In the previous post, I looked at how a lot of the common arguments used by advocates of the Paleo diet are actually quite flawed, and reminiscent in many ways of the typical cases given for vegetarianism and veganism.
Now, before I get hunted down by an angry mob with clubs and spears, I should point out that I am not necessarily saying that the Paleo Diet itself is flawed, nor that we should abandon it in favour of KFC Buckets and Krispy Kreme Donughts, only that some of the arguments used to advocate it are dubious at best.
The Problems with Nutritional Science
There is so much conflicting advice out there when it comes to diet, it is enough to make your head explode.
The government tell you to eat healthy whole grains and avoid red meat, but there are respected scientists and doctors advocating a Paleo diet which says exactly the opposite.
There are low fat diets, low carb diets, raw vegan, food combining, calorie restriction, the list goes on and on. All appear to have evidence to back up their arguments, but in all cases there seems to be contradictory evidence also.
Why is this the case, and what can we do about it?
Good science starts with an observation, which leads to a hypothesis, which must then be tested through experimentation in order to assess its validity.
This sounds great on paper, but the problem when it comes to the science of nutrition and human health, is that testing most hypotheses is extremely difficult, if not impossible.
Almost everything we “know” about nutrition comes from observational data. Observational data comes from studying populations or groups of people, gathering data about them, and then drawing conclusions based on correlations between observed behaviours and outcomes.
A recent case of observational data causing waves, both in the mainstream and all across the paleoshpere was the “Red Meat Causes Heart Disease, Cancer and Death” débâcle. For a thorough debunking of this particularly bad example of observational science at its worst, I’d recommend checking out this article by Denise Minger.
Now this was a very badly conducted observational study. Both the data collection, and categorisation were done in a laughably meaningless manner. Not all observational studies are quite so slap dash, as there are better ways of collecting and analysing data. No matter how well designed an observational study is, however, all you can possibly derive from the data is a hypothesis. There are simply always too many variables and confounding factors to draw any meaningful conclusions.
If you are familiar with nutritional science you may well have heard of “The lipid hypothesis”, “The diet-heart hypothesis” or more recently “The Carbohydrate Hypothesis”. From other fields of science you have more than likely heard of “The Theory of Evolution” and “The Theory of Relativity”.
In everyday language, the terms hypothesis and theory are used pretty much interchangeably. In science, however, the two terms have very specific meanings. A hypothesis is just that – Little more than an idea, or belief, held by a lone scientist or small group of scientists based on their observations and experiences. For a hypothesis to become a theory, these scientists must construct well designed experiments to test the validity of their hypothesis following the scientific method. These experiments must then be reviewed and replicated by their peers, until there is sufficient evidence and agreement within the scientific community to declare a new scientific theory.
In pretty much any other field of science, you would never even hear of the hypothesis until it came anywhere near becoming a unified theory – In the field of nutrition on the other hand, government policies, which affect lives of billions of people are formulated on the back of them – How can this be?
The problem with nutritional science hypotheses is that they are incredibly difficult to test.
The golden standard when it comes to scientific testing is the double-blind placebo controlled trial.
In such a study you would have two cohorts (or groups) of substantial size, one of which was eating x, the other a placebo, and neither the subjects nor the scientists new which was which until after the completion of the study. The problem is, when it comes to nutrition, this kind of study is pretty much impossible.
Take for example the hypothesis that red meat increases the risk of heart disease, cancer and all cause mortality.
To test this theory using the above method, you would need to get two large groups of people, have one eat meat, the other to abstain, while keeping all other variables (calories, macro and micronutrients etc) matched, somehow without either them, or the scientists, knowing whether they were eating the meat or not.
It would then be necessary to follow them for either their entire lives, or at least until a significant number of individuals from either one or both groups had developed heart disease, cancer or died!
Perhaps such a test could be theoretically possible by feeding the subjects some kind of “meat shake” and a veggie alternative, matched for nutritional value, but then you’d also need to run another similar experiment to see if the processing affected the outcome!
As you can imagine, conducting such an experiment is simply not possible – Even if you could find enough willing participants, the time and money involved would be astronomical.
Though it is not really possible to blind either the scientists nor participants, it is certainly still possible to design controlled human studies.
Even the best designed human clinical study has its limitations however:
Too many variables – Cut out a food, and you’ll reduce the number of calories the participants are eating, but replace the food with something else and you’ll inevitably alter the macro-nutrient ratio. Also, how does one group foods? “Red Meat” is a very broad category, how do we know beef, lamb, pork all have the same effects? What about different farming methods and preparation techniques? What about different combinations of foods?
The bell curve effect – Things are not always as straight forward as linear causality (i.e. none of something is best, lots of it is worse or vice versa). For example, no animal products at all is certainly very bad, but perhaps too many animal products are also bad, and the “sweet spot” lies somewhere in the middle. Unless you have subjects eating on every point of the meat axis, it would not be possible to spot this relationship.
Time – It is neither ethical, nor practical to enlist people into clinical studies for extended periods of time. For this reason most are conducted over a few months, or years at most. Scientists then measure “markers of health”, intended to predict long-term health outcomes. Unfortunately many of these so-called markers later turn out to be very unreliable (cholesterol being a prime example).
An alternative to human studies are animal models.
These have advantages over human studies because:
You can control the diet of the animals completely
You can study the animal for its entire lifespan in short space of time
You don’t have to pay the animals or get them to sign a disclaimer!
Unfortunately, though animal studies can be very useful in creating new hypothesis, this is all that they can do, as while we might share many common strands of DNA, there is no way of knowing whether the results can be extrapolated to humans.
Things are further confounded by the fact that lab rats are not being fed little miniature burgers and pizzas, but various kinds of “rat chow” that bears very little resemblance to anything ever actually eaten by humans.
Despite their many flaws, observational studies actually have some advantages over either human or animal clinical trials, in that they provide data from human subjects, in large numbers, over entire lifetimes.
Our decisions with regards to what to eat, therefore, have to be based on a combination of observational data, short-term human trials, and animal models. Three notoriously flawed sources of information!
No wonder we’re so confused over what to eat!
In the next instalment, I’m going to look at the problems with Paleo Diet “Science”, and how best to make decisions about what to eat based on the available information.