ANDC: an early warning score to predict mortality risk for patients with Coronavirus Disease 2019

Abstract

Background: Patients with severe Coronavirus Disease 2019 (COVID-19) will progress rapidly to acute respiratory failure or death. We aimed to develop a quantitative tool for early predicting mortality risk of patients with COVID-19.

Methods: 301 patients with confirmed COVID-19 admitted to Main District and Tumor Center of the Union Hospital of Huazhong University of Science and Technology (Wuhan, China) between January 1, 2020 to February 15, 2020 were enrolled in this retrospective two-centers study. Data on patient demographic characteristics, laboratory findings and clinical outcomes was analyzed. A nomogram was constructed to predict the death probability of COVID-19 patients.

Results: Age, neutrophil-to-lymphocyte ratio, D-dimer and C-reactive protein obtained on admission were identified as predictors of mortality for COVID-19 patients by LASSO. The nomogram demonstrated good calibration and discrimination with the area under the curve (AUC) of 0.921 and 0.975 for the derivation and validation cohort, respectively. An integrated score (named ANDC) with its corresponding death probability was derived. Using ANDC cut-off values of 59 and 101, COVID-19 patients were classified into three subgroups. The death probability of low risk group (ANDC < 59) was less than 5%, moderate risk group (59 ≤ ANDC ≤ 101) was 5% to 50%, and high risk group (ANDC > 101) was more than 50%, respectively.

Conclusion: The prognostic nomogram exhibited good discrimination power in early identification of COVID-19 patients with high mortality risk, and ANDC score may help physicians to optimize patient stratification management.

Keywords: COVID-19; Mortality; Nomogram; Risk factor; SARS-Cov-2.

Fig. 1

Flow chart of study participants in the derivation and validation cohort

Fig. 2

Nomogram to predict the death probability of patients with COVID-19. The nomogram was constructed based on the following variables: age, NLR, D-dimer and CRP. Locate the values of a patient’s age, NLR, D-dimer, and CRP and draw four vertical lines for each of the four predictors to reach the “Points” axis, respectively. The intersections between the vertical lines and the “Points” axis are the corresponding score for the predictors. The summation of the scores from four predictors (named ANDC) could be converted to death probability of patients with COVID-19 by drawing another vertical line from the “Total points” axis to the “Death probability” axis. COVID-19, coronavirus disease 2019; NLR, neutrophils-to-lymphocytes ratio; CRP, C-reactive protein

Fig. 3

Calibration plot comparing predicted and actual death probability of patients with COVID-19. These two figures show actual against predicted death probability of patients with COVID-19. a represents the internal validation. b Represents the external validation. Dotted curve represents the apparent curve without bootstrapping correction. The solid curve represents the 1000-times repeated bootstrapping-correction curve. The dashed curve represents the ideal fit. COVID-19, coronavirus disease 2019

Fig. 4

Decision curves analysis comparing different models to predict the death probability of patients with COVID-19. The net benefit balances the mortality risk and potential harm from unnecessary over-intervention for patients with COVID-19. Full model incorporates the following predictors: age, NLR, D-dimer and CRP. COVID-19 coronavirus disease 2019, NLR neutrophils-to-lymphocytes ratio, CRP C-reactive protein