ORIGINAL PAPER Spatiotemporal statistical analysis of influenza mortality risk in the State of California during the period 1997–2001 Kyung-Mee Choi Æ Hwa-Lung Yu Æ Mark L. Wilson Published online: 7 July 2007 Ó Springer-Verlag 2007 Abstract Using the Bayesian maximum entropy (BME) method of spatiotemporal statistics, the present study examines the geographical risk pattern of influenza mor- tality in the state of California during the time period 1997–2001. BME risk analysis is of considerable value, since influenza is the largest contributing factor to wintertime mortality increases in the US. By incorporating age-adjusted mortality data collected at the county level, informative influenza mortality maps were generated and composite space-time influenza dependences were assessed quantitatively. On the basis this analysis, essential risk patterns and correlations were detected across the state during wintertime. It was found that significantly high risks initially occurred during December in the west-central part of the state; in the following two weeks the risk distribution extended in the south and east-central parts of the state; in late February significant influenza mortalities were detected mainly in the west-central part of the state. These findings, combined with the results of earlier works, can lead to useful conclusions regarding influenza risk assess- ment in a space-time context and, also, point toward promising future research directions. Keywords Influenza  Mapping  Mortality  Risk  Spatiotemporal statistics  BME  California 1 Introduction Influenza epidemics typically occur during wintertime and have been responsible for an average of 36,000 deaths per year in the US during the period 1990–1999 (Harper et al. 2004). On a global scale, the World Health Organization (WHO) has suggested that there is currently a diminishing window of opportunity to stop the first massive outbreak of influenza in the twenty-first century. Influenza A H1N1, A H3N2 and B viruses have been in global circulation since 1977. Most studies suggest that H3 changes more rapidly than H1, while influenza B evolves more slowly than either A subtype (Ferguson et al. 2003). Influenza epidemics continue to occur regularly with serious effects on the population health and the national economies, and an accurate quantitative assessment of these influenza effects is an important issue. Evaluating influenza effects in terms of the population mortality distribution across space and time is an important matter (Lui and Kendal 1987). There has been evidence in the past that geographical space is, indeed, relevant to the pattern of influenza spread, and that the disease propagates in the form of spatial waves originated at epidemic centers (Cliff et al. 1986; Cliff and Haggett 1988; Cliff 1995; Haggett 2000). There is a disagreement, however, on whether the local dynamics of an epidemic characterized by its infection force or basic reproduction number depend on population density heterogeneity (May and Anderson K.-M. Choi Department of Environmental and Occupational Health, University of North Texas Health Science Center, 3500 Camp Bowie Blvd., Fort Worth, TX 76107-2699, USA H.-L. Yu (&) Department of Bioenvironmental Systems Engineering, National Taiwan University, 1, Sec 4, Roosevelt Rd, Taipei 10617, Taiwan e-mail: hlyu@mail.sdsu.edu M. L. Wilson Department of Epidemiology, School of Public Health, University of Michigan, 109 Observatory St., Ann Arbor, MI 48109-2029, USA 123 Stoch Environ Res Risk Assess (2008) 22 (Suppl 1):S15–S25 DOI 10.1007/s00477-007-0168-4