Editor’s note: The Salt Lake Tribune is providing free access to critical stories about the coronavirus. Sign up for our Top Stories newsletter, sent to your inbox every weekday morning. To support journalism like this, please donate or become a subscriber.
Testing has taken on a gigantic role in the story of this pandemic. It’s the subject of genuine confusion and grandstanding claims, and because it provides the basis of our knowledge about the virus, it’s legitimately important. But it’s funny: despite how vital tests are, I don’t think most people know how they work.
Let’s address that. While we’re at it, let’s also look at how long tests take in Utah, why testing numbers have changed in the state (it’s not a conspiracy), and what might be coming down the pipeline.
How do these tests work?
The vast majority of coronavirus tests administered in the United States are called RT-PCR tests. Basically, we want to know whether or not a person has the virus’ RNA in their upper respiratory tract. So we get a sample — that could be saliva, cells from the nasopharyngeal tract (gathered with those long cotton swabs), or just from the nostril. From there, we use chemicals to break open the cells in the sample and filter it so we get just the bits that are the size of RNA.
To figure out what kind of RNA it is, we use a technique called polymerase chain reaction, or PCR. We’ve known how to do PCR for only 35 years, but it’s become so important for so many applications that we did it in my high school chemistry class 12 years ago.
Basically, what PCR does is amplify the DNA in a sample from just a few copies to millions or billions of copies, simply by adding a number of ingredients into a test tube and changing its temperature.
For the coronavirus, we have to turn the RNA we sifted out of our sample into DNA by adding an enzyme called reverse transcriptase. Then, we add DNA polymerase to the soup, along with DNA building blocks, fluorescent probes and coronavirus-specific primers.
Then it’s all about the temperature.
We heat the mixture to 200 degrees to split open the DNA into strands, then cool it to 140 degrees. That temperature allows the primers to potentially bind to the open DNA, but only if they match the genetic code of the coronavirus. The probes attach too.
Then, we heat it up to 161 degrees. This is when the rest of our soup goes to work. The DNA polymerase assembles building blocks floating in the soup to recombine into that beautiful double-helix you picture when you think of DNA. The probes get knocked off the chain and begin to glow.
By repeating this cycle 25-35 times, we can turn one copy of this DNA into billions of copies. And at that point, the solution can fluoresce so much that it can be detected. But remember, it only will light up if the coronavirus RNA was there in the first place — otherwise, the primers won’t attach, and no copying will happen at all.
This is very impressive stuff, but PCR takes 2-4 hours and requires a specialized oven to quickly heat and cool this solution to the needed temperatures. It also requires a lot of specialized material: for example, the DNA polymerase has to be heat-resistant, so we get it from microorganisms that usually hang out near geothermal heat vents in the bottom of the ocean. Of course, we need those coronavirus-specific primers, too. This stuff can and has been subject to shortages in the pandemic.
How long does it take for people to get their test results back?
Of course, just because it takes less than a day to actually complete the test doesn’t mean people get their results that quickly: samples have to be driven to labs, then prepared once they get there. Afterward, test results have to be recorded, sent to a medical provider, and then the patient has to be contacted.
Thanks to the Utah Legislative Auditor General, we now know how long this typically takes in Utah. On average, it takes 1.6 days from the time a sample is collected to when the lab notifies the medical provider.
While 1.6 days was the average, there was significant variation between labs. One lab got the results back within 1.3 days on average, but another major lab took 3.7 days to get the test results back. The labs weren’t identified in the audit, instead they were called lab A, lab B and so forth.
And there was tremendous variation in between individual tests, even at the fast labs. The quickest lab still took seven days on a few tests, and one of the labs that performed well even had a batch that took 21 days to process — 21 days!
The audit found that testing times increased as labs dealt with more tests. It also found that tests outside of Salt Lake County were more likely to take longer, presumably because of the driving time required to get them to a lab.
We don’t know how long it usually takes medical providers to tell people whether they are positive or negative. Sometimes, it seems people are contacted in a matter of minutes or hours. But sometimes, it takes longer than that: the audit notes that one major provider just stopped notifying patients who had tested negative in June and July when testing volumes increased.
Here’s the issue: a coronavirus test that takes more than a few days to come back is nearly worthless from a public health standpoint. People who are tested are relying on the results to decide whether or not to quarantine themselves — you’d hope that anyone taking the test would quarantine, but the reality is that some will do the minimum until they hear the word “positive.”
Some businesses require a positive test before COVID-19 sick leave can begin, and nearly all businesses would wait until a positive test to inform close contacts. When there’s a delay, all of those secondary interventions are delayed too, allowing times for the virus to spread.
The audit included several suggestions, all of which the Utah Department of Health agreed to work on. They want to set a goal for test processing times. They want to collect data on how long each test takes, and publish the results to coronavirus.utah.gov. They want the labs to share best practices and resources on how to get these tests done more quickly. And they want the labs to work together when there’s a pileup at one lab and capacity at another.
Why has testing slowed in Utah?
Clearly, the number of tests performed has declined in Utah over the past month, as has the rate of positive cases.
This has led some to believe Utah is fiddling with the system — intentionally reducing the number of tests taken in order to reduce case counts, making it look like the state is recovering when it isn’t. This would be rather evil.
I don’t see a lot of evidence for that theory, though.
Those who work at testing sites have noticed fewer cars driving through than before. It might be due to the heat, it might be because people are on summer vacations or because they are just less interested in getting tested. Those are all relevant theories covered by my colleagues Erin Alberty and Sean Means this week.
The biggest factor, though? I think fewer people are getting sick.
We now have sewage system monitoring of coronavirus levels throughout the state. And just as cases have gone down, so too have detectable levels of coronavirus in our sewage systems.
The past three weeks, the system has found pretty low levels of coronavirus in the sewers, much more comparable to the relative lull in mid-May.
I suppose it’s possible that the government is cooking the books on the numbers of tests performed, that they instruct random testing site employees to lie about drive-thru traffic and hospital administrators to lie about their tests gathered, and are feeding artificial data to the wastewater monitoring system. But that strikes me as exceptionally unlikely, the kind of conspiracy that would get exposed in 0.2 seconds flat by somebody somewhere with eyeballs and a brain.
In other words — I think this good news is real. That doesn’t mean we’re out of the woods. We’re still experiencing high case numbers, so it isn’t time to go to birthday parties and hug everyone you see again. But we’re trending in a positive direction.
What’s next in the world of testing?
The last time I wrote about testing, I looked at ways to make it easier for everyone involved. And most of those recommendations have been adopted. We’re doing widespread sewage testing. Many test sites now get a saliva sample, and some just swab the shallow part of your nostril, not the part deep inside your skull. And many — but not all — businesses have incentivized their employees to stay home if they’re sick. That’s good!
I think we can still do better, though. In particular, I think we need to put more emphasis on rapid testing, an evolving field that has the potential to cut transmission of the disease to near-zero.
Our current testing setup doesn’t do much to prevent the virus’ spread. People are most contagious in the day or two before and the day or two after they first exhibit symptoms, so testing them once they show symptoms and then giving them the results a couple days later doesn’t mitigate the peak of the spread. Ideally, we’d want to test even healthy people, especially in high-risk or high-importance areas.
The good news: we now have tests that can respond within minutes. These are antigen tests that are described as a “lab on a swab.” Essentially, you get a nasal swab or saliva sample, mix it with a buffer liquid, and then drop it on one end of a test strip. Capillary action — the same force that causes your whole paper towel to get wet even if you only expose one corner to water — drags the solution across the strip.
At the other end of the strip lay coronavirus antibodies. If the virus is present, the antibodies bind to the virus, just as they would inside of your body. And that binding triggers a marker that changes color. Essentially, it works almost like a pregnancy test: if you see the strip light up, you have coronavirus.
Now, this process takes minutes and is really easy to administer: sample, mix it, drop it on a test strip. These tests can cost about $1 or $2 each, where the PCR tests described earlier usually cost about $100.
They’re also way less accurate.
Remember how PCR turns one coronavirus copy into millions or billions, making it easy to spot? The antigen tests don’t do that, so they’re relying on some amount of chance to make sure the virus interacts with the antibodies on the strip. Antigen tests have a success rate of 50% to 90%. PCR, when done right, is about 98%. These rapid tests are going to tell a lot of people who really do have the disease that they don’t, and that causes problems, to be sure.
But that downside is mitigated by how quick and cheap they are. Let’s say you own a business that wants to test people every day, and your test has an 80% success rate. If each test is independent (and early indications say that’s largely the case), the odds that someone with the coronavirus will test positive at least once after two tests is 96%. After three tests, it’s 99.2%. As Bettina Fries, chief of infectious diseases at Stony Brook University, told Science magazine, “Even if the sensitivity [of antigen tests] is not perfect, if you test over and over you will pick up those cases.”
Remember, this test will catch positives before symptoms occur, so even with three days of tests, you’re likely going to get a result more quickly than someone waiting for their cough to get bad and drive to a testing facility, then wait for the PCR test to be processed and the results to be reported back.
There are plenty of situations in which you’d want the higher accuracy of the PCR tests. But it’s also easy to imagine cases in which you’d want the fast, cheap test. For the general public, catching 50% to 90% of coronavirus cases early would decimate the effective transmission rate, or Rt, and drive our case numbers way down. “We need to let go of the notion that all the tests have to be perfect,” Fries said.
Perfect can be the enemy of the good, and good repeated over time is enough to decimate the virus. Rapid tests are good.
Andy Larsen is a Salt Lake Tribune sports reporter who covers the Utah Jazz. During this crisis, he has been assigned to dig into the numbers surrounding the coronavirus. You can reach Andy at firstname.lastname@example.org or on Twitter at @andyblarsen.