Eating Dangerously Read online

  Colorado became a “Food Center of Excellence” in 2012, one of five states researching the best ways to handle outbreak investigations. It’s another federal designation that boosts the state’s investigative power, although it doesn’t come with funds to hire outbreak investigators. The likelihood that budgets for food safety investigation will expand is slim. Instead, cuts threaten to knock the level of safety down a notch. The head of the CDC, Dr. Tom Frieden, warned in 2013 that budget cuts are likely to result in lower funding for state teams on guard for patterns in foodborne illnesses, the first signs of outbreak. He predicted a loss of $300 million, not just for detective work in foodborne illness, but in preventing foreign diseases from entering and spreading through this country.2

  The 2011 cantaloupe investigation was a model effort that might not have been solved as quickly in another state without Colorado’s expertise. Unlike Colorado, some states do not have the capability to perform genetic tests that would match a Listeria strain found in contaminated food to an infection in a patient’s body, for example. The massive, near-round-the-clock investigation began with questionnaires. State epidemiologists directed county health authorities to interview patients and their families, all using the same fifteen-page questionnaire Colorado uses when an outbreak is suspected. Each state has a preferred questionnaire to find out what food poisoning victims ate and where they bought it; the forms vary only slightly from one another. They are detailed, in depth, and memory jarring.

  Did you eat at a cafeteria? Did you buy hummus? In the last four weeks, did you eat watermelon more than five times a week? The interviews are tedious and time-consuming for victims and their families. Sometimes, people get annoyed: How are they supposed to remember what they ate five weeks ago when they don’t remember what they had for breakfast yesterday?

  In the cantaloupe outbreak, the majority of victims were elderly, as they often are in foodborne illness outbreaks. The elderly, the young, and those with compromised immune systems are more susceptible to severe food poisoning. “The average age is in the 80s and they are quite ill,” Cronquist said. “Their family members are at their bedside, and we are asking them to remember food that they ate a month ago. They are actually very difficult interviews.”

  At first, epidemiologists suspected the usual Listeria culprits—deli meats, hot dogs, dairy products. Several of the patients had eaten those foods. Scientists pulled some of them—along with whole and uncut cantaloupes—from refrigerators and brought them to the state food lab in Denver. Cantaloupe wasn’t on the radar immediately; the 2011 outbreak was the first deadly one that would link Listeria with cantaloupe. Melon has a long, sordid history of causing sickness, but before 2011, cantaloupe was nearly always linked to Salmonella.

  As one set of health authorities was questioning patients, another group was collecting blood samples and isolating the Listeria bacteria. Those samples, called “isolates,” went to the state lab in a brick building on the east side of Denver. There, state microbiologist Hugh Maguire’s team deconstructed the DNA profiles in the bacteria to see if the patients had the same strain.

  The DNA profiles are uploaded to a national database kept by the Centers for Disease Control and Prevention. The database—called PulseNet—is a key investigative tool that federal officials say has significantly cut the amount of time it takes to find the source of an outbreak. In rapid time, federal epidemiologists at the CDC in Atlanta can see whether the same strain of bacteria sickening people in Colorado is wreaking havoc elsewhere.

  In the cantaloupe outbreak of 2011, more than one strain was identified, and those strains were not just sickening and killing people in Colorado, but Nebraska, Texas, and New Mexico, among many others. The outbreak would eventually spread to twenty-eight states.3

  Within five days of suspecting an outbreak, the Colorado state lab had determined that two patients had matching strains, and another two patients had a different matching strain. Already, there were two separate strains identified within the outbreak. With hundreds of strains of Listeria, two or more can sit side by side on the same food. The fact that scientists were now tracking two did not necessarily mean there were two sources. “It just means we have a deeper mystery,” Cronquist said. In the end, five different strains were linked to the same cantaloupe farm.

  Epidemiologists in the counties where the patients lived set out for another round of interviews. One epidemiologist interviewed two patients with the same strain of Listeria to listen better for common patterns; another interviewed the other matched pair.

  That same day, the state health department sent out a Listeria alert by fax and email to doctors, emergency rooms, hospitals, and labs. And the public received the first notice of an outbreak, a warning for those susceptible to the illness—the elderly, the pregnant, and those with compromised immune systems—to avoid deli meats and unpasteurized cheese. Cantaloupe was not mentioned. The health emergency was dire enough that state health officials needed to warn the public, but they didn’t know yet what to warn them about. They were guessing. “We were telling people not to eat deli meats. We had no concrete message to give,” Cronquist said.

  The public health warning with a vaguely mentioned culprit—maybe cheese, maybe meat—was a rare move. Colorado issues, on average, only one public warning per year regarding an outbreak. The state health department puts out an average of thirty-three notices per year warning the public of food that has been recalled, from peanut butter to crackers to ground beef. The recalls most often come before there has ever been a report of anyone ill. They are typically the result of routine testing at production plants, and reasons vary from pathogen contamination to mislabeling to bits of metal or plastic that inadvertently fell into the food.4

  It would take one more week for Colorado officials to tell the public it was cantaloupe that was making people deathly sick. Epidemiologists at the state lab were beginning to zero in on melon. Cantaloupe is on the patient questionnaire, but not emphasized. At first, the fact that several patients had eaten cantaloupe didn’t stand out. After all, the melon hadn’t been linked to a Listeria outbreak before, and it was August, the peak of the melon season in Colorado, so the fact that people were eating it wasn’t unusual.

  Cronquist turned to the CDC database created by the FoodNet survey, the nation’s most accurate gauge on what the general population is eating. It’s where epidemiologists go to check the baseline for healthy people. How likely is it, for example, for an American to eat sushi on any given day? Certainly a lower percentage than eating at McDonald’s, but if half of the people who got sick in one outbreak had eaten tuna sushi, a Japanese restaurant is likely where investigators would look first. Or, if epidemiologists were tracking a summer outbreak on the West Coast among people who reported eating salsa, they might query the database to find out how common it is that West Coasters in general eat tomatoes in July. How likely are American women to eat spinach on a weekly basis? Or peanut butter on toast? The database is maintained by the CDC’s enteric disease epidemiology branch, which surveys people in different regions of the country aiming to capture America’s diverse eating habits. When national investigators are tracking an outbreak, the FoodNet team can answer a query within a matter of hours.

  As useful as the database can be, however, it is badly out of date; the last eating surveillance took place in the early 2000s. Imports of seafood and other suspect foods have risen sharply since then, while food processing techniques also have changed dramatically. Eating habits evolve quickly in this country. Grocery stores didn’t routinely offer Mediterranean olive bars and sushi trays back in 2002, for example. The survey is expensive, and there hasn’t been money in the budget to keep it up to date.

  In Colorado that summer of 2011, Alicia Cronquist wanted to know how likely it was for Listeria victims to eat cantaloupe. The answer from FoodNet was 65 percent. There was her “aha” moment. Of the patients Cronquist had been tracking, 100 percent
reported eating cantaloupe. And Cronquist knew one of them still had it in the refrigerator.

  The patient had purchased two melons, cutting one into pieces and putting it in the refrigerator. The second cantaloupe was still whole. State investigators gathered those cantaloupes and brought them to the lab for testing. Meanwhile, two state employees went grocery shopping around metro Denver, buying fifteen cantaloupes. They tested the rind, as well as the flesh of the melons, for Listeria bacteria.

  The lab work was rushed. If protocol was to incubate the swabs of the cantaloupe for four to sixteen hours, scientists incubated the specimens for “four hours on the dot,” Maguire said.5 His lab fast-tracked the genetic matching to find out whether the bacteria strains from patients’ blood samples matched the strains in the melon. To finish the work as fast as possible, the lab set aside routine lab work, such as tests that make sure milk is properly pasteurized, or others that track bacteria count on raw meat.

  Five cantaloupes from one store tested negative for Listeria. But all ten melons from the other store were contaminated with the bacteria. What’s more, the strains of Listeria found on those ten melons matched strains found in the ill patients. “Everything we were seeing in the patients, we were seeing in the cantaloupe,” Maguire said.

  One week after the first public warning, health authorities announced they had linked the source of the poisoning to cantaloupe. The cheese, deli meats, and dairy industries were in the clear this time, and while those officials exhaled in relief, the produce industry braced for the onslaught of attention—the bad kind.

  As more patient blood samples arrived at the state lab, they fell into three distinct strains, their DNA pattern matching side by side on a state computer screen like barcodes from a grocery store. Cantaloupe taken from patient refrigerators—both cut and uncut—had the same strains. But the confiscated cantaloupe did not have a sticker naming the farm.

  In interviews, though, patients were saying they had eaten cantaloupe from Colorado’s most well-known melon-growing region, a town called Rocky Ford. “Patients were spontaneously telling us the cantaloupe said ‘Rocky Ford’ on it,” Cronquist said. “That was coming out loud and clear. Some patients called it ‘extra sweet,’ which we all know is why it is so delicious.”

  The CDC refined its public health warning, this time not just naming cantaloupe, but specifically melon from the renowned Rocky Ford area of southeastern Colorado. It was a warning that would cause backlash from Rocky Ford’s farmers, who suffered as cantaloupe consumption dropped after the scare—up to an 80 percent drop, according to national estimates. Rocky Ford farmers claimed their reputation was tarnished unfairly, especially since the cantaloupe was later linked to a farm in Holly, some ninety miles away from Rocky Ford. The farm at fault—Jensen Farms—had borrowed the famed Rocky Ford name on its cantaloupe for marketing purposes.

  As the food tracking continued, investigators knocked on the door of Herb Stevens, the eighty-four-year-old patient who woke up shaking uncontrollably, had a fever of 102.7, and couldn’t stand up. Stevens lay seriously ill in a hospital bed in the Denver suburb of Littleton. Investigators wanted to talk to his wife, eighty-one-year-old Elaine. What did she serve to eat over the last month? Elaine couldn’t recall all of that, but she had something better than her memory: all of her King Soopers credit card receipts. She saved them to make sure she got all the frequent flier miles she earned by using her credit card. Health officials were also able to get her loyalty card information for the grocery store, so they could check her purchases and see for themselves exactly what the Stevenses had been eating.6

  Grocery store records gathered from patients showed which cantaloupe they purchased and on what days. Then, the FDA was able to track the melon purchases back to the distribution trucks, then back to the farm. The FDA narrowed the focus of the investigation to two farms, including Jensen Farms near Holly. Federal investigators descended on the farm, taking samples from soil and machinery.

  Just two days passed between the state health department’s warning to avoid Rocky Ford cantaloupe and a new warning. They had pinpointed the farm. Jensen Farms recalled the killer melons on September 14, 2011, about two weeks after the first reports of sick people began surfacing.

  The next morning, Colorado grocery stores launched automated tools to spread more warnings about cantaloupe. King Soopers has the phone numbers of loyalty card customers, and the store scanners recognize whether they bought cantaloupe. Upon news of the recall, grocery officials pushed the button on a robocall system that contacted each cantaloupe consumer and urged them to heed the recall. Among those who received a robocall was the Stevenses’ daughter, Jeni Exley. She, too, had bought a cantaloupe from the contaminated farm back in August. The melon never made Jeni sick, but the one her dad had eaten had thrown the family’s lives into turmoil.

  Herb Stevens was so ill from the cantaloupe he had to move into a nursing home. He never rebounded, never regained the strength to garden or walk without help. He died in July 2013.

  Laura Gieraltowski, an epidemiologist on the CDC’s foodborne outbreak investigation team in Atlanta, thinks of herself as a detective of sorts. Each week, she tracks up to twenty or thirty “clusters”—groups of people across the nation with the same strain of bacteria making them sick, more people than what epidemiologists would normally expect. Foodborne illness bacteria collected from patient blood samples is uploaded to the CDC’s PulseNet, which spits out periodic reports for Gieraltowski and her teammates.7 Many of the clusters fade out soon after they are noticed, without public warning or a product recall. That wasn’t the case with a cluster of patients, mostly children, who were showing up in doctor offices and hospitals across the country with matching strains of Salmonella bredeney in the fall of 2012.

  Gieraltowski was the lead investigator on that outbreak, which began, as usual, mysteriously. The early reports were that many of the sick children’s families shopped at Trader Joe’s, and based on interviews, these kids weren’t eating the standard American kid fare of frozen chicken nuggets and macaroni and cheese. They were eating organic snacks and drinking soy and almond milk.

  She held conference calls with local investigators in the twenty states where kids were sick, stricken with fever, diarrhea, and stomach cramps. The brainstorming first focused on milk and eventually wound around to organic peanut butter. Almost every family had shopped at Trader Joe’s, but only 60 percent of them remembered buying and eating peanut butter. Among those who reported eating peanut butter, though, several recalled the particular kind of peanut butter: Trader Joe’s Valencia. “That was the red flag,” Gieraltowski recalled.

  She wasn’t deterred by the fact that not everyone recalled eating that particular peanut butter, or any peanut butter at all. Rarely do 100 percent of people report eating the contaminated food that made them all sick. “Most people have trouble remembering what they ate yesterday,” Gieraltowski said, and it takes two or three weeks for Salmonella infections to make people sick enough to seek medical treatment. Plus, this was a case involving kids. “Maybe they ate a Trader Joe’s peanut butter sandwich at a friend’s house. Maybe you interview the mom but the mom doesn’t know what the dad gave them,” she said.

  A few of the states were able to collect some of the peanut butter from the pantries of sick children. In Washington, lab tests matched the strain found in the peanut butter to the blood samples collected from patients. Almost immediately after hearing the brand name—Valencia—CDC investigators notified the FDA so trackers there could begin finding out which plant makes Valencia. The search would lead them to Sunland, Inc., in Portales, New Mexico.8

  Sunland recalled hundreds of products, including peanut butter, almond butter, and cashew butter. The FDA confiscated peanut butter from the Sunland nut butter plant and found the same strain of Salmonella bredeney.

  Among the forty-two people reported ill, the average age was seven. Ten of those
victims were hospitalized. No one died, perhaps in part because of the detective work that led Gieraltowski and her team to the source of the outbreak within several days. Gieraltowski calls it “shoe leather epidemiology.” Evolving technology—databases including PulseNet and FoodNet and computer records like grocery store frequent shopper cards—speeds investigative work, but much of it still happens in interviews with patients and brainstorming sessions among epidemiologists that connect the dots. Those skills are evolving with the ever-changing American food chain. They’ve learned to ask more complex questions. “It’s just on the radar to ask, ‘What spices do you use when you cook?’” Gieraltowski said. At first, it might seem it’s the salami that’s sickening people, but on closer look, it’s the peppers inside the salami.

  It was after the Sunland peanut butter scare that the FDA for the first time used its new enforcement authority to halt operations at a food plant. The federal agency suspended production at the country’s largest organic peanut butter processor on powers gained through the 2011 Food Safety Modernization Act. The law gives the FDA authority to intercede on company operations when there is “reasonable probability” of causing serious health problems or death, an action that prior to the 2011 law would have required the FDA to seek court action.9 The plant reached a deal with federal authorities to reopen about a month after the forced shutdown. Included in the deal was a requirement that Sunland hire an independent expert to develop a sanitation plan.10

  The 2012 peanut butter outbreak, like the 2011 Listeria outbreak, ended with fingers pointed directly at the source of the problem. But that’s not always how it goes. Sometimes, foodborne illness investigators know exactly which manufacturer or food store chain shipped the contaminated food to grocery stores or handed it through the window in the drive-thru lane, yet they don’t tell us.