BackgroundThe prolonged manifestation of concurrent leukopenia, thrombocytopenia and normal C-reactive protein (CRP) (named as SFTS prediction score) in febrile diseases is not usual and may be used to make an initial differential diagnosis, which is a characteristic finding of severe fever with thrombocytopenia syndrome (SFTS).MethodsThe dynamics of SFTS prediction scores was investigated in SFTS patients. The study subjects for the comparison were febrile patients aged ≥ 16 years with SFTS scores of 2 (S2) or 3 (S3) who visited an emergency room for a 4-year study period. The dynamic distribution of S2 and S3 at presentation with regards to onset of illness, the characteristics of responsible diseases and the predictability of SFTS in both groups were described.ResultsIn 104 patients with SFTS, the daily proportion of S2 or S3 ranged from 58.3 to 100% from day (D) 1 to D12 after the onset of illness. The S2 subtype of ‘leukopenia plus thrombocytopenia’ and S3 represented 72.7–100% of all scores. In contrast, for the 130 patients in the febrile cohort, 73.8% of evaluations were distributed between D1 and D4 after the onset of illness, and 68.8% of patients had the S2 subtype of ‘leukopenia plus normal CRP’. Upper respiratory infection was the most frequent (50.0%) cause of diseases. Pneumonia (13.8%) and urosepsis (6.2%) initially presented with either S2 with normal CRP or S3 but had poor prognosis. The presence of S2 or S3 predicted SFTS with sensitivity and specificity of 0.85 (0.42–0.99; 95% CI) and 0.98 (0.98–0.98; 95% CI), respectively.ConclusionThe temporal distribution and composition of S2 or S3 were unique in several febrile diseases including SFTS, and the SFTS prediction score may be useful for differentiating febrile diseases in primary care settings of SFTS endemic areas.