In raw milk samples spiked with high amounts of bird flu virus, small amounts of infectious virus were still detectable after treatment with a standard pasteurization method, researchers said today.

The findings reflect experimental conditions in a laboratory and should not be used to draw any conclusions about the safety of the U.S. milk supply, according to the authors of the study from the U.S. government’s National Institute of Allergy and Infectious Diseases (NIAID) Rocky Mountain Laboratories.

The research was published in the New England Journal of Medicine.

Compared to the spiked raw milk with virus floating freely used in the study, raw milk from cows infected with H5N1 influenza may have a different composition or contain virus inside of cells, which may impact heat effects, the researchers said.

U.S. dairy cows were found to be infected with bird flu in March. The U.S. Food and Drug Administration surveyed pasteurized retail samples of milk and estimated that a fifth of the U.S. milk supply contained strands of virus. The agency has said that pasteurized milk is safe to drink.

The virus used in the experiments had been isolated from the lungs of a dead mountain lion, mixed with raw, unpasteurized cow milk samples, and heat-treated at 63 degrees C (145.4 degrees F)and 72 degrees C (161.6 degrees F) for different periods of time.

After treatment at 72 degrees C for 20 seconds – five seconds longer than the industry standard for pasteurization at that temperature – very small amounts of infectious virus were detected in one of three samples, the study found.

“This finding indicates the potential for a relatively small but detectable quantity of H5N1 virus to remain infectious in milk after 15 seconds at 72 degrees C if the initial virus levels were sufficiently high,” the authors note.

Within 2.5 minutes, treatment at 63 degrees C caused a marked decrease in infectious H5N1 virus levels, indicating that standard industry pasteurization of 30 minutes at that temperature would eliminate infectious virus, the researchers said.

The researchers said that their experimental conditions are not identical to large-scale industrial pasteurization processes for raw milk and that their findings need to be replicated with direct measurement of infected milk in commercial pasteurization equipment.

It remains unknown whether ingesting active H5N1 virus in milk could cause illness in people, the researchers added.