The “Body” of Research About Scary Diet Drinks

Just in time for Halloween, JACC has published another scary study about diet drinks. The study finds an association with cardiovascular disease. It’s great fuel for the health news machine, but not so helpful for informing people about real health risks.

CNN says this is part of a “growing body of research” suggesting something scary about these drinks. But so long as this “body of research” is just a string of redundant correlation studies, it’s more of a zombie tale than a real live problem.

“What Is It About These Diet Drinks?”

CNN quotes Yasmin Mossavar-Rahmani wondering why diet drinks have this scary association. He did an earlier study with a similar finding:

“Is it something about the sweeteners? Are they doing something to our gut health and metabolism? These are questions we need answered.”

In other words, there must be something wrong with these beverages. The bottom line here is that diet drinks arouse strong emotions. Humans have been drinking sweet beverages for thousands of years. And all the while, scolds have taken the opportunity to opine about the moral hazards those sweet beverages present. Sugar, after all, is the starting point for alcohol.

Thus, diet drinks seem like cheating. Something sweet that doesn’t have sugar or calories? It’s got to be wrong. Cardiologist Andrew Freeman explains that there are unresolved questions about causality, but still, he tells his patients that unsweetened beverages are the only correct choice:

“I tell them that the perfect beverage for human consumption remains water, probably always will be. And maybe with a very close second of unsweetened tea and unsweetened coffee. And the rest probably should not be consumed regularly – if at all.”

A Modest Correlation

If you look at the data in this study, you will see that the correlation they’ve found is not especially striking. For high consumers of diet drinks they found a hazard ratio of 1.32 – a 32 percent increase in cardiovascular disease risk. But the 95 percent confidence interval was 1.00 to 1.73. Ordinarily, one might conclude that this finding was not significant because of that.

However the authors, without specifying their statistical methods, say that the trend for increasing risk with increasing consumption is significant, with a p-value of 0.03. Perhaps this was a pre-specified analysis plan, though the authors never say so. Thus, they leave us unimpressed.

Beating a Correlation to Death

Really, these redundant correlation studies do a great disservice to the reputation of nutritional epidemiology. No matter how many times you repeat the study, finding a correlation does not prove causality. It only builds a bias of familiarity.

Click here for the study and here for more from CNN.

Zombie Halloween Cupcakes, photograph © jamieanne / flickr

Subscribe by email to follow the accumulating evidence and observations that shape our view of health, obesity, and policy.

October 27, 2020