Randomised controlled trials (RCTs)--essentials.

Correspondence to: Y H Chan Tel: (65) 6317 2121 Fax: (65) 6317 2122 Email: chanyh@ cteru.gov.sg The prospective, randomised, controlled, blinded (if possible), sample-size calculated study with a preplanned statistical analysis and trial monitoring is undoubtedly the gold standard for a therapeutic or interventional clinical study. RCTs carried out recently have discredited some treatments deemed effective by observational studies include (to mention a couple), hormone replacement therapy to prevent coronary heart disease events(1) and arthroscopic surgery for osteoarthritis of the knee(2). The top-down research process in the conduct of an RCT could be broadly categorised into three stages (see Table I). Each stage of the process must be carefully planned and carried out. A poorly designed, poorly conducted and poorly reported trial is a violation to the rights of the subjects who gave consent to participate in a study; this is not ethical (ICH E6, Guidelines on Good Clinical Practice(3)). Thus it is not only the sole responsibility of the principal investigator of the trial but all the researchers/collaborators (including patients!) involved in the trial to make sure all the stages are being carried out with the highest standards. The main objective of a clinical trial is to determine the differences (if any) between groups in outcomes of interest. However, these differences could be due to bias (imbalances between groups unrelated to treatment but related to outcome) or to chance alone (random error). In this article, we shall discuss the techniques (in general, for each stage of the research process) to limit the error of wrongly detecting a non-existent difference.