Lies, damned lies, and statistics

If you missed the media storm that descended upon us on the 5th of June, where were you hiding? Anaesthesia published what is surely going to be one of the landmark papers not just in our beloved specialty, but across all academic medicine. Our very own John Carlisle, one of the editors of this journal, has spent years designing a statistical technique to analyse the baseline patient characteristics of randomised controlled trials, and applied this to more than 5000 published studies over a 15-year period across six of the largest anaesthetic journals, as well as the New England Journal of Medicine and the Journal of the American Medical Association. By comparing the reported to the expected distribution of variables such as age, gender, height or weight, the Carlisle Method gives the probability of these characteristics coming from a truly randomised population sample. He showed that 1.6% of published RCTs use data that is either erroneous or fabricated. The Carlisle Method demonstrated a higher frequency of non-random sampling in retracted studies, but found no differences between the anaesthetic and general medical literature. There are numerous reasons for the data to be erroneous, including simple mistakes by both authors and journals, as well as data reporting and analysis, and the publication and method used will now be subject to further scrutiny.

Loadsman and McCulloch contextualise Carlisle’s paper, both applauding and expressing reservations. Perhaps one of the more interesting takes on the issue is that fraudsters may adapt to find newer ways of overcoming the Carlisle Method. They take the example of software that is currently being used to overcome the issue of plagiarism in literature, and state that there are now accessible methods to circumvent the software. Whilst a massive undertaking is called for to clean up the potential mess that John Carlisle has picked up, the authors are not clear how and when this methodology will become the industry standard.

Of course, followers of this blog and of Anaesthesia will be familiar with the Carlisle Method, as it has been used to pick up non-random sampling in 31 trials published by Yujhi Saitoh after previous investigative successes with Yoshitaka Fujii. Anaesthesia has since decided to apply the Carlisle Method do all submitted RCTs to the journal with the aim of increasing the quality of published studies, and combat scientific misconduct.

In the June edition of the Journal, another high-impact paper was published looking at the impact of  implementing the Lifebox pulse oximeter in Malawi. These simple yet game-changing devices are not broadly available in low- and middle-income countries. Introducing them initially requires staff training before the monitors could be demonstrated to improve safety. That is precisely what Albert et al assessed, and found that understanding and knowledge retention of pulse oximetry increased. Perhaps more importantly, they also found a 36% reduction in oxygen desaturation episodes after training staff with the device.


Scott and McDougall’s accompanying editorial suggests that the data is a testament to dealing with a real-life practical problem. Looking beyond pulse oximeters, the authors remind us that 77,000 operating theatres did not have one, but the simple introduction of the WHO Surgical Safety Checklist led to tangible outcome improvements. The key, argue Scott and McDougall, is an effective education programme that aids in attainment and retention of knowledge, which is what was beautifully demonstrated in Alberts study. It is not enough to merely donate equipment, but training in its use is just as important.

Moving back from a global stage to the UK, exciting trainee research networks are beginning to produce high-quality and practice-changing data. The PAINT Study is one such paper, which looked at how often physicians document (if it isn’t documented, it didn’t happen!) assessing pain in critical care patients. In a 24-h snapshot study, they assessed documentation from all adult critical care patients across 45 centres in London and the South-East of England. They found that 21.2% of the 750 patients had no documented pain assessment by anyone, 28.6% had no documented pain assessment by a nurse, and 64.5% had no documentation of pain being assessed by a physician. This included many patients that were receiving opioid infusions, and even patients where changes in analgesic regimen were concurrently implemented. This is certainly an area that all clinicians, not just critical care physicians, must work hard to improve. I know I’ve changed my practice since this paper!

At an institutional level, El-Boghdadly et al prospectively assessed awake fibreoptic intubation (AFOI) practice at their tertiary centre. They reported that the most common indication for AFOI was limited mouth opening, and less than 1 in 100 were truly ‘awake’ intubations (I know I would rather have some sedation if I needed AFOI!). Interestingly, three-quarters of AFOIs were done by trainees, and the success rate was independent of training grade, but dependant on practice. Only 1% of AFOIs were not successful, but there were no episodes of severe complications, CICV or hypoxia. They have progressively taken up high-flow, heated, humidified nasal oxygenation (remember the big hitter of 2015: THRIVE!) during AFOI.

Murphy and Howes critically appraised this study in an interesting accompanying editorial. They question the generalisability of data from a single-institution study, particularly as the training opportunities afforded in that institution seem higher than most, alluding to a previous editorial suggesting an increasing role for videolaryngoscopy rather than AFOI. Additionally, the editorial points to the increased use of THRIVE did not reduce the rate of complications, and if anything increased the incidence of over-sedation. Are we now over-relying on THRIVE without the evidence to support it? Lots of interesting questions asked here, and it is worth considering where AFOI sits in current practice.


The June edition of the Journal was exciting and varied. We published interesting ultrasound data demonstrating delayed gastric emptying in patients with renal failure, a brilliant bench study revealing that different spinal needles have variable flow characteristics, a game-changing Cochrane review demonstrating the superiority of suxamethonium over rocuronium for RSI, and a terrifying case report of airway ignition with THRIVE. If you pick it up this month, you won’t be able to put it down.


Kariem El-Boghdadly

Trainee Fellow, Anaesthesia

Andrew Klein


One thought on “Lies, damned lies, and statistics”

Leave a Reply

Please log in using one of these methods to post your comment: Logo

You are commenting using your account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

%d bloggers like this: