To the Editor:
We suggest that several clinical and statistical aspects of the article by Lim et al.1 (comparing three regimens of patient-controlled epidural analgesia [PCEA] dosing as part of a combined spinal and epidural technique) should be clarified.
First, the final line of the abstract states “Demand-only PCEA (5-mL bolus, 15-min lockout interval) resulted in … lower maternal satisfaction, when compared with PCEA with background infusion.”1 However, we find no support for the abstract’s conclusion regarding maternal satisfaction in the body of the article: “no difference in … maternal satisfaction among groups (Table 2).” This is concerning because it has been our experience that many readers do not read beyond the abstract of an article, and, if so, this could inadvertently mislead a portion of the readers. Table 2 in the article shows a P value of 0.06 for the satisfaction data that seem very similar on inspection, with values in each group having the same median and with similar ranges: 90 (50–100), 90 (50– 100), and 90 (60–100) for Groups 0, 5, and 10, respectively. (Because we do not have access to the raw data, it is impossible to verify the P value because a nonparametric test was used to analyze these data.) Thus, the conclusion of the abstract seems to contradict the data presented,1 recent work,2 and two reviews referenced in the article.3,4
Second, in the text, the final statement of the introductory paragraph states “[PCEA] results in a higher patient satisfaction …” Neither of the two referenced articles supports this assertion.3,4 Although Saito et al. found a trend toward improved satisfaction with PCEA,5 statistical testing does not appear to have been completed; patient satisfaction was not measured in their 2009 study.6
Third, the PCEA dosing regimens chosen for comparison seem counterintuitive and not clinically relevant in that the regimens with larger basal infusions also allowed more frequent bolus infusions. Although the allowable theoretical maximum of 20 mL/h was the same for all regimens, it would be very difficult to obtain 20 mL/h with the demand-only regimen; one would have to time PCEA demands precisely every 15 min, whereas just two demands separated by 10 min would lead to the maximum dosing in an hour under the largest basal rate regimen. Thus, the demand-only group was not able to use as much as local anesthetic, as shown in Table 2. In contrast to the authors’ conclusions, one could instead draw the conclusion that either more frequent availability of PCEA boluses or a regimen that, in practice, allowed more local anesthetic consumption caused the improved outcomes they observed.
Fourth, data for the number of supplemental boluses in Table 2 are difficult to interpret. It would be more informative to provide the actual numbers of patients receiving 0, 1, 2, and 3 boluses, rather than median and range data. We are left to trust that “0 (0–2)” in Group 5 is significantly different (P < 0.001) than “0 (0–3),” with little information about the actual frequency of supplemental boluses.
Fifth, it is unclear from the description of statistical tests in the article which test was used to analyze the “Duration of second stage” and “Effective analgesia” duration variables in Table 3. The data are presented as mean ± sd, which suggests a parametric test was used (if not, as with other nonparametric data in their article, data are usually presented as median [interquartile range]), but it is stated in the Methods section that a “back-up Mann– Whitney U” test was performed on the duration of effective analgesia variable. The Mann–Whitney U-test is a nonparametric test, and so it is unclear which type of testing (parametric or nonparametric) was actually used.
Sixth, Table 3 likely has at least two typographical errors. The P value for the duration of second stage (min) between Group 0 and Group 5 is listed as 1.00, which appears unlikely given that the means are quite different. Total duration of labor (min) in Group 5 is listed as “392 ± 03,” suggesting an implausibly small coefficient of variation.
Lastly, a recent editorial noted that mixed-effect models should be used for analgesic trials involving repeated measures over time.7 Lim et al. have not attempted to develop a mixed-effect model of analgesic effect, which has likely resulted in loss of statistical power.
Timothy P. Turkstra, MD, M. Eng, FRCPC
Philip M. Jones, MD, FRCPC
Department of Anesthesiology and Peri-operative Medicine
University of Western Ontario
London Health Sciences Centre, University Hospital
London, Ontario, Canada
1.Lim Y, Ocampo CE, Supandji M, Teoh WH, Sia AT. A randomized controlled trial of three patient-controlled epidural analgesia regimens for labor. Anesth Analg 2008;107:1968–72
2.Vallejo MC, Ramesh V, Phelps AL, Sah N. Epidural labor analgesia: continuous infusion versus patient-controlled epidural analgesia with background infusion versus without a background infusion. J Pain 2007;8:970–5
3.van der Vyver M, Halpern S, Joseph G. Patient-controlled epidural analgesia versus continuous infusion for labour analgesia: a meta-analysis. Br J Anaesth 2002;89:459–65
4.Hodnett ED. Pain and women’s satisfaction with the experience of childbirth: a systematic review. Am J Obstet Gynecol 2002;186:S160–72
5.Saito M, Okutomi T, Kanai Y, Mochizuki J, Tani A, Amano K, Hoka S. Patient-controlled epidural analgesia during labor using ropivacaine and fentanyl provides better maternal satisfaction with less local anesthetic requirement. J Anesth 2005;19:208–12
6.Okutomi T, Saito M, Mochizuki J, Amano K, Hoka S. A double-blind randomized controlled trial of patient-controlled epidural analgesia with or without a background infusion following initial spinal analgesia for labor pain. Int J Obstet Anesth 2009;18:28–32
7.Shafer SL, Struys MM. Mixed effect modeling in analgesia trials. Anesth Analg 2008; 107:9–10