Anesth Analg 2001;93:514
© 2001 International Anesthesia Research Society
LETTERS TO THE EDITOR
A Randomized Controlled Trial on Local Anesthetics in the Wound: What do the Numbers Mean?
Stephan K.W. Schwarz, MD, Dr med
Centre for Anesthesia & Analgesia, Departments of Anesthesia and Pharmacology & Therapeutics, The University of British Columbia, Vancouver, BC, Canada, Dr. Rowlingson did not wish to respond.
To the Editor: The recent article by Fredman et al. (1) and the accompanying editorial by Rowlingson (2) raise concerns about the validity of the conclusions drawn from the results (1) and the indirect inference (2) about the lack of utility of randomized controlled trials. In their article, Fredman et al. (1) make no mention of a post hoc power analysis, necessary to determine whether the failure to detect a difference is simply a result of an insufficiently low sample size. Using the data on "rescue" morphine administration (total morphine, mg/6 h; mean ± SD: 16 ± 17* [n = 25] vs 18 ± 14 [n = 22]), such an analysis yields a power of <10% to detect the 5-mg difference for  = 0.01 (two-tailed), and <20% for = 0.05 (two-tailed). Moreover, the results imply that there were patients with negative morphine consumption (*), raising further questions about the data and/or the need for a log transformation. The only valid conclusion that can be drawn is that the results are inconclusive and repetition of the trial with a sufficiently large sample size would be necessary. For example, assuming a SD of 14 for "rescue" morphine (mg/6 h), n = 100 patients per group would be required to detect a difference of 5.58 mg at 80% power and = 0.05 (two-tailed). Dr. Rowlingson writes: "So much for randomized, controlled trials, eh?" (2). Randomized controlled trials remain the gold standard for experimental design in clinical research and provide the crucial basis for the increasing trend toward evidence-based practice in anesthesiology (3). In order to be able to utilize these powerful research tools appropriately, however, it is imperative to carefully consider what the numbers actually mean.
References
-
Fredman B, Zohar E, Tarabykin A, et al. Bupivacaine wound infiltration via an electronic patient-controlled analgesia device and a double-catheter system does not decrease postoperative pain or opioid requirements after major abdominal surgery. Anesth Analg 2001; 92: 189–93.[Abstract/Free Full Text]
-
Rowlingson JC. How can local anesthetic in the wound not help? Anesth Analg 2001; 92: 3–4.[Free Full Text]
-
Pronovost PJ, Berenholtz SM, Dorman I, et al. Evidence-based medicine in anesthesiology. Anesth Analg 2001; 92: 787–94.[Free Full Text]
Brian Fredman, MB BCh,
Edna Zohar, MD, and
Robert Jedeikin, MBChB, FFA(S.A.)
Department of Anesthesiology and Critical Care, Meir Hospital, Kfar Sava, Israel
In Response: When analyzing the efficacy of a postoperative analgesic technique, it is important to remember that acute pain is influenced by many subjective variables. First, the patient’s cultural and psychosocial background will affect pain perception. Second, relief at having completed the surgical "ordeal" as well as euphoria following surgical delivery may decrease the centrality of pain in the immediate postoperative period. Finally, anxiety, depression, or learning that a specific postoperative pain regime does not provide pain relief or intrinsically induces pain may affect patient perception of postoperative pain. This, in our opinion, is the message of Dr. Rowlingson’s editorial.
It should be remembered that prestudy power analysis is a limited tool because it is based on results of a similar (but by definition different) study. Furthermore, post hoc power analysis is used to define the likelihood of a type I or type II statistical error. Because the entire battery of tests used in our study (visual analog scale scores at rest, on coughing, and after leg raise, "rescue" opioid administration, patient satisfaction) showed no difference between the groups, there is no clinically significant reason to suspect the occurrence of a statistical error.
This study was associated with an 80% power to detect a 11.3 mg/6h ( = 0.05, two-tailed) and a 14.5 mg/6h ( = 0.01, two-tailed) difference in "rescue" morphine administration. Because this difference was over a 6-h period, our instillation system did not significantly improve patient care. Furthermore, we do not believe that a smaller difference in "rescue" morphine administration is required to persuade the Health Care Provider that this specific postoperative analgesic technique does not work.
|
