Clinical and Laboratory Predictors of In-Hospital Mortality in 305 Patients with COVID-19: A Cohort Study in Wuhan, China
17 Pages Posted: 5 Mar 2020More...
Background: COVID-19 has caused a large number of deaths in a short period and lacks a specific treatment. Early identification of poor prognosis patients may facilitate doctors to offer proper supportive treatment in advance. This study aimed to develop death prediction models for COVID-19 patients.
Methods: In this cohort study, participants were COVID-19 patients who had been hospitalized in the First People’s Hospi predictive models tal of Jiangxia District in Wuhan from January 7, 2020 to February 11, 2020. Clinical characteristics and laboratory data of patients were collected. We selected the baseline variables at admission through ensemble XGBoost model, stepwise Akaike information criterion (AIC) and clinical significance. Two predictive models of mortality were built, one with clinical characteristics and the other one further included laboratory data.
Findings: Before February 12, 2020, 305 COVID-19 patients were enrolled, of whom 22 (7.2%) died during hospitalization and 283 (92.8%) had been cured. The mean age was 47.8 years and 53.4% were female. Baseline Neutrophil count was the strongest predictor of death, followed by age, plasma D-dimer, lymphocyte count, hsCRP, APTT, WBC, platelet count, history of hypertension and fever. The clinical model showed good discriminatory power (n=305, AUC 0.85, 95% CI 0.78–0.92, sensitivity 88.89%, specificity 73.98%, NPV 98.91%). The addition of laboratory test data to the clinical model significantly (p=0.0493) improved the discriminatory power of the model (n=264, AUC 0.92, 95% CI 0.84–0.96, sensitivity 94.44%, specificity 76.02%, NPV 99.47%).
Interpretation: In this study, we developed two predictive models for in-hospital mortality of COVID-19 patients in Wuhan. They can help to effectively predict the poor prognosis of COVID-19 patients at an early stage, and may provide practical decision-making suggestions on which patients should be paid close attention and given high-level treatments.
Funding Statement: This study was supported by a grant from National Natural Science Foundation of China (Grants: 81671386 and 81974222).
Declaration of Interests: The authors declare no competing interests.
Ethics Approval Statement: The study protocol was approved by the Medical Ethics Committee of the First People’s Hospital of Jiangxia District and complied with the Declaration of Helsinki. We verbally informed the patients that their data would be used anonymously for medical studies and had their permission.
Keywords: COVID-19; in-hospital mortality; predictive model
Suggested Citation: Suggested Citation