This story was supported by the Science Fund for Investigative Reporting. Please help Science pursue ambitious journalism projects.
On the morning of 13 July, more than 20 COVID-19 experts from across the U.S. government assembled in a conference room at the Department of Health and Human Services, steps from the Capitol. The group conferred on how best to gather key data on available beds and supplies of medicine and protective gear from thousands of hospitals. Around the table, masks concealed their expressions, but with COVID-19 cases surging out of control in some parts of the country, their grave mood was unmistakable, say two people who were in the room.
Irum Zaidi, a top aide to White House Coronavirus Task Force Coordinator Deborah Birx, chaired the meeting. Zaidi lifted her mask slightly to be heard and delivered a fait accompli: Birx, who was not present, had pulled the plug on the Centers for Disease Control and Prevention’s (CDC’s) system for collecting hospital data and turned much of the responsibility over to a private contractor, Pittsburgh-based TeleTracking Technologies Inc., a hospital data management company. The reason: CDC had not met Birx’s demand that hospitals report 100% of their COVID-19 data every day.
According to two officials in the meeting, one CDC staffer left and immediately began to sob, saying, “I refuse to do this. I cannot work with people like this. It is so toxic.” That person soon resigned from the pandemic data team, sources say.
Other CDC staffers considered the decision arbitrary and destructive. “Anyone who knows the data supply chain in the U.S. knows [getting all the data daily] is impossible” during a pandemic, says one high-level expert at CDC. And they considered Birx’s imperative unnecessary because staffers with decades of experience could confidently estimate missing numbers from partial data.
“Why are they not listening to us?” a CDC official at the meeting recalls thinking. Several CDC staffers predicted the new data system would fail, with ominous implications. “Birx has been on a monthslong rampage against our data,” one texted to a colleague shortly afterward. “Good f---ing luck getting the hospitals to clean up their data and update daily.”
When Birx, a physician with a background in HIV/AIDS research, was named coordinator of the task force in February, she was widely praised as a tough, indefatigable manager and a voice of data-driven reason. But some of her actions have undermined the effectiveness of the world’s preeminent public health agency, according to a Science investigation. Interviews with nine current CDC employees, several of them senior agency leaders, and 20 former agency leaders and public health experts—as well as a review of more than 100 official emails, memos, and other documents—suggest Birx’s hospital data takeover fits a pattern in which she opposed CDC guidance, sometimes promoting President Donald Trump’s policies or views against scientific consensus.
The agency’s loss of control over hospital data is emblematic of its decline in nine short months. Since the pandemic began, CDC has foundered (see sidebar, below). It has committed unforced errors, such as shipping out faulty coronavirus tests, and has been squelched or ignored amid continual political interference.
CDC employees with whom Science spoke—who requested anonymity because they fear retaliation—along with other public health leaders, say Birx’s actions, abetted by a chaotic White House command structure and weak leadership from CDC Director Robert Redfield, have contributed to what amounts to an existential crisis for the agency. And her disrespect for CDC has sent morale plummeting, senior officials say. During a May task force meeting, The Washington Post reported, Birx said: “There is nothing from the CDC that I can trust.”
CDC scientists and others say Birx’s record echoes her approach as head of the President’s Emergency Plan for AIDS Relief (PEPFAR) since 2014. Although that program is widely praised, people who worked on it for years say her draconian management and unrealistic data demands damaged morale and disrupted fieldwork and patient services.
Through the office of Vice President Mike Pence, who heads the task force, Birx declined to be interviewed or to respond to written questions, including whether the White House pressured her to use TeleTracking’s system. (TeleTracking’s majority owner, real estate developer Michael Zamagias, has donated to Republican candidates and has ties to Trump businesses through colleagues, according to an NPR report.)
Redfield also declined to be interviewed or to respond to written questions. In a statement, he said: “I will do everything in my power to support [CDC scientists] as I maintain my commitment to lead this agency grounded in data, science, and public service.”
Birx’s admirers, including Emory University epidemiologist Carlos del Rio, credit her with sometimes pushing back against antiscientific White House policies. And Anthony Fauci, respected head of the National Institute of Allergy and Infectious Diseases and fellow task force member, has consistently praised Birx as someone who picks her battles to exert positive influence in the long run. He has suggested her recent trips to advise state health leaders have helped reduce COVID-19 cases.
“Birx is in a horribly difficult position,” says Nancy Cox, former director of CDC’s influenza division and now an affiliated retiree of the agency. “She wants to stay in the good graces of the president and the rest of the administration while trying to do the right thing with respect to public health. Do I view her as a good scientist who gets things done? Yes.”
But Birx’s lack of background in respiratory disease outbreaks has left her with blind spots, as has her management style, Cox says. “Do I view her as someone who is willing to railroad people into doing what she has decided is the right thing to do? Yes. Does she listen real well? Not necessarily. One result was the debacle with the hospital data.”
The interviews and documents obtained by Science show Birx replaced a functional, if imperfect, CDC data system—well understood by hospitals and state health departments—with an error-ridden and unreliable filter on hospital needs that sometimes displays nonsensical data, such as negative numbers of beds. Such problems could hamper effective distribution of federal resources during an anticipated fall and winter spike in COVID-19 and flu cases, CDC officials say.
“This is the surreal part of it: They are attempting to replicate something we built over 15 years. And they are failing,” says a high-level CDC official with personal knowledge of the system. “Either Birx isn’t looking at the data, or she’s looking away—because it’s an absolute disaster.”
When Trump installed Birx to coordinate the task force on 27 February, she was widely regarded as a strong choice, even by the president’s critics. Public health scientists hoped her “data driven” discipline, political adroitness, long-standing ties to Redfield, and military bearing—she spent 28 years in the Army—could insulate the pandemic response and CDC against some of Trump’s damaging impulses. Her new role gave her the ability to exert substantial authority over CDC (see organizational chart, below).
At that point, the agency was already struggling. Among other mistakes, CDC had botched the rollout of its initial COVID-19 tests in early February. And the Trump administration had attacked or muzzled agency leaders for speaking basic truths and repeatedly forced CDC to soften its scientific advice.
Birx appeared to have the experience to bolster the agency. She had worked as a fellow in cellular immunology under Fauci and later became lab deputy to Redfield at the Walter Reed Army Institute of Research (WRAIR), where he directed AIDS vaccine research. The pair co-authored numerous scientific papers during the late 1980s and early ’90s. When Redfield was investigated for possible misconduct after presenting overly rosy data on a “vaccine therapy” approach to AIDS treatment at a 1992 Amsterdam conference, Birx defended him to Army investigators. (The investigation found errors but cleared Redfield of misconduct charges.)
Soon after, Redfield retired from the military to co-found the Institute of Human Virology at the University of Maryland. Birx assumed his job at WRAIR, then rose to direct the U.S. Military HIV Research Program. She next moved to CDC, directing its global AIDS project for 9 years. In 2014, then-President Barack Obama appointed her as global AIDS ambassador and head of the Department of State’s PEPFAR. That project has been seen by many people as an exemplary foreign aid program. It is credited with saving millions of lives in the developing world since its founding in 2003 and sharply lowering HIV infection rates among girls and young women, among other achievements.
Birx set out to strengthen PEPFAR data systems and move funds to where they could save the most lives, those familiar with the program say. “PEPFAR had a mandate of doing more with the same budget,” says del Rio, who chairs PEPFAR’s Scientific Advisory Board. Birx succeeded by “squeezing out inefficiencies, and being very data driven,” and by cutting drug costs, he says.
Sten Vermund, dean of the Yale School of Public Health, led PEPFAR programs in Africa under Birx and her predecessor, Eric Goosby. Vermund praises Birx’s “industry and science” and notes that his Mozambique program got more funding under her leadership. Birx also deftly shepherded PEPFAR into the Trump era with no budget cuts, and her plan to refocus PEPFAR made sense, says Amanda Glassman, executive vice president of the Center for Global Development, a Washington, D.C.– and London-based research group.
But Glassman and Vermund also agree that Birx made constant, burdensome, urgent demands for data on HIV/AIDS cases, treatments, and other factors. Vermund says those demands sometimes disrupted services for patients. He describes some of Birx’s data demands as “almost whimsical” and their purposes opaque, calling her leadership style “authoritarian.”
That description is backed by a blistering audit of PEPFAR’s work in four African countries, issued in February by the Department of State’s inspector general (IG). Most of the PEPFAR staff auditors interviewed in 2019 said their input was ignored and that program heads, led by Birx, set unachievable benchmarks.
One PEPFAR staffer told auditors that a target for the number of people on antiretroviral therapy in one country “was actually greater than the number of people living with HIV.” Some staff described PEPFAR management as “autocratic” or “dictatorial.” One said high-level technical staff adopted an approach of “just obey and move on. … Working in fear and a space where nothing is negotiable.”
Programs that missed data quotas could have their funding cut off, a situation a third staffer described as “a recipe for cooking data.” The data targets “put a lot of pressure on the [local] partners,” a fourth staffer told the IG investigators. “Sometimes, you are not even sure that the numbers are true. Especially when you go to the field and look at the [patient] registers. You cannot verify that they are real patients.” A CDC PEPFAR manager told Science that “Countries need the money,” so program staff manipulated performance data.
Vermund says his program never falsified data, but “we knew for a fact that others did not necessarily tell the truth, … [using] exaggerations to make themselves look better.” He says some programs double-counted patients who entered treatment, dropped out, and then returned. “Perverse incentives were created based on the data-driven outcomes.” Despite the pressure, the audit showed, Birx’s data targets were often missed.
Amid the constant distraction of data demands, services to patients sometimes suffered, one staffer told the IG. That problem occurred in many PEPFAR nations, the CDC manager says.
In response to the IG report, Birx promised some reforms and “clearer, transparent dialogue.” She said local PEPFAR teams would set their own targets, although funding would be “adjusted to the presented level of ambition.” Before those actions were fully in place, Trump appointed her to coordinate the Coronavirus Task Force.
Del Rio isn’t surprised that some PEPFAR staff members were unhappy. “She’s a no-bullshit kind of person,” he says. “She’s not running a Montessori school.”
And Glassman notes that many women who attain powerful jobs face extra criticism or get tagged as authoritarian, whereas men with comparable leadership styles are simply accepted. “Is [Birx] getting pounded partially because she’s a woman?” she asks.
But Glassman concedes that despite good intentions, Birx’s style was “a disaster” at PEPFAR. And tying data targets to funding without independent verification does invite misreporting, she says. “Her desire to get those results and show them to the world, I think, just overcame everything,” Glassman says.
Now, some CDC staffers say, Birx is applying the PEPFAR playbook to the new pandemic, and the dismantling of CDC’s COVID-19 hospital data system is a consequence.
Hospital data debacle
Agency insiders concede that CDC’s National Healthcare Safety Network (NHSN)—the system used for 15 years to gather crucial data from hospitals—was far from perfect. The network, which collects data from about 37,000 hospitals and other health care facilities, has been underfunded for years. All the same, five times weekly, NHSN reliably produced actionable COVID-19 data such as available hospital beds, intensive care occupancy, and ventilators used, according to CDC sources and internal reports obtained by Science. CDC staffers used long-tested statistical algorithms to impute missing data.
When NHSN was shut down for hospital COVID-19 data in July, more than 100 public health and patient advocacy groups, along with scientific and medical societies (including AAAS, which publishes Science), warned that the switch could degrade crucial data reporting. Attorneys general for 21 states and the District of Columbia echoed their concerns. (The system still collects COVID-19 data from nursing homes and on issues such as on health care–associated infections and resistance to antimicrobial therapies such as remdesivir, which has helped some severely ill COVID-19 patients and was given to Trump.)
But del Rio says Birx viewed NHSN in a similar light as the data system she inherited at PEPFAR in 2014—woefully inadequate. Birx was exasperated, del Rio said, that NHSN could not provide daily, comprehensive data, for example on supplies of remdesivir, to guide the government’s efforts.
Either Birx isn’t looking at the data, or she’s looking away—because it’s an absolute disaster.
In a spring meeting, Birx seemed fixated on applying the lessons of HIV/AIDS in a small African nation to COVID-19 in the United States, says a CDC official who was present. “Birx was able to get data from every hospital on every case” in Malawi, the official says. “She couldn’t understand why that wasn’t happening in the United States” with COVID-19. Birx didn’t seem to see the difference between a slow-moving HIV outbreak and a raging respiratory pandemic. “[CDC Principal Deputy Director] Anne Schuchat had to say, ‘Debbi, this is not HIV.’ Birx got unhappy with that.”
Birx insisted every hospital update 100% of its data every day, including detailed patient demographics. She added new data categories, such as patient age and supplies of remdesivir. CDC officials told her 100% daily compliance was virtually impossible, but said NHSN statisticians could accurately extrapolate from partial data, providing results “in near real time,” one agency official says.
Another CDC official charged with responding to Birx’s demands calls her “fixation and fetishization of those daily count numbers” deeply frustrating. Birx’s top assistants accused CDC employees who pushed back of being callous about COVID-19 deaths. The process assumed the tone of a military command structure, the CDC official says: Obey without question. Echoing PEPFAR employees, the official adds that it seemed “designed to make you feel like you are failing every day.”
Birx “doesn’t really understand data,” says James Curran, public health dean at Emory University, who led CDC’s HIV/AIDS unit for 15 years before Birx’s time at the agency. “I don’t think she’s asking the right questions. It doesn’t mean that the CDC is always right. But you should have a partnership with people.”
Instead, say CDC sources with direct knowledge of the events, Birx’s team made a “take it or leave it” demand: Immediately collect case data by age, or NHSN would be replaced.
When the switch to TeleTracking was announced, Redfield applauded it as a way to “streamline reporting.” He also tweeted that the decision was made at “working levels” in CDC and didn’t rise to his level. Many people in CDC saw his comments as a betrayal.
Birx’s imperative of 100% of hospital data every day has proved elusive. The Department of Health and Human Services (HHS) data hub for hospital capacity, including inpatient beds occupied overall and by COVID-19 patients, now draws on data collected by TeleTracking, a for-profit company with nearly 400 employees, and on data submitted by state health departments and hospital associations. As with NHSN, nearly all data are collected manually rather than automatically from electronic patient records. Some hospital associations and health departments combine data from hospitals into spreadsheets and send them by fax or email to HHS. TeleTracking also offers a web interface for hospitals or their contractors to enter data.
The underlying data tables are updated daily but run 3 to 4 days behind—less efficient than NHSN before it ceased operating. CDC officials and public health experts blame several factors for those problems: Hospitals aren’t used to TeleTracking’s system and the additional data points (such as age) added work. Also, TeleTracking has long-standing relationships with fewer hospitals than NHSN; such relationships can speed troubleshooting.
In a 7 October written reply to questions, an HHS spokesperson acknowledged the time lag but called it “a good practice to provide the most complete dataset” because hospitals might not be able to respond on weekends, for example. But the department plans to reduce that lag, the spokesperson wrote.
At a 6 October press briefing, Birx said 98% of hospitals were reporting at least weekly and 86% daily. In its reply to Science, HHS pegged the daily number at 95%. To achieve that, the bar for “compliance” was set very low, as a single data item during the prior week. A 23 September CDC report, obtained by Science, shows that as of that date only about 24% of hospitals reported all requested data, including protective equipment supplies in hand. In five states or territories, not a single hospital provided complete data.
HHS said the 23 September analysis was “filled with inaccuracies, misunderstandings, and errors,” without providing details, and questioned its authenticity.
But Science also obtained a report prepared by CDC data experts for use by the agency and Birx’s team dated 19 May, back when TeleTracking and NHSN were both offered as options for data submission. NHSN showed 3% to 6% missing data for such items as COVID-19 inpatient bed occupancy and ventilator use. TeleTracking showed 36% to 57% missing data. Those numbers were mostly unchanged, with significant improvement in only one category, in the 23 September update, after NHSN was no longer used for hospital reporting.
Like NHSN, the HHS system—based on data provided by hospitals or others to TeleTracking, as well as on data reported to HHS directly by the states—estimates missing data, but HHS declined to release details for independent analysis. CDC staffers say the system has other problems; for example, many hospitals share ID numbers, making it difficult to differentiate between each one’s needs. (TeleTracking wrote in an email that it issues separate IDs for each hospital. But within the HHS system, the ID problem has not been resolved, according to a CDC official with direct knowledge of the issue.) CDC critiques obtained by Science and data displayed on the HHS data hub also show the system has consistently reported nonsensical numbers. For example, it showed negative numbers and estimates of occupied hospital beds, as well as more than 15,000 beds for a single California hospital. In nearly 1500 cases, it showed more beds filled than total beds in a hospital.
TeleTracking referred questions to HHS, which said it plans to boost automation to reduce errors. For now, the HHS spokesperson wrote: “HHS made a conscious decision to take a different approach on data collection. … Rather than reject incorrect data outright, HHS allows it to flow into our system” and then attempts to manually fix detected errors. A CDC data expert calls that “an admission of faulty data practices.”
In a 25 September memo from Birx to HHS Secretary Alex Azar, obtained by Science, Birx made a major concession to reality: She gave up on elements of the daily 100% compliance rule that had motivated the switch to TeleTracking. For example, Birx instead asked that inventories of supplies, such as personal protective equipment, be provided weekly because more frequent reporting had proved infeasible.
During the 6 October press briefing, Birx said the moves “ensure that we are not adding additional burden” for hospitals. The key is “valid and timely data,” she said.
But in that same briefing, Birx and Centers for Medicare & Medicaid Services Administrator Seema Verma announced a new and stringent requirement reminiscent of PEPFAR: Funding will be tied to reporting compliance. Hospitals will be disqualified from Medicare and Medicaid reimbursements if they fail to submit required data after being warned of lapses. The memo from Birx to Azar, marked “not releasable to the public,” shows Birx pushed for the change.
The move is “sledgehammer enforcement” that could needlessly divert time and money from patient care, Federation of American Hospitals President and CEO Charles “Chip” Kahn said in a statement after the announcement.
As at PEPFAR, CDC officials say this requirement could create “perverse incentives” to supply false data. Medicare funding is a survival issue for hospitals, so many are likely to submit the requested numbers regardless of whether they are accurate, say three high-level CDC officials with personal knowledge of agency data systems.
HHS cited safeguards against such fraud, including assistance from state health departments, and “systematic logic and error checking.” The department’s IG also identifies fraud aggressively, the spokesperson wrote.
However, the potential long-term impact keeps CDC staffers up at night, one says. “I worry, is this going to damage the whole process of how public health data are collected down the road?”
Lines of power
CDC officials say Birx acquired her outsize influence over the agency in part because of how power was allocated in the federal pandemic response. An organizational chart obtained by Science, marked “for official use only,” shows Birx coordinates the task force and co-chairs two key bodies: the unified coordination group, which manages the response from HHS and therefore CDC; and Operation Warp Speed, the vaccine development effort. A physician advisory group, comprising Fauci, Redfield, National Institutes of Health Director Francis Collins, and others—is off to the side (see organizational chart, above). Senior CDC people say those advisers have been reduced to “window dressing,” with little ability to mediate in the deteriorating relationship between Birx and CDC.
She calls into question the science of the agency.
The hospital data system is perhaps the most calamitous flashpoint in that relationship. But CDC officials say that, in other instances, Birx flouted science and undermined the agency to placate the president. For example, she responded with silence to Trump’s suggestion that ingesting disinfectants might cure COVID-19. And according to the nonpartisan FactCheck.org, in March she understated the pandemic’s spread by “misleadingly” portraying states with few cases as “almost 40% of the country,” although those states make up only about 7% of the population.
“Dr. Birx, what the hell are you doing? What happened to you? Your HIV colleagues are ashamed,” tweeted Yale epidemiologist and AIDS expert Gregg Gonsalves in response.
And she pressured CDC to tone down its guidance on school openings, according to The New York Times; it published an email she wrote asking Redfield to take a more permissive approach.
Several CDC leaders say Birx’s distrust and rejection of input from CDC data experts has created enormous animosity. “She calls into question the science of the agency,” says a current senior CDC official. “We’re not perfect … but in the midst of a crisis, to indicate that one of your chief arms for responding to a very severe pandemic can’t be believed” has been disastrous.
Birx’s supporters say she has done as well as anyone can working for Trump. “She can navigate science and politics,” del Rio says. “I don’t think anybody can navigate science and Trump.” And Birx has resisted some of the president’s outrageous claims, even persuading Pence to sometimes wear a mask. In August, she described the pandemic to CNN as “extraordinarily widespread,” losing some favor with Trump.
Even critics within CDC give Birx qualified credit. Her consistent push for testing has “gotten crosswise with the White House,” says a top CDC official, who adds: “At her core, she’s a scientist.”
But that official and others also see Birx as a cautionary tale of how an ostensibly well-meaning expert can cause great harm by working in the style of the Trump administration. “Bullying and threatening is a last resort for our usual way of operating, whereas that’s the modus operandi for this administration and the White House task force,” the official says. “‘We need people who think like we do.’ I’ve heard that stated in multiple instances” by Birx’s top assistants, the official adds. “It’s not scientific.”
Many executives and midcareer professionals who represent the future of the agency plan to leave if Trump wins reelection, several sources say. It’s another reason that has public health authorities wondering whether CDC has already passed a tipping point, from which it will struggle to recover no matter who is elected.
The implications of a discredited CDC for the COVID-19 pandemic are grave, says Thomas Frieden, who led the agency under Obama. If the public doesn’t trust government guidance to take vaccines when available, he says, the pandemic could rage indefinitely. “Breaking that trust could cost our economy—and American lives.”
Senior career executives at another beleaguered agency, the Food and Drug Administration, recently called for preservation of their scientific independence in a USA Today editorial. At CDC, leaders below Redfield are talking privately about whether to take a similar public stand against the destruction of their agency by the Trump administration. One says: “The longer we don’t speak out, the harder it will be to regain our credibility.”
*Clarification, 16 October, 5:30 p.m.: This story has been clarified to reflect the role of TeleTracking Technologies Inc. in the current HHS data hub. TeleTracking declined to comment before publication but later said the company was not responsible for negative values in the HHS data hub, and that it issued unique identifiers to each hospital. Science has verified that the negative values likely derive from other data sources used by HHS. Within the HHS data system, the hospital ID problem is ongoing, according to a CDC official.