM
Rubino et al (20) (Italy)
Prevention 30 aortic dissection + CPB
Clonidine 0.5 mg/kg IV bolus at weaning off mechanical ventilation, 1–2 mg/kg/hr infusion over weaning phase
Placebo
DSM
Shehabi et al (21) (Australia)
Prevention 360 (≥ 60 yr) CABG ± valve replacement/ repair + CPB
CAM-ICU Dexmedetomidine Morphine (10–70 μg/ 0.1–0.7 μg/kg/hr with kg/hr) with open-label open-label propofol on propofol on arrival to ICU arrival to ICU and titrated to and titrated to a target a target sedation level sedation level
Tagarakis et al Treatment (22) (Greece)
80 elderly on-pump cardiac surgery with delirium
Haloperidol 5 mg IV
Ondansetron 8 mg IV
Resolution of delirium by 4-point scale
Yapici et al (23) Treatment (Turkey)
72 elective CABG ± valve replacement
Dexmedetomidne Midazolam CAM-ICU 0.3–0.7 μg/kg/hr infusion 0.05–0.2 mg/kg/hr until 1–2 hr after extubation infusion until 1–2 hr after extubation
CPB = cardiopulmonary bypass, CAM-ICU = Confusion Assessment Method for the ICU, DSM = Diagnostic and Statistical Manual for Mental Disorders, ICDSC = Intensive Care Delirium Screening checklist, CABG = coronary artery bypass graft.
Critical Care Medicine
www.ccmjournal.org
197
Mu et al
Table 2.
Risk of Bias Assessment of Included Studies Random Sequence Generation
Allocation Concealment
Blinding of Participants and Personnel
Blinding of Outcome Assessment
Incomplete Outcome Data
Selective Reporting
Other Bias
Unclear
Unclear
Low
Unclear
Low
Low
Low
Dieleman et al (12)
Low
Low
Low
Low
Low
Low
Low
Gamberini et al (13)
Low
Low
Low
Low
Low
Unclear
Low
Hakim et al (14)
Low
Low
Low
Low
Low
Low
Low
Hudetz et al (15)
Unclear
Low
Unclear
Low
Unclear
Low
Low
Maldonado et al (16)
Unclear
Unclear
High
High
Low
Low
Low
Mardani et al (17)
Low
Unclear
Unclear
Unclear
Low
Low
Low
Prakanrattana et al (18)
Low
Low
Low
Low
Low
Low
Low
Royse et al (19)
Low
Low
High
Low
Low
Low
Low
Rubino et al (20)
Unclear
Unclear
Low
Unclear
Low
Low
Low
Shehabi et al (21)
Low
Low
Low
Low
Low
Low
Low
Tagarakis et al (22)
High
Unclear
Unclear
Unclear
Unclear
Low
Low
Unclear
Unclear
Unclear
Unclear
High
Low
Low
Reference
Atalan et al (11)
Yapici et al (23)
The terms “low,” “unclear,” and “high” risk of bias follow the criteria defined by the Cochrane Risk of Bias Methods (29).
There was a subgroup difference effect across the seven different pharmacologic agents (p = 0.04): clonidine (20) (RR, 1.20; 95% CI, 0.47–3.09), dexamethasone (12, 17) (RR, 0.63; 95% CI, 0.32–1.26; I2 = 53%), dexmedetomidine (16, 21) (RR, 0.26; 95% CI, 0.07–1.00; I2 = 50%), ketamine (15) (RR, 0.11; 95% CI, 0.02–0.82), propofol (19) (RR, 0.60; 95% CI, 0.25–1.45), risperidone (14, 18) (RR, 0.38; 95% CI, 0.22–0.66; I2 = 0%), and rivastigmine (13) (RR, 1.08; 95% CI, 0.62–1.87). Severity of Delirium. The Delirium Detection Score was used to measure the severity of delirium in one trial (20). Clonidine was associated with less severity of delirium than placebo group (mean DDS, 0.6 ± 0.7 vs 1.8 ± 0.8, respectively; p < 0.001) (20). The Mini Mental State Examination (MMSE) measures cognitive symptoms only and not severity of delirium per se (37). Prophylactic dexamethasone was associated with improvements in cognition function because the MMSE scores were higher on postoperative days 1 (p = 0.04), 2 (p = 0.04) but not 3 (p = 0.36) than placebo (17). In contrast, the median MMSE scores during the first 6 days after surgery were similar between prophylactic rivastigmine and placebo groups (p > 0.60) after adjusting for type of operations in a separate study (13). Duration of Delirium. Four trials (13, 14, 16, 21) measured the duration of delirium. In the study of Maldonado et al (16), only one patient in the dexmedetomidine group had delirium, and it was not possible to estimate the MD in delirium duration against the combined propofol and midazolam control groups. There was no difference in the duration of delirium associated with rivastigmine (13) (MD, –0.50 d; 95% CI, –1.34 to 0.34) or with risperidone (14) (MD, 0.00 d; 95% CI, –1.15 to 1.15) when compared with placebo. However, dexmedetomidine reduced the duration of delirium (MD, –3.00 d; 95% CI, 198
www.ccmjournal.org
–6.93 to 0.93) when compared with morphine (21). Overall, the intervention group had similar duration of delirium to the control group (MD, –0.41 d; 95% CI, –1.14 to 0.33; I2 = 9%) when pooled (13, 14, 21). Need for Rescue Pharmacologic Agents. There was no reduction in the proportion of patients needing rescue haloperidol associated with prophylactic pharmacologic agents (RR, 0.79; 95% CI, 0.34–1.88; I2 = 0%) (Fig. 5). These results were the same as the sensitivity analysis on low risk of bias trials (14, 21). Similarly, there was no reduction in the proportion of patients needing rescue lorazepam between intervention and control groups (RR, 0.64; 95% CI, 0.25 to 1.67; I2 = 30%) (13, 16). Length of Stay in ICU and Hospital. Pharmacologic agents reduced the length of stay in ICU (MD, –0.30 d; 95% CI, –0.57 to –0.04), but there was high heterogeneity (I2 = 85%) between the 10 trials (n = 5,572) (Fig. 6). There was no subgroup difference across the different pharmacologic agents (p = 0.71). Sensitivity analysis on low risk of bias trials (12, 14, 18, 21) showed no effect of intervention on ICU length of stay (MD, –0.25 d; 95% CI, –0.78 to 0.28) with a high level of heterogeneity (I2 = 93%). For length of hospital stay, there were nine trials involving 5,547 patients. There was no difference between groups (MD, –0.22 d; 95% CI, –0.85 to 0.40) and there was high level of heterogeneity between studies (I2 = 83%) (Fig. 7). There was a subgroup difference across the six types of pharmacologic agents (p < 0.00001); dexamethasone (12, 17) (MD, –0.95 d; 95% CI, –1.18 to –0.71; I2 = 0%), dexmedetomidine (16, 21) (MD, –0.68 d; 95% CI, –1.78 to 0.43; I2 = 49%), ketamine (15) (MD, 1.00 d; 95% CI, –0.82 to 2.82), propofol (19) (MD, 1.00 d; 95% CI, 0.45–1.55), risperidone (14, 18) (MD, 0.03 d; 95% January 2015 • Volume 43 • Number 1
Review Articles
Table 3. Summary of Findings for the Effect of Prophylactic Pharmacologic Agents on Delirium After Cardiac Surgery Estimated Absolute Effects Outcomes
Delirium event
No. of Patients (Studies)
5,572 (10 studies)
Relative Effect (95% CI)
Assumed Risk or Value With Control
RR, 0.57 Study population (0.40–0.80) 145 per 1,000 100 per 1,000a (low risk)
43 fewer per 1,000 (from 20 to 60 fewer)b
a
129 fewer per 1,000 (from 60 to 180 fewer)b
500 per 1,000 (high-risk) 125 (3 studies)
Need for rescue haloperidol
400 (2 studies)
Length of stay in intensive care unit (d)
5,572 (10 studies)
Length of stay in hospital (d)
5,547 (9 studies)
Hospital mortality
5,300 (6 studies)
RR, 0.89 (0.57–1.38)
Quality of the Evidence (GRADE)
⊕⊕⊕⊖ Moderate because of publication bias
215 fewer per 1,000 (from 100 to 300 fewer)b Mean duration was 0.41 d lower (1.14 lower to 0.33 higher)
⊕⊕⊕⊖
12 fewer per 1,000 (from 37 fewer to 49 more)b
⊕⊕⊕⊖
Mean duration ranged from 1 to 4 d
Mean duration was 0.30 d lower (0.57–0.04 lower)
⊕⊕⊖⊖
Mean duration ranged from 6 to 14 d
Mean duration was 0.22 d lower (0.85 lower to 0.40 higher)
⊕⊕⊖⊖
Two fewer per 1,000 (from 7 fewer to 6 more)b
⊕⊕⊕⊕
Mean duration ranged from 3 to 5 d RR, 0.79 (0.34–1.88)
62 fewer per 1,000 (from 29 to 87 fewer)b
a
300 per 1,000 (moderate risk)
Duration of delirium (d)
Risk Difference or Reduction in Value With Intervention
56 per 1,000
16 per 1,000
Moderate because of imprecision Moderate because of imprecision Low because of risk of bias and inconsistency Low because of risk of bias and inconsistency High
GRADE = Grading of Recommendations Assessment, Development and Evaluation, RR = relative risk. a Estimated risk from PREdiction of DELIRium in ICu patients risk prediction model (38). b Absolute risk difference calculated by assumed risk – (assumed risk × relative risk).
CI, –0.66 to 0.71; I2 = 0%), and rivastigmine (13) (MD, 0.00 d; 95% CI, –2.57 to 2.57). Sensitivity analysis on low risk of bias trials (12, 14, 18, 21) showed no effect of intervention on hospital length of stay (MD, –0.33 d; 95% CI, –1.05 to 0.39) and a high level of heterogeneity (I2 = 77%). Hospital Mortality. In the study of Maldonado et al (16), two patients died in the propofol control group. Overall, there was no difference in the risk of hospital mortality between intervention and control groups (RR, 0.89; 95% CI, 0.57–1.38) (Fig. 8). Sensitivity analysis on low risk of bias trials (12, 14, 21) showed that there was no increased risk of hospital mortality associated with pharmacologic agents (RR, 0.90; 95% CI, 0.57–1.42), and this effect was consistent between trials (I2 = 0%).
Figure 2. Network plot of randomized controlled trials.
Critical Care Medicine
Treatment of Delirium Because the three treatment trials (11, 22, 23) in 205 patients reported different clinical outcomes, meta-analysis was not www.ccmjournal.org
199
Mu et al
Figure 3. Prophylactic effect on risk of delirium.
Figure 4. Contour enhanced funnel plot.
undertaken. There was a discrepancy in the number of patients remaining delirious reported in the text and result table in one dexmedomidine treatment trial (23). Attempts to contact the authors for clarification failed. The proportion of patients remaining delirious despite haloperidol treatment was similar to the ondansetron control group (15% vs 17.5%; RR, 0.86; 95% CI, 0.32–2.33) (22). In another trial, there was no difference in the mean (sd) duration of hyperactive delirium between haloperidol and morphine groups (34 ± 17 vs 32 ± 17 hr; p = 0.61) (11). The haloperidol group was more likely to require rescue lorazapam than the morphine group (30.7% vs 3.7%; RR, 8.31; 95% CI, 1.12–61.87) (11). Notably, the risk of reintubation (for unspecified reasons) appeared higher in the haloperidol group than in the morphine group (23.1% vs 3.7%; p = 0.05) (11). However, the risk of hospital mortality was similar between the haloperidol and the morphine groups (7.7% vs 3.7%; RR, 2.08; 95% CI, 0.20–21.55) (11). 200
www.ccmjournal.org
Adverse Effects Four trials (13, 14, 21, 23) described the adverse side effects associated with pharmacologic agents. When compared with placebo, rivastigmine had similar risks for nausea (RR, 1.29; 95% CI, 0.96–1.74), vomiting (RR, 1.16; 95% CI, 0.77–1.77), and diarrhea (RR, 1.21; 95% CI, 0.43–3.38) (13). Patients taking risperidone had a similar risk of extrapyramidal side effect with those in the placebo group (3.9% vs 2.0%; RR, 1.96; 95% CI, 0.18–20.94) (14). The risk of postoperative nausea and vomiting was similar between dexmedetomidine and morphine groups (RR, 0.83; 95% CI, 0.49–1.42) (21). When compared with morphine, dexmedetomidine was associated with a lower risk of systolic hypotension (23.0% vs 38.1%; RR, 0.62; 95% CI, 0.43–0.88) but a higher risk of bradycardia (16.5% vs 6.1%; RR, 2.74; 95% CI, 1.32–5.68) (21). In another study (23), dexmedetomidine was associated with lower mean heart rate at 12 hours (MD, –14 bpm; 95% CI, –20 to –8) and at 24 hours (MD, –18 bpm; 95% CI, –21 to –14) but not at 6 hours (MD, –2 bpm; 95% CI, –9 to 6) after the start of infusion when compared with midazolam.
DISCUSSION This systematic review and meta-analysis of pharmacologic agents found moderate quality of evidence from 10 RCTs (n = 5,572) to support the prophylactic use of pharmacologic agents. The PREdiction of DELIRium in ICu patients risk score (38) defines a risk of 30% (20–40%) as a moderate risk of delirium and 50% (40–60%) as a high risk. The estimated effect of pharmacologic prophylaxis in patients at moderate (30%) and high (50%) risk of delirium would be a corresponding reduction in delirium events of 13 and 22 per 100 patients (Table 3). The majority of RCTs that examined the effect of prophylactic pharmacologic agents on delirium were conducted with low to moderate risk of bias. It should be noted, however, that most of the evidence was derived from a high-quality large multicentered January 2015 • Volume 43 • Number 1
Review Articles
Figure 5. Rescue haloperidol.
Figure 6. Prophylactic effect on ICU length of stay.
Figure 7. Prophylactic effect on hospital length of stay.
trial that showed a significant effect of high-dose intraoperative dexamethasone on the risk of delirium (12). The result of the subgroup analysis on different types of pharmacologic agents on the risk of delirium suggests that Critical Care Medicine
there is probably a credible subgroup effect using the recent criteria outlined by Sun et al (39). In considering whether the type of pharmacologic agent used to prevent delirium is clinically important, additional consideration should be given to www.ccmjournal.org
201
Mu et al
Figure 8. Prophylactic effect on mortality.
the following issues. Although the risperidone subgroup analysis showed that it was effective in preventing delirium, a limitation is that it can only be used after extubation in oral (14) or sublingual form (18). The effect of ketamine may be overestimated in the trial of Hudetz et al (15) because the total fentanyl dose per body weight was not taken into account in the analysis (40). The post hoc analysis by Burkhart et al (41) of the trial of Gamberini et al of rivastigmine (13) showed that when fentanyl was adjusted by body weight, there was a significant association between intraoperative fentanyl use and postoperative delirium (adjusted odds ratio, 4.9; 95% CI, 1.7–13.8 per 10 μg/ kg increase). Dexmedetomidine is a selective α2-receptor agonist used widely in the early postoperative period for sedation and analgesia (42). A recent retrospective cohort study of 1,134 patients undergoing cardiac surgery showed that dexmedetomidine given after cardiopulmonary bypass and continued for less than 24 hours after surgery was associated with lower odds of delirium (adjusted odds ratio, 0.53; 95% CI, 0.37–0.75) (42). In a previous metaanalysis examining factors associated with delirium after cardiac surgery, the authors concluded that “sedation with dexmedetomidine and fast-track weaning protocols may be effective in reducing the risk of delirium”(43). Although our results suggest that dexmedetomidine may reduce the risk of delirium, caution is warranted because the relatively wide confidence interval suggests that the estimate was imprecise. To date, a direct comparison between dexamethasone and dexmedetomidine for the prevention of postcardiac surgery delirium has not been undertaken. We found imprecise estimates for the role of prophylactic pharmacologic agents in reducing the duration of delirium and in the need for rescue haloperidol and lorazepam. Although there was a significant reduction in ICU length of stay (0.30 d; 95% CI, 0.04–0.57) associated with prophylactic pharmacologic agents, the clinical importance of the finding may depend on individual ICU settings because there was a wide variation in the control group length of stay of 1 to 4 days. The different pharmacologic agents in the subgroup analysis did not explain the large heterogeneity in the ICU length of stay data. There 202
www.ccmjournal.org
was low quality of evidence to support a lack of association between all prophylactic pharmacologic agents and hospital length of stay. However, the subgroup analysis found a significant difference in the effect between individual pharmacologic agents. Rates of reported transient minor adverse effects and the risk of hospital mortality were low. However, none of the RCTs included in this systematic review followed up patients beyond hospital discharge or 30-day survival. Lengthening the duration of follow-up reporting in future studies may be indicated because a recent retrospective cohort study reported that dexmedetomidine was associated with a reduction in mortality 1 year after surgery (adjusted odds ratio, 0.49; 95% CI, 0.31– 0.70) (42). The three small RCTs (11, 22, 23) that examined the role of pharmacologic agents for treating established postcardiac surgery delirium were conducted with moderate to high risk of bias. The data reported in studies examining the effectiveness of haloperidol for treating delirium were inconclusive because both trials were underpowered (11, 22). These results support the 2013 American College of Critical Care Medicine guidelines that found no evidence to support the use of haloperidol for delirium treatment (10). The results of this systematic review may be limited because of the methodology used for the diagnosis of delirium. The DSM method is the gold standard used by psychiatrists and was used in four of the 13 RCTs included in this systematic review. The CAM-ICU and ICDSC methods were used in four and two trials, respectively. A metaanalysis of the diagnostic accuracy of CAM-ICU and ICDSC showed that these two screening tools had similar negative likelihood ratios (0.26 vs 0.28), but that CAM-ICU had a higher positive likelihood ratio than ICDSC (16.3 vs 3.06) (28). Thus, there is a possibility that not all delirium events were identified in the RCTs using these screening instruments. Unfortunately, such detection bias is almost unavoidable in delirium studies but is likely to be minimal in this systematic review because the CAM-ICU has excellent interrater reliability (κ = 0.89) (44). Finally, despite our efforts to conduct a thorough search for eligible RCTs, publication bias was detected. January 2015 • Volume 43 • Number 1
Review Articles
Thus, the summary estimates associated with the prophylactic use of pharmacologic agents may be an overestimate.
CONCLUSIONS Moderate to high-quality evidence supports the use of pharmacologic agents for the prevention of delirium, but the conclusion is based largely on one dexamethasone trial (12). Individual pharmacologic agents differ in their effect on the risk of delirium, but the observation of the apparent effectiveness of prophylactic dexmedetomidine requires caution because the results are imprecise. No robust comparison between dexamethasone and dexmedetomidine has been performed. Finally, the evidence for treating postcardiac surgery delirium with pharmacologic agents is inconclusive.
ACKNOWLEDGMENTS We thank Ms Lok Kuen Wong, medical librarian, for her help with developing the search strategies used in the electronic databases and Professor Wilson Tam for statistical review of this article.
REFERENCES
1. American Psychiatric Association: Diagnostic and Statistical Manual of Mental Disorders. Fifth Edition. Arlington, VA, American Psychiatric Association, 2013 2. Patel RP, Gambrell M, Speroff T, et al: Delirium and sedation in the intensive care unit: Survey of behaviors and attitudes of 1384 healthcare professionals. Crit Care Med 2009; 37:825–832 3. Katznelson R, Djaiani GN, Borger MA, et al: Preoperative use of statins is associated with reduced early delirium rates after cardiac surgery. Anesthesiology 2009; 110:67–73 4. Rudolph JL, Jones RN, Levkoff SE, et al: Derivation and validation of a preoperative prediction rule for delirium after cardiac surgery. Circulation 2009; 119:229–236 5. Ely EW, Gautam S, Margolin R, et al: The impact of delirium in the intensive care unit on hospital length of stay. Intensive Care Med 2001; 27:1892–1900 6. Lin SM, Liu CY, Wang CH, et al: The impact of delirium on the survival of mechanically ventilated patients. Crit Care Med 2004; 32:2254–2259 7. McPherson JA, Wagner CE, Boehm LM, et al: Delirium in the cardiovascular ICU: Exploring modifiable risk factors. Crit Care Med 2013; 41:405–413 8. Saczynski JS, Marcantonio ER, Quach L, et al: Cognitive trajectories after postoperative delirium. N Engl J Med 2012; 367:30–39 9. Koster S, Hensens AG, Schuurmans MJ, et al: Consequences of delirium after cardiac operations. Ann Thorac Surg 2012; 93:705–711 10. Barr J, Fraser GL, Puntillo K, et al; American College of Critical Care Medicine: Clinical practice guidelines for the management of pain, agitation, and delirium in adult patients in the intensive care unit. Crit Care Med 2013; 41:263–306 11. Atalan N, Efe Sevim M, Akgün S, et al: Morphine is a reasonable alternative to haloperidol in the treatment of postoperative hyperactive-type delirium after cardiac surgery. J Cardiothorac Vasc Anesth 2013; 27:933–938 12. Dieleman JM, Nierich AP, Rosseel PM, et al; Dexamethasone for Cardiac Surgery (DECS) Study Group: Intraoperative high-dose dexamethasone for cardiac surgery: A randomized controlled trial. JAMA 2012; 308:1761–1767 13. Gamberini M, Bolliger D, Lurati Buse GA, et al: Rivastigmine for the prevention of postoperative delirium in elderly patients undergoing elective cardiac surgery–a randomized controlled trial. Crit Care Med 2009; 37:1762–1768
Critical Care Medicine
14. Hakim SM, Othman AI, Naoum DO: Early treatment with risperidone for subsyndromal delirium after on-pump cardiac surgery in the elderly: A randomized trial. Anesthesiology 2012; 116:987–997 15. Hudetz JA, Patterson KM, Iqbal Z, et al: Ketamine attenuates delirium after cardiac surgery with cardiopulmonary bypass. J Cardiothorac Vasc Anesth 2009; 23:651–657 16. Maldonado JR, Wysong A, van der Starre PJ, et al: Dexmedetomidine and the reduction of postoperative delirium after cardiac surgery. Psychosomatics 2009; 50:206–217 17. Mardani D, Bigdelian H: Prophylaxis of dexamethasone protects patients from further post-operative delirium after cardiac surgery: A randomized trial. J Res Med Sci 2013; 18:137–143 18. Prakanrattana U, Prapaitrakool S: Efficacy of risperidone for prevention of postoperative delirium in cardiac surgery. Anaesth Intensive Care 2007; 35:714–719 19. Royse CF, Andrews DT, Newman SN, et al: The influence of propofol or desflurane on postoperative cognitive dysfunction in patients undergoing coronary artery bypass surgery. Anaesthesia 2011; 66: 455–464 20. Rubino AS, Onorati F, Caroleo S, et al: Impact of clonidine administration on delirium and related respiratory weaning after surgical correction of acute type-A aortic dissection: Results of a pilot study. Interact Cardiovasc Thorac Surg 2010; 10:58–62 21. Shehabi Y, Grant P, Wolfenden H, et al: Prevalence of delirium with dexmedetomidine compared with morphine based therapy after cardiac surgery: A randomized controlled trial (DEXmedetomidine COmpared to Morphine-DEXCOM Study). Anesthesiology 2009; 111:1075–1084 22. Tagarakis GI, Voucharas C, Tsolaki F, et al: Ondasetron versus haloperidol for the treatment of postcardiotomy delirium: A prospective, randomized, double-blinded study. J Cardiothorac Surg 2012; 7:25 23. Yapici N, Coruh T, Kehlibar T, et al: Dexmedetomidine in cardiac surgery patients who fail extubation and present with a delirium state. Heart Surg Forum 2011; 14:E93–E98 24. Lonergan E, Luxenberg J, Areosa SA: Benzodiazepines for delirium. Cochrane Database Syst Rev 2009;CD006379 25. Lonergan E, Britton AM, Luxenberg J, et al: Antipsychotics for delirium. Cochrane Database Syst Rev 2007;CD005594 26. Overshott R, Karim S, Burns A: Cholinesterase inhibitors for delirium. Cochrane Database Syst Rev 2008;CD005317 27. Siddiqi N, Stockdale R, Britton AM, et al: Interventions for preventing delirium in hospitalised patients. Cochrane Database Syst Rev 2007;CD005563 28. Neto AS, Nassar AP Jr, Cardoso SO, et al: Delirium screening in critically ill patients: A systematic review and meta-analysis. Crit Care Med 2012; 40:1946–1951 29. Higgins JP, Altman DG, Gøtzsche PC, et al; Cochrane Bias Methods Group; Cochrane Statistical Methods Group: The Cochrane Collaboration’s tool for assessing risk of bias in randomised trials. BMJ 2011; 343:d5928 30. Guyatt G, Oxman AD, Akl EA, et al: GRADE guidelines: 1. IntroductionGRADE evidence profiles and summary of findings tables. J Clin Epidemiol 2011; 64:383–394 31. Balshem H, Helfand M, Schünemann HJ, et al: GRADE guidelines: 3. Rating the quality of evidence. J Clin Epidemiol 2011; 64: 401–406 32. Walter SD, Yao X: Effect sizes can be calculated for studies reporting ranges for outcome variables in systematic reviews. J Clin Epidemiol 2007; 60:849–852 33. Higgins JPT, Green S: Cochrane Handbook for Systematic Reviews of Interventions Version 5.1.0 [updated March 2011]. The Cochrane Collaboration, 2011. Available at: http://www.cochrane-handbook. org. Accessed November 12, 2012 34. Higgins JP, Thompson SG, Deeks JJ, et al: Measuring inconsistency in meta-analyses. BMJ 2003; 327:557–560 35. Peters JL, Sutton AJ, Jones DR, et al: Contour-enhanced meta-analysis funnel plots help distinguish publication bias from other causes of asymmetry. J Clin Epidemiol 2008; 61:991–996 36. Chaimani A, Higgins JP, Mavridis D, et al: Graphical tools for network meta-analysis in STATA. PLoS One 2013; 8:e76654 www.ccmjournal.org
203
Mu et al 37. Grover S, Kate N: Assessment scales for delirium: A review. World J Psychiatry 2012; 2:58–70 38. van den Boogaard M, Pickkers P, Slooter AJ, et al: Development and validation of PRE-DELIRIC (PREdiction of DELIRium in ICu patients) delirium prediction model for intensive care patients: Observational multicentre study. BMJ 2012; 344:e420 39. Sun X, Ioannidis JP, Agoritsas T, et al: How to use a subgroup analysis: Users’ guide to the medical literature. JAMA 2014; 311:405–411 40. Tse L, Schwarz SK, Bowering JB, et al: Pharmacological risk factors for delirium after cardiac surgery: A review. Curr Neuropharmacol 2012; 10:181–196
204
www.ccmjournal.org
41. Burkhart CS, Dell-Kuster S, Gamberini M, et al: Modifiable and nonmodifiable risk factors for postoperative delirium after cardiac surgery with cardiopulmonary bypass. J Cardiothorac Vasc Anesth 2010; 24:555–559 42. Ji F, Li Z, Nguyen H, et al: Perioperative dexmedetomidine improves outcomes of cardiac surgery. Circulation 2013; 127:1576–1584 43. Lin Y, Chen J, Wang Z: Meta-analysis of factors which influence delirium following cardiac surgery. J Card Surg 2012; 27:481–492 44. Luetz A, Heymann A, Radtke FM, et al: Different assessment tools for intensive care unit delirium: Which score to use? Crit Care Med 2010; 38:409–418
January 2015 • Volume 43 • Number 1
