In Reply:—
We thank Dr. Scott et al. for their interest in our article.1In their eagerness to defend their use of thoracic epidural anesthesia/analgesia (TEA) in all cardiac surgery patients, Scott et al. attempt to invalidate the results of our study. They claim that our study “is grossly underpowered to confirm or refute the findings of any previous study on postoperative outcome” and that our study has “no statistical merit in its design.” We strongly disagree. For the purpose of our study, our population size is adequate. We had no intention to evaluate the effect of patient-controlled thoracic epidural analgesia (PCTEA) on cardiac, renal, pulmonary, or neurologic complications as primary postoperative outcome variables. Our study was designed to evaluate the potential effects of PCTEA versus patient-controlled intravenous analgesia (PCA) on length of hospital stay and patient-perceived quality of recovery, as also stated in the title of our article. We performed a power analysis (clearly described in our article) and found that an estimated 40 patients in each group were needed at a statistical power of 80% and a significance level of 5% to detect a clinically meaningful, 2.5-day reduction in length of hospital stay. Allowing for failures to complete the study, more than 100 patients were recruited.
Scott et al. raise concerns on “the conduct of the epidural regimen” in our study. They suggest that “patients in the TEA group did not have an effective block.” They even give the statement that the epidural failure rate of our study “is more likely to be closer to 100%.” Furthermore, they imply that we did “not establish” that our epidurals were “truly working.” Our group has a documented 20 yr of experience in providing high TEA in patients with severe coronary artery disease and unstable angina pectoris.2–7We have repeatedly shown that 4 ml bupivacaine, 0.5%, induces a blockade corresponding to Th1–Th7, i.e. , a complete cardiac sympathetic blockade.2–7We carefully tested that our epidurals were working, i.e. , caused a cardiac sympathetic blockade, on the day of insertion. Perioperatively, the patients received 8 ml bupivacaine, 0.5%, as a bolus and 8 ml/h as epidural infusion, which is an even higher dose than that administered by Scott et al. ,8who reported that they induced a neuraxial block from Th1 to Th10 in their study. More importantly, the catheter function was also tested after surgery in all patients before discharge from the intensive care unit. Therefore, 93% of patients allocated to PCTEA treatment did receive effective pain treatment with PCTEA during the time period when major endpoint variables were collected. Other evidence that our epidurals were “truly working” were that the patients were extubated earlier, they received 20–30% less propofol and remifentanil during surgery compared with controls, and the mean visual analog scale score was less than 1 on the 3 postoperative days of the study, as seen in table 3 of our study.
Scott et al. propose that the fact that the PCTEA and the PCA groups “are so similar in virtually all of the parameters measured . . . is strongly suggestive that [we] were comparing like with like and that patients in the TEA group did not have an effective block.” Surprisingly, it seems that the opinion of Scott et al. on the usefulness of TEA in cardiac surgery is so biased that it never occurred to them that our findings that PCTEA offers no major advantage with respect to hospital length of stay (LOS) and quality of recovery could be explained by the possibility that PCA might be a pain treatment that is as effective as PCTEA. Instead, they insinuate that the lack of effects of TEA on primary endpoints of our study1and in the study by Fillinger et al. 9is caused by malfunction of the epidural catheters.
Regarding the statement from Scott et al. that “this technique is not dangerous,” we would like to remind them that there are at least two published case reports of epidural hematomas related to TEA in cardiac surgery patients.10,11
We also thank Drs. Pivalizza and Warters for their constructive comments on our article. They raise questions mainly regarding the presentation of our data. Fifty-five patients in the PCTEA group received a successfully placed epidural catheter, and 55 patients were allocated to PCA treatment (fig. 1). We decided to present the preoperative data of all these patients in table 1, irrespective of whether they were finally analyzed. They correctly point out an error in table 2. In one of the patients in the PCTEA group, surgery was postponed, and the patient was transported (with the epidural catheter) to another hospital for surgery. Obviously, intraoperative data from this patient were not obtained. However, intraoperative data from the patient who died in the operating room after end of surgery were obtained. Thus, intraoperative data were obtained from 54 patients in the PCTEA group and not 55 patients as stated in table 2. In tables 3 and 4, data on postoperative visual analog scale and mobilization scores, as well as lung function, were obtained only from patients who were finally analyzed according to figure 1.
Yes, 7 patients had malfunctioning epidural catheters when assessed before intensive care unit discharge. Three of these patients had functioning epidural catheters replaced before discharge, whereas the remaining 4 received PCA. This is the clinical reality. Perioperative treatment with TEA will never be 100% successful. There will always be patients with suboptimal function of the epidural catheter. That is why we used an intention-to-threat analysis; it gives a more reliable estimate of true effect because it replicates what we actually do in routine practice. In our study, 51 of 55 patients (93%) allocated to treatment with PCTEA did receive PCTEA during the time period when the primary and secondary endpoint variables were collected.
The six patients who had postoperative stroke or confusion were not able complete quality of recovery, visual analog scale, or mobilization scores or lung function tests and were not finally analyzed according to figure 1.
It is difficult to compare various institutions with respect to hospital LOS after cardiac surgery. This variable is dependent on many factors not related to, for example, pain treatment itself. To circumvent this problem, the time to fulfillment of prospectively defined criteria for hospital discharge was assessed for each patient by observers blinded to treatment. Furthermore, one must define how LOS is assessed. In our calculation of actual hospital LOS, we included the day of admittance, i.e. , the day before surgery and the day of discharge. Actual mean hospital LOS in our study on a mixed population of cardiac surgery patients, including patients undergoing combined procedures, were 7.5 ± 3.3 and 7.9 ± 2.8 days for the PCTEA and PCA groups, respectively. In recent studies on patients undergoing low-risk coronary artery bypass surgery, hospital LOS varies between 5 and 7 days.
*Sahlgrenska University Hospital, Gothenburg, Sweden. sven-erik.ricksten@aniv.gu.se