How Modern Science and Old-fashioned Detective Work Identified Tomatoes


The contamination of tomatoes with a rare strain of salmonella has led to the largest outbreak of food-borne illnesses since E. coli in spinach killed five and sickened hundreds almost two years ago.

At least 277 people of all ages in 28 states and the District of Columbia have been sickened; 43 have been hospitalized. A nationwide recall of round, plum and Roma tomatoes has dealt a sharp blow to the $2.7 billion fresh-tomato market, costing the food industry tens of millions of dollars.

KEY DATES: How the outbreak unfolded

But it could have been a lot worse if a red flag hadn't been raised early in the outbreak last month by a public health nurse with good instincts in one of the nation's poorest, most remote regions.

Indeed, health officials say that because the first cluster of patients surfaced on the Navajo Nation in New Mexico, where they are served by a small, close-knit medical community, federal investigators were able to quickly identify the contaminated foods and take steps to contain the outbreak the past two weeks.

After being the first to recognize the signs of an emerging outbreak, the federal Indian Health Service staff played a key role in the search for the tainted food. "It was 21st-century molecular epidemiology and old-fashioned boot leather," says John Redd, the infectious disease branch chief with the Indian Health Service in Albuquerque. "You've got to get out from behind your desk and hit the road sometimes."

Kimberlae Houk has 24 years of experience in public health nursing in the Navajo Nation, the largest reservation in the USA, with lands extending into Arizona, Utah and New Mexico.

Her Shiprock Indian Health Services Unit provides medical care to more than 45,500 American Indians, mostly Navajo, in an area that covers 23 communities in the three states. Homes can be extremely isolated, and many are without telephones.

Houk knew something was up on Monday, May 19, when four people very sick with diarrhea, fever and abdominal cramps showed up at the Northern Navajo Medical Center in Shiprock, N.M.

"A lot of time with these kinds of diseases you get your babies and your grandmas in the hospital," she says. "But in this one we had fit 30-year-olds. And we just don't get 30-year-olds in the hospital with dehydration."

And these people weren't just dropping in at the doctor's office. "We serve a very rural population. They have to drive an hour to the clinic and an hour back. So it's a big deal to come in," Houk says.

With previous experience with outbreaks of measles, whooping cough, hantavirus and even the plague, Houk immediately went into outbreak mode. "We literally drop everything when there's a communicable disease, to protect people."

That day, "We all just ran," says Houk. "We can really get on top of things quickly because all our nurses, our doctors, our clinics, our labs, we're all under the same roof."

Salmonella is a bacteria found in animal intestines that can be transmitted when foods are contaminated by animal feces. It's a reportable condition, meaning if it pops up anywhere, state and federal health departments must be notified.

Suspecting that bacteria could have sickened their patients, Houk and her colleagues sent stool samples to New Mexico's state laboratory. The tests showed that the patients had saintpaul — one of 2,300 different strains of salmonella, and a very rare one.

Next the state lab did a genetic fingerprint on the bacteria to make sure that all were indeed the exact same strain.

"Our lab called us on May 21, which was a Wednesday," says Paul Ettestad, who works in the state's infectious disease epidemiology bureau. The salmonella saintpaul cases all had the exact same DNA fingerprint. It was all one outbreak.

Over the next two days, cases appeared in several New Mexican counties. Some patients were Navajo, some were not. All were very ill. "That really starts ringing the bells," Ettestad says. He contacted Ian Williams, chief of the outbreak team at the Centers for Disease Control and Prevention.

Cases were starting to pop up in Texas, too. "We had two cases reported out of Houston on the 22nd, and an additional 12 on the 23rd. Things moved pretty quickly," says Linda Gaul, head of the food-borne illness team at the Texas Department of State Health Services.

Then New Mexico posted the genetic fingerprints of its cases onto PulseNet, the CDC's computer disease-tracking network. Within hours, matches began to show up. The outbreak wasn't just in New Mexico and Texas, it was all over the country.

The common thread

Now came the challenge: What connected a patient on remote Navajo lands with the other patients throughout the nation?

Even though it was Memorial Day weekend, everyone mobilized to work. New Mexico, Texas, the CDC and the Indian Health Service began holding daily conference calls. As other states got patients, they joined in. New Mexico started the calls, but eventually CDC took over hosting them because they "can accommodate 100 people on their lines," Houk says.

In a case like this, epidemiologists, the doctors who study outbreaks, pull out what they call a "shotgun survey." It's a long — in this case 22 pages — survey that covers just about anything a person might eat, drink or be exposed to that could cause such an illness.

"Shotgun, because it's like shooting in the dark to see what's there," says Texas' Gaul.

Enter the Indian Health Service again, charged with the task of administering the survey in the Navajo Nation, which by sheer chance seemed to have gotten the most cases in New Mexico.

"Our Indian Health nurses would drive two or three hours to try to find these people and when they couldn't find them, they'd have to go back," he says.

Also, each nurse had to painstakingly reconstruct everything their patient had eaten in the previous two weeks. "Nurses pulled out calendars for clients and said, 'Where were you? Who were you with? What meal did you have with that event? What did you do before the event, and what did you eat later in the day?' " says Houk.

Not only that, but the surveys had to be given in three languages: English, Spanish and "English with Navajo clarification," Redd says.

"The folks in New Mexico really did an amazing job," says CDC's Williams. "Extraordinary."

By Saturday more than a dozen of these questionnaires were completed. Epidemiologists and public health workers pored over them, looking for patterns. Jessica Jungk, a master of public health who also helped track spinach in the 2006 E. coli outbreak, got called in to help analyze the data. But while tomatoes were high on the list of foods eaten, they weren't a strong enough presence to be isolated as the problem.

Ettestad got on the phone with CDC's Williams, who urged really zeroing in on what people were eating. "Get them to open their refrigerators, their pantries" while they talk, he suggested.

To do that, a nurse was sent into a patient's home "and she literally pointed at every shelf on the refrigerator and every cabinet and asked, 'Did you eat anything on this shelf?' It's a difficult thing to do. It makes people feel anxious," Houk says.

But it did the trick. Even people who swear they didn't eat raw tomatoes remember they might have when asked about salsa or guacamole or a slice of tomato on a hamburger. On Saturday, with surveys coming in not only from New Mexico but a few other states as well, the percentage of patients who'd eaten fresh tomatoes stood at 75%, compared with an average of 68% random Americans. By Sunday, with more cases analyzed, the share shot up to 83%, Ettestad says.

But that wasn't enough to nail tomato as the culprit. Next came the scientific gold standard, a case control survey to look at whether people who didn't get sick ate significantly less of the suspect food than people who did. For this, the investigators employed a tried and true tool for random selection: spin-the-bottle.

The key to a good case control is randomness. Investigators want to compare healthy people with those who got sick, but they want them to be similar in every other way, and they don't want to bias who gets chosen. Today epidemiologists use computerized phone books and an Internet randomizing program.

But a lot of the people they needed to talk to didn't even have phones.

So they want back to the old techniques. "You go out to the house of someone who got sick. You take a bottle and put it on the ground. You spin it and you go in the direction it points until you hit a house," Ettestad says. "And that's just as random as the Internet."

Redd of the Indian Health Service didn't even have a bottle: "I was spinning a government-issue pen."

By Saturday, May 31, New Mexico was ready to start warning people. "We felt we had enough evidence and we needed to protect our citizens," Ettestad says.

Three days later FDA issued a warning for Texas and New Mexico. Five days later, with cases appearing across the nation, FDA made the warning national.

The agency was able to quickly announce where the tomatoes didn't come from because most tomato-growing areas weren't harvesting in late April when the first cases showed up. It soon became clear that mid-Florida and Mexico were the only major growing areas selling tomatoes at that point.

Investigation continues

Efforts to pinpoint the source of the contaminated tomatoes are ongoing, and the FDA is still identifying clusters of cases and backtracking to the suppliers.

Food producers say the financial impact could surpass that of the spinach contamination. At least 50% of Americans eat tomatoes regularly, but only 4% eat spinach, according to the NPD Group, a market research firm.

Tomato sales have fallen amid reduced demand from grocers and restaurants, and prices for varieties affected by the outbreak have likewise dropped, says Mark Munger, vice president of marketing for Andrew & Williamson Fresh Produce.

Major restaurant chains such as McDonald's are once again using tomatoes, carefully sourced to make sure they came from regions that weren't implicated in the outbreak. They're also back in supermarkets, often with notes posted above them about where the tomatoes were picked and why those areas were safe.

But consumers will remain wary, especially if the FDA can't locate the source of the outbreak, says Tom Nassif, CEO of Western Growers, which represents growers in Arizona and California.

The outbreak is a reminder that consumers pay a price for the vast agricultural production and distribution system that supplies cheap, plentiful produce year round, says Kenneth Albala, a food historian at the University of the Pacific in Stockton, Calif.

"In the past most food was produced and consumed locally, you wouldn't have much trouble figuring out where (tainted) food came from," he says.

Today, he says, having fresh, ripe, cheap tomatoes available in salads nationwide in April — inconceivable two generations ago — also means "the distribution is so broad that something (contaminated) can show up in 13 states the next day."

Contributing: Julie Schmit