Energy drinks and beer cause increased littering (Causation vs. Association)

(Disclaimer : all newsletter content is for interest and informational purposes only, and should not be considered medical advice. If you are interested in any topics please research further and discuss with your personal physician)

Views & opinions are personal, and do not necessarily reflect Swiss Re's.

"Follow a healthy diet" : I smile each time I hear or read that someone needs to follow a healthy diet. What does that even mean? And still harder to answer, how does anyone actually know which food is "healthy"?

Different types of nutrition evidence exist (and will be considered in a subsequent article), but by far the most commonly used are so called nutritional (observational) epidemiological studies.

?"By definition, epidemiology is the study (scientific, systematic, and data-driven) of the distribution (frequency, pattern) and determinants (causes, risk factors) of health-related states and events (not just diseases) in specified populations (neighborhood, school, city, state, country, global)." (CDC)

So what does this actually mean in terms of nutrition?

Typically these studies determine eating or nutritional patterns/quantities in many individuals by using dietary questionnaires (which themselves are difficult, very complex and open to misrepresentation), and then follow up these individuals for a number of years to determine health outcomes. These could be changes in weight, cardiovascular risks, events (e.g. heart attacks), diagnoses (e.g. cancer), or mortality. If individuals eating more (or less) of a certain food, and have a significantly different health outcome, then an association is made between the food (type, pattern etc.) and the health outcome. So imagine a made up example - people eating more popcorn have statistically more heart attacks than those eating less or no popcorn. There is a clear association between eating more popcorn, and having more heart attacks.

Does this mean (eating) popcorn causes heart attacks?

No. (Phew – for those eating popcorn)

Association does NOT imply causation. Why? Because statistical confounders can never be completely removed. Health or unhealthy user effect means that the health outcome may be due to some other unknown health or behavioural determinant and NOT due to the associated determinant or cause.

Two examples are given below:

?·????????Ice cream sales are associated with increased drownings. Clearly this is not causal. Sunny days means more people buy and eat ice cream, and as more people are swimming in the sea, the risk of drowning increases as well. Association sure. Causality clearly not.

·????????Ambulances are more often seen at the site of automobile accidents. Gosh, so ambulances cause accidents? Of course not, they are there because of the accident. Finding something somewhere associated with a disease, does not mean it is necessarily a cause of said disease (there is a great study on triglyceride lowering I will write about in future which really highlights this).

A further example, but with a twist. Behavioural differences associated with the determinant may be causal, but not the actual determinant being studied.

·????????Young adult smokers have an increased risk of accidental death compared to young adult non-smokers. Does smoking cause accidental death? No, but the young adults who smoke are more risk tolerant. They continue to use a product they know is very harmful, with many warning labels. They (generally, on average) take more risks than non-smokers, and in aggregate are more likely to be involved in dangerous or risky situations e.g. dangerous driving. While smoking per se is not causing the accidents, the behaviours/personalities related to those who do smoke lead to more accidents.

?

Then my homegrown example. Every few months I walk along a beautiful stretch of road leading out of our village and pick up litter. Being in Switzerland, litter is a rare thing, and it irks me to see it in such a beautiful countryside. 80% of what I pick up are bottles and cans, and easily 80% of those are various brands of energy drinks and beer. So I went to my youngest daughter and told her what I had found, and said that its clear energy drinks and beer cause people to litter. She looked at me, thought about it, and immediately said that it wasn't that they caused littering, but that the type of people who drank those were perhaps behaviourally not that concerned with littering. Younger, trying to show-off males (probably) racing along above the speed limit (and possibly smoking). Again a really beautiful example of (un)healthy user effect. It has nothing to do with the drinks, but it is the associated behaviour of (some of) the people who drink energy drinks and beer that leads to the outcome (littering).

?

So how is this relevant in nutritional studies? Well, most people have been told similar things as to what is healthy (salad for example) and what is not healthy (donuts and French fries). So the people that eat salad every day may reasonably be assumed to be trying to be healthy, and those eating donuts and French fries every day, may reasonably be assumed to not really care all that much about their health. The reason this matters, is the donut/French fry eating group likely does other unhealthy things, and the salad munchers probably do a bunch of other healthy things. And so when a study simplistically shows (an association) that donut eating has worse outcomes compared to salad eating, it is definitely NOT causal, as how can one know it’s the donuts or salad that are bad or good, rather than the other health behaviors of the donut vs. salad eaters.

This is the(un) healthy user effect which, particularly in nutritional epidemiology, makes it nigh impossible to attribute causation.

Sir Bradford Hill in 1965 gave the President's Address at the Royal Society of Medicine titled "The Environment and Disease: Association or Causation". His 9 Bradford Hill criteria have become a cornerstone for epidemiologists. While he was not talking specifically about nutrition, here is an excerpt from his introduction:

?

?A post I did some time ago, had such a great example of the healthy user effect - shown wonderfully in data. Associations of habitual fish oil supplementation with cardiovascular outcomes and all cause mortality: evidence from a large population based cohort study | The BMJ

The outcome of the study is not relevant (its association only right!). What is relevant is the difference in fruit and vegetable consumption, the amount of physical activity, and the % taking mineral and other dietary supplementation.

These are not the same people.

And as much as we can try to adjust for these "known" confounders or factors, we can never adjust for everything. And so a nutritional association remains just that – an association.?