Abstract:
|
Early escape designs were developed as a modification of placebo-controlled clinical trials in order to reduce the length of time subjects remain on an assigned treatment under which they appear to be deteriorating, and instead receive additional rescue medication. Although early escape designs alleviate ethical concerns, they lead to complications with statistical analysis and interpretation. Comparisons between the groups assigned to different treatments are no longer interpretable as simple comparisons of received treatments because the treatment actually received may have changed from that assigned. Receipt of rescue medication is a post-treatment outcome that might also affect clinically relevant outcomes. We approach this problem with the principal stratification framework, in which subjects belong to latent groups based on post-treatment variables, here, receipt of rescue medication. For only one principal stratum can the estimands of causal effects of assignment be interpreted as the causal effect of receipt of active treatment versus placebo. We present the results from a placebo-controlled clinical trial of treatment for relapsing-remitting multiple sclerosis in order to
|
ASA Meetings Department
732 North Washington Street, Alexandria, VA 22314
(703) 684-1221 • meetings@amstat.org
Copyright © American Statistical Association.