After China relaxed the world’s most stringent COVID-19 restrictions in December, the virus exploded. Hints of the surge were everywhere: Hospitals turned away patients. Crematories were overwhelmed with bodies.

But China’s official COVID death toll for the entire pandemic remains strikingly low: 83,150 people as of Feb. 9. Researchers believe that number is a vast undercount, in part because it only includes infected people who died in hospitals.

While a precise accounting is impossible, public health researchers have been working to piece together the mystery of the outbreak that accelerated in December. Four separate academic teams have converged on broadly similar estimates: China’s COVID wave may have killed from 1 million to 1.5 million people.

All of the researchers consulted by The New York Times cautioned that without reliable data from China, the estimates should be understood as informed guesses, with significant uncertainty — although they fit the evidence far better than the official figures do.

The differences between China’s figures and researchers’ estimates are large. The official numbers would give China the lowest death rate per capita of any major country over the entirety of the pandemic. But at the estimated levels of mortality, China would already have surpassed official rates of death in many Asian countries that never clamped down as long or as aggressively.

At the same time, China would rank below Germany, Italy, the United States and other countries where outbreaks accelerated before vaccines became available.

Two of the estimates were in papers published in academic journals or posted for peer review, while two other analyses were shared by public health researchers in response to queries from the Times.

“If the data say what we think they say, this was an explosive wave,” said Lauren Ancel Meyers, a professor of biology and statistics at the University of Texas at Austin.

Why official data underrepresents China’s outbreak

China has a narrow definition of what counts as a COVID-19 death.

As crematories were inundated in December, Chinese officials only announced deaths that involved respiratory failure, leaving out infected people who died of liver, kidney or cardiac failure — an omission that was met with widespread skepticism. In mid-January, the government started releasing data on other deaths, but the figures are still incomplete.

Most glaringly, they exclude people who died outside hospitals. While it is impossible to know exactly how many deaths at home have been missed, from 2018 to 2020, only around one-fifth of all deaths in China occurred in hospitals.

The official figure is “certainly an underreport of all COVID deaths,” said Yong Cai, a demographer at the University of North Carolina at Chapel Hill who studies mortality in China. “There’s no question about that.”

While government data shows that China has doubled the number of intensive care beds since 2020, hospitals were still overloaded during the recent surge. Experts believe hospital deaths probably still account for only a small proportion of total deaths.

(BEGIN OPTIONAL TRIM.)

“With such a rapid spread, the ICU beds definitely were not enough to cope with the peak,” said Shengjie Lai, a public health researcher at the University of Southampton.

The number of people infected is unknown, which further complicates understanding the reach of the epidemic. After two years of widespread testing and quarantining, the Chinese government in December shuttered once-ubiquitous testing centers and made the reporting of self-test results voluntary.

Other data is missing. At least nine cities in different parts of China, including Beijing, have stopped publishing quarterly cremation totals.

(END OPTIONAL TRIM.)

An earlier estimate, based on the Shanghai outbreak

One estimate, published last year by scientists largely at Fudan University in Shanghai, used a previous omicron outbreak in Shanghai to estimate how quickly the virus might spread in mainland China.

The virus spread through the city early last year, before lockdowns and other social restrictions had a chance to slow it down. The researchers used data from that period to inform a disease model that estimated how a future outbreak might play out if strict control measures were removed.

The researchers made a number of assumptions: how many ICU beds would be available, when a lockdown would end and how quickly people would receive additional vaccines.

(BEGIN OPTIONAL TRIM.)

But if anything, the estimate might be conservative, said Bruce Y. Lee, an infectious disease modeler at City University of New York who was not involved in the research.

The study assumed an outbreak during the spring and summer, when more people are outdoors, meaning the rate of transmission would be relatively slow. But the virus took off in China in the winter.

“The evidence is that this virus is demonstrating seasonality,” Lee said. “If you had to guess, you would expect the reproduction rate to increase during the winter.”

(END OPTIONAL TRIM.)

The focus of the paper was on how treatment, vaccination and other measures might be able to slow the wave and reduce the toll. But the work was unwavering in its ultimate conclusion: Ending the “zero-COVID” policy was likely to overwhelm the health care system, producing an estimated 1.6 million deaths.

A more recent estimate, based on travel patterns

The toll of China’s outbreak would also have been influenced by the age, and the movements, of those infected.

In a more recent paper, three scientists at the University of Hong Kong estimated deaths by looking at how many people in each age group died during previous outbreaks in other countries, and adjusting the data for China’s demographics. Several researchers made similar calculations.

The Hong Kong researchers also modeled how increased travel around China’s Lunar New Year, the busiest travel period of the year, would help to spread the virus. They estimated that the surge might kill about 970,000 people by the end of January.

A retrospective estimate of a 90% infection rate

A third team of researchers shared another estimate with the Times, using information that became available after the worst of the outbreak had passed.

The researchers — Meyers at the University of Texas and Zhanwei Du, a public health researcher at the University of Hong Kong — found a unique way into another crucial question: How many people were infected? Even after China eliminated its mass testing program, health officials continued to test hundreds of thousands of people from around the country between mid-December and mid-January in an effort to track infection rates, according to a report from the Chinese CDC.

Based on that data, they inferred that 90% of the population was infected in little more than a month.

While the figure is high, public health researchers who were not involved with the project said such a rate was plausible. And in January, a leading government public health researcher said on Weibo, a Chinese social media platform, that 80% of the population had been infected.

When the researchers incorporated the timing of the outbreak, estimated fatality rates and the effect of vaccinations into a statistical model, they found that the outbreak may have killed about 1.5 million people. Given the uncertainties — like how quickly the vaccines took effect — a plausible range for the estimate was 1.2 million to 1.7 million deaths, Meyers said.

Numerous factors could affect how accurately the sampling program in China gauged the true number of infections, Meyers cautioned. She called those figures “highly uncertain” and pointed out that any inaccuracies would influence the estimate.

A back-of-the-envelope calculation based on fatality rates in the United States

Even the simplest calculations by disease modelers found that the number of deaths was very likely to be an order of magnitude higher than the official tally.

Jeffrey Shaman, a public health researcher and professor of environmental health sciences at Columbia University, started with a simple assumption, that the fatality rate for people infected in China was roughly the same as it presently is in the United States: 0.15%, or about 1 in 650 people.

Various factors could balance out, Shaman said. China uses different vaccines from those used in the United States. But China’s population had been less exposed to the virus by the time the outbreak hit, making it more susceptible. At a fatality rate similar to America’s, if 40% to 65% of China’s population was infected — a conservative estimate — then 900,000 to 1.4 million people may have died, he said.

(STORY CAN END HERE. OPTIONAL MATERIAL FOLLOWS.)

Ben Cowling, a public health researcher at the University of Hong Kong, arrived at a similar death toll by considering only the 82 million people in China ages 60 and older who were unvaccinated or had received fewer than three vaccine doses as of late November. If 80% of that group were infected, he would expect more than 1 million of them to have died, given their limited immunity and exposure to the virus, he told the Foreign Correspondents’ Club of China last week.

China is, after all, the only country in the world that faced its first major wave of infections without making any attempt to slow it, resulting in what Cowling conjectured was the fastest spread of a respiratory pandemic virus in modern history.

This article originally appeared in The New York Times.