Identifying the transmission route of influenza A(H1N1) 2009 in Japan by geographic, effective and phylogenetic distances

Abstract

Background: A large outbreak of influenza A(H1N1) was observed in 2009 and a huge number of people was infected and died during this pandemic event. The natural hosts of influenza virus are wild birds, however, the infection spreads mainly by human-to-human transmission once the viruses are adopted to humans. To develop health countermeasures, it is therefore important to identify how the viruses spread during the outbreak. In this talk, the transmission route of influenza A(H1N1) in 2009 in Japan was estimated based on three different information, namely geographic distance, domestic transportation network and genome information. Methods & Materials: H1N1 viruses were imported to 5 prefectures from abroad and 9 domestic importation cases were observed from the end of May the middle of June in Japan. The corresponding genome sequences were taken from GenBank and the geographic data and the human travel data used were from Geospatial Information Authority of Japan and Ministry of Land, Infrastructure, Transport and Tourism, respectively. The mutation of influenza virus is quite fast and, firstly, we calculated the differences among genome sequences of detected viruses and the phylogenetic distances between each pair of the viruses were measured. Recently, it was shown that there is a strong correlation between the effective distance and arrival time of viruses/infection. Applying this idea to domestic human travel data, the frequencies of human migration from one city to another were computed. Geographic distance plays an important role in the spread of infectious diseases and the distances among prefectures were also computed. Three distances were described by likelihood function and the parameter values were estimated. Results: All pairs of transmission network were evaluated by constructed likelihood functions using the influenza importation cases from abroad and the domestic transmission case in each prefecture. The estimated transmission trees were compared by AIC and the most possible transmission route was identified. Conclusion: Recent development of big data analysis gives us new insights in modelling the transmission route identification. Our approach using non-epidemiological data gives us more plausible results than epidemiological data alone.