In the last post, I introduced the different levels of research, experiments, and evidence in nutrition science. So when we hear about some nutrition related information, it is a good idea to first judge the quality of the evidence being provided. However, even if we follow this step, we may still misunderstand a few things while interpreting the results of a study. What if there is contradictory evidence out there? Is it more important to eliminate unhealthy foods or to add healthy foods? How do we find out if someone making claims on social media about nutrition is spreading misinformation? In this post, I will explain common misconceptions about nutrition and strategies to prevent them.
Does the current evidence agree?
In order to fully answer a question related to nutrition, we should not just focus on one study or one type of evidence. If we searched hard enough, we could find studies that support almost any opinion. For example, there are studies that claim that smoking cigarettes in certain circumstances is good for your health.1 So if we just focus on these studies and ignore the other evidence, we might be misled. Different types of experiments have different advantages and disadvantages, so the best approach is to analyse all of the evidence together. Does a wide range of different types of evidence point to similar results, trends, or conclusions? If we achieve similar results through different methods, we can be more confident about those results. Meta-analyses and systematic reviews are the most reliable pieces of evidence, but they are not perfect and their conclusions should not automatically be trusted. It is better to consider all of the evidence together.
One of the disadvantages of randomised controlled trials is that we cannot control participants with in a study for long periods of time due to both ethical and economic reasons. That is why these studies are not suitable for showing the long-term effects of diet. However, the advantage of randomised control trials is that we can better control the confounding variables, making it easier to show cause and effect. In contrast, the disadvantage of epidemiology is that it is less effective at showing cause and effect than controlled trials. However, the advantage of epidemiology is that we can we collect results over a long period of time, look at much larger populations, and can show the long-term effects of diet.
We can also propose hypotheses about which foods or nutrients cause a health outcome of interest through mechanistic studies, but we cannot prove that these variables have a large effect on human health as a whole. The body is very complex, and there are many chemical and biological processes that occur simultaneously. So, no matter what kind of food we eat, and what kind of effect one mechanism may have, other mechanisms may have an opposite or stronger effect. For example, a food may raise cholesterol to a certain extent, but at the same time, other mechanisms arising as a result of eating that food can produce other positive effects. If we focus only on a small increase in cholesterol, we may judge the food incorrectly. That is why the overall effect is most important.
The body is similar to a black box. If you put an input inside the black box, an output will come out. In the case of nutrition, the black box is all of the biological mechanisms within our body, the input is the food we eat, and the output is the health outcome of interest. The schematic below shows this black box model. We can try to understand the processes that occur within that black box through mechanistic studies. This information is useful for us to increase our understanding, and to develop more efficient treatments and medicines. Mechanistic evidence can provide support for the degree of cause and effect in a correlation found in other studies.2 However, as you can see in the schematic, the food we can lead to many different outcomes due to many different mechanisms. Some mechanisms lead to an overall health outcome of interest, but many do not. Overall health outcomes are the most important and relevant piece of information to us when making food choices. We can get this kind of information from randomised controlled trials and prospective cohort studies.
If a mechanism’s affect agrees with the results of prospective cohort studies and randomised control trials, then our confidence about the importance of that mechanism may increase. But even then, we should not fully believe our hypothesis. On the other hand, even if a food initiates a seemly dangerous mechanism, if the effect of that mechanism disagrees with other higher quality studies, the importance of that mechanism is probably not very significant. If the only evidence we have about a topic is mechanistic, we should carefully make decisions based on that evidence. However, if there is mechanistic evidence and higher quality evidence available, and they disagree, then we should be more suspicious of the mechanistic evidence.
The Buddhist teaching about the blind men and the elephant illustrates the problem of focusing only on mechanistic evidence. I just want to point out that this blog is not religious in nature, but this teaching provides a relevant metaphor. The gist of the teaching is that several blind men who have never seen an elephant before try to describe what an elephant is by only touching its body. One blind man grabs the elephant’s nose and concludes that an elephant is like a snake. Another other blind man grabs the elephant’s leg and concludes that an elephant is like a tree. These blind men were wrong about the overall nature of an elephant because they only has information about one part. Similarly, if we focus only on a few mechanisms and ignore the overall system and health outcomes, we will have limited knowledge. Nutrition is like a puzzle, and different types of evidence are the pieces. If we focus on just one piece, we cannot see the whole picture. If we focus on where they overlap, we will be able to gradually solve the puzzle piece by piece.
In the absence of scientific evidence on a topic, we have to make a best guess or make mechanism based decisions. In that case, a cautious approach is usually good, and we should have low confidence in our decision. However, if there is scientific evidence for a subject, our assumptions, thoughts, opinions, or hypotheses about the subject cannot replace that evidence. We should always strive to make decisions based on high quality evidence.
Can we dismiss epidemiology?
Some argue that because epidemiology has flaws, the results of that research should be discarded. However, stating that a study has flaws is automatically correct, no matter the study. All studies are flawed. However, we should not disregard entire studies just because this is true. Instead, we have to figure out what important information we can gather, while taking the shortcomings and other forms of evidence into account. In fact, if we ignored all epidemiological studies, we would not have enough information to create nutrition guidelines. High-quality epidemiologic research is a necessity for public health.
Should we wait for the publication of a perfect randomised controlled trial on a massive number of participants living in a controlled environment from birth to death before we make decisions about nutrition? In an ideal world, we would have that information. But perhaps that world would not be so ideal. To do this, many people’s lives would have to be controlled. So, this would be both unethical and impractical. It would also be unethical to force people to eat or do something that we suspect is unhealthy. Randomised controlled trials tend to look more and more like epidemiological studies as the study duration increases, and the subjects become less likely to follow the experiment’s prescribed diet. In fact, epidemiological studies are the reason we know that smoking causes negative health outcomes.3 We did not randomly assign participants to smoke a pack of cigarettes every day for 20 years. Only epidemiology can provide this kind of information by observing existing smokers.
The healthy user bias is potential issue related to epidemiological research. People who want to participate in research are probably healthier than the general population. These people seem to be more likely to have an interest in health, have healthier habits, and have a higher socioeconomic status. So then, is epidemiological research garbage? Researchers can reduce the effective of this bias by designing experiments well, including participants from similar populations, or looking at sufficiently large populations. Also, if we look at with different populations and observe similar trends, we can be more confident in the results.
Similarly, are vegans generally less likely to develop a disease simply because they tend to have a higher socioeconomic status, care more about their health, or exercise more? Is it possible that people who eat meat are more likely to develop disease simply because they are more likely to eat more processed foods, exercise less, smoke more, or be less concerned about their health? Well, researchers are well aware of this potential problem. In order to control for these confounding variables, researchers statistically adjust the final result. If, after adjusting for all relevant confounding variables, the food still statistically significantly affects the health outcome, then our confidence that the correlation we observed is causal increases. Of course, it is impossible to control for all variables, and we do not know every single variable involved, but researchers are always improving and refining their results. The effect of any confounding variables that scientists have not yet considered are likely to be small. An epidemiological study is of lower quality if the results are not adjusted for confounding variables as described.
To assess the quantity of food eaten by participants, researchers will usually use a food frequency questionnaire. But since an individual may not provide answers that are representative of their typical diet, how can we trust these surveys? Scientists are well aware of this issue and first validate the survey. For example they weigh the quantity food eaten by several participants. If the survey results and weight measurements do not match, the researcher adjusts the survey until they get it right. Furthermore, these days good epidemiological research tests the blood work of some participants to see if the nutrients measured in the blood match the reported food. Blood doesn’t lie. Prospective cohort studies usually look at large populations, so weighing everyone’s food is too expensive and impractical. So, although it is a little less accurate to use a validated survey, it would be near impossible to observe a large population without this method. We can get still extract a lot of useful information from this research. An epidemiological study is of lower quality if they do not validate their questionnaires as described.
Reverse causation is a phenomenon in which an assumed effect is actually the cause of the assumed cause. For example, if people with really low cholesterol are more likely to die, should we conclude that having a low cholesterol is dangerous? In reality, this correlation can occur because cancer and other diseases lower cholesterol, and people with these diseases are more likely to die.4 It is not the low cholesterol that is causing them to die. So we have to interpret these results with caution and take into account the direction of the causality. High-quality prospective cohort studies will address these misconceptions. Overall, a focus on high-quality prospective cohort studies can reduce the common biases and shortcomings of epidemiological research and can provide useful information. We cannot put all of epidemiology into the same basket.
Replacement
If we look further at the results and methods of seemingly contradictory studies, we can begin to see that there is in fact no contradiction. However, if we only read the title or the abstract of a study, we cannot easily see this. In the next sections, I will talk about these nuances.
In order to examine the effects of a food, participants usually have to replace it with another food. The results may occur due to the fact that a food was removed or that the replacement food was added. So both the replacement food and the replaced food are important. Because of this, the comparisons made in studies are important. Even if some foods are relatively healthier than others, we cannot claim that they are absolutely healthy. For example, if a hypothetical study suggests that people who eat a lot of red meat are relatively healthier compared to people who eat foods high in processed carbohydrates as a replacement, do these results prove that red meat is a healthy food? Well actually no. We can only conclude from these results that consuming that quantity of red meat is healthier than consuming that quantity of processed carbohydrates. However, another hypothetical study may find that people who eat a lot of red meat are less healthy compared to those who eat more whole grains or legumes as a replacement. This result may occur due to a combination of the participants eating less of certain nutrients in red meat and eating more of certain nutrients in the whole grains or legumes. From these results, we could conclude that the that quantity of whole grains or legumes is healthier than that quantity of red meat. As we interpret the results of a study, we must consider its substitutions and can only draw conclusions that are relevant to the context in which the experiment was conducted.
Dose
Studies usually compare people who consume a high dose of a certain food with people who consume a low dose. But what are high doses and low doses? In one population, some doses may be relatively high, while in another population, that same dose may be relatively low. For example, if you look at a population in Asia, a high dose of soy may be a lot larger than a high dose of soy in the United States. Hence, a study from Asia may conclude that soy consumption has a significant impact on health. However, studies in the United States may not show this effect. So, if we compare studies, we have to compare the investigated dosages. We also have to take into account the overall diet of each population. Some populations may already eat a relatively healthy diet, so including some unhealthy foods may not have as much of a significant impact on their health.
If we don’t take the dosage range into account, we may inadvertently draw incorrect conclusions. For example, let’s consider a hypothetical study that shows that eating less than 50 grams of candy daily does not increase the risk of developing a disease. Let’s also consider another hypothetical study that compares people who eat 40 grams of candy per day with those who eat 10 grams of candy per day. If the probability of developing a disease is the same between those two groups, should we then conclude that eating more candy does not cause disease? That conclusion is only correct for the tested dosage range. However, this conclusion is not representative of all doses. If another study compared people that eat 100 grams of candy per day to those who eat 40 grams of candy per day, and the probability of developing a disease is higher, we can then conclude that eating more than 40 grams of candy can increase risk of a disease. Sometimes, in order to observe a change in the probability of developing a disease, the dosage must be above a certain threshold. This dosage threshold depends on the food. A study may not observe any difference between groups if participants who eat below that threshold and participants who eat above that threshold are not investigated at the same time. If we are not aware of this, we can mistakenly make conclusions that are too general.
In addition, the relationship between dose and results are not always linear and can be S-shaped, exponential, logarithmic, or anything in between. Increasing the intake of a food by 10% does not always change in the probability of developing a health outcome by 10%. We can get a better sense of the overall effects of a food by combining the different results from different tested dosage ranges.
Similarly, it does not matter if a food simply contains a certain “good” or “bad” nutrient or not. The dosage of that nutrient is what is important. Even if a food contains unhealthy nutrients, the food can still be good for us overall because of a variety of other nutrients and mechanisms. That is why it is most important to look at the overall effects of a food. For example, even if walnuts have saturated fat, the amount of saturated fat is low, and the amount of unsaturated fat is relatively high. When we look at the effects of walnuts on health, we can conclude that they are generally healthy.5 However, if we had the attitude that we absolutely want to avoid saturated fats, we would miss out on the benefits of walnuts. Similarly, we need oxygen to survive. However, if we breathe in too much oxygen, it can be poisonous. An appropriate and balanced dose is the most important. For some foods, the appropriate dose is high and for others, the appropriate dose is low.
Should we focus on nutrients, foods, or our overall diet?
Nuritionism is an approach that focuses on the nutrients within a food instead of considering a food as a whole. If we just eat the good nutrients and not the bad ones, will our health definitely improve? This approach is a bit too reductionist. In fact, the nutrients in food can interact and affect our bodies differently compared to eating each nutrients in isolation. Nutrients can synergise or oppose each other.
In addition, if a study were to test the health effect of only one food at a time, we would generally only see a small effect. However, if a study tests the health effects of a whole dietary pattern, we are likely to see larger effects. While it is important to know about healthy nutrients and health foods, the most important thing is to focus on eating an overall health dietary pattern. A healthy diet can include certain doses of “unhealthy” nutrients or foods. We do not need to completely eliminate nutrients and foods that are somewhat unhealthy in order to be healthy. As I mentioned in the last section, the dose is most important. Instead, studies that experiment with diets as a whole offer more general and useful healthy eating principles.
What if I don’t have time to check the quality of nutrition information?
I get it. We are all busy and nutrition can be complicated, so we may not have time to read a lot or become an expert before making dietary decisions. That is why listening to dietary guidelines is a good first step. There are also evidence-based podcasts out there that you can listen to while commuting. There are also evidence-based blogs, YouTube channels, and websites. As mentioned in the previous post, this kind of media is not high quality primary evidence itself, but if we can find media that consistently cites high-quality evidence, is nuanced, and utilises the principles mentioned in this post, then we are on the right path. I recommend the podcasts and videos by The Proof, Nutrition Made Simple, and Sigma Nutrition as a good starting point.
Common characteristics of people who spread low-quality information
Below I summarise the common approaches of people who spread misinformation:
- Anti-science. Do they oppose science as a whole? Science is the best kind of proof we have. Without science, we just have anecdotes, opinions, and stories.
- No sources. If someone does not provide sources to support their claims, that is a red flag. A person’s thoughts are not evidence.
- Providing low-quality evidence. If a person includes a source, we can judge the quality of their source using the evidence hierarchy pyramid. Citing scientific sources does not automatically mean that the evidence provided is useful. Do they focus only on low quality mechanistic evidence? When a person ignores higher quality evidence, that is a red flag.
- Not comparing all of the evidence. Does the person compare their sources to other sources? Do they provide different types of evidence-based information? Do they focus on only on studies that support their opinion? If so, they may be a victim of confirmation bias. Whether a person does this or not, we should verify this information ourselves using different kinds of evidence. When discussing the quality of a study, a trustworthy source will usually discuss why there is contradictory evidence.
- Focusing on old research. Does the person focus mainly on studies that were published a long time ago? If a person has to focus on really old studies to support an opinion, that is a red flag. In some cases, studies published in the past are still valid and relevant in the present. In many cases, however, there are many more recent studies that build on older studies. In particular, if other more recent studies could not replicate the results or oppose the results of the original study, it may be misleading to focus on the older study. Current evidence likely removes or improves upon the flaws and understanding provided by older research.
- Focusing on researchers instead of the research. Does the person focus on the author of a study? Attacking the person making a claim rather than talking about the claim itself is the classic ad-hominem fallacy. This strategy can distract us from the real problem. There are corrupt or evil scientists out there. But this does not prove that their work is automatically wrong. A scientist’s personal life or personality is irrelevant and we should solely focus on the quality of the study.
- Not considering the details. If a person only focuses on the abstract of a journal article, they may miss out on more specific information, the discussion of the nuances and flaws of the study, and the limitations of the results contained within the article.
- Absoluteness. A wise man once said, “Only a sith deals in absolutes” (sorry to those who don’t like Star Wars). Does the person make absolute claims? For example, if a person claims that any dose of food or nutrient in any population is bad, that is a red flag. Or, if they claim that a certain food will solve all of our problems, that is also a red flag. When talking about nutrition, there is almost no absolute advice out there. There are almost always nuances. However, the media and people tend to not like uncertainty and want simple solutions, explanations, and something to blame for a problem, so they try to find absolute information. Many media capitalise on this by sending incorrect absolute messages in order to increase views. People may prefer incorrect absolute messages instead of reality, which is uncertain and nuanced.
- Certainty. Does the person use language such as “probably” and “maybe” or do they use “certainly” an “will”? A person who is thinking like a scientist would be aware of the uncertainty regarding what they are talking about and will admit what we don’t know. Scientific sources can distinguish between hypotheses and reality.
- Too confident. Is the person too confident about what they are claiming? If a person is confident, that means that they know what they are talking about, right? Not quite. Someone can state information with confidence, but at the same time, what they are saying may be completely wrong. Does the person challenge their beliefs? We have to accept that we all have the ability to spread misinformation. If a person does not accept this and is too confident, they may be more likely to send a biased message. We should continually challenge our beliefs and should not avoid or ignore those who have other beliefs. Instead, if we listen to their perspective and discuss it, we can get closer to the truth.
- Blaming “them”. If a person blames one group or one profession for all of our problems, that is a red flag. For example, it is common for people to blame doctors, the food industry, farmers, the rich, the government, or any other authority figure or group for all of our problems without evidence. Usually this involves conspiracy theories. It is interesting to think about conspiracy theories, and it is important to question everything. However, being suspicious of everyone without evidence is a dangerous mindset. In many cases, people do not have ill intentions. Although it is often easier to understand a complex problem when we reduce it down to a set of simple causes, this is often not representative of the complex and nuanced nature of reality.
- Stubbornness. If a person makes a mistake, do they admit it and correct it? Do they change their opinion if provided with high quality opposing evidence? If they do, the they demonstrate that they value maintaining the facts. Whether we are a doctor, a researcher, or anyone else, we all have the ability to be wrong, and we need to be able to change our opinions after learning about new high quality information. If our ego is too big to admit this, we may continue to spread misinformation.
- Bias. Does the person have biases or conflicts of interest? Many influencers or people who work in the health field have made a career out of spreading misinformation regarding health. We are all biased, and this in and of itself is not a reason to not trust a person. However, a person is a bit suspicious if they do not admit that they have biases or conflicts of interest. For example, my bias is that I am vegan and I wish for a world where we do not exploit animals. I am aware of this bias, so when I talk about nutrition I endeavour to interpret the evidence objectively. If there is evidence that shows that in a certain context an animal product is healthy or sustainable, I am happy to admit that.
Even well-educated people can sometimes take these irrational approaches. Some vegans are also guilty of using these kinds of strategies. We are all human beings, and we can all spread misinformation unintentionally or deliberately. The most important thing is our attitude. If we have a desire to continuously learn the facts, we will approach the truth. We should also view content from people who do not align with our lifestyle or opinions. Even if the information is wrong, we should at least confirm that it is wrong and not automatically dismiss it. If we don’t have an open mind, we won’t have the opportunity to learn. Even if you don’t agree with another other person, at least we will be able to understand their point of view better.
Conclusion
Below are the most important steps to avoid nutrition misinformation:
- Assess the quality of the evidence provided.
- Check that a source of information agrees with other types of high quality evidence.
- Check the food replacements and comparisons used in the study.
- Check the population and doses tested in the study.
- Instead of focusing on one food or nutrient, focus on entire dietary patterns.
- Check that the person making a claim has not demonstrated any of the red flags mentioned above.
The concepts in this post can also be used areas outside of nutrition. The most important thing is to keep an open mind and to always try to learn more. Could your current beliefs be wrong?
References
1. Wolf R, Orion E, Matz H, Maitra S, Rowland-Payne C. Smoking can be good for you. Journal of Cosmetic Dermatology. 2004;3(2):107-111.
2. Clarke B, Gillies D, Illari P, Russo F, Williamson J. Mechanisms and the evidence hierarchy. Topoi. 2014;33:339-360.
3. Fagerstrom K. The Epidemiology of Smoking. Drugs. 2002;62:1-9.
4. Fiorenza AM, Branchi A, Cardenà A, Molgora M, Rovellini A, Sommariva D. Serum cholesterol levels in patients with cancer. Relationship with nutritional status. International journal of clinical & laboratory research. 1996;26(1):37-42.
5. Rock CL, Flatt SW, Barkai H-S, Pakiz B, Heath DD. Walnut consumption in a weight reduction intervention: effects on body weight, biological measures, blood pressure and satiety. Nutrition Journal. 2017;16(1).