Purpose: To predict the risk of developing severe pneumonia among mild novel coronavirus pneumonia (mNCP) patients on admission.
Methods: A retrospective cohort study was conducted at three hospitals in Shanghai and Wuhan from January 2020 to February 2020. Real-time polymerasechain-reaction assays were used to detect COVID-19. A total of 529 patients diagnosed with NCP were recruited from three hospitals and classified by four severity types during hospitalization following the standards of the Chinese Diagnosis and Treatment of Pneumonia Caused by New Coronavirus Infection (eighth version). Patients were excluded if admitted by ICU on admission (n=92, on a general ward while meeting the condition of severe or critical type on admission (n=25), or there was insufficient clinical information (n=64). In sum, 348 patients with mNCP were finally included, and 68 developed severe pneumonia.
Results: mNCP severity prognostic index values were calculated based on multivariate logistic regression: history of diabetes (OR 2.064, 95% CI 1.010-4.683; p=0.043), time from symptom onset to admission ≥7 days (OR 1.945, 95% CI 1.054-3.587; p=0.033), lymphocyte count ≤0.8 (OR 1.816, 95% CI 1.008-3.274; p=0.047), myoglobin ≥90 mg/L (OR 2.496, 95% CI 1.235-5.047; p=0.011), and D-dimer ≥0.5 mg/L (OR 2.740, 95% CI 1.395-5.380; p=0.003). This model showed a c-statistics of 0.747, with sensitivity and specificity 0.764 and 0.644, respectively, under cutoff of 165.
Conclusion: We designed a clinical predictive tool for risk of severe pneumonia among mNCP patients to provided guidance for medicines. Further studies are required for external validation.
Keywords: novel coronavirus pneumonia; predicting score; severe pneumonia.
© 2020 Guo et al.