Extending the Capture-Recapture Methodology To Estimate Disease Subpopulation Sizes in the Presence of Cross-Classification

• Activity Number: 238
• Type: Topic Contributed
• Date/Time: Tuesday, August 8, 2006 : 8:30 AM to 10:20 AM
• Sponsor: Section on Health Policy Statistics
• Author(s): Ulysses Diva*+ and Dipak Dey and Timothy Morse
• Companies: University of Connecticut and University of Connecticut and University of Connecticut Health Center
• Keywords: Bayesian modeling ; capture-recapture ; maximum likelihood approaches ; MSD in Connecticut ; population and subpopulation sizes

Abstract

This paper attempts to extend the capture-recapture methodology to answer questions about the sizes of subpopulations, particularly in situations where there is cross-classification. Three methods of estimating the sizes of subpopulations based on the multinomial likelihood are presented. The first is a conditional maximum likelihood estimator, while the second is an integrated maximum likelihood estimator. The third is a hierarchical Bayesian implementation, which parameterized the table probabilities to implement the independence assumption. The methods were evaluated using simulations and applied to the data on work-related musculoskeletal disorders in Connecticut. We estimated about 149,900 (se=6226) work-related MSD cases in CT from 1995 to 2001, with about 41,600 (se=1781) Carpal Tunnel Syndrome and 109,900 (se=4554) cases having other types of MSD.