Comparing national and global data collection systems for reporting, outbreaks of H5N1 HPAI

Determining if outbreak data collected by regional or international organizations can reflect patterns observed in more detailed data collected by national veterinary services is a necessary first step if global databases are to be used for making inference about determinants of disease maintenance and spread and for emergency planning and response. We compared two data sources that capture spatial and temporal information about H5N1 highly pathogenic avian influenza outbreaks reported since 2004 in four countries: Bangladesh, Egypt, Turkey, and Vietnam. One data source consisted of reports collected as part of each country's national veterinary services surveillance program, while the other data source included reports collected using the Emergency Prevention System for Priority Animal and Plant Pests and Diseases (EMPRES-i) global animal health information system. We computed Spearman rank-order correlation statistics to compare spatial and temporal outbreak distributions, and applied a space-time permutation test to check for consistency between the two data sources. Although EMPRES-i typically captured fewer outbreaks than detailed national reporting data, the overall similarity in space and time, particularly after 2006, reflect the ability of the EMPRES-i system to portray disease patterns comparable to those observed in national data sets. Specifically, we show that the two datasets exhibit higher positive correlations in outbreak timing and reported locations after 2006 when compared to December 2003 through 2006. Strengthening the capacity of global systems to acquire data from national and regional databases will improve global analysis efforts and increase the ability of such systems to rapidly alert countries and the international community of potential disease threats.