Spatial statistical methods to study the spread of Tularemia
Independent thesis Advanced level (professional degree), 20 credits / 30 HE creditsStudent thesis
Tularemia is a notifiable disease caused by the bacterium Francisella tularensis. Mosquito bites are believed to be to the most common way of human contamination in Sweden. Other ways of contracting the disease can be through for example ingestion of contaminated food or water and exposure to infected animals. The incubation period is 2-10 days and the symptoms vary depending on the infectious route. Common symptoms includes high fever, headache and nausea. The bacterium is highly virulent and has been ranked as a possible bioterrorism agent by the World Health Organization (WHO). In order to prevent future outbreaks more knowledge about the disease is needed. The purpose of this thesis is to answer questions about the geographical distribution of the disease. To our help we have two datasets with tularemia cases in Sweden. One of the datasets contains cases from whole Sweden during the period 1984 -2012 while the other is restricted to cases near Örebro from 2000 to 2010. The early cases of tularemia occurred almost exclusively in the northern part of Sweden. Today cases are reported from various places all over Sweden which suggests that the disease has moved south. By dividing the country into subareas and considering observations from different time periods we were able to statistically prove that the disease had moved south during the study period. The bacterium Francisella tularensis has besides subspecies also been separated further into genetic clades or subtypes. Information about clades involved in clinical cases was given for the observations from the Örebro region. By dividing the study area into subareas with help of a new method that uses hierarchical clustering, we were able to test for differences in distribution between these subtypes. The results showed no significant difference in distribution between the subtypes. An association study was conducted to see if tularemia cases could be linked to water. A comparison between mean distance to water for observed and simulated cases showed a clear connection to water. Test were also performed to see if it was possible to connect the different subtypes to specific streams in the Örebro area. The result showed no significant connection. With help of scan statistics it was possible to localize high risk areas at different levels. The size of the areas together with the choice of parameters had big impact on the result.
Place, publisher, year, edition, pages
Probability Theory and Statistics
IdentifiersURN: urn:nbn:se:umu:diva-88432OAI: oai:DiVA.org:umu-88432DiVA: diva2:715677
Master of Science Programme in Engineering Physics
Sjöstedt de Luna, Sara