Intention-to-treat Analysis Controls the Bias When Suffer From Noncompliance and Missing Outcomes in Randomized Controlled Trial Study
The randomized controlled trial is regarded as the strongest research design to evaluate the effects of health interventions. Randomization is the key point to proper trial design because it removes the potential bias in the allocation of patients to different interventions, and guarantees the accuracy of statistical tests of significance used to interpret the results.1 Dr. Kang pointed the limitation of preoperative clinical staging in selecting the enrolled patients before randomization in this study.2 We are appreciated the comment. Indeed, even by detail preoperative evaluation, making an absolutely accurate tumor stage was still difficult before operation.3,4 There are small number of patients who were evaluated as stage II or stage III cancer before operation, but were diagnosed with stage I tumor after paraffin pathology, and even distant metastasis or peritoneal dissemination (stage IV) during surgical exploration.
How should investigators analyze study data if some patients have not adhered to the allocated management strategy, such as noncompliance or missing? In this study, we used intention-to-treat analysis, which may provide an unbiased assessment of the efficacy of the different interventions at the level of adherence observed in the trial.5,6 In addition, before randomization, both carefully preoperative evaluation and the exploratory surgery during operation were done, which excluded patients with peritoneal dissemination or distant metastasis (stage IV, approximately 2% as our experience). After that patients were randomly assigned into 2 arms interoperation.
For stage I patients, conventional endoscopic resection is a widely accepted treatment for early colorectal neoplasia. Endoscopic mucosal resection and endoscopic submucosal dissection are well established, safe, and effective for lesions without submucosal invasion.7,8 Endoscopic full-thickness resection was also used to treat T1 to T2 tumor for colon cancer.9 In some cases, endoscopic treatment alone is insufficient for disease treatment, and laparoscopic surgery is required. The combination of laparoscopic surgery and endoscopic resection will represent a new frontier in cancer treatment.10 Owing to the strong support from endoscopic center, patients diagnosed with early colon cancer were given preference to think about endoscopic resection or combination of laparoscopic surgery treatment first in our hospital. This may be the reason why no stage I colon cancer patients were enrolled in this study.
Generally, randomized assignment undergone before operation was a good alternative approach. In that case, when the patients were diagnosed with stage I or stage IV after randomization, then these patients would be analyzed in the intention-to-treat population and excluded from evaluable population.