Severe versus common COVID-19: an early warning nomogram model

Introduction

Coronaviruses are single-stranded, positive-sense RNAs consisting of four genera, including alpha-coronavirus (α-CoV), beta-coronavirus (β-CoV), delta-coronavirus (δ-CoV) and gamma-coronavirus (γ-CoV), which are viruses that have caused severe global infections, such as severe acute respiratory syndrome (SARS) and Middle East respiratory syndrome (MERS) [1, 2]. As a novel beta-coronavirus, severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) is characterized by an enveloped RNA virus and pleomorphic shape with a diameter of 60–140 nm [3]. It is known as the cause of coronavirus disease 2019 (COVID-19), a highly infectious disease that has been declared a global public health emergency by the World Health Organization (WHO). The COVID-19 outbreak has rapidly transitioned into a worldwide pandemic owing to the major transmission of respiratory droplets as well as contact transmission (such as inhalation of aerosols in certain activities (e.g., singing, intubation or the use of nebulizers) or even contamination of inanimate surfaces). Up to April 26, 2020, the COVID-19 pandemic caused more than 2,719,897 cases and 187,705 deaths around the world, with over 200 countries affected [4].

According to the latest version of the guidelines on the Diagnosis and Treatment of COVID-19 by the National Health Commission of China, patients infected with COVID-19 are clinically defined as mild, moderate, severe and critical [3]. Among them, mild patients present light symptoms without signs of pneumonia, and moderate patients start to show fever, respiratory symptoms and pneumonia by imaging. For severe patients, most of them develop dyspnea and/or hypoxia within one week, and even critical cases progress rapidly to acute respiratory distress symptom (ARDS), septic shock, irreversible metabolic acidosis, coagulation dysfunction and multi-organ dysfunction. Laboratory indexes show normal or reduced peripheral white blood cell (WBC) count, total lymphocyte count, increased erythrocyte sedimentation rate (ESR), and C-reactive protein (CRP) and partial reduction of heparinase (Hpa), lactate dehydrogenase (LDH), creatase, and myoglobin in moderate patients. Severe patients have decreased D-dimer and progressive reduction of peripheral lymphocytes or elevated inflammatory cytokines in severe and critical patients. At present, most patients are mild and moderate at diagnosis, but a few patients can progress rapidly to severe, critical or even death [3]. Since there is still no consensus on the factors that lead to the progression from common (mild and moderate) to severe (severe and critical) disease, the clinical diagnosis and treatment of COVID-19 is difficult. Therefore, an early warning consensus of severe patient is needed to guide early and timely treatment.

Inspired by the use of predictive models for prognosis in other diseases, we retrospectively reviewed the medical records of 1059 COVID-19 patients, conducted logistic regression analysis for factors related to severe versus common type and established an early warning nomogram model including these high risk factors of severe disease occurrence in COVID-19 patients. Then, the early warning nomogram model was validated in another cohort including 123 COVID-19 patients.

Results

Discussion

There is a large difference in prognosis between common patients and severe patients. A study on COVID-19 patients from Jinyintan Hospital and Wuhan Pulmonary Hospital reported that severe and critical patients show significantly higher mortality than general patients [5]. The in-hospital observations also show high mortality in patients admitted to extracorporeal membrane oxygenation (ECMO) and a rapid drop in SaO in severe patients. Therefore, investigating the influential factors that underlie the transition of common to severe disease is extremely important for the early identification of patients at risk as well as time management. We believe that the progression of common to severe disease occurs earlier physiologically. For instance, some patients had sharp deterioration soon after slightly increased physical activity. This indicated that the disease had already progressed over the common threshold to severe before the patient started to show severe symptoms. As analyzed by existing evidence, there are risk factors for mortality in COVID-19 patients, including older age, high D-dimer level and high Sequential Organ Failure Assessment (SOFA) score at admission [5]. In addition, the signs for ICU admission, including dyspnea, low WBC count, high neutrophil count and D-dimer level, are also critical references for the diagnosis of severe disease [6]. According to these risk factors for ICU admission as well as mortality, it is presumable that analyzing patients’ characteristics at hospital admission could help with identifying common to severe progression and may aid appropriate supportive care and prompt access to the ICU if necessary. Considering the clinical feasibility, we comprehensively extracted key clinical features of 1059 COVID patients, including sex, age, weight, symptoms and laboratory tests, conducted logistic regression analysis and successfully constructed an early warning model for severe disease occurrence.

Herein, we discovered that age (≥60 years V < 60 years), dyspnea (yes V no), lymphocyte count (low V normal), CRP (high V normal) and IL-6 (high V normal) were independent factors to infer the severe disease occurrence out of common type of COVID-19, in which age was the most important contributor. In terms of age (≥60 years V < 60 years), older patients featured a poor physical state and decreased immune function, which made them vulnerable to infections and difficult to quickly relieve in the face of COVID-19, subsequently prolonging the disease course to progress rapidly to severe or critical or even death [7, 8]. In addition, older patients usually have several comorbidities (including hypertension [9], diabetes [10] and cardiovascular disease [11]. Considering that lung infections caused by this novel coronavirus might increase the burden on the heart and cause high blood sugar, the characteristics of multisystem disease coexisting in the elderly lead to complicated and complex diseases. Multiple diseases affect each other, and the difficulty of treatment is greatly increased. Thus, elderly patients are prone to multisystem organ dysfunction and even failure, thereby progressing rapidly to severe or critical or even death [7]. Older age could mostly infer the severe disease occurrence out of common type of COVID-19.

Regarding dyspnea (yes V no), SARS-CoV-2 entered the lungs via respiration and then activated immune cells, cytokines and other pathogen-resistance systems [12]. With an increase in virus virulence, patient immunity might decrease during the struggle between the virus and the human host, which might result in congestion and edema in lung tissue, thickening of the interstitial lung and increased exudation in the alveolar space to form a transparent membrane-like structure [12]. These severe pathological changes in lung tissue might directly make patients feel dyspnea. Hence, dyspnea could be an independent factor to prewarn the severe disease occurrence out of common type of COVID-19.

In terms of lymphocyte count (low V normal), after SARS-CoV-2 infection, lymphopenia is a typical laboratory abnormality, which is similar to other highly pathogenic coronavirus infections (including SARS-CoV [13] and MERS-CoV [14] infections). Lymphocyte subsets play an important role in cellular immune regulation, and SARS-CoV-2 infection has been reported to primarily impact T lymphocytes (especially CD4+ T and CD8+ T cells), which means that T lymphocytes might be highly involved in the pathological process of COVID-19 and provide an important defense against COVID-19 [12, 15]. Additionally, its magnitude of decrease suggested the extent of impairment of the immune system by viral infection [12, 16]. Thus, lymphocyte count might act as a powerful factor for independently indicating the severe disease occurrence out of common type of COVID-19.

CRP and IL-6, two common proinflammatory factors, have been confirmed to be associated with severe lung injury and adverse clinical outcomes of SARS-CoV or MERS-CoV infection [14, 17–19]. After SARS-CoV-2 infection, CRP and IL-6 also increased rapidly in a short time [12], which is in line with the concept of a “cytokine storm,” and the magnitude of the cytokine storm is associated with disease severity [16]. Therefore, increased CRP and IL-6 might be related to the disease severity of COVID-19. In addition, CRP and IL-6 might directly accelerate protein degradation and indirectly influence important metabolic pathways to make patients frail [20], thereby causing patients to progress rapidly to severe or critical infection or even death. CRP and IL-6 might serve as independent factors prewarning the severe infection occurrence. However, considering that the understanding of COVID-19 is still insufficient and the detailed reasons for the role of these factors in COVID-19 patients is unclear, further studies are needed.

The widespread use of COVID-19 in the global population is currently the most rigorous health threat. Although most cases are mild, the apparently large number of infections poses a considerable burden to the national medical resources in affected countries. In addition, since the window of common to severe transition is short and the clinical outcomes of severe patients are extremely poor, an increasing number of lives have been lost due to delay of life support and treatment. Early warning biomarkers are urgently needed to assist with timely treatment and early prevention of disease aggravation. The nomogram model showed good agreement between prediction and actual probability in the primary cohort as well as the validation cohort. Patients who are high risk based on the nomogram model could be closely monitored for clinical index during disease course, and early intervention could be initiated to shorten the time interval between the onset of severe disease and treatment, which increases the survival possibility.

There are some limitations of our study. First, the sample size for nomogram validation was relatively small. More cases as well as a prospectively designed study are needed to further prove the accuracy of our nomogram model. Second, the phylogenetic analysis revealed three different variants of SARS-CoV-2 that might present distinct disease severity, whereas the virus variant was not differentiated in this study, which might influence the accuracy of the early warning model [21]. Third, since different centers were involved and the treatment concept in different centers might differ due to the relatively urgent medical situation, the clinical outcomes might vary. Furthermore, a comprehensive and in-depth understanding of COVID-19 is still lacking regarding the pathology and virus source; therefore, the current findings need supportive data from broader and global analyses.

In conclusion, the nomogram (including age, dyspnea, lymphocyte count, CRP and IL-6) objectively and accurately prewarns the severe disease occurrence out of common type COVID-19 patients. Our findings provide good evidence to accurately estimate changes in the patient’s condition, thereby aiding prevention and advance treatment.

Methods

Supplementary Materials

Supplementary Material