Glenfield General Hospital Leicester

with error bars of 2 SD, the results may be seen in perspective (fig. 1, lower panel). The investigators described repeat testing for differences between the groups by Student's unpaired t test or by Mann— Whitney U test, and stated that they then confirmed the differences by analysis of variance or by Kruskal-Wallis. However, it is invalid to test repeatedly by Student's or by Mann-Whitney unless a difference between groups has first been shown by analysis of variance or by Kruskal-Wallis. The investigators' final assertion is unproven. They imply that a patient who has already had a pressor response to cannulation may have an exaggerated response to a further stimulus, and that this is the reason that high risk patients should receive local anaesthesia before cannulation. The investigators did not describe subsequent laryngoscopy, so do not know that a response would be exaggerated. Patients truly at high risk would almost certainly receive some prophylactic treatment before laryngoscopy, and "will never be included in the control group of an outcome study examining the efficacy of haemodynamic control at... intubation!" [2].