Hi everyone,
For this Friday, we will have Dr. Elizabeth Rose Mayeda come speak to us about using simulation platform for quantifying survival bias in research on determinants of cognitive decline.
Please read the following abstract to prepare for her talk. Thank you.
Quantifying survival bias in research on determinants of cognitive decline with a simulation framework
Survival bias is a major concern for researchers studying determinants of cognitive decline because there is substantial evidence that cognitive decline predicts mortality. Biased estimation of the effect of an exposure on rate of cognitive decline can occur when mortality is influenced by both the exposure and cognitive decline or a determinant of cognitive decline. However, it is difficult to anticipate the degree of bias in empirical studies. We developed a flexible simulation platform that corresponds with assumptions of substantive researchers to quantify expected degree of bias in studies of determinants of cognitive decline. Using this simulation platform, we evaluate the potential survival bias in a longitudinal study examining the effect of a binary exposure on rate of cognitive decline when implementing naïve analyses under several selective survival scenarios. The degree of bias is strongly influenced by the baseline mortality rate. In high mortality scenarios, substantial bias is easily induced when mortality is influenced by the exposure and cognitive decline. However, when the mortality is influenced by the exposure and a determinant of cognitive decline, substantial bias only arises when a strong interaction on the hazard ratio scale between the exposure and the determinant of cognitive decline is present. This simulation platform provides a flexible platform for evaluating biases in studies of determinants of cognitive decline.