|
?:abstract
|
-
OBJECTIVE To develop and validate a nomogram using on admission data to predict in-hospital survival probabilities of COVID-19 patients. METHODS We analyzed 855 COVID-19 patients with 52 variables. The least absolute shrinkage and selection operator regression and multivariate Cox analyses were used to screen significant factors associated with in-hospital mortality. A nomogram was established based on the variables identified by Cox regression. The performance of the model was evaluated by C-index and calibration plots. Decision curve analysis was conducted to determine the clinical utility of the nomogram. RESULTS Six variables, including neutrophil (hazard ratio [HR] 1.088, 95% confidence interval [CI] [1.0004-1.147], p < 0.001), C-reactive protein (HR 1.007, 95% CI [1.0026-1.011], p = 0.002), interleukin-6 (HR 1.001, 95% CI [1.0003-1.002], p = 0.005), D-dimer (HR 1.034, 95% CI [1.0111-1.057], p = 0.003), prothrombin time (HR 1.086, 95% CI [1.0369-1.139], p < 0.001), and myoglobin (HR 1.001, 95% CI [1.0007-1.002], p < 0.001), were identified and applied to develop a nomogram. The nomogram predicted 14-day and 28-day survival probabilities with reasonable accuracy, as assessed by the C-index (0.912) and calibration plots. Decision curve analysis showed relatively wide ranges of threshold probability, suggesting a high clinical value of the nomogram. CONCLUSION Neutrophil, C-reactive protein, interleukin-6, D-dimer, prothrombin time, and myoglobin levels were significantly correlated with in-hospital mortality of COVID-19 patients. Demonstrating satisfactory discrimination and calibration, this model could predict patient outcomes as early as on admission and might serve as a useful triage tool for clinical decision making. This article is protected by copyright. All rights reserved.
|
![[This page as RDF]](https://research.tib.eu/covid-19/static/rdf-icon.gif){kind=icon}