It’s a Disposable LMA, Just Cut It Shorter— for Fiberoptic Intubation

world. It should be noted, for example, that the commonest morbid endpoint in our trial was sepsis, occurring in 45% of our high-risk patients across the trial—many anesthesiologists would remove an epidural catheter in these circumstances. Our mean duration of epidural analgesia was 74 h. The occurrence of inadequate analgesia and the attrition rate for epidural duration in the trial are consistent with the published experience of other large observational studies of the use of epidural analgesia (2,3), and so, as was our objective, it reflects routine clinical practice. Distortion in interpretation of results from a trial can occur, however, where data are manipulated in variance to the intention to treat analysis. Attempts to stratify patients according to duration of epidural catheterization, as suggested by Gogarten et al., are a good example. Selecting only those patients in whom an epidural infusion was continuing at 72 h immediately introduces serious survivor bias into the comparison with the control group (for instance, those who die within 72 h of surgery do not have the opportunity of receiving 72 h of epidural analgesia). We rejected this approach early in our examination of our findings, especially since a quarter of deaths in the trial occurred in the first 48 h postoperatively. However even if these early deaths are excluded there is still no trend in favor of improved survival in the epidural group (deaths after 48 h:14 in control group, 18 in epidural group).