Activity Number:
|
76
- Contributed Poster Presentations: Section on Statistics in Epidemiology
|
Type:
|
Contributed
|
Date/Time:
|
Monday, August 3, 2020 : 10:00 AM to 2:00 PM
|
Sponsor:
|
Section on Statistics in Epidemiology
|
Abstract #313800
|
|
Title:
|
A Multiplicative Weighted Cox Model for Relative Mortality Analysis
|
Author(s):
|
John Michael Williamson* and Hung-Mo Lin and Hae-Young Kim
|
Companies:
|
Centers for Disease Control and Prevention and Icahn School of Medicine at Mount Sinai and New York Medical College
|
Keywords:
|
Cox proportional hazards;
relative mortality;
excess survival
|
Abstract:
|
Relative survival methods are used to compare the survival experience of a cohort with that of the background population. We propose a multiplicative model for the underlying hazard to the population hazard with a "weighted" Cox-type model where we weight each individual's contribution to the usual Cox score equation, and not the risk set itself. The covariate effects follow the proportional hazards assumption. The weight adjusts for expected mortality in the population and is the inverse of the instantaneous probability that a demographically-similar individual in the population "fails", given that an individual in the population "fails." The population weights are estimated as the ratio between the population mortality of a group of demographically-similar individuals versus the population mortality of a specified baseline group. The proposed method does not require estimation the underlying hazard for the reference group in the cohort of interest. We present simulation studies with log-logistic distributed data to detail the performance of our approach, as well as an illustrative analysis of colon cancer data from a Finnish study.
|
Authors who are presenting talks have a * after their name.