Pandemic Data Are About to Go Sideways

States are likely to report fewer coronavirus cases, but not because things are getting better.

A family sits down to Thanksgiving dinner, while a red graph line dips and then shoots up.
Getty / The Atlantic
Editor’s Note: The Atlantic is making vital coverage of the coronavirus available to all readers. Find the collection here.

Recently, over the course of just one week, the Houston Health Department received more than 110,000 lab reports of COVID-19 test results. In a city of 2.3 million people, “it’s quite a high volume,” says Beau J. Mitts, the department’s bureau chief. Less than two-thirds of those lab reports flow automatically into the health department’s electronic system, according to Mitts. Another 35 percent arrive in digital form but must be imported into the city’s database, and the remainder arrive via fax.

All over the country, health departments are facing such influxes, and many are struggling to keep pace. The latest surge has earned the terrible distinction of having the highest number of daily cases and hospitalizations since the pandemic began. Now data-reporting delays caused by the Thanksgiving holiday and long weekend may provide a veneer of comfort—a seeming dip in cases—when the actual course of the pandemic in the coming days will almost certainly be much bleaker than the reported numbers show.

Our colleagues at the COVID Tracking Project have monitored how holidays and weekends can affect the daily reporting of tests, cases, hospitalizations, and deaths to public-health departments. As the project’s managing editor, Erin Kissane, wrote this week, the number of new COVID-19 cases reported on a Sunday or Monday—data that are actually collected on Saturday and Sunday—is typically several percentage points below the weekly average. These delays have several causes: Doctors’ offices may be closed or public-health departments may be short-staffed during the weekend, resulting in fewer tests being conducted, fewer positive cases being identified, and less data making their way onto the internet.

But though the reporting infrastructure slows over weekends, the virus does not. Later in the week, the data “catch up,” with higher-than-average numbers on Thursday, Friday, and Saturday.

Holidays tend to bring the same delays as weekends, and Thanksgiving combines a long weekend with a holiday at the pandemic’s worst point yet. If the patterns that the COVID Tracking Project has documented over other holidays hold, in the next week, reporting will slow for a few days, then spike. While no one can say exactly how long labs will take to work through the holiday backlog, the current picture, of rapidly rising testing, cases, and deaths, could be blurred for days at a critical point in this third surge.

On top of the holiday-related data delays, the offices that report coronavirus data could be dealing with a larger volume of paperwork simply because of the size of this surge—which means more people getting tests, more people going into the hospital, and more people dying. While testing has grown appreciably—in the past week, the U.S. reported an average of 1.8 million tests a day, more than double the figure from three months ago—those tests can be slow to process. That’s especially true now, as new cases exceed 170,000 a day and Americans have flocked to testing centers under the misguided notion that a negative test would magically render large Thanksgiving dinners safe. This week Quest Diagnostics, which operates labs and makes COVID-19 testing equipment, said that higher demand is resulting in delays in processing results, to two to three days for most patients.

Jessica Guernsey, the public-health director for Multnomah County, Oregon, says her county began to see changes in the data before Thanksgiving. “It has more to do with the health department being overwhelmed because of sheer volume,” Guernsey says. “The system we’ve set up could be more nimble—to have the data processed and make its way into the statewide system is a fairly laborious process.”

Even with bumps in the data, Guernsey warns, the numbers are going up. “I don’t think most people looking at the data would think we’re in the clear, and none of our messaging reflects that,” she says. “We’re in each other’s care right now, and we need to make decisions that are sacrificial. We’ll have to hunker down for a while to protect other people.”

Because COVID-19 data can vary significantly from day to day, averages over the longer term paint a clearer picture of the disease’s trajectory. Decreases in the numbers of cases or deaths in the days surrounding Thanksgiving won’t be particularly meaningful until we have a wider view of the data, in the form of weekly averages. As Kissane notes, hospitals don’t get days off, so data about hospitalized patients are less volatile than other metrics. Nearly 90,000 people are hospitalized with COVID-19 in the United States, and that number has grown 89 percent since November 1.

Any immediate post-holiday spike, driven by data delays, should not be conflated with the rise in cases that could follow still later, driven by new infections incubated by holiday travel and gatherings. Public-health officials have mostly implored Americans to stay home for the holiday, but for those who do visit Grandma, a COVID-19 infection acquired this weekend could take more than a week to appear in a public-health dashboard, says Angela Rasmussen, a virologist at the Georgetown Center for Global Health Science and Security. “The median time that people test positive is five days after exposure, but it could take up to two weeks.” A person who gets infected late this week might not display symptoms for several days (if at all); scheduling a test, and getting results back, could take several more days. “I would expect to see increases in cases one to two weeks after Thanksgiving. I would expect to see a two-to-three-week lag for hospitalizations, and then another two to three weeks for mortality to increase,” Rasmussen says.

After Halloween, it took about a week to see an increase in cases, says Olivia Kasirye, the public-health officer in Sacramento County, California. “A week from this Thursday is when we’ll see an impact from the Thanksgiving holiday.” In Canada, where fewer than 12,000 people have died of COVID-19, cases rose 33 percent in the three weeks after the country celebrated Thanksgiving in mid-October.

Rasmussen cautions that the current stress on the medical-care system now makes any hospital trip more dangerous. “When facilities and people are put under this amount of strain, it’s not just COVID that is a risk; it’s a heart attack, pancreatitis that requires emergency surgery, a car accident,” Rasmussen says. “You may find yourself in a situation where you are not able to access the kind of care that you usually would. It’s really scary.”

If there is a rise in deaths following Thanksgiving, the evidence of it could be delayed even further, because of the grim math of the disease at this time of year. “Reporting of death is even more challenging,” says Ellie Murray, a Boston University epidemiologist. The bulk of Thanksgiving-related COVID-19 deaths will likely come around Christmas, she says. Confirming and reporting a death takes time, and, Murray points out, the people working in the systems that track these numbers may step back for rest and reprieve during the holiday season. “That could potentially delay the reporting even more,” she says. “We may not really have a full picture of what happened post-Thanksgiving until sometime in January.”