Science changes over time: do scientists? A bibliographic exploration of views on adherence adjustment among clinical trialists

Background: When some randomized trial participants do not adhere to assigned treatments, the intention-to-treat effect represents the effect of assigning participants to the study treatment, rather than of receiving the treatment. This latter effect is the per-protocol effect, and is often of more interest to patients. Advances in statistical and epidemiological methods allow us to validly estimate the per-protocol effect. However, many trialists and clinicians are unfamiliar with these new methods and are skeptical adherence adjustment in randomized trials. Much of this skepticism stems from an analysis of the Coronary Drug Project (CDP) placebo arm published in 1980. Objective: To assess the impact of the 1980 CDP analysis on the use and scientific opinion of adherence adjustment over the last 40 years, and to determine how opinion has changed over time.Methods: We obtained copies of all publications indexed in Google Scholar which cited the 1980 CDP paper. Articles were reviewed to determine why the authors cited the CDP paper, and categorized by theme. To assess updating of beliefs, we also obtained copies of all papers citing two recent re-analyses of the CDP data and assessed citation themes over time.Results: The 1980 CDP placebo arm analysis continues to be used as evidence against adherence adjustment, with little change in citation practices over time. The impact of re-analyses on researchers’ beliefs appears to be minimal. Conclusion: The impact of the 1980 CDP paper continues to be strong, despite recent methodological advances. Changing