fb-pixel Skip to main content
Globe Magazine

Why is it so hard to track the source of a food poisoning outbreak?

As a cluster of food poisoning cases grew, Boston’s public health system mobilized to discover the source of an unusual toxin.

Enterohemorrhagic E. coli, EHEC for short, comprises strains of bacteria that disrupt the GI tract and cause bleeding.Getty Images

Get the best of the Magazine’s award-winning stories and features right in your Inbox every Sunday. Sign up here.

THE YOUNG MAN in the darkened hospital room was moaning; his eyes were closed as morphine flowed into an IV inserted into his right arm. When I examined his belly, pressing just to the right of his navel, he arched his back.

It was an April afternoon, nine months into my fellowship in gastroenterology at Boston’s Brigham and Women’s Hospital. The 19-year-old, a student at Northeastern University, had been brought to the hospital that morning after passing bloody stools for several days. The shades were drawn so that he could sleep, but it wasn’t dim enough to obscure the worried look on his mother’s face.


“We’ll figure this out,” I promised, taking her hand with my gloved fingers. The young man didn’t seem to hear me. I asked the nurse to send samples of his blood and stool down to the laboratory.

Though frightening for the sufferer, passing blood is not that unusual. In some, its origin may be infectious: vacationers returning from the tropics with stowaway amoebas in the gut, for instance, or churchgoers at a Sunday picnic eating undercooked barbecued chicken infested with Campylobacter jejuni.

What was unusual was that a couple of hours later, my pager rattled again at my waist, summoning me to the emergency room for an eerily similar case: a 21-year-old college student complaining of passing bloody stools over the past day.

It was early evening by then. A fire alarm was going off somewhere, and the lights in the hallway began to pulse. Orderlies and nurses swept past me toward a more pressing emergency. I found the young woman sitting on a stretcher in the hallway looking forlorn.

“Where does it hurt?” I asked.


She winced and pointed across her abdomen. Then I noticed another odd similarity on her medical chart.

“You’re from Northeastern, too?” I asked.

This was a bit strange. Yes, Northeastern is huge, with a sprawling campus, and the chances that I would see two students from the same college on the same day weren’t exactly low. But two Northeastern students within the span of a few hours — each passing a large volume of blood — gave me pause.

Could they have eaten in the same cafeteria? I wondered. Did they sit beside each other in the same class? Had they stopped at the same bar near campus?

As I walked out of the emergency room, I couldn’t shake the feeling that the two cases might be connected.

LIKE MANY PEOPLE, I love the idea of hidden patterns. We want to believe in an underlying structure to the events that befall people — even sickness, even misfortune. We crave stories that neatly explain the puzzles we encounter, and we draw circles to group like things together. A sense of design, instead of mere randomness, gives us the possibility of reason, and when there’s reason, there’s an accompanying hope that there can be resolution, too.

In medicine, that desire for reason drives us to search for signals in the noise. An example: In February 2005, epidemiologists noticed strange blips on the Boston Syndromic Surveillance System, a network routinely used by public health officials to scan for diseases spreading through the city. Over the course of a single weekend, the number of visits to local hospitals for stroke symptoms had apparently doubled. Even more unusual, most of these patients were young men. Perplexed physicians investigated them for common infections, toxins, or a drug that they all might have taken. Upon review of the cases, a strange commonality emerged: The majority of them had come into the hospital just a day after New England Patriots linebacker Tedy Bruschi had been hospitalized for a stroke. Most of the young men did not actually suffer strokes; presumably, seeing Bruschi sick, they had started to misinterpret their own benign symptoms. Constellations appeared where none really existed. Today, it’s known as Tedy Bruschi syndrome.


According to the Centers for Disease Control and Prevention, two cases are enough to constitute an outbreak. But when does it make sense to draw real relationships among a cluster of cases, and when should we dismiss them as an illusion of circumstance and noisy data? Were my two Northeastern student cases related, or was I seeing a bogus correlation, as in the case of the football fans?

Truthfully, real outbreaks are rare in medicine, but when they do occur, they capture the public’s attention and imagination in a way that other phenomena of disease do not. With their mysterious large-scale mechanics and potentially terrifying consequences, outbreaks lend themselves to spooky storytelling: Where did this all start? Why is this happening? Who is at risk?

Hollywood films like Outbreak and Contagion, which portray disease outbreaks in tidy narratives, feed the belief that every epidemic has an identifiable culprit, and that we generally have the tools to track and corral its spread. But it’s simply not that easy. Even now, we don’t know for sure who Patient Zero was in the HIV epidemic. (Gaetan Dugas, long believed to have triggered the outbreak in the United States, was recently, and posthumously, genetically exonerated.) Likewise, the pathways of Zika transmission have proved more labyrinthine than expected, and the origins of the horrifying 2014 Ebola epidemic are controversial (no one has yet found the virus in the suspected host animal, African bats).


The true story of outbreaks is that cases are often scattered and splintered, with only subtle patterns in all of the noise. The search for an explanation can be frustrating for epidemiologists, doctors, and the sickened.

When the laboratory paged me the next day with the young man’s stool sample results, I had my first piece of the puzzzle. “He’s positive for Shiga toxin,” I was told.

I had rarely seen this type of infection detected in the lab. Shiga toxin can wreak havoc on the human body, sometimes leading to sepsis and organ failure. It is generally produced by one of two types of bacteria that invade the gut from contaminated food: Shigella and a specific strain of E. coli called O157:H7. The latter was at the root of the 1993 Jack in the Box epidemic that killed four children and sickened hundreds throughout the Pacific Northwest and California. Once in the body, the bacterial toxin causes a bloody diarrhea and occasionally shuts down the kidneys, especially in people who are treated with antibiotics.


E. coli O157:H7 was the bacteria that caused the 1993 Jack in the Box epidemic that killed four children and sickened hundreds. AP

To be certain of the cause of my patient’s Shiga toxin, I needed to know what kind of bacteria was growing in his stool. I had my answer when I contacted the hospital lab that evening: It was E. coli O157:H7. The bacteria belongs to a larger group of enterohemorrhagic E. coli, which disrupt the GI tract and cause bleeding.

Later that night, I learned my young female patient had O157:H7, too.

I texted colleagues at other hospitals around the Boston area to tell them what I was seeing.

“We have a confirmed E. coli O157:H7 here also,” one replied. Another told me he had diagnosed two more cases.

We hardly see a single case in a year, and now we had five cases of toxin-producing E. coli in three days.

I didn’t know how many people O157:H7 had infected in Boston nor where the bug had come from, but I felt compelled to learn as much as I could. Questions flooded my mind. Was this a small outbreak or something larger? Where had it begun, and how might we contain it? How would health officials go about tracking it?

On the first Tuesday in May, a few weeks after I had first examined my two Northeastern patients, I finished my morning hours at the clinic at Brigham and Women’s — where we focus on protecting the health of individuals — and crossed the lawn to the Harvard School of Public Health, where faculty train students to trace the effects of disease on our larger community. I hoped William Hanage, who researches bacterial epidemiology there, could help answer my first question — whether the five patients we had seen were linked.

The problem a lot of epidemiologists struggle with, Hanage told me as we chatted in his sleek, modern office on the school’s fifth floor, is how relaxed their assumptions should be. When are two things associated and when are they just noise? These days, if we want to know precisely whether two infections in two separate patients are related, we sequence the DNA of the organisms responsible for their infection, Hanage said. If the DNA is a close match, it’s more likely they picked up the infection from the same source, akin to incriminating a burglar who left the same set of fingerprints at two separate crime scenes.

But food outbreaks in particular, he said, are tricky to define.

“In some ways, they’re harder to track because they’re not usually communicable,” he said.

By the time a restaurant is identified that links the stricken patients, the bacteria has often disappeared from kitchens, refrigerators, and the food in question. In the summer of 2013, for example, dozens of salmonella cases sprang up in Boston. When the city conducted detailed interviews with those who had fallen ill, many said they had eaten either a chickpea fritter or an egg and eggplant sandwich at Clover food trucks. The city’s health inspector and the health commission swept the company’s food trucks and its restaurant plants. They looked at the tahini, the hummus, the tomatoes, the cucumbers, and other ingredients. They tested employees for infections to see if they were salmonella carriers. They documented refrigerator temperatures, evaluated kitchen protocols, and checked food samples for the culprit bacteria. But the results of every test came back negative.

In food outbreaks, “there are lots of dead ends,” Hanage told me. “Many are never solved.”

HOUSED IN AN UNASSUMING brick building on Massachusetts Avenue in Dorchester, the Boston Public Health Commission is a kind of nerve center that detects changes in our population’s health and helps track outbreaks. Though names are not supplied, the commission receives daily, real-time information about every medical visit to the city’s emergency rooms and some acute-care facilities, down to the patient’s chief complaint, ZIP code, age, race, ethnicity, and body temperature.

That information is uploaded to the Boston Syndromic Surveillance System, where it can help public officials discover abnormal patterns. Health officials have used the data to track, for example, a spate of carbon monoxide poisonings to a single housing complex, a rise in influenza over a nasty winter, and the status of victims transported to hospitals after the Boston Marathon bombing.

Anita Barry was  director of the commission’s Infectious Disease Bureau at the time, and I made an appointment to talk with her a few weeks after I met with Hanage. If a communicable disease had spread through Boston, Barry would know about it.

“We knew that something strange was going on,” Barry said, after inviting me to take a seat in her office.  “A hospital reported a positive Shiga toxin E. coli to us on April 5. Later on that same day, we had another one, and the next day we heard about two more.”

The number of local cases was probably much higher than that. The Centers for Disease Control and Prevention has said that for every diagnosed case of E. coli, 26 go unreported. Plenty of people stricken with E. coli or other infections see the doctor and don’t provide a stool sample or get a positive test result; many never even go to the doctor. Using the CDC’s figures, there could have been 130 more cases within the Boston area, potentially from a single source.

In the days after the suspect cases began popping up on Barry’s radar, a fleet of public health nurses sprang into action. Their first objective was to call each of the patients whose stool samples had tested positive, including the two students I had seen. The nurses had an extensive list of questions that targeted food sources known to harbor E. coli.

“Did you handle raw beef?” they asked. “Did you eat steak at home?” They ran through a list of food and drink, from bison, venison, elk, and boar to artisanal cheese, cider, sprouts, and soy-nut butter. They asked where patients had ordered food and where they had dined out.

Once finished with their interviews, the team compared answers. They circled common food sources and highlighted overlapping locations. They came up with a number of hypotheses from the primary data, Barry said.

“A lot of the cases were geographically clustered around Northeastern,” she noted. “I wondered about Boston water around that area. Could there be a contaminated water source? A sewer problem? But none of that panned out. Then we found that a lot of people had eaten avocados.” The avocado possibility didn’t lead anywhere either, she explained, because people had gotten them at various places, from restaurants to food markets.

No Chicken & Rice Guys facilities tested positive for E. coli O157:H7. Joanne Rathe/Globe Staff/File/Globe Staff

About two weeks after I first met the young woman from Northeastern in the ER, I called to check on her. She was doing much better. In addition to the nurses’ list of questions, health officials had asked her to try to remember what she had eaten in the week before she had gotten sick, and she had prepared a food diary for them.

Dietary recalls are hard to do; most people can’t remember what they ate for dinner the night before. But the illness had spurred her memory, and she shared her list with me. At Kigo Kitchen, she’d had chicken and noodles; at Whole Foods, a California roll; at Caffebene, a grilled-cheese waffle sandwich with spinach; at Rebecca’s Cafe, beef chili; at Bon Me, pork with soba noodles; at Chicken & Rice Guys, chicken and rice; at Oakleaf, a Boston Creme Pie pop tart; at Wollaston’s, a grilled chicken, bacon, spinach, and guacamole sandwich. She had eaten food she bought at Trader Joe’s and Whole Foods, too.

Armed with this information, I signed on to the CDC database and searched for cases of O157:H7 tied to Massachusetts. I was preparing a report for my department looking at whether the patients we had identified were linked to one another or to other cases in the state. The CDC’s data contained no mention of any of the restaurants the student had visited. But I did find a multistate outbreak of O157:H7 that occurred along the West Coast between January and April of this year. There was only one case reported in Massachusetts.

Most of the sickened people were as young as my patients. Of the 32 cases, nine had developed into kidney failure. But the CDC had tied the cases to a brand of soy-nut butter, which the female student hadn’t eaten.

I needed a way to search for infections that were happening right now.

I checked Twitter, PulseNet (a national network run by the CDC that looks for food outbreaks using DNA signatures), and iwaspoisoned.com, a kind of crowdsourcing successor to negative Yelp reviews that specifically collects reports from people who become sick after eating out. I searched for some of the restaurants my female patient had mentioned. Nothing turned up.

I had spoken with the male student from Northeastern before he was discharged from the hospital. He had taken longer to get better, spending almost a week on the wards, where we had given him intravenous fluids and treated his pain. Eventually, he was able to eat again. We had asked him to recall what he might have eaten in the days before he got sick. Like the female student, he had ordered from Chicken & Rice Guys.

As the city’s health department conducted interviews on the growing number of E. coli cases, the chain’s name popped up repeatedly. I knew the place and was a fan of its food, having eaten from one of the Chicken & Rice Guys’ bright yellow food trucks, scooping up flavored rice and chicken in a pita while I sat by the Copley Square fountain.

On April 11, about a week after we diagnosed the flurry of E. coli cases, Chicken & Rice Guys posted on its Facebook page that some customers had not been feeling well after visiting some of the food trucks. The company had decided to voluntarily shut down operations and meet with the city’s health inspector.

“If you think you may have become ill recently after eating at a CNR location,” Chicken & Rice Guys wrote on April 12, “we absolutely want and need to hear from you.”

In my quest to learn everything I could about the outbreak, I met with William “Buddy” Christopher, the commissioner of Boston inspectional services, in early May. Responsible for checking up on the restaurant itself, his office scoured the site for health violations, looking at everything from refrigeration temperatures to kitchen protocols.

“The whole thing was an anomaly because of the multiple sites,” Christopher told me. Chicken & Rice Guys had three locations in the city in addition to three food trucks that operated in Boston. Officials had to figure out how the food ended up at each location.

“Where was the rice coming from?” Christopher pondered. “Where was the chicken coming from?”

Two of the locations — one in Allston and a food truck that drove around near Northeastern — had repeatedly shown up in the health commission’s investigation. A kitchen in Somerville where Chicken & Rice Guys prepared its food appeared to be central to most of the cases, Christopher told me.

By April 14, more than a week after the first case was discovered, a total of 16 people who had tested positive for E. coli had been reported to health officials, with 10 of the patients hospitalized.

The fact that 12 of the sickened patients ate at the same chain, that the ingredients could be tracked back to the Somerville kitchen, and that all 12 grew out O157:H7 bacteria in culture suggested that Chicken & Rice Guys played a role in its dispersal. But six months later, despite the investigation by the Public Health Commission and the Inspectional Services Department, the culprit bacteria hasn’t been detected at Chicken & Rice Guys.

The restaurants reopened on April 22; the food trucks were out of service for longer — two of them are operating again in Boston. Chicken & Rice Guys identified two products that were common to all of the dishes the sick patients consumed and has discontinued its relationship with the supplier for one of those products. None of the company’s employees, food, or facilities tested positive for E. coli. However, Chicken & Rice  Guys is undergoing more audits by an external food safety consultant it hired to review practices at all of its locations. The company also now provides more advanced employee certification through a program administered by the National Restaurant Association, and it uses monitoring technologies, including cloud-based temperature sensors, and disinfectant light technology. It no longer uses the commercial kitchen in Somerville.

The Chipotle chain experienced repeated outbreaks of food poisoning in 2015.AP

I VISITED CHICKEN & RICE GUYS in Allston in July, a few months after the restaurants reopened. It’s a small spot with a few tables, a block from a row of Korean eateries. I ordered a plate of chicken and an orange Fanta and sat with my notebook as diners came in and out.

Lightning does not typically strike the same restaurant twice — the Chipotle chain’s repeated outbreaks in 2015 are a recent exception. Restaurants implicated in cases of food poisoning often take financial hits and their reputation suffers, even when investigations are inconclusive. However, the right moves after such events can lead to positive transformation: Following the E. coli outbreak at Jack in the Box in the 1990s, for example, the chain went on to set new standards for food safety, encouraging other companies to follow suit.

Food outbreaks are unlikely ever to be completely preventable; indeed, they are our most pointed reminders of communal risk. Even when they’re over, we continue to share exposures — via food, roadways, air, sunlight, water, travel, public space — which means we will continue to share diseases. We like to think it is our individual behavior that dooms us or saves us from illness, but knowledge of and investment in the health of our communities might prove an even more effective protection.

In the months since the outbreak, I haven’t encountered any more cases of O157:H7 . But I keep my eyes open around the patients I see for clues to our mutual risks, like footprints left behind for us to follow.

Sushrut Jangi is a gastroenterology fellow and instructor in medicine at Brigham and Women’s Hospital. Send comments to magazine@globe.com. Follow us on Twitter @BostonGlobeMag.