After blaming cucumbers, backpedaling on the cucumbers and blaming bean sprouts, then backpedaling on the sprouts, German authorities have now concluded that bean sprouts are, in fact, to blame for the spread of E. coli O104:H4, which has sickened more than 3,000 people and killed 31. Patients with the most severe cases have suffered kidney and neurological damage.

This morning, authorities announced in Berlin that epidemiologic evidence, rather than laboratory results, pointed to bean sprouts from an organic farm in Lower Saxony as the source of the outbreak. The New York Times’ Alan Cowell explains how investigators reached this conclusion:

[Reinhard Burger, head of Germany's disease control agency] said investigators had examined 112 people, 19 of whom had been infected with E. coli during a group visit to a single restaurant, and had examined recipes for the food they had eaten, spoken to the chefs and even examined photographs they had taken of one another with their choice of food on the table.

The aim was “to discover exactly how each meal prepared, which ingredients went into it,” he said. The result was that customers who ate sprouts were found to be almost nine times more likely to be infected than other diners. It was this trail that led health inspectors to the organic farm where the sprouts originated.

If you want a more detailed explanation about how such outbreak investigations work, CDC has a great writeup (which includes an example I encountered in my first public health class, the famous outbreak of gastroenteritis following a church supper in Oswego, New York, in 1940).

To an audience accustomed to CSI dramas, laboratory tests linking a food item to the bacteria will generally be a more sataisfying form of evidence. A few hours after the Berlin announcement, officials in North Rhine-Westphalia said they’d finally identified the E. coli strain in a package of bean sprouts from the suspect farm.

How did the bacteria contaminate the sprouts in the first place? Tara C. Smith at Aetiology has raised a few possibilities in her posts on the outbreak: Animals in the area might be carrying the bacteria; water on the farm might have been contaminated (by sewage, for instance); the bacteria might have originated in the seeds; or a sick worker might have spread it to the sprouts. CIDRAP cites a statement by a farm spokesperson that three women who helped package the sprouts had diarrhea during the first half of May – but that doesn’t tell us whether the infection originated with an employee or whether the workers were sickened through contact with the sprouts. I’m sure investigators will keep pursuing this question, but they might never get a satisfactory answer.

The Importance of Surveillance
The number of new cases being reported has fallen, and the sale of produce from the farm in question has been suspended. Now, public attention will probably turn more fully to the shortcomings in the official outbreak response. An editorial in The Lancet states, “communication surrounding the outbreak has been haphazard at best, dismal at worst.” Farms and vegetable sellers have suffered enormous losses while consumers, not knowing the source of the outbreak, avoided whole swathes of fresh produce.

Outbreak investigations in countries like Germany and the US will inevitably be complicated by the fact that most of us eat a wide variety of foods from a wide variety of sources. Over the past week, I’ve eaten food from a multi-vendor farmers’ market, two different grocery stores, and two restaurants. Some of it had been sitting in my cupboard or freezer for several weeks, and the packaging has probably been hauled away already. The restaurants probably get ingredients from multiple suppliers. If an illness has a relatively long incubation period or I’ve already recovered, I’ll have to remember even further back, and my memory will be hazy.

Even with these complications, though, it’s possible to create a public-health infrastructure that can identify and respond quickly to outbreaks of foodborne illness. Der Spiegel reports on some shortcomings in the German response:

Eighteen valuable days were lost between when the first patient came down with diarrhea around May 1, and when the RKI [Germany's disease control agency] was finally alerted.

While part of this may be because those infected waited until their cases became severe to see a doctor, the delay also indicates how unprepared German doctors were to properly report and handle the outbreak. Many fail to consider the possibility of an E. coli infection at all, sending patients with diarrhea home with a prescription and failing to take stool samples. But the larger problem is that local German health authorities are given abundant time to relay news of E. coli infections in their area. Infections, evidence of E. coli sources and even resulting deaths must be reported to state authorities just once a week, at the latest on the third working day of the week following initial identification. State authorities then have another entire week before they must inform the RKI. “Why can’t this just be directly sent electronically?” asked Wieler of the Institute of Microbiology and Epizootics in Berlin.

A good disease surveillance system will ensure that three important things happen quickly: 1) doctors get stool samples from patients reporting symptoms of foodborne illness; 2) laboratories process the stool samples; and 3) a central agency receives reports from multiple labs and monitors them for evidence of outbreaks. The more quickly an outbreak is identified, the more quickly the patients can be interviewed and the source traced.

Germany’s surveillance system didn’t work as well as it should have. In the US, it seems to be patchy. We’ve got CDC’s FoodNet, which uses 650 clinical laboratories to test stool samples for a range of pathogens and issues regular reports on its findings; the latest one was just released. However, as the New York Times’ Gardiner Harris pointed out in a 2009 article, the quality of surveillance varies widely from state to state:

The importance of a few epidemiologists in Minnesota demonstrates the problem. If not for the Minnesota Department of Health, the Peanut Corporation of America might still be selling salmonella-laced peanuts, Dole might still be selling contaminated lettuce, and ConAgra might still be selling dangerous Banquet brand pot pies — sickening hundreds or thousands more people.

In these and other cases, epidemiologists from Minnesota pinpointed the causes of food scares while officials in other states were barely aware that their residents were getting sick. From 1990 to 2006, Minnesota health officials uncovered 548 food-related illness outbreaks, while those in Kentucky found 18, according to an analysis of health records.

… Some delay is inevitable. Most people sickened by food do not bother to see a doctor. Many of those who do are not asked to provide a stool sample, and when asked, some refuse.

When patients are willing, laboratories may not be. In Utah, for instance, only 18 of the state’s 1,388 medical laboratories process stool tests, said Dr. Pat Luedtke, director of the Utah public health laboratory. Well-meaning doctors who wish to send stool samples sometimes must pay the postage because insurers often refuse to pay for a test that largely serves a public health function; many doctors do not bother.

I hope Minnesota’s still funding its surveillance system during the ongoing economic crisis (Update, 6/20/11: Maryn McKenna reports that the state’s health department’s work is in jeopardy). At the federal level, as I noted last week, a House appropriations subcommittee has cut the FDA’s food-safety budget. These are the kinds of decisions that might not feel too painful until another outbreak strikes.