Lost in Translation: Missing Data Theory and Practice

• Activity Number: 395
• Type: Contributed
• Date/Time: Tuesday, August 5, 2014 : 2:00 PM to 3:50 PM
• Sponsor: WNAR
• Author(s): Melanie Bell*+ and Paul Hsu and Mallorie Fiero
• Companies: University of Arizona and University of Arizona and University of Arizona
• Keywords: missing data ; intention-to-treat ; randomized controlled trials ; sensitivity analyses

Abstract

Missing data in randomized controlled trials threatens the validity of the trial results, and makes an intention to treat (ITT) analysis difficult. Any analysis requires assumptions, but when data are incomplete the assumptions may be strong and unverifiable. In these situations, sensitivity analyses, which examine the robustness of results to assumptions, are essential. Our study reviews all the randomized controlled trials published in the latter half of 2013 in the top 4 medical journals to assess the handling of missing data; the labeling of the authors as to whether the analysis was ITT; whether this labeling was correct; and whether a sensitivity analysis was performed. We compare our results with previous reviews to ascertain whether there is evidence of change over time.