The failure by the Centers for Disease Control and Prevention to quickly produce a test kit for detecting the novel coronavirus was triggered by a glaring scientific breakdown at the CDC’s central laboratory complex in Atlanta, according to scientists with knowledge of the matter and a determination by federal regulators.
The CDC facilities that assembled the kits violated sound manufacturing practices, resulting in contamination of one of the three test components used in the highly sensitive detection process, the scientists said.
The cross contamination most likely occurred because chemical mixtures were assembled into the kits within a lab space that was also handling synthetic coronavirus material. The scientists also said the proximity deviated from accepted procedures and jeopardized testing for the virus.
The Washington Post separately confirmed that Food and Drug Administration officials concluded that the CDC violated its own laboratory standards in making the kits. The substandard practices exposed the kits to contamination.
The troubled segment of the test was not critical to detecting the novel coronavirus, experts said. But after the difficulty emerged, CDC officials took more than a month to remove the unnecessary step from the kits, exacerbating nationwide delays in testing, according to an examination of federal documents and interviews with more than 30 present and former federal scientists and others familiar with the events. Many of them spoke on the condition of anonymity because they were not authorized to comment publicly.
This account confirms for the first time the contamination’s role in undermining the test and the CDC’s failure to meets its lab standards.
The development and rollout of the original kits are subjects of an investigation led by the Department of Health and Human Services, federal officials said.
The CDC – America’s premier institution for combating the spread of catastrophic disease – declined to make available for interviews those involved in the test design or manufacturing. A spokesman, Benjamin Haynes, provided a statement Friday that acknowledged substandard “quality control” in its manufacturing of the test kits.
Those efforts “were not sufficient in this circumstance,” the statement said. The agency also said it has “implemented enhanced quality control to address the issue.”
The CDC said the problems with the test kits might have resulted from “a design and/or manufacturing issue or possible contamination.”
Haynes also defended the CDC’s work, saying that earlier troubles were eventually ironed out.
“As of March 23, more than 90 state and local public health labs in 50 states, the District of Columbia, Guam, and Puerto Rico verified they are successfully using [the] diagnostic kits,” Haynes said in the statement.
Shortcomings with the tests were first noticed in late January, after the CDC sent an initial batch to 26 public health labs across the country. According to those with knowledge of what unfolded, false-positive reactions emerged at 24 of the 26 labs that first tried out the kits in advance of analyzing samples gathered from patients.
“Only two of them got it right,” said a senior federal scientist who reviewed the development of the kits and internal test documentation, and who concluded that the false positives were caused by contamination that occurred at the CDC.
The false positives arose during testing of “negative control” samples that contained highly purified water and no genetic material. That aspect of testing was essential to confirm that the test results were reliable and not because of contamination.
“The bottom line is, if you have a negative sample, and it’s coming up positive, the only way for that to happen is cross contamination. . . . There is no other explanation for it,” the scientist said.
Experts said the kits were contaminated before they were shipped out to the state health labs.
Stephen Morse, a retired senior CDC microbiologist, said the circumstances as reconstructed by The Post point to contamination as the cause of the false positives.
“With a negative control, there’s nothing there to be amplified unless there was some contamination present,” Morse said. “If your negative control is giving you a positive reaction, that’s indication of contamination.”
The possibility of contamination in the CDC lab was raised by Axios in a story in early March. “The big question: It was not immediately clear if or how possible contamination in the Atlanta lab played a role in delays or problems with testing,” the story said.
The CDC’s delay in changing course after the test problems has hindered efforts to contain the novel coronavirus, which emerged in China in late 2019. It grew to a regional outbreak and, ultimately, a pandemic that has wrought widespread death and an unprecedented shuttering of the U.S. economy. As of Saturday, the virus has infected at least 723,493 Americans and killed at least 34,214.
The failure with testing kept the public health labs from performing disease surveillance intended to predict and minimize harm before the virus became widely established in the United States. The impact has been magnified by the nation’s inability to rapidly expand the availability of testing.
There remains no proven cure or vaccine to prevent the onset of the virus, which scientists suspect jumped from an animal species to humans in Wuhan, China. Until effective medical countermeasures emerge, diagnostic testing is crucial to assessing the spread of the virus and containing it.
The FDA’s examination of the CDC’s test kit exonerated its design – but concluded that the problem was caused by substandard manufacturing practices, according to an FDA statement. The FDA regulates the safety and effectiveness of medical devices, including the test kits manufactured to diagnose covid-19 disease in humans.
The “CDC did not manufacture its test consistent with its own protocol,” the FDA told The Post.
The FDA declined to elaborate on its findings, but those with knowledge of the matter said the problem involved contamination in the manufacturing process.
“It’s critical that the tests used work, because false results can also contribute to the spread of covid-19,” the FDA said.
In late February, after an FDA official visited the CDC’s lab complex in Atlanta, he advised the CDC to discontinue manufacturing the kits, the regulatory agency said. In response, the CDC turned to an outside contractor to manufacture the remaining kits it had intended to make for public health labs.
The CDC’s performance with the test kits marks an unparalleled low in its often-proud, 74-year history.
“I was just saddened and embarrassed when this test didn’t work out,” said James Le Duc, a virologist and former CDC official who now is director of the Galveston National Laboratory in Texas. “It’s really a terrible black mark on the CDC, and the impact was devastating to the country.”
– – –
On Jan. 12, Chinese authorities made public a vital piece of medical information: the genetic sequence of the new coronavirus that was raging in the metropolitan area of Wuhan, about 500 miles west of Shanghai.
Because of the volume of global travel, cases would almost surely emerge in the United States. The genetic sequence was what scientists at the CDC needed to design a test kit for detecting coronavirus infections.
At the CDC’s labs in Atlanta, scientists went to work. Officials there assigned responsibility for designing the test kits to the CDC’s Division of Viral Diseases, whose stated mission is to “prevent disease, disability and death.”
Those familiar with the events said the design efforts were led by Stephen Lindstrom, an accomplished respiratory virus specialist who was a co-inventor of seven earlier CDC tests for strains of the flu. Lindstrom, who did not respond to a request for comment, was responsible for designing but not manufacturing the kits, scientists told The Post.
For reasons that have remained unexplained publicly, the CDC scientists chose complexity over simplicity in the test’s design.
The test kits featured two components that focused on separate regions of the virus’s genome, a standard approach. However, the CDC also outfitted the kits with a third component, a pan-coronavirus segment. That addition sought to identify a wider family of coronaviruses, of which covid-19 is the most recent strain to be observed in humans. Tests that were being developed abroad under sponsorship from the World Health Organization did not include this extra feature.
With the additional test component, the CDC’s scientists may have hoped to bolster the kits’ reliability in distinguishing covid-19 from other coronavirus strains.
One of them, severe acute respiratory syndrome, or SARS, originated in China in 2003 and killed 774 people worldwide, though none in the United States. SARS jumped from an animal species to humans, as was the case with the coronavirus. The CDC test component also sought to detect coronaviruses that are carried by bats.
The CDC’s extra test component was not essential to detecting the novel coronavirus and it complicated the test when speed was critical, many experts said.
“Either the CDC didn’t know it was a crisis or they should have pulled the plug on that faster,” said Paul Keim, a Northern Arizona University geneticist whose institute is testing for the virus. “They didn’t have a test that worked.”
Officials at the CDC chose to have the test kits manufactured in-house, instead of by an outside contractor. The CDC facilities are typically staffed by experienced microbiologists and technicians, and the labs had successfully made test kits for other pathogens. Producing reliable test kits requires rigorous quality control.
The kits were developed in a specialty lab that focuses on disease research and were assembled at the CDC’s Biotechnology Core Facility Branch, located at the agency’s headquarters in Atlanta.
On Jan. 17, five days after the Chinese made public the genetic sequence, Nancy Messonnier, director of CDC’s National Center for Immunization and Respiratory Diseases, said in a news briefing that agency counterparts in Japan and Thailand had already used DNA testing to detect coronavirus cases.
“We at CDC also have the ability to do that today, but we are working on a more specific diagnostic,” she added.
Messonnier apparently was referring to the extra, pan-coronavirus component that the CDC was designing into its test kit.
The tests used in Japan and Thailand had been developed by the WHO without the third component. A growing number of countries were turning to that test without difficulties.
Messonnier also said it was likely that the virus would appear in North America.
“It’s highly plausible that there will be at least a case in the United States,” she said.
On Jan. 21, Messonnier announced that a few days earlier the CDC had “finalized development” of its test and used it to confirm the first coronavirus infection in the United States, a man in Washington state who had traveled from the Wuhan region.
“In the coming weeks, we anticipate sharing these tests with domestic and international partners,” she told reporters.
While human-to-human spread had been confirmed in China, Messonnier said, “we continue to believe the risk of this novel coronavirus to the American public at large remains low at this time.”
After using the test to successfully diagnose the first U.S. patient, CDC technicians began a second phase of development – manufacturing the batch of kits that would go to the 26 public health labs. It was during this phase that the chemical materials for the kits became contaminated, according to a scientist with knowledge of what happened.
“The first lot, they did not find any issues,” the scientist said. “They used the same [genetic] sequence for the second lot. . . . The second lot they manufactured ended up getting cross-contaminated.”
In the third week of January, the CDC shipped out the kits to more than two dozen public health labs scattered across the country, from Albany, New York, to Richmond, California.
As designed, the kits required the labs to use a highly sensitive molecular technique called polymerase chain reaction, or PCR.
The testing relies on a multistep regimen that starts when a six-inch cotton-tipped swab is used to gather a sample of mucus from a person’s nasal passage or throat. That sample is delivered to a lab in a sealed container.
At the lab, nucleic acid is extracted from the sample and placed into a small tube, along with solutions of various chemical reagents, including an enzyme that converts viral RNA, if present, into DNA.
Once the DNA is made, portions of the solution are transferred to tiny plastic cups, containing additional reagents to help detect whether the virus is present. The cups are placed into the PCR machine, which roughly resembles a midsize office photocopier.
The process seeks to copy and amplify targeted regions of the coronavirus genome. If the virus is present in the original sample, a detectable, fluorescent dye is released.
The CDC provided most of the necessary materials for each of the kits’ original three components.
The labs were instructed by the CDC to demonstrate that the test would work before analyzing samples from patients.
But when those facilities began using the kits to analyze a negative control sample – highly purified water supplied by each lab and free of any genetic material – the tests wrongly signaled the presence of the coronavirus.
“It’s not the water that’s contaminated,” said the senior federal scientist who reviewed what went wrong with the kits. “It’s one of the reagents.”
The precise means of contamination may not be knowable.
Scientists experienced with such lab work pointed to several possibilities, including inadequate decontamination of an enclosed area called a hood, where technicians may have worked with the synthetic coronavirus material. Improper handling of reusable lab devices also could contaminate the reagents.
The synthetic, or man-made, viral material that was used reduced the chance of infecting lab workers.
The widespread false positives point to a central source of contamination – the CDC’s manufacturing and assembly of the kits, the senior scientist and others said. The problems were observed in the test’s pan-coronavirus component.
Those familiar with what unfolded when the kits were tried out also said the contamination appeared to be at low levels.
For instance, they said, the false signals emerged only after the molecular testing had run for 34 to 36 cycles. Detection more typically occurs at 25 to 30 cycles, in which all of the test ingredients are heated, cooled and reheated from about 160 to 204 degrees.
“On known negative samples, two out of the three [segments] were negative as they should be, but there was a little bit of reactivity with the third one,” said a supervising scientist at one of the state labs that had a false-positive result with the pan-coronavirus segment.
This disqualified the entire test, said the scientist, who spoke on the condition of anonymity because he had been instructed not to comment publicly.
Even a trace of coronavirus-like material in lab spaces at the CDC where the kits were assembled could have caused the contamination, those familiar with the matter said.
Had the kits been used to analyze patient samples: “That means when you amplify that [sample], you have no way of distinguishing whether it’s coming from a real covid-19 source or not,” said a longtime federal microbiologist who spoke on the condition of anonymity to be candid. “That’s really bad.”
Rigorous validation and record-keeping should have detected the contamination before the CDC distributed the kits, he and other scientists said.
The CDC’s lab standards are based on federal Clinical Laboratory Improvement Amendments. The protocols are designed to catch errors in the manufacturing process, build in corrective measures and ensure that scientists keep an exhaustive record of their work.
“My question is – where was the adult supervision?” a former CDC lab chief said. “A competent laboratory would not have that problem. I don’t really understand how the kits got out without detecting a problem.”
– – –
The first public hint of trouble with the test came during a Feb. 12 press briefing in which the CDC’s Messonnier mentioned unspecified “issues” bedeviling the public health labs. At the time, most American clinics and hospitals remained unable to test for the coronavirus.
“Some of the states identified some inconclusive laboratory results,” said Messonnier, speaking to reporters by phone.
Messonnier suggested that the cause of the unexpected results remained elusive. The CDC’s goal, she said, was to make sure “that the laboratory results are correct.”
“We have multiple levels of quality control to detect issues just like this one,” she said. “We’re looking into all of these issues to understand what went wrong, and to prevent these same things from happening in the future.”
A reporter pressed Messonnier to elaborate.
“We think that the issue at the states can be explained by one reagent that isn’t performing as it should consistently, and that’s why we are remanufacturing that reagent,” she said.
At the public health labs, officials struggled to figure out what was wrong. Some labs determined that the test would work without the third component. But under the CDC’s emergency instructions, health officials had to use the test as it had been designed.
As the lack of reliable testing for the virus persisted deep into February, FDA officials based in Silver Spring, Maryland, were unable to get a satisfactory explanation from the CDC of what was wrong with the test, according to the regulatory agency and individuals familiar with the events.
By Feb. 23, the number of Americans who were confirmed as infected by the virus had climbed to 53, spanning six states, according to the CDC. The World Health Organization reported 78,811 cases globally.
That weekend, Timothy Stenzel, a top FDA official for regulating diagnostic devices used for medical treatment, traveled to Atlanta to meet with the CDC’s scientists and to see firsthand the lab areas where the kits had been developed and assembled.
According to the FDA, Stenzel for nearly a month could not determine, based on information provided by the CDC, whether the kits were failing because of a “design or manufacturing issue.” With demand for testing surging, some of the state and local labs were using the original kits to analyze samples drawn from patients, on the condition that results would be confirmed by additional testing by the CDC.
Stenzel would evaluate whether the CDC was suited to continue making coronavirus test kits in-house, according to interviews and written responses from the FDA. He also would assess whether companies should be allowed to use the CDC’s design to make and distribute higher volumes of the test kits.
Hired in August 2018 as director of the FDA’s Office of In Vitro Diagnostics and Radiological Health, Stenzel was trained as both a physician and a Ph.D. microbiologist/immunologist. He had founded a molecular diagnostics lab at Duke University and, during 15 years as an industry executive, helped develop dozens of sophisticated tests, including an FDA-approved assay for detecting pancreatic cancer.
During his visit in Atlanta, Stenzel determined that the problems with the coronavirus test were caused by the CDC’s manufacturing, not the design, according to the FDA. The shortcomings with the test kits were attributable to what the FDA described as a “manufacturing issue.”
Stenzel advised CDC officials to stop making the kits in-house.
The CDC was “expected to make a quality product” and was required to comply with sound manufacturing practices, the FDA said.
Stenzel declined to be interviewed.
In response to questions, the FDA said Stenzel “worked with CDC to facilitate the production and quality control processing of test kits,” made ultimately by the contractor, Iowa-based Integrated DNA Technologies. Stenzel also worked with the CDC to “expedite test kit distribution” to public health and commercial labs.
“The test manufactured by IDT was distributed and has encountered no issues, thus supporting the conclusion” that the CDC’s manufacturing had caused the original kits to fail, the FDA said.
By this point officials at the public health labs widely viewed the extra, pan-coronavirus component of the CDC’s test kit as unreliable. Amid those concerns, the FDA on Feb. 26 informed the CDC by email that the labs could begin testing samples while skipping the third component.
On Feb. 28 – 47 days after the Chinese distributed the virus’s genetic sequence – Messonnier announced that “labs can start testing with existing CDC test kits.”
In the news briefing, Messonnier also said that the CDC had “established that the third component . . . was the cause of the inconclusive results” and “can be excluded from testing without affecting accuracy,” she said.
Messonnier said nothing about the FDA’s recommendation that CDC stop making the test kits in-house.
“We are working as quickly as we can to get CDC test kits to state and local public health authorities,” Messonnier said. “To date, our strategies have been largely successful.”
That week, the CDC reported that 1,007 people had been tested nationwide. That compared with more than 420,000 tests that had been performed worldwide.
The next day, on Feb. 29, the CDC announced the nation’s first death from the virus, a man in his 50s in Washington state.
Citing “unfolding situations” in other states, a CDC news release said that “preliminary information raises the level of concern about the immediate threat of COVID-19 for certain communities in the United States.”
On March 2, the FDA endorsed the release of the newest kits – assembled by Integrated DNA Technologies. Still, patients and health-care providers struggled to secure testing and have continued to wait for many days or even weeks for results.
CDC officials have been tight-lipped regarding what went wrong with the test kits.
At a March 3 news briefing, Messonnier was asked about potential contamination.
“Contamination is one possible explanation but there are others,” she said. “And I really can’t comment on what is an ongoing investigation.”
The failure with the test kits was highlighted at a congressional hearing on March 11 that examined the government’s preparedness and response to the virus.
“The Trump administration’s testing for the coronavirus has been severely inadequate,” said the chairman of the House Oversight and Reform Committee, Rep. Carolyn Maloney, D-N.Y. “If you don’t test people, then you have no idea how many people are infected.”
“We don’t know where to direct resources,” she said. “We are operating in the dark.”
In testimony, CDC Director Robert Redfield described in general terms what may have caused the kits’ failure.
“The third control did not perform the way we wanted it to perform,” he said, adding that the cause was either “a contamination” or an unspecified “biologic” factor that caused the test materials to malfunction.
Asked about Redfield’s testimony, the federal scientist who reviewed the internal test data said the kits steadily amplified nucleic acid within what should have been the DNA-free negative control samples. That pattern of amplification, he said, could only have been caused by contamination, not by any other design or manufacturing flaw.
When a committee member, Rep. Raja Krishnamoorthi, D-Ill., asked about contamination, Redfield said: “This is currently under an investigation at this point, and I think I’m going to leave it there.”
Krishnamoorthi excoriated the CDC’s performance.
“When we don’t test as rapidly as we should, the virus spreads and people die,” he said, noting that South Korea, Italy and other nations had tested far greater percentages of their populations.
Haynes, the CDC spokesman, said Messonnier and Redfield were not available to comment.
The CDC’s refusal to promptly jettison the problematic first test kit puzzled many who were seeking prompt, reliable testing.
“They just kept doubling down on what they knew was a poor performing assay, and that has really bit us in the butt,” said James Lawler, a physician at the University of Nebraska Medical Center who has treated covid-19 patients.
Keim, whom the FBI relied on for testing during the bureau’s investigation of the 2001 anthrax letter attacks, noted that although the additional test segment was apparently intended to help distinguish covid-19 from the other coronaviruses, it wasn’t needed: covid-19 has a distinct genetic sequence.
This made searching for the other strains superfluous.
Among the known coronaviruses, covid-19’s nearest genetic neighbor is SARS, Keim said. Although SARS and covid-19 are 85% identical when they are analyzed with the amplifying powers of PCR molecular testing, that gap is an unmistakable distinction, Keim said.
“Fifteen percent is a massive difference when it comes to PCR,” Keim said. Covid-19, because of its dissimilar genome, “is like the easiest target in the world. . . . It’s not a hard thing to develop an assay to.”
– – –
The Washington Post’s Alice Crites contributed to this report.