My co-authors and I stand by the results of our study  but wish to point out some concerns about the study by Wittels et al.  as well as to respond to the letter.
We never claimed that addition of epidural butorphanol (BU) to epidural morphine (MS) reduces postoperative analgesia. However, Wittels et al. claim that overall analgesia from MS, as expressed by a cumulative VAS score, is superior when BU is added. Unfortunately, the significant P value reported by them does not apply to an analysis comparing the two groups in question. They used a one-way analysis of variance; we repeated this test using the means and standard deviations reported, obtaining a P value of 0.335 (or 0.236 if an adjustment is made for the unequal variation across the four groups). More importantly, a mean difference of 0.5 cm is not clinically significant in our view.
Table 1of their letter is misleading because the percentage values applied to VAS pain score > 3 relate to percentage of patients in our study, but percentage of all pain score evaluations in their study. By including all pain score evaluations, a bias toward the production of inappropriately liberal detection limits is created, which is not valid.
The letter also states that the pain scores we reported are unusually high. We do not agree, since our results are similar to others  and there is no basis for the descriptive labels they applied to VAS score ranges.
The treatment of pruritus in our study did not differ among groups, but Wittels et al. claimed a significant decrease in the need to treat pruritus when BU was added (25% vs 52%). We represented degree of pruritus on a 10-cm VAS scale at each assessment and found no differences among groups at any time point. We question why their data concerning the magnitude of pruritus are missing. They stated that request for treatment was the more important parameter. However, other factors such as nursing availability and selection bias are important determinants.
In conclusion, Wittels and colleagues have not demonstrated any benefit from the addition of BU to MS because serial measurements of data were not suitably analyzed. An effective treatment of the side effects caused by MS still eludes us.
David R. Gambling, MB, BS, FRCPC
Rick Risser, MS
Departments of Anesthesiology and Academic Computing Services, University of Texas Southwestern Medical Center, Dallas, TX 75235-9068
Jonathan Berkowitz, PhD
Berkowitz and Associates, Statistical and Mathematical Consultants, Vancouver, Canada V6N 3S2
1. Gambling DR, Howell P, Huber C, Kozak S. Epidural butorphanol does not reduce side-effects from epidural morphine after cesarean birth. Anesth Analg 1994;78:1099-104.
2. Wittels B, Glosten B, Faure EAM, et al. Opioid antagonist adjuncts to epidural morphine for postcesarean analgesia: maternal outcomes. Anesth Analg 1993;77:925-32.
3. Chadwick HS, Ready LB. Intrathecal and epidural morphine sulfate for post-cesarean analgesia. Anesthesiology 1988;68:925-9.