The sewer is the last stop for Bozeman, Montana’s waste, but lately, it’s the first one for Blake Wiedenheft’s work. An associate professor of microbiology and immunology at Montana State University, Wiedenheft has joined other virologists, epidemiologists and immunologists as a member of the university’s COVID-19 task force.
Back in March, a colleague mentioned testing wastewater for evidence of COVID-19 in human waste. The next day, Wiedenheft drove down to Bozeman’s wastewater treatment plant to see if he could grab a sample. Given how few cases there were in the area at the time, and that 6 million gallons of water flow through the plant daily, he wasn’t sure if the virus would be detectable. But Wiedenheft immediately found evidence of it — and it kept appearing in the four samples he analyzed over the next 10 days. Wiedenheft worked with local officials to continue regular testing as Bozeman became one of the first cities in the world to look to sewage for answers. Now, cities across the Western United States are sampling wastewater for evidence of SARS-CoV-2 as a potential “early warning” system for outbreaks. Scientists estimate that up to 45% of people infected with the virus show no symptoms. Given that asymptomatic people are less likely to get tested, many cases may go undetected. With many areas experiencing substantial delays in swab test results, daily wastewater testing can give scientists an idea of community infection nearly immediately, Wiedenheft said.
The autosampler looks like two mini fridges stacked vertically. The top compartment houses a tube and pump, which hoovers up half-cup samples and deposits them into a plastic jug in the cool bottom compartment. Every 24 hours, a sample from this jug is taken to Wiedenheft’s lab for testing. Wiedenheft’s lab employs the same methods used to analyze COVID-19 nose swab tests: Researchers take RNA — whether floating in a liter of wastewater or smeared on a swab — and perform a procedure that amplifies the genetic signature of the SARS-CoV-2 virus. That allows researchers to identify the virus.
The challenge of analyzing wastewater, as opposed to a swab, is that RNA samples are such a tiny part of the overall jug of wastewater. “Imagine you’ve got a jar full of jelly beans, and you’re dipping into the jar to see if you can find a red jelly bean,” said Wiedenheft. “If your jar is small, it’s more likely you’re going to scoop up that red jelly bean, but if your jar is big, it’s less likely, and you’re going to have to collect more of the sample to find it.” The virus is like the red jelly bean; to get an accurate glimpse into how much virus is in a city’s water, scientists have to make sure the samples they collect are representative of its overall water flow. That means taking more frequent samples at peak times — in the mornings, when people are starting their routines, or in the evenings, when they are doing chores — as well as less frequent samples when the water flow is lower, as it is very early in the morning. Bozeman’s autosampler collects around three dozen samples daily.
The upshot of Wiedenheft’s lab analyses — whether samples are virus positive or negative — are posted on the county’s public health website. In the 26 samples taken between June and mid-August, the coronavirus was detected every time.
French said they are now experimenting with sampling from specific areas of town, like the hospital and Montana State’s campus. Engineers have identified their main sewer flow lines and placed mobile samplers just beneath manhole covers, where they collect samples one day a week. As the school year kicks off, MSU’s data could track viral spread on campus, while the hospital data could help scientists better understand how their results map on to community spread. “We know how much viral protein (Wiedenheft) is detecting in his lab, but how many individuals is that reflective of?” asked French.
Researchers have yet to arrive at a definitive answer. Converting wastewater data into an estimate of positive cases requires a key metric scientists are still learning about: how much virus a sick person sheds. That depends on the stage of illness the person is in, as well as how severe the illness is. Because those variables are hard to nail down, Wiedenheft has intentionally avoided reporting such estimates. “We don’t feel confident enough to make that translation,” he said.
But in some cities, officials have released estimates using wastewater data to indicate how many community members are infected. For instance, analyses by Biobot, a wastewater testing startup, estimated that levels of SARS-CoV-2 found in Moscow, Idaho’s sewage corresponded to 1,800 cases. At the time of the estimate, Latah County, where Moscow is located, had only 46 known cases. (Biobot declined to be interviewed, and its hired communications firm did not answer questions about how the company calculates case estimates. A recent study published by Biobot’s co-founders and colleagues reported that assumptions about individuals’ viral load can massively affect these estimates. According to their calculations, assuming infected people have a low viral load leads to an estimate that 5% of the population is infected, whereas assuming a high viral load will lower that estimate to 0.1%.)
“It was somewhat reassuring to have another backstop indicator of what was happening.”
So far, many local officials have looked to wastewater analyses more as a way to corroborate their knowledge about community viral spread than as a way to estimate case numbers. “We won’t know how many people in the community have COVID-19 from the methodology, but this data will tell us if trends are going up or down,” says Nicole Rowan, clean water program manager at the Colorado Department of Public Health and Environment, which has launched a state-wide wastewater testing effort.
Matt Kelley, health officer at the Gallatin City-County Health Department, said that when his county went through a period of few cases in May, the sewage data gave additional confirmation that spread had slowed. “It was somewhat reassuring to have another backstop indicator of what was happening,” he said — and when positive COVID-19 swab tests rose again, so did levels of SARS-CoV-2 in the water. Kelley said that once cases fall again, it will be helpful to track wastewater as an indicator of undetected community spread. “If we’re not seeing tests in the traditional medical testing, and also not seeing them in the wastewater, that’s a validating factor for us.”
As more areas launch wastewater tracking programs, Wiedenheft said there’s still one thing needed: a central repository for this data, which could provide a bigger-picture view of viral spread. Over the last few months, Montana has developed several monitoring sites, and some states, like California, Colorado and Wyoming, created their own networks. It’s difficult just to compile a list of cities conducting testing, since some key data is private; Biobot said it is working with 400 facilities, but declined to provide additional information, including how many cities that corresponds to. Wiedenheft points to Johns Hopkins’ popular coronavirus tracker as an example of a well-organized health surveillance tool. “It would be nice to have a wastewater surveillance website that does the same thing, where you could look at a geographic map and look at what’s happening,” he said.
In mid-August, the U.S. Centers for Disease Control and Department of Health and Human Services announced plans for a National Wastewater Surveillance System, a portal to which health departments can submit their wastewater data. After the federal government’s changes to a portal tracking hospital data lead to widespread reporting delays and data inconsistencies, it’s unclear how the database will fare — and whether this new initiative provides funding to areas conducting testing. Wiedenheft’s funding goes through August, and he’s unsure where the money will come from after that. “If wastewater sampling is important — and I think it is, since there’s plenty of evidence to suggest it’s really the only real-time indicator in the community — then it seems important to have funding to support this ongoing effort,” he said. “That’s imperative to be able to manage a pandemic.”
Note: The section of this story on assumptions about how viral loads impact estimates of community spread has been updated to reflect that assuming a low viral load would mean 5% of the population has SARS-CoV-2 and that a high viral load could mean as little as 0.1% are infected.
Jane C. Hu is a contributing editor for High Country News and an independent journalist who writes about science, technology and the outdoors. She lives in Seattle. Email her at jane.hu@hcn.org or submit a letter to the editor. Follow @jane_c_hu
This article appeared in the print edition of the magazine with the headline Raw data.
Republish our articles for free, online or in print, under a Creative Commons license.
