A few weekends ago, while trying to convince my 4-year-old son that pants are still required for afternoon hikes, my pocket buzzed with a text message that canceled everything.
“COVID19 Community Tracing Collaborative: We have information about the status of your test. We need to speak with you and will call you back.”
Having written about efforts to trace people exposed to Covid-19, I knew what it meant. Sure enough, a few hours later, a contact tracer called to deliver the bad news: I’d tested positive. She told me I needed to isolate, and she asked for the names and phone numbers of those I’d been in close contact with so that they could be alerted.
Read all of our coronavirus coverage here.
It was worrying of course—but also baffling. I had no symptoms, to my knowledge I hadn’t been near anyone sick, and I’m always careful about mask-wearing, handwashing, and social distancing.
I’d barely left the house in weeks, in fact. As I explained to the tracer, the only interaction I’d had with anyone outside my family bubble in the past week was meeting an old friend, but that was outside, both of us wearing masks and staying at least 6 feet apart. My son is in daycare, and we’ve been in a “bubble” with another family whose child also attends. But his school has introduced all sorts of precautions, with teachers and parents voluntarily testing regularly, a new air filtration system, and countless cleaning and safe-distancing protocols. It all seemed like a shocking reminder of how sneaky the virus really is.
After a few days pacing my hotel room, however, I was less sure. By then, my wife and son had both received several negative results; my friend and the other family had too, along with about two dozen parents and kids at the daycare.
I took a second test three days after the first, and the results came back overnight: negative. At my doctor’s suggestion, I took a third, three days after that at a different location. That too came back all clear. As per the guidelines at the time from the Centers for Disease Control and Prevention, I remained in quarantine for two weeks. But increasingly it felt like maybe something had gone wrong.
I began to wonder what it means to test positive. A Covid test is not a binary thing. There is no single, standard way to detect the virus; different labs set their own thresholds for signaling a positive result. Some experts now think that the sensitivity of a test, and how much virus it detects, should be factored into behavioral guidelines and the public health response.
My initial test was at a drive-through site run by the city of Cambridge, Massachusetts. A swab from my nose went from there to the Broad Institute, a biomedical research center created by Harvard and MIT that converted its genomics lab into a Covid-19 testing facility in March.
The Broad uses a technique known as polymerase chain reaction (PCR) to detect viral genetic material in a sample. A PCR test typically takes a day or more to produce a result, but it is considered the gold standard of Covid testing, because it’s so good at picking up minuscule fragments of the virus. Rapid tests, which detect specific proteins on the surface of the virus, are cheaper and faster, but they are less accurate than PCR and work best when someone has high levels of the virus.
A PCR Covid test involves preparing a sample using chemical reagents to isolate fragments of RNA and enzymes to generate complementary strands of DNA. The lab then amplifies this DNA by adding compatible strands carrying fluorescent markers that break off and activate after binding. This process is repeated over and over. If the virus is present, then the chamber containing the sample should start to glow.
The number of cycles required to trigger a result is crucial. The more virus someone is carrying, the fewer cycles needed; more cycles mean the patient likely has only a low level of infection. Labs generally do not disclose the number of cycles required to get a result—only whether or not there is one. Broad uses 40 cycles as the limit for its tests, as recommended by the CDC. If the sample doesn’t glow by 40 cycles, the result is considered negative. But some other labs use different thresholds.
Michael Mina, an assistant professor of epidemiology at Harvard, believes there should be more focus on the so-called cycle threshold, the number of PCR amplification cycles required to produce a positive result. A high number of cycles suggests a low viral load. This may indicate that someone is at the beginning of an infection—or the end of one. If a person tests positive but is symptom-free, and a subsequent test shows a lower viral load, then they might not need to quarantine for as long. Public health experts increasingly are debating this idea. Some believe that a more nuanced picture could help control the spread of the disease by pinpointing the most infectious people and reducing the burden of quarantine for those who are not infectious. Others warn that the cycle threshold may not be a sufficiently reliable gauge of viral load, since it is influenced by the equipment, the chemistry, and the quality of the sample in each test.
The way a patient’s viral load rises and falls over the course of a few days means that the virus can be missed if someone is tested too early or too late.
Mina says I may have been exposed and infected without realizing it. “And by the time you got the first potentially positive PCR result, you were right on the verge of becoming negative again,” he says.
Gigi Gronvall, a senior scholar at the Johns Hopkins Center for Health Security, has been collecting details on Covid-19 tests to help people judge their reliability. Her data shows significant variation in the minimum amount of virus that a test can nearly always detect—from as few as 10 copies of the virus per milliliter to 10,000 copies per milliliter.
Then there are so-called false positives, when a sample is contaminated because of human error or misread by a machine. The chances of a true false positive might be one in many thousand.
But across millions of tests, that means many people who are told they tested positive may in fact not have had the virus. Matthew Binnicker, a professor at the Mayo Clinic who studies the diagnosis and management of viral diseases, says this means we should not test everyone, but only people who are symptomatic or in high-risk groups.
Here's all the WIRED coverage in one place, from how to keep your children entertained to how this outbreak is affecting the economy.
By Eve Sneider
In a review paper published in November, Binnicker calculates that if a quarter of all school students in the US were tested three times per week, that would result in over 14 million false positives per year. In areas where the virus is rare, he says, it doesn’t make sense to screen lots of healthy people. Otherwise, there is an increased risk of wrongly diagnosing people and forcing them, as well as all of their close contacts, to isolate unnecessarily.
Even where the virus is widespread and broader testing is necessary, Binnicker says it is important to understand that testing does not work on its own.
“Testing is not a foolproof strategy to allow anyone to go about their normal daily lives without also using the other mitigation strategies,” he notes. “If you layer [testing] with masking, and then physical distancing, then you create a strategy where together it significantly reduces the chance of transmission.”
Plenty of others disagree. Mina says we need new approaches to testing that involve screening as many people as possible as often as possible. Although rapid tests are less reliable than PCR ones, he argues they can capture a more accurate picture over time. He suggests using rapid tests widely, with PCR used to confirm positive results. “If you take a step back from thinking of rapid tests as a medical device and think about them as a public health device,” Mina says, “then they are actually more sensitive, because they give you real-time results.”
To get to the bottom of my own test, I spoke to Niall Lennon, a leader of the Broad testing program. He says using the number of cycles needed to trigger a positive result to inform clinical decisions is fraught, partly because it is unclear how accurate a representation of viral load this really is.
Although the Broad does not tell people the number of cycles involved in their test, Lennon agreed to look up mine—35.5. Each of the experts I spoke to said this indicates a low viral load, consistent with the early or, more likely, late stages of an infection. If the viral load were dropping, Lennon says, it’s also slightly more likely that one or both of the follow-up tests might miss the last traces. It doesn’t quite solve the case, but it’s probably as close as I’m going to get.
“Your example is exactly why we need to be thinking about” the number of cycles to trigger a result, Mina says. He says my follow-up negative tests could indicate I was in the final stages of defeating the virus. “I’d say tail end makes most sense. But of course—impossible to know for sure.”