Eligible travelers could volunteer to provide a self-collected nasal swab at the airport. Swabs from five to 25 travelers — from the same flight or country — were added to a single tube and then tested for the virus using P.C.R. testing. Positive samples were then sequenced to determine which version of the virus was present.

Between Sept. 29 and Jan. 23, 10 percent of 161,000 eligible travelers enrolled in the study, and 1,454 sample pools were tested for the virus. Despite the preflight testing requirement, more than 15 percent of the pools were positive.

This relatively high positivity rate may indicate that travelers were early in the course of their infections — and thus had viral loads too low for some tests to detect — when they took their predeparture tests, or that they contracted the virus in the time between being tested and landing in the United States, the researchers say. People may also have submitted fraudulent test results.

Before Nov. 28, nearly all of the positive sample pools contained the Delta variant, the researchers found. (The only exception was a positive sample whose exact genetic lineage could not be determined.)

After that, however, Omicron quickly came to dominate; from Nov. 28 to Jan. 23, two-thirds of the positive samples were Omicron. Most of the Omicron samples were the BA.1 subvariant, which was initially the most prevalent version worldwide. BA.1 remains the most common lineage in the United States, though BA.2 has been gaining ground and now accounts for 35 percent of infections, according to C.D.C. estimates.

But the researchers found the BA.3 subvariant in a pool of samples collected from travelers arriving from South Africa on Dec. 3. They reported the finding in GISAID, an international repository of viral genomes, on Dec. 22. It was the first reported case of BA.3 in the United States; it would be more than a month before the next was reported.

The airport program also detected the first known U.S. case of BA.2, in samples collected from South African travelers on Dec. 14. The researchers reported the finding a week before the next U.S. case of BA.2 was reported.

“This is a new tool in the C.D.C. tool kit that works, and we’ve shown it’s effective and it can be layered with all of our other mitigation measures,” Dr. Friedman said.