A large new study provides some of the strongest evidence yet that vaccines reduce the risk of developing long COVID.

Scientists looked at people in the United States infected during the first two years of the pandemic and found that the percentage of vaccinated people who developed long COVID was much lower than the percentage of unvaccinated people who did.

Medical experts have previously said that vaccines can lower the risk of long COVID, largely because they help prevent severe illness during the infection period and people with severe infections are more likely to have long-term symptoms.

But many individuals with mild infections also develop long COVID, and the study, published Wednesday in the New England Journal of Medicine, found that vaccination did not eliminate all risk of developing the condition, which continues to affect millions in the United States.

"There was a residual risk of long COVID among vaccinated persons," Dr. Clifford Rosen, a senior scientist at the MaineHealth Institute for Research, who was not involved in the study, wrote in an accompanying editorial. Because of that, Rosen added, new cases of long COVID "may continue unabated."

The study evaluated medical records of millions of patients in the Department of Veterans Affairs health system. It involved nearly 450,000 people who had COVID between March 1, 2020, and Jan. 31, 2022, and about 4.7 million people who were not infected during that time.

The veterans study population was significantly less diverse than the general American population. Nearly three-fourths of the participants were white, about 91% were male, and their average age was 64.

The researchers analyzed the health records to estimate the percentage of COVID patients who had long COVID one year after being infected. The lowest rate of long COVID in the study, 3.5%, was among vaccinated people who were infected during the latest period in the study, between mid-December 2021 and January 2022.

That compares with a rate of 7.8% for unvaccinated patients in the study who were infected during the same period.

"We found that much of the decline is attributable to vaccination," said the study's lead author, Dr. Ziyad Al-Aly, the chief of research and development at the VA St. Louis Health Care System and a clinical epidemiologist at Washington University in St. Louis.

Still, he said, "vaccine effectiveness wanes considerably with time, and people are not keeping up with yearly vaccine shots."

He added, "We cannot have our cake and eat it, too. We cannot say that long COVID is down because of vaccines and then abandon vaccination. This will result in the rise of cases again."

To rule out other possible causes, the researchers factored in comparisons between uninfected people who developed similar symptoms, Al-Aly said.

For example, major long COVID symptoms such as fatigue and brain fog can also affect patients with cancer and other conditions, so the authors subtracted the rate of those symptoms in the uninfected population from the rate in infected people to calculate the percentage attributable to long COVID, he said.

The study covered the time period from the initial emergence of the coronavirus through the arrival of two increasingly contagious variants — delta and omicron — after the rollout of vaccines. The authors compared results among vaccinated and unvaccinated patients, but did not compute a rate for both groups together.

Researchers found that among unvaccinated people infected between June 19 and Dec. 18, 2021, when delta was the dominant variant, the rate of long COVID a year later decreased slightly to 9.5% from 10.4% among those infected in the first 15 months of the pandemic.

The rate decreased further — to 7.8% — among unvaccinated people infected between Dec. 19, 2021, and Jan. 31, 2022, during the omicron wave.

Among vaccinated people who had been infected, the rates of long COVID were markedly lower. Differences in variants and other aspects of the delta and omicron periods played a role, the authors said, but they attributed about 72% of the decrease to vaccines.

About 5.3% of those infected during the delta period had long COVID a year later, and 3.5% of those infected during the omicron period did.

"This is lower than earlier phases, but it is not low," Al-Aly said. "Multiplied by the huge number of people who continue to get infected and reinfected, 3.5% per 100 adults infected will translate into millions of additional cases of long COVID."

The researchers did not look at later time periods, but a recent survey by the Centers for Disease Control and Prevention reported that about 5.3% of adults in the United States — about 13.7 million people — currently have long COVID.

The authors said that long COVID symptoms in most categories, including cardiovascular and kidney problems, decreased over the pandemic's first two years, but gastrointestinal, metabolic and musculoskeletal problems increased during the omicron era in unvaccinated people, probably reflecting changes in the virus and other factors.