Careful With Correlations

Every day, Haley’s comet careens further from the earth. Every day, my cat gets more ornery.

These two things are perfectly correlated, and yet have absolutely nothing to do with each other. Science 101 teaches you that two things moving together, no matter how tightly in lock step, are not necessarily tethered. They may be, but correlation only implies cause, it doesn’t prove it.

It’s surprising then, to hear scientists routinely confuse cause and effect like this. Just this week, we read that “If salt intake was reduced by half in all children … there would be an average reduction … in the consumption of sugar-sweetened soft drinks … of 2.3 soft drinks per week.” This is according to Dr. Feng J. He, M.D, at St. George’s University of London, England.

What Dr. Feng’s study showed was not that reducing salt intake reduced soda consumption. What he actually found was that the kids who ate more salty snacks also drank more soda. The authors of the study only guessed that the two are linked.

Incidentally, they also found that kids ate more snack foods as they got older: increasing age is correlated with increasing salt intake. But no one is stating that “If salt intake was reduced by half in all children … there would be an average reduction … in the aging of our children … of 2.3 years per child.”

But, so what? Why niggle like some picky scientific method wonk?

No one is suggesting that sodas are good for you, nor that salty junk food snacks are in any way healthy. Further, we all agree that we should reduce the consumption of both, if our goal is optimal health. That said, this research report states quite flatly that you can reduce your child’s consumption of soda by reducing the salty snack foods.

So the reason to give this study a Yellow Card is simply because it is misleading. The authors are telling you that Haley’s Comet is causing my cat to be grumpy, and if we could somehow reverse its meteoric ellipse, I would get a pliant lap cat who is 99% hiss-free. In other words, cut out salty snacks and your kids will give up sodas on their own, because these two are causally linked.

But this is an error we have seen before. Do you remember when we were told that coffee was bad for us? One of the reasons for that conclusion came from the very same Science 101 snafu. In the 1950s, researchers identified a correlation between coffee consumption and cancer.

Therefore, they surmised, coffee causes cancer. However, later, they figured out that the folks who were drinking their morning joe were also smoking cigarettes. Cancer and coffee were perfectly correlated; perfectly unrelated.

To reduce the consumption of foods that fuel our childhood obesity epidemic (and the adult version as well), a great first solution would be to do your own correlation analysis research.

Measure the consumption of salty snacks and sodas when you have them around the house, compared to when you just don’t buy them. I hypothesize that you will see statistically significant decreases in consumption, when they are not in your home. You might even write a research paper on it!

The study they quoted was very well done, showing that kids who consumed low levels of salty snacks also drank fewer sodas. In fact, each reduction of 1 gram/day of salt was correlated with 27 g/day less of sugar-sweetened soft drinks.

The researchers suggested that reducing children’s salt intake would cause them to consume fewer calories in soft drinks, therefore helping to reduce childhood obesity. It might also improve their cardiovascular health directly, since salt intake is correlated with blood pressure.

Dr. He and colleagues determined that mean salt intake rose with age, from 4.6 g/day (SD 1.5) among 4-year-olds to 6.8 g/day (SD 2.1) for 18-year-olds. The researchers did not count salt added during cooking or at the table. “The majority of salt intake is from salt already hidden in food, i.e., added by the food industry,” they wrote.

The researchers found that soft drinks accounted for more than half of fluid consumption among most age groups for girls and boys, except for 17-year-olds. Taking all of the participants together, 56% of total fluid intake was in the form of soft drinks, of which 55% were sugar-sweetened and 45% were low-calorie.

Consumption of low-calorie soft drinks fell by about half as children grew older and consumption of soft drinks containing sugar increased slightly.

The researchers also found a highly significant association between salt intake and total fluid consumption. They calculated a partial correlation coefficient of 0.33 and a beta value of 100 g/day (SD 7) of fluid consumption per gram per day of salt intake (P<0.001), after adjusting for age, sex, and body weight. When the researchers compared salt intake with sugar-sweetened soft drink consumption, they also found a significant correlation, with a coefficient of 0.12 and beta value of 27 (SD 5) (P<0.001), after adjusting for age, sex, and body weight. In turn, they added, the reduced soft-drink intake would translate to 61 fewer grams of sugar and 244 fewer kilocalories each week. Per-capita consumption of soft drinks is similar in Britain and the United States, the researchers said. Dr. He and colleagues said reductions of 10% to 20% in the salt content of processed foods “cannot be detected by the human salt taste receptors and do not cause any technological or safety problems either.” They said prospective studies and other research have demonstrated that changes in salt intake lead to altered beverage consumption, including sugary drinks. There are direct cardiovascular benefits to reduced salt intake as well, the researchers said.
They noted that, in Finland, systolic and diastolic blood pressure has declined by 10 mm Hg since the country began a national effort to reduce salt intake by its citizens in the 1970s.

Moreover, they said, stroke and coronary heart disease mortality have fallen by 75% to 80%, even though average body mass index and alcohol consumption have risen.

Great Britain now has a policy to reduce salt content of processed foods. However, Dr. He and colleagues said “some members of the food industry outside the U.K.” have resisted compliance, in part to protect soft drink sales.

“Some soft drink companies own large snack companies that specialize in highly salted snacks,” they said.

“However, they should not be allowed to stand in the way of a reduction in salt intake, because this reduction would have major benefits to the health of the whole population and, particularly, to children in potentially preventing the development of high blood pressure and obesity, thereby reducing the appalling burden of cardiovascular disease later in life,” the researchers concluded.

In an accompanying commentary, Myron H. Weinberger, M.D., of Indiana University Medical Center in Indianapolis, echoed the sentiment. He wrote that “prevention of obesity, rises in blood pressure, and the other precursors of cardiovascular disease must begin in childhood.”

He also agreed that reductions in salt and soft drink consumption among children could help avert health problems in adulthood.

But diet is not solely responsible for the rising prevalence of cardiovascular problems and obesity in children and adolescents, Dr. Weinberger said. Lack of physical activity and exercise play an important role as well.

He also noted that, in the U.S. at least, making foods healthier is a major challenge. “Motivating food preparers and commercial sources of food and drink to reduce salt and calorie content of their foods has been very challenging and relatively ineffective,” he said.

He called the conclusions of Dr. He and colleagues “provocative but largely speculative based on currently available evidence.”