The debate about the findings underscores just how much remains to be known about the coronavirus, which has killed at least 187,000 people in the United States, according to a Washington Post analysis.

“This is consistent with the growing body of data that suggests that there’s been community spread much earlier than we had anticipated,” said study author Joann G. Elmore, a doctor and epidemiologist at the David Geffen School of Medicine at the University of California at Los Angeles.

The researchers examined six years of electronic health records, representing nearly 10 million patients, at the UCLA health system from July 2014 through February. That included patient visits to three UCLA hospitals and to nearly 200 associated outpatient clinics.

Health agencies have surveillance systems in place to detect the early signals of disease outbreaks, such as a rise in patients with fever checking into hospitals. But medical records were an under-tapped resource, Elmore said. “People weren’t paying attention to the outpatient setting,” she said.

The study authors searched outpatient and emergency department reports that used the word “cough,” and tallied the number of people hospitalized for acute respiratory failure.

That approach revealed an uptick in patients that began the week of Dec. 22 and remained elevated for 10 weeks. The number of extra people exceeded the researchers’ predictions by 50 percent, totaling about 1,000 more patients compared with the previous five flu seasons.

Influenza cannot be ruled out as a cause of the increase, Elmore said. “And, you know, we did see a bad bout of flu this year,” she said. But what gave her pause was the consistent, weeks-long trend found only in this most recent season and not others.

Some experts said they doubted that coronavirus infections were the likely cause of respiratory problems in California so far back in time. “The data countywide would suggest that it really began to spread in March,” said Brad Spellberg, chief medical officer at the Los Angeles County+USC Medical Center, who was not involved with the new research.

Although the virus may have infected a small number of people sooner than previously reported, Spellberg said he doubted that “meaningful transmission” occurred in December or January.

Using data from emergency departments that reported patients with flu-like illnesses, Spellberg and his colleagues observed two peaks in patients in December and February, as they reported in JAMA this spring.

Those increases were consistent with a severe flu season, Spellberg said. Los Angeles’s third spike in flu-like illnesses, this time caused by the coronavirus, came later.

What’s more, between March 2 and March 18, only 5 percent of 131 patients with flu-like illnesses tested positive for the coronavirus in the JAMA study. Spellberg said that if the virus had an earlier foothold in California, he would have expected that percentage to be higher. “You would have seen an explosion of cases,” he said.

Understanding how long the virus circulated within a population helps refine epidemiological models of transmission. Infectious-disease scientists and doctors in many pockets of the world are eager to uncover when the coronavirus first spread outside of China.

In late December 2019, Chinese health officials identified clusters of viral pneumonia in Wuhan. Researchers sequenced the culprit’s genome, describing the new coronavirus strain, in early January. The first officially reported U.S. case of coronavirus, a man who traveled home from Wuhan, occurred two weeks later.

A few observations indicate that the virus may have traveled farther, earlier, before it flared into a global pandemic. A study of Italian sewage revealed traces of the virus in December. When researchers retested a nasal swab from a man hospitalized near Paris dating to Dec. 27, they detected the coronavirus.

Study author Judith Currier, a UCLA infectious-disease physician, said that when it comes to people who wonder whether they were exposed to the virus many months ago, she does not recommend “antibody testing for people who never had a symptomatic illness,” citing guidelines from the Centers for Disease Control and Prevention.

“If someone had a compatible clinical illness but never had testing for covid during that time, antibody testing could help to confirm,” she said. “Although we don’t know how long the antibodies last, so it would not be definitive."