Dong YM, Sun J, Li YX, Chen Q, Liu QQ, Sun Z, Pang R, Chen F, Xu BY, Manyande A, Clark TG, Li JP, Orhan IE, Tian YK, Wang T, Wu W, Ye DW
Clin Infect Dis - (-) - [2020-07-10; online 2020-07-10]
The outbreak of coronavirus disease (COVID-19) in 2019 has spread worldwide and continues to cause great threat to peoples' health as well as put pressure on the accessibility of medical systems. Early prediction of survival of hospitalized patients will help the clinical management of COVID-19, but such a prediction model which is reliable and valid is still lacking. We retrospectively enrolled 628 confirmed cases of COVID-19 using positive RT-PCR tests for SARS-CoV-2 in Tongji Hospital in Wuhan, China. These patients were randomly grouped into a training cohort (60%) and a validation cohort (40%). In the training cohort, least absolute shrinkage and selection operator (LASSO) regression analysis and multivariate Cox regression analysis were utilized to identify prognostic factors for in-hospital survival of patients with COVID-19. A nomogram based on the three variables was built for clinical use. Areas under the ROC curves (AUC), concordance index (C-index) and calibration curve were used to evaluate the efficiency of the nomogram in both the training and validation cohorts. Hypertension, higher neutrophil-to-lymphocyte ratio and increased NT-proBNP value were found to be significantly associated with poorer prognosis in hospitalized patients with COVID-19. The three predictors were further used to build a prediction nomogram. The C-index of the nomogram in the training and validation cohorts was 0.901 and 0.892, respectively. The AUC in the training cohort was 0.922 for 14- day and 0.919 for 21-day probability of in-hospital survival, while in the validation cohort was 0.922 and 0.881, respectively. Moreover, the calibration curve for 14- day and 21-day survival also showed high coherence between the predicted and actual probability of survival. We managed to build a predictive model and constructed a nomogram for predicting in-hospital survival of patients with COVID-19. This model represents good performance and might be utilized clinically in the management of COVID-19.