Skip to main content

Association between biomarkers and COVID-19 severity and mortality: a nationwide Danish cohort study

Our take —

This nationwide retrospective study of 1310 adults 18 years and older who were hospitalized in Denmark with laboratory-confirmed SARS-COV-2 infection revealed a significantly higher age- and sex-adjusted risk of admission to intensive care units and mortality among those with elevated levels of several clinical biomarkers, including D-dimer, Procalcitonin, serum troponin, leukocyte counts, and C-reactive protein, as well as among those with stage III and IV chronic kidney disease. The large sample size in addition to access to well-characterized registry data and standardized measurement of serum biomarker concentrations is a notable strength of this study. However, lack of information about the timing of biomarker measurement relative to symptom onset could have resulted in a spurious association if the levels of these biomarkers were more strongly related to the timing of disease onset. Furthermore, given that the study focused on hospitalized individuals, findings may not be generalizable to those with less severe COVID-19 or those in outpatient settings.

Study design

Prospective Cohort

Study population and setting

This retrospective cohort study evaluated the association between different clinical inflammatory biomarkers and the composite risk of intensive care unit admission (ICU) or death among adults (18 years and older) hospitalized with laboratory-confirmed SARS-CoV-2 infection hospitalized between February 27 and May 1, 2020. The study used data obtained from four large national registries in Denmark that are linked via unique patient identification numbers. Clinical biomarkers included serum CRP, leukocyte count, ALT, Ferritin, D-Dimer, Troponin, Procalcitonin, urea and glomerular filtration rate (eGFR), along with a variety of clinical characteristics obtained within 24 hours of presentation and 30-day composite outcome of intensive care unit admission or death. Biomarker concentrations were categorized as normal, and mildly, moderately and severely elevated using clinically-defined cutoffs, and the absolute risk of the composite outcome (death or 30-day ICU admission) was calculated for each category adjusted for age and sex using standard survival analysis and bootstrapping to obtain confidence limits and test statistical significance.

Summary of Main Findings

1310 participants were included in the analysis. The median age was 73.6 years old (54.6% male). A total of 352 (26.9%) had the composite outcome and 263 (20.1%) succumbed to their illness.

In univariate analysis, patients who experienced ICU admission or death within 30 days were more likely to have pre-existing comorbidities (diabetes, chronic obstructive pulmonary disease, atrial fibrillation, hypertension, heart failure or lower eGFR) and were more likely to take the following medications: aspirin, beta-blocker, loop diuretics, calcium channel blockers, angiotensin receptor blockers or spironolactone.

Participants with moderately elevated CRP (below 99 mmol/l) and severely elevated CRP (above 100 mmol/l) had significantly higher absolute risk of ICU admission or death of 21.5% (95% CI: 18.1%-24.9%) and 39.2% (95% CI: 35.6%-43%) respectively, compared to an absolute risk of 5% (95% CI: 0-12%) among those with normal CRP. Patients with moderately elevated leukocyte counts (below 15 cells /L), and severely elevated leukocyte count (15 cells/nanoL-30 cells/ηL respectively) had absolute risks of 34.5% (95% CI: 29.5%-39.4%) and 46.6% (95%CI: 38.5%-54.6%) for having the composite outcome respectively, compared to 23.2% (95% CI: 20.4%-22.2%) for patients with normal white blood cells count. Patients with CKD stage III (Glomerular filtration rate (eGFR) between 30-60 ml/min) and those with eGFR< 30 ml/min had significantly higher absolute risk for the composite outcome compared to those with eGFR>60. ( 41.1%(95% CI:35.1%-48%) for those with eGFR 30-60ml/min, 45.9% (95% CI: 34.9%-56.8%) for those with eGFR<30ml/min, and 30.4% (95% CI: 26.7%-34.1%) for those with eGFR>60ml/min) .

Patients with elevated D-dimer (above 0.5 mcg/ml) had higher absolute risk for 30-day mortality or ICU admission compared to those with normal D-dimer (31.8% (95% CI 26.7%-36.8%) vs. 17.5% (95% CI 10.9%-24.1%)).

Patients with elevated cardiac troponin had a significantly higher risk for 30-day mortality and ICU admission compared to those with normal cardiac troponin, with higher risk for those who’s level above 2 times the upper limit of normal (>2 ULN, 57.3%(95% CI43.3%-71.3%), above ULN but <2 ULN, 27.7%(95% CI 16.5%-38.9%), and 9.4%(95% CI(4.2%-14.5%) for those with normal troponin level). Furthermore, those with Procalcitonin level above 0.5 mg/L had an absolute risk of 52.1% (95% CI 41.5%-62.6%) for 30-day mortality or ICU admission compared to 28% (95% CI: 21.1%-34.9%) among those with normal Procalcitonin.

While elevated urea and Alanine aminotransferase (ALT) levels were associated with a trend toward greater absolute risk of either ICU admission or death, the risk difference between those with elevated and normal levels of each did not reach statistical significance. Elevated ferritin was associated with a higher risk for having the composite outcome until 300μg/L, after which the trend declines (an inverted U-shaped association), although this difference did not reach statistical significance.

Study Strengths

The study used a large, nationwide cohort of adults with laboratory-confirmed SARS-CoV-2 infection who were admitted to a hospital in Denmark. The data included in the study were identified using 4 large Danish databases that are well ascertained, which would minimize the risk for misclassification bias.

Limitations

Only about 250 (19%) of the total included participants in the study had laboratory biomarkers for ferritin, D-dimer, troponin and Procalcitonin. Therefore, the study outcomes pertaining to these variables might have been biased if the rest of the cohort had significantly different values from the measured ones. Also, the data lacks important clinical variables pertaining to hospital admission such as hypertension, diabetes, BMI, and smoking history. Moreover, the biomarkers were measured within 24 hours of the patient’s admission to the hospitals, without including information that relates to the time of symptom onset. It is plausible that different levels of study biomarkers among the cohort could be partially attributed to the differences in time from symptoms onset to medical seeking. Results were only age- and sex-adjusted, which limits the independent casual inference between each biomarker and the composite outcome. The study population consisted of those who were admitted to the hospital; therefore, results may not be applicable to those with less severe clinical disease and those in outpatient settings.

Value added

The study adds critical knowledge to the literature regarding the relationship between different levels of clinical biomarkers and the risk of ICU admission or death within 30 days of seeking medical attention.

This review was posted on: 27 January 2021