- New cases of a deadly coronavirus running through China surged at the start of February, then they leveled off, then they jumped by more than 19,000 overnight.
- Those swings might not have as much to do with the reality on the ground as they do how Chinese authorities count a "confirmed case."
- While the methodology changes have bred confusion, epidemiologists say it wasn't likely a deliberate attempt to manipulate the data.
New cases of a deadly coronavirus running through China surged at the start of February, then they leveled off, then they jumped by more than 19,000 over 24 hours.
But those swings might not have as much to do with the reality on the ground as they do how Chinese authorities define a "confirmed case."
China's National Health Commission has now revised national guidelines on responding to the COVID-19 outbreak at least six times since Jan. 22. Two of those changes implemented after Feb. 12 have substantially changed daily case counts.
The World Health Organization said their field team that returned Monday from China's Hubei province, the epicenter of the outbreak, and other areas of China concluded that the virus "peaked and plateaued" between Jan. 23 and Feb. 2. But independent researchers wonder about the reliability of China's data and are questioning whether the peak is valid.
When the outbreak first emerged in Wuhan two months ago, China's system of reporting new cases was straightforward enough. There were lab-confirmed cases, which had been verified with one of the government-authorized nucleic acid tests, and there were suspected cases, which included those who exhibited symptoms but had not tested positive.
On Feb. 12, China's National Health Commission quietly re-categorized patients who exhibited symptoms, as determined by a chest scan, as "confirmed cases" even if they had not tested positive in the lab. The new definition only applied to Hubei province, the epicenter of the outbreak. (Separately, China's health commission on Feb. 14 said it removed 108 deaths from its total count to correct some double-counting in Hubei province, which doesn't appear to have stemmed from the methodology change.)
The case definition was changed, according to a CNBC translation of the official announcement's Chinese text, so a broader set of patients could receive the same treatment as confirmed cases. WHO officials applauded the decision as a move toward transparency. The change took a few days to show up in China's reported numbers, but when it did, it sent shock waves throughout the globe. China's total cases soared by more than 19,000 — from 51,174 on Feb. 16 to 70,635 the next day, according to WHO data.
A few days later, on Feb. 20, China's National Health Commission reversed the change, removing some previously confirmed cases from its tally and causing its case count to level off as reports of new cases fell sharply. On Friday, state news agency Xinhua reported that an official for the province's health commission directed officials to stop retroactively deleting confirmed cases from the province's official count.
While revising diagnostic methodology makes sense when tracing a rapidly evolving outbreak, China's lost a lot of credibility and faith in the accuracy of its data with its swift and frequent changes, international researchers say.
It serves no scientific purpose, Dr. Eric Feigl-Ding, a global health economist at the Harvard Chan School of Public Health, said of China's frequent revisions.
"It's OK to switch the system once, like last week, but now to switch it back and then switch it back again, it makes comparison very difficult," he said. "Being consistent is really important. For comparison, you want an apples-to-apples comparison as much as possible. Here, there's a lot of apples and oranges."
Part of the problem with the Feb. 12 change that sent total cases soaring, Feigl-Ding said, is that there were no dates associated with the cases involved in the spike. Chinese officials said that part of those 19,000 new cases were older cases that had been clinically diagnosed, but not confirmed in a lab. Feigl-Ding said that meddled the outbreak's trajectory.
While the methodology changes have bred confusion, Feigl-Ding said he doesn't believe it's necessarily a deliberate attempt to manipulate the data. However, he said that the frequency of the changes only gives fodder to those who say China is purposefully under reporting the numbers.
The White House doesn't "have high confidence in the information coming out of China" regarding the count of coronavirus cases, a senior administration official told CNBC last week. And the regular changes in reporting methodology don't help matters, Feigl-Ding said.
"This kind of confusion breeds distrust," he said. "You should stick to a standard even if you have caveats."
The original change in reporting last week signaled that Chinese health providers were likely struggling to efficiently administer the official COVID-19 test kits, director of global health studies at Seton Hall University Yanzhong Huang said. These tests are a form of nucleic acid test, which are often used to detect viruses like Hepatitis and Ebola. Such tests can have as low a rate of accurate diagnosis as 50%, Huang said, and that figure could be even worse in a city like Wuhan that has such a stressed healthcare system amid the outbreak.
"When you're facing an outbreak like what is happening in Wuhan, it can be so overwhelming that it may basically overwhelm their capability to provide even the testing services," he said. "If you don't handle it well, or you are not adequately trained, that kind of test leads to a false negative or positive."
If healthcare workers across Hubei province don't have the resources or time to properly administer the tests, Huang said it's better to rely on symptoms for official counting so that more people can receive treatment.
On the other hand, the reversal of the policy on Feb. 20, he said was likely a sign that national authorities have bolstered resources and health workers in Hubei province. An example of this would be the two hospitals built recently in Wuhan to treat COVID-19 patients.
"It probably means that Wuhan has improved their testing capacity," he said. "More health-care providers have been mobilized from other parts of China, so there's more people to do more tests to make it more accurate. Also, I think they're better trained to accurately read the tests."
Huang's point was echoed by director-general of the World Health Organization Tedros Adhanom Ghebreyesus on Friday when reporters asked about the changes.
"This may indicate the system in Wuhan has regained ability to test all suspected cases," he said. "We're glad that China has come back to that kind of counting and this will bring clarity, actually."
Walter Zhang, assistant president of Liferiver Bio-Tech, a Shanghai-based company that quickly developed a test kit for COVID-19 that is employed by Chinese health officials, stands by the accuracy of his company's tests. However, he acknowledged that the test requires proper training and equipment to yield accurate results.
Zhang said his company's tests are able to directly detect the COVID-19 pathogen, but require a special machine called a "thermocycler," which can cost thousands of U.S. dollars. He acknowledged also that the "pathogen is not detectable when its volume is very low."
"This detection requires reagents, machines and skilled personnel," he said in an emailed statement, adding that if properly equipped, medical providers can complete the test in just a couple of hours.
Lieven Gevaert, owner and CEO of Genprice, another biotechnology company that has developed a coronavirus lab test, echoed that point.
"Thermocyclers are expensive instruments in terms of investment ... and need qualified researchers to work with," he wrote in an email. "Not all hospitals have a thermocycler available in China."
CNBC's Evelyn Cheng contributed to this article.