Utility of Procalcitonin Levels in Identifying Bacterial Infections in Children With and Without COVID-19

Abstract

Rationale: Elevated procalcitonin levels are associated with increased rates of bacterial infection in children. Observational studies have reported high levels of procalcitonin in COVID-19 and correlated procalcitonin level with illness severity. Data on rates of bacterial coinfections in COVID-19 are sparse;small studies suggest a low coinfection rate. We aim to quantify the positive predictive value (PPV) of procalcitonin in identifying bacterial infection in pediatric patients with and without COVID-19. Methods: A retrospective chart review was performed of 215 pediatric patients age 1 month to 21 years admitted to our tertiary children's hospital between February 1, 2013 and July 15, 2020 who had procalcitonin levels measured within 48 hours of admission. Confirmed bacterial infection was defined as positive blood, urine, or CSF culture, positive endotracheal/sputum culture with significant leukocytosis on gram stain, or CXR consistent with pneumonia. Contaminated blood cultures were excluded. Suspected bacterial infection was defined as either confirmed bacterial infection or administration of therapeutic antibiotics for >48 hours. Results: Of the 215 patients, 73 were PCR and/or antibody positive for SARS-CoV-2 (53% admitted to PICU, 29% required vasoactive medications) and 142 had alternative diagnoses (32% admitted to PICU, 9% required vasoactive medications). Median procalcitonin level was 1.28ng/mL in the COVID-19 cohort and 0.37ng/mL in the control cohort. There were 10 suspected and 5 confirmed bacterial infections in the COVID-19 cohort and 69 suspected and 44 confirmed bacterial infections in the control cohort. Procalcitonin was greater than 1ng/mL in 39 (53%) of COVID-19 subjects and 44 (31%) of control subjects. PPV of procalcitonin level over 1ng/mL for identifying suspected bacterial infection in the COVID-19 cohort was 10% (CI 1-20%) and in the control cohort was 75% (CI 62%-88%). PPV of procalcitonin level over 1ng/mL for identifying confirmed bacterial infection in the COVID-19 cohort was 3% (CI 0%-8%) and in the control cohort was 55% (CI 40%-69%). Conclusions: Our retrospective review demonstrated a low PPV of elevated procalcitonin level above the established threshold of 1ng/mL in identifying either culture-confirmed or suspected bacterial infection in pediatric patients hospitalized with COVID-19. Despite 29% of the COVID-19 cohort requiring vasoactive support, indicating a high pretest probability of bacterial sepsis, procalcitonin performed poorly as a marker of bacterial infection. Our results suggest that COVID-19 infection is associated with an elevated procalcitonin level independent of presence of bacterial infection. Given the importance of antibiotic stewardship, alternative means of identifying bacterial coinfection in COVID-19 are needed.