Statistics GIDP Oral Defense
Missing Data in Non-Inferiority Clinical Trials
In clinical trials, patient dropout and non-compliance with treatment protocol may complicate analysis. When the data planned for collection are compromised or incomplete, estimates for treatment effect are often biased and trial conclusions may not be generalizable. Non-inferiority trials aim to show an experimental treat-ment is therapeutically no worse than existing treatments. If the new treatment is preferred for reasons such as cost, convenience, or safety, a non-inferiority design may be appropriate for investigating whether the treatment is as effective as the standard of care within a pre-defined margin. Non-inferiority trials are by nature less conservative than superiority and placebo-controlled studies, and many of the chal-lenges in their analysis and interpretation are exacerbated by missing data. I present results from a systematic review of non-inferiority trials demonstrating current prac-tices for handling missing data. Next, I use Monte Carlo simulations to compare some common, ad-hoc imputation approaches with statistically principled modeling methods under various missing data mechanisms. Finally, I develop a sensitivity analysis for missing data assumptions using a pattern-mixture model approach im-plemented in a multiple imputation framework. Given the increasing popularity of the non-inferiority design, the persistent challenge of missing data and patient com-pliance, and the reliance of regulators and clinicians on trial results, there is a critical need for robust analyses. Better practices have potential for patients’ easier access to new treatments and for minimizing risk of exposure to treatments that are in-effective.