Federal officials are cracking down on the proliferation of unauthorized antibody tests for COVID-19 in the United States, as lab directors say the embattled form of testing still has an important role to play in public health.

A Minneapolis company, Premier Biotech, is one of four firms under investigation by a U.S. House subcommittee over concerns their tests are not as accurate as what’s claimed on their labels. The company stands behind the China-made test and says its accuracy has been validated by outside researchers. An independent researcher said in an interview that the Premier Biotech test is widely used because it is accurate.

But the House Subcommittee on Economic and Consumer Policy is relying on non-peer-reviewed data to lambaste the tests, which are supposed to tell whether someone has had COVID-19 by detecting antibodies to the virus. The subcommittee took credit last week for prompting the Food and Drug Administration to reverse course and tighten up policies that had allowed more than 150 unauthorized antibody tests for COVID-19 to flood the U.S. market.

Under the revised guidelines, all commercial manufacturers of antibody tests have until May 18 to submit the results of a clinical study on their tests to the FDA, an FDA official confirmed. The FDA wants commercially manufactured tests to score 90% accuracy in positively identifying a single kind of antibody in a sample and 95% accuracy in identifying samples that don’t contain that antibody.

Because of the error rates, antibody tests can’t provide definitive evidence that someone has had COVID-19. But public health officials say high-quality, well-characterized tests can still be valuable in determining how widely the virus has spread in a given community.

The body naturally makes antibodies to fight off a COVID-19 infection, so antibody tests can be used to tell who has had the disease. Theoretically, if enough people in a community had antibodies to COVID-19, then “herd immunity” could protect those who lack the antibodies.

Governors such as Minnesota’s Tim Walz and New York’s Andrew Cuomo have talked about using antibody tests to measure antibody prevalence to help decide when it’s safe to open shuttered businesses. But such hopes have been dashed by a lack of strong evidence backing the tests, and the major unanswered question of how well antibodies protect against re-infection.

On Sunday, the Minnesota Department of Health reported 481 new confirmed cases of COVID-19 in the state, bringing the state’s tally to 11,271.

But such tallies don’t represent every case in the state, because of limited supplies of the sophisticated lab-based RT-PCR tests that can diagnose a case by detecting a virus’ genetic material. Such tests rely on the availability of special chemicals called “reagents” to drive the biological reactions that can reveal the virus’ presence. Last week public health officials changed course and began to urge people with symptoms to contact their doctors and get tested.

Antibody tests like the ones distributed by Premier Biotech have been seen as a way to address a lack of widespread diagnostic testing.

The commercially available tests are quick, simple and inexpensive, and they use blood samples instead of long swabs inserted into the nose or throat. Some antibody tests can only be done on special lab equipment, while others are “lateral flow” tests that produce visible lines on disposable cassettes, like a pregnancy test.

Their use became controversial following a series of decisions by the FDA in March that allowed scores of tests to enter the U.S. market without any authorization from regulators. FDA officials said they allowed these “serology” tests to hit the market so quickly because the tests were never intended to be able to diagnose an active case of COVID-19, but they’ve since reconsidered that approach.

“In mid-March, it was critical for the FDA to provide regulatory flexibility for serology test developers,” FDA Deputy Commissioner Dr. Anand Shah and FDA devices director Dr. Jeff Shuren wrote in a May 4 commentary. “However, flexibility never meant we would allow fraud. We unfortunately see unscrupulous actors marketing fraudulent test kits and using the pandemic as an opportunity to take advantage of Americans’ anxiety.”

But which ones are fraudulent? Only a handful of independent studies so far have sought to objectively measure the accuracy of the tests, and the one that attracted the House subcommittee’s interest is a draft manuscript posted online by researchers at the University of California, San Francisco, and University of California, Berkeley.

The study, which has not been peer-reviewed by independent researchers, evaluated 12 antibody tests using 238 positive and negative human specimens and found some did not live up to the performance listed on the package insert. The manuscript carried a disclaimer that said the data were preliminary and should not be relied on by doctors or reported as “established information.”

The House subcommittee sent letters demanding explanations to four companies, including Premier Biotech, whose test showed “drastically lower sensitivities” to the antibodies abbreviated as IgG and IgM. The letter, signed by subcommittee Chairman Raja Krishnamoorthi of Illinois, called the findings “deeply troubling.”

Premier Biotech CEO Todd Bailey said in an e-mail that the test, made by Hangzhou Biotest Biotech Co. in China, has been sent to universities and the National Cancer Institute for performance assessments, though the results are not yet public. The company, which has about 100 employees in Minnesota, stands behind the accuracy of the test.

“Independent tests have shown strong product performance for a product that is dearly needed,” Bailey’s statement said. “We have always conducted our business with the greatest integrity and we will continue to meet those high standards in the days ahead.”

Dr. Eran Bendavid, an infectious disease physician and researcher with Stanford Health Policy, said the Premier Biotech test has been validated by at least four independent laboratories and has been used in studies that documented surprisingly high rates of antibodies in communities, including in Santa Clara. (That study report was not peer-reviewed.)

Epitope Diagnostics, a San Diego company whose test also didn’t do well in analysis relied on by the House subcommittee, strongly criticized the UCSF and USC researchers. Epitope said the researchers didn’t follow the instructions for the kit, which affected the results. And they say the way samples were heated likely corrupted the results.

Dr. Alex Marson, a UCSF immunology researcher and one of the corresponding authors on the paper, said the team stands by its study and its methodology and plans to submit it for full peer review and publication.

Some researchers in clinical laboratories say an important point is being lost in the accuracy debate: A well-validated test can be useful in diagnosing the health of a community, even though an individual can’t rely on the same test as the sole source for a diagnosis.

That’s because while all tests can provide false results, if the error rates are well-characterized, the prevalence in a given population can be adjusted statistically to find the true result. The same adjustment can’t be made for a specific case.

“If we’re talking at a population level, we don’t need to know exactly which person had it, we just need to know the average number,” Bendavid said.