Soda-Heart Attack Study Uses Fizzy Math

Related articles

Screen Shot 2015-09-01 at 3.17.52 PM"Carbonated drinks linked with out-of-hospital cardiac arrest of cardiac origin"

This must be a joke right? It looks like something from The Onion but it's no parody. Professor Keijiro Saku, Dean and professor of cardiology at Fukuoka University in Japan presented it at the European Society of Cardiology.

It will become obvious (below), that this conclusion required a whole lot of data dredging and a lot more hand waving. Who would buy such nonsense? Well, The Washington Post did, and that is why you should be reading the American Council on Science and Health instead.

For analysis, we turned to Dr. Stan Young, a member of the Council's Board of Advisors who is, among other things, an expert in statistical analysis of clinical trials. Dr. Young and I took the paper apart and here are our comments on some of the most egregious errors and biases in the paper.

Study: "Some epidemiological studies have shown a positive correlation between the consumption of soft drinks and the incidence of cardiovascular disease (CVD) and stroke"

Dr. Young: "Once a person has a stroke, there can be examination of hundreds of foods and other things, exercise, sex, air pollution, etc. It can be a grand data dredge. If you find nothing, you don t write a paper. If you find anything and have nothing better to do you write a paper."

Study: "[O]ther reports have demonstrated that the intake of green tea and coffee reduced the risk and mortality of CVD."

Dr. Young: "This is not new for Saku, here's another of his pieces: 'Associations between the consumption of different kinds of seafood and out-of-hospital cardiac arrests of cardiac origin in Japan.' Suematsu Y, et. al... So, you get the picture. He is testing possibly hundreds of things and writing a paper each time he gets a statistically significant finding. Likely all of them are statistical multiple testing false positives. He (or someone with that name) appears to have over 400 papers."

Study: "Carbonated beverages, or sodas, have frequently been demonstrated to increase the risk of metabolic syndrome and CVD, such as subclinical cardiac remodeling and stroke. However, until now the association between drinking large amounts of carbonated beverages and fatal CVD, or out-of-hospital cardiac arrests (OHCA) of cardiac origin, was unclear...The study compared the age-adjusted incidence of OHCAs to the consumption of various beverages per person between 2005 and 2011 in the 47 prefectures of Japan. It included 797 422 patients..."

Dr. Young: "So he has access to a very large data base and he asks one question after another."

Study: "The analysis focused on the 785 591 OHCA cases that received resuscitation, of which 435 064 (55.4%) were of cardiac origin and 350 527 (44.6%) were of non-cardiac origin. Those of non-cardiac origin included cerebrovascular disease, respiratory disease, malignant tumour, and exogenous disease (4.8%, 6.1%, 3.5%, and 18.9%, respectively)."

Dr. Young: "OK, now he is slicing and dicing (see: Walt Frazier). There are multiple ways to look at any particular finding. He could try males and females separately, for example."

Study: "The researchers found that expenditures on carbonated beverages were significantly associated with OHCAs of cardiac (r=0.30, p=0.04) (Figure 1), but not non-cardiac origin (r=-0.03, p=0.8)."

Dr. Young: "So here he asks at least two questions, so his p=0.04 should be doubled. It becomes non- significant."

Study: "Expenditures on other beverages, including green tea, black tea, coffee, cocoa, fruit or vegetable juice, fermented milk beverage, milk and mineral water were not significantly associated with OHCAs of cardiac origin."

Dr. Young: "We can start to count the number of questions. Nine are listed."

Study: "Carbonated beverage consumption was significantly and positively associated with OHCAs of cardiac origin in Japan, indicating that beverage habits may have an impact on fatal CVD. The acid in carbonated beverages might play an important role in this association."

Dr. Bloom: "Now he's at the special level of logic we see Dr. Oz use so often. Acid?? He must be kidding. How about vinegar? It has a pH of 2.4, which is rather acidic. The pH of 25 different sodas has been measured. ZERO of them are more acidic than vinegar, which you wouldn't think twice about putting on your salad. Here's a list of sodas that are significantly less acidic than vinegar: Diet Sprite (8-times less acidic than vinegar), Diet Coke (17-times), A&W Root Beer (223-times). He mentions that fruit juices were non-significant but those, like orange, are significantly more acidic than soda. Lemons are about as acidic as vinegar, and limes are about 1.5-times more so. If Saku's theory about the acidity of soda causing cardiovascular disease is correct (it isn't), you better get rid of your lemon, limes, and vinegar, and fast, or your heart will explode."

Study: "Our data on carbonated beverage consumption is based on expenditure and the association with OHCA is not causal. But the findings do indicate that limiting consumption of carbonated beverages could be beneficial for health."

Dr. Young: "After scaring everyone, he throws in a 'non-causal' statement, but says you should still be scared."

Finally, it would be wrong not to show that "best" part of the study. No, you do not need new glasses:

Screen Shot 2015-09-01 at 3.06.59 PM

Just to show you how stupid this is, here is a graph that is an example of a nearly perfect correlation of two events.

Screen Shot 2015-09-01 at 1.09.42 PM

Yep there is a perfect correlation (r= 0.997) between sales of organic food and increasing autism. Do they have anything to do with each other? Of course not. But at least the numbers are good. This is far more than what can be said about this study.

So enjoy that carbonated beverage. It is not causing a heart attack.