U.S. Created a Data Disaster With Its Uneven Covid Response
More than a month into a resurgence of the novel coronavirus that has besieged Sun Belt states, flooded hospitals and strained public-health infrastructure, the U.S. still lacks a complete picture of the on-the-ground reality.
That’s no surprise to public-health experts following the country’s response, since the U.S. doesn’t have an accessible, real-time system to track the virus’s spread. At times, even the federal government has had to rely on third-party databases.
The gap is due to decades of neglect of technological infrastructure, exacerbated by the country’s sprawling size and a state-by-state approach to collecting public health data. It has left not only government officials hunting for reliable data, but kept the public in the dark as well.
The challenges were highlighted earlier this month when hospitals were told to send virus data to the U.S. Department of Health and Human Services instead of the Centers for Disease Control and Prevention. The agencies said the goal was to streamline the system, which helps inform where to send medical supplies, but critics worried it could impede transparency.
“We do have gaps in data reporting; we hope that by making this public, we will drive improved data reporting among hospitals and states across the country,” an HHS spokeswoman said in a statement.
State and local areas also report virus data on online dashboards, with approaches to presenting information varying widely. The result is an “inconsistent, incomplete and inaccessible” system, according to Tom Frieden, a former CDC director who now leads the nonprofit Resolve to Save Lives.
In May, some states acknowledged they’d released data mixing diagnostic and antibody tests, muddying how much testing was actually going on. Georgia backdated new cases to the first symptoms or when a test was taken – other states count them from when the results arrive – making its current figures look low and helping to justify reopening. In Florida, the health department recently had to correct a daily death toll after erroneously combining two days of data, and didn’t report current Covid hospitalizations until this month.
On Monday, the death toll in Texas rose by 675, an apparent 13% leap. However 631 were additional cases added after analyzing death certificates going back to March.
After analyzing how states tracked 15 essential data points, including numbers of new cases, deaths and hospitalizations, Resolve to Save Lives found further discrepancies. The report called it an “information catastrophe.”
Metrics like new cases and hospitalizations were “surprisingly inconsistent,” the review concluded. The one thing every state tallied was daily deaths by confirmed cases, but that too provides limited information since many people may have died of the virus without being diagnosed, and as seen in Texas, not all localities track Covid deaths the same.
HHS has set federal reporting standards, the spokeswoman said, pointing to guidance outlining how hospitals should submit information on capacity and other metrics.
“States have always collected local data,” she said.
Nicholas Reich, an associate professor at University of Massachusetts, Amherst, is working with the CDC to better predict the virus’s trajectory. The work has been impeded by inconsistent hospitalization figures, and the HHS reporting change has cost him the data he’d relied on the CDC for.
“The lack of a coordinated effort at a national level has hurt our model’s ability to assist with a response,” Reich said.
“Since the data collection change was implemented, HHS has made an even larger amount of hospital data public to researchers,” the spokeswoman said, by publishing raw data “for everyone to access and analyze.”
A further complication is that each metric has its flaws. Hospitalizations and deaths lag, but even confirmed case counts don’t track outbreaks in real time. Symptoms take days to develop, and even if a test is performed, waits for results mean that by the time an increase is evident, it could be weeks late.
“Without good data on your situation, states are really operating with one hand tied behind their back,” said Jaline Gerardin, an assistant professor of preventive medicine in epidemiology at Northwestern University Feinberg School of Medicine. She and a colleague are working with Illinois’s public-health department to use ongoing testing to track whether Covid is on the rise.
A CDC spokesperson said the agency is working with states to track illnesses that resemble the flu and coronavirus and sending them twice-weekly reports on the virus’s trajectory.
Monitoring how people feel can also provide early warnings, which the CDC and HHS are incorporating in their tracking efforts.
In March, volunteers from tech companies built CovidNearYou, an online crowdsourcing tool that’s collected data from nearly two million people on relevant symptoms like body aches and shortness of breath. Local and state officials as well as the CDC use it to help capture the national picture.
“People feel symptoms before they eventually go to their doctors,” said John Brownstein, who leads the project and is chief innovation officer at Boston Children’s Hospital.
Kinsa, a maker of smart thermometers, has long used them to track fevers during flu seasons. The devices could also help track Covid outbreaks in real time, founder Inder Singh said.
“We have the technology to talk to people before they enter into the health-care system,” Singh said. He looked into federal funding for surveillance efforts, but was told they’re focused on early detection in individuals rather than across whole populations.
HHS is, however, using Kinsa’s data in its response efforts, according to the spokeswoman.
Symptom screenings are becoming more commonplace, as employers deploy them broadly in the workforce before sending individuals back to the office, schools and other settings. But Meredith Matone, the scientific director of PolicyLab at Children’s Hospital of Philadelphia, cautioned that without standardized questions and a method of reporting answers, it could be just another missed opportunity.
“None of that data’s going anywhere,” she said. “If we know folks are going to be doing this, why would we not roll that up into an effective, robust surveillance strategy?”
©2020 Bloomberg L.P.