Background
Sedation and analgesia are frequently administered to patients in critical care units to facilitate therapies, such as mechanical ventilation and other invasive procedures, with the objectives of ensuring patient safety, comfort and sometimes amnesia. Commonly used sedatives are propofol, benzodiazepines and alpha-2 agonists (clonidine and dexmedetomidine). Guidelines favour sedation strategies using agents such as propofol or dexmedetomidine over strategies using benzodiazepines to improve clinical outcomes in mechanically ventilated patients [
1]. Previous surveys conducted in the United Kingdom (UK) [
2,
3] and elsewhere [
4] indicate a shift away from benzodiazepines toward propofol for sedating patients in critical care. Guidelines on sedation also include recommendations for pain management, with an emphasis on assessing and treating pain first before administering sedative medication [
1].
The Intensive Care National Audit & Research Centre (ICNARC) was commissioned by the National Institute for Health Research Health Technology Assessment Programme to design and conduct a study to establish current sedation practice in UK adult critical care. The overall aim of the study was to provide baseline data on current practice to inform evidence synthesis and potential future primary research. Specific objectives were to establish (1) through a survey of all UK adult general critical care units, reported current sedation practice; and (2) through a point prevalence study (PPS) in a representative sample of UK adult general critical care units, the current prevalence of use of sedative agents and regimens. This combination of study designs was selected to give the most complete picture of current practice in terms of both the breadth and the depth of information and to examine reported versus actual clinical practice.
Methods
National survey
National Health Service (NHS) adult general critical care units were identified from databases maintained by ICNARC, and the Scottish Intensive Care Society Audit Group [
5]. A general critical care unit was defined as an intensive care unit (ICU) or combined ICU-high-dependency unit (HDU). Stand-alone HDUs and speciality units (e.g., neurosciences, cardiothoracic) were excluded.
A review of previous surveys [
2,
3,
6‐
10] informed development of the survey tool (Additional file
1: Figure S1), which captured information on management of sedation as well as the use of specific sedative and analgesic agents. With respect to analgesia, the focus was on analgesic agents with a sedative effect, such as intravenous opioids, rather than on oral or regional analgesia. Following piloting, the survey was sent via email, in January 2014, to the clinical directors of all UK adult general critical care units, who were asked to complete the survey either online (via SurveyMonkey®), electronically or on paper. Non-responders were followed up, with a second email and then a telephone call, until May 2014.
Point prevalence study
The PPS was nested in the Case Mix Programme (CMP), the national comparative clinical audit for adult critical care in England, Wales and Northern Ireland, coordinated by ICNARC. At the time, 95 % of all possible adult general critical care units were participating in the CMP.
It was estimated that a sample size of 50 units would be required to give representation in terms of unit characteristics and to allow for variation in sedation practice. Based on CMP data, the mean number of patients in an average-sized unit at 1400 on a midweek day is 9; therefore, the projected total sample size was 450 patients, of whom it was anticipated that approximately 45 % (200 patients) would be sedated. Because the aim was to recruit 50 units, the invitation to take part was restricted to 97 actively participating units that were the most up-to-date with data submission and validation at the time. Those that agreed to take part were asked to complete a short data collection form (Additional file
2: Figure S2) for every patient in the unit at 1400 on 11 December 2013. For sedated patients (i.e., those who had received a sedative and/or analgesic in the previous 24 h), staff were asked to provide information on management of sedation as well as sedative and/or analgesic agents received. Completed data collection forms were returned either electronically or on paper.
Statistical analysis
For the analyses, we used data from the CMP database, which contains pooled case mix and outcome data on consecutive admissions to participating units and has been independently assessed to be of high quality [
11]. Responses to categorical questions in the survey were summarised as the number (percentage) of respondents selecting each response. Continuous data were summarised as the mean, SD and range (minimum to maximum) across the responses.
Patients in the PPS were compared with all patients in CMP adult general critical care units at 1400 on 11 December 2013 for case mix, outcome and length of stay. Numbers (percentages) of sedated patients assessed using a sedation scale/score were summarised in addition to, among patients who had been in the unit for at least 24 h, those considered for and/or receiving a sedation hold during the previous 24 h. The number (percentage) of patients receiving each sedative and analgesic and combinations of agents during the previous 24 h were summarised together with, for agents received by five or more patients, the mean (SD) of the highest rate of infusion (in milligrams per hour) and the total dose (in milligrams), including both infusions and boluses. With respect to analgesia, analgesia administered orally or regionally (e.g., epidural) was excluded from the analysis.
Survey responses were compared with PPS data, where both were available, to investigate the degree to which reported practice reflected actual practice. For units that reported daily sedation holds, the number (percentage) of sedated patients who had been in the unit for at least 24 h and who had been considered for a sedation hold was summarised. Across all units, the median proportion and IQR of sedated patients considered for a sedation hold were calculated.
For each sedation scale/score reported, the number (percentage) of sedated patients in the PPS who were assessed using the reported scale/score was summarised. The median (IQR) proportion of patients assessed using the same sedation scale/score as reported was calculated across all units. For patients who had received a sedative agent, the proportion that received the unit’s reported first-choice agent was summarised. The median (IQR) proportion of patients receiving any sedative agent who received the reported first-choice agent was calculated across all units. The same approach was repeated for analgesic agents.
All data analyses were conducted using Stata/SE version 13.0 software (StataCorp LP, College Station, TX, USA). Support for the collection and use of patient-identifiable data without consent for the CMP was obtained under section 251 of the NHS Act 2006 (approval PIAG 2-10(f)/2005).
Discussion
Of the 235 UK NHS adult general critical care units identified, a high proportion (91.1 %) responded to the survey. In addition, a representative sample of 52 units also participated in the PPS. Data derived from the two studies indicated that reported practice does not necessarily reflect actual practice. Most (88 %) PPS patients were assessed using the same sedation scale/score reported in the survey; however, there was some variation across units. Furthermore, although a high proportion (94 %) of units reported using daily sedation holds, overall only half of sedated patients in the PPS who had been in the unit for 24 h or more had been considered for a sedation hold during the previous 24 h.
Both the survey and PPS indicated that propofol is the preferred sedative and alfentanil, fentanyl and morphine the preferred analgesics in UK critical care. Most units (83 %) reported frequently/very frequently administering sedatives in combination with analgesics, and around two-thirds (68.6 %) of patients in the PPS had received a combination of sedatives and analgesics, most frequently propofol combined with either alfentanil or fentanyl.
The use of guidelines or protocols for management of pain, agitation and delirium is strongly recommended [
1,
15]. Only 57 % of units reported having a written sedation protocol, similar to the findings of a recent Internet-based survey of UK critical care pharmacists in which 55 % of respondents reported use of sedation guidelines [
16]. However, this is considerably lower than previously reported in the United Kingdom. In a telephone survey conducted in 2011 [
17] and a postal survey [
3] conducted in 2007, 82 % of units reported having a written sedation policy and 80 % having a sedation guideline, respectively. The different responses could be related to how the question was phrased and the mode of administration. The rate of protocol implementation reported in our survey is similar to rates that have been reported in other countries, including Australia (54 %) [
18], Germany (52 %) [
8] and the United States (64 %) [
19].
The RASS [
12] is recommended [
1] as one of the most valid and reliable subjective sedation scales for measuring depth of sedation [
20]. The reported use of this sedation scale has increased considerably in the UK since 2007 [
3], from 5 % to 65 % of units, with a general shift away from using the RSS [
13]; a quarter of units reported using this scale/score, compared with 67 % in 2007 [
3].
The practice of daily sedation holds is recommended [
1,
15,
21] and has been incorporated into the ventilator care bundle in the UK [
22]. The benefits of minimising sedation include less time on a mechanical ventilator, fewer complications and reduced length of stay in critical care [
23‐
25]. The proportion of UK units that reported practicing daily sedation holds was higher (94 %) than previously reported in 2007 (78 %) [
3] and more recently in 2014 (80 %) [
16]. Data derived from the PPS suggest that overall compliance is possibly much lower. Of the patients who had been in the unit for at least 24 h, around half (53 %) had been considered for a sedation hold in the previous 24 h, and of these, 77 % had their sedation withheld.
Previous surveys in the UK [
2,
3,
16] and elsewhere [
4] indicate a shift from benzodiazepines to propofol for sedating patients. Both the survey and the PPS indicated that propofol is by far the most widely used sedative agent in the UK. Although around one-third of units reported using midazolam, very few reported it to be their first choice. Use of the alpha-2 agonist clonidine has increased in the UK, with around one-third of units reporting very frequent/frequent use, whereas use of dexmedetomidine is rare. It seems that uptake of dexmedetomidine in UK critical care has been slow since it was licensed for use in 2011. Cost may be a factor, with clinicians preferring to use established and often cheaper alternatives.
Intravenous opioids are recommended as first-line agents for non-neuropathic pain in critically ill patients. Authors of previous UK surveys [
2,
3] have reported that alfentanil, fentanyl and morphine are the most frequently used opioids. Our findings are similar, although both the survey and the PPS suggest a trend away from morphine toward agents such as alfentanil and fentanyl as the first choice for analgesia. Authors of surveys done elsewhere have reported a similar trend, although morphine is still widely used [
4]. Even so, authors of a recent survey of UK critical care pharmacists reported morphine to be the most common first-line prescription in almost half (49 %) of units [
16].
Surveys are frequently used for establishing current clinical practice in a variety of healthcare settings, their advantage being that they are relatively cheap, quick and simple to conduct, although achieving high response rates can be challenging. A high response rate (91 %) was achieved in our survey, comparing favourably with some previous UK surveys with response rates of 79 % [
2], 64 % [
3], 78 % [
17] and 60 % [
16]. Elsewhere, response rates have ranged from 20 % [
9] to 84 % [
7]. One limitation of surveys, however, is that reported practice does not necessarily reflect the reality of actual clinical practice at the patient level. A major strength of the present study is the combination of a national survey of adult general critical care units and a PPS, which enabled us to examine both
reported sedation practice across the UK and
actual practice at the patient level in a representative sample of UK units. We were unable to investigate the reasons for the discrepancies observed between reported practice and actual practice because of limited resources for the study. However, it is likely that these reasons are multifactorial. For example, there may be a lag between a unit policy being initiated and staff at the bedside changing their clinical practice, or the discrepancy may reflect the difficulties in applying universal policies on sedation and analgesia to individual patients, such as that the choice of agent depends on a number of individual patient factors. Another possible explanation is that the person completing the survey is influenced by her own practice when reporting unit practice, which may not necessarily reflect that of her colleagues or may not even be in line with the unit policy. Our results indicate a need for improved auditing of clinical practice within units to ensure that agreed unit policy on clinical practice is being followed and, if not, to identify the reasons. Less than half of units reported in the survey that they audit compliance with policies, such as sedation holds, and among those that did, there was considerable variability in compliance with unit policy.
A possible limitation of the present study is that the PPS and the survey were not conducted contemporaneously. However, both were completed within a 5-month period, and it seems unlikely that during this time there were any major changes in policy within units that might explain the discrepancies in clinical practice observed. Furthermore, a potential strength of not conducting the two studies contemporaneously is that the PPS provided a snapshot of actual clinical practice before units were asked about their clinical practice in the survey. Nesting the PPS within the CMP allowed us to conduct a detailed examination of case mix and outcomes of patients who were in a CMP unit on the day of the study. Patient populations in CMP units that did and did not participate in the PPS were very similar with respect to case mix and outcomes. Furthermore, comparison of the survey data from 51 units that participated in the PPS with all other responding units did not reveal any differences in reported sedation practices, suggesting that units that participated in the PPS were a representative sample.
A PPS is relatively quick and simple to conduct. However, a limitation is that it will only provide a snapshot of clinical practice at a single point in time. It is also more likely to capture those patients who stay longer in the critical care unit. Sedation and analgesia regimens will vary according to the expected duration of sedation and analgesia, which is an important factor in determining the choice of agents.
Acknowledgements
We express our sincere gratitude to the staff at the critical care units that participated in the national survey and the point prevalence study. We also thank our colleagues at ICNARC for their support, in particular Prof Kathy Rowan and Dr David Harrison for their helpful comments and suggestions.