Dexmedetomidine as a Sedative Agent in Critically Ill Patients: A Meta-Analysis of Randomized Controlled Trials

Introduction The effect of dexmedetomidine on length of intensive care unit (ICU) stay and time to extubation is still unclear. Materials and Methods Pertinent studies were independently searched in BioMedCentral, PubMed, Embase, and the Cochrane Central Register of clinical trials (updated February first 2013). Randomized studies (dexmedetomidine versus any comparator) were included if including patients mechanically ventilated in an intensive care unit (ICU). Co-primary endpoints were the length of ICU stay (days) and time to extubation (hours). Secondary endpoint was mortality rate at the longest follow-up available. Results The 27 included manuscripts (28 trials) randomized 3,648 patients (1,870 to dexmedetomidine and 1,778 to control). Overall analysis showed that the use of dexmedetomidine was associated with a significant reduction in length of ICU stay (weighted mean difference (WMD) = −0.79 [−1.17 to −0.40] days, p for effect <0.001) and of time to extubation (WMD = −2.74 [−3.80 to −1.65] hours, p for effect <0.001). Mortality was not different between dexmedetomidine and controls (risk ratio = 1.00 [0.84 to 1.21], p for effect = 0.9). High heterogeneity between included studies was found. Conclusions This meta-analysis of randomized controlled studies suggests that dexmedetomidine could help to reduce ICU stay and time to extubation, in critically ill patients even if high heterogeneity between studies might confound the interpretation of these results.


Introduction
Dexmedetomidine was approved by the Food and Drug Administration (FDA) at the end of 1999 as a short-term medication (,24 hours) for analgesia and sedation in mechanical ventilated intensive care unit (ICU) patients. In 2008, the FDA approved a new indication in non intubated patients requiring sedation before and/or during surgical and non-surgical procedures. Dexmedetomidine is a highly selective a2-adrenergic receptor agonist, which binds to transmembrane G proteinbinding adrenoreceptors in the periphery (a2A), brain and spinal cord (a2B, a2C) tissues [1]. In contrast to other sedative agents, dexmedetomidine, by acting on a2 receptors in the locus caeruleus [2], has potential analgesic effects [3] without respiratory depression [4,5]. Only one meta-analysis of randomized controlled trials (RCTs) [6] was published so far: Tan and Ho reported a reduction in length of ICU stay, but not in duration of time to extubation when dexmedetomidine was compared with alternative sedative agents.
Since several RCTs [7][8][9][10][11][12][13][14], including two large ones [8], were recently published, and one further RCT [15] was not included in the previous meta-analysis [6] we decided to perform an updated meta-analysis of all the RCTs ever performed on dexmedetomidine versus any comparator in the ICU setting to evaluate time to extubation, ICU stay and survival.

Search Strategy
Pertinent studies were independently searched in BioMedCentral, PubMed, Embase, and the Cochrane Central Register of clinical trials (updated February 1 st 2013) by four trained investigators. The full PubMed search strategy aimed to include any RCTs ever performed in humans with dexmedetomidine in Table 1. Description of the 28 trials included in the meta-analysis. any clinical setting and is presented in the supplemental material (Text S1). In addition, we employed backward snowballing (i.e., scanning of references of retrieved articles and pertinent reviews) and contacted international experts for further studies with no language restriction.

Study Selection
References were first independently examined at a title/abstract level by four investigators, with divergences resolved by consensus, and then, if potentially pertinent, retrieved as complete articles. The following inclusion criteria were used for potentially relevant studies: random allocation to treatment (dexmedetomidine versus any comparator with no restrictions on dose or time of administration); studies involving patients who required mechanical ventilation in an ICU. The exclusion criteria were duplicate publications (in this case we referred to the first article published while retrieved data from the article with the longest follow-up available), non-adult patients and lack of data on all of the following: ICU stay, time to extubation and mortality. Two investigators independently assessed compliance to selection criteria and selected studies for the final analysis, with divergences resolved by consensus.

Data Abstraction and Study
Baseline, procedural, and outcome data were independently abstracted by four trained investigators (table 1 and table 2). If a trial reported multiple comparisons [25,34], the comparators were aggregated as a single control group. At least two separate attempts at contacting original authors were made in cases of missing data. The co-primary endpoints of the present review were the length of ICU stay (days) and time to extubation (hours from randomization to extubation).
The secondary endpoint was mortality rate at the longest followup available. Adverse effects (hypotension and bradycardia as per author definition) were also analysed. Further endpoints included the number of patients requiring rescue doses of analgesic (opioids) or sedative (propofol, benzodiazepines, or any antipsychotics) drugs and the number of patients completely comfortable during ICU stay.
The internal validity and risk of bias of included trials was appraised by two independent reviewers according to the latest version of the ''Risk of bias assessment tool'' developed by The Cochrane collaboration [16], with divergences resolved by consensus. Publication bias was assessed by visually inspecting funnel plots and scatter plots and by analytical appraisal based on the Egger's linear regression test and on the Peters' test for asymmetry. According to the Egger [17] or Peters [18] methods for publication bias evaluation, a two-sided p value of 0.10 or less was regarded as significant.

Data Analysis and Synthesis
Computations were performed with Stata release 11, College Station, TX) and SAS 2002-08 program (release 9.2, SAS Institute, Inc, Cary, NC). Hypothesis of statistical heterogeneity was tested by means of Cochran Q test, with statistical significance set at the two-tailed 0.10 level, whereas extent of statistical consistency was measured with I 2 , defined as 100%6(Q-df )/Q , where Q is Cochran's heterogeneity statistic and df the degrees of freedom. Binary outcomes from individual studies were analysed to compute individual and pooled risk ratio (RR) with pertinent 95% confidence interval (CI), by means of inverse variance method and with a fixed-effect model in case of low statistical inconsistency (I 2 ,25%) or with random-effect model (which better accommodates clinical and statistical variations) in case of  Collaboration and Preferred Reporting Items for Systematic Reviews and Meta-Analyses guidelines [16,19,20] (Checklist S1).

Quantitative Data Synthesis
Further subanalyses with the different comparators (propofol, midazolam, placebo and morphine) are detailed in supplemental material ( TableS2 Table S3, Table S4 and Table S5) but were not informative with respect to ICU stay or time to extubation due to the paucity of trials included.
Visual inspection of funnel and scatter plots (figures 4 and 5; figures S3 and S4) did not identify a skewed or asymmetrical shape for the co-primary endpoints. Quantitative evaluation did not suggest a presence of publication bias, as measured by the Egger's test (p = 0.4 for the length of ICU stay and p = 0.5 for time to extubation) and Peters' test (p = 0.6 for the length of ICU stay and p = 0.9 for time to extubation). Since the funnel plots identified three outlier studies [14,32,33] we repeated the analyses removing   them and found that the statistically significant difference in ICU stay and time to extubation was maintained (p = 0.005 and p = 0.003 respectively ) and the heterogeneity reduced (I 2 = 28% and I 2 = 39% respectively ) (

Sensitivity analyses
Estimate results from both random and fixed effect models were extremely similar (table 3); hence we excluded a considerable small study effect. Sensitivity analyses performed by sequentially removing each study and reanalysing the remaining dataset (producing a new analysis for each study removed), did not determine major changes in direction or magnitude of statistical findings, confirming the pooled effect of each co-primary endpoints (all SWD,1) and the statistical significance (all p of effect

Discussion
Our meta-analysis confirmed that dexmedetomidine is associated with a reduction in ICU stay and suggested that it might reduce the time of extubation when compared to other sedative or hypnotic agent. Even if dexmedetomidine is associated with an increase in the risk of bradycardia and with a trend toward an increased risk of hypotension, no detrimental effects on mortality were detected.
The ideal sedative agent should provide anticipated, predictable effects, rapid onset, and quick recovery. It should be easy to administer with no adverse events, no interaction with other drugs, no accumulation of metabolites and no withdrawal effects at the end of infusion. Unluckily an ideal sedative agent that can suit the need of all patients does not yet exist.
Dexmedetomidine is one of the most recently released intravenous agents for sedation in the ICU, though the drug started to be investigated more than 20 years ago. It was introduced in clinical practice in the United States in 1999 while the European Medicine Agency authorised its use for all 27 European member states in September 2011. It is an alpha2agonist and produces sedation acknowledged as ''cooperative'' or ''arousable'', which is different from the sedation ''clouding of consciousness'' induced by drugs acting on GABA receptors, such as midazolam or propofol [39]. Tan and Ho, in a previous metaanalysis updated on December 2009 [6] reported that when dexmedetomidine was compared with alternative sedative agents it was associated with a statistically significant reduction in length of ICU stay, but not in duration of mechanical ventilation. We updated their findings on February 2013 identifying eight recently published manuscripts [7][8][9][10][11][12][13][14] and one trial that was not identified in their systematic search [15], thus increasing the number of patients by 50% (up to 3,648 overall randomized patients included in our meta-analysis) and providing more robust safety data. By adding more patients data we were able to show, for the first time in a meta-analysis, that dexmedetomidine increases the rate of bradycardia when all trials are pooled together and also shows a trend towards an increase rate of hypotension. However, these side effects were not associated with differences in mortality (200/1499 [13%] in the dexmedetomidine group vs 173/1409 [12%] in the control group, p = 0.9 with 20 studies included).
Dexmedetomidine decreases sympathetic nervous system activity and is therefore associated with an increase in cardiovascular adverse events. These effects may be most pronounced in patients with decreased autonomic nervous system response such as the elderly, diabetic patients, patients with chronic hypertension or severe cardiac disease such as valve stenosis or regurgitation, advanced heart block, severe coronary artery disease, or in patients who are already hypotensive and/or hypovolemic [40]. Therefore, in patients who depend on a high level of sympathetic tone or in patients with reduced myocardial function who cannot tolerate the decrease in sympathetic tone, loading doses of  Visual inspection of funnel plots did not identify a skewed or asymmetrical shape for the coprimary endpoints. Quantitative evaluation did not suggest a presence of publication bias, as measured by the Egger's test (p = 0.5) and Peters' test (p = 0.9). SE = standard error; SMD = standardized mean difference. doi:10.1371/journal.pone.0082913.g005 dexmedetomidine should be avoided. On the other side, the characteristics of dexmedetomidine to provide an ongoing sedation and sympathetic block could be beneficial in reducing early postoperative ischemic events in high-risk patients [41][42].
Intravenous administration of dexmedetomidine exhibits the following pharmacokinetic parameters: a rapid distribution phase with an half-life (t 1/2 a) of 6 min, a terminal elimination half-life (t 1/2 b) of 2 hours, and a steady-state volume of distribution (Vss) of 118 litres. It presents linear kinetics when infused in the range of 0.2-0.7 mg/kg/h for no more than 24 hours and undergoes almost complete biotransformation through direct glucuronidation and cytochrome P450 metabolism. Consequently it can accumulate in patients who are on P450 enzyme inhibitors, some of which are commonly used in ICU. Metabolites of biotransformation are excreted in the urine (95%) and faeces [43].

Limitations
We acknowledge that this study has several limitations. The quality of the included studies is not high since only 13 of them were blind. Moreover we noted high heterogeneity between the included studies. The heterogeneity remained when sensitivity analyses on studies with low risk of bias where performed. It was abolished only removing three outliers studies cited above. Nonetheless we excluded the possible influence of small-study effects on the results of our meta-analysis comparing the fixed-and random-effects estimates of the treatment effect (table 3). The overall reduction in ICU stay and time to extubation may appear clinically modest, but it should be acknowledged that the largest study [8] had very conservative imputation rules (to worst outcome) and this might have softened our results.

Conclusions
Dexmedetomidine for sedation in mechanically ventilated critically ill adult patients seems to help to reduce time to extubation and ICU stay. The known side effects (increased incidence of bradycardia and a trend toward an increased risk of hypotension) had no effect on the overall mortality in this metaanalysis of all the RCTs published so far.
Larger, multicentre, randomized clinical trials, especially in long term sedated patients requiring mechanical ventilation, would be welcome to confirm these findings.

Supporting Information
Checklist S1 PRISMA checklist. (DOC) Figure S1 Forest plot for the length of ICU stay using standard mean difference (days) instead of weighted mean difference (absolute value with no units of measurement). Overall analysis showed that the use of dexmedetomidine was associated with a significant reduction in length of ICU stay (SMD = 20.48 [20.78 to 20.18], p for effect = 0.002, p for heterogeneity ,0.001, I2 = 91% with 17 studies and 2,424 patients included). ICU = intensive care unit; CI = confidence interval; SMD = standardized mean difference; N = number; SD = standard deviation. (TIF) Figure S2 Forest plot for the time to extubation using standard mean difference (days) instead of weighted mean difference (absolute value with no units of measurement).. Overall analysis showed that the use of dexmedetomidine was associated with a significant reduction of time to extubation (SMD = 20.39 [20.