“Evidence for compromised data integrity in studies of liberal peri‐operative inspired oxygen.” Myles P. S. et al. Anaesthesia 2019

Background

• The World Health Organization (2016) made a strong recommendation that adult patients undergoing general anaesthesia for surgical procedures should receive an 80% fraction of inspired oxygen to reduce the risk of surgical site infection based upon a meta-analysis of 15 randomised controlled trails
• Five of these trials have been retracted, two of which were by the Mario Schiertoma’s group
• The merits of liberal peri-operative inspired oxygen have been disputed

Aim

• To identify compromised data by Mario Schietroma’s group
• To undertake a meta-analysis of other randomised trials of liberal peri-operative inspired oxygen

Methods

• Literature search for papers with Dr Mario Schietroma as author, included RCTs and observation series published in English
• Use of ‘Carlisle method’ to calculate probability of distribution of baseline characteristics for participants use RCTs.
• Checked calculations of p values represented in any paper.
• Compared figures for duplication
• Meta-analysis of liberal trials using Cochrane Collaboration software

Results

• Identification of 40 papers that’s fulfilled criteria: 24 RCTS with 5064 participants and 16 observational studies with 1847 patients
• Evidence that data intergrity was compromised in 38 of the 40 analysed papers
• Meta-analysis of 20 other randomised trials of liberal peri‐operative inspired oxygen in 8357 participants. Pooled results without trials by Schietroma et al. and not, did not show an effect of liberal peri‐operative inspired oxygen on wound infection (OR (95%CI) 0.89 (0.73–1.08); p = 0.23).
• Liberal inspired oxygen reduced the rate of postoperative wound infection if trials by Schietroma et al. were included (OR (95%CI) 0.77 (0.61–0.96); p = 0.02 ).

Discussion
Strengths

• Extreme p values for baseline distributions identified
• Identification of replication of graphs and data amongst papers
• Only systematic and randomised trials were used in the meta-analysis. All of the reported data in each of the relevant studies were checked. This resulted in the identification of numerous discrepancies

Weakness

• Inherent problems of
  o Carlisle method:
  – Inherent with statistics
  – error may not be secondary to fraud
  o Meta-analysis
• The WHO has already revised its guidelines on liberal peri-operative inspired oxygen following a series of retractions by the Schietroma group.
• Ethics behind using Carlisle method to discredit data in a public domain

---

Summary by Dr Rose Beesley