Scoring Model for Predicting the Occurrence of Severe Illness in Hospitalized Patients with Severe Fever with Thrombocytopenia Syndrome.

Xuemin Wei,Lirui Tu,Ling Qiu,Mengting Chen,Yao Wang,Mengyu Du,Haopeng Kan,Qing Dong,Xiaoying Xu,Haowen Yuan,Li Zhao,Hongling Wen
DOI: https://doi.org/10.7883/yoken.jjid.2021.716
2022-01-01
Japanese Journal of Infectious Diseases
Abstract:Severe fever with thrombocytopenia syndrome (SFTS) is an emerging hemorrhagic fever with high mortality. Severe cases progressed rapidly, with deaths occurring within 2 weeks. Therefore, constructing a model to predict disease progression among hospitalized patients plays an important role in clinical practice. The development cohort included 121 patients with SFTS, 25 with severe SFTS, and 96 with mild SFTS. Two of the 64 variables were independent risk factors, including neurological symptoms (odds ratio [OR], 12.915; 95% confidence interval [CI], 3.342-49.916; P < 0.001) and aspartate aminotransferase/alanine aminotransferase levels (OR, 1.891; 95% CI, 1.272-2.813; P = 0.002). The model's area under the curve (AUC) was 0.882 (95% CI: 0.808-0.956). The mean AUC value obtained from the internal validation was 0.883 (95% CI: 0.809-0.957). The AUC in the external validation cohort was 0.873 (95% CI: 0.775-0.972). This model can be used to identify severely ill patients as early as possible with high predictive value, stability, and repeatability. This model can help clinicians with their treatment plans.
What problem does this paper attempt to address?