鈥淭he first infections came out in December. And so we really don鈥檛 have any way to know ahead of time how long immunity鈥檚 going to last,鈥� said Deepta Bhattacharya, an associate professor in the Department of Immunobiology at UA鈥檚 College of Medicine.

In the worst case, based on studies of previous coronaviruses, such immunity is 鈥減robably not lasting for too much longer than a year,鈥� he said.

鈥淎 year ain鈥檛 too bad, though, I guess is what I would say,鈥� he added. 鈥淔rom an epidemiological standpoint, if that鈥檚 what this confers, that鈥檚 not terrible.鈥�

The UA announced April 14 that it is developing a test for COVID-19 antibodies and will make 250,000 such tests available to the state, beginning about May 1.

The first of the tests will go to front-line medical personnel and first responders, officials said. UA also plans to offer optional testing to its faculty, staff and students.

Bhattacharya acknowledged a recent study finding that a majority of tests now on the market have accuracy rates that make them effectively useless.

But he said the test the UA is rolling out will be more accurate in determining if there are antibodies present than some of what鈥檚 on the market now.

He said the test is designed to make it 鈥渧ery unlikely鈥� to return a false positive. In fact, he said, it is crafted to err on the side of telling someone who may have some antibodies that they do not have protections.

鈥淲e decided that was probably the lesser of two evils given that we don鈥檛 want to give people a false sense of security,鈥� Bhattacharya said.

He said the belief that COVID-19 antibodies provide some level of immunity is based on what he called the 鈥済arden-variety coronaviruses,鈥� which have been around and for which there are studies. Those include MERS 鈥� Middle East respiratory syndrome 鈥攁nd severe acute respiratory syndrome, or SARS.

The World Health Organization earlier this month said there was no evidence that antibodies prevent reinfection.

鈥淎nd then I think they heard from many irritated scientists such as myself saying, 鈥榃hat do you mean by no evidence?鈥欌€� Bhattacharya said. He said it might have been more accurate for WHO to say that they 鈥渘eed more evidence.鈥�

WHO later backed down, Bhattacharya said, modifying its statement to say that most people will generate antibodies when they get infected, and that those antibodies are 鈥渆xpected to generate some degree of protection.鈥�

鈥淣ow, that鈥檚 vague,鈥� he acknowledged. 鈥淏ut that鈥檚 essentially the data on the ground.鈥�

A lot of the research, Bhattacharya said, is occurring with scientists extracting the plasma from people who have recovered from COVID-19 and giving it to people in intensive-care units who are having trouble controlling the virus.

He said the number of such tests is limited, meaning the sample size may not be enough to draw any major conclusions.

鈥淏ut at least in those small studies it seems like that actually has quite a bit of benefit,鈥� Bhattacharya said.

There have been some nonhuman studies in primates like macaques where researchers infected the animals intentionally and then were unable to reinfect them later, he said.

Bhattacharya and other scientists have done experiments with antibodies and cell cultures. 鈥淎nd what we鈥檙e seeing is that people who have antibodies have at least some degree of neutralizing the virus, meaning preventing it from getting into cells,鈥� he said.

All of those things come with a degree of uncertainty.

鈥淚 would never be comfortable saying, 鈥榃ell, you have an antibody test, go do whatever you want,鈥欌€� he said. But he said everything he has seen so far leads him to believe that 鈥渟ome degree of immunity is conferred.鈥�

Research funded by the Chan Zuckerberg Biohub found that some of the tests now on the market have an accuracy rate of less than 90 percent.

鈥淭hat was highly concerning,鈥� Bhattacharya said. But he said some of it comes down to the type of test being administered.

Some are 鈥渇inger-prick鈥� antibody tests. Their advantage is they involve a small amount of blood and yes-or-no results can come back within an hour.

鈥淔rom a logistical standpoint and getting many people tested, there鈥檚 a lot of upside to having a test like that,鈥� Bhattacharya said. But as the research found, 鈥渁 majority of them are really quite bad.鈥�

It鈥檚 complicated by people analyzing the results, which he said can be like reading a home pregnancy kit.

By contrast, he said, tests run by central labs 鈥� like the kind UA will be doing 鈥� are more accurate.

That requires a blood draw,鈥� Bhattacharya said. 鈥淎nd so there are some logistical issues with that in testing large numbers of people.鈥�

Still, he said, it produces a number that is related to the antibodies, not a color code on a stick. 鈥淵ou鈥檙e not relying on the human eye,鈥� he said.

That, in turn, allows scientists to set 鈥渧ery strict cutoffs鈥� for telling people whether they have a positive antibody test.

鈥淲hat that means, also, is we err on the side of false negatives, meaning someone who may have some low levels of antibodies, our tests might call them negative,鈥� so as not to give people a false sense of security.