Prospective Pooling for Discrete Survival Outcome

• Activity Number: 38
• Type: Contributed
• Date/Time: Sunday, July 31, 2011 : 2:00 PM to 3:50 PM
• Sponsor: ENAR
• Author(s): Paramita Saha Chaudhuri*+ and Clarice Weinberg and David Umbach
• Companies: National Institute of Environmental Health Sciences and National Institute of Environmental Health Sciences and National Institute of Environmental Health Sciences
• Keywords: pooling ; discrete survival ; time to pregnancy

Abstract

Pooled exposure analysis has become a very useful technique especially when the exposure assay is expensive or limited volume of specimen is available for assaying. Saha and Weinberg (2010) extended the pooled exposure analysis for a discrete-time survival outcome such as time-to-pregnancy. Two limitations exist. First the pooled exposure cannot be reused for another disease. Subjects concordant for cancer may not be concordant for heart disease, hence pooling needs to be done afresh with each new disease studied. Moreover, the analysis assumes a logisitic model that may not hold in practice. We introduce a multiple imputation approach that uses a prospective pooling where subjects are grouped at the outset of the study without restricting the grouping within same outcome stratum addressing the first limitation. A flexible modeling approach is employed to account for a general risk model. We show that for a discrete-time survival outcome, this approach can be employed to test for the exposure effect on the time to outcome and can be used to estimate the exposure effect under certain conditions. We demonstrate this approach via extensive simulation studies and real data example.