At least 14 people in six states have fallen ill in an E. coli O145 outbreak that killed a Louisiana toddler on May 31st.

And while public health investigators have identified the deadly bacteria strain, they have yet to figure out where it’s coming from.

Now the hunt is on. As state health departments coordinate with the U.S. Centers for Disease Control and Prevention to pinpoint the outbreak’s source, experts say it’s most likely food.

Food Safety News spoke with Dr. William Keene, senior epidemiologist for Oregon Public Health, to get an idea of how public health officials in the affected states might be investigating the outbreak. Though he’s not involved in this investigation, he has one of the country’s strongest track records of cracking outbreaks.

As Keene puts it, it all starts with the most useful tool in an epidemiologist’s arsenal: The questionnaire.

culturemedium-406.jpgKeene: “In Oregon, for example, we have a standardized, hypothesis-generating questionnaire with about 500 or 600 questions on it. That’s a lot. And in effect, we ask a bunch of questions about where people get food.”

As best they can, investigators try to cover every possible place the victims might get food, from the everyday places like grocery stores and restaurants to the not-so-everyday, like a friend’s house, a sporting event or a movie theater.

Keene: “A lot of places where we eat or obtain food are kind of forgettable, so we try to jog people’s memories.”

Pretty soon, they try to get into specific foods people ate, but there are always limitations. Not everyone can spare the time to list everything they’ve eaten for the past few weeks.

Keene: “You have to cut corners and make compromises, and I think it’s more efficient to get data about more foods in less detail. To the extent that you start asking what brands and where you shop and how you prepare the food, that’s going to slow you down and you’ll have to drop some things from your questionnaire.

“And that’s one of the big trade-offs in the way people do it around the country. Some people like to ask about fewer foods to get more detail up front, and I like to ask about more foods with — generally speaking — less detail, and then if necessary we’ll go back and re-interview people.

“But no matter how you do it, you’re going through a list and trying to find out what the cases have in common, and to what extent those patterns look abnormal to you.”

Of course, the cookie-cutter questionnaire is not the only way to go.

Keene: “Another approach that’s also successful is to have people with a lot of skill and experience with this doing more open-ended interviews where they’re sort of getting people to talk about the foods they eat for breakfast or lunch or dinner.

“We have friendly arguments about what’s the best way to do it, and I think the answer is that what’s best for some may not be best for everybody. But we all agree that the answer is to get out there and do these comprehensive interviews as quickly as possible.”

Most of the time, they’ll find the source, but often the question is whether they can find it fast enough, and no investigation is ever guaranteed to find it at all.

Keene: “There certainly are well-investigated outbreaks of this size — or much larger — where at the end of the day we had no idea what it was. And, again, there are some vehicles that are just intrinsically difficult to identify. Of course, the problem of second-guessing why we struck out is difficult when you don’t know what the product was.”

One of the best ways to narrow down an investigation is to look at who’s getting sick.

Keene: “So one of the key things to look at are the demographics of the cases. Just the simple facts of ‘how old are they,’ ‘what sex are they,’ often gives you very, very important clues into the type of product it is.

“If you have a bunch of cases and most of the cases or all of the cases are children — just making up an example — you’d know it’s not going to be vegetables. Kids don’t eat lettuce more commonly than adults do.”

In this outbreak, ages range from 1 to 79 years old, with a median age of 33. 79 percent of the cases are females.

Keene: “If the profile is adult females, you start looking at things like sprouts or another salad bar item. If the profile is one of mostly adult men, I hesitate to even speculate on what you might think about, but beef jerky, beer — something like that.

“Geographic distribution of cases: If cases are all in one region, you’re looking for a product that’s produced or distributed in that region. If cases are spread out over months and months and months, you’re looking at something with a long shelf life or it’s a continuing problem at a plant like the ground turkey outbreak of the last year.

“If it’s just a short spike and then the cases peter out within a few days or weeks, that could be something like a produce item that gets harvested from one field and happens to have some contamination on one corner of it. If that works its way though the market and through consumers, you get a wave of illnesses that’s delayed by a week or three, depending on how it takes to get the product to market. But then it’s gone once that corner of the field is harvested and gone.

“Again, all those clues get added to the questionnaire data and usually help us cut through the fog. Usually, once the right hypothesis emerges. It takes relatively little time to confirm it to our satisfaction.

“The end game of these outbreaks can be astonishingly fast — just hours sometimes from when the idea first comes up to when we’re moving ahead with product recalls or public announcements or something like that.

“You can really be making no progress, no progress, and then suddenly you know the answer. I hope we’re going to find the answer to this soon and either identify at least what already happened or prevent ongoing cases if it’s an ongoing source.”