Canada has just released its new nutritional guidelines and as with all the guidelines these are based on expert consensus and systematic review of the literature on nutrition and food intake. We've been studying nutrition for ages now, but, still, the majority of nutritional studies are inherently flawed and it's quite difficult to see the forest behind the trees, especially if you are not a researcher in this field. I always do my research when writing my blogs and preparing my YouTube videos and I try to base them on real, high-quality research. I must say, that as a researcher I struggle enormously with finding unbiased data. Today, I'd like to share my observations with you – what seems to be problematic with nutritional research and why we have to be careful when interpreting nutritional data.
The Ideal Study:
I'd like to start with a description of an ideal study – the one that would definitely answer lots of questions with scientific rigor. And I'll tell you why it's unrealistic to have such a study, by deconstructing this concept element by element. First of all, such a study would be a randomized triple-blind clinical controlled trial of several well-defined dietary approaches – hundreds if not thousands of participants would be randomized into one of the groups, they would receive prepackaged foods that would not be distinguishable from each other, so they wouldn't know what kind of diet they are receiving. The caloric intake would be controlled for as well as the level of physical activity, smoking, drinking, and number of sociodemographic and medical parameters. Finally, these people should be followed for extended period of time; we're talking years, if not decades. We need large cohort of patients and we need to follow them for a long time as most of the long-term outcomes are relatively rare and we need lots of person-years to see a meaningful amount of events such as heart attacks, cancer etc.
What we actually have:
At this point, I think, it's clear why it's almost impossible to have such a study – it would be too costly and restrictive, but, quite honestly, it's the only way to establish causal relationship with the highest degree of certainty. So, what do we have instead?
1.No blinding (participants do not know what type of intervention they are receiving) – this is one of the most difficult elements to create in nutritional research when we are talking about diets. It still can be done with supplements. Effectively, it would require a special food manufacturing plant and a huge lab of food engineers to create the rations to provide participants with. In reality, we have to rely on the foods that we have. Also, there are quite a few psychological aspects of nutrition that should not be ignored.
2. Clinical trials – randomized or not, unblended – we actually do have a few, but with their own problems:
a) Almost no long-term clinical trials – it's nearly impossible to maintain the rigor of a clinical trial for extended periods of time, so what we have is short-term studies ranging from several months to a year, rarely longer (just a follow-up in a while at best).
b) Randomization does happen in some cases, but there are some inherent problems with it as subjects are not blinded and they might not like their group assignment. Say, someone who is a vegetarian is assigned to a group of people consuming animal products – that would be unethical and practically we wouldn't expect such a participant to stay in the study and to adhere to its design.
c) Finally, with small number of participants and relatively short duration of these trials you can't really expect to see a meaningful number of major events like cancer or stroke occurrence, which is a huge limitation.
Positive aspect: These studies are quite rigorous and they can provide us with a good snapshot of the short-term effects of specific diets such as changes in cholesterol level, hunger, weight gain or loss etc.
3. Observational studies (the most widely used approach):
a) No randomization or blinding possible by default as the studies are observational in nature.
b) Nutritional assessment is often done using food frequency questionnaires (FFQ), which are difficult to fill out for lots of people. Do you remember how many servings of fruit you had last week? What about past 3 months? Personally, I don't and I see how problematic this approach is.
c) Nutritional assessments are done very infrequently, sometimes only once or twice and diets change over the course of time so whatever information you get at one point of time can completely change within the course of weeks or months let alone years or decades.
d) Multiple confounding variables – the direct result of non-random assignment. For example, someone who reports eating lots of red meat might be more likely not to care about their doctor's advice, to smoke and to drink lots of beer, whereas someone who sticks to "balanced" diet is likely to be very health-minded and, accordingly, to exercise more and not to smoke. Thus, if we see that red meat eaters having higher risk of dying prematurely we are not sure if this happens due to the red meat consumption or due to other factors. That makes these confounders very important and often there is no way to account for them other than mathematically.
So, in the end of the day, what we have is a limited number of seriously flawed studies – either short-term clinical trials that cannot track long-term effects and spot major medical events or some long-term large-scale observational studies that are so methodologically flawed that the data are nearly meaningless. That's the reality of the nutritional research on diets and my final point would be to be aware of these problems and to take all these new sensational scientific breakthroughs with a grain of salt.
These were my thoughts on the matter. I hope it was helpful and, as always, feel free to ask questions, to make comments and to explore my website and to subscribe to my YouTube channel for updates.
Sincerely Yours,
Dr.Sam