152 • CID 2007:44 (1 January) • CORRESPONDENCE Reprints or correspondence: Dr. Bertrand Lebouche, Ho ˆpital Ho ˆtel-Dieu, Service d’He ´patogastroente ´rologie, 1 Place de l’Ho ˆpital, 69002 Lyon, France (bertrand.lebouche@chu-lyon.fr). Clinical Infectious Diseases 2007; 44:151–2  2006 by the Infectious Diseases Society of America. All rights reserved. 1058-4838/2007/4401-0033$15.00 Nitazoxanide versus Metronidazole for Clostridium difficile– Associated Colitis To the Editor—Reports of increased incidence and severity of Clostridium dif- ficile–associated colitis have stimulated in- terest in the search for additional treat- ment options for this clinical problem. Musher et al. [1] recently reported the re- sults of a noninferiority study that com- pared metronidazole with nitazoxanide for the treatment of C. difficile colitis. The authors concluded that “nitazoxanide is as effective as metronidazole” (p. 425) for this infection. Scrutiny of some aspects of the study design and analysis calls this conclusion into question. Notably, there was an unexplained ex- clusion of a significant number of the ini- tially enrolled patients. The investigators compared the baseline characteristics and severity of illness in an “intent-to-treat” population; however, 32 enrolled and ran- domized patients were excluded before any efficacy analysis. Therefore, this was not an intent-to-treat analysis. The 110 subjects included in the primary efficacy analysis may have been very different than the initial 142 enrolled patients, and some effects of randomization were possibly lost. In addition, it is not revealed specif- ically what adverse events occurred among the excluded patients, and the cause of death was not revealed for any patient. The reader is told that 9.1% of the original 142 patients died, “indicating the severity of underlying disease(s) and/or the colitis” [1, p. 424]. This implies that some deaths were due to complications of C. difficile– associated colitis. This point should have been clarified. If adverse events or deaths were even peripherally related to colitis, the exclusion of these patients brings the study results into question. There were other concerns regarding this study. (1) The investigators did not analyze the minimum number of subjects required in each group and, therefore, did not achieve adequate statistical power to reject the null hypothesis. (2) When us- ing an active control with no placebo- only group, one should employ a double- dummy design to maintain true double- blinding. Metronidazole and nitazoxanide do not resemble one another and have dif- ferent dosing schedules. Although the ran- domization was blinded, it appears that the study drugs were not administered in a strictly double-blind fashion. (3) The reader is not told the proportion of subjects in each group who continued to receive the original offending antibiotic(s) that presumably led to C. difficile–associated colitis. With such a small study, one cannot rely on randomi- zation to eliminate the effects of this potential confounder. On the basis of in vitro susceptibility data, a hamster model of disease [2], and the clinical trial by Musher et al. [1], it appears that nitazoxanide does have some efficacy in treating C. difficile–associated colitis. However, one should not conclude that it is as effective as metronidazole on the basis of this study. If noninferiority to metronidazole had been shown appropri- ately, the significantly higher cost of ni- tazoxanide makes this drug difficult to rec- ommend as an alternative treatment. The role of nitazoxanide in current clinical practice remains undefined. Acknowledgments Potential conflicts of interest. All authors: no conflicts. Jeremy D. Young, Julie E. Mangino, Kurt B. Stevenson, and Susan L. Koletar Division of Infectious Diseases, The Ohio State University Medical Center, Columbus References 1. Musher DM, Logan N, Hamill RJ, et al. Ni- tazoxanide for the treatment of Clostridium dif- ficile colitis. Clin Infect Dis 2006; 43:421–7. 2. McVay CS, Rolfe RD. In vitro and in vivo ac- tivities of nitazoxanide against Clostridium dif- ficile. Antimicrob Agents Chemother 2000; 44: 2254–8. Reprints or correspondence: Dr. Jeremy D. Young, N-1129 Doan Hall, 410 W. 10th Ave., Columbus, OH 43210 (jeremy .young@osumc.edu). Clinical Infectious Diseases 2007; 44:152  2007 by the Infectious Diseases Society of America. All rights reserved. 1058-4838/2007/4401-0034$15.00 Reply to Young et al. To the Editor—We thank Young et al. [1] for their interest in our article. We believe that we can answer their queries and leave them confident in the validity of our conclusion—namely, that nitazox- anide is at least as effective as metroni- dazole for treatment of Clostridium diffi- cile–associated colitis. First and most importantly, the blinded, prospective nature of the study precluded bias. Exclusions and evaluations of adverse events or outcome were determined on the basis of predefined criteria without knowledge of treatment group assign- ment. No study is truly blinded if the in- vestigators can determine the treatment group by examining the pills, and we, of course, used a “double-dummy” design. Each patient took 1 yellow tablet (nita- zoxanide or a matching placebo) twice daily and 1 white tablet (metronidazole or a matching placebo) 4 times daily. The senior author (D.M.M.) preferred not to use the term “double-dummy” in the final article because it is jargon. In our article, the text and figure 1 pro- vided the reasons for excluding 32 pa- tients; the reasons included death (4 pa- tients); other adverse events, such as transfer of status to “nothing by mouth” (10 patients); dropping out (12 patients; generally because medications were not given when the patients were transferred from ward to ward); other medications for treatment of C. difficile infection ordered by residents (3 patients); and noncompli- ance (3 patients). The exclusions were equally distributed among the groups. Deaths specifically due to colitis did not occur. Characteristics of the remaining 110 patients, including the proportions who continued to receive the offending antibiotic(s), were similar among the 3 by guest on February 21, 2016 http://cid.oxfordjournals.org/ Downloaded from