Multivariate Disease Mapping (MVDM)

• Activity Number: 471
• Type: Contributed
• Date/Time: Wednesday, August 5, 2009 : 10:30 AM to 12:20 PM
• Sponsor: Section on Statistical Graphics
• Author(s): Caroline Jeffery and Al Ozonoff*+ and Marcello Pagano
• Companies: Harvard School of Public Health and Boston University School of Public Health and Harvard School of Public Health
• Keywords: multivariate ; disease mapping ; surveillance

Abstract

BACKGROUND: MVDM combines several diseases mapping of the same region A, in order to gain information from the joint spatial distribution of n distinct diseases. METHODS: Using electronic health data in New England, we construct risk maps for 2 syndrome groups, respiratory (RESP) and gastro-intestinal (GI), at 2 study periods separated by 1 year in time, for a total of 4 maps r1, r2, g1 and g2. We extend our one-disease mapping approach to a joint mapping F and estimate its 4x4 correlation matrix S. RESULTS: Coefficients from S ranged from 0.47 to 0.58, except corr(R1,R2) = 0.86 (95% CI 0.84-0.87). A map of Mahalanobis-adjusted joint disease scores show areas of distinguished spatial heterogeneity. CONCLUSION: MVDM is an exploratory method for analysis of multiple spatial data streams. Results suggest that RESP group demonstrates more persistent spatio-temporal patterns than GI group.