I read with interest the paper by Taha and colleagues [1], which investigated the onset of oxygen desaturation during apnoea following rapid sequence induction of anaesthesia, when using either suxamethonium or rocuronium. This is an interesting area of research particularly as rocuronium may begin to infiltrate the privileged position that suxamethonium has previously held, when used for rapid sequence induction.

However, I believe that the authors should have also included a group who received propofol and rocuronium as a direct comparison to the SO group (who received only propofol and suxamethonium). Without this extra group, I believe that their conclusion ‘when suxamethonium is administered for rapid sequence induction of anaesthesia, a faster onset of oxygen desaturation is observed during the subsequent apnoea compared with rocuronium' requires qualification in conjunction with the statement ‘after pre-treatment with 1.5 mg.kg^-1 lidocaine and 2 µg.kg^-1fentanyl'.  By including this fourth group, the authors may have been able to quantify the potential protection against desaturation by pre-treatment with lidocaine and fentanyl, when using propofol and rocuronium, during a rapid sequence induction. 

Furthermore, I have re-plotted the authors' box and whisker plot from the data, provided in their results section (Fig 1). This shows a larger range (longer whiskers indicated by the shaded area) for both the S and SO groups. I would like to ascertain if the statistical analysis was undertaken on the raw data, or on the data represented by the box and whisker plot, or indeed, are the shorter whiskers simply a typographical error. A widerspread data set could influence the degree of significance of the authors' findings. My interpretation of the spread of the data in the original figure is that it may be skewed. The authors described median values (a non-parametric indicator), but their statistical method appears to be a parametric one (analysis of variance). I wonder if the authors tested the data for normality, and if not why the authors selected a parametric analysis forseemingly non-normally distributed data. Despite analysis of variance being a robust statistical test, the application of an inappropriate statistical test has the potential to result in incorrect conclusions being drawn.

JS Dawson

Nottingham University Hospitals

Nottingham, UK.

E-mail: james@dawson.me.uk

No external funding and no competing interests declared

References

1.       Taha SK, El-Khatib MF, Baraka AS et al. Effect of suxamethonium vs rocuronium on onset of oxygen desaturation during apnoea following rapid sequence induction. Anaesthesia 2010; 65:358-61.

Figure 1 Time to reach S_pO₂ of 95% during apnoea following induction of anaesthesia with lidocaine/fentanyl/propofol/rocuronium (Group R), lidocaine/fentanyl/propofol/suxamethonium (Group S), or propofol/suxamethonium (Group SO). Horizontal line, median; box, IQR; whiskers, range. ^*p< 0.001 compared with Group R and Group S; †p = 0.003 compared with Group R. Shaded extensions to whiskers (range) to reflect the actual numerical data documented by the authors in the results section of the paper.