A 'Burning Point' Is Found Before the Composite End Point Event Happened in Critically Ill Patients with COVID-19: A Multicenter Retrospective Study

Abstract

Background: The ongoing outbreak of novel coronavirus disease (COVID-19), which started in Wuhan, China in December 2019 has developed into a global pandemic. Among critical patients, death seems likely once a composite end point event (CEPE), such as acute respiratory distress syndrome (ARDS) requiring mechanical ventilation, other organ failure needing an admission to the intensive care unit (ICU) or shock occur. However, until now, little is known about the change pattern of critical cases, the crucial turning points and high-risk period before the CEPEs occur.

Methods: In this observational multicenter study, we examined 411 severe and critical COVID-19 patients admitted from Jan 12 to Feb 20, 2020, and their outcomes were followed up until Mar 16, 2020. These patients were divided into two groups (Group 1: critical patients who developed CEPE; Group 2: severe patients without CEPE and discharged). In order to predict the likelihood of CEPE during hospital stay, a baseline nomogram, based on multivariate logistic regression analysis, was constructed and validated. The linear mixed model (LMM) was used in critical patients to identify the burning point and the burning-point-related indicators (included in an early warning system), and their weights were calculated on the basis of their standardized changes after normalization.

Findings: We constructed a baseline nomogram containing five independent risk factors (blood urea nitrogen [BUN]>5·3mmol/L, D-dimer>0·97μg/ml, lactate dehydrogenase [LDH]>354U/L, C reactive protein [CRP]>21·4mg/ml, and direct bilirubin [DBIL]>4·6mmol/L) to identify high-risk patients who might develop CEPE. In the patients, the receiver operating characteristic (ROC) values in the training set and the validation set were 0·919 and 0·864 respectively. Importantly, we identified a burning point, from which some indicators started to change dramatically and CEPE is about five days away. Six most significant laboratory indicators (CRP, BUN, DBIL, platelet [PLT], neutrophil-to-lymphocyte ratio [NLR], and LDH) were selected as burning-point indicators and were integrated into the early warning system. Their continuous change estimates were 13·22 mg/L, 1·03 mmol/L, 0·77 mmol/L, -6·05 x10⁹/L, 8·57, 128·0U/L respectively, and weights of them were 37·71%, 15·28%, 15·09%, 12·04%, 11·73%, and 8·14% respectively. The time interval between the burning point and CEPE was deemed a high-risk period of CEPE.

Interpretation: The baseline nomogram model can be employed at admission to identify the high-risk patients who might develop CEPE. During hospitalization, by monitoring the changes in the burning point indicators in the early warning system, clinicians can combine their respective weights to determine whether the patient has crossed the burning point. If so, CEPE will occur within five days. Hopefully, by using this early warning system, clinicians could intervene proactively before a CEPE develops, thereby avoiding irreversible disease progression and minimizing the likelihood of fatality.

Funding Statement: This paper was supported in part by the National Natural Science Special Foundation of China for COVID-19(NO.82041018), the Independent Innovation Research Fund for Huazhong University of Science and Technology (2020kfyXGYJ).

Declaration of Interests: The authors declare no competing interests.

Ethics Approval Statement: This study was approved by the institutional review board of Medical Ethics Committee of Union Hospital, Tongji Medical College, Huazhong University of Science and Technology (NO.0036). Written informed consent was exempted by the Ethics Committee of the designated hospital for this emerging infectious disease based in Union Hospital, Wuhan, China.

Keywords: Novel Coronavirus Disease (COVID-19); Severe and critically ill patients; Composite end point event (CEPE); Burning point; Early warning system; High-risk period; Change pattern; Pseudo-improvement point; Nomogram; Linear mixed model (LMM)