|
Activity Number:
|
23
- Statistical Considerations for Epidemiologic Studies of Radiation Risk
|
|
Type:
|
Topic Contributed
|
|
Date/Time:
|
Sunday, August 7, 2022 : 2:00 PM to 3:50 PM
|
|
Sponsor:
|
Section on Statistics in Epidemiology
|
|
Abstract #323424
|
|
|
Title:
|
Excess Relative Risk and Excess Absolute Rate Models in (Radiation) Dose-Response Modeling
|
|
Author(s):
|
Dale L. Preston* and Daniel O. Stram
|
|
Companies:
|
Hirosoft International, Eureka, CA and Keck School of Medicine, University of Southern California
|
|
Keywords:
|
dose-response;
risk modeling;
radiation;
excess relative risk;
rate differences
|
|
Abstract:
|
Concerns in the analyses of dose effects on disease rates include: the dose response shape; effect modification; and the nature of risk factor interactions. Description of rates in terms of rate ratios (relative risks) and rate differences (excess absolute rates) are both important. These issues have prompted the development of rate models and Poisson-regression-based estimation methods that are better suited to the characterization of dose response effects on disease rates than the commonly used partial-likelihood methods and loglinear proportional hazards (Cox regression) models . The models include additive and multiplicative excess relative risk models like T_0 [ 1 + ERR_1(d) + ERR_2 + …] and T_0 [1 + ERR_1(d)] [1 + ERR_2] …, respectively and excess rate models like T_0+ EAR_1(d) + EAR_2 + … in which the excess terms are described using products of linear and log-linear functions such as ExR(d, age, b) = (b_1 d + b_2 d^2) exp[b_3 log(age)] We use examples from analyses of the atomic bomb survivors and the Russian Mayak and Techa River cohorts to describe some of these models, to discuss challenges in their use and to note where further work is needed.
|
Authors who are presenting talks have a * after their name.
