Background: Tick-borne encephalitis (TBE) is a viral infection with no available treatment. Due to its non-specific symptoms, TBE tends to be under-diagnosed and under-reported. We aimed to identify factors predicting TBE diagnosis to develop a diagnostic algorithm for use by physicians.
Methods: We conducted a case-control study using data routinely collected in Poland during 2009-2010. We included patients admitted to hospitals, who were assigned an International Classification of Disease (ICD) code indicating aseptic meningo-encephalitis. Cases were confirmed by detection of specific IgG and IgM antibodies. Patients that tested negative for TBE were included as controls. We used logistic regression to determine associations and recursive partitioning to build a diagnostic algorithm based on 70% of the dataset, and validated the algorithm using the remaining 30%.
Results: Of 774 patients, 273 (35%) were TBE-positive. Cerebrospinal fluid protein levels and presence of a tick bite were key decision points in the algorithm, while living in a TBE endemic area was not important. Application of the algorithm to the validation dataset yielded a sensitivity of 89% and specificity of 37%.
Conclusions: TBE should be included in routine diagnostic protocols for all cases admitted to hospitals with meningitis or encephalitis. However, in resource-limited settings and in regions with unknown TBE endemicity status, our algorithm could indicate which cases should be tested for TBE.
Keywords: Poland; Tick-borne encephalitis; algorithm; diagnosis; recursive partitioning.