Edwards et al. are to be applauded for their efforts to provide a sedation service delivered by non-anaesthetists, using propofol and opioids.  There is no doubt that there is increasing pressure on anaesthetists to “relinquish” control of what is seen to be less than general anaesthesia.  However, there are problems with this, not the least that most anaesthetists would consider sedation a more challenging task than general anaesthesia.  Indeed, mortality data would bear this out. Quine’s seminal paper of 1995 [1] which studied > 14 000 gastroscopy sedations produced an overall 30 day mortality of 1:2 000; 2 of the deaths were attributed to the sedation, giving a 1:3 500 sedation mortality  (cf. mortality for general anaesthesia of approximately 1:100 000 [2]) – quite risky in comparison.  Are patients sedated by non-anaesthetists told at consent that their risk of death is at least 30 times greater than it would be for a general anaesthetic administered by a trained anaesthetist?  The problem is that most patients, and health workers, wrongly consider sedation to be safer than general anaesthesia.

There are many published series of sedations by non-anaesthetists administering propofol in the literature, but it is important to remember the “rule of 3” when attempting to work out the mortality in a group where no mortality has occurred.  This is that risk= 3/No of observations [3], i.e. Edwards series of 4 342 cases without a death, means that he can be 95% confident that the mortality of the service no worse than 1:1 447. This is quite an achievement, but no better than Quine’s series in 1995, and still at least 70 times the likely risk of a  general anaesthetic given by a trained anaesthetist.

I believe that sedation can be safe in the very controlled circumstances described in this study, but the danger is “mission creep”: such that others dumb down the rigidly controlled system in Glasgow, so that propofol sedation given by non-anaesthetist for a wide range of procedures becomes an accepted norm, which would undoubtedly lead to unnecessary deaths.  In my own institution, it is suggested that sedation of this nature should be given for electrophysiological procedures such as AF ablation, where the patient must remain completely still for > 4 hours, as papers have been published from Germany and the USA "demonstrating" the safety of such a technique. Should the minimum training for the use of anaesthetic induction agents for sedation remain the FRCA?  I believe that this is the safest course, and that we should not be forced into lowering standards of training and practice by the very low risks that these very things have achieved.  A risk of 1:100 000 is small (and yet to be achieved by any sedation series),  but even at that level it is still twice the 1:200 000 risk.

1.       Quine MA et al. A prospective audit of upper gastrointestingal endoscopy in two regions of England: safety, staffing and sedation methods. Gut 1995; 36:462-7.

2.       NHS Choices http://www.nhs.uk/conditions/anaesthetic-general/Pages/Definition.aspx

3.       Ho AM, Chung DC, Joint GM. Neuraxial Blockade and Haematoma in Cardiac Surgery. Estimating the risk of a rare adverse event that has not (yet) occurred. Chest 2000; 117:551-5.

We thank Dr Hunter for the interest in our article [1]. We are gratified that he sees our attempts to deliver a somewhat controversial service within an adequate clinical governance framework as admirable. We wholeheartedly agree, as stressed in the article, that it is the whole package of care, not just the protocol that have given us the results we have obtained. There is no guarantee that others would achieve similar results in different institutions. Furthermore, we agree that these extended roles are worthy of discussion before there is more widespread managerial pressure to replicate such a service without the safeguards that we have attempted to put in place.

Regarding the discussion of estimates of our (likely) mortality rates we would like to raise the following points:

1. We deliberately avoided quoting mortality rates as we thought we had inadequate data to make reasonable inferences, in contrast to morbidity, which is more common. Therefore we think our confidence intervals are useful in benchmarking our service with others published.
2. We are well aware of Hanley's formula, and indeed referenced the original article in our manuscript [2]. Obviously. we discussed it with regards to morbidity. It should not be forgotten that only quoting the upper limit is useful in reminding us that although we have no mortality in our series, the rate could be as high as 4342/3= 1 in 1447. Poisson modelling gives us a more accurate estimate of 95% confidence of a mortality rate as 0 to 3.6889/4342= 0-0.85/1000. Hence with equally probability, the mortality rate could be as low as the lower limit i.e. zero. The reality is that it is probably somewhere in between. The estimate is of course imprecise until we actually have encountered several deaths. Given that the service has been running for 5 years with no actual patient harm, let alone any (anaesthetic) morbidity or mortality, it will take a very long time to get a meaningful estimate of mortality rate.
3.  Comparing our sedation to that from the Quine et al [3] is not helpful. Firstly, this study was performed in the mid 90s when the sedation techniques (in this series) involved high doses of sedatives, low rates of monitoring, limited intravenous access, little use of supplementary oxygen and restricted recovery facilities. Secondly, most of the deaths were determined as being because of other predominantly surgical factors, not related to sedation. Hence, the sedation related mortality rate was calculated as 2/14000 so actually 0.14/1000 patients. Thirdly, the demographics of the patients who suffered mortality was quite different to our patients; they were primarily in their 70s with ASA 3. So this mortality rate should be compared to that for general anaesthesia of patients with similar ASA status i.e ASA 2= 0.7%, ASA 3= 3.5% [4]. When we do this, sedation seems far less risky. In summary, our patients were younger, fitter, better monitored, received supplementary oxygen, intravenous access, and adequate recovery facilities.

If comparisons must be made from two totally different cohorts of patients, it would be better to compare morbidity, which has a higher incidence rate and thus better estimate of true rate. Despite the 2 airway/respiratory events we described in our study, none of our patients actually suffered any arterial oxygen desaturation, let alone measurable morbidity. Not with standing our incident rate of 0.5/1000 patients is 10 fold less than that morbidity rate quoted by Quine et al. of  5/1000. We have no reason to believe that our rate of mortality would not be significantly different also .

We are not complacent but realistic about our results. Others may try to replicate our service, but the real value of this report is to highlight the lack of guidance in developing such a service and to stimulate debate on sedation practice in general, and in particular who can and should do it.

References:

1. Edwards JA, Kinsella J, Shaw A, Evans S, Anderson KJ. Sedation for oocyte retrieval using target controlled infusion of propofol and incremental alfentanil delivered by non-anaesthetists. Anaesthesia 2010; 65: 453-461.
2. Hanley JA, Lippman-Hand A. If nothing goes wrong, is everything alright? Journal of the American Medical Association 1983; 259: 1743-5.
3. Quine MA, Bell GD, McCloy RF, Charlton JE, Devlin HB, Hopkins A. Prospective audit of upper gastrointestinal endoscopy in two regions of England: safety, staffing, and sedation methods. Gut 1995; 36: 462-467
4. Walters, Wolf, Stutzer, Scroeder. ASA classification and perioperative variables as predictors of postoperative outcome. British Journal of Anaesthesia  1996; 77:217-222.