News & Events

Analysis of electronic health records suggests COVID-19 hit Los Angeles as early as December 2019

A recently published study suggests COVID-19 may have been circulating in the Los Angeles area as early as December 2019, months before the first official case in the U.S. was identified. Kathleen Kerr, a professor of biostatistics at the University of Washington School of Public Health, is a member of the research team.

Kathleen Kerr
Kathleen Kerr

“Although the study does not provide conclusive proof that the SARS-CoV-2 virus was circulating in U.S. communities prior to the first confirmed case in the U.S., it adds to a growing body of evidence that this was the case,” says Kerr.

Led by researchers at the University of California-Los Angeles (UCLA), the study analyzed electronic health records and found a significant increase in patients with coughs and acute respiratory failure at UCLA Health hospitals and clinics beginning in late December 2019 and continuing through February 2020.

Researchers focused not only on hospitalization data but also on data from outpatient settings. Kerr’s role was to advise and direct the data analysis.

“The similar findings in data from outpatient clinics, emergency rooms, and hospitalizations is consistent with how the COVID virus behaves,” says Kerr. “While our study was retrospective, it points to a way that electronic health record data could be used for real-time public health surveillance to detect disease outbreaks earlier, enabling a quicker response.  Since an infectious disease like COVID spreads exponentially, every bit of lead time in a response can help prevent many infections.”


Michael Pfeffer, a study co-author and chief information officer for UCLA Health, adds that “The pandemic has really highlighted our need for agile health care analytics that enable real-time symptom and disease surveillance using electronic health records data. Technology, including artificial intelligence powered by machine learning, has further potential to identify and track irregular changes in health data, including significant excesses of patients with specific disease-type presentations in the weeks or months prior to an outbreak.”

The study appears in the peer-reviewed Journal of Medical Internet Research.

Related article: UCLA Newsroom