Article Text

Original research
Inconsistent relationship between depth of sedation and intensive care outcome: systematic review and meta-analysis
  1. Leanne M Aitken1,
  2. Kalliopi Kydonaki2,
  3. Bronagh Blackwood3,
  4. Laurence G Trahair4,
  5. Edward Purssell1,
  6. Mandeep Sekhon5,
  7. Timothy S Walsh6
  1. 1 School of Health Sciences, City, University of London, London, UK
  2. 2 School of Health and Social Care, Edinburgh Napier University, Edinburgh, UK
  3. 3 Wellcome-Wolfson Institute for Experimental Medicine, Queen's University Belfast Faculty of Medicine Health and Life Sciences, Belfast, UK
  4. 4 Adelaide Medical School, The University of Adelaide, Adelaide, South Australia, Australia
  5. 5 School of Population Health & Environmental Sciences, Faculty of Life Sciences and Medicine, King's College London, London, UK
  6. 6 Department of Anaesthesia, Critical Care & Pain Medicine and Usher Institute, The University of Edinburgh, Edinburgh, UK
  1. Correspondence to Dr Leanne M Aitken, School of Health Sciences, City University of London, London EC1V 0HB, UK; leanne.aitken.1{at}


Purpose To determine the effect of depth of sedation on intensive care mortality, duration of mechanical ventilation, and other clinically important outcomes.

Methods We searched MEDLINE, Embase, Cochrane Register of Controlled Trials, Cumulative Index to Nursing and Allied Health Literature, PsycINFO from 2000 to 2020. Randomised controlled trials (RCTs) and cohort studies that examined the effect of sedation depth were included. Two reviewers independently screened, selected articles, extracted data and appraised quality. Data on study design, population, setting, patient characteristics, study interventions, depth of sedation and relevant outcomes were extracted. Quality was assessed using Critical Appraisal Skills Programme tools.

Results We included data from 26 studies (n=7865 patients): 8 RCTs and 18 cohort studies. Heterogeneity of studies was substantial. There was no significant effect of lighter sedation on intensive care mortality. Lighter sedation did not affect duration of mechanical ventilation in RCTs (mean difference (MD): −1.44 days (95% CI −3.79 to 0.91)) but did in cohort studies (MD: −1.52 days (95% CI −2.71 to −0.34)). No statistically significant benefit of lighter sedation was identified in RCTs. In cohort studies, lighter sedation improved time to extubation, intensive care and hospital length of stay and ventilator-associated pneumonia. We found no significant effects for hospital mortality, delirium or adverse events.

Conclusion Evidence of benefit from lighter sedation is limited, with inconsistency between observational and randomised studies. Positive effects were mainly limited to low quality evidence from observational studies, which could be attributable to bias and confounding factors.

  • critical care

Data availability statement

No data are available.

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

Key messages

What is the key question?

  • Does depth of sedation effect intensive care mortality and duration of mechanical ventilation, as well as secondary physiological, hospital mortality, resource use, adverse event and life impact outcomes?

What is the bottom line?

  • Evidence of the effect of sedation depth is limited, with inconsistency between observational and randomised studies.

  • Positive effects from lighter sedation were mainly limited to low to very low quality evidence from observational studies.

Why read on?

  • Depth of sedation appears to have differential effect on various outcomes.

  • We need to build on the current evidence to determine how to optimise patient outcomes, both within and beyond intensive care.


Mechanically ventilated patients in intensive care receive sedation and analgesia to manage their discomfort. Although these medications are considered important for many patients, there is recognition that both the amount and type of sedation that patients receive are potentially related to patient outcomes.1 Various proposals and guidelines recommend alternative ways of administering sedation or using different sedative agents to improve outcomes from critical illness.1–3 Although interpretation of this literature is challenging due to inconsistent and problematic definitions, evidence suggests lighter sedation is probably beneficial.1 Despite this, recent reports show many intensive care unit (ICU) patients worldwide continue to be deeply sedated.4–6

In a recent review of outcomes associated with sedation depth in the first 48 hours of mechanical ventilation across the emergency department (ED) and ICU lighter sedation was associated with reduced mortality, mechanical ventilation and ICU stay days.7 Given many critically ill patients remain heavily sedated for longer than 48 hours, it would be useful to know if this relationship between sedation depth and patient outcomes extends across patients’ entire ICU stay and relates to a range of patient outcomes or only the short-term outcomes of mortality and duration of mechanical ventilation and ICU stay. The effect of lighter sedation on selected outcomes was also examined in the pain, agitation/sedation, delirium, imobility and sleep (PADIS) guidelines, however the included meta-analysis incorporated only studies where sedation depth was defined a priori,1 with inconsistent evidence identified. These reviews provide some insights into the evidence to guide sedation practice, but both reviews focused on specific subgroups of studies. We therefore considered a review of a wider range of relevant studies appropriate and important.


To systematically examine the effect of depth of sedation in ICU patients on patient outcomes that extend across the ICU stay and beyond. ICU mortality and duration of mechanical ventilation were co-primary outcomes selected because ICU mortality is patient-focused and duration of mechanical ventilation reflects sedation practice. Secondary outcomes from the five domains of the outcome taxonomy proposed by Dodd et al 8 were selected and included hospital mortality, physiological outcomes (time to extubation, ventilator-free days (Vfd) to day 28), resource use (ICU and hospital length of stay (LOS)), adverse events (incidence of delirium, self-extubation, reintubation and tracheostomy, ventilator-associated pneumonia (VAP)) and life impact outcomes (memories, anxiety, depression and symptoms or diagnosis of post-traumatic stress disorder); these latter outcomes mirror those identified as important to patients and family members in a research priority setting exercise.9


The protocol for this systematic review was registered on PROSPERO (CRD42018092554; Additional detail is available in online supplemental materials.

Supplemental material

Search strategy

MEDLINE, Embase, Cochrane Register of Controlled Trials (CENTRAL), Cumulative Index to Nursing and Allied Health Literature and PsycINFO were searched with the following strategy: (intensive care OR critical care OR critically ill) AND (sedat* OR midazolam OR propofol) AND (length of stay OR mortality OR outcome assessment OR physical function OR psychological OR cognitive OR memories).

We searched for publications reporting randomised controlled, quasi-experimental and before-after trials, and cohort studies (prospective and retrospective) published in English between January 2000 and February 2020.

Types of participants

We included studies in adult patients receiving invasive mechanical ventilation in ICU, including patients who commenced their ventilation in another location, for example, ED, operating room. We excluded studies: (i) in patients receiving non-invasive ventilation and mechanically ventilated patients not admitted to ICU; (ii) where the intervention included different sedative agents. Studies testing the effect of different sedative agents were excluded because it is not possible to determine if any difference in outcome was due to effect of the different agent or different depth of sedation. We defined our exposure as deeper sedation at any time throughout the period of mechanical ventilation in the ICU. Our classification of depth of sedation as either ‘lighter’ or ‘deeper’ did not need to be (but could be) predefined by study authors, but was based on published information incorporating any objective measures of sedation depth including assessment using a validated sedation assessment instrument, hourly or daily doses of sedatives. To clarify, studies that tested any intervention (eg, goal or protocol directed sedation, no sedation), other than different sedative agents, were eligible for inclusion if one group of patients received lighter sedation than another group of patients in the study. Studies were not excluded on the basis of which sedative agent they used, and no attempt was made to control analgesic use, although it is recognised that many have a secondary sedative effect. Only the Richmond Agitation Sedation Scale (RASS) and the Riker Sedation-Agitation Scale (SAS) were accepted as validated instruments.10

Study selection

Titles and abstracts were screened independently by two researchers, with full text of included studies reviewed by two authors to assess eligibility. Studies where separation of depth into ‘lighter’ and ‘deeper’ sedation could not be identified were excluded. Studies including >2 groups based on sedation depth were not included in the meta-analysis but were retained in the additional analyses. Sedation was defined as the use of pharmacological agents that have the primary purpose of calming or inducing sleep, and alternative agents such as analgesics were not included despite acknowledging that secondary effects of sedation are often present. We did not include different outcomes from the same patient cohort, reported in multiple papers, twice in any analysis but this relationship was noted.

Data extraction

Two authors extracted data on study design, population and setting, patient characteristics, study interventions, measure of depth of sedation (methodology and results) and relevant outcomes.

Assessment of bias

The domains of bias for RCTs and cohort studies were assessed consistently with current guidance.11 12 Relevant confounding factors were not identified a priori, but were based on the study method and cohort and included demographic, clinical and treatment variables with the potential to influence relevant outcomes. No studies were excluded on the basis of quality assessment.

Data analysis

Two authors extracted data on study design, population and setting, patient characteristics, study interventions, measure of depth of sedation (methodology and results) and relevant outcomes. All studies that contained data suitable for inclusion in at least one meta-analysis were included in the quantitative analysis. Continuous data were analysed as means and SD. Where the median and IQR was reported, these were converted to mean and SD using a standard method.13 Dichotomous data were analysed as risks and relative risks. Random effects meta-analyses were undertaken with the meta package14 in R.15 This allowed for both within and between studies variance to be calculated, the latter being reflected in a statistical test of heterogeneity. Cohort studies and RCTs were analysed separately based on an a priori decision. The quality of evidence was rated using Grades of Recommendation, Assessment, Development and Evaluation (GRADE).16 For outcomes where significant methodological differences occurred (eg, different instruments or time points), results were combined descriptively.

Sensitivity analysis

Categorisation of patients into ‘lighter’ and ‘deeper’ sedation groups could be based on either a difference in RASS or SAS scores, a difference in average dose of sedation over time (hourly/daily) or a combination of both. Due to the potential differential effect of sedation amounts on patients’ sedation levels, a post hoc decision was made to repeat meta-analyses incorporating only those studies where categorisation was based on RASS or SAS scores alone or in combination with sedation dose, that is, to exclude studies where categorisation was based solely on sedation dose. Similarly, a post hoc analysis of cohort studies to examine the influence of the temporal nature of the design (ie, prospective or retrospective) was conducted.


After removal of duplicates, 3390 articles were identified (figure 1), with full text of 116 articles assessed. Ninety were excluded: 69 met exclusion criteria; and 21 because, although patients were in groups, levels of sedation did not differ between the groups.

Figure 1

Preferred Reporting Items for Systematic Reviews and Meta-Analyses flow diagram.

Twenty-six articles reporting the results of 23 studies incorporating 8575 patients remained for descriptive synthesis with 17 articles (7027 patients) included in a meta-analysis for at least one outcome.17–33 The included papers reported results prospective (n=16; n=5534) and retrospective (n=2; n=2028) cohort studies and randomised controlled trials (n=8; n=1534) published between 2001 and 2020 conducted across Asia (n=2), Australia and New Zealand (n=3), Europe (n=10), MiddleEast (n=1) and North (n=7) and South America (n=4) (online supplemental table S1). Depth of sedation was measured either using sedation assessment instruments or average doses of sedatives or a combination of both (table 1, online supplemental table S2). The level of sedation that constituted ‘lighter’ or ‘deeper’ sedation was inconsistent across studies.

Table 1

Criteria used in studies to separate ‘deeper’ versus ‘lighter’ sedation*

Risk of bias was highly variable in the cohort studies. In the RCTs risk of bias was more consistent, with lack of blinding being the main source of bias. Blinding of participants and personnel was not possible and blinding of outcome assessors was rare (online supplemental figure S1) (online supplemental table S3). There was infrequent incorporation of relevant confounding factors into analysis in cohort studies (online supplemental figure S2) (online supplemental table S4).

Supplemental material

Included studies addressed both our primary outcomes, and secondary outcomes within the five domains of mortality, physiological outcomes, resource use, adverse events and life impact outcomes8 (online supplemental table S5), with most outcomes assessed in meta-analyses (table 2). Outcomes within the life impact domain could not be pooled, but a descriptive synthesis of results related to memory and psychological function is provided (table 3). Studies not included in the meta-analyses are synthesised under additional analyses.

Table 2

Summary of findings

Table 3

Life impact outcomes

Primary outcomes

ICU mortality

When comparing lighter versus deeper sedation, we found no difference in ICU mortality in either RCTs or cohort studies (figure 2, table 2).

Figure 2

Forest plots for primary outcome: (A) intensive care unit (ICU) mortality; (B) duration of mechanical ventilation. RCT, randomised controlled trial.

Duration of mechanical ventilation

We found no difference in duration of mechanical ventilation in the RCTs comparing lighter versus deeper sedation, but identified reduced duration of mechanical ventilation with lighter sedation in cohort studies (MD −1.52 days (95% CI −2.71 to −0.34), I2=87%, 8 studies, 3304 participants) (figure 2, table 2).

Secondary outcomes

Hospital mortality

Pooled data from five RCTs and five cohort studies showed no difference between lighter and deeper sedation on hospital mortality (table 2, online supplemental figure S3).

Physiological outcomes

Pooled data from four RCTs and two cohort studies showed no difference between lighter and deeper sedation on 28-day Vfd (table 2, online supplemental figure S4). There was no difference in time to extubation in a single RCT, but cohort studies reported reduced duration with lighter sedation (MD −3.77 days (95% CI −5.49 to −2.06), I2=98%, 2 studies, 2132 participants). Pooled data from four RCTs and four cohort studies showed no difference between lighter and deeper sedation on incidence of delirium (table 2, online supplemental figure S4).

Resource use

Pooled data from six RCTs showed no difference between lighter and deeper sedation on ICU LOS or hospital LOS, but a significant reduction favouring lighter sedation was identified in ICU and hospital LOS in cohort studies (eight and six studies, respectively; table 2, online supplemental figure S5). Lighter sedation had no effect on frequency of tracheostomy (four RCTs, two cohort studies; table 2).

Adverse events

We found no difference between lighter and deeper sedation on self-extubation (two RCTs, three cohort studies) or reintubation (five RCTs, two cohort studies) (table 2, online supplemental figure S6). Lighter sedation had no effect on risk of VAP in one RCT, although data from two cohort studies showed a reduced risk with lighter sedation (RR 0.56 (95% CI 0.33 to 0.96), I2=51%, 1906 participants) (table 2, online supplemental figure S6).

Sensitivity analyses

Meta-analyses, incorporating only those studies where RASS or SAS data were available to categorise patients as lighter or deeper sedation, were repeated on outcomes where studies existed. Results were largely similar, although fewer significant differences were identified (online supplemental table S6).

Meta-analyses examining the influence of the temporal nature of the design in cohort studies, that is, prospective or retrospective, were conducted. Results were largely similar to the overall results, although analysis of only the prospective studies substantially reduced the heterogeneity when examining ICU and hospital mortality and hospital LOS but had no effect on heterogeneity in relation to other outcomes (online supplemental table S7).

Additional analyses

Nine studies met the inclusion criteria, but were excluded from all meta-analyses for reasons detailed in the ‘Methods’ section.34–42 The main reasons were single group cohort studies with multivariable regression analysis35 39 or more than two groups of patients not able to be combined based on sedation depth,34 40 as well as variable time points and methods for outcome measurement. In addition, some studies (where the primary outcome has been incorporated in meta-analyses above) incorporated life impact outcomes as secondary measures, however differences in methods of outcome assessment precluded a meta-analysis of life impact outcomes. A descriptive synthesis is provided here.

Mortality, physiological outcomes and adverse events

A positive relationship between deep sedation and increased mortality35 39 and increased duration of mechanical ventilation39 40 was reported in cohort studies, but depth of sedation was not associated with mechanical ventilation duration across different stages of implementation of a sedation protocol and education intervention.34 A relationship between deeper sedation and both delirium39 and VAP34 was identified.

Life impact

Outcomes reflecting the impact of sedation depth on a person’s life focused only on memories and psychological health measured in 10 studies using a variety of instruments at different times (table 3). There was some evidence of a relationship between sedation depth and presence or type of memories that patients reported. In a cohort study of 128 Brazilian patients, those who received any sedation reported less real memories (21 (24%) vs 29 (69%)), more illusionary memories (7 (8%) vs 0) and more amnesia (16 (19%) vs 4 (10%)) than patients who received no sedation.40 In a cohort study of 313 Swedish patients, increased time deeply sedated was associated with having no recall of ICU (OR 1.60, 95% CI 1.35 to 1.91).37 In further analysis of the same cohort, patients who spent more time awake were more likely to remember the endotracheal tube (OR 1.45, 95% CI 1.29 to 1.62) and be bothered by memories of stressful ICU experiences (OR 1.37, 95% CI 1.13 to 1.67), but sedation depth was not associated with nightmares during recovery.38 In contrast, in 289 patients in Canada and the USA, patients with no recall of ICU received lower daily doses of midazolam (26.9 (SD 63.7) vs 82.5 (SD 314) mg), but delusional memories were not associated with higher sedative doses (OR 1.18, 95% CI 0.37 to 3.81).41 No difference in frequency or type of memories was reported in two studies27 36 or in studies exploring the relationship between psychological distress and sedation depth.18 24 30 40 42


In this systematic review of data from 26 studies incorporating just under 8000 adult patients there was inconsistent and inadequate evidence of the relationship between sedation depth and patient outcomes. Moderate-level evidence from RCTs was identified in relation to the primary outcomes of ICU mortality and duration of mechanical ventilation, as well as secondary outcomes including hospital mortality, time to extubation, Vfd, ICU LOS, incidence of delirium and tracheostomies, however no benefit of lighter sedation was identified in any of these outcomes. Outcomes where benefit of lighter sedation was shown in cohort studies included duration of mechanical ventilation, time to extubation, ICU and hospital LOS and VAP; the evidence was assessed as very low level for all these outcomes. Reasons for low levels of evidence were multifactorial but included inconsistency and imprecision, frequently with very high levels of heterogeneity, likely occurring as a result of differences in the primary aim and design of included studies as well as variation in interventions used to achieve lighter sedation. The multidimensional nature of factors that influence each of the outcomes also likely influences the inconsistency in results. High levels of heterogeneity potentially occurred as a result of the different designs (RCTs as well as prospective and retrospective cohort studies), the intent of the project (eg, primarily as a quality improvement project) and the level of sedation and intervention fidelity achieved. The heterogeneity shown in this review highlights the issue of sedation being a complex healthcare invention influenced by multiple factors including agent chose, patient characteristics, protocols and practices, contextual issues within ICUs and individual clinician values and beliefs. These issues increase the relevance of the possible uncertainty highlighted in our review.

There was little evidence of effect of sedation depth on life impact outcomes. There was no evidence that anxiety, depression or symptoms of post-traumatic stress were related to sedation depth.18 24 30 40 42 There was, however, inconsistent evidence of whether, and how, sedation depth might influence the presence and type of memories.18 27 36–38 40 41 The role of memories after critical illness, and the relationship with psychological health, is inconsistent, with some suggestion that intrusive, persecutory or delusional memories may be more harmful than real memories,43 with the possibility that more frightening memories might be associated with greater psychological trauma.44 No evidence of a relationship between sedation depth and delirium was identified in this review, however any potential relationship between sedation, delirium and memories requires further investigation.43

Few of the included studies identified an a priori aim related to sedation depth. Instead, many studies examined the effect of interventions to improve sedation practice, or explored the relationship between sedation and outcomes. Labelling of groups as ‘deeper’ and ‘lighter’ sedation in this review may not be appropriate given that ‘deeper’ sedation in one study could be similar to ‘lighter’ sedation in another study or setting. For example, RASS score of −3 indicated moderate sedation in one study40 and deep sedation in others,17 20 while one prepost study achieved ‘lighter’ sedation with a median first RASS score of −4 postintervention.17 No studies targeted RASS 0 to −1 (alert and calm to drowsy), with the exception of work from Scandinavia examining ‘no sedation’.25 29 42 The diversity of clinical practice strategies to achieve lighter sedation also presented challenges. We aimed to summarise whether strategies, whatever their design or content, that targeted deeper sedation avoidance were effective in changing outcomes relative to the comparator.

Recently, a Peruvian multicentre observational cohort study examining the relationship between benzodiazepine dose and mortality was published.45 In this study, benzodiazepine dose was associated with a higher risk of mortality and a significant decrease in Vfd, although it should be noted that 98% of participants were deeply sedated at some point during the study and depth of sedation was assessed using either the Glasgow Coma Scale, Ramsay Sedation Scale or RASS. The primary results of the Sedation Practice in Intensive Care Evaluation (SPICE-III) study comparing dexmedetomidine with usual sedation are also published.6 SPICE-III compared different sedatives and was therefore ineligible for this review. However, it is worth noting that although the dexmedetomidine group had a slightly higher proportion of patients with lighter RASS scores (56.6% vs 51.8%), no difference in outcomes was observed. In two French studies also not meeting our inclusion criteria, one multicentre study found no difference in Vfd or mortality with the introduction of an oversedation prevention strategy,46 while a single-centre study found reduced duration of mechanical ventilation by stopping sedation immediately after ICU admission.47 The most recent relevant study published was the Danish NONSEDA study where a strategy of no sedation was compared with light sedation.25 In this high-quality RCT with clear separation in sedation levels, a non-significant trend towards higher mortality in the non-sedated group was identified, emphasising the need for a strong body of evidence to illuminate the effect of sedation depth on a range of patient outcomes.

The reasons for reporting the effects of sedation depth on clinical outcomes from cohort studies alongside those from RCTs deserves attention. Changing sedation practice frequently requires an integrated or bundled approach to sedation assessment and management to achieve cultural change of clinician behaviour.2 48 Cohort (before and after) studies are more amenable to achieve practice change than randomised studies. Once a shift in clinicians’ sedation management behaviour has been learnt, it can be difficult to apply earlier (usual care) practices when patients are randomised. The RCTs in this review all randomised at the patient level. So, although cohort studies provide lower quality evidence than RCTs, in the area of sedation practice they have provided a pragmatic method for studies designed to modify sedation depth. To improve the quality of evidence, we recommend cluster randomised trials to address the weakness of intervention contamination in patient level randomisation and improve the quality of evidence. We have also provided ratings of evidence using the GRADE criteria,16 although we note the limitations of this system in that it is based on subjective judgements and does not take into account the benefits of various study methodologies as outlined above.

There have been multiple calls in clinical guidelines and opinion papers for lighter sedation in ICU patients1 2; these calls have been based on subsets of the available evidence7 or individual studies.28 49 In response to these calls, multiple strategies have been proposed to achieve lighter sedation including protocols,50 expert staffing patterns51 and daily interruption of sedation.52 To date, systematic reviews have not identified consistently useful strategies,53 54 although reviews are ongoing.55

This review represents the most comprehensive description of the current evidence related to sedation depth and patient outcomes. Despite the use of liberal inclusion criteria, and a wide range of outcomes examined, the certainty of evidence remains low and inconsistent. Additionally, the findings are limited by the variable nature of how ‘lighter’ and ‘deeper’ sedation were determined in the studies, the lack of control of analgesic agents and the frequent lack of determining this differentiation a priori or indeed stating it as an aim. In some studies, the only measure of sedation depth was average dose of sedation, which may not reflect sedative effect on the individual patient. Ideally validated sedation scores such as RASS or SAS should be used to indicate the actual depth to which a patient is sedated. Yet, despite a sensitivity analysis of studies where the difference in sedation depth was based on RASS or SAS, the lack of consistency in effect on patient outcomes remained. The review only included studies that used sedation assessment scales validated for use in the ICU environment in international practice guidelines,10 and thus may have had the effect of biassing the meta-analysis. The review was also limited by including English language publications and published data only. The preponderance of cohort studies including those using two groups of patients before and after a behaviour change intervention, and the implicit limitations of them, represents a limitation of this body of evidence. There was also no examination of the effect of sedation depth on related activities such as early mobilisation or on infrequently measured adverse events such as thromboembolic events.

Based on the low certainty of evidence, there is an urgent need for systematic evaluation of the effect of sedation depth on patient-centred outcomes to provide direction for sedation management. Studies addressing this question should use an RCT design, ideally with randomisation at cluster level to achieve cultural change in clinician behaviour. Studies should incorporate a priori identification of target ‘light’ sedation levels, based on individual patient need, and the effect on a range of patient-centred outcomes56 57 should be assessed.

Despite inconsistency in results, all clinical benefits identified in this review were related to lighter sedation, and importantly this review did not identify any harm related to lighter sedation. In this context, strategies to embed lighter levels of patient sedation in critical care are warranted. The challenging and multidimensional nature of sedation practice has been identified,58 and additional evidence-based strategies are urgently needed to optimise sedation and related areas of care such as early mobilisation.


Despite a considerable body of evidence discussing the relationship between sedation depth and various outcomes, we identified low to very low quality evidence suggesting that lighter sedation may be beneficial in some patient outcomes. The inconsistency of this evidence is exacerbated by the variable risk of bias in included studies, the different evidence of impact between RCTs and cohort studies, the inconsistent evidence of benefit across different outcomes and the inconsistent methods used, preventing combining data in meta-analyses. Future studies using rigorous controlled trial designs measuring patient centred outcomes, with randomisation occurring at the cluster level, are needed to understand the benefits associated with lighter patient sedation across a range of patient outcomes.

Data availability statement

No data are available.

Ethics statements

Patient consent for publication


The authors would like to thank Dr Maggie Tarling for assistance with some of the data extraction.


Supplementary materials

  • Supplementary Data

    This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.


  • Twitter @Leanne_Aitken, @Claire Kydonaki@CKydonaki, @Mandy_sekhon

  • Contributors Review concept and design: LMA, KK, BB and TSW. Study identification, critical appraisal and data extraction: LMA, KK, BB, LGT and MS. Statistical analysis: EP. Interpretation of analysis: all authors. Drafting of manuscript: LMA. Critical revision of the manuscript for important intellectual content: all authors.

  • Funding The authors have not declared a specific grant for this research from any funding agency in the public, commercial or not-for-profit sectors.

  • Competing interests None declared.

  • Provenance and peer review Not commissioned; externally peer reviewed.

Linked Articles