Yesterday, as I was standing in line at my local grocery store, waiting to pay for some fruits and veggies, my gaze unconsciously fell upon the headlines on the front page of yesterday’s newspapers, which were conveniently placed directly beside the counter of the store. One of the headlines in particular grabbed my attention. It said that new research shows that the consumption of alcohol is beneficial in that it helps reduce type-2 diabetes risk.
If you’ve been paying attention, you’ve undoubtedly noticed that new studies investigating the health effects of alcohol consumption are published “all the time”. You’ve probably also noticed that the results of the research in this area are very heterogeneous, in the sense that there is no clear pattern as to the direction in which they point. Some studies seem to indicate that it’s beneficial to consume alcohol; others suggest that alcoholic beverages such as wine are neutral, in the sense that they don’t do any harm, but not much good either; whereas others again have found a link between alcohol consumption and various adverse health outcomes. If you drop by ScienceDaily.com and proceed to do a search for alcohol, you can clearly observe these discrepancies.
All of this conflicting information can make anyone confused. In today’s article I thought I’d briefly try to explain why the studies in this area are very incongruous with respects to their findings, as well as why you should think twice before you believe what newspapers have to say about what “the latest research” investigating the link between alcohol consumption and human health shows.
Nutritional research is complicated
The conduction of a nutritional study is not a simple, straight-forward process. Confounding variables, biases, type-1 errors, chance effects, and compliance-related issues are just some of the many things scientists have to consider when they carry out a nutritional study.
It’s not just scientists who should know about these things. Everyone who’s somehow involved in research should. Unfortunately though, this isn’t the case. Many journalists and other people who bring scientific research out into the public know little to nothing about how the scientific process works. They don’t know why one should be very cautious about drawing causal conclusions from observational studies, what a meta-analysis is, or what separates randomized controlled trials from other studies. This is problematic, because it’s very difficult to correctly interpret scientific research if one doesn’t know anything about research design.
From a research design perspective, it’s generally a lot more complicated to conduct a nutritional study than a drug study. In a drug study, one can simply separate the study participants into two groups and proceed to give the participants in one group capsules containing the drug and the participants in the other group placebo pills. This is not something that one can do in a nutritional study, unless the goal is to investigate how a dietary supplement (e.g., vitamin pills) of some sort affects human health.
Unlike drugs, foods contain a lot of energy in the form of calories. Moreover, bananas, apples, nuts, and so forth are not designed by man, but rather by nature. They aren’t composed of just one or a couple of substances; rather, they contain a mix of many different nutrients, phytochemicals, vitamins, and so forth, with different food products containing different quantities of each. Finally, we humans eat a great variety of different foods. We don’t just eat apples or bananas. If you add one thing to a person’s diet, you also have to take something out, or else the person will suddenly be taking in more calories than he used to.
It’s not difficult to understand that this can make nutritional research complicated.
Correlation doesn’t equal causation
Many of the studies that have looked into the health effects of alcohol consumption are observational studies. It’s not difficult to understand why this is the case. Diseases such as type-2 diabetes, heart disease, and osteoporosis take a long time to develop. They don’t come into existence over night. Hence, in order to elucidate how different dietary practices affect our risk of developing these diseases, one has to be patient and conduct a long study.
In a randomized controlled trial lasting a couple of months, one can assess various disease-related biomarkers; however, it can be difficult to get some useful data pertaining to hard end points. This is particularly true if the purpose of the study is to investigate how people are affected by consuming small-moderate quantities of something like wine or meat, seeing as the health effects induced by such a behavior is not going to be acute or dramatic.
One of the major limitations of observational studies is that they are very prone to confounding. Not only that, but the investigators typically have very little control over their participants. The participants in observational studies are out in the world living their lives; they aren’t required to meet up at a medical clinic every week to get instructions regarding their diet/lifestyle or take in a set amount of a specific food every day. This largely helps explain why the results of different observational studies that look into similar phenomenons can be all over the place, in the sense that the findings don’t overlap.
Let’s say that you, with the help of a group of experienced researchers, set out to do an observational study in order to investigate the link between alcohol consumption and chronic disease risk. You fix your gaze on a large, specific group of people who you follow for 3 years. During that time, you keep track of the participants’ diet, lifestyle, and health status. You have them fill out various questionnaires along the way and also make a note of any incidents related to the three following chronic diseases: type-2 diabetes, heart disease, and colon cancer.
At the end of the study, you and your fellow researchers carry out various statistical analyses in an attempt to elucidate if there’s a link between the participants’ consumption of alcohol and their risk of developing the aforementioned three chronic diseases. You find that the incidence of chronic disease was markedly lower among the participants who consumed little to no alcohol during the study period than those who consumed moderate-large quantities of alcohol. However, after you control for variables such as smoking and physical activity levels, you find that the result is only statistically significant for type-2 diabetes.
Does that mean that you can firmly conclude that it’s better to consume little to no alcohol than moderate-large quantities of alcohol? No…
There are many reasons why this is the case. First of all, it’s virtually impossible to stay on top of all covariates in a study such as this one. The participants who consumed no or little alcohol may differ in several respects from those who consumed a lot of alcohol. They may perhaps have been healthier at the onset of the study, spend more time in nature, or eat a healthier diet. This is something it can be very difficult to fully control for.
Second, the data that you collected via your questionnaires are not perfect. People who participate in research studies misreport and forget things all the time. Third, the fact that you found that the incidence of type-2 diabetes was lower among the participants in your study who consumed little to no alcohol than among those who consumed moderate-large quantities of alcohol doesn’t necessarily mean that it’s better, across the board, to consume little or no alcohol. For example, one can’t exclude the possibility that the incidence of another chronic disease, such as osteoporosis, was lower among the participants who regularly drank wine, beers, and/or other alcoholic beverages.
This is one of the major limitations of all scientific research: There’s a limit to how many outcomes one can investigate. Experimental studies such as randomized controlled trials are obviously “stronger” than observational studies; however, they are not bulletproof. Far from it.
This is why it’s so important to be careful when interpreting the results of scientific research. Moreover, it’s why one should always consider the evolutionary evidence before jumping to conclusions.
To avoid making this article excessively long, I won’t get into the question of whether it’s actually healthy to consume alcohol or not. That’s a topic for another day. What I would like to say though is that ethanol is not a compound that agrees well with the human body. This isn’t surprising, seeing as it’s only very recently – on an evolutionary time scale – that alcoholic beverages such as beer, wine, and vodka became a part of the human diet. Our Paleolithic ancestors may have occasionally consumed small quantities of ethanol in the form of fruits that had started undergoing natural fermentation processes; however, they obviously didn’t sit around a fire every night drinking beer. This is not to say that you should necessarily shun all alcoholic beverages like the plague; however, it is something to keep in the back of the mind the next time you’re traveling through your local grocery store, making decisions about what you are going to eat and drink for the coming days.