And small clusters aren't uncommon; in any year Wisconsin may have around 200 cases. Davis adopted a wait-and-see approach.
But there was one confusing twist. Graff, who would become the second fatality, hadn't gone to the fair.
By midweek, there was an outbreak in another county, and a pattern was emerging. On Thursday, Sept. 7, the day Graff died, the director of the BloodCenter of Wisconsin told Davis he'd gotten requests for plasma for five people at five hospitals, all of whom had HUS. "That was very, very unusual," Davis says.
Wisconsin confirmed that all the hospitalized people had the same strain of E. coli O157:H7 and posted three of the test results Friday to PulseNet — a national database launched in 1998 that allows public health officials nationwide to track food-borne illnesses.
Davis then called Chris Braden, chief of the Outbreak Response and Surveillance Team at the CDC, to alert the federal agency.
By themselves, the Wisconsin cases didn't mean much. "There's always a certain number of background cases," says Robert Tauxe, chief of the food-borne and diarrheal diseases branch at CDC. In August, for example, PulseNet had had nine different E. coli strains from nine states.
The Wisconsin cases were different from the usual, unrelated cases but no one knew that yet. But they were now out there, waiting to see if anyone, anywhere else in country, was also infected with the identical E. coli strain.
By Tuesday, two other states had posted matches, PulseNet records show.
On Wednesday, Sept. 13, Robert Brackett, head of the FDA's Center for Food Safety, got an e-mail from Wisconsin asking if there was contamination reported in lettuce.
As lettuce had been the major source of more than a dozen O157:H7 outbreaks, it was a good question. But the FDA wasn't tracking anything in lettuce, so the answer was "no."
Oregon had a little E. coli cluster going, as well. But it wasn't initially viewed as a big deal — until Sept. 13. Melissa Plantenga, an investigator, told senior epidemiologist William Keene that of the six people who were sick, five said they'd eaten bagged spinach, although they named four different brands.
Keene wasn't surprised: Fresh spinach was a very plausible vehicle for E. coli O157:H7 because there is no guaranteed "kill step" in readying it for consumption. Bagged salads are washed and re-washed in processing plants, making them safe to eat straight from the bag, companies say. But the wash doesn't eradicate all bacteria.
Keene might have reacted more forcefully if he'd been able to check PulseNet and see that other states were reporting E. coli cases, too.
But he couldn't do that because the state had been bounced off PulseNet. A CDC-issued device that generates the security codes that allow states to log into the computer network had expired and the new one hadn't yet arrived in the mail.
Keene called officials in California, where most of the nation's leafy greens are grown, and learned that it isn't unusual for multiple brands of packaged bags to come from the same plant.
It was all falling into place. Keene pounded out a somewhat cryptic e-mail to the CDC: Oregon had a cluster of O157:H7 cases, and he wondered if anyone else was seeing anything similar. Ten minutes later, the CDC e-mailed him back, saying yes, there were a lot of other cases out there.