clock menu more-arrow no yes

Filed under:

It’s impossible to count everyone with COVID-19

The best public health experts can do is an accurate estimate

Illustration by Alex Castro / The Verge

Zack Moore, state epidemiologist in North Carolina, can’t say how many people in the state have COVID-19. He knows how many people test positive for the coronavirus each day, and how many people are in the hospitals across the state, and how many people go to the emergency room with an illness that could be COVID-19. None of those numbers alone pins down how many people are infected. But together, they help sketch an outline of the pandemic.

“It’s never about one single data source. Every source of data is helpful, but they all have their own limitations. So it’s about using them together,” he says.

There’s no way for public health experts to actually count every single person who has any illness at a given time, even well-known illnesses like the flu. Instead, they combine different sources of data to make the best-possible estimate of what disease outbreaks look like. The goal is to understand the virus well enough to make those estimates as accurate as possible.

“People think we know how many people had the flu every year, and we don’t,” Moore says. “We have to use these surveillance tools and indirect measures to get a full understanding.”

Public health officials are measuring the COVID-19 pandemic in a handful of ways. The number of positive test results is an important one, says John Brownstein, an infectious disease epidemiologist and chief innovation officer at Boston Children’s Hospital. “That data is incredibly valuable, because it’s confirmed illnesses,” he says. But it’s an imperfect tool because most communities in the United States don’t have enough tests available to test everyone who feels sick.

Because everyone who’s sick won’t get a test, experts also look to data collected by hospitals on the people who show up to an emergency room with symptoms of COVID-19. Those measurements also have limitations because many people who have these symptoms might just have the flu or another type of respiratory disease. This approach also misses people who may feel sick but don’t feel sick enough to go to the hospital.

Zooming out even further, some researchers (including Brownstein) have developed programs that ask people to report how they’re feeling each day. They hope that the data will help reveal hotspots where people are starting to feel sick but aren’t getting tested for COVID-19 yet. Another big-picture strategy is measuring levels of the coronavirus in sewage to predict how many people in a community are infected. Antibody tests, which can check if someone was exposed to the virus in the past, are also critical tools. They can help find people who never felt sick and so wouldn’t be caught by any of the surveillance methods that screen for symptoms.

“We’re constantly trying to bring together different pieces of information, and it all provides a different lens on an outbreak,” Brownstein says.

In North Carolina, Moore looks at all of their data sources daily and weekly and focuses on the trends in that data. If the number of people coming to the emergency room with symptoms of COVID-19 is going down, but the hospitalization rate and number of positive test results are still going up, it might just mean that people are avoiding the emergency room — not that fewer people are sick. “It has to be viewed in the bigger context,” he says. “The trends are what we’re focused on, because that’s what helps us in terms of making decisions about control measures.”

Moore says they’re looking closely at the surveillance tools they’re not already using to see what might be a useful addition to their data sets. “We want to understand whether they’re really value added,” he says. “We don’t want to create more noise — surveillance can already be confusing enough.”

Ongoing research on the virus and how it spreads also help experts figure out the best way to interpret the data they’re collecting. Studies that collect blood samples to test for antibodies in segments of the population, for example, will help identify the percentage of people who catch the coronavirus without ever feeling sick. With that data, public health officials will be able to more accurately estimate the number of people who actually have the virus using the information they gather on people who report symptoms. Moore says he uses an iceberg metaphor: “What we can see is the people who come out and seek medical care. The part under the water is the people who don’t, who we need to learn more about.”

With well-known diseases, public health experts can look at the number of confirmed cases and predict how many cases went undetected. “When we see a single case of hepatitis C, I can tell you pretty confidently that there were 14 other people who probably also had it and weren’t identified,” Moore says. “We’re trying to get that level of understanding for COVID-19.”

Brownstein thinks we’re getting closer. “The further this goes on, the more we’ll be able to get a better estimate.”