Computing the Standard Error of the Relative Risk of Dementia versus Death in a Multistate Nonstationary Markov Chain

• Activity Number: 257
• Type: Contributed
• Date/Time: Monday, August 2, 2010 : 2:00 PM to 3:50 PM
• Sponsor: Section on Statistics in Epidemiology
• Author(s): Lei Yu and Richard J. Kryscio*+
• Companies: Rush Alzheimer's Disease Center and University of Kentucky
• Keywords: relative risk ; multistate Markov chains ; categorical longitudinal data

Abstract

We investigate the problem of determining the standard error associated with the relative risk (RR) of absorption between two competing absorbing states in the context of modeling transitions from cognitive intact to dementia with death treated as a competing event. A submatrix with rows depending on the initial state contain the probabilities of eventual absorption into these states and the RR is defined as the ratio of entries in these rows. The submatrix is defined by an infinite matrix series in a nonstationary discrete time multi-state Markov model. We use simulations to compare a bootstrap resampling technique with a first order Taylor approximation for estimating the standard error. Results are illustrated with an application to the Nun Study, a cohort of 424 nuns each followed from one of three baseline nondemented cognitive states to dementia or death whichever came first.