Admission and demography data
A total of 23,253 COVID-19 patients were admitted to our hospital from December 1, 2022 to January 31, 2023. Patients were excluded because they did not have complete records (n=465), did not meet criteria for critical COVID-19 (n=22,230), or below 18 years of age (n=133) (Figure 2). Finally, 424 patients were enrolled, with a median age of 75.46 years ranging from 21 to 98 years, and a major proportion of male (291/424, 68.63%). Comorbidities were present in nearly all of patients, with hypertension being the most common comorbidity (n=247, 58.25%), followed by diabetes (n=138, 32.55%) and cerebrovascular disease (n=73, 17.22%) (Table 1).
Although all patients were treated according to the “Diagnosis and Treatment Scheme of New Coronavirus Infected Pneumonia[20]”, the total in-hospital mortality of critical COVID-19 patients was 43.40% (184/424) (Table 1).Compared with the survival group, the non-survival group had higher proportions of older age (>75 years) (61.41% vs. 51.25%, P=0.037), more comorbidities of diabetes [38.59% vs. 27.92%, P=0.020), cerebrovascular disease (22.28% vs. 13.33%, P=0.044], chronic renal failure (15.76% vs. 7.92%, P=0.012), obesity (16.85% vs. 7.50%, P=0.003) (Table 1).
Laboratory and imaging characteristics
In comparison with survival group, patients in non-survival group displayed a higher value of creatinine [178.05(71.25-188.20) vs. 138.71(65-150.75) μmol/L, P=0.029], procalcitonin [6.59(0.20-2.79) vs. 4.99 (0.12-1.80) ng/mL, P=0.002], C-reactive protein (111.22±75.72 vs. 95.74±79.12 mg/L, P=0.011), interleukin-6 [749.48(33.70-285.50) vs. 349.68(14.11-137.98) pg/mL, P<0.001], SOFA score [10.41(8-13) vs. 7.60(4-11), P<0.001], and lymphocytopenia [135(73.37%) vs. 125(52.08%), P<0.001] on admission day (Table 2). The main imaging feature was bilateral pneumonia [198(82.50%) vs. 145(78.80%), P=0.489], followed by partial patchy shadows [137(57.08%) vs. 110(59.78%), P=0.523], multiple patchy shadows [119(49.58%) vs. 97(52.72%), P=0.577], and hydrothorax [119(49.58%) vs. 85(46.20%), P=0.489]. Of note, significant higher score of CT-SS on admission day was observed in non-survival group than that in survival group [18.99(13-25) vs. 10.70(5-15), P<0.001].
Table 1. Population characteristics of hospitalized patients with severe COVID-19a,b.
Variable
|
Overall
(n=424)
|
Survival
(n=240)
|
non-survival
(n=184)
|
P Value
|
Age(>75years old)
|
236(55.66%)
|
123(51.25%)
|
113(61.41%)
|
0.037b
|
Gender
|
|
|
|
0.786b
|
Male, n(%)
|
292(68.87%)
|
164(68.33%)
|
128(69.57%)
|
|
Female, n(%)
|
132(30.90%)
|
76 (31.67%)
|
56 (30.43%)
|
|
BMI
|
|
|
|
|
Comorbidities
|
|
|
|
|
Obesity, n(%)
|
49(11.56%)
|
18(7.50%)
|
31(16.85%)
|
0.003b
|
Diabetes, n(%)
|
138(32.55%)
|
67 (27.92%)
|
71 (38.59%)
|
0.020b
|
Hypertension, n(%)
|
247(58.25%)
|
135(56.25%)
|
112(60.87%)
|
0.339b
|
Chronic heart failure, n(%)
|
21 (4.95%)
|
8(3.33%)
|
13 (7.07%)
|
0.079b
|
Cerebrovascular disease, n(%)
|
73 (17.22%)
|
32 (13.33%)
|
41 (22.28%)
|
0.044b
|
COPD, n(%)
|
32 (7.55%)
|
13 (5.42%)
|
19 (10.33%)
|
0.058b
|
Chronic renal failure, n(%)
|
48 (11.32%)
|
19 (7.92%)
|
29 (15.76%)
|
0.012b
|
Note: aMann-Whitney U test, bPearson χ2 test.
Abbreviations: BMI: body mass index; Obesity, (BMI>30); COPD: chronic obstructive pulmonary disease.
Treatment and outcomes
Compared with the survival group, non-survival patients had higher proportion of mechanical ventilation (92.39% vs. 77.92%, P<0.001), usage of dexamethasone (86.96% vs. 77.92%, P=0.017), need of CRRT (32.07% vs. 11.67%, P<0.001) on admission day. It also had longer length of hospitalization in non-survival group when compared with the survival group [24.74±16.94 vs. 21.32±17.30 d, P<0.001] (Table 2).
Table 2. Laboratory/imaging feature data on admission and post resuscitation care.
Variable
|
Overall
(n=424)
|
Survival
(n=240)
|
non-survival
(n=184)
|
P Value
|
Total bilirubin (μmol/L)
|
19.19 (9.50-18.98)
|
17.13(9.20-18.73)
|
21.88(9.83-19.65)
|
0.319a
|
Creatinine (μmol/L)
|
155.75(69.10-159.50)
|
138.71(65-150.75)
|
178.05(71.25-188.20)
|
0.029a
|
WBC (×109/L)
|
10.10±10.13
|
10.05±9.22
|
10.18±75.72
|
0.836a
|
Procalcitonin (ng/mL)
|
5.68(0.15-2.18)
|
4.99(0.12-1.80)
|
6.59(0.20-2.79)
|
0.002a
|
C-reactive protein (mg/L)
|
102.46±77.95
|
95.74±79.12
|
111.22±75.72
|
0.011a
|
Interleukin-6 (pg/mL)
|
523.18(18-183)
|
349.68(14.11-137.98)
|
749.48(33.70-285.50)
|
<0.001a
|
PaO2/FiO2 (mmHg)
|
224.35±135.14
|
230.12±131.75
|
216.83±139.43
|
0.933a
|
D-Dimer elevation, n(%)
|
404(95.28%)
|
225(93.75%)
|
179(97.28%)
|
0.089b
|
Lymphopenia, n(%)
|
260(61.32%)
|
125(52.08%)
|
135(73.37%)
|
<0.001b
|
SOFA score, median(IQR)
|
8.82 (6-12)
|
7.60(4-11)
|
10.41(8-13)
|
<0.001a
|
Imaging feature data
|
|
|
|
|
bilateral pneumonia
|
343(80.90%)
|
198(82.50%)
|
145(78.80%)
|
0.489a
|
partial patchy shadows
|
247(58.25%)
|
137(57.08%)
|
110(59.78%)
|
0.523a
|
multiple patchy shadows
|
216(50.94%)
|
119(49.58%)
|
97(52.72%)
|
0.577a
|
hydrothorax
|
204(48.11%)
|
119(49.58%)
|
85(46.20%)
|
0.338a
|
CT-SS, median(IQR)
|
14.29(7-21)
|
10.70(5-15)
|
18.99(13-25)
|
<0.001a
|
Post resuscitation care
|
|
|
|
|
Hormones use, n(%)
|
347(81.84%)
|
187(77.92%)
|
160(86.96%)
|
0.017b
|
MV, n(%)
|
317(74.76%)
|
147(61.25%)
|
170(92.39%)
|
<0.001b
|
CRRT, n(%)
|
87 (20.52%)
|
28 (11.67%)
|
59 (32.07%)
|
<0.001b
|
Length of hospitalization (d)
|
22.81±17.21
|
21.32±17.30
|
24.74±16.94
|
<0.001a
|
Note: aMann-Whitney U test, bPearson χ2 test. D-Dimer elevation: D-Dimer>5000μg/L; Hormones use: use Dexamethasone; Lymphopenia(ly-mphocyte<0.8*10⁹/L).
Abbreviations: TBil: total bilirubin; Cr: creatinine WBC: white blood cell; PCT: procalcitonin; CRP: c-reactive protein; IL-6: Interleukin-6; PF: PaO2/FiO2, ratio of the partial pressure of oxygen to the fraction of inspired oxygen; SOFA: sequential organ failure assessment; IQR: interq-uartile range; BP: bilateral pneumonia; PPS: partial patchy shadows; MPS: multiple patchy shadows; MV: mechanical ventilation; CRRT: continuous renal replacement therapy; LOH: length of hospitalization.
Multivariate logistic regression analysis for in-hospital mortality
In the multivariate logistic regression, older age (OR 3.049, 95%CI 1.761-5.279, P<0.01), obesity (OR 3.177, 95%CI 1.417-6.855, P<0.01), diabetes (OR 1.820, 95%CI 1.053-3.145, P<0.05), SOFA score (OR 1.113, 95%CI 1.005-1.234, P<0.05), CT-SS (OR 1.412, 95%CI 1.102-1.184, P<0.01), mechanical ventilation (OR 3.030, 95%CI 1.243-7.386, P<0.05), and lymphocytopenia on admission day (OR 2.094, 95%CI 1.417-6.855, P<0.01) independently predicted in-hospital mortality of patients with critical COVID-19 pneumonia (Table 3).
Table 3. Univariate analysis and multivariate logistic regression for in-hospital mortality
Variable
|
Univariate analysis
|
Multivariate analysis
|
|
OR
|
95%CI
|
P value
|
OR
|
95%CI
|
P value
|
Age(>75years old )
|
1.514
|
1.025-2.236
|
0.037
|
3.049
|
1.761-5.279
|
<0.001
|
Creatinine (μmol/L)
|
1.001
|
1.000-1.003
|
0.025
|
|
|
|
Procalcitonin (ng/mL)
|
1.006
|
0.994-1.017
|
0.332
|
|
|
|
C-reactive protein (mg/L)
|
1.003
|
1.000-1.005
|
0.043
|
|
|
|
Interleukin-6 (pg/mL)
|
1.000
|
1.000-1.000
|
0.146
|
|
|
|
length of hospitalization (d)
|
1.012
|
1.000-1.023
|
0.046
|
|
|
|
Lymphopenia
|
2.535
|
1.676-3.833
|
<0.001
|
2.094
|
1.226-3.578
|
0.007
|
SOFA score, median(IQR)
|
1.234
|
1.164-1.308
|
<0.001
|
1.113
|
1.005-1.234
|
0.040
|
CT severity score, median(IQR)
|
1.154
|
1.119-1.189
|
<0.001
|
1.412
|
1.102-1.184
|
<0.001
|
Comorbidities
|
|
|
|
|
|
|
Diabetes, n(%)
|
1.622
|
1.077-2.443
|
0.021
|
1.820
|
1.053-3.145
|
0.032
|
Cerebrovascular disease, n(%)
|
1.692
|
1.010-2.833
|
0.046
|
|
|
|
Chronic renal failure, n(%)
|
2.176
|
1.178-4.021
|
0.013
|
|
|
|
Obesity, n(%)
|
2.499
|
1.349-4.628
|
0.004
|
3.117
|
1.417-6.855
|
0.005
|
Post resuscitation care
|
|
|
|
|
|
|
Hormones use, n(%)
|
1.889
|
1.116-3.199
|
0.018
|
|
|
|
Mechanical ventilation, n(%)
|
7.682
|
4.201-14.049
|
<0.001
|
3.030
|
1.243-7.386
|
0.015
|
ROC curves and predictive model with regards to mortality
For prediction of in-hospital mortality, lymphocytopenia had an AUC of 0.623 (0.570~0.676) with a cut-off value of 0.755, with a sensitivity and specificity of 92.4% and 46.7%, respectively; SOFA score on admission with a cut-off value of 8.5 had an AUC of 0.706 (0.657~0.754), with a sensitivity and specificity of 81.5% and 50.4%, respectively; Similarly, CT-SS on admission with a cut-off value of 12.5 had an AUC of 0.790 (0.748~0.833), with a sensitivity and specificity of 77.7% and 67.9% , respectively. Notely, once we combinated the SOFA score and chest CT-SS, a significant increase in AUC of 0.826, with a cut-off value of 18.19, a sensitivity of 81.0% and specificity of 70.4% was observed in comparison with them alone (Figure 3). The distribution of SOFA score combined with CT-SS for in-hospital mortality showed that the in-hospital mortality of COVID-19 patients significantly increases with the rising score when the combined score of SOFA and CT-SS ≥18 (Figure 4).