Elevated glucose level leads to rapid COVID-19 progression and high fatality

We established a retrospective observational study cohort, based on 3,059 cases admitted to the Huoshenshan hospital in Wuhan between February 4 and April 15, 2020. The exclusion criteria were: (1) Patients who were not confirmed by a positive result of severe acute respiratory syndrome coronavirus 2 detection in respiratory specimens by the reverse transcriptase polymerase chain reaction assay, or in serum by the specific IgM and IgG antibody detection; (2) Patients who referred to other medical institution during hospitalization; (3) Patients who were admitted to the hospital multiple times; (4) Patients were younger than 18 years old; (5) Patients without laboratory data included in this study within the 24 h after admission. This study tracked the progression of COVID-19 patients from admission until one of the endpoints (discharged or death).

Demographic, clinical, laboratory, treatment, and clinical outcome data were obtained from the hospital’s electronic clinical medical records. At the first clinical consultation, demographic, clinical and laboratory data were collected within the first day after admission. Treatment data and clinical outcomes (including the event of disease progression, time of each disease stages, fatality, duration of hospitalization and endpoint status) were also collected during the course from admission to the study endpoints. We defined the event of disease progression as a mild or moderate patient at admission would progress to severe or critical stage at the first time during hospitalization.

The severity of COVID-19 was defined according to the Guidance 7th edition [20]. Patients were classified as ‘mild’ if there was no evidence of pneumonia on imaging nor any of the features for moderate or higher severity; as ‘moderate’ if they had evidence of pneumonia on imaging but no features of severe or higher severity; as ‘severe’ if they meet any of the following criteria: (1) respiratory distress (≥ 30 breaths/min); (2) oxygen saturation ≤ 93% at rest on room air; (3) arterial partial pressure of oxygen (PaO2) or fraction of inspired oxygen (FiO2) ≦ 300 mmHg (l mmHg = 0.133 kPa); and as ‘critical’ if they required mechanical ventilation, had a septic shock or required admission to ICU. Comorbidities were defined according to ICD10-CM code [21]. Detailed definitions for clinical symptoms were provided in the supplemental materials. We considered a patient progressing to a severe or critical disease stage when the individual had none of the severe or critical stages at admission but developed these stages for the first time during hospitalization.

Patients had to meet all the following criteria before being discharged: (1) body temperature returned to normal (< 37.5 °C) for three consecutive days; (2) respiratory symptoms improved substantially; (3) pulmonary imaging showed an obvious absorption of inflammation; and (4) two consecutive negative nuclei acid tests, each at least 24 h apart.

We presented continuous variables as the median and interquartile range (IQR) and examined the differences between disease severity groups using the Kruskal–Wallis one-way ANOVA We presented categorical variables with the corresponding percentage and examined the differences using χ² test or Fisher’s exact test. We conducted survival analyses on disease progression and fatality based on a competing risk framework. The outcome variables included: (1) the event of the first time disease progression to severe or critical disease states among mild or moderate patients at admission, and (2) in-hospital fatality among patients with severe or critical at admission. Discharge from the hospital was considered as a competing risk event. Five clinical indicators (interleukin-6, natriuretic peptide type B, supersensitive troponin I, myoglobin and procalcitonin.) with more than 30% of entries missing were excluded from the analysis. Data imputation was performed if missing percentage < 30% using Multivariate Imputation by Chained Equations. Statistically significant variables in the univariate analysis were ranked and further selected using LASSO regression [22]. The number of variables was defined as the number of variables when λ = λmin in LASSO, or the total number of events divided by 10 (event per variable > 10 rule)[23], whichever is smaller. Variables, which particularly reported in previous literatures were included in final analysis. The pooled set of variables were then included for the final multivariable cause-specific Cox proportional hazard model. Cumulative incidence curves were plotted to demonstrate the incidence of differences between different risk levels of key variables. A p value of < 0.05 was considered statistically significant. Statistical analyses were conducted using the R software (version 3.6.1).

After excluding 214 patients who were only diagnosed clinically according to the Guidance 7th edition [20], 46 patients who referred to other medical institution, 16 patients who were admitted to the hospital multiple times, six patients were younger than 18 years old. Further, we excluded 328 patients without laboratory data included in this study within the 24 h after admission, we included 2,433 COVID-19 patients in the final analysis. Fifty patients died during hospitalization, and 2,383 were discharged, corresponding to a case-fatality ratio of 2.1%. Patient’s median age was 60.0 years (IQR 50.0–68.0), and 50.2% were male (Table 1). The most common symptoms or signs on admission were cough (55.7%), fatigue (38.9%), and shortness of breath (25.0%). Hypertension being the most common comorbidity (31.6%), followed by diabetes (14.3%) and coronary heart disease (6.7%). During hospitalization, 847 (34.8%) patients received antibiotics, 1233 (50.7%) received antivirals, and 68 (2.8%) received non-invasive mechanical ventilation and 42 (1.7%) received invasive mechanical ventilation.

Among 25 mild patients at admission, 19 retained mild and were discharged after 6.0 (5.0–11.0) days; six patients who progressed to moderate severity in 3.0 (1.8–5.5) days, but all discharged after another 8.0 (6.8–8.8) days. Of 1,733 moderate patients at admission, 1,259 patients retained moderate and discharged after 11.0 (7.0–16.0) days. In contrast, 474 patients progressed to the severe state in 3.0 (1.0–7.0) days, but all recovered and were discharged after another 12.0 (6.5–18.0) days, and 9 patients deceased after 9.0 (2.5–19.0) days. Of 635 severe patients at admission, 604 patients regressed to moderate severity in 7.0 (5.0–11.0) days and were discharged after another 6.0 (4.0–11.0) days. Thirty-one patients progressed to critical severity after 3.0 (1.8–8.0) days, and of whom only 10 patients were discharged after 15.5 (14.0–30.5) days, and 21 died after 6.0 (0.0–10.5) days. Of 40 critical patients at admission, 20 patients regressed to moderate severity after 10.5 (8.3–15.8) days and were discharged after another 11.5 (6.3–19.5) days, and the remaining 20 patients died after 6.5 (4.0–16.3) days (Figs. 1 and 2).

Across all patients, it required a median of 3.0 (1.8–5.5) days to progress from mild to moderate, 3.0 (1.0–7.0) days from moderate to severe, 3.0 (1.0–8.0) days from severe to critical and 6.5 (4.0–16.3) from critical to fatality. In contrast, it required 7.0 (5.0–11.0) days to regress from severe or critical to moderate severity and 6.5 (4.0–11.0) days from moderate to discharge. The median admission-to-discharge time among mild, moderate, severe and critical patients on admission were 7.0 (5.5–12.0), 13.0 (9.0–19.0), 15.0 (8.0–22.0), and 25.0 (15.3–34.8) days, respectively.

Of 1,758 mild and moderate patients at admission, 474 (27.0%) progressed to severe or critical severity during hospitalization. Multi-variable cause-specific Cox proportional hazard model (Table 2) identified that patients with age 60–74 years (HR = 1.26, 95%CI 1.02–1.56), > 74 years (1.44, 1.02–2.03), respiratory rate > 20 times/min (1.28, 1.05–1.57), fever (temperature ≥ 37.5 °C) (1.93, 1.21–3.08), chest tightness (1.47, 1.12–1.92), blood glucose > 6.1 mmol/L (1.58, 1.25–1.98), c-reaction protein > 4 mg/L (1.45, 1.12–1.87), lactate dehydrogenase > 250 IU/L (1.63, 1.20–2.20), direct bilirubin > 8 μmol/L (1.51, 1.03, 2.21), albumin < 40 g/L (1.38, 1.07–1.77) and lymphocyte count < 1.1*10⁹/L (1.44, 1.15–1.81) were risk factors for disease progression to severe and critical stage (Table 2). The 21-day cumulative incidence of progression was 47.8% in > 74-years age group, followed by 32.4%, 19.8% in age groups of 60–74 and < 60-years respectively. The cumulative incidence of disease progression at day 21 was also much higher in patients with blood glucose ≥ 6.1 mmol/L (40.8%) than blood glucose in range of 3.9–6.1 mmol/L (21.4%, Fig. 3a, b).

Of 675 severe or critical patients at admission, 634 (93.9%) were discharged and 41 (6.1%) died during hospitalization. Table 2 showed that patients with age > 74 years (3.41, 1.07–10.89), blood glucose > 6.1 mmol/L (3.22, 1.54–6.73), platelets count < 125*10⁹/L (4.39, 2.02–9.54), fibrinogen < 2 g/L (6.48, 1.46–28.67) and creatine kinase-MB > 24 IU/L (6.29, 2.51–15.80) were risk factors for in-hospital fatality (Table 2). The 21-day cumulative incidence of fatality was 13.1% in > 74 years age group, 4.6%, 1.4% in age groups of 60–74, < 60 years respectively. The incidence of fatality at day 21 was four times higher in patients with blood glucose > 6.1 g/L (11.8%) than blood glucose in range of 3.9–6.1 mmol/L (2.7%, Fig. 3c, d).