|
|
| Volume 46, Number 2, Fall 1998 |
by Dan Wilson
About 9,000 Americans die every year from food
poisoning…
—The Associated Press, December 17, 1997
When food-borne illness became a major public policy issue, propelled by a series of high-profile outbreaks over the past four years, the press naturally wanted to know how many die from it. A figure of 9,000 Americans per year emerged. But where did the figure come from?
A Nexis check shows that some stories attributed it to a variety of sources, including the U.S. Department of Agriculture, the Centers for Disease Control and Prevention, the General Accounting Office, and congressional testimony. Many more, if not most, did not attribute it at all, as if the number were one of those accepted truths that require no attribution, like “squirrels have bushy tails.”
But it isn’t. My editor told me to locate the source and determine its veracity. “If that many people are dying of food poisoning,” she asked, “why haven’t we heard of a case locally?”
I found dozens of stories that used the same figures—a range of 6.5 million to 33 million people are sickened and approximately 9,000 die each year. I focused on the fatality figure because it is a number that can be quantified. Deaths are recorded.
I started with the USDA. A spokesman said the department uses the number, but that it came from the CDC in Atlanta. A spokesman there said the number comes from a scientific paper known as the Altekruse report. She faxed it to me.
And there it was, that same number, only footnoted. The footnote revealed the number came from something called the Council for Agricultural Science and Technology (CAST), a think tank located in Ames, Iowa. In 1994 it issued a report, “Foodborne Pathogens: Risks and Consequences,” which I obtained.
A task force created by the think tank specifically to study the food-borne pathogen issue had relied upon two researchers—Dr. Ewen Todd, a microbiologist, and Dr. John Bennett, medical doctor—to come up with annual estimates of illnesses and deaths caused by food poisoning. They produced wildly different numbers. Todd: 12,581,630 cases resulting in 522.7 annual deaths. Bennett: a lower number of total cases, 6,485,755, but resulting in a higher 8,982 deaths. Both scientists had used mathematical models to arrive at an estimate. Neither model was based on known cases.
When its report was issued with an accompanying press release, CAST adopted Bennett’s 9,000 figure. Todd’s fatality estimate was not used.
Before getting on the phone, I called up the CDC’s Web site for the annual Summary of Notifiable Diseases, which lists the numbers of recorded deaths by category—hard numbers based on reports from public health officials around the country. The differences between those numbers and Bennett’s were striking.
According to Bennett, for example, 1,000 people die annually from trichinosis, a pork parasite. According to the CDC, only one trichinosis death has been recorded in the past ten years. Bennett posited 28 deaths per year from typhoid fever, which is carried by shellfish; the CDC recorded a total of 21 over a ten-year period. (The CDC has not been tracking E. coli deaths long enough to have a number.)
I put all this to Dr. Tanya Roberts, co-chairman of CAST’s task force. “Notifiable deaths are horribly under-reported,” she said. What the CDC has, “is a legal record, not a medical record.”
Roberts concedes that some of Bennett’s numbers may be inflated: “Until we do a good analysis I would say we don’t know for sure. I don’t know where the truth lies, and I don’t think anyone else does. I said up to 9,000 deaths (in the press release). I don’t think Todd is accurate and I don’t think Bennett is accurate. The truth is somewhere else, or in between.”
Still, Roberts said she leans towards the higher number, because “Bennett’s science is the best to date.”
Nonetheless, numbers that were based on one researcher’s best guess have achieved the status of unassailable truth simply by being run through several spin cycles until they were adopted without attribution by many reporters.
Somewhere in the cycle comes a new phenomenon; I call it unattributed-numbers bracket creep. This from U.S. News & World Report, November 24: “Each year up to 81 million Americans suffer a food-borne illness; 9,100 die.” And this from USA Weekend, January 23: “Deaths from tainted food topped 10,000 last year…” In neither story were sources cited for the rising numbers.
Reprinted from the May/June 1998 Columbia Journalism Review. © 1998 Graduate School of Journalism, Columbia University.
Wilson is a reporter for the Post-Crescent (Appleton, WI).