Behind Georgia’s Covid-19 dashboard disaster

The Georgia Department of Public Health saw its reputation scorched as a result of the state’s ridiculed Covid-19 dashboard. But as it turns out, the health department had little control over the troubled site.

698

Research for this story was supported by the Fund for Investigative Journalism.

On Tuesday, April 28, eight days after Brian Kemp sent shock waves nationwide as the first governor to announce he would reopen his state during the pandemic, a quiet storm was brewing over another of Kemp’s decisions. State officials were sending flurries of emails about the previous day’s launch of Georgia’s new Covid-19–tracking dashboard—the primary tool that business owners would use to decide when or whether to reopen, now that they could. The launch was supposed to mark an improvement over the state’s preexisting Covid-19 webpage. But it was not going well.

Nancy Nydam, director of communications for the Georgia Department of Public Health, forwarded to two of her colleagues an email she’d received listing constituents’ complaints about the dashboard: deaths by county and demographic had disappeared; age and gender information had vanished; the color scheme was difficult to see for some readers; numbers on the page contradicted each other. At least one state agency reached out with an urgent need for data that were no longer on the page—an office manager from the Georgia Emergency Management and Homeland Security Agency (GEMA) wanted answers from the health department “like ASAP” to a list of questions about missing demographic information regarding hospitalizations and deaths, as well as some other metrics. “I wanted to see if you guys have the information listed below in an easy to share format?” she wrote.

That day’s hitches were not the first indication of the dashboard’s potential problems; as recently as the weekend before its launch, the state’s lead epidemiologists noted that Dougherty County, where the virus’s scorching arc through low-income Black communities had rendered Albany the city with the second-highest number of Covid-19 cases per capita in America, was absent from the as-yet-unpublished dashboard’s list of “top five” counties.

Nor would the dashboard operate smoothly in the weeks and months to come. That much would become clear both to state officials firing off frantic emails and to bewildered Georgians trying to interpret the dashboard’s data in an attempt to decide whether to visit a restaurant, attend religious services, or send their children to summer camp or daycare.

What remained unclear to the public, however, was who exactly was pulling the strings behind the state’s maligned Covid-19 dashboard. Although by all accounts it would appear that it was operated by the Georgia Department of Public Health, some skeptics felt that the fingerprints of the state’s public-health experts were conspicuously absent from the dashboard bearing the agency’s name.

In May, health department commissioner Dr. Kathleen Toomey abruptly ended an interview with a WABE reporter when he raised a question to that effect.

“Who is making the call about what information the Department of Public Health is displaying on [its data dashboard] page?” reporter Sam Whitehead asked. “Is that being made within your agency?”

“Listen, I’m gonna have to run,” Dr. Toomey responded, in what came across as an almost comical attempt to avoid the question. “I actually can’t answer this right now because I’m getting called by the Governor’s office.”

The answer to Whitehead’s question proved more elusive than it should have. The Atlanta Journal-Constitution reported in July that the health department had not fulfilled any of the dozens of open records requests seeking emails relating to the state’s handling of the Covid-19 pandemic since March. In August, the AJC reported that GEMA had redacted enormous amounts of information from Covid-19–related records requests it had fulfilled—and presented the newspaper with a bill for nearly $33,000 to fulfill additional requests.

Atlanta was able to obtain emails illuminating the inner workings of the state’s Covid-19 dashboard not from the state’s Department of Public Health but from the Governor’s Office of Planning and Budget. Why would the office that handles Kemp’s and the state’s budgetary affairs have been the custodian of emails about what ostensibly belongs in the state health department’s domain? Because that office had outsourced the dashboard to a private company—and had assumed what public-health experts describe as an unusually expansive role in overseeing the project.

A series of open records requests Atlanta filed to the Governor’s Office of Planning and Budget yielded thousands of emails concerning the state’s new Covid-19 dashboard, sent between employees of that office and those of the health department—as well as those of the third-party vendor tasked by that office with creating the dashboard. An examination of those emails revealed the health department had limited input into and no real oversight over the dashboard during its creation and in the months after its launch. Additionally, the sidelining of the health department allowed for errors in the analysis, interpretation, and visualization of the state’s Covid-19 data, while simultaneously costing the state tens of thousands of dollars—and time it did not have to spare.

Other open records requests for emails to and from a different state agency showed that at the same time the Covid-19 dashboard was suffering from very public problems, health department officials were working in collaboration with that agency to create a different dashboard—and that after its launch, they were unsuccessful in their attempts to make its existence widely known.

Furthermore, when the dashboard elicited public outrage, the health department shouldered the blame for errors over which it had no control, damaging the relationship between the agency and the community it serves.

“This is the type of information that you make informed decisions on—decisions that impact millions of people in a jurisdiction,” says Dr. Syra Madad, an infectious-disease epidemiologist and special pathogens preparedness expert in the New York City hospital system, in reference to state-run Covid dashboards. Because the impact of dashboards on those decisions is so outsized, authorities must take great care in determining who oversees them, according to Dr. Madad. “It’s okay to bring in outside individuals or contract with other entities as long as it’s in collaboration,” she says. “But if this [outsourcing of the dashboard] was based on a political decision and not in collaboration with public-health people that actually know what they’re doing, then that’s a recipe for disaster.”

In her April 28 email, Nydam, the health department’s communications director, particularly had been concerned about an inquiry from the AJC in relation to the one-day-old dashboard: “The most pressing is this email from the AJC,” she had written to two health department employees. “Someone must talk to them or we are going to get dragged through the dirt for something that we did not do.”

In response to Atlanta’s detailed questions about the contents of the emails—including why the health department didn’t have more control over the dashboard on its own site and whether its epidemiologists were given enough input into the dashboard—the governor’s press secretary, Cody Hall, only responded: “We are referring comment to DPH here.” When Atlanta pointed out that the questions concerned the actions and decisions of the Governor’s Office of Planning and Budget, Hall would only state: “As the media contact for the Governor’s Office my comment is: ‘I am referring this media request to the Department of Public Health.’”

Similarly detailed questions to the health department were met with this statement from Nydam: “Throughout the COVID-19 pandemic, the Georgia Department of Public Health has worked and continues to work closely with Governor Kemp’s office, the Georgia Department of Community Health and the Georgia Emergency Management and Homeland Security Agency to provide data that is accurate and transparent. We continually review and update features of the dashboard with our vendor . . . to ensure we are providing as complete a picture as possible of COVID-19 in Georgia.”

Several experts on American public-health infrastructure told Atlanta it’s not uncommon for health departments to have a contractual arrangement with a third party to help with certain aspects of data management or with special, time-limited projects like surveys. But it’s unusual to completely outsource a public-health data analysis that shows up on a health department’s site while failing to give the health department oversight of that analysis, says Janet Hamilton, executive director of the Council of State and Territorial Epidemiologists, a nonprofit organization representing public-health epidemiologists. She points out that a state’s team of epidemiologists is uniquely equipped to interpret, analyze, and visualize public-health data.

“That is the job of an epidemiologist, to not just produce a report—a biostatistician can do that—but [to carry out] the ‘ground truthing’ of it,” says Hamilton. That is, tethering the data to real events rather than the projections of policy experts. “It’s just so critical that you do have the right epidemiologists that are leading the efforts and able to see inside the work.”

In Georgia, those epidemiologists existed; they were employed by the Department of Public Health. But they were not leading the efforts.

On Monday, March 16, the novel coronavirus had begun to wreak havoc on Georgians’ lives. The night before, Atlanta mayor Keisha Lance Bottoms had declared a state of emergency, and it was the first day of remote learning for students in many school districts statewide. The Department of Public Health’s daily Covid-19 status report—at that time, a bare-bones page consisting of no more than a case density map of the state, a list of cases by county, and a couple of pie charts—counted 99 cases and one death due to the virus.

That morning, Chavis Paulk, the division director of analytics in Governor Kemp’s Office of Planning and Budget, sent an email introducing himself and his team to Theresa Do, a Washington, D.C.–based epidemiologist and manager at SAS, a data-analysis software and consulting company headquartered in North Carolina. The email mentioned an Excel file containing the details of each suspected Covid-19 infection in the state, which Paulk’s team had just uploaded to a secure server.

It was an innocuous enough introduction, but it opened the door to a protracted and consequential barrage of emails between SAS, the governor’s planning and budget office, and, eventually, the health department.

SAS has been around since the 1960s, when it was known as the Statistical Analysis System, a computer program for analyzing agricultural data. Later incorporated in Raleigh, the company has since evolved into a multinational software and data analysis consulting corporation with more than 14,000 employees. Its software is widely used in health-services research and in public health, including at the Centers for Disease Control and Prevention (CDC); the Morbidity and Mortality Weekly Report—the agency’s flagship publication—often notes use of the company’s software.

The relationship between the governor’s office and SAS was relatively new. In an annually renewable contract initially signed in August 2019, the company agreed to provide software and consulting services to the Governor’s Office of Planning and Budget at a total cost of nearly $3.7 million over five years. But OPB’s director since early 2019, Kelly Farr, who also had worked for Kemp back when the governor was the secretary of state, already knew SAS well: From 2017 to 2019, Farr had worked for the company as an account executive.

The data in the Excel file that the governor’s planning and budget office sent to SAS on March 16 were similar to the data the health department was using to make its own Covid-19 webpage, then only four days old. Over the next six weeks, as the health department continued to maintain its Covid page, the team at SAS would develop an entirely different one using its own software and analysts.

Well before the launch of the SAS dashboard, the Covid-19 webpage managed by the health department had its own problems. As the SAS team worked on its prototype—and as novel coronavirus infections surged in Georgia—the health department scrambled to keep its webpage updated with the flood of information coming its way. Its efforts were complicated by the massive influx of inaccurate and incomplete data pouring in via antiquated reporting processes managed by a decentralized and underfunded public-health system. The effects of these problems only would be amplified once the state’s public-health authorities no longer had control of how Covid-19 data was presented on its own website.

“I have no access to the site and no real awareness of who is responsible for the details behind this . . .”

The pressures of a public-health emergency can create intense demand for frequent, real-time reporting that may exceed a health department’s capacity, according to Hamilton, with the Council of State and Territorial Epidemiologists. But when outside data analysts responsible for quality control don’t see a dataset through a public-health lens, the high-pressure environment can lead to errors, she says. “I don’t necessarily want to say that [any errors are] malicious—I think that they’re being driven in part by unrealistic expectations that data is coming in in a way that is much cleaner” than it is, she says.

On April 11, Farr, director of the governor’s planning and budget office, sent an email to Lorri Smith, Governor Kemp’s chief operating officer, and Dr. Toomey, the health-department commissioner, with two links to the SAS team’s work in progress: one with “high level information that could be incorporated as [a] website” and another with “additional information and insights.”

Four days later, health-department epidemiologist Laura Edison responded to an email from Anand Balasubramanian, the governor’s technology advisor, in which he’d asked about “some concerns” she had with the dashboard prototype. “I think this is a great display,” she wrote back, “and just have some nuances to discuss.” In a conference call summarized in a subsequent email, Edison and her colleagues noted that in some places, the dashboard used inappropriate terminology and lacked sufficient explanatory text; in others, key metrics and tables were absent, or existed where they didn’t belong; the graph showing the daily case count did not use shading to indicate a 14-day “pending period” to account for the lag time between a person’s onset of symptoms and the confirmation of their positive test result by the state. SAS epidemiologist Do summarized health-department staffers’ recommendations in a table spanning three pages. (SAS would make nearly all the changes, she wrote.)

But Karl Soetebier, the director of the health department’s informatics office, later would make plain just how little input he’d had into the SAS dashboard.

“My only real involvement to date has been to provide the data to OPB [the governor’s planning and budget office] and a few discussions with the folks from SAS about the data itself,” he wrote to Balasubramanian. “I have no access to the site and no real awareness of who is responsible for the details behind this or what process is needed to have changes made.”

Behind Georgia’s Covid-19 dashboard disaster
Governor Brian Kemp at his April 27 Covid-19 press conference

Photograph by Kevin C. Cox/Getty Images

When Kemp announced on Monday, April 20, that he’d soon allow nail salons, hair salons, and bowling alleys—followed by restaurants and movie theaters—to resume serving customers, Georgia did not yet meet the criteria to reopen as set forth in White House guidelines (namely, a downward trajectory of documented cases within a 14-day period). President Trump himself criticized Kemp for reopening the state prematurely. The following weekend, the day after the first businesses reopened their doors, SAS’s Georgia team lead Albert Blackmon wrote to Aaron Cooper with the governor’s planning and budget office and several others, saying: “I know that there is a desire to go live with the site very soon.”

Blackmon acknowledged minor inconsistencies between SAS’s and the health department’s analyses of the state’s Covid-19 data and noted that, if there were still concerns about SAS’s numbers, his team would need to get on the phone with the health department immediately and attempt to reconcile any discrepancies before SAS’s new dashboard was unveiled.

Two days later, on the morning of Monday, April 27, Kemp’s technology advisor Balasubramanian wrote in an email to his colleagues and to SAS that the governor’s office wanted the SAS dashboard to go live that afternoon. The launch would come a day ahead of schedule—and an hour and 15 minutes in advance of a press conference at which Governor Kemp, with health-department commissioner Dr. Toomey at his side, discussed how restaurants would safely reopen for dine-in customers effective immediately. Kemp also took a few moments to introduce the new data dashboard: “We realized as a team that we can provide a more unified, user-friendly platform for Georgians in every corner of our state.”

The next day, the health department’s Soetebier vented to his higher-up, Dr. Toomey: “As you know we were given a new website for the public yesterday for which we have had little input on to date and for which we no longer have direct control.” He also made clear that SAS should take responsibility for any dashboard problems. “I have asked them to own the ongoing list of issues that are identified with the dashboard and to commit to reviewing their progress on them with us regularly,” he wrote.

The public reaction to the dashboard was negative and swift. An AJC article two days after the dashboard’s launch noted that it “confused ordinary Georgians as they decide whether Gov. Brian Kemp was right to begin reopening the state’s [businesses]” and was “making it difficult for the public to determine if Georgia is meeting a key White House criteria for reopening.”

“A lot of people are now accusing us of trying to hide data and/or misrepresenting . . .”

Three days after the dashboard’s launch, Megan Andrews, the health department’s director of government relations, forwarded a roundup of constituent complaints to SAS’s Blackmon, asking for assistance in responding to the concerns expressed in the constituents’ emails.

Blackmon replied, “We will get you answers ASAP.” Four days later, Andrews’s deputy, Emily Jones, sent a follow-up email: “We are really in need of some answers for constituents,” she wrote on May 4. “A lot of people are now accusing us of trying to hide data and/or misrepresenting, so getting them information quickly is important.”

Particularly worrying to Jones was the concern several constituents had raised about perceived manipulation of the data to artificially show a decrease in cases. They “believe that these graphs are intentionally designed to show a downward trend and are wondering if a better explanation of the methodology can be given,” she wrote.

SAS’s Blackmon seemed to think the existing explanation on the dashboard was enough: “There is a clear asterisk under the chart” explaining that the last 14 days in the chart may be missing cases, he wrote. “That is what I have been telling people,” replied Jones, “but I wanted to make you aware that we are getting several of these inquiries a day.”

The next Saturday, May 9, a Twitter user called out an egregious graphic on the dashboard. “I’m sorry but I have to curse your twitter feeds with this nightmare graph from @GaDPH,” she tweeted. “The X axis shows dates, BUT not in chronological order for some godforsaken reason.” In an attached image captured from the dashboard, cases descended from left to right, at first glance suggesting a downward trend as time progressed—but as the out-of-order dates indicated, time was not actually progressing but jumping all over the place.

Behind Georgia’s Covid-19 dashboard disaster
A screenshot of the chart published on Georgia’s Covid-19 dashboard in May that falsely showed a decrease in the state’s infections—by rearranging the order of the dates at the bottom.

Other Twitter users were quick to speculate about the explanation for the chart’s unusual configuration: “Oh, we know the reason. A clear attempt to make the data say what they want it to say, rather than just letting it speak,” wrote one. Journalists also were perplexed: “Only in Brian Kemp’s Georgia is the first Thursday in May followed immediately by the last Sunday in April,” a Washington Post columnist quipped. Pete Corson of the AJC tweeted that the graphic had been “the subject of much head scratching” at his publication.

In a response to Corson, Kemp’s director of communications, Candice Broce, implied the health department was to blame: “The graph was supposed to be helpful,” she tweeted, “but was met with such intense scorn that I, for one, will never encourage DPH to use anything but chronological order on the x axis moving forward.”

Over the next two weeks, a volley of errors emerged from the dashboard: A chart showing Covid-19 cases by race mistakenly included a diagnosis date in 1970, making it unreadable; the total case number inadvertently included—then abruptly expunged—231 serology test results, resulting in a confusing decrease in positive cases between reporting periods; and data points went missing from charts depicting individual counties’ daily case numbers.

A May 19 AJC article explored multiple explanations for the mistakes, quoting Broce as saying of the health department: “We are not selecting data and telling them how to portray it, although we do provide information about constituent complaints, check it for accuracy, and push them to provide more information if it is possible to do so.” Although the story noted a Kemp aide had blamed “a software vendor” for the widely ridiculed nonchronological graph, it did not give further detail on the extent or nature of the vendor’s responsibility.

The next morning, the Columbus Ledger-Enquirer reported that the dashboard’s misstep with the serology tests “artificially lowers the state’s percentage of positive tests.” (Emails indicate that the dashboard’s errors stemming from the tests were due to the health department misclassifying them. “This is not a technical issue per se with the website,” Soetebier wrote to his health-department colleagues and Kemp’s chief management officer, Caylee Noggle.)

Amid the fresh wave of public rancor in the wake of the story, the health department’s Edison warned in an email the next day to Noggle that “by rushing through data analyses, we run the risk of making errors.” Edison proposed that a clarifying footnote be added to the dashboard. “It takes time to work through these complicated and far from perfect data.”

Four hours later, Edison sent Noggle and several other Kemp staffers a four-page data FAQ of sorts to post to the site. Balasubramanian, the governor’s technology advisor, forwarded the document to the SAS team with a request to post it to the website—but two minutes later, he walked that request back: “Hold on, don’t POST,” he wrote. “Please review and let me know if you have any suggestions.” (Kemp staffers later stripped almost all of the explanatory content from the data FAQ the health department team had written.)

At a May 21 press conference, Kemp addressed some of the public derision related to the dashboard. Citing his administration’s commitment to transparency and honesty, he praised Dr. Toomey and the health department: “They are taking massive amounts of data from all sources, putting them into accessible format under a global spotlight, all at breakneck speed,” he said. “Please afford them some patience, and please steer clear of personal attacks.”

But Kemp did not mention his own team’s role in creating much of the pressure the health department was under, nor the fact that some of the highest-profile mistakes had not been the health department’s errors at all.

“It’s a fair point that it could look like we’re ‘moving the goalposts’ . . .”

Emails also show that when health-department staffers sought potential fixes with SAS, their requests were not treated with a sense of urgency. In early June, Leslie Onyewuenyi, a newly hired interim director of informatics who was brought on to work above Soetebier and improve data quality at the health department, asked SAS’s Blackmon for a 30-minute call to review SAS’s quality-control process.

“I don’t believe that there is a need for a call unless [the health department’s] Karl [Soetebier] would like for us to convene,” Blackmon responded.

“We need a high level overview of process flow on your end,” Onyewuenyi wrote back. “Are there any quality control checks on your end before the data is published? The aim of this exercise is to reduce the risk of publishing inaccurate data whether from DPH side or from your end.”

After Onyewuenyi appeared not to get a response to this email or to a follow-up one he sent three days later reiterating his request, the governor’s planning and budget office intervened to set up a call between Onyewuenyi and Blackmon, noting that Blackmon was on vacation.

“We’ll respond on email first,” a SAS project manager wrote. “We can then follow-up as needed.”

At around the same time, Balasubramanian forwarded to SAS a media question that had been sent to the health department about a county map: Why was the threshold for a county to be shaded red—indicating the highest case rates in the state—changing from day to day?

SAS responded by forwarding an explanation from one of its systems engineers: “It’s a fair point that it could look like we’re ‘moving the goalposts’, it might be something we could revisit.” But the method behind the color-coding would remain unchanged until, more than a month later, a viral tweet pointed to it as an example of how the health department “is violating data visualization best practices in a way that’s hiding the severity of the outbreak.”

Trent Smith, a senior external communications specialist with SAS, responded to a series of Atlanta’s questions about its work on Georgia’s Covid-19 dashboard by stating: “We can’t share customer names without their permission.” Smith also wrote: “SAS has been used for decades in public-health departments, from local to state to national governments and is currently in all U.S. state health departments.”

As the health department was publicly battered for mistakes over which it had little control, its leadership was well aware of the need to improve the dashboard and the magnitude of the fallout from its problems. On the Fourth of July, after reviewing examples of other states’ data dashboards, Dr. Toomey asked health-department staff to request that the SAS team add certain metrics to the dashboard and noted the negative public perception of her agency: “I am getting complaints from the public as well as other officials that we are deliberately not being transparent.”

Atlanta coronavirus update
Health-department commissioner Dr. Kathleen Toomey answers questions during Governor Kemp’s Covid-19 press conference on April 27.

Photograph by Kevin C. Cox/Getty Images

Some of the state’s public-health experts felt Georgians deserved Covid-19 analysis and insight beyond what the SAS dashboard ultimately offered, and they tried— unsuccessfully—to offer that info on the health department’s site.

Back in March, at the same time SAS began what would be a six-week effort to build its dashboard, a team from another government agency was creating other Covid-19 dashboards for internal use.

Susan Miller, who leads the Georgia Geospatial Information Office (GIO), began working on maps to assist other state agencies in allocating pandemic-related resources in March. Her team used a product made by the California-based company Esri. No other mapping platform on the market is as “comprehensive, holistic, or stable,” as Esri, she says. (Miller worked for the company as a product engineer in the early 2000s.)

In mid-April, the health department’s Edison asked Miller’s team to create a report aimed at providing the governor—and, possibly, the general public—with the data Georgians would need to decide when it was safe to reopen for business. She asked if these could take the same form as one of the internal dashboards the team already had created.

Once Miller’s team got started on the project, it took less than a week for a prototype to come together. The GIO’s Esri dashboard, compared with the SAS dashboard, had “increased functionality, such as ZIP code level data, death demographics by county/zip, and downloadable data,” wrote the health department’s Edison in an email to Miller and colleagues at other agencies on April 28, one day after the SAS dashboard launched. “I do not think the SAS Dashboard has the functionality that the ESRI one has and I think they can be used in tandem to complement each other.”

Esri’s software and the use of its consulting services weren’t free: The contract Miller’s parent agency signed with Esri’s Disaster Response Program in May totaled $265,000. But those dollars went toward Esri’s work on multiple mapping projects for a variety of agencies.

Health-department officials were hopeful about sharing the Esri dashboard on their agency’s website. “My goal is at a minimum to make this accessible from a link on the page,” Soetebier wrote on April 29 to health-department colleague Edison and staffers at GIO and Esri, “though we should be able to get a new page put together to properly house it.”

On May 13, Edison forwarded to Miller an announcement about a CDC partnership with Esri aimed at enabling all states—at no cost to them—to build or enhance data dashboards using the software. The next day, Edison exclaimed in an email to Miller, Soetebier, and an Esri employee: “We have some traction!” She wrote that two people from the governor’s office “are going to pitch the dashboard!”

But the Esri dashboard would not end up being included or even noted anywhere on the health department’s site. It was published on the GIO’s Covid-19 website, but it wasn’t publicized until Miller’s office published a blog post about it three months later, in mid-August—and even then, the existence of the dashboard remained largely unheralded for several more weeks.

Eventually, one government agency would find value in the multiple Esri dashboards Miller’s team had produced and published on GIO’s Covid-19 website. In September, GEMA replaced its daily Covid-19 situation report with that website, calling it “a one-stop shop for all of the data in a format that is more easily accessible.”

At the most concrete level, the problems with the state’s Covid-19 dashboard made it unreliable as a tool for Georgians simply trying to figure out how to safely go about their lives. As Georgia planned to reopen its doors for business in late spring, the health department fielded an onslaught of questions and complaints from people confused about how to interpret what they were seeing on the dashboard. The lead pastor at a church in Cobb County wrote for assistance understanding how rampant the virus was locally in the hopes of helping his church determine when to reopen for in-person worship. The assistant superintendent of a school district south of Seattle requested an explanation of conflicting case numbers in the hopes of advocating to reopen his own state; “I would like my state open, and Georgia serves as a bellwether,” he wrote. “Please explain the data so that I can advocate correctly and not put my community at risk.”

In the next three months, Georgians celebrated Memorial Day and the Fourth of July, and Governor Kemp squashed mayors’ efforts to enact local mask mandates and other protective measures. Also in that time, more than 155,000 Georgians were infected with the novel coronavirus, of whom 2,551 died.

Beginning in late July, the dashboard stopped attracting as much negative attention as it had early on. Although two public-health experts recently told Atlanta they would like to see additional data on the dashboard, such as case information by zip code and information related to school outbreaks, public outrage over the dashboard’s appearance has largely ebbed.

But public-health experts say the damage to the health department’s reputation caused by the dashboard’s pattern of problems may have lasting effects. In a statewide survey the health department conducted in late July, only 55 percent of respondents perceived the agency as credible. Amber Schmidtke, a volunteer advisor to the state’s Covid-19 Data Task Force who until recently was an assistant professor of microbiology at Mercer University in Macon, recalled several fumbled efforts at transparency on the state’s Covid dashboard, concluding: “So, yeah, I think it does harm people’s trust.”

Melanie Thompson, an Atlanta doctor and researcher who coauthored two July letters protesting Kemp’s handling of the pandemic that were signed by thousands of healthcare workers, says the contents of her inbox made the public’s loss of faith plain: “The emails and things that I got from a variety of people made me feel that there is no trust in the governor to do the right thing scientifically,” she says, “and that extends to the Department of Public Health, because [its] commissioner basically serves at the pleasure of the governor and does not contradict him at all.”

When public trust in an institution is sufficiently eroded, it can be hard to recover, says Joseph Cappella, a specialist in health communication at the University of Pennsylvania’s Annenberg School for Communication. “It’s the old idea of poisoning the well,” he says. When public-health institutions lose credibility as a consequence of one misstep, he says, the resulting lack of trust can impact their ability to effectively carry out other public-health activities, like vaccine distribution.

Clarity about who’s doing the work on state websites is important, too, says Laura Harker, a senior policy analyst at the Georgia Budget and Policy Institute. When a consulting company’s work is presented on an agency’s website, “having that made clear somewhere—at least the name on the bottom of who the outside contractor is, or some type of contact information for the data managers—is always, I think, important to have for transparency purposes,” she says.

The state of Georgia slashed the health department’s epidemiology budget during the lean years of the recession—from $6 million annually in 2009 to less than $4 million in 2011—and that budget was never fully restored. Georgia’s public-health funding lags well below the national average.

“People are thinking that public health has failed society,” says Dr. Madad, the New York City–based epidemiologist and preparedness expert. “No. Society has failed public health because we didn’t invest and see the value of it. And we’re seeing the consequences today.”

The early chaos of the Covid-19 dashboard shows how Georgia squandered the chance to shine a light on the merits and necessity of a public-health department, says Thompson. “This was an opportunity for DPH to shine, . . . to come into its own, and to really teach the public what public health is all about, to really engender trust.”

On April 28, the day after the SAS dashboard launched, health-department epidemiologist Edison and GIO head Miller exchanged emails about the difficulty of getting the best Covid-19 data to the public and the need for a more collaborative effort among government agencies. “My head is spinning,” Edison wrote. “I just want to share the damn data.”

Miller responded: “We can either feed the real data to Georgians, the country and the world . . . or let them fend for themselves. . . . I will back you on getting the data out until the end of time!!!!”

This article appears in our December 2020 issue.

Advertisement