CDC to cut back on Covid data it collects as public health emergency ends

The ending of the public health emergency for Covid-19 will force the Centers for Disease Control and Prevention to cut back on the data it collects to analyze how much illness the SARS-CoV-2 virus is causing in the future, the CDC announced Friday.

But some of the data metrics the CDC will no longer be able to collect are of lesser value now than they were at earlier points in the pandemic, agency officials explained in a press briefing Thursday in advance of Friday’s release of two articles in the CDC’s online journal Morbidity and Mortality Weekly Report that detailed the changes.

The CDC will no longer try to track all Covid infections, an effort long since made futile by ubiquitous rapid tests done at home. And hospitals are no longer being asked to report suspected cases of Covid, a useless metric in an era of widespread availability of Covid tests.

“We have the right data for this phase of Covid-19, though our data going forward will be different,” said Nirav Shah, CDC’s principal deputy director. “They will continue to provide timely insights for CDC or local health officials, as well as for the public to understand Covid-19 dynamics at the community level.”

The long-running federal public health emergency was first declared on Jan. 31, 2020, the day after the World Health Organization announced that the new virus that was spreading out of China constituted a Public Health Emergency of International Concern. The U.S. public health emergency has been renewed multiple times in the intervening three-plus years, but will formally end on Thursday.

With its demise go some powers the CDC has been using to require states and territories to supply some Covid metrics.

Some data sources have been drying up for a while; already some locations do not report as frequently as they did previously. And the problem is not simply an American one: Countries around the world have been scaling back for some time on the amount of Covid-related data they collect, Maria Van Kerkhove, the World Health Organization’s technical lead on Covid-19, told STAT in an interview.

“What we have been trying to do is have higher quality information, even if it’s less,” she explained. “And that’s a challenge, because what we want to focus on right now are hospitalization trends, ICU trends, death trends. And at one point we had … I think 130 member states providing information on hospitalizations. And I think in the last month it was around 50.”

As a consequence, the WHO’s Covid hospitalization tracker now posts updates monthly rather than weekly, she said.

In many respects, the CDC’s transition to a more concise basket of metrics is merely a reflection of the reality that the Covid pandemic is in a different phase.

By this point through vaccination, infection, or both, most people have acquired some degree of protective immunity — enough immunity, at any rate, that for most a bout of Covid is no longer life-threatening.

Cécile Viboud, an infectious diseases epidemiologist at the National Institutes of Health’s Fogarty International Center, agreed that data collection needs have changed.

In an interview last month, Viboud said her group, which has modeled predictions of where Covid activity is going, switched from cases to hospitalizations as the basis of its Covid projections about a year ago, because counting people in hospitals who test positive for Covid is a more reliable metric at this point.

That is the way the CDC is now headed, with hospitals across the country reporting hospitalizations on a weekly basis, rather than the daily reports currently required. Shah said focusing on how many people are being hospitalized for Covid is a “strong indicator of Covid-19 status at the local level.”

Other surveillance approaches the CDC will continue to use going forward are wastewater testing (which is not being conducted in all areas), emergency department visits for Covid, and testing positivity rates from a network of 450 laboratories that feed information to the CDC on a variety of respiratory and intestinal infections.

“Our epidemiological understanding of Covid-19 after the end of the public health emergency will be deeper than what we have for other viruses like [respiratory syncytial virus] and influenza,” Shah said.

He pointed to the hospitalization data as an example. For influenza, hospitalizations are estimated, not counted.

Deaths are another area where Covid data are and will continue to be more in depth than those available for flu and RSV. The latter two are estimated through mathematical modeling. But Covid deaths are counted, using information from death certificates.

Bob Anderson, chief of the mortality statistics branch at CDC’s National Center for Health Statistics (NCHS), said despite some earlier reports that coroners in some parts of the country underreported Covid deaths, he is confident the death data the agency amass are solid.

“In talking with coroners broadly throughout the United States — coroners and medical examiners — we really don’t think that that is a widespread practice by any stretch,” Anderson said in an earlier interview with STAT.

“The information flow from death certificates is pretty good, and it’s pretty timely,” he said. “We don’t think we’re missing many.”

There is some lag, though, in the death data reporting. Some states file promptly — daily — with about a three- to five-day lag, but others file their data with a lag of as long as eight weeks, Anderson said. “But generally we were getting pretty timely information. Timely enough that we were running our surveillance, the death surveillance based on death certificates, with about a one-week lag.”

Viboud acknowledged there is a bit more of a lag in death reporting in the NCHS system, “but not a lot. A week or two,” adding that the data are “really, really good, and really nice to have.”

While the CDC’s approach to Covid data is evolving, Shah said the agency is still working to strengthen what he called its data architecture. The CDC was hamstrung in the early days of the pandemic, needing to negotiate data sharing agreements with 64 states, tribes and territories — each of which can take weeks. Without a better system in place, the agency would face the same constraints when the next emergency hits.

Better systems are badly needed, agreed Sheri Lewis, a global disease surveillance expert with the Johns Hopkins Applied Physics Laboratory. The Hopkins group created and ran the university’s Covid data tracker, the go-to data site until the tracker was mothballed in March.

“One of the biggest lessons learned [during the pandemic] is that we have over 2,500 — and that’s obviously an approximation — state local health departments that are collecting data in a non-standardized way,” Lewis told STAT last month. “And so why not think about how we can go about standardizing into … shared definitions, common data categories, so that we can have some type of commonality, if you will. That was the biggest challenge three years ago.”

“I think the Hopkins team will say this over and over: Nobody wants to be where we were three years ago, where we were building the plane as we flew.”