Background
Several studies have found that coagulation-related markers are correlated with a poor prognosis in patients were diagnosed COVID-19 infection. However, the findings of these studies are constrained by methodology and data analysis. Furthermore, there would be little evidence to report trends in coagulation-related indicators over time all through the disease's progression.
Methods
From January 7, 2020 to February 28, 2020, we collected 663 participants in a Chinese hospital in a non-selective and consecutive manner. The association between baseline coagulation-related indicators and death was investigated using a standardized and two-piecewise Cox regression model. The GAMM (generalized additive mixed model) was used to examine the time-varying trend of indicators in non-survivors and survivors.
Results
Of the six coagulation-related indicators, basal-Log2D-dimer (D-D) was independently positively correlated with the risk of death .The area under the ROC (receiver operating characteristic curve) curve is 0.7937 and the optimum cut-off value is 0.4650mg/L. In the GAMM model, we found that thrombin time (TT ) (interaction coefficient 0.0071; 95%CI 0.0034to 0.0108) and activated partial thromboplastin time (aPTT) (interaction coefficient 0.0217; 95%CI 0.0160 to 0.0273) of non-survivors have been increasing progressively over time during the entire disease course compared with no fluctuations among survivors. An inverse U-shaped time-vary trend was observed in D-D, prothrombin time (PT) and international normalized ratio (INR) among non-survivors, while a steady trend was found among survivors.
Conclusion
Baseline D-D were independently positively correlated with the risk of death and have the potential to be used for risk assessment early on.