EHR Reveals Portrait of Youths Hospitalized With COVID-19

An analysis of electronic hospital records (EHR) provides a clearer picture of the characteristics of children severely sickened by the novel coronavirus around the world, including laboratory test results and clinical findings that will hopefully advance medicine’s understanding of the course of COVID-19 in the pediatric population.

A multinational team led by Florence Bourgeois, MD, MPH, associate professor of pediatrics at Boston Children's Hospital, and Paul Avillach, MD, PhD, assistant professor of biomedical informatics at Harvard Medical School and assistant professor of pediatrics at Boston Children's Hospital, examined electronic health data culled from 27 hospitals in France, Spain, Germany, the UK, the US, and Singapore. Patients, all of whom had a positive PCR test for COVID-19, ranged in age from under 2 to 20 and had been hospitalized between February and October of 2020.

The facilities involved in this study were participants in the Consortium for Clinical Characterization of COVID-19 by EHR, or 4CE, a global effort encompassing hundreds of adult and pediatric hospitals that collects EHR information on 39,200 patients hospitalized with confirmed Covid-19. The consortium’s data has informed various studies on the clinical complications of COVID-19, the predictive value of laboratory tests when it comes to disease severity, and differences in COVID-19 patients from country to country. The authors of this study aimed to use 4CE’s EHR data on pediatric patients hospitalized with COVID-19 to discern patterns and identify typical markers of disease in this population, which has been less impacted by COVID-19 than adults and thus generated more limited data.

The 671 patients in the study were almost evenly split between male and female (52% to 48%). The highest proportion of patients were either 2 years old or younger (30%) or 12 to 17 years old (25%). The study participants typically had abnormal lab results upon admission, with the mean C-reactive protein level at 83 mg/L, ferritin at 417 ng/mL, and procalcitonin at 1.45 ng/mL, all suggesting elevated levels of inflammation and/or coagulation. White blood cell and neutrophil counts, however, were normal. Dimerized plasmin fragment D, or D-dimer, which can be used to judge the risk of blood clots, was elevated to a mean of 0.78 ug/mL. The mean level of fibrinogen, also a factor in clotting, was 477 mg/dL. Fifty-nine patients had high cardiac troponin levels (mean 0.032 ng/mL), suggesting heart injury, and 15% experienced cardiac arrhythmias. Viral pneumonia was present in 13.3% of patients and respiratory failure in 10.5%.

In general, abnormally high inflammation and coagulation markers declined rapidly within the first week of hospitalization, although some children experienced elevated kidney and liver function markers during their second week of hospitalization. With few exceptions, children were not treated with therapeutics administered to adults hospitalized for COVID-19 because these remedies were not fully approved and were prescribed based on limited data. Other drugs were tested in trials closed to children.

In the US and the UK, study patients most often were white or of an unspecified “other” race, followed by Black, Asian, and—in the US—Latino. Singaporean patients typically were Asian, while race was not recorded in France, Germany, or Spain.

Studies frequently have revealed differences in COVID-19 outcomes by race, something into which the authors were hoping to delve more deeply. “What we realized is that for the multinational work we are conducting, new ontologies are needed to meaningfully classify race across countries and regions,” Bourgeois told Contagion. “For example, the usual classification we use in the U.S., which includes Native American and Pacific Islander, is not applicable to populations in the UK. Conversely, patients in countries such as Singapore have a number of ethnic and racial subgroups that are not available in our US-based racial classifications.” The team is aiming to use different methodologies that will help them and others in the field capture the nuances of racial data across countries.

Overall, the study provides support for the use of EHR data to help understand the trajectory of COVID-19 in the under-21 population. “‘[We have demonstrated] the feasibility of using large-scale informatics-based approaches to integrate data across healthcare systems and provide multinational analyses of the epidemiological and clinical features of COVID-19 in children,” Bourgeois said.