Trend Estimation Using Summarized Dose-response Data

• Activity Number: 145
• Type: Contributed
• Date/Time: Monday, August 5, 2013 : 8:30 AM to 10:20 AM
• Sponsor: Section on Statistics in Epidemiology
• Author(s): Junshan Qiu*+
• Companies: FDA
• Keywords: Trend estimation ; Dose-response ; Summarized data

Abstract

Meta-analysis of trend estimation based upon summarized data has been widely used to obtain a robust estimate of dose-response relationship. In practice, exposure level for each subject or study unit is reported as an interval rather than as a single value. This raises a well-recognized statistical problem: How can variation of exposure levels be integrated into trend estimation in a statistically sound manner? Shi and Copas (2004) recommended using observed frequency of subjects in each exposure category, reported bounds of exposure intervals and even prior knowledge from historical studies to estimate the distribution of exposure. In the light of the method by Shi and Copas (2004), we developed a new strategy for trend estimation using summarized information from literature. The new method features estimating the distributions of both exposure and probability of being a case in each exposure category.