1. Spatio-temporal changes of tick-borne encephalitis (TBE) incidence in the Baltic States: possible causes? D.Sumilo supervised by Prof S.E. Randolph Department of Zoology, University of Oxford, UK EDEN PhD meeting, 8th June, 2006, CIRAD, Paris Estonia Latvia Lithuania

2. EstoniaLatvia Lithuania UK Spain Italy German y Poland Romani a Hungar y Czech Republi c Slovaki a Sloveni a

3. Estonia Lithuania Poland Latvia Recorded annual cases of TBE Hungary Czech Republic Slovakia Slovenia Italy Germany no data Changes in the recorded annual cases of TBE over time 1985-90low 1993-98 high 1999-04 mod high Randolph SE (2001) Phil. Trans. Roy. Soc. Lond. B 356, 1045-56

4. 1993 - 1998 1985 - 1990 1999 - 2004 Spatial heterogeneity of TBE within the Baltics Data source: Official Public Health statistics, Estonia, Latvia, Lithuania EstoniaLatvia Lithuania

5. The driving factors?  Just an artefact – change in public health activities? General consensus: General consensus: possibly exaggerated, but nevertheless real increase  Biologicalabiotic - climatic & habitat conditions biotic - tick & host populations  Non-biologicalsocio-economic changes human behaviour changes increased human-tick contact rates

6. Spatial and temporal analysis  Spatial Regression analysis to identify abiotic predictors of TBE incidence at the ‘county’ level across all Baltic States

7. Variables for spatial analysis (1993-1998)  Abiotic - climatic and habitat conditions Surrogate climate variables (LST, NDVI, MIR) Surrogate climate variables (LST, NDVI, MIR) extracted from meteorological satellites Temporal Fourier analysis used to yield summary statistics Elevation - digital elevation model (DEM) Elevation - digital elevation model (DEM) Land cover – CORINE database Land cover – CORINE database Classified in according to suitability for the tick- habitat

8. Spatial analysis – linear regression Baltics (1993-1998) Dependent variable Independent variables Variables selected R 2 (n=85) Mean annual TBE incidence per ‘county’ ElevationElevation LST Fourier factorsLST Fourier factors NDVI Fourier factorsNDVI Fourier factors MIR Fourier factorsMIR Fourier factors Land cover typeLand cover type 3 LST variables3 LST variables 2 MIR variables2 MIR variables ElevationElevation 0.470

9. Spatial analysis – linear regression Baltics (1993-1998)  Interaction between tick habitat and human contact with that habitat  Precise seasonal patterns of vegetation and temperature are important Dependent variable Independent variables Variables selected R 2 (n=85) Mean annual TBE incidence per ‘county’ ElevationElevation LST Fourier factorsLST Fourier factors NDVI Fourier factorsNDVI Fourier factors MIR Fourier factorsMIR Fourier factors Marginal suitabilityMarginal suitability tick habitat 2 LST variables2 LST variables 2 MIR variables2 MIR variables 1 NDVI variable1 NDVI variable 0.555 Land cover typeLand cover type

10. Spatial analysis (1993 – 1998) Ranks to No statistical difference between observed and predicted TBE incidence (p>0.05) TBE incidence observed observed predicted

11. Tick suitability habitat map Tick habitat suitability: high (woodland/forests) marginal (green urban areas, pastures etc.) zero (built-up areas, water bodies, arable land etc.) Data source: CORINE Land Cover national databases Currently working with EDEN High Resolution RS team to integrate land- cover structure variables in the model

12. Conclusions from spatial analysis  Abiotic factors (temperature, vegetation index, land cover) are important  Account for 55% of variation in mean annual TBE incidence across the Baltic States  There appear to be other important factors e.g. hosts, public health, human factors

13. Temporal analysis  Have these predictors changed at the right time, place and direction to explain the upsurge in TBE incidence since 1993? Biologically most significant increase in temperature occurred in April Analysis of temperature data by 10-day periods from 1970 to 2004 Šiauliai Jelgava Gulbene Rezekne TBE relative increase (log scale) Liepaja Dagda Biržai Utena Klaipėda Kaunas Vilnius Jõgeva Jõhvi Stende Lääne-Nigula Pärnu Sõrve Tallinn-Harku Tartu-Tõravere Võru

14. April, last 10-day period Temporal changes in TBE and temperature LarvaeNymphs Temperature thresholds for tick acitvity

15. Changes in precipiation  Dryer summers during 1993-1998 might have encouraged more leisure activity outside

16. Changes in average I.ricinus nymph abundance (I.ricinus monitoring site, Riga’s county, Tireli) ticks per 1 km Data source: Latvian Public Health Agency ticks per 1 km

17. Changes in other factors Data source: National Statistical offices and Forest agencies in Latvia, Lithuania and Estonia % of area sown for field crops Roe deer per 1000 ha

18. Change in recorded TBE incidence over time Data source: Official Public Health statistics, Estonia, Latvia, Lithuania

19. TBE incidence and vaccination coverage by counties in Lithuania and Latvia Relative change (log scale) of mean TBE incidence 1993-98 to 2002-04 % of fully immunised people 2002-04 Data source: Latvian and Lithuanian official public health statistics 1.0 0.5 0.0 -0.5 0.0 5.0 10.015.0 20.0

20. Who goes to the forest? Data source: SKDS Research Centre, Latvia  Secondary analysis of cross-sectional survey data of 18-74 year olds in Latvia in 2001 Who are more likely to go? 69% (701/1022) had been to the forest at least once during tick- activity season <55 years of age With the lowest and the highest income Living in rural areas and small towns From those who go, who are more likely to go frequently? Males With the lowest income Who are more likely to be vaccinated? 35 – 54 years of age With higher income Those who go to the forest

21. Next steps  Further analyses of biotic factors, socio-economic conditions and human behaviour to explain the epidemiological patterns in the Baltics more fully Thank you!  To test the findings on the pan- European scale

22. Acknowledgments I would like to thank EDEN-TBD partners in the Baltic States: Antra Bormane, Public Health Agency (Latvia) Loreta Asokliene, Centre for Communicable Diseases Prevention and Control, (Lithuania) Veera Vasilenko, National Institute for Health Development and Health Protection Inspectorate (Estonia) for their help with providing data for this study