UGA professor develops tool to quickly predict COVID-19 severity

A University of Georgia researcher has developed tools to help clinicians quickly determine which COVID-19 patients are likely to need intensive care.

While some of these types of tools, known as clinical prediction rules, exist for COVID-19, they require laboratory tests that take a long time to get back or may only be available in an emergency room.

“You want to be able to make this triage decision quickly,” said project lead Mark Ebell, an epidemiologist in UGA’s College of Public Health.

“Right now as we get into the winter where in some states, like California, they’re really getting overwhelmed in the hospitals, and they have to evaluate a patient in the emergency department and make a decision – is this person safe to go home, is this somebody we need to admit and watch for 23 hours, or is this somebody we want to admit and possibly be sent directly to the ICU?” said Ebell.

Ebell and his collaborators aimed to develop rules that used either no labs or labs that could be run in under an hour.

Using de-identified patient data from six major university hospitals across the U.S., the team collected information on patient age, co-morbidities, basic biomarkers, and how the patient fared with COVID-19.

The team ran statistical analyses to learn which of these variables were the best independent predictors of a patient having a severe case. Age was by far the dominant predictor of severe cases, said Ebell. Having asthma was also a strong predictor.

Breathing rates and oxygen saturation are key predictors for both simple lab and no lab rules. Additional lab tests that assess inflammation, kidney function, and white blood cell count are features of the simple lab rule.

Ebell believes these new rules may be particularly useful to clinicians who are managing patients via telemedicine.

“The physician doesn’t have a chance to listen to their lungs or examine them, and that’s why the rule that we developed that doesn’t require labs may be particularly relevant for that setting because it can help a physician identify patients who may be at risk for deteriorating and might need to be referred to the emergency department for a more comprehensive work up,” he said.

The rules have been internally validated, meaning that 60% of the patient data were used to develop the rule, and the remaining 40% to test its accuracy. Ideally, the rules still need to be evaluated using a new group of patients.

Ebell and his team are applying for funding to support this next step.

A paper outlining the new clinical decision rules was published in the Journal of the American Board of Family Medicine in January. Read it online here:

– Lauren Baggett

Posted on January 20, 2021.