Abstract:
|
In the presence of missing data in clinical trials, sensitivity analyses should explore the potential impact of the missingness on the reliability of key results. One approach, coined a tipping point analysis in recent literature, is to vary assumptions about the missing outcomes in the experimental and control arms in order to identify scenarios under which there is no longer evidence of a treatment effect. Then, the plausibility of those assumptions can be discussed. Previous tipping point approaches (e.g., Yan et al., 2009; Liublinska and Rubin, 2014) relied on single or multiple imputation of outcomes in patients with missing data. We introduce a simple, alternative tipping point approach in which inference on the treatment effect is based on the observed data and the sensitivity parameters, with minimal assumptions and no need for imputation. The sensitivity parameters to be varied are the mean differences between outcomes in dropouts and outcomes in completers on each of the two treatment arms. We derive the asymptotic properties of the proposed statistic and illustrate the potential utility of this approach in the regulatory setting.
|
ASA Meetings Department
732 North Washington Street, Alexandria, VA 22314
(703) 684-1221 • meetings@amstat.org
Copyright © American Statistical Association.