People with coronavirus infections of the omicron variant often have significantly different viral levels in their noses, throats and saliva, and testing just a single type of sample is likely to miss a large share of infections, according to two new papers, which analyzed omicron infections over time in a small number of people.

The papers, which have not yet been published in scientific journals, suggest that coronavirus tests that analyze nasal and throat swabs would pick up more omicron infections than those that rely on just a nasal swab. Although these combined tests are common in other countries, including Britain, none are yet authorized in the United States.

“You could get a lot more bang for your buck if you use these mixed specimen types,” said Rustem Ismagilov, a chemist at the California Institute of Technology and senior author of both papers.

Both papers are based on data collected during a study of household coronavirus transmission conducted in the Los Angeles area between Nov. 23 and March 1, when omicron was spreading rapidly.

The first paper focuses on 14 people who enrolled in the study before or at the same time that their infections began, allowing the researchers to capture the earliest stage of infection.

The researchers found significant differences in the viral load of different sample types from the same individuals.

In most participants, the virus was detectable in saliva or throat swabs before it was detectable in nasal swabs. But later, when the viral load did spike in the nose, it rose to higher levels, on average, than in either of the oral samples, the researchers found. Even then, however, there was significant variability.

Because of this variation, during the first four days of infection, “no single specimen type” will reliably catch more than 90% of infections, even with a highly sensitive PCR test, the data suggests.

In the second paper, the researchers assessed the performance of the Quidel QuickVue At-Home antigen test, which uses a nasal swab, in 17 participants who enrolled in the study early in the course of their infections.

The researchers found that even when people had viral loads high enough to be considered infectious in at least one type of specimen, the antigen tests were positive just 63% of the time.