Online Program

How to Generalize or Transport a Clinical Trial Finding to a Target Population of Interest? (305291)

Manisha Desai, Stanford University 
Kris Kapphahn, Stanford University 
Albee Ling, Stanford University  
*Maria Montez Rath, Stanford University  

Keywords: generalizability, missing data, propensity score methods,

Randomized clinical trials (RCTs) are the gold standard for evaluating treatment effects with high internal validity. However, the estimated treatment effects may not be generalizable to target populations of interest (Cole & Stuart, 2010; Goldstein et al., 2019; Lesko et al., 2017; Najafzadeh et al., 2018; Olsen et al., 2013). There are significant costs, both ethical and financial, associated with initiating new clinical trials for each target population of interest. Making use of existing data from trials can be enormously valuable. With the increasing importance of practicing evidence-based medicine, there is a critical need to generalize trial findings to relevant target populations. Recent research has led to promising methods to generalize trial findings to target populations (Buchanan et al., 2018; Cole & Stuart, 2010; Dahabreh et al., 2018; Hartman et al., 2015; Kern et al., 2016; Rudolph & van der Laan, 2017; Stuart et al., 2011; Tipton, 2013; Westreich et al., 2010, 2017). While there are sound methods available, practical challenges exist. We aim to provide guidelines to use these methods to generalize specific trial findings to populations of interest.