Establish A Novel Model For Predicting Significant Progression In COVID-19 Imaging And Comparison Of Clinical Features With Human H7N9 Avian Influenza

Part 1:Establish a novel model for predicting significant progression in COVID-19 imagingObjective:To investigate the predictive value of four pneumonia scores about PSI,CURB-65,MuLBSTA and COVID-GRAM for significant imaging progression in patients with COVID-19,and to establish a novel risk prediction model for significant imaging progression in patients with COVID-19 based on multi-factor analysis of clinical features of viral pneumonia.Methods:1.A retrospective study was conducted to collect clinical data of 87 patients admitted to the Infectious Diseases Hospital of Soochow University with confirmed COVID-19 from January 2020 to March 2020,and the significant progression of pulmonary imaging(>50%)within 72 hours of admission as the study endpoint,and divided into significant progression and non-significant progression groups.2.The worst values of PSI,CURB-65,MuLBSTA and COVID-GRAM scores within 24 hours of admission were calculated for COVID-19 inpatients.The risk groups,including low-risk,intermediate-risk and high-risk groups were assigned according to the scores,and the rates of significant pulmonary imaging progression were compared between the groups.3.Plot the subject operating characteristic curves(ROC),compare the area under the curve(AUC),assess the clinical efficacy of each scoring tool,and determine the sensitivity and specificity at the optimal cutoff value.4.Multi-factor logistic regression analysis of clinical characteristics of all patients was performed to identify independent risk factors and to establish a novel risk prediction model for significant progression in imaging.Results:1.The types of imaging progression in COVID-19 inpatientsAmong all of 87 patients hospitalized with COVID-19,20(22.99%)patients showed significant progression of pulmonary imaging within 72 hours of admission,31(35.63%)patients showed imaging progression but not more than 50%,27(31.03%)patients showed absorption of imaging and 9(10.34%)showed no change.2.The analysis of severity of COVID-19 inpatients using different pneumonia scoring toolsUsing the PSI score,there were 84(96.55%)cases in the low risk group and 3(3.45%)cases are in the intermediate risk group;Using the CURB-65 score,there were 82(94.25%)cases in the low risk group and 5(5.75%)cases were in the intermediate risk group,there were no patients in the high risk group in both scores;Using the MuLBSTA score,56(64.37%)cases were in the low risk group,22(25.29%)cases in the medium risk group and 9(10.34%)cases in the high risk group;Using the COVID-GRAM score,7(8.05%)cases were in the low risk group,70(80.46%)cases in the medium risk group and 10(11.49%)cases in the high risk group.3.The relationship between different severity groupings obtained by different pneumonia scoring tools and imaging progression in COVID-19 patientsSignificant imaging progression was seen in 22.62%of patients in the low risk group with PSI score(19/84),33.33%of patients in the intermediate risk group(1/3)and no patients in the high risk group,with no statistically significant difference between the groups(P>0.05);21.95%of patients in the low risk group with CURB-65 score showed significant imaging progression(18/82),and 40.00%of patients showed significant imaging progression(2/5)and there were no patients in the high risk group,with no statistically significant difference between the groups(P>0.05);16.67%of patients in the MuLBSTA score of low and intermediate risk group(13/78)and 77.78%of patients in the high risk group(7/9),with a statistically significant difference between groups(P<0.001);18.18%of patients in the COVID-GRAM score of low and intermediate risk group(14/77)and 60.00%of patients in the high risk group showed significant imaging progression(6/10),with a statistically significant difference between groups(P=0.002).4.The analysis of the predictive value of four scoring tools for significant progression of COVID-19 imaging using ROC curvesFor the PSI score,taking the best cutoff value of 46.50,the sensitivity was 64.20%,the specificity was 60.00%,and the area under the ROC curve(AUC)was 0.588(95%CI 0.451-0.725);For the CURB-65 score,the sensitivity was 74.60%,the specificity was 60.00%,and the area under the ROC curve(AUC)was 0.678(95%CI 0.550-0.806)when the optimal cut-off value of 1.50 was taken;For the MuLBSTA score,taking the best cut-off value of 6.00,the sensitivity in predicting significant progression of COVID-19 imaging was 67.20%,the specificity was 95.00%,and the area under the ROC curve(AUC)was 0.882(95%CI 0.810-0.955);For the COVID-GRAM score,when taking the best cut-off value of 104.15,the sensitivity to predict significant progression in COVID-19 imaging was 86.60%,the specificity was 80.00%,and the area under the ROC curve(AUC)was 0.866(95%CI 0.813-0.959).The above results suggest that there were differences in the clinical efficacy of the four scores for predicting significant progression of COVID-19 imaging:COVID-GRAM>MuLBSTA>CURB-65>PSI.5.Establish a novel risk model for predicting significant progression of COVID-19 imaging based on multi-factor analysisThrough univariate and multi-factor logistic regression analysis,the Age,C-reactive protein(CRP),oxygenation index(OI),and neutrophil-to-lymphocyte ratio(NLR)were identified as independent risk factors for significant progression of COVID-19 imaging.Based on the above independent risk factors,a new risk prediction model was established based on the regression coefficient(β)of each risk factor ACON=(0.11×CRP+0.12×NLR)/(0.06×OI+0.05×Age).Taking the best cut-off value of 0.088,it had a sensitivity of 85.1%and a specificity of 95.0%for predicting significant progression in COVID-19 imaging,with an AUC of 0.949(95%CI 0.900-0.998)and higher predictive accuracy than the other four scores.Conclusions:1.The traditional pneumonia scoring tools PSI score and CURB-65 score perform poorly in terms of clinical predictive efficacy for significant progression on COVID-19 imaging and are not suitable for risk assessment of imaging progression on COVID-19.2.The MuLBSTA score and the COVID-GRAM score have good predictive value for significant COVID-19 imaging progression,both of them have similar predictive accuracy and can effectively screen for patients at high risk of significant progression.3.The novel risk prediction model(ACON)based on age(Age),C-reactive protein(CRP),oxygenation index(OI),and neutrophil-to-lymphocyte ratio(NLR)had better predictive efficacy for significant progression of COVID-19 imaging than the MuLBSTA score and COVID-GRAM score.Part 2:Comparison of clinical features of COVID-19 and human H7N9 avian influenzaObjective:To systematically compare the clinical features of patients with severe pneumonia caused by COVID-19 and human H7N9 avian influenza in order to facilitate better differentiation between them by clinicians.Methods:A retrospective study was conducted to collect clinical data of 23 patients with severe COVID-19 and 23 patients with severe human H7N9 avian influenza admitted to the Infectious Disease Hospital of Soochow University.The differences in demographic characteristics,clinical manifestations,laboratory findings,imaging features,severity,treatment and clinical regression between the two groups were compared.Results:1.The analysis of patient demographics and underlying diseaseA higher proportion of patients with severe human H7N9 avian influenza developed underlying disease than patients with severe COVID-19(56.52%vs 26.09%,P<0.05),and more than half of these patients with severe human H7N9 avian influenza had hypertension(52.17%vs 8.7%,P<0.01).2.The analysis of differences in clinical presentationDyspnea(86.96%vs 30.43%,P<0.01),muscle aches(47.83%vs 17.39%,P<0.05),and nasal congestion(30.43%vs 10.00%,P<0.05)occurred in a higher proportion of patients with severe human H7N9 avian influenza than in patients with severe COVID-19,and fever,cough,sore throat,gastrointestinal symptoms occurred in comparable proportions in both groups(P>0.05),and hemoptysis was not present in either group.3.The analysis of laboratory testsBoth patients with severe COVID-19 and severe human H7N9 avian influenza showed abnormalities in coagulation and lymphocytes,but severe human H7N9 avian influenza had increased D-dimer(5447.00±1094.00 vs 1914.00±426.50 μg/ml,P<0.01)and prothrombin time(PT)was prolonged(14.67±0.74 vs 12.47±0.36 s,P<0.01)and lymphopenia was more pronounced than in severe COVID-19(0.44±0.39 vs 0.71±0.06×109/L,P<0.01).4.Imaging analysisGround glass shadows(43.48%vs 60.87%,P>0.05)and solid lung lesions(60.87%vs 100.00%,P<0.01)were common in both patients with severe COVID-19 and patients with severe human H7N9 avian influenza,but paving stone signs were more frequent in patients with severe COVID-19(26.09%vs 0.00%,P<0.05),less frequent pleural effusion(0.00%vs 47.83%,P<0.01).5.SeverityThe oxygenation index was lower in patients with severe human H7N9 avian influenza than in patients with severe COVID-19(133.70±11.18 vs 228.30±11.01 mmHg,P<0.05),but the time from onset to the onset of ARDS was not significantly different(8.39±0.59 vs 8.09±0.72 days,P>0.05);Patients with severe human H7N9 avian influenza had worse SOFA scores(5.35±0.77 vs 3.57±0.31 points,P<0.05)and PSI scores(89.96±7.72 vs 61.22±3.00 points,P<0.01)than patients with severe COVID-19 and had a higher rate of developing acute myocardial injury(69.57%vs 4.35%,P<0.05).6.Time to viral RNA conversionPatients with severe COVID-19 required a longer time to viral RNA regression compared with severe human H7N9 avian influenza(13.91 ± 1.56 vs 9.86±0.81 days,P<0.05),but imaging remission required a shorter time than patients with human H7N9 avian influenza(9.13±0.70 vs 11.64±0.83 days,P<0.05).7.Clinical regressionA significantly higher proportion of patients with severe human H7N9 avian influenza received noninvasive mechanical ventilation throughout treatment than patients with severe COVID-19(69.57%vs 21.74%,P<0.05),with a higher proportion of secondary Gram-positive bacterial infections(34.78%vs 4.35%,P<0.05);three cases of severe human H7N9 avian influenza deaths and no fatal patients in COVID-19.no patients with COVID-19 deathsConclusions:1.Compared with patients with severe COVID-19,patients with severe human H7N9 avian influenza had a higher proportion of comorbid underlying disease,a higher proportion of respiratory catarrh symptoms,worse disease severity scores,lower oxygenation index,a higher proportion of ARDS,acute cardiac injury,and more frequent secondary Gram-positive bacterial infections;patients with severe human H7N9 infection had abnormal coagulation and lymphopenia were more pronounced than in patients with severe COVID-19.2.The crazy-paving signs were seen on imaging in a greater proportion of patients with severe COVID-19 compared with patients with severe human H7N9 avian influenza,but a smaller proportion had combined pleural effusions and solid lung changes,and their imaging was easily improved;patients with severe COVID-19 had a prolonged time to viral RNA conversion compared with patients with severe human H7N9 avian influenza.3.With the better understanding of viral pneumonia,patients with severe COVID-19 can achieve lower mortality due to the individualized treatment and refined management.