TCS Daily

Loopy Links

By Sandy Szwarc - September 30, 2004 12:00 AM

You are about to learn of a beverage so dangerous, that we must ban or restrict its sales, or at least enact tax penalties on it to deter consumption. Here's what the research shows:

• Every American who drinks it dies.

• It's been linked to obesity: in fact, bigger people drink the most of it.

• It's associated with type 2 diabetes and all diabetics drink it in especially large amounts.

• All heart attack victims drink it and it's a known factor in heart failure.

There are been hundreds of studies finding these correlations -- correlations so strong they make the evidence irrefutable. This is bad stuff.

Everything you've just read is true. What is it?


Of course, you could have filled in the blank with anything that today is frequently blamed for obesity, type 2 diabetes, heart disease or premature death: sodas, high fructose corn syrup, dietary fat, carbs, high cholesterol, prediabetes, fast food, snacking, trans fats, watching television and all sorts of things others want to fix in us. And they're all just as spurious as water.

This illustration demonstrates just how easy it is to think that correlations (links between things) mean anything at all. Just because certain lifestyle or dietary habits, laboratory values or numbers on the scale, rise or fall in synch or appear together, doesn't mean they have anything to do with each other. Yet, we hear assertions made every day by mainstream scientists and medical professionals, reputable healthcare organizations, public policy makers and, most of all, media in which correlations are used as proof of a cause. These are taken as facts, not because of any sound evidence, but because they seem intuitively correct and match what "everybody knows."

But correlations taken as cause become even more nonsensical ... and dangerous ... when the link is turned backwards to say:

"Therefore, restricting or eliminating water ("it") will prevent or cure obesity, heart disease or type 2 diabetes."

Please don't try that at home. It's clearly a preposterous and groundless cure.

We should all be concerned by how correlations found in "studies" or even simply incorrectly assumed, are being used to support healthcare guidelines and public regulations, with absolutely no proof that such solutions work. Even worse, they completely disregard the harm that can result. For instance, people at risk for type 2 diabetes, believing sodas and sweets are the cause, might change their diets but fail to do the very thing that averts, minimizes and even reverses the condition: physical activity. This mistake could cost them their lives, vision or limbs. People might restrict their calories, fats or carbs (dieting) in futile attempts at weight loss, but fail to do the one thing that would avert, minimize and even reverse supposed "obesity-related" health concerns: physical activity. This could significantly increase their risks for premature death, heart disease, high blood pressure, stroke, type 2 diabetes and cancers.

To protect yourself from making unsound health choices for you or your children, or putting your support behind costly public health solutions, learn to identify "data dredge" studies -- where correlations frequently come from -- and to differentiate them from evidence you can trust to mean something. Data dredges, are among the weakest types of epidemiological studies upon which we can base any meaningful conclusions about our own health.

Data Dredge of the Week

Last week, a study led by Barry Popkin, PhD at the University of North Carolina at Chapel Hill, was released which claimed soda consumption had increased 135% since 1977 and since rates of type 2 diabetes and obesity were rising, too, that was evidence that "consuming these [drinks] increase weight gain in children and adults."

Based on that correlation alone, they then leapt in reverse to conclude, "reduced soft drink and fruit drink intake ... would seem to be one of the simpler ways to reduce obesity in the United States."

Did you catch the fallacies in this example? Just because consumption of a certain food goes up or down among an entire population does not demonstrate that only fat people are eating that food or that that food is the cause of obesity or type 2 diabetes. Such correlation-generated claims rely on the belief that fat people eat differently. But consumption of sodas and sweets, for instance, have been shown to actually be as high or higher among thinner, more active people. Such claims also rely on the belief that sugary foods and beverages cause obesity and type 2 diabetes. But sugar has been studied probably more than any other food ingredient in history and it's been repeatedly found to not cause obesity, type 2 diabetes or any chronic disease. In fact, a surprising number of studies have demonstrated an inverse relationship between dietary sugars and obesity.

Popkin cited a study led by David Ludwig of Boston Children's Hospital in 2001 to support sweet beverages' role in obesity, which Popkin said "showed the effect of increased consumption of sugar-sweetened beverages on increased energy intake and obesity among U.S. teens." But Ludwig's study actually found no difference in the BMIs of children consuming the most and least amounts of sugar and the researchers noted "there is no clear evidence that consumption of sugar per se affects food intake in a unique manner or causes obesity."

The Popkin study was a "meta-analyses," lumping together five different dietary surveys (telephone surveys to questionnaires) gathered over the decades from a total of 73,345 random individuals. These one- and two-day population dietary surveys were all done using different methods and also underwent significant redesigns over the years to probe for more complete information and lessen under-reporting, meaning the earlier surveys would be more likely to under-estimate how much people actually ate and using them would accentuate perceived increases. Like all meta-analyses, when researchers combine data from several different sources trying to create something bigger and more convincing, their results are actually more untenable. I call them Rorschach2 studies. That might explain why sounder studies, such as those at the University of Michigan led by Youngme Park which closely following the diets for weeks at a time for years of a total of 12,000 children, have found no increase in soda consumption and no evidence that sodas were reducing milk consumption.

Of the thousands of foods and beverages people consume, this study chose sodas. But in typical data dredge fashion, Popkin could have mined that databank and pulled out anything...and has. For example, in a previous study he found grains, legumes and low-fat milk intake up among adults since 1965, along with significant decreases in calories and percentages of dietary fat. Yet he didn't tie these overall "healthful" eating trends to rising rates of obesity or type 2 diabetes. Why, that wouldn't have made sense.

© 2004 Sandy Szwarc. All rights reserved


TCS Daily Archives