Jason is a data scientist. He’s also the parent of two young children who aren’t yet eligible for Covid-19 vaccines. Since the pandemic began, he’s used his data skills to try to assess what, exactly, the risks are for his family. Are his children at higher or lower risk than his vaccinated parents? Is it safe to send the kids to daycare? What if their teachers are high risk?
Like many, he’s frequently been flummoxed. “I’m a frickin’ data scientist—I do this for a living—but I still have been really surprised by just how hard understanding the basic risks [can be],” he said. (Jason asked to be identified by first name only so he could talk openly about health decisions without being harassed.) “I can see a stat that says, OK, the death rate [for kids] over the past year has been 0.9 out of 100,000 or something like that, but what does that actually mean for my everyday living? How does that compare to other risks that I face on a regular basis? It is overwhelming trying to figure that out.”
In theory, these are the questions the U.S. Centers for Disease Control and Prevention was set up to answer. Yet the agency, while it continues to do vital scientific work, does not seem to be up to the immense task that has fallen on it—especially when it comes to communicating with the public.
Over the past two years, the CDC has had several high-profile missteps. At the outset of the pandemic, faulty and inaccessible Covid tests meant untold numbers of people were never diagnosed—“a huge own goal,” as Jason put it. Then the agency was hesitant to offer advice on masks, leading to persistent unsafe policies in some health systems. The CDC was also slow to acknowledge that the virus spreads through the air. In May 2021, the CDC gave the OK for fully vaccinated people to remove masks indoors, even as the delta wave was emerging. In December 2021, it shortened the guidelines on isolation to five days based on data from the delta variant, even though omicron tends to have a significantly longer infectious period. Most recently, it changed community risk guidelines—which govern recommendations like when masks should be used and when more tests need to be done—to focus on hospital capacity rather than cases, even as thousands of people are dying each week.
These errors in communication can have massive policy implications. “I will literally never forget the day in May of 2021 when President Biden said if you’re vaccinated you can take your mask off,” said Jessica Malaty Rivera, the former science communication lead of the Covid Tracking Project and senior adviser for the Rockefeller Foundation’s Pandemic Prevention Institute. “I wanted to throw my laptop and my phone, everything, out of the window, because I couldn’t believe we had gotten to a place where people were looking at mitigation as one effort canceling out another.”
The new guidance felt political and shortsighted: “It’s almost like we treated the American people like toddlers who needed a cookie for doing one hard thing,” Rivera added, by rewarding those who got vaccinated with going mask-free. “And we are still reeling from that.”
Since the May 2021 mask-dropping guidance, masks have become an even more divisive issue, even among those who support measures like vaccination. Rather than offering a reward for vaccine holdouts, the guidance seemed to confirm what some anti-maskers had long argued: Those still masking were completely irrational. Some states have now banned schools from requiring masks even as new surges and potential variants hover on the horizon.
“Trust is a huge determinant of people’s public health behavior,” Rivera said. “And the trust that continues to be fractured and not repaired only adds to people making their own decisions not based on science or public health guidance.”
When Jason dived back into his risk analyses recently, he noticed something was off. The mortality data for some age groups offered by the National Center for Health Statistics was much different from that of the Covid Data Tracker, a more public-facing page that analyzes demographic trends like race and age. He wrote to the CDC asking about the change.
Journalists, including myself, were also reaching out to the agency with the same questions. Why did one part of the CDC cite 894 pediatric deaths, for instance, while another put the number at 1,755? Which was more accurate, and why did the differences exist?
The CDC responded last week by quietly changing the number of deaths in the demographic tracker for all ages. In a footnote, the agency attributed the change to a “coding logic error.” The more accurate number was the lower one provided by NCHS, a spokesperson told me, even though those records face significant lags because of the lengthy process of confirmation.
“But why would you publish something that’s not accurate?” Jason asked when I told him of the CDC’s response. “That’s the primary thing—if you’re googling ‘CDC data,’ you’re going to come to the Data Tracker, and most people are probably never going to find the NCHS data.”
It’s incredibly difficult to report accurate numbers in real time during a massive outbreak, particularly since public health has been underfunded for decades. “I’m sympathetic in some ways, because I feel like they have an impossible task and they’re never going to please everyone. But I’m also not sure that they’ve done a great job, either,” Jason said. The mistake, while unfortunate, isn’t the main problem, he emphasized: The problem is that the CDC didn’t explain publicly what went wrong and how they’re keeping errors like that from happening again. Without the CDC explaining fully what happened, and how they reach the estimates in the Data Tracker, errors like these can further erode trust, which in turn empowers purveyors of misinformation.
“If there’s a vacuum, basically conspiracy theories are going to fill the void,” Jason said. “People who were on all sides have been like, ‘Seriously, you’re not going to explain this at all? You’re just going to drop a footnote and be like, oh, yeah, it was a coding error?’”
This issue adds to an already fraught conversation about how to count deaths, which has sometimes been politicized by states. Determining causes of deaths is an incredibly complicated process that can take months. “It is incredibly messy. Death reporting is not standardized, and it’s something that I think we will probably be talking about for many, many years post-pandemic,” Rivera said. “There are so many more people who will never be counted in this death count because of the lack of standardization in how this process is completed.”
Even once a death count is confirmed, it’s not always accurate. “Death counts overall, pediatric or adult, are a severe undercount,” Rivera said. Even when the CDC was showing more than 1,700 pediatric deaths in the Data Tracker, she said, “I never held tightly to that number to begin with, because I know it’s higher than that.”
So how could the CDC do better? One problem is, the CDC usually functions more like an academic institution, excelling at producing detailed reports months after an outbreak or episode. It wasn’t built to provide real-time analyses or communicate complicated, fast-moving science to the public. Aspects of the CDC’s communication have also been politicized. Some of that is inevitable; it’s a federal agency, and public health is inherently political. But the evidence itself should be as free from political influence as possible.
The agency changed its isolation guidelines, for instance, after politicians and industries expressed fears about the economic fallout of the more transmissible omicron variant. When the metrics for community risk changed, it similarly seemed to prioritize politics over health. “What they were trying to do was to provide receipts for a number of jurisdictions making those policy changes,” Rivera said. Changing the community risk metrics, in other words, offered state and local officials a rationale for changes they already wanted to make despite case counts remaining high.
These changes had a pretty big effect on public perception of the pandemic, said Dr. Jason Salemi, an associate professor of epidemiology at the University of South Florida College of Public Health. “CDC is obviously a relatively trusted source, although it’s been politicized during the pandemic. But people pay a lot of attention,” he told me. “People are paying attention to this map, which has just been bleeding red across the entire United States for much of December, January, and February. And then within one day, they see this transition.” Suddenly, many counties with high transmission were now ranked as low risk. The CDC, for its part, put out a guidance document explaining the changes—but that filtered down to few people in the public.
“This is a tool—one of the tools that the public can use and people can use to gauge risk,” Salemi said. “But it has always been imperfect, the data are imperfect, and there are so many other indicators that are being collected and reported that hopefully enough people are paying attention to all of the indicators in every community, all of the information that is accessible to them, to again get out in front of any sort of surge that emerges.” He worries, though, that when another variant emerges, people will be slow to take up precautions again—especially if they mistakenly believe they’re safe now.
“There’s a lot more that I would like to understand better, and I just don’t have the time and it’s frustrating,” Jason said. “I need to understand the risks so I can make the best decisions for our family.” His trust in public institutions has eroded. “They’re just not really set up for providing real-time information or public guidance in the way that we really needed in the pandemic,” he said. “I really don’t want to be too critical because I think they do have such a hard job. But from a blunt point of view, they’re not getting the job done.”
The experts I talked to basically agreed with Jason. While there’s a limit to how quickly any agency can process the deluge of data created by Covid-19, these experts highlighted the need to improve communication, specifically. “For the longest time, the federal government has treated science communication as an afterthought,” Rivera said. One example is vaccinations—the unfortunately named Operation Warp Speed gave the impression of rushed science, while the initial campaigns to get shots in arms were left up to poorly funded state and local health departments.
“It’s maddening,” said Rivera, “because there’s no expectation that this data has to be perfect, and this is where I think science communication really has an opportunity to shine here because part of science communication is explaining to people how to interpret things.” Humans are also really bad at evaluating risk and need guidance on interpreting it, she said. “We still have not learned how to communicate risk to the public, and that is probably one of the biggest failures.”
When a scientific agency is offering evidence or data, it needs to be clear about the nuances and disclaimers around those resources. Without context and caveats, data can be weaponized and used to explain policies based not on health but on political or ideological aims. With the Data Tracker, for instance, it’s important to note that the numbers are best estimates, not established facts—and it would help to have a section on methodology and access to raw numbers, experts said.
Acknowledging and explaining mistakes can actually build credibility, rather than diminish it, Rivera said. “It’s not just the science evolving. It is the fact that humans are fallible and humans make errors, and it is OK to admit when you’re wrong. There is trust that can be built and restored when you say, ‘Hey, I messed up.’”
There is still an important role for agencies like the CDC, now more than ever. “I’m not willing to throw the baby out with the bathwater because I still recognize these are experts in the space, that they are a premier public health agency, but they are overworked and underutilized in a public health emergency with other fires to put out at the same time,” Rivera said.
That work is only growing in importance, as Covid-19 continues to spread and evolve, and other pandemic-potential pathogens are likely to emerge in coming years. Because it’s impossible to make informed, moral decisions without data, the role of the CDC has grown in our lives, even as its shortcomings have been revealed. The biggest challenge we will face, moving forward, isn’t tracking data or following evidence; it’s building trust. That process can—and must—start now.