Diet is one of the most important factors shaping health, but it’s surprisingly difficult for researchers to figure out what people eat.
Researchers’ usual tools, like food frequency questionnaires or asking people to recall what they ate the day before, are tedious and expensive, said Christopher Gardner, a nutrition scientist at Stanford. Such efforts depend on people being candid and accurate; people forget about 15 to 20 percent of what they ate after a few days.
It’s one of the reasons that nutrition research is so difficult and why the best evidence comes from small metabolic chamber studies, Dr. Gardner said. In such studies, researchers essentially “incarcerate someone,” he added. “They can’t leave. You cook the food for them, you slide it under the door, you know if they didn’t finish it.”
But a new study, published today in the Proceedings of the National Academy of Sciences, points to a treasure trove of nutrition data: sewage. Using wastewater in North Carolina, researchers searched for traces of food DNA that had survived digestion. For less than a penny a person, they identified hundreds of foods that had been consumed — like chocolate, lettuce, turkey, blueberries, black pepper — and could even infer the socioeconomic profile of a community.
During the Covid-19 pandemic, wastewater monitoring became a familiar tool in public health, helping cities track viral surges without testing every resident. But scientists are increasingly using this same infrastructure to measure illicit drug use, antibiotic resistance, stress hormones and, now, diet.
With such data, researchers could assess if nutritional needs are being met in a community and how policy changes, like food stamp cuts, alter eating behavior in near real time. But the technology also raises tricky questions around privacy, surveillance and risks from potential commercialization.
“Every day, eight billion people do roughly three diet experiments,” said Lawrence David, a microbiologist at Duke University and a senior author of the new study. “And none of that data is captured.”
People generally ingest up to a gram of food DNA every day. Most of it is broken down during digestion, but a small fraction is excreted, Dr. David said, leaving behind millions of fragments that researchers can analyze.
In the new study, researchers examined wastewater samples from 2.1 million people in North Carolina, about a fifth of the state’s population. These samples had been collected and concentrated early in the pandemic, and some were preserved in the lab’s freezer, said Rachel Noble, an environmental microbiologist at the University of North Carolina at Chapel Hill and the other senior author of the study.
After thawing this wastewater concentrate, the researchers sequenced the food DNA inside, looking for genetic material from animal mitochondria and plant chloroplasts, the cellular machinery behind photosynthesis. They found that food patterns varied by community: Black-eyed peas tracked with rurality, beer ingredients with affluence, mangos and coconuts with large immigrant populations.
However, sequencing has its limits, since some foods leave stronger DNA signals than others. For example, leafy greens are packed with more chloroplasts than potatoes, said Mengyi Dong, a food scientist at Virginia Tech who led the study, while refined sugar contains little DNA from its original source, sugar cane or sugar beets. It’s also difficult to distinguish between the many industrial afterlives of a single food — like corn syrup, corn chips or corn on the cob.
That’s why some researchers are taking a different approach, looking for chemical signatures of diet like fiber compounds, vitamin traces and other byproducts of digestion and metabolism. For example, instead of assessing beer intake through hops and barley, researchers have analyzed ethyl sulfate, a chemical formed when the body breaks down alcohol and flushed out in urine. While this approach can’t pinpoint specific foods, it can provide helpful observations about specific nutrients and toxins that matter for health, said Devin Bowes, a professor of environmental health sciences at the University of South Carolina.
The trade-off is that it’s quite labor intensive, and it can take a little “detective work,” to determine what chemical is meaningful and measurable, Dr. Bowes said.
No matter which signals researchers track, sewage is still a messy data source. As food DNA or chemicals enter the pipes, some gets degraded or trapped, Dr. Noble said, especially in older sewer systems with terracotta or concrete pipes. Wastewater also carries more than human poop: livestock feces, restaurant scraps and fallen leaves, all of which blur the signal, said Kyle Bibby, an environmental engineer at the University of Notre Dame.
The researchers tried to account for such noise by processing samples “as quickly as possible” to limit degradation, Dr. Noble said. They also excluded DNA from plants and animals that people don’t eat (like grass clippings or wolves), but these nonfood sources accounted for only a quarter of the plant DNA and 1 percent of the animal DNA sequenced.
The researchers also compared the wastewater findings with a small set of individual stool samples, finding that the two sources produced similar DNA patterns. Still, more work is needed to anchor wastewater diet surveillance to a “ground truth,” said Jordan Peccia, an environmental engineer at Yale, similar to how Covid-19 wastewater data was validated against local case counts and hospital admissions. But even then, he added, wastewater would probably be better at detecting patterns over time instead of absolute measures of food intake.
Despite the limits, scientists are hopeful wastewater can help transform diet research.
For example, instead of recruiting millions of people and asking them to complete repeated surveys, researchers could cheaply monitor dietary signals over time, said Dr. Gardner, the nutrition scientist at Stanford. That could reveal how policy changes like tariffs and free school lunch programs affect nutrition. Linked with medical data, those signals could also help researchers connect dietary patterns with community health and chronic disease trajectories.
For now, most wastewater diet surveillance happens at municipal treatment plants, where a single sample might represent thousands or hundreds of thousands of people. But Dr. David’s team is starting to look toward neighborhood-level data, beginning with Durham, N.C.
They avoid sampling areas smaller than about 500 people for privacy reasons. Still, people don’t necessarily think about what they flush away as valuable data and, as wastewater surveillance moves into more intimate areas like diet, may object to their stool being analyzed.
Dr. David’s answer has been to try to fold communities into the process. For the past year, his team has met with community boards and neighborhood leaders in Durham, asking how this data could be useful to them.
When Erin Dooley, a founder of the nonprofit BLK South, first heard about the research, it gave her pause. “We’re living in a time where our data is a gold mine,” Mrs. Dooley said. She worried about how this information could be extracted from her community, sold to advertisers and used to shame residents for their diets.
But she and her husband, Kendall, also saw an opportunity. Their neighborhood, Hayti — one of America’s historic Black Wall Streets and now a food desert — has been trying to raise $6 million to bring in a grocery store. So the Dooleys have collaborated with Dr. David and the Grant Street Community to help determine what her community is eating and what it lacks, turning this data into a case for the grocery store and greater fresh food access.
“It’s helpful information,” Mrs. Dooley said, “but only if it’s to the betterment of the people in the neighborhood.”
