[oldalum]

BC is correct about potential reasons for well-intentioned and executed mid-course changes, but it should be realized that there is a history in medical research of such changes being made improperly, compromising the integrity of the study. In general they should be infrequent, justified by changes in scientific knowledge outside of the trial, independent of knowledge or analysis of interim trial data, determined by an independent data monitoring committee, and clearly and publicly specified. See, e.g., article. There are good reasons why the data to be collected and a detailed analysis plan are required to be specified before a trial starts. If the changes are needed to salvage the chance to produce generalizable knowledge from a trial, pitfalls must be avoided and the changes and their justification should be clearly published.

[oldalum]

BC is correct about potential reasons for well-intentioned and executed mid-course changes, but it should be realized that there is a history in medical research of such changes being made improperly, compromising the integrity of the study. In general they should be infrequent, justified by changes in scientific knowledge outside of the trial, independent of knowledge or analysis of interim trial data, determined by an independent data monitoring committee, and clearly and publicly specified. See, e.g., article. There are good reasons why the data to be collected and a detailed analysis plan are required to be specified before a trial starts. If the changes are needed to salvage the chance to produce generalizable knowledge from a trial, pitfalls must be avoided and the changes and their justification should be clearly published.