Some studies show that drinking coffee everyday can reduce your risk of cancer. Others suggest that it will increase it. There are scientific studies that seem to back up both. It’s the same story for all kinds of food. In fact, there’s a great 2013 meta-analysis that went through a cookbook. The researchers found that of the first 50 ingredients mentioned, 40 of them had at least one study linking it to either an increased or decreased risk of cancer. How is this possible?
As it turns out, the world of nutritional science is much messier and more complicated than it seems. The gold standard for testing treatments or medications on humans is the double blind controlled study. This involves:
- Gathering a large number of people with diverse backgrounds
- Randomly splitting them into two groups
- Give one group the treatment or medication
- Give the other group a placebo
To be double blinded, neither the participants nor the researchers know who will be given the medication. (Obviously, that information is kept somewhere to be looked at after all the data is collected, or you wouldn’t gain any information.) A single blinded study consists of the researcher knowing who is given what, but the participants themselves don’t know. Both the researchers and participants know who has received what in an unblinded study. Blinding is important because both the researchers and participants can unconsciously change their behavior if they know what they want to happen. Participants generally want to please researchers, which can cause them to have a more positive reaction if they know they got the treatment. If researchers want their hypothesis to be true, they can accidentally affect the analysis of the data.
For obvious reasons, nutritional studies can’t be blinded. You can’t give someone a placebo hamburger, and someone else a real one. There have been studies where people stay in a facility and only eat what is given to them, but these method are very expensive and can only be done over a short term. It’s hard to do an experimental study of any kind outside of a lab, because when told to follow a particular diet, people tend to not obey the rules.
So, we mostly rely on what people self report that they eat over a long period of time, but that’s hard too. Do you remember what you ate yesterday, and how big the serving sizes were? I certainly don’t. When people do remember, they often aren’t 100% truthful. When logging my food into MyFitnessPal I know I tend to overestimate how big my salad was and underestimate how many cookies I ate. Even if people remember what they ate and logged it perfectly, just the act of tracking what you’re eating can cause you to change the way you eat. After all, that’s why people use calorie counters like MyFitnessPal.
Making the issue even more confusing are our old friends correlation and causation. Say that the data I’ve gathered from this survey shows that people who eat more fish have a lower risk of heart disease. Is it the fish itself that protects the heart, or are fish eaters more likely to spend time exercising? Maybe people tend to eat their salmon with veggies, meaning they get more vegetables overall than people who don’t like seafood. It can be very hard to figure out where the actual effect is coming from.
If this weren’t enough, here’s another confounding factor: the same kind of food can have wildly varying nutritional profiles. An apple can have between 50 and 100 calories, depending on the size and type. A hamburger you cook at home is going to be healthier than one you get from McDonalds. Finally, even if all of this could be accounted for, people still digest food differently. I’ve already talked a little about this, but basically, the same food can produce different responses in different people, which makes nutritional science very difficult.
All of this can make a person feel like there is no way to truly know anything about what makes up a healthy diet, which can lead to nutritional backlash. When there is so much contradictory evidence flying at your face, it’s easy to ignore all of it and eat whatever you want. But luckily there is some hope: reproducible studies can be good evidence that they’re on to something, as are different kinds of studies pointing toward the same result. There’re always systematic reviews; a study that combines all of the studies done on a subject. These methods are how we know that pregnant women need more folic acid and drinking too much soda can lead to diabetes.
There are some basics that (almost) all nutritional scientists agree on*. A healthy diet should be varied, with lots of fruits, veggies, whole grains, legumes, and nuts. Red meat, refined grains and refined sugar should be eaten in small amounts. Everything in moderation, as they say, nothing needs to be entirely cut out of a diet to be healthy.
Instead of worrying about how chocolate affects your risk of cancer, you should focus on lifestyle changes that have good evidence for avoiding cancer: wear sunscreen, don’t smoke, exercise, and drink alcohol in moderation.