Implementation of a Kronecker Product Correlation Structure for the Analysis of Live-Donor Kidney Transplant Rates

• Activity Number: 473
• Type: Contributed
• Date/Time: Wednesday, August 4, 2010 : 8:30 AM to 10:20 AM
• Sponsor: Health Policy Statistics Section
• Author(s): Arwin Thomasson*+ and Hanjoo Kim and Justine Shults and Russell Localio and Harold Feldman and Peter Reese
• Companies: University of Pennsylvania and Forrest Labs and University of Pennsylvania and University of Pennsylvania and University of Pennsylvania and University of Pennsylvania
• Keywords: renal transplant ; quasi-least squares ; Kronecker product ; correlated data

Abstract

Medical studies often yield data with multiple sources of correlation. For example, a quality-of-care measure applied to multiple transplant centers over multiple years could be correlated in two ways--the scores for each center could be correlated over time, and the center measurements could be correlated within regions. We present an approach for analysis of unbalanced multi-level correlated data that implements a Kronecker product (KP) structure with quasi-least squares (QLS). While previous researchers primarily implemented the KP structure for analysis of balanced data with a constant number of measurements per subject, our approach allows for consideration of unbalanced data. We implement our approach in an analysis of a standardized live-donor transplant rate that identifies characteristics of transplant centers that do not fully utilize live-donor kidney transplants.