Methods From January 17 to February 14, 2020, 116 patients with COVID-19 admitted to Wenzhou Central Hospital were selected. According to their clinical manifestations, 116 patients were divided into a mild group (n = 4), a common group (n = 90), a severe group (n = 18) and a critical group (n = 4). The hospitalization time and complications of all patients were recorded, and the general data and clinical indicators were compared among the 4 groups. Risk factors for the prognosis of patients with COVID-19 were obtained by multivariate Logistic regression analysis. Then a visual regression nomogram was established using R language software. Finally, efficacy of the nomogram was detected by a receiver operating characteristic (ROC) curve.
Results The age [(39 ± 11), (43 ± 12), (53 ± 13), (60 ± 8) years; F = 5.815, P = 0.001], C-reactive protein [1.7 (0.6, 7.1), 7.9 (2.9, 21.6), 28.4 (13.9, 42.5), 61.7 (44.7, 79.8) mg/L; H = 8.424, P < 0.001], hematocrit [(36 ± 5)%, (41 ± 4)%, (39 ± 4)%, (37 ± 5)%; F = 4.344, P = 0.006], platelet count [318.0 (251.0, 409.0) × 109/L, 180.5 (140.0, 225.5) × 109/L, 162.0 (130.0, 222.8) × 109/L, 108.5 (82.0, 103.0) × 109/L; H = 7.225, P < 0.001], aspartate aminotransferase [16.0 (15.5, 19.5), 23.0 (19.0, 31.0), 34.5 (26.3, 55.0), 39.5 (29.0, 82.3) U/L; H = 6.159, P = 0.001], albumin [(44 ± 6), (43 ± 16), (39 ± 3), (33 ± 4) g/L; F = 9.508, P < 0.001], and lactate dehydrogenase [142.5 (107.8, 189.3), 198.0 (159.5, 238.0), 295.0 (251.0, 323.0), 369.5 (295.2, 436.3) U/L; H = 14.225, P < 0.001] of patients with COVID-19 in the 4 groups were statistically significantly different. Then the age, C-reactive protein, hematocrit, platelet count, aspartate aminotransferase, albumin, and lactate dehydrogenase were incorporated into multivariate Logistic regression analysis. It showed that the albumin [odds ratio (OR) = 0.756, 95% confidence interval (CI) (0.581, 0.982), P = 0.036] and lactate dehydrogenase [OR = 1.019, 95%CI (1.007, 1.032), P = 0.002] were risk factors for the prognosis of patients with COVID-19. In the meantime, a reliable and intuitive nomogram was obtained by R language software. ROC curve analysis showed that this nomogram had excellent predictive ability for severe and critical patients [area under the curve (AUC) = 0.903, 95%CI (0.831, 0.975), P < 0.001], for critical patients alone [AUC = 0.974, 95%CI (0.932, 1.000), P < 0.001] and for severe patients alone [AUC = 0.848, 95%CI (0.759, 0.937), P < 0.001].