Citizen statistician keeps COVID data
Early in the pandemic, volunteers filled the information void
Sometimes the heroes who suddenly rise to a crisis are easy to spot. With Harvey, you couldn’t miss the Cajun Navy. They roared up in boats, chainsawing people out of flood debris.
Fighting COVID, Judith Oppenheim is quieter. From her house in Friendswood, she wields spreadsheets.
Last year Oppenheim, then 63, was a chemical engineer with time on her hands. She holds a dozen patents for things like a “carbon capture cooling system and method” and a way to do “process stream desulfurization,” and she’s used to wrangling teams spread across the globe. But the price of natural gas had plummeted even before COVID hit, leaving research and development projects few and far between.
So in May, itching to do something useful, she volunteered with the ad-hoc COVID Tracking Project. Consider it the Cajun Navy for nerds: When the federal government wasn’t providing the data it should have, Oppenheim and hundreds of other volunteers swooped in to help. Nobody expected the gig to last long.
Like five days before Katrina
Back when the novel coronavirus still felt novel, before the outbreaks had coalesced into a pandemic, Robinson Meyer and Alexis Madrigal, journalists at the Atlantic, were shocked to realize how little U.S. data was available about the threat. “Imagine we’re reporters on the Army Corps of Engineering beat, five days before Hurricane Katrina,” Meyer famously told Madrigal. “Like, what the f--- are we doing here? We should do something.”
They did what journalists do: They wrote an article. They noted that even as Congress was demanding to know how many Americans had been tested for the new coronavirus, nobody had a good answer — not the Centers for Disease Control and Prevention, not the U.S. Food and Drug Administration, and not the White House. And importantly, they went a step further: They contacted the public health departments of all 50 states, gathered the messy data and produced the best estimate available anywhere at that point. That, they noted pointedly in their article, was a job the CDC would normally do.
Afterward, similarly frustrated journalists and data wizzes got in touch. They stitched together an ad-hoc crisis-response team of volunteers to gather, clean and analyze data from the states each day, until the U.S. public health system got its act together. The COVID Tracking Project, they called themselves. The Atlantic supported it. Everyone figured it’d last a couple of weeks.
Hundreds of people signed up to help — some for a single day, others for the long haul. Volunteers gathered, cleaned or double-checked numbers. They tended the website. They organized other volunteers. They made graphs and charts and explained things.
And suddenly, instead of figuring out, say, how to manufacture nylon intermediates, Oppenheim found herself crunching Texas COVID numbers, and asking the Department of State Health Services pointed questions about its stats’ completeness and quality.
The COVID calculations and spreadsheets were easy, she said, compared to the torrent of data a chemical plant throws off. For her, the new and exciting parts were dealing with the project’s journalists (writing! deadlines! fact-checking!) and the data-visualization people (charts and graphs that normal human beings can understand!).
The volunteer gig lasted longer than she or anyone else had expected. Instead of disappearing after a couple of weeks, the United States’ data problems persisted, and in response, the COVID Tracking Project grew in size, ambition and influence. Oppenheim, a small cog, was impressed by the big machine — by the way its findings clearly influenced people such as Dr. Anthony Fauci.
The project was cited in nearly 900 academic papers, including the New England Journal of Medicine, Nature and JAMA. Its stats were used in more than 7,000 news articles, and even by the U.S. government itself: By both the Trump and Biden administrations, as well as the CDC, Health and Human Services and the FDA.
Earlier this month, after a year, the project shut down — finally satisfied, Madrigal and Erin Kissane wrote, “that the federal government is now producing enough data to replace many of our metrics.”
Oppenheim, though, isn’t done. Texas, she thinks, still needs help understanding what’s going on.
Baselines, plural
Recently, without the COVID Tracking Project’s data-entry volunteers, Oppenheim began uploading DSHS data into her spreadsheets. She was anxious to establish baselines, showing where the state stood before it lifted the mask requirement and reopened for business. If Texas’ previous reopening serves as a guide, she notes grimly, that will be significant.
And that’s baselines, plural — not a single baseline for the entire state. As much as Oppenheim enjoyed working with the COVID Tracking Project, she sometimes grew frustrated by the project’s national scope. Her job was to analyze Texas as a whole, but Oppenheim found that the state’s enormous size often hides what’s going on.
In the PowerPoint she uses to make her case, there’s a graphic that shows how Austria, Switzerland, Slovenia, the Netherlands, the Czech Republic, Belgium, Macedonia, Slovakia, Hungary and Luxembourg could all fit, without overlapping, inside Texas’ borders. Why, she asks, would it make sense to average COVID-19 numbers for a place that big? If you averaged the expected high temperatures for Lubbock and Houston, you’d get a number that’s misleading for both places.
But on the other hand, you wouldn’t want to go too far in the opposite direction — down to the tight focus of Texas’ 254 counties. Nobody can get their head around 254 dashboards.
So to track COVID going forward, Oppenheim has broken the state into five pieces: the Panhandle, the Border, Central Texas, North Texas and the Gulf Coast. That last, which includes the Houston area, runs from Orange County down to Kleberg County.
Each of those places had its own peaks and patterns of COVID hospitalizations. The Gulf Coast had twin peaks, its summer and winter surges both around 3,600 hospitalizations a day. North Texas, which includes Dallas, had a small hump followed by a big one: a summer surge under 2,400, followed by a winter surge over 4,800.
Oppenheim is anxious to see what comes next. She publishes her findings on her personal Facebook page, but soon plans to start a blog; she’d like to find institutional partners to get her word out. On April 12, she’ll show her PowerPoints at an online seminar sponsored by
Rice University’s Baker Institute for Public Policy. Rice health economist Vivian Ho, who invited her, is a fan.
Oppenheim isn’t sure how long she’ll continue tracking COVID in Texas. She guesses that vaccines will be available to everyone here by May, and maybe, soon after that, her self-assigned gig will be over.
Maybe. She’s been wrong about that before.