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.
Utah is testing fewer and fewer people for the coronavirus.
Over the past five days, we’ve tested an average of 2,758 people per day. During our best week, in late April, we tested an average of more than 4,700 people per day.
We have the capacity for more in Utah. So why is there this decline, even as testing in other states increases?
Here are some possibilities:
1. Fewer people are getting the coronavirus. This would be a more convincing explanation if last week hadn’t actually featured Utah’s highest number of cases so far, albeit not by a huge amount. But maybe doctors are doing a better job of identifying who has the virus so we don’t test as many negatives.
2. Fewer people are getting diseases with similar symptoms. This makes sense to me: Flu season has ended, so fewer people are getting the normal flu and thinking it’s COVID-19.
3. Coronavirus fatigue. People are thinking less frequently about the virus than they were in April and are less likely to want to get tested if they have mild symptoms. It is hard for a community to sustain its attention for months on end.
4. More people are back at work. When people are at work, they don’t have time to get tested. Some worry about significant financial downsides of testing positive — they can’t go to work and make money — so why get tested at all? Some might work through early or mild symptoms, or go to work even though they have a sick family member at home.
5. TestUtah worries. The Salt Lake Tribune has published some significant and worrying reporting on the TestUtah situation. For a while, TestUtah’s medical teams were testing asymptomatic people before the state Department of Health told them to knock it off, because they’d get a skewed sample. Even among symptomatic tests, there were concerns about accuracy. Maybe some Utahns just lost trust in the TestUtah program and were less likely to use it.
6. Word of mouth. As more people took the tests, they probably talked to their friends about it. Most of them tested negative, which probably somewhat dissuaded their friends from getting tested. Some probably told their friends about the uncomfortable nature of the nasopharyngeal swab, and their friends weren’t excited about it.
Reasons No. 1 and No. 2 would indicate that we were getting better at figuring out who has the coronavirus, which would mean the decline in tests taken isn’t a problem. But reasons three through six would mean we’re getting less adept at figuring out who has the virus, and that is a problem.
When we miss coronavirus cases, that means we’re going to be less effective in managing the spread of it. Contact tracers can’t do their jobs if they don’t have known cases to trace. Sick people may not quarantine. We won’t know which places the virus has been. High-risk people won’t even know they’ve been exposed, which means they’re going to seek medical attention later than if they’re watching for symptoms.
Now, one possible solution to this is through repeated messaging about the importance of testing and tracing. Maybe articles like this one, public statements from Dr. Angela Dunn or Gov. Gary Herbert, or public service announcements could spread the word. This was the early strategy, in fact; daily news conferences, asking people to download the Healthy Together app, ads and signs all over the state.
But the past month has made one fact clear: If you tell a population of Americans to do something, there’s going to be some percentage of rebels who will now be more inclined to not do that thing than ever before. Regardless if it’s wearing a mask, getting a vaccine, or an Oprah-style new car giveaway for every human being, there are going to be people who somehow make not participating in the common good a proud part of their identity. And the percentage of people who just aren’t paying attention is probably even greater.
If we actually want to catch a higher percentage of our coronavirus cases, those in charge are going to have to widen the goal posts. We’re going to have to significantly lower and/or eliminate the barriers to testing.
First, we should note that antibody testing isn’t what we’re looking for here. Antibodies generally take about 24 days to be created after symptom onset; by then, people probably aren’t contagious. Widespread antibody testing programs like the one run by the University of Utah allow us to know a lot about what the state of the virus was three weeks ago, but won’t be very nimble in allowing us to manage ongoing hot spots.
Instead, we’re looking to more effectively find out which people have gotten the coronavirus recently. Here are a few ways we could do that.
Change the economic incentives
Even when coronavirus tests are free, there are economic barriers that might prevent people from getting tested.
Right now, Congress mandates that all businesses with under 500 employees must give workers up to 80 hours of full paid leave if they must quarantine due to COVID-19. In addition, employees get up to 80 hours of two-thirds pay sick leave if they’re unable to work because of a sick family member or child who has to stay at home because schools are closed. These benefits are paid for through federal payroll tax breaks. This is good.
But Congress didn’t mandate that big businesses do the same. While the rationale was that big businesses can pay for these benefits themselves, the follow-through has been spotty. Without paid sick leave, people are incentivized to work even if they are ill or have a family member who is ill. Those people are unlikely to want to get tested.
In Utah, businesses have also been asked to conduct temperature checks on employees before they begin their shifts. While the state created the PPE Push Pack program to get protective gear in the hands of small businesses, there wasn’t a similar push to get touchless thermometers in the hands of businesses.
Finally, there’s just the fact that people have to take time out of their day to get tested. Even with free tests, it’s a hassle. Some economists have suggested incentivizing people to take the test, even paying them to do so. Freakonomics author Steve Levitt suggested running a coronavirus testing lottery — take the test, be entered to win megabucks!
I’m not sure the lottery idea would fly in Utah, but Utahns sure do love giveaways. Maybe the private sector could get involved in donating freebies to those who take the tests.
Making the test easy to take
Of course, people might still pass on a lottery ticket or giveaway if it meant that they’re going to have to suffer through what a coronavirus test consists of now. Most tests in the U.S. are still administered by essentially sticking a Q-tip up your nose and way inside your head.
I’ve taken one of these tests. It’s not horrendous — it felt like I had just had some aggressively spicy, sinus-clearing hot sauce — but it is uncomfortable. It certainly left my eyes watering for the next 10 to 15 minutes. That might be enough to turn people away, or at least make them think twice.
We use this testing method because for other viruses it’s been shown to give the best results. But as we learn more about this coronavirus, we’re learning that the back of our nasal cavity isn’t always a place of high viral load.
In fact, one recent study showed that saliva is more sensitive for coronavirus detection than the nasopharyngeal swab. The study took the swabs from 44 people with coronavirus, then took regular old spit from them, too.
OK, these figures look confusing at first, but I can explain them. The graph on the left shows how much virus each test picked up. See how the blue bubbles — the saliva bubbles — are higher on the graph than the green ones? That means the saliva tests are detecting more virus than the nasal swabs.
The middle graph shows how the paired samples performed. Most of the lines are trending upward from left to right, meaning that for most people, the spit test picked up more virus than the nasal one did. Finally, the right graph shows that as well, with more matched samples on the high side for saliva — and also shows that the nasopharyngeal tests missed more positives than the saliva test did.
The study also later said it found less “temporal variability” with saliva tests than nasopharyngeal ones. In other words, people who tested positive with saliva continued to test positive. With the nasopharyngeal tests, positive people sometimes tested negative.
Now, this paper has yet to be peer-reviewed, so we do have to be a little bit skeptical. But it is significant evidence that saliva tests are better.
Another interesting study found that self-collected tests were just as good as physician-administered ones. One concern about the TestUtah program was how many of the tests were self-administered, but it appears that didn’t necessarily need to be a worry — their low results were probably for other reasons.
These results, when verified, obviously change the testing game, and many countries have already begun changing their protocols to take advantage. We can, too. It’s easy to imagine a prework spit test — while it might not return results until a day later, detecting viral load still should beat the onset of fever. We could also send Ancestry-style saliva testing kits to people’s homes if they had exposure to an outbreak.
Self-collected saliva testing allows us to be really creative about our approaches.
Fun with sewage sludge
Here’s another creative approach: wastewater testing.
Essentially, there is detectable virus in the fecal matter of people who have COVID-19. Some of that virus is just dead, but sometimes it can be infectious. (Bathroom cleanliness is always a concern, but especially in these times.)
The upshot is that if you have access to sewage lines and wastewater treatment facilities, as city governments do, you can test that sludge for the virus. You may have seen this from a Yale study posted last week that did exactly that in Connecticut; and found a remarkable correlation between the tested viral concentration in sludge and new cases nearly a week later.
Now, it turns out that the prettiness of the graph above is really, really suspect. There’s nearly a perfect correlation (R = 0.994, for those interested) between those two lines, and in science, whenever a correlation is that perfect, it raises red flags. Real life is messy, with all sorts of confounding variables: You could test people’s heights from one week to the next and not get that good of a correlation.
As it turns out, the data had been cooked a bit. It wasn’t probably intentional, but the smoothing algorithm the Yale team used for the lines actually incorporated future data and removed outliers. This is why we have a peer-review process in normal times, to prevent this from happening. In this emergency, amateurs have to catch these sorts of errors, and we caught this one.
The good news is that, even once you correct for the ugly stat errors, there is a real signal in the sewage data. This more accurate analysis of the Yale team’s data found that the correlation between sewage and hospitalizations is still significant with a one-day lead time, and still exists with a two or three-day lead time. Basic sewage surveillance could still warn us of impending outbreaks, and give us a day or two to get resources — like hospital beds — ready in the right places.
It’s an idea we’re trying in Utah, too. Three universities — Utah State University, the University of Utah and Brigham Young University — are working to collect data at wastewater facilities up and down the Wasatch Front. They estimate 30% of the state’s population are covered by these tests, which should give us even a better idea of how the virus’s spread is happening if the concept works here.
Even beyond larger trends, you can start to imagine all sorts of more fanciful — a word I’d never thought I’d use in a conversation about sewage — targeted possibilities for this approach. You might start testing the sewer lines in every neighborhood, and then if that neighborhood spikes, test every street, allowing you to narrow the source of the outbreak. You could then call the people on that street, seeing if anyone is sick.
Heck, you could then add in other testing advances. You could mail saliva test kits to people who live on that street to identify asymptomatic carriers and tell them to stay home, for example. If the sewage of a business tests positive, you could deliver saliva test kits to the door and get samples back right away. We’d need to minimize the procedural test time for this to really work, but that is possible — I got the results of my coronavirus test, conducted on March 11, in about 10 hours.
Right now, Utah has a declining test count, despite an overabundance of coronavirus tests. Whether it’s by making testing more economical, less uncomfortable, or even hidden beneath the surface, we need to figure out how to best test our population in the mid- to long term, so that we can spot and eliminate these outbreaks before they grow.
Doing that would allow us to have the best of both worlds: It would allow us to completely open the economy while minimizing the risk of infection for everyone. Sweet, sweet normality — or at least something like it.
More coronavirus stories by Andy Larsen
Andy Larsen is a 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.