UBC Theses and Dissertations
Intention to treat : a systematic review on recommendations and how to use it appropriately Kennedy, Dawn
Intention-to-Treat (ITT) has long been understood as a method to ensure data is analyzed according to the original assignment and hence to preserve the benefits of randomization in randomized controlled trials (RCTs). The two main concerns ITT deals with are non-compliance and missing outcome data. There are many strategies to deal with these issues, and the context in which they are employed is paramount. The purposes of this thesis are twofold: (1) to synthesize systematic review studies, which were published between 1990 and 2020, in order to understand the trend of ITT practices in RCTs; and (2) to investigate how ITT is defined and what ITT methods and practices are recommended in methodology articles from 2010 to 2020 using a systematic review. A total of 1,281 articles published 1990-2020 were identified in the first search. Then we separated them into two pools, systematic review papers (1990-2020) and methodology papers (2010-2020). A total of 30 articles for the synthesis of systematic reviews and 53 methodology articles for the systematic review were included in the final review. Both of our synthesis and systematic review studies suggest that a variety of definitions have appeared in the literature in addition to the traditional widely cited definition, “once randomized, always analyzed.” Modified ITT (mITT) has become a popular trend and attracted a lot of attention in the last decade, but there is not agreement in how to define mITT. Our studies have also identified a variety of statistical methods and techniques adopted and recommended for handling missing outcome data and non-compliance in the literature. Interestingly, last outcome carried forward was found to be the most popular method for missing outcome data in our synthesis study, though multiple imputation was highly recommended by methodologists. For dealing with non-compliance issues, maximum likelihood based Complier Average Causal Effect and Instrumental Variable methods were the most popular techniques in our synthesis study, which was also echoed in our systematic review study.
Item Citations and Data
Attribution-NonCommercial-NoDerivatives 4.0 International