Seventy-six million people Americans sick with foodborne illness, 323,000 hospitalizations, 5,200 deaths.
Those grim numbers have been bandied about the Internet, the popular media, and more recently around congressional hearing rooms – like unemployment rates or major league batting averages. Public health advocates have cited them repeatedly while trying to push new food safety legislation through the House and Senate.
Critics claim the numbers, cobbled together in 1999 by the Centers for Disease Control and Prevention (CDC), are so much, well, diarrhea.
“76 million foodborne illnesses … really??” asks Dr. Richard Raymond, former U.S. Department of Agriculture undersecretary for food safety, who blogs frequently on the subject. He dismisses the stats as “fuzzy numbers” and “political doublespeak.”
Either way, those familiar statistics are about to become obsolete. The CDC has conducted a new study and reached new estimates of foodborne illness in America, and those statistics will be released any day now.
But don’t expect new estimates to resolve anything. The CDC will depict them as the best estimates possible, based on what they know. But they will be estimates, all the same.
So where do those numbers come from? And what do they mean?
The source is a 1999 study by Paul S. Mead and others at the CDC. It used data from a dozen or more sources, including CDC’s own FoodNet program, the Public Health Laboratory Information System, hospital discharge statistics, and a variety of studies published, mostly in the 1990s. Then it ran those statistics through a complicated formula to estimate the actual impact of foodborne illness in the US.
The study does NOT suggest that 76 million Americans each year are hauled off to hospitals, and hooked up to kidney dialysis machines. On the contrary, Mead and associates estimated that fewer than one-half of one percent are hospitalized. And of those, about 5,200 die.
The CDC’s standard for foodborne illness is essentially one person experiencing three or more bouts of diarrhea in a 24-hour day lasting one day or more. Uncomfortable, even painful, but not usually life-threatening.
As a result, most of those illnesses are never reported. The sick people do not feel sick enough to see a doctor. Even if they do, the doctor is unlikely to take stool specimens, let alone report the illnesses to state or local public health officials.
Based on previous studies, the CDC estimates, for example, that for every case of Salmonella infection that is reported, there are about 38 that are not reported. More severe illnesses, such as those caused by Listeria, were far more likely to be reported. In both cases, the researchers attempted to correct for the estimated underreporting.
Here’s a rough look at CDC’s method:
Based on their multiple sources, CDC estimated that 39 million Americans per year were sickened by “known pathogens.” These included notorious bugs such as Salmonella, Campylobacter, Shigella and E. Coli. But 23 million of those sicknesses were attributed to the less-lethal “Norwalk-like” viruses.
But far more cases of “acute gastroenteritis,” especially the less severe cases, cannot be traced to a specific pathogen, the study pointed out. The Mead study estimated those illnesses at a staggering 211 million per year.
But only about one-third of those illnesses are caused by food. The others are blamed on airborne, person-to-person or other forms of transmission. So divide by three.
And the bottom line: 76 million illnesses, 323,000 hospitializations, 5,200 deaths.
At least, that was the bottom line a decade ago. Since then, many state and local health departments have greatly improved their monitoring of food poisoning and other illnesses. CDC’s FoodNet has a decade more data and experience. DNA analysis has enabled health officials to link outbreaks to specific sources. The reported incidence of many illnesses has declined.
Perhaps most importantly, norovirus illnesses are more likely to be attributed to causes other than food.
Raymond, who carefully monitors CDC research publications, believes the new estimates will be markedly different, and should be much lower, than the familiar 1999 conclusions.
How different? Stay tuned.