This editorial policy1 was formulated after discussion with key researchers in observational study design and implementation, including discussion at panels of meetings of the Multicenter Perioperative Outcomes Group in 2013 and 2015. There are two reasons why we chose not to require prior public registration of observational studies as we do for randomized clinical trials. First, many observational studies utilize publicly available data, and there was concern that publication of a study, analysis plan, and identification of the database could result in others with more resources taking these ideas, performing the analysis, and publishing the results before the original investigator. This is less of a concern with clinical trials, which often require many months to even initiate. Second, clinical trial registration was formed in large part to avoid underreporting of negative results, especially trials supported by industry. Multicenter Perioperative Outcomes Group investigators and editors of the Journal felt that underreporting of negative results in observational studies was likely a minor problem. Rather, lack of a priori definition of an analysis plan would more likely yield positive results due to the multiplicity of analyses that could be performed.
Dr. Eisenach has received consultant fees from Teva Pharmaceuticals and Adynxx and has received salary support from the American Society of Anesthesiologists as Editor-in-Chief of Anesthesiology.