Risk factors for adverse clinical outcomes with COVID-19 in China: a multicenter, retrospective, observational study

Abstract

Background: The risk factors for adverse events of Coronavirus Disease-19 (COVID-19) have not been well described. We aimed to explore the predictive value of clinical, laboratory and CT imaging characteristics on admission for short-term outcomes of COVID-19 patients. Methods: This multicenter, retrospective, observation study enrolled 703 laboratory-confirmed COVID-19 patients admitted to 16 tertiary hospitals from 8 provinces in China between January 10, 2020 and March 13, 2020. Demographic, clinical, laboratory data, CT imaging findings on admission and clinical outcomes were collected and compared. The primary endpoint was in-hospital death, the secondary endpoints were composite clinical adverse outcomes including in-hospital death, admission to intensive care unit (ICU) and requiring invasive mechanical ventilation support (IMV). Multivariable Cox regression, Kaplan-Meier plots and log-rank test were used to explore risk factors related to in-hospital death and in-hospital adverse outcomes. Results: Of 703 patients, 55 (8%) developed adverse outcomes (including 33 deceased), 648 (92%) discharged without any adverse outcome. Multivariable regression analysis showed risk factors associated with in-hospital death included ≥ 2 comorbidities (hazard ratio [HR], 6.734; 95% CI; 3.239-14.003, p < 0.001), leukocytosis (HR, 9.639; 95% CI, 4.572-20.321, p < 0.001), lymphopenia (HR, 4.579; 95% CI, 1.334-15.715, p = 0.016) and CT severity score > 14 (HR, 2.915; 95% CI, 1.376-6.177, p = 0.005) on admission, while older age (HR, 2.231; 95% CI, 1.124-4.427, p = 0.022), ≥ 2 comorbidities (HR, 4.778; 95% CI; 2.451-9.315, p < 0.001), leukocytosis (HR, 6.349; 95% CI; 3.330-12.108, p < 0.001), lymphopenia (HR, 3.014; 95% CI; 1.356-6.697, p = 0.007) and CT severity score > 14 (HR, 1.946; 95% CI; 1.095-3.459, p = 0.023) were associated with increased odds of composite adverse outcomes. Conclusion: The risk factors of older age, multiple comorbidities, leukocytosis, lymphopenia and higher CT severity score could help clinicians identify patients with potential adverse events.

Keywords: COVID-19; Coronavirus; Mortality; Pneumonia; Risk factor.

Figure 1

Flowchart of This Study. ICU: intensive care unit; IMV: invasive mechanical ventilation support.

Figure 2

Representative CT Images of confirmed COVID-19 pneumonia. Panel A: A 46-year-old male patient presented with fever, cough, and fatigue without a clear exposure history. Irregular GGO (red arrow) can be seen in the middle lobe of the right lung, and large patchy irregular GGO with consolidation (blue arrow) can be seen in the lower lobes of both lungs. Panel B: A 56-year old male patients presented with fever, cough, fatigue and headache without a clear exposure history. Extensive diffuse pneumonia can be seen in both lungs with multiple ILD (red arrow). Panel C: A 62-year-old male patients presented with fever, cough, myalgia as well as diarrhea, and had a close contact with confirmed COVID-19 patients. In the context of GGO with consolidation, crazy-paving pattern (red arrow) can be seen in the right lung. COVID-19: Coronavirus Disease-19; GGO: ground-glass opacity; ILD: interstitial lung disease.

Figure 3

Predictors of In-hospital Death and Adverse Outcomes. Positive estimated effect sizes of variates in multivariable Cox regression for death (Panel A: 31/545) and adverse outcomes (Panel B: 49/552). The forest plot displays estimated effect sizes of regression coefficients with 95% CI (x-axis). Panel A: Variates associated with a significant increase in in-hospital death were ≥ 2 comorbidities, leukocytosis, lymphopenia and higher CT severity score (> 14). Panel B: Variates associated with a significant increase in adverse outcomes were higher age (> 60), ≥ 2 comorbidities, leukocytosis, lymphopenia and higher CT severity score (> 14). HR: hazard ratio; CI: confidence interval.

Figure 4

Kaplan-Meier Survival Curve. Panels A-D: Kaplan-Meier survival curve for in-hospital death group according to risk factors. Panels E-I: Kaplan-Meier survival curve for adverse outcomes group according to risk factors. No.: Number.

Figure 5

Prognostic Value of Clinical, Laboratory and CT Imaging Characteristics. Comparison of time-dependent ROC curves of each as well as combination variable in death (Panel A: at median follow-up 14 days) and adverse outcome (Panel B: at median follow-up 4 days) groups. Panel A: Model 1: ≥ 2 comorbidities; Model 2: leukocytosis; Model 3: lymphopenia; Model 4: CT severity score > 14; Model 5: leukocytosis + lymphopenia; Model 6: ≥ 2 comorbidities + leukocytosis + lymphopenia; Model 7: ≥ 2 comorbidities + CT severity score > 14; Model 8: leukocytosis + lymphopenia + CT severity score > 14; Model 9: ≥ 2 comorbidities + leukocytosis + lymphopenia + CT severity score > 14. Compared with Model 3, the combination of leukocytosis and lymphopenia (Model 5) improved the performance in predicting death patients (AUC: 0.68 vs. 0.87, p < 0.001). Similarly, compared with Model 5, the combination of ≥ 2 comorbidities, leukocytosis, lymphopenia and CT severity score > 14 (Model 9) improved the performance in predicting death patients (AUC: 0.87 vs. 0.93, p = 0.02). Panel B: Model 1: age > 60 years; Model 2: ≥ 2 comorbidities; Model 3: leukocytosis; Model 4: lymphopenia; Model 5: CT severity score > 14; Model 6: age > 60 years + ≥ 2 comorbidities; Model 7: leukocytosis + lymphopenia; Model 8: age > 60 years + ≥ 2 comorbidities + leukocytosis + lymphopenia; Model 9: age > 60 years + ≥ 2 comorbidities + CT severity score > 14; Model 10: leukocytosis + lymphopenia + CT severity score > 14; Model 11: age > 60 years + ≥ 2 comorbidities + leukocytosis + lymphopenia + CT severity score > 14. Compared with Model 10, the combination of age > 60 years, ≥ 2 comorbidities, leukocytosis, lymphopenia, CT severity score > 14 (Model 11) improved the performance in predicting adverse outcome patients (AUC: 0.82 vs. 0.88, p = 0.01). ROC: receiver operating characteristic; AUC: area under curve.