Simple Estimator for a Shared Frailty Regression Model

Jason P. Fine
Departments of Statistics and Biostatistics and Medical Informatics
University of Wisconsin

4:00-5:00 pm
Thursday, February 19, 2004
208 Patterson Hall, NCSU Campus

We propose a simple estimation procedure for a proportional hazards frailty regression model for clustered censored survival data in which the dependence is generated by a positive stable distribution. Inferences for the frailty parameter can be obtained with output from Cox regression analyses in standard software. The computational burden is substantially less than that for the other approaches to estimation, for example, maximum likelihood. The large sample properties of the estimator are sketched and simulations show that the approximations are appropriate for use with realistic sample sizes. The methods are motivated by genetic epidemiology studies of familial associations in the natural history of diseases. Their practical utility is illustrated in a correlation analysis of stroke and diabetes incidence in sibpairs from Beaver Dam, Wisconsin.


Return to Biostatistics Working Group