As COVID-19 vaccines became widely available in the U.S. this year, some people who had put life plans on hold earlier in the pandemic decided not to wait any longer. One of them was Scientific American senior health editor Tanya Lewis, who got married in August. But in the weeks leading up to the wedding day, infections of the novel coronavirus, or SARS-CoV-2, had started to rise again nationwide. So the wedding was held outdoors and limited to fewer than 40 guests, with all the adults confirming they were vaccinated—and Lewis handed out over-the-counter antigen tests for the coronavirus just before the ceremony, then asked guests to take them. These relatively low-cost tests return results within 15 to 30 minutes. “I wanted to make sure that I had one extra layer of protection,” she says.

Her strategy seems to have worked out well. None of the tests—performed by putting a nasal swab and some reagent drops on a test card or test cassette that quickly displays two lines for positive or one for negative—returned a positive result, and no one reported infections in the following days.

But the accuracy of antigen tests varies. These assays correctly identify a SARS-CoV-2 infection in 72 percent of people with symptoms and 58 percent of people without them, according to a review study published in March. And timing matters. The tests detect an average of 78 percent of cases in the first week of symptoms but only 51 percent during the second week, the researchers found. If antigen tests had been Lewis’s only layer of defense (beyond the setting and hosting a mini wedding with all adults vaccinated), this strategy would have held the potential to disrupt her important day with misinterpreted or false test results. How should people use over-the-counter antigen tests? And if they do, what should they be wary of?

At-home antigen tests are expected to become more widely available later this autumn: on Wednesday the Biden administration committed to address shortages with the purchase of hundreds of millions of the assays. The tests are useful as a quick, initial screen for SARS-CoV-2 infection prior to traveling, attending an event, or even going to work or school, particularly if one is experiencing mild or moderate symptoms. Repeated antigen testing at frequent intervals is ideal to increase the chances of spotting an infection if more accurate polymerase chain reaction (PCR) tests are not available. One small study found that antigen testing every three days is 98 percent accurate at detecting SARS-CoV-2 infections, but there is no magic number for how often concerned individuals should take these tests, experts say. People who test positive (or “detected”) should take the result seriously and seek health care. A negative test can ease anxieties, at least for the time being—but people with symptoms should still follow up with a more accurate test.

Antigen tests, which detect pieces of the virus’s proteins, are considered less sensitive to low amounts of virus than the more accurate PCR tests. The latter, for which results can take a day or more to come back from a lab, detect pieces of the virus’s genetic code. If a person with very low viral levels in their nose took both tests at the same time, they would be more likely to receive a positive or detected result on a PCR test than on an antigen test. A person who had been recently infected might slip past an antigen test because the virus would not have had much time to replicate in the nose.

In that regard, antigen tests arguably are more likely than PCR tests to only return a positive result when a person’s case reaches the threshold of infectious—not when they are just infected. For instance, the accuracy of Abbott’s BinaxNOW clinical antigen test increases from about 85 percent to 95 percent among symptomatic people with higher amounts of virus in their nose, the company states. This feature of the rapid tests can hold some public health benefits, says Monica Gandhi, an infectious disease physician at the University of California, San Francisco. Antigen tests are “good for detecting the amount of virus in your nose that’s usually associated with transmission,” she says, “which is actually exactly what you want to know.” Gandhi describes PCR tests as “too sensitive” when it comes to determining infectiousness and says antigen tests are often superior in this area.

Many experts agree that infected people with low levels of virus in their nose (usually described as a low viral load) typically do not spread the virus. So if the goal is simply to make sure that test takers are less likely to infect anyone else—rather than identifying every infected person even if they are relatively unlikely to spread the virus—an antigen test often fits the bill, Gandhi says. Lewis says this was part of her rationale for using them.

But Omai Garner, a clinical microbiologist at the University of California, Los Angeles, cautions against assuming that antigen tests rely on the correct threshold of infectiousness. “I am unaware of a study that ties infectiousness to antigen-test positivity,” he says. Garner adds that antigen tests pick up too few infections in people who are experiencing no symptoms. One type of antigen test detected SARS-CoV-2 infections in only 41 percent of infected people without symptoms, according to a Centers for Disease Control and Prevention study published in January.

If picking up all possible infections is a concern, why bother with antigen tests at all? One answer is that speedier results can at least help quickly flag many or most of the infectious test takers, allowing them to receive care sooner and to isolate before infecting others.

Any infection test can only capture a snapshot in time. With antigen tests, that captured moment is only 15 to 30 minutes prior to results, so they are capable of revealing a sufficient viral load before an infected individual would likely have much time to interact with many others. Because the more sensitive PCR results take longer, any virus present in an infected person’s nose could multiply while they wait—or a person who was uninfected at testing time could catch the virus.

A rapid test an hour or two before going to school, work or some other gathering provides an up-to-date (if imperfect) answer on whether the test taker could spread the coronavirus that day, says Clare Rock, an epidemiologist and infectious disease specialist at the Johns Hopkins University School of Medicine. “You are getting that real-time information,” she says. For multiday situations such as going to work or school throughout the week, such tests would ideally be taken daily (or at least randomly) to potentially detect getting infected in the window between tests and to spot infections in which the viral load increased to the point that it triggered detection that slipped past the initial test. At about $20 per test, however, the costs mount quickly.

Antigen tests also run the risk of a false-positive result, particularly in areas with moderate or low transmission. But false-negative results are more common. Incorrectly swabbing one’s nose—or reading the test results before or after the specified time window—can also yield inaccurate results.

Antigen tests clearly have limitations. In short, if a person with no symptoms tests positive, especially in an area of low transmission, that result is on shaky ground. People in these cases should also follow up with a PCR test, Gandhi says. And the CDC recommends that symptomatic people with a negative antigen test should follow up with a PCR test within 48 hours.

One way to view rapid at-home tests is as an extra precaution—not a license to throw caution to the wind. “They’re not a stand-alone tool to use and say, ‘Okay, I’ve tested myself, and I don’t need to do any of the other prevention pieces,’” Rock says.

This is how Lewis treated the antigen tests at her wedding. Layered measures relieved her of constantly worrying about COVID-19 at her small outdoor gathering. “I felt reasonably safe,” she says. “I mean, as safe as you can feel.”