The majority of responding ICUs (88%) in our survey use a sedation scoring system. This has increased considerably since the UK survey in 2000, when 67% of hospitals used a scoring system 22. Despite this increased uptake since the last survey and favourable comparison with other countries, an evidence-based approach is still not universally followed.

Numerous sedation assessment tools have been developed to minimize this risk for over-sedation. Some sedation scales have been validated against other scales in patients (for example, Riker Sedation-Agitation Scale 2425, Motor Activity Assessment Scale 26, Vancouver Interaction and Calmness Scale 27, and more recently the Richmond Agitation-Sedation Scale 2829, published after the American College of Critical Care Medicine (ACCM) guidelines 12. The latter has shown an excellent performance, not only with regard to inter-rater reliability and validity, but it is also the first score to detect changes over time in the critically ill patient. However, no consensus yet exists in an international guideline regarding which assessment tool to use, and validation itself is problematic because of the lack of a standard to validate against that is not based on opinion.

The uptake of the Richmond Agitation-Sedation Scale has been slow in previous surveys. It did not feature in the 2005 German survey 17 or in the Canadian survey conducted in 2006 18, with one unit reporting its use in the 2007 German update survey. The French survey in 2007 21 and our study are the first reports of its use being more widespread.

The sedation scale used most commonly in this survey was the Ramsey Sedation Scale 23, with 66.5% of ICUs using this sedation assessment. Furthermore, most of the other scales used are adaptations of the Ramsay Sedation Scale (for example, the UK Intensive Care Society sedation scale). The choice of sedation scoring tool has not changed since the last UK survey, with the Ramsay Sedation Scale score being used most commonly then (40 out of 142 units stating that they used a sedation scoring system in 2000) 22.

The widespread use of the Ramsay Sedation Scale is in contrast to the ACCM guidelines 12, which recommend use of the validated assessment scores, such as the Motor Activity Assessment Scale, Riker Sedation-Agitation Scale, and Vancouver Interaction and Calmness Scale. Its advantages appear to be familiarity to staff and simplicity, and it is the scale that has been most commonly used historically despite its clinical limitations. However, the Ramsay Sedation Scale lacks clear discrimination and exhibits considerable inter-rater variability 30.

The practice of sedation assessment in the UK differs from that in other countries. In Germany in 2005 only 51% of responding ICUs report sedation monitoring, with the Ramsay Sedation Scale used 'almost exclusively'17, and in Canada in 2006 only 49% of responding ICUs utilized sedation monitoring, with 69% of ICUs using the Ramsay Sedation Scale 171820. Our results are consistent with a 2001 sedation survey conducted in European ICUs 28, which showed that ICUs in the UK use sedation scales more frequently than do those in all other participating European countries (72% of UK ICUs).

It has been shown in many but not all studies over the past decade that an ICU sedation protocol results in fewer days on the ventilator, a shorter stay in the ICU and reduced costs 31323334353637. Despite the mostly good evidence and a comparably easy and cheap means of improving care, not all hospitals have implemented a formal sedation guideline.

In this survey, 80% of the responding hospitals have an operating sedation guideline, which has increased sharply since 2000, when 43% of participating hospitals stated that they had a written guideline. A German survey conducted in 2007 20 revealed that 46% of hospitals used a sedation guideline, and the Canadian survey in 2006 18 reported that 29% of ICUs used a sedation protocol.

Our high rate of ICUs reporting use of a sedation guideline could reflect reporter bias, because units with more interest in sedation may be more likely to respond to this questionnaire. However, there has been increased utilization of sedation guidelines in Germany in the recent years, suggesting that our results may reflect actual change in practice 20.

In 2000, Kress and coworkers 711 showed that daily withholding of sedative agents led to reduced length of ICU stay, less ventilator time, fewer ICU complications and fewer neurological investigations. Subsequent studies by the same group demonstrated daily sedation withholding to be safe in patients with ischaemic heart disease, and that it reduces the psychological sequelae of critical illness 1038. A different group, however, raised safety concerns in a trial including a high percentage of patients (around 30% to 40%) with alcohol and other drug use disorders, emphasizing that patient selection and an individualized approach is important 39.

A recent pilot trial addressed the issue of safety and feasibility of daily interruption of sedation with simultaneous use of protocolized sedation 40. The authors concluded that in their pilot daily trial sedation practice was not associated with an increased incidence of adverse events.

Sedation withholding is now part of the 'ventilator care bundle', as outlined by the UK Department of Health 33, and recommended by the Surviving Sepsis Campaign 41.

In our survey, 78% of the ICUs state that they practice daily sedation holding. By comparison, the reported proportion of ICUs practicing sedation withholding in Canada is 40%, in Denmark it is 31% and in Germany it is 34% 181920. However, our study revealed that only 53% of ICUs audit their use of daily sedation holding, which suggests that the number of ICUs practicing sedation holding effectively is probably lower than the 78% stated.

There may be many reasons why ICUs are not practicing daily sedation holding. Devlin and coworkers 42 surveyed American clinicians in 2004 and found that some ICUs were not adopting sedation holding because of a lack of nursing acceptance of this practice, a potential increase in patient self-harm, potential for respiratory compromise and concern about patient comfort.

The high rates of sedation holding reported in this study may reflect increasing awareness and acceptance of the technique and most subsequent studies supporting its safety and benefit.

The sedation guideline published by the ACCM 12 recommends use of fentanyl or morphine for analgesia, midazolam or propofol for short-term sedation, and lorazepam for longer term sedation. The practice in the UK differs greatly from these guidelines. Alfentanil is used more commonly than fentanyl or morphine, likely because of its lesser degree of accumulation and shorter duration of action. For patients expected to require sedation for longer than 24 hours, morphine and midazolam were most frequently chosen, whereas lorazepam is rarely chosen.

Our survey illustrated that the duration of action of the sedating agent was a more important factor in choice of sedating agent than its cost. This concurs with the German sedation survey 20. It will be interesting to observe whether newer short-acting but more expensive agents (for instance, remifentanil and dexmedetomidine) are chosen for sedation in the future.

Muscle relaxing agents are infrequently used in UK ICUs. The few ICUs using neuromuscular blocking agents (NMBAs) in more than 10% of patients were mostly neurological ICUs. The infrequent use of NMBAs may reflect the increasing emphasis on lighter levels of sedation and the concerns regarding critical illness neuropathy and myopathy 43.

This study shares the limitations of all surveys in that reporter bias cannot be excluded. Furthermore, only the head of the department was addressed; the answers may therefore only reflect the individual's practice, and may not be representative for the entire unit. Past surveys, however, would have faced similar limitations, and given our good response rate, comparison with surveys in the past and in other countries can be made. The wide range of units responding make it likely that our results reflect actual UK practice appropriately, within the constraints of self-reporting practice.