Fig 2. For the early years (2001–2009), an autumn decrease of CNS and myelitis was detected.

For the later years (2010–2018), however, there is a steep increase in CNS and myelitis occurrence in autumn. Predicted probabilities and corresponding 95% confidence intervals (shaded area) based on the results of logistic regression models with the time point in year at which TBEV infection related symptoms were detected included as a continuous but non-linear covariate. Non-linearity was modelled using restricted cubic splines with knot locations based on Harrell's recommended percentiles [14].