PHOTOILLUSTRATION BY EMILY SCHERER/GETTY IMAGES
It’s 6:30 am and your third alarm is ringing. Another 10 minutes and you’ll have to skip breakfast before you go to work. You get up, brush your teeth, shower, and set out a pot of coffee while you watch the morning news. After the station meteorologist tells you it’s sunny with a high of 70 degrees, a health reporter follows the COVID-19 forecast: red alert today with a 75 percent chance of exposure in rooms with at least 50 people.
You respond to the risk by taking a quick test – taking a photo of the negative result for your colleagues – before you leave the house – and grabbing your N95 mask.
This is a hypothetical future, but one that a new team at the Centers for Disease Control and Prevention is working on. That Center for forecasting and outbreak analysis received initial $200 million funding from the American Rescue Plan to hire data scientists, epidemiologists, and science communicators who will create forecasts for the public and drive data-driven decision-making. CFA aims to operate”like the National Weather Service, but for infectious diseasessaid epidemiologist Caitlin Rivers, associate director of science at the new center, in an April article in the Washington Post.
Which would be great – except the forecasters now have less data to work with than they’ve had in years. On the one hand, COVID-19 case data is becoming less and less reliable thanks to rapid tests. In addition, hospital data may lag behind transmission trends and may become unavailable in the coming months. And new sources like wastewater monitoring are not yet ready to replace clinical data. Imagine making a weather forecast without reliable temperature or humidity readings.
Two converging trends are causing case data to become less useful. First, more Americans are using rapid home tests, and second, fewer Americans are using lab-based PCR tests.
According to Mara Aspinall, a diagnostic testing expert and public health consultant, home testing is about six times as common as PCR testing in the United States. Using data from test manufacturers and retailers, Aspinall has been closely monitoring testing capacity in the United States throughout the pandemic, and at this time last year Americans were performing significantly more PCR testing than home testing.
Unlike PCR test results, which are automatically reported by the labs that process the tests, the vast majority of home test results are not reported to health authorities. In recent months, PCR tests have also become less available. states like Vermont and Colorado close free public testing sites, instead directing patients to home testing and private healthcare providers — even as government funding for testing for uninsured individuals has expired.
“The official numbers we hear about on a daily basis in terms of official COVID-19 cases are becoming a growing understatement of the true number of infections circulating,” said Jason Salemi, an epidemiologist at the University of South Florida College of Public Health.
This growing undercount is creating more uncertainty in the COVID-19 forecast, said Marc Lipsitch, an infectious disease expert at Harvard and interim director of science at CFA. As the quality of case data deteriorates, “we rely more on the lagged indicators like hospitalizations and deaths,” he said.
CFA scientists are currently focusing on hospitalization data, which US modelers believe to be more reliable than case counts: Lipsitch described it as the best source that is “reported routinely and consistently.”
Hospital admissions — the number of new COVID-19 patients admitted for treatment — can be particularly useful, said Lauren Ancel Meyers, director of the University of Texas at Austin’s COVID-19 Modeling Consortium. This metric provides “a more stable signal” than case counts, although it usually takes a week or two to show an increase after an increase.
However, when health authorities rely on hospital data as the main source for policy decisions, “they lose valuable time trying to prevent morbidity and mortality,” Salemi said. Essentially, by the time hospitalizations show a meaningful increase, it is already too late to prevent a spike that will lead to more people becoming seriously ill – and it will lead to a similar spike in long COVID, which is not visible at all in the hospital data.
Lipsitch agreed that this delay is an issue. Hospitalization data reflect recent spread of COVID-19; When data about the past is used to identify trends in the present, “the present becomes more ambiguous because the most timely signal was also the lowest quality.”
Additionally, Salemi said, hospital admission metrics face a geographic issue. While people living in any US county can contract SARS-CoV-2, not every county has a hospital. For example, Alachua County, Florida has a large health care system that serves Floridians from several smaller, nearby counties. Most hospitalizations from this region occur in Alachua County, not the surrounding areas.
Even the imperfect hospital data may not always be available. When the Biden administration ends the national public health emergency for COVID-19, hospitals and local health departments may no longer be required to report their data to federal agencies. The emergency declaration will continue beyond at least July 15but it’s unclear how long the administration will extend it.
And if the data becomes even less reliable, what then? CFA is looking for alternatives, and effluent monitoring is a top priority. Our sewers can provide more recent data than hospitalizations and less biased data than cases. But wastewater is a long way from a national surveillance system: COVID-19 levels are not being tracked in water systems in much of the country, it takes time and resources to set up that tracking, and public health officials are not yet sure how to do it should use the data as soon as they have it.
Lipsitch also wants to find a way to pinpoint who in a community, with data at an individual level, not at a population level like sewage. Ideally, he would like to see the CDC conduct population surveys and interview a representative subset of Americans to estimate who has COVID-19. Such surveys could help provide accurate estimates of how many people are currently infected in a given state or county, and provide projections of a person’s likelihood of exposure when they go to work or school. British modelers are considering similar surveys carried out by the UK Office for National Statistics much more reliable than case numbers, Lipsitch said. But there is no comparison for these polls in the US
The CFA’s talks about potential public health data sources and actions are largely theoretical at this point, as the center still is a team of only four people – three of them are “borrowed” from their academic positions, as Lipsitch put it. Publishing job descriptions for the center’s eventual 100 scientists and communicators “is taking longer than we had hoped,” he said. From the beginning of June just a job posting is open for applications.
As it works to scale, CFA faces a major challenge: the CDC’s lack of authority over state and local health officials. “Outside of a public health emergency, the CDC has no authority to require states to share data,” Lipsitch said. And even during the COVID-19 emergency, the agency was unable to request disclosure of some key metrics, such as: breakthrough cases and deaths.
The CDC also has limited authority over Americans as a whole, and trust in the public health system is waning. A recent survey found that nearly a third of the country believes the pandemic is already overthroughout Congress consistently fails to allocate new funds for COVID safety measures. These could be worrying signals for the future of the new prediction center: even if the CFA is able to provide timely, accurate infectious disease forecasts like we have for the weather, how many people will actually act on the information?
