Evaluation of Treatment Effect on Gap Time

• Activity Number: 559
• Type: Topic Contributed
• Date/Time: Wednesday, August 3, 2011 : 2:00 PM to 3:50 PM
• Sponsor: Section on Government Statistics
• Author(s): Bruce Chow and Julia Y. Lin*+ and Mei-Chiung Shih and Ying Lu
• Companies: VA Cooperative Studies Program Palo Alto Coordinating Center and Palo Alto Coordinating Center and Veterans Administration and Palo Alto VA CSP Coordinating Center/Stanford
• Keywords: Gap Time ; Clinical Trials ; Survival Analysis ; Causal Inference ; Informative Censoring ; Selection Bias

Abstract

Times of sequential event are observed in some clinical trials, and the time from the first to the second event is the gap time. In a randomized control trial of smoking cessation treatments, 943 smokers with military-related PTSD were randomized to smoking cessation treatment integrated with mental health care (integrated care; IC) or the usual care of referral to smoking cessation clinics (SCC), and they were followed up for 18 to 48 months (McFall et al. 2010). We are interested in estimating the causal effect of the IC on time to quitting smoking, and then the causal effect of the IC on the gap time of quitting to relapse, compared to the SCC. The treatment effect on relapse suffers from informative censoring and selection bias due to its conditional nature on the occurrence of quitting. We utilize a principal stratification framework (Frangakis & Rubin 1999, 2002) that precludes the confounding when stratifying on observed post-randomization behavior (i.e., observed quitting).