Healthy Lives

DEEP DIVES

Nutritional science is complex. Everything is connected to everything, and cause and effect are easily confused. Over the past decades, this has led to widespread misunderstanding and bad health advice.

Interpreting scientific information is tricky, and nutritional science especially. There are many factors that make nutritional research more or less useful. Scientists usually acknowledge these factors truthfully, but popular articles about science are rarely written by scientists and the nuances often get lost in translation. So be wary of the articles you read, and look out for these common pitfalls:

Correlation vs causation
This is perhaps the most common scientific research pitfall. When two things correlate to each other, our instinct is to assume one causes the other. In a field as complex as human nutrition, this is rarely the case. For example, while a high HDL level may be an indication for a healthy cardiovascular system, taking drugs to increase HDL does not improve your cardiovascular health. In reality, there are other factors that increase both your HDL and your cardiovascular health. HDL is just a marker for good health, it is not the cause of it.

Self-reporting
Many nutritional researches are based on data gathered by asking people to keep track of what they eat during a day. Or worse, by asking people to remember what they ate yesterday, last week or in extreme cases all of last year. This type of data gathering is notoriously unreliable, but this is seldom mentioned in the results (especially in the popular article about the results).

Relative percentages
When you read a headline like “animal fats cause a 20% increase in colon cancer” it’s important to realize that they usually refer to relative percentages. The average likelihood of anyone getting colon cancer during their lifetime is around 5%. This means that with a 20% increase, this 5% chance increases to 6%. In absolute terms, this would be a 1% increase, but that wouldn’t make such a great headline. For perspective, smoking increases your relative chance of lung cancer by 750-5000%.

Conclusions from extrapolation
Sometimes, conclusions are drawn about things that were not part of the research. As an example of this, let’s look at fiber research. A recent study (conducted over decades) found that people that ate large amounts of fiber were less likely to suffer from diabetes and heart disease. The researchers then went on to conclude that the current low carb trend is dangerous because low carb means low fiber. However, the research didn’t include any participants on a low carb diet. It just compared people that ate refined grains to people that ate whole grains (and consequently more fibers). But low carb diets are so different from both of these diets that extrapolating from these results is meaningless if not downright misleading.

Some tips to navigate scientific articles
Clearly not everything written about nutrition can be taken at face value. But how do you decide what’s true and what’s false? Here are some tips:

  • Instead of reading a one-sided article, try to find a discussion about the topic on the internet. That way, you can read both sides before making up your mind.
  • Be wary of anyone using some variation of “trust me, I’m a doctor”. If they can’t explain why they are correct, they are most likely wrong. This is especially true if rather than provide an argument, they attack the other side’s qualifications.
  • Be wary of scientists with a second agenda. For example, many nutritionists looking at meat consumption are concerned about animal welfare and the environmental impact of the meat industry. While those are legitimate concerns, they should not impact the conclusions of nutritional research.

Our Deep Dives feature current scientific insights on important health & fitness subjects in easy-to-read articles. As with all science, these are not set in stone; new insights may change our view on certain subjects. We will update our Deep Dives whenever convincing new evidence comes up.

  • Grassland indicating the end of the page

  • Grassland indicating the end of the page