To the Editor:—
We read the recent article in your journal by Hansdottir et al. 1In it, they described a small (113 patients), prospective, randomized controlled trial on the use of patient-controlled thoracic epidural anesthesia/analgesia (TEA) for several types of cardiac surgery that did not show any difference on postoperative outcome comparing this technique with patient-controlled analgesia morphine.
In their article, the authors quote the article from our department in 2001 that showed significant improvements in outcome but not hospital discharge using TEA for coronary artery bypass surgery.2In this article, we took statistical advice from the outset to achieve a 90% power at a 1% level of significance for a reduction in the incidence of cardiac arrhythmias from 25% (considered to be the normal incidence in non-TEA patients) to 10%. We were advised that for one operation, we would need to investigate 420 patients, which we did in an open prospective manner. In doing so, we also found significant differences in several other outcome measures, including chest infection, acute renal failure, and postoperative confusion. At the time, our statistician calculated that we would need almost twice as many patients to determine true differences in the incidences of myocardial ischemia and infarction, given the much lower incidence for these after coronary artery bypass surgery. In the subsequent meta-analysis, Liu et al. 3calculated that 6,000 patients are needed in a randomized controlled trial to make definitive statements about the effects of different therapies on myocardial ischemia.
In simple terms, the study by Hansdottir et al. is grossly underpowered to confirm or refute the findings of any previous study on postoperative outcome had they investigated only one procedure. For the authors to include five different operations in their analysis of only 113 patients and for none of these subsets to have similar numbers of patient-controlled analgesia and TEA patients suggests that the study has no statistical merit in its design. Finally, we are not told how many valve procedures are aortic or mitral, or whether they were repairs or replacements.
Of greater concern is the conduct of the epidural regimen. In our opinion, there are two main reasons for using TEA, more so than analgesia: First is sympathetic blockade, with its consequent beneficial effects on postoperative organ dysfunction—this we were able to confirm in our study—and second, by avoiding moderate to severe pain, is the ability to avoid the use of systemic opioids in patients receiving TEA. This is a major target for modern acute pain services. It is also the reason that in our hospital we also use clonidine, rather than an opioid, in our epidural infusion because it too has both analgesic and sympatholytic properties.
In a clinical arena of such a highly controversial nature, to insert an epidural catheter and not establish that it is truly working when the perceived risk of complications is so high is surely inappropriate. Therefore, establishment of the block before both surgery and anesthesia is a fundamental requirement of the technique both from a philosophical and practical viewpoint.
The fact that the two groups are so similar in virtually all of the parameters measured by the 40-item quality of recovery questionnaire is strongly suggestive that Hansdottir et al. were comparing like with like and that patients in the TEA group did not have an effective block. This seems to be borne out by closer inspection of their methodology. The epidural was sited the day before surgery between T2 and T5, and 4 ml lidocaine, 1%, was given. This would merely confirm that the catheter was not intrathecal and is too little local anesthetic to confirm an effective epidural block. There is also no science to a loading dose or infusion rate based on ml/kg or ml · kg−1· h−1, and neither guarantees an effective block. At the end of surgery, a bolus of 0.1 ml/kg was apparently given to all patients in the epidural group. Why was this necessary if the epidural was effective? Therefore, no data are presented to confirm adequate placement of the catheter, spread of the local anesthetic, and effectiveness of the block.
For these reasons, we cannot agree that the failure rate was only 5.2% as the authors claim, and given the amounts of local anesthetic used intraoperatively and the need for an immediate top-up postoperatively, it is more likely to be closer to 100%. Moreover, to rely on a patient self-administering morphine as a means of determining a successful block is not appropriate because many patients will have good analgesia without an epidural catheter or, in patients who do have an epidural, without a demonstrable block.
The results of this study are similar to those of Fillinger et al. ,4who demonstrably failed to provide an adequate and effective block, thus ensuring no differences in outcome between groups. That study, too, allowed surgeons unblinded to the techniques to control the primary endpoint of their study, namely time to hospital discharge.
The literature on this topic does not need any more studies of this kind, which are too small to detect significant differences. Ideally, a large multicenter study should now be approved, but for a variety of reasons, this is unlikely to happen. In its stead, the only reasonable alternative is a thorough audit by centers that are experienced in the use of the technique. To that effect, we can confirm that since our study was completed, we have converted to the routine use of TEA for all coronary artery bypass surgery patients. In total, we have performed 2,700 TEAs for coronary artery bypass surgery, with a 28-day mortality of 0.9% and no incident of epidural hematoma. We are aware of similar data from other European centers with a total of around 15,000 patients, confirming that this technique is not dangerous, and we have now started to use it for valve patients as well, with additional precautions with regard to the use of warfarin.
*Golden Jubilee National Hospital, Glasgow, Scotland. firstname.lastname@example.org