key: cord-286544-ipmcqz8n authors: Cheng, Biao; Hu, Jiahao; Zuo, Xiuran; Chen, Jian; Li, Xiaochao; Chen, Yuchen; Yang, Guoliang; Shi, Xiaowu; Deng, Aiping title: Predictors of progression from moderate to severe COVID-19: a retrospective cohort date: 2020-07-02 journal: Clin Microbiol Infect DOI: 10.1016/j.cmi.2020.06.033 sha: doc_id: 286544 cord_uid: ipmcqz8n OBJECTIVE: Most coronavirus disease 2019 (COVID-19) cases were identified as moderate, which is defined as having a fever or dry cough and lung imaging with ground-glass opacities. The risk factors and predictors of prognosis in such cohorts remain uncertain. METHODS: All adult patients with COVID-19 of moderate severity diagnosed using qRT-PCR and hospitalized at the Central Hospital of Wuhan, China, from Jan 1 to Mar 20, 2020 were enrolled in this retrospective study. The main outcomes were progression from moderate to severe or critical condition or death. RESULTS: Among the 456 enrolled patients with moderate COVID-19, 251/456 (55.0%) had poor prognosis. Multivariate logistic regression analysis identified higher NLR on admission (OR =1.032, 95%CI 1.042-1.230, P = 0.004) and higher CRP on admission (OR =3.017, 95%CI 1.941-4.690, P < 0.001) were associated with increased odds ratios of poor prognosis. The area under the receiver operating characteristic (ROC) curve (AUC) for NLR and CRP in predicting progression to critical condition was 0.77 (95% CI 0.694-0.846, P < 0.001) and 0.84 (95% CI 0.780-0.905, P < 0.001), with a cut-off value of 2.79 and 25.95 mg/l, respectively. The AUC of NLR and CRP in predicting death was 0.81 (95% CI, 0.732-0.878, P < 0.001) and 0.89 (95% CI 0.825-0.946, P < 0.001), with a cut-off value of 3.19 and 33.4 mg/l, respectively. CONCLUSIONS: Higher levels of NLR and CRP at admission were associated with poor prognosis of moderate COVID-19 patients. NLR and CRP were good predictors of progression to critical condition and death. As of Apr 19, 2020, there have been 2,241,359 confirmed cases of COVID-19 worldwide, including 152,551 deaths reported by WHO [1] . The outbreak of COVID-19 has become an international public health emergency [2, 3] . The prognosis of COVID-19 patients with different severities at admission is significantly different. Most mild or moderate patients that receive basic medical care at Fangcang shelter hospitals, which are large-scale, temporary hospitals rapidly built since Feb 5 in China, have better prognosis [4] . Relative to the moderate cases, severe or critical patients have a higher probability of being admitted to intensive care units (ICU), have longer stays [5, 6] , and are more likely to die [7, 8] . Identification of which initially mild or moderate patients will deteriorate into severe or critical illness is useful, as it would allow for earlier treatment to prevent worsening outcomes and save medical resources for other patients. In this study, we focus on the clinical features and outcomes of patients with moderate COVID-19 treated at a single institution and explore the factors and indicators associated with their prognosis. All adult patients with moderate cases of COVID-19 hospitalized at the Central Hospital of Wuhan from January 1 to March 20, 2020, were enrolled in this retrospective cohort study. This is a tertiary hospital located in the central area of Wuhan, China, and is one of the designated hospitals for treating COVID-19 patients. The data cutoff for this study was March 31, 2020. The flowchart of confirmed patients enrolled in this study is shown in Fig S1. All patients were diagnosed with COVID-19 based on positive SARS-CoV-2 qRT-PCR using throat swab samples, in accordance with the Diagnosis and Treatment Protocol for Novel Coronavirus Pneumonia recommended by the National Health Commission (NHC) of China (version 7.0) [9] . This study was approved by the Central Hospital of Wuhan Hospital Ethics Committee (No. 2020-75). Written informed consent was waived by the Ethics Commission of the designated hospital for emerging infectious diseases. Epidemiological, demographic, clinical, laboratory, treatment, and outcome data (progression to severe/critical/death) were reviewed and extracted from electronic medical records using a standardized data collection form by experienced clinicians and independently reviewed by two researchers. Fever was defined as an axillary temperature of at least 37.3°C. Disease severity grading (mild, moderate, severe, or critical) of COVID-19 was defined according to the Diagnosis and Treatment Protocol for Novel Coronavirus Pneumonia. Mild grade was defined as few symptoms (low fever, fatigue) and without lung computed tomography (CT) findings. Moderate grade was defined as fever, respiratory symptoms (dry cough, chest distress, and shortness of breath after activities), and lung CT findings (i.e. ground glass opacity, multiple small patchy shadows, and pulmonary consolidation). Severe grade was defined as respiratory frequency ≥ 30/min, blood oxygen saturation ≤ 93%, oxygenation index < 300 mmHg, and/or lung infiltrates > 50% within 24 to 48 hours. Critical grade was defined as respiratory failure, septic shock, and/or multiple organ dysfunction or failure. Poor prognosis refers to progression from moderate to severe grade, critical grade, or death. Categorical variables are reported as number (%). Normally distributed continuous data were reported as mean ± standard deviation (SD) and non-normally distributed continuous data were reported as median (interquartile range [IQR] ). Categorical data were compared using the χ2 test or Fisher exact test. Independent t-tests were used to compare normally distributed continuous data, while the Mann-Whitney U-test or Exact Mann-Whitney rank sum test was used to compare non-normally distributed continuous data. To adjust for the risk factors associated with illness progression inhospital, univariable and multivariable logistic regression models were used. Considering the total number of prognoses (n=251) in our study and to avoid overfitting of the model, 12 variables were chosen for multivariable logistic analysis on the basis of univariable logistic analysis results and clinical significance. Multivariable Cox proportional hazards regression analyses were used to further adjust the risk factors associated with survival. Considering the total number of deaths (n=46) in our study and to avoid overfitting of the model, four variables were chosen for Cox regression analysis on the basis of multivariable logistic analysis results and clinical significance. Receiver operating characteristic (ROC) curves were used to evaluate the potential predictive value of risk factors on prognoses in-hospital. The Hosmer-Lemeshow test was used to calibrate the ROC curves. The Net Reclassification Index (NRI) was used to determine which indicators of ROC curves analysis were better at predicting outcomes, in line with previously published methods [10] . P value less than 0.05 was considered statistically significant. Statistical analysis was performed using SPSS (version 19.0) and GraphPad Prism (version 8.0) software. A total of 456 (100%) moderate cases were recruited in this study (Table 1) The laboratory data of all moderate cases on admission are shown in Table 1 . Numerous variables were significantly associated with outcome, and cases with poor prognoses generally had lower lymphocyte counts, and higher levels of C-reactive protein (CRP), neutrophil/lymphocyte ratio (NLR), and procalcitonin. Treatment and outcome data are presented in Table 2 . As indicated, antiviral treatment (i.e. ribavirin, arbidol and lopinavir/ritonavir) was the most common treatment method for moderate cases (437/456, 95.83%), followed by antibiotic treatment (i.e. ephalosporins and quinolones; 369/456, 80.92%) and glucocorticoid treatment (226/456 patients, 49.56%). Glucocorticoid treatment and intravenous immunoglobin were more commonly used for patients with poor prognoses than patients that did not progress. The median time of illness onset to admission was 7 days (IQR 4.25-14) in all moderate patients and did not differ significantly between two groups (p > 0.05). were associated with increased odds ratios of poor prognoses. Furthermore, we calculated the odds ratio for the different of prognoses in more detail (Table S1) . Briefly, older age, male gender, and NLR and CRP levels at admission greater than 6.0 mg/l were associated with increased odds ratios of severe progression. Male gender, NLR, CRP greater than 6.0 mg/l on admission were associated with increased odds ratios of progression to critical condition. Older age, male gender, NLR, procalcitonin greater than 0.5 ng/ml, and CRP greater than 6.0 mg/l on admission were associated with increased odds ratios of death. These results are consistent with our Cox regression analysis (Table S2) . To explore risk factors that can predict prognosis of patients with moderate COVID-19, we used ROC curve analysis. The ROC curve of NLR and CRP in predicting the total poor prognoses and severe progression is shown in Figure 1a Table 4 . In this retrospective study, the major symptoms of moderate COVID-19 were fever and cough and these symptoms did not differ between the two outcome groups (Table 1 ). Therefore, predicting prognosis based on symptoms is not possible. Using comparative and multivariable analyses of basic patient characteristics, we found that comorbidities in moderate cases are not a risk factor for poor prognosis, which is consistent with recent studies [11] . However, older age, male gender, and NLR and CRP levels on admission were significantly associated with poor prognoses in patients with moderate COVID-19. In our study, the AUC of both NLR and CRP in predicting progression to critical condition and death was more than 0.75 (Table 4) , which suggests that NLR and CRP may act as predictors of progression. Compared with NLR, the NRI of CRP was greater than 0 in predicting progression to critical condition and death, indicating that CRP is a better predictor, which is consistent with AUC results. Additionally, although the AUC of PCT in predicting death was also more than 0.75, the P value of ROC curve of the Hosmer-Lemeshow test for PCT was less than 0.001 (Table 4) , which suggests poor calibration of the ROC curve. Hence, the difference between the predicted value and the true value cannot be explained by chance. Thus, these results indicate that PCT is not a good predictor of death in moderate COVID-19 cases in our study. Additionally, multivariable logistic analysis revealed that antibiotic, intravenous immunoglobin, and glucocorticoids treatments were not associated with prognosis (Table S3 ), suggesting that these medications did not improve prognosis when given to patients with moderate COVID-19. As most COVID-19 cases are mild or moderate and medical resources are limited, these findings are clinically significant for taking appropriate treatment options and utilizing medical resources in a cost-effective way. However, randomized controlled trials (RCTs) are required to confirm the impact of drug treatment on moderate COVID-19 patients. There are several limitations of the study. First, this is a single center, retrospective study. Second, most moderate COVID-19 patients that were enrolled in this study were older and had multiple comorbidities, and thus were more likely to have adverse outcomes. Hence, the rate of disease progression in our study may not reflect the true rate. In conclusion, age, gender, and NLR and CRP levels at admission are associated with poor prognoses of patients with moderate COVID-19. NLR and CRP levels on admission tend to be a good predictor of critical progression and death. The authors declare that they have no conflicts of interest. Total poor prognoses, moderate cases progress to severe, critical cases or death; NLR, neutrophil-lymphocyte ratio. CRP, C-reactive protein. Coronavirus disease (COVID-19) outbreak situation Characteristics of and Important Lessons From the Coronavirus Disease 2019 (COVID-19) Outbreak in China: Summary of a Report of 72314 Cases From the Chinese Center for Disease Control and Prevention Fangcang shelter hospitals: a novel concept for responding to public health emergencies Clinical progression of patients with COVID-19 in Shanghai Clinical course and risk factors for mortality of adult inpatients with COVID-19 in Wuhan, China: a retrospective cohort study Coronavirus Disease 2019 in elderly patients: characteristics and prognostic factors based on 4-week follow-up Diagnosis and Treatment Protocol for Novel Coronavirus Pneumonia recommended Discrimination and Calibration of Clinical Prediction Models: Users' Guides to the Medical Literature Prevalence of comorbidities in the novel Wuhan coronavirus (COVID-19) infection: a systematic review and meta-analysis None. This work was supported by the Natural Science Foundation of Hubei Province of China (2019CFA426).