Differential diagnosis and prospective grading of COVID-19 at the early stage with simple hematological and biochemical variables

Song, Lin; Liang, En-Yu; Wang, Hong-Mei; Shen, Yan; Kang, Chun-Min; Xiong, Yu-Juan; He, Min; Fu, Wen-Jin; Ke, Pei-Feng; Huang, Xian-Zhang

We evaluated simple laboratory variables to discriminate COVID-19 from bacterial pneumonia or influenza and for the prospective grading of COVID-19. Multivariate logistic regression and receiver operating characteristic curve (ROC) were used to estimate the diagnostic performance of the significant discriminating variables. A comparative analysis was performed with different severity. The leukocytosis (P = 0.017) and eosinopenia (P = 0.001) were discriminating variables between COVID-19 and bacterial pneumonia with AUC of 0.778 and 0.825. Monocytosis (P = 0.003), the decreased lymphocyte-to-monocyte ratio (LMR) (P < 0.001), and the increased neutrophil-to-lymphocyte ratio (NLR) (P = 0.028) were predictive of influenza with AUC of 0.723, 0.895 and 0.783, respectively. Serum amyloid protein (SAA), lactate dehydrogenase (LDH), CD3(+) cells, and the fibrinogen degradation products (FDP) had a good correlation with the severity of COVID-19 graded by age (≥50) and NLR (≥3.13). Simple laboratory variables are helpful for rapid diagnosis on admission and hierarchical management of COVID-19 patients.