Junk Science: Food Nannies’ Halloween Cancer Scare
The latest food scare was announced, appropriately enough, on Halloween. But the science behind the scare is about as believable as are ghosts and goblins.
"Landmark Report: Excess Body Fat Causes Cancer; Panel Also Implicates Red Meat, Processed Meat and Alcohol" blared the media release about a new report from the American Institute for Cancer Research (AICR).
The massive 537-page tome "assembled over five years by nine independent teams of scientists, hundreds of peer reviewers and 21 international experts who reviewed over 7,000 large-scale studies” purports to be the “most comprehensive ever published on the evidence linking cancer risk to diet, physical activity and weight."
"The most striking finding in the report is that excess body fat increases risk for numerous cancers... Even small amounts of excess body fat, especially if carried at the waist, increase risk," proclaimed the media release.
The report advises limiting the intake of hamburgers, French fries, milk shakes, pastries and soft drinks. It says that there is "no safe level of consumption" of processed meats a hysterical claim that is not even true for the most poisonous substances.
This certainly is a landmark report never before have so many scientists labored so long to embarrass themselves and their academic disciplines.
There’s not enough room in this column to debunk each and every claim made in the AICR report, but we’ll look at some examples after considering some fundamental facts and principles.
First, scientists don’t really understand carcinogenesis very well. It’s known that the risk of cancer increases with age possibly because of the deterioration of DNA repair mechanisms and a few well-documented risk factors, such as family history of cancer, heavy smoking, and exposure to certain viruses and some exposures to radiation. Outside of those and perhaps a few other risk factors, the occurrence of cancer is largely inexplicable.
Significantly, not a single case of cancer among the tens of thousands studied in the "7,000 large-scale studies" was definitively linked with any specific dietary factor. The AIRC report’s claim to link diet with cancer largely amounts to post-facto guesswork abetted by statistical hijinks and imagination run amok.
A cardinal principle of epidemiology is that it is a very useful methodology when looking for linkage between high rates of rare diseases the sort of relationship classically found, for example, in outbreaks of food poisoning.
But epidemiology is wholly incapable of identifying low risks of relatively common diseases or conditions, such as most cancers. The reason for this is simple: the margin of error in study data due to inaccurate and incomplete data collection is typically far greater than the size of any statistical relationship that may exist or be detected.
Accordingly, the rule of thumb in epidemiology, as famously espoused by the National Cancer Institute, is that, "In epidemiologic research, [increases in risk of less than 100 percent] are considered small and usually difficult to interpret. Such increases may be due to chance, statistical bias or effects of confounding factors that are sometimes not evident."
Further, just because a reported risk is greater than 100 percent, that does not necessarily indicate a cause-and-effect relationship. Such reported risks may be statistically insignificant (indicating they could have occurred by chance) or have wide margins of error (indicating flaky data). And, of course, for any statistical risk to have meaning, it must be backed up by biological plausibility.
With these concepts in mind, let’s consider the AICR report.
The vast majority of the results from individual studies between every type of food and every type of cancer cited in the report are either significantly below 100 percent and/or statistically insignificant. The relatively few cited risks that exceed 100 percent are typically not statistically significant or have wide margins of error.
Consider the data presented for processed meat, which the AICR report claims to be too dangerous to eat.
Of the 17 study results concerning processed meat and colon cancer comparing high consumption to low consumption 15 are way below, and one is at the 100 percent-risk threshold. Thirteen studies aren’t statistically significant. Not only is the lone study claiming a risk above 100 percent (a reported 250 percent increase in risk) barely statistically significant, it has a margin of error four times the size of the reported risk.
Of the seven studies reporting a cancer risk per serving of processed meat, all reported risks are substantially below the 100 percent threshold. Four results are clearly not statistically significant and two are borderline insignificant.
On the basis of these dubious statistical results, the AICR report concludes that “processed meat is a convincing cause of colorectal cancer.” This is an appalling and unsupported conclusion.
In the end, the AICR report isn’t really science at all it’s more of bloody crime scene where science got violently mugged by hoods costumed as health and nutrition experts and wielding statistical pepper spray. In some ways, this shoddy science isn’t surprising when one considers that the AICR also pitches cranberry recipes and other culinary snake oil as a means of reducing cancer risk.
The AICR advocates against consuming fat, salt, sugar and alcohol-- an agenda worth $37 million in charitable donations in 2006. So we shouldn’t be surprised when the food police issue a “report” advancing such a lucrative agenda.