Abstract
BACKGROUND: The global outbreak of COVID-19 is a significant threat to public health. Among COVID-19 cases, critically ill patients account for most in-hospital deaths. Given the pressing clinical needs, identification of potential prognostic factors that would assist clinicians to determine appropriate therapeutic interventions is urgently needed. METHODS: A retrospective analysis of 171 critically ill COVID-19 patients from two medical centers in Wuhan was conducted. The training and validation cohorts were comprised of 77 and 94 patients, respectively. Univariate and multivariate Logistic regression analyses were used to identify independent prognostic factors, and the linear prediction index was established and externally validated. RESULTS: Blood urine nitrogen (BUN) and high-sensitive C-reactive protein (hs-CRP) were independent factors negatively correlated with patient survival in the training cohort. A linear prediction model, named as the CB index (hs-CRP combined with BUN), was established and logistic regression analysis showed that this was associated with a 13% increase in death rate, with high sensitivity (86.7%) and specificity (89.7%). Patients were then divided into a high-risk group (CB index >32) and low-risk group (CB index <32) and the high-risk group showed a 56.3-fold risk of death compared with the low-risk group. Importantly, these findings were readily recaptured in the validation cohort. The efficacy of the CB index in predicting prognosis in real-world patients was then determined, which showed that patients with a higher CB index had an increased risk of death in comparison to those with a lower CB index. CONCLUSIONS: The CB index may be an important prognostic factor in critically ill COVID-19 patients.