The following article originally appeared in FierceBiotech and was written by Rae Ellen Bichell, Kaiser Health News

Carol Wilusz’s mornings now often start at 4 a.m., scanning the contents of undergraduates’ feces. Specifically, scanning the data on how much coronavirus they flushed into the shadows, destined to be extracted from 17 manholes connected to dorm buildings on Colorado State University’s (CSU’s) Fort Collins campus.

“There are quite extensive numbers of poop jokes,” said Wilusz, a CSU molecular biologist.

Emerging research suggests infected people start shedding the coronavirus in their poop early in their infection, and possibly days before they begin shedding it from their mouths and noses. “It means that we can catch them before they’re actually spreading the infection,” she said.

In normal times, Wilusz studies stem cells and muscular dystrophy. Now, her team is on the front lines of defense against the massive COVID-19 outbreaks that, for a campus with more than 23,000 undergraduates alone, always seem to be lurking around the corner. The sewage review is part of a multipronged attack that includes the usual weapon of contact tracing plus a specialized “paired pooling” form of testing saliva samples. So far, the school has had about 500 cases since the semester started, about half that of the only somewhat bigger University of Colorado-Boulder.

Amid fluctuating scientific recommendations and a virus that still holds uncertainties, colleges across the country are taking a choose-your-own-adventure approach to COVID-19. For those holding in-person classes, the adventure includes an extra puzzle: how to concentrate a lot of people into one place without an outbreak tearing through the student body and spilling into the community, all without safety precautions that would break the bank. Testing is at the core of those plans.

“A lot of these institutions started testing just symptomatic students. And that is really not good, to put it bluntly, because as we’ve seen over the past couple of months, students tend to be asymptomatic,” said Chris Marsicano, an assistant professor at Davidson College in North Carolina who is leading an initiative tracking how universities are responding to the pandemic. “The institutions that have been the most successful are ones that are testing every student at least once a week.”

According to data collected in mid-September, only about 6% of large universities with in-person classes are routinely testing all students, according to an NPR analysis of his group’s data. The University of Illinois at Urbana-Champaign has been leading the pack, testing about 10,000 students each day using a streamlined spit-testing method. But it’s pricey. Despite driving down the cost of an individual test to about $10, Paul Hergenrother, a chemist leading the effort, said the school is still spending about $1 million a week.

At Colorado State University, Lori Lynn, co-chair of the school’s pandemic response team, said initially the school was paying $93 a pop to test students using the usual nose swab method.

“We quickly spent several million dollars on testing,” said Lynn, who added that cost is just one limiting factor. “We can’t test everybody in the community, you know, weekly or twice a week.”

Instead, Mark Zabel, a CSU molecular biologist and immunologist who typically studies neurodegenerative diseases, said his group recently figured out how to screen saliva for less than $20 a person. It involves pooling drool samples in a strategic way reminiscent of the children’s game Battleship.

Traditionally, pooling involves mixing samples from multiple people and testing them all in one go, to save time and materials. If the pool comes back negative for the virus, everyone in the pool can be considered negative. If it’s positive, samples from each person in that pool must be retested. If there are high rates of infection, that means a lot of retesting.

Instead of pooling samples willy-nilly, Zabel and his colleagues are doing something he calls paired pooling: They start with an eight-by-eight grid of saliva from 64 people, arrayed almost like a Battleship board. Each person’s spit sample gets divided up and analyzed in two pools, one pool for the row it sits in and one for the column it sits in, for a grand total of 16 pools per grid.

If the test containing samples in Row A and the test containing samples from Column One appear positive, that would indicate that the person whose spit is in the A-1 slot is a positive case.

“So, it’s super easy if we’ve got one positive among 64,” said Zabel. In that case, they’ve screened 64 people with just 16 tests. No retesting necessary.

Limited retesting is needed only if at least four pools come back positive.

They’re also using a different kind of PCR test than usual, in an effort to avoid competing for limited reagents, whose shortages have hampered labs nationwide.

Zabel said it takes between eight and 24 hours for results. However, some drawbacks exist. If retesting is necessary, total turnaround time could extend to three days. And if the outbreak were to grow beyond a certain point, in which at least 5% of people tested are positive, the process would become more cumbersome because they’d have to add more layers of testing.

It’s a shifting target and the university is continually reevaluating its testing strategy, but Zabel expects his lab could test up to 3,000 people a day, which would enable testing the entire student body every other week.

According to other researchers, that might not be enough.

Daniel Larremore and others writing in The New England Journal of Medicine said it’s time to ditch any approach that relies on highly accurate tests, and instead embrace antigen tests, which are cheap and quick—albeit less accurate—and can be administered frequently.

“You have the science of testing, which says if you’re testing everybody twice a week, you should basically have zero cases,” said Larremore, a computational biologist at the University of Colorado-Boulder, referring to modeling studies from his lab and others.

But then, there’s reality. And no testing system alone will solve the problem, Larremore said, “because there are humans involved.”

Wilusz, the CSU professor, knows how difficult this is. Often people continue shedding virus in their poop long after they’ve recovered, so over the course of the semester more and more dorms have started to yield virus-positive sewage.

“And then there’s also, we can’t stop students pooping in the wrong dorm. So one could poop in this dorm one day and then next door on the other day,” she said, making it hard to know which dorm to screen with saliva tests.

Also, only about 5,000 of the school’s 28,000 enrolled students live in dorms, though Wilusz said those close quarters create a high risk for spreading the disease because “they’re essentially like nursing homes for young people.”

She wonders how long students will remain game to spit into tubes before they get bored. Michigan State University researchers experimenting with paired pooling and saliva have made a habit of double-checking that students have submitted spit instead of something else. (Chewing tobacco and something the color of blue Gatorade have sullied a few CSU samples so far.)

But the shifting, multifaceted approach does seem to be helping at Colorado State. Back in September, Wilusz noticed a concerning spike in the amount of virus in the sewage connected to two dorms that collectively housed about 900 students. The university put the dorms on lockdown and tested everyone inside, revealing nine positive cases that hadn’t been found using other methods.

Now, with pooled-spit screening, Zabel said the team has been able to identify positives without locking down entire dorms, and can then use subsiding levels in sewage to confirm no infections slipped through the cracks.

The goal is to make it to Thanksgiving, when students return home. But then comes 2021. “We’ll see if we can keep on top of it,” Zabel said, knocking on his desk for luck.