Yang Bai1,2, Enxin Wang3, Shoujie Zhao4, Jing Li5, Yejing Zhu4, Yongchao Zhang3, Liang Cao6, Haitao Liu7, Yushu Dong2, Fang Wang8,9, Guobiao Liang2, Lei Liu1. 1. Department of Gastroenterology, Tangdu Hospital of Fourth Military Medical University, Xi'an, China. 2. Department of Neurosurgery, General Hospital of Northern Theater Command, Shenyang, China. 3. Department of Medical Affairs, Air Force Hospital of Western Theater Command, Chengdu, China. 4. Department of General Surgery, Tangdu Hospital of Fourth Military Medical University, Xi'an, China. 5. Department of Digestive Disease, Shanxi Bethuen Hospital of Shanxi Academy of Medical Sciences, Taiyuan, China. 6. Department of Traditional Chinese Medicine, Tangdu Hospital of Fourth Military Medical University, Xi'an, China. 7. Department of Cardiology, Xijing Hospital of Fourth Military Medical University, Xi'an, China. 8. Department of Dermatology, The First Affiliated Hospital of Sun Yat-sen University, Guangzhou, China. 9. Division of Dermatology, Department of Medicine, Washington University School of Medicine, St. Louis, MO, United States.
Abstract
Background: Although laboratory tests have become an indispensable part in clinical practice, its application in severity classification and death risk stratification of COVID-19 remains unvalidated. This study aims to explore the significance of laboratory tests in the management of COVID-19. Methods: In 3,342 hospitalized patients with COVID-19, those of mild or moderate subtype were categorized into the non-severe group, while those of severe or critical subtype were categorized into the severe group. Initial laboratory data were analyzed and compared according to disease severity and outcome. Diagnostic models for the severe group were generated on risk factors identified by logistic regression and receiver operating characteristic (ROC) analyses. Cox regression and ROC analyses on risk factors were utilized to construct prognostic models. Results: In identification of patients in the severe group, while age, neutrophil-to-lymphocyte ratio, and α-hydroxybutyrate dehydrogenase were identified as independent predictors, the value of combination of them appears modest [area under the curve (AUC) = 0.694]. Further ROC analyses indicated that among patients in the severe group, laboratory indices had a favorable value in identifying patients of critical subtype rather than severe subtype. For death outcome, IL-6, co-existing cerebrovascular disease, prothrombin time activity, and urea nitrogen were independent risk factors. An IL-6 single-parameter model was finalized for distinguishing between fatal and recovered individuals (AUC = 0.953). Finally, a modified death risk stratification strategy based on clinical severity and IL-6 levels enables more identification of non-survivors in patients with non-critical disease. Conclusions: Laboratory screening provides a useful tool for COVID-19 management in identifying patients with critical condition and stratifying risk levels of death.
Background: Although laboratory tests have become an indispensable part in clinical practice, its application in severity classification and death risk stratification of COVID-19 remains unvalidated. This study aims to explore the significance of laboratory tests in the management of COVID-19. Methods: In 3,342 hospitalized patients with COVID-19, those of mild or moderate subtype were categorized into the non-severe group, while those of severe or critical subtype were categorized into the severe group. Initial laboratory data were analyzed and compared according to disease severity and outcome. Diagnostic models for the severe group were generated on risk factors identified by logistic regression and receiver operating characteristic (ROC) analyses. Cox regression and ROC analyses on risk factors were utilized to construct prognostic models. Results: In identification of patients in the severe group, while age, neutrophil-to-lymphocyte ratio, and α-hydroxybutyrate dehydrogenase were identified as independent predictors, the value of combination of them appears modest [area under the curve (AUC) = 0.694]. Further ROC analyses indicated that among patients in the severe group, laboratory indices had a favorable value in identifying patients of critical subtype rather than severe subtype. For death outcome, IL-6, co-existing cerebrovascular disease, prothrombin time activity, and ureanitrogen were independent risk factors. An IL-6 single-parameter model was finalized for distinguishing between fatal and recovered individuals (AUC = 0.953). Finally, a modified death risk stratification strategy based on clinical severity and IL-6 levels enables more identification of non-survivors in patients with non-critical disease. Conclusions: Laboratory screening provides a useful tool for COVID-19 management in identifying patients with critical condition and stratifying risk levels of death.