|Home | About | Journals | Submit | Contact Us | Français|
We found the article of Sharma et al (Gut 2006;55:1233–9) of great interest. They have looked into 2‐year data comparing the efficacy of achieving complete reversal (endoscopic and histological) between multipolar electrocoagulation (MPEC) and argon plasma coagulation (APC) in patients with Barrett's oesophagus.
This was an important study if one accepts the hypothesis that the reversal between MPEC and APC may reduce the risk of adenocarcinoma of the oesophagus. However, we believe that the authors have misinterpreted their power calculation for the study and failed to address the potential limitations of both methods if they were to be adopted into standard clinical practice and shown to be of clinical benefit.
Firstly, there were no results on the practical aspects of both procedures for the readers to know whether they would be acceptable to the patients and feasible from the endoscopist point of view. We suspect that these frequent and time‐consuming methods have resulted in difficulty to recruit patients for this study. Only 35 patients from two large hospitals had agreed to participate in the study over a 4‐year study period. A participants' flowchart as recommended by the Consolidated Standards of Reporting Trials statement in 2001 would help to understand the acceptance by patients of these potential useful methods. The average duration for both procedures would also help the endoscopist to assess the feasibility in an already “demand out‐stretching capacity” endoscopy unit.
Secondly, we thought that the study was grossly underpowered and incorrectly accepted the null hypothesis (type II error)—that is, no difference in outcome between the two methods. Our calculation using Stata V.8 showed that 51 patients are needed to detect a significant difference with 80% power for each group. Therefore, the total number needed is 102, based on the assumption of 90% ablation of Barrett's oesophagus in the MPEC group and 65% ablation of Barrett's oesophagus in the APC group with 80% power and type I error of 5%.
We recommend that the authors accept the possibility for type II error in their results and conclusion.
Competing interests: None.