Existing models for measuring health impacts of the human diet are limiting our capacity to solve obesity and its related health problems, claim two of the world's leading nutritional scientists in their newest research.
In the latest edition of Annual Review of Nutrition, Professor David Raubenheimer and Professor Stephen Simpson from the University of Sydney's Charles Perkins Centre call for a radical rethinking of human nutrition science through a new framework called -nutritional geometry' – the culmination of more than 20 years of research in the field.
-Nutritional geometry' considers how mixtures of nutrients and other dietary components influence health and disease, rather than focusing on any one nutrient in isolation. It is hoped this new model will assist health professionals, dietitians and researchers to better understand and manage the complexities of obesity.
'Our framework throws down the gauntlet to the whole field of human nutrition. It shows that the prevailing focus on single nutrients is not able to help us understand complex chronic diseases, and that an approach based on nutrient balance can help solve the problem," said Professor Stephen Simpson, Academic Director of the Charles Perkins Centre.
Human nutrition science has historically focused on a single-nutrient approach, which is predicated on a lack of resources or micronutrient deficiency. For instance, the absence of vitamin C in human diets is a known cause of scurvy.
But this traditional approach is no longer useful in the face of modern nutrition-related diseases, the authors argue, which are driven by an overabundance of food, an evolved fondness for foods containing particular blends of nutrients, and savvy marketing by the packaged food industry which exploits these preferences.
'Conventional thinking which demonises fat, carbohydrate or sugar in isolation as causes of the obesity crisis – dubbed the single nutrient approach – has now run its course. We've provided a framework for not only thinking about but also experimentally testing issues around dietary balance. Much like the invention of the telescope or microscope, this framework offers a new tool with which to look at complex dietary problems and bring them into focus," said Professor Simpson.
'Our new approach provides a unique method to unify observations from many fields and better understand how nutrients, foods and diets interact to affect health and disease in humans," said Professor David Raubenheimer, who heads the Nutrition Theme at the Charles Perkins Centre.
'The -nutritional geometry' framework enables us to plot foods, meals, diets and dietary patterns together based on their nutrient composition, and this helps researchers to observe otherwise overlooked patterns in the links between certain diets, health and disease."
The new model enables complex problems like obesity to be viewed from a variety of perspectives, from the impacts of nutrients on metabolism and the health of individuals, through to the sustainability of global food systems.
'Although at face value more complex than the single-nutrient model, our -nutritional geometry' framework can simplify the study of human nutrition in the long run by helping to identify those subsets of factors and their interactions that are driving negative health and environmental outcomes in our rapidly changing environments," said Professor Simpson.
To illustrate the power of the approach, Professor Raubenheimer and Professor Simpson plotted data for the composition of 116 diets, compiled from previous published studies examining macronutrient ratios (carbohydrate, fats and protein) and energy intake in humans.
Their model shows that protein was the strongest driver influencing diet, regulating the intake of fat and carbohydrate. This finding is consistent with the previously observed -protein leverage' phenomenon, in which the strong human appetite for protein leverages the intake of fats, carbohydrates and total energy.
The paper, -Nutritional Ecology and Human Health' is published in the latest edition of Annual Review of Nutrition; complementary full text access is available online.
David Raubenheimer is Professor in nutritional ecology in the Charles Perkins Centre at the University of Sydney. His research applies theory from ecology and evolution to understand how animals adapt to the foods and diets available to them. The many species that he has studied – among them insects, spiders, fish, birds, grizzly bears, giant pandas, elephants, snow leopards, monkeys, baboons, gorillas, and orangutans – all have one thing in common: they are experts at choosing combinations of food that provide a nutritionally balanced diet.
Ten years ago David and his long-term collaborator Professor Stephen Simpson, Academic Director of the Charles Perkins Centre, started to wonder about our own species. If other animals select a balanced diet, then can we too? And if so, why are health problems that result from a bad diet so common – for example obesity, diabetes, and some cancers? If anything, we should do better than other species, because agriculture, science and technology have provided unprecedented opportunities to create an abundant and healthy diet.
David, Stephen and their collaborators have applied the theory of nutritional ecology to understand why. Brooke spoke with David about how a modelling approach that they invented, called nutritional geometry, can help to understand and solve the problems of human nutrition.
Question: What do you consider to be the biggest challenge for human nutrition?
Professor David Raubenheimer: Tremendous progress has been made in understanding various aspects of how diet relates to health. Tens of thousands of scientific papers are published each year, on topics from detailed nutritional physiology to large-scale issues like trends in global food supply and many between. The field is not lacking for information.
And yet urgent nutritional challenges that we face in the modern world have not been solved. A good example is obesity and its associated problems, like diabetes and heart disease. These are on the rise, at immense cost in human suffering and healthcare dollars.
Why have we not solved these problems when we know so much about them? We believe that the problem is not lack of information, or of the ability to collect more information. Rather, we need new ways of thinking about the problem; we need new models for interpreting the mass of information that we already have, and deciding what are the most important unanswered questions at which to direct our research.
Question: What are the issues with the existing models for measuring health impacts of the human diet?
Professor David Raubenheimer: There are several, but two stand out as particularly important:
1. Too much emphasis is placed on identifying which specific nutrient causes a particular health problem. Take for example the argument over whether fats or carbs cause obesity. It began over half a century ago, and despite the best that science has thrown at it the debate remains unresolved. Some researchers argue that fats are to blame, others that carbs are to blame; to further complicate issues, yet other researchers suggest that eating too much protein or too little fiber is the cause.
Our work with non-human animals suggests that obesity is very unlikely to be caused by a single nutrient. Rather, many nutrients interact with each other to influence obesity and its health consequences. To solve the problem we need to change the question that we ask from 'which nutrient causes obesity and how" to 'which combinations of nutrients are associated with obesity, and how do the different components of these interact to cause the problem".
2. There is no clear idea whether nutrients, foods or diets are the right level at which to approach questions of how nutrition influences health. The confusion that has resulted from trying to pin the blame for obesity on a single nutrient has led many scientists to conclude that nutrients are the wrong level at which to approach the problem. Rather, they argue, we should be directing our efforts at understanding which foods and diets are to blame for the obesity crisis and other nutrition-related health issues.
In reality, nutrients, foods, meals and diets are all important - each provides the best level for understanding particular parts of the problem. For example, there are strong associations between eating junk foods such as doughnuts and obesity. But foods like doughnuts do not themselves interact with our physiology to make us fat; rather, the nutrients that they contain do. Nutrients, and not foods, therefore are the right level at which to think about the physiological causes of obesity.
But to understand how those nutrients actually get into our mouths, stomachs, blood and cells to affect our physiology, we need to understand why we eat doughnuts. This can involve nutrients – for example, we might eat doughnuts because we like the sweetness from their high sugar content. But it will also involve other things, including price, advertising, availability, and so forth. If doughnuts were more expensive, or more difficult to find, we'd eat less of them. Foods are therefore an important level at which to understand what fills our shopping carts, and in turn enters our mouths.
The issue doesn't stop at foods. Even though doughnuts could make us fat, they won't necessarily do so. In some circumstances they might have no effect, for example if eaten only occasionally as a treat, or they might even prevent disease if fat and carbohydrate are lacking in the diet. To understand the role of foods in health, we therefore need to know how they combine with other foods to form the overall diet – are they a disproportionate part of the diet, are they an occasional treat with no health consequences, or are they necessary to help balance an otherwise imbalanced diet?
We should not therefore be debating whether nutrients, foods or diets are the right level for understanding the links between eating and health. Rather, they should all be included in our models, so that we can relate each to its specific roles in the problem.
Question: How have these existing models limited our capacity to rectify the obesity crisis?
Professor David Raubenheimer: Not only has the single-nutrient model failed to manage the obesity crisis, but might actually have contributed to it. For example, in the 1960's and 1970's when obesity first emerged as a serious problem, too much dietary fat was widely considered to be the cause. Official advice was, quite logically, to reduce the amounts of fat in the diet. But in the 1980's it became clear that this wasn't working: there was no sign of the obesity epidemic slowing, let alone reversing.
So were people not following the dietary advice? It turns out that they were, but it was bad advice. It was bad not because fats aren't associated with obesity – as we said above, the jury is still out on this – but because the focus on a single nutrient had unintended consequences. Rather than reduce the total amount of energy eaten, which very likely would reduce obesity, people simply replaced fat in the diet with carbs.
It's not difficult to understand why they did this. Replacing fats with carbs was easier, because it enabled people to continue to eat the amount they usually did, while providing the comfort of sticking within official dietary advice. It was also aided, in no small measure, by the processed foods industry. Sensing an opportunity, they quickly began marketing foods that were conspicuously labelled 'low-fat". What the labels didn't say is that these foods were also 'high-carb".
Many have concluded that if fat isn't responsible, and obesity has continued to rise with increased carb intake, then carbs must be to blame for the obesity epidemic. It might be true that excessive carb intake has played a role in the obesity epidemic (probably together with fat), but there are again suggestions that singling them out as 'the cause" is leading to problems. First, the shift of blame from fats to carbs has not reduced the problem of obesity. Second, just as demonizing fats led to increased carb intake, so too is the vilification of carbs leading to nutritional imbalances with unintended health consequences.
It seems quite obvious that if carbs are the cause of obesity, then the solution is to eat a low-carb diet. This is exactly what many people are doing – for example the paleo, Atkins and Banting diets. But recalling the fat-carb debacle, if carbs are reduced then it is likely that something else will replace them, and that something turns out to be protein. Although low-carb/high-protein diets likely do lead to reduced energy intake (for reasons discussed below), evidence now suggests that they might also have nasty side-effects. They accelerate the onset of age-related diseases such as cancers, and shorten lifespan.
The message is clear. Attempting to solve health problems nutrient-by-nutrient is like herding cats: as soon as one nutrient is under control, another slips out of line. And even if a particular nutrient does turn out to be the main culprit in the obesity epidemic, this information is only of academic interest unless we can find ways of preventing people from over-eating it. To do that, we need to know which foods are providing the nutrient in excess, why those foods are purchased and eaten in the quantities they are, and which other foods they are combined with in the diet overall. This is where nutritional geometry comes in.
Question: What is nutritional geometry?
Professor David Raubenheimer: Nutritional geometry is a modelling approach to examine how mixtures of nutrients (and other components of foods, such as fiber) influence health. Rather than ask whether obesity is caused by carbs, for example, nutritional geometry enables us to examine how different mixtures of carb:fat:protein relate to obesity and other health issues such as the rate of ageing. This is important, because a high carb diet will have different effects on health depending on what carbs are high relative to: high relative to fat, relative to protein or relative to both.
Thinking about diet in this way can help to solve many issues in nutrition, including the debate about whether fat vs. carbs causes obesity. Those arguments have not gone away because studies keep producing contradictory results. Some show that carbs but not fats are associated with obesity, others show that fats but not carbs are, and yet others show that both are associated with obesity. It is likely that there is nothing to argue about at all. Sometimes fats do and sometimes don't result in obesity, and likewise for carbs, with the particular outcome depending on the levels of other nutrients in the diet. For example, fat might cause obesity if the dietary level is high relative to protein, but not if high relative to carbohydrate.
By modelling nutrient mixtures, nutritional geometry can therefore help understand links between diet and health. But modelling mixtures also helps in another respect. It avoids the argument of whether nutrients, foods, meals or diets are the right level for managing the nutrition-related health, because it enables all of these levels to be included in a model. The trick here is that each of these levels is itself a mixture: foods are mixtures of nutrients (and other things, like fiber), meals are mixtures of foods, and diets are mixtures of meals. By using the mixture-ready approach of nutritional geometry, we can therefore move seamlessly between the levels, and identify how each plays a role in the problem.
Nutritional geometry therefore provides a means to avoid both of the basic problems with other nutritional models mentioned above. It deals with mixtures of nutrients, as well as all other levels of dietary mixtures including foods, meals, and diets. And it does this in a common language that speaks directly to the physiological systems that determine health and disease: nutrients. This has enabled us to develop a new model of obesity, the Protein Leverage Hypothesis.
Question: Can you talk us through the new human nutrition model you propose?
Professor David Raubenheimer: In the Protein Leverage Hypothesis, neither fats nor carbs on its own is specifically to blame for obesity, but neither are they entirely exonerated. Rather, both nutrients interact with protein to cause the problem.
To understand how this interaction takes place we need to take a step back and think about certain aspects of human biology. Like most other animals, humans have separate appetites for fats, carbohydrates and protein. This makes sense, because the body must know which nutrients it needs at a given time and ensure that it eats enough of those. In humans (and many other animals) the appetite for protein is particularly strong. By 'strong protein appetite", is meant that humans remain hungry until they have eaten enough protein, and also that they will do their best to avoid eating too much protein. Marksman-like, our appetites target a certain amount of protein, and to the best of their ability avoid both over- and under-shooting that target.
This tight regulation of protein intake has important functions. Without enough dietary protein we 'digest" our own bodies, drawing protein from muscle to produce life-supporting enzymes and other essential protein-derived molecules. At the same time, too much dietary protein is toxic and shortens life. A finely-honed protein appetite ensures that we remain between the rock and the hard place of self-cannibalizing and poisoning our bodies.
But strong protein regulation also has a downside. If we eat a diet that is low in protein relative to fats and carbs (for example the doughnuts mentioned above), in order to avoid under-shooting our protein target we will need to over-eat fats and carbs. Our new model of obesity – the Protein Leverage Hypothesis – suggests that the obesity epidemic is a result of this strong protein appetite colliding with a reduction in the proportion of protein in the human diet (caused by social and other factors). To maintain our target protein intake, we now need to eat more carbs and fat than previously.
Does this mean that the solution to the obesity crisis is 'increase the protein content of our diets"? This is certainly the implication, and there is a lot of evidence that high protein diets do help to reduce energy intake and obesity. But, unfortunately, in nutrition things are seldom as simple as they seem. If the proportion of protein in the diet is too high – like the 'low carb" weight management diets discussed above – fat loss is likely to come at the cost of accelerated ageing and shortened lifespan. The message, therefore, is eat a BALANCED diet: one that is neither too high, nor too low in protein, fats or carbs.
Question: How will this method help us understand and manage the complexities of obesity?
Professor David Raubenheimer: Nutritional Geometry might seem more complicated than the traditional approach of trying to unravel the problem one nutrient at a time. But it actually simplifies and streamlines the search for the causes of obesity and helps to identify solutions.
One way that it simplifies is by helping to make sense of otherwise uninterpretable or contradictory information. We gave an example above - the arguments over whether fat or carbs is to blame for obesity. Nutritional geometry can resolve that argument. It does this by digging deeper into the problem, where it can find the point at which the results do not contradict each other. In the Protein Leverage Hypothesis, we strike gold when our excavation reaches the depth at which protein connects both nutrients – either nutrient is or is not associated with over-eating energy, depending on whether it occurs in the diet with low or high levels of protein. At this point the anti-fat and anti-carbs camps cease to be adversaries, and become part of the same explanation for the obesity epidemic.
Imposing order in this way on what is otherwise a tangled mess of information has benefits beyond peace-keeping. It can help to identify the most effective targets for research, and design practical strategies for combatting the problem. For example, if protein dilution is the cause of the obesity epidemic then we need to find out how and why the protein content of our diets has become diluted to cause the modern epidemic of energy over-consumption. Answers to that question will help us to deal with the problem.
Question: Why do you believe we have an obesity crisis?
Professor David Raubenheimer: The obesity epidemic has arisen so rapidly – within a single generation - that it can only be due changes in our environment. The question is: 'what has changed in our environment to cause us to eat too much energy?"
We suspect that a reduction of protein in the human diet has made an important contribution to the problem. Although a good deal of research still needs to be done, there already is respectable evidence that this is the case – in many countries the proportion of protein in the diet has gone down as energy intake and obesity has increased. It has also been shown in several laboratory experiments that humans eat more energy when restricted to low protein diets compared with high protein diets, as expected if the Protein Leverage Hypothesis is true.
It would, though, be a mistake to conclude that this closes the question – the obesity crisis is due to protein leverage. Protein leverage is a nutrient-level explanation, but as discussed above, to really understand and manage a nutritional crisis like the obesity epidemic we also need to approach the issue from other levels, such as foods and diets. For example, we need to know which foods are causing the dilution of protein in the human diet, and why these foods are being included in diets at damaging levels. We already have answers to both of these questions.
The foods involved are what are technically called ultra-processed foods – more commonly known as junk foods. These are mass-produced, industrial products that are made from highly processed ingredients, and usually marketed by big multi-national companies. They are typically high in fats, carbohydrates, and salt, and low in protein, fiber, vitamins and minerals. Examples include crisps, ice cream, highly processed breads, processed meats and pizza. Ultra-processed foods have spread like a plague across the globe over the past few decades. Although it has happened at different times for different cultures, the result is always the same – when these kinds of foods hit the market, traditional diets are displaced, and obesity rises. So common and so predictable is the sequence that it has its own name – a 'nutrition transition".
Why are ultra-processed foods eaten in such large quantities that they transform traditional diets and lead to obesity? The short answer is that they are designed to. The processed ingredients are assembled into products that are precision-engineered to achieve one goal: high sales. Among the strategies to achieve this, is to combine the ingredients in proportions that taste best: in the industry jargon, which hit the 'bliss point". Experiments - and product sales - suggest that macronutrient mixtures that most closely approach the bliss point have high fat and/or high carbs, combined with – you guessed it - low protein. Think ice cream, chocolate, doughnuts, soda … etc.
The low protein content of ultra-processed foods not only increases their taste appeal, but serves another function. Protein is an expensive nutrient, and it is therefore a fortuitous gift to the processed food industry that we have a particular fondness for foods that don't contain much of it. It means that protein-dilute ultra-processed foods can also have a price advantage. How much better can it get than tasty and cheap? It's easy to see why ultra-processed foods so readily distort traditional diets to dilute their protein content and cause fats and carbs to be over-eaten.
This also explains another well-known but poorly understood association. Many studies have shown that obesity is higher in lower socioeconomic groups. If you think about it, this is the opposite of what we should expect: why would low-earners, who can least afford it, actually eat more than higher-earners? The protein leverage hypothesis provides an answer. What they eat more of is fats and carbs from cheap, low-protein, ultra-processed foods. Why they eat more if it is they are compelled by their marksman-like protein appetites, in order to avoid under-shooting the protein target. If this were true, then without cheap ultra-processed foods lower-earning groups should not be fatter - if anything leaner - than higher-earning people. This is exactly what we find in cultures that have not yet been gripped by, or are in the early stages of the nutrition transition.
Nutritional geometry can therefore help to explain the obesity crises as an outcome of evolved human appetites interacting with changed food environments to produce excess body fat. Obesity is just one example of how this approach can help to link nutrition to health. Other diseases, and other nutrients, for example specific types of fats, fiber and vitamins, can be analysed in the same way. The broader point is that modelling nutrition in terms of mixtures, whatever the components, rather than single nutrients, can help to clarify the relationships between diet and health like no other approach can.
Question: How will nutritional geometry allow you to find links between certain diets and disease?
Professor David Raubenheimer: There are two main approaches to studying the relationships between diets and disease. The first is the epidemiological method. This involves measuring what people eat in their everyday lives, and testing for relationships between diet and disease. Nutritional geometry provides a powerful tool for doing this. The reason is obvious: as we have seen above, foods, meals and diets ARE mixtures of nutrients, and understanding them in this way helps to see things that otherwise would be missed.
The second approach for finding links between diet and disease is the experimental method. In this approach, the responses of people (or sometimes laboratory animals, such as mice) to fixed diets of known composition are compared. Nutritional geometry plays more than one role in the experimental method.
Most obviously, it helps to design the mixtures (i.e. diets) whose effects we want to measure. Another, less obvious role is that it clarifies an important difference that can otherwise be overlooked in nutritional studies. At the nutrient level, the word 'diet" is often used rather vaguely to mean one of two things. First, it can refer to the composition of the foods that are provided to different groups of animals or people in the experiment. For example, we might say 'one experimental group was given a diet low in protein and high in carbohydrate". Second, the word 'diet" is used to describe the mix of nutrients that is actually eaten by the different groups.
Now you are almost certainly confused and thinking '… if an animal is given only a low-protein/high-carb -diet' (sense one above), how the heck can the -diet' that it has eaten (sense two) be anything but low protein/high-carb?". And you are right to be confused; this is the point. In one sense, the compositions of diet sense 1 and diet sense 2 must be the same: if all you have to eat is a low-protein/high-carb diet, then all you CAN eat low-protein/high-carb diet. But nutritional geometry helps us to see that the two 'diets" need not be the same thing. This can make a big difference to how we understand the links between diet and disease.
The trick is that in nutritional geometry, diets are not defined only as a simple chemical composition. Rather, they are defined in relation to the needs of the animal. In this sense, if an animal is restricted to foods with a lower protein/carbohydrate ratio than it needs, then that is a low-protein/high-carb diet in sense 1. But here's the magic: simply by regulating how much of that food it eats, the animal can transform low-protein/high-carb foods into a diet (sense 2) that is not low in protein at all, but exactly meets its protein needs. We saw an example of this above, in the protein leverage effect: humans eating low-protein diets (sense 1) eat enough of these to ensure they hit their target for protein, but in the process must over-eat fats and carbs. So it is NOT a low-protein diet in sense 2; it is a normal protein diet, with too much carbs and fat.
This might sound like academic nit-picking, but it most certainly is not. And here's why. If we see that the low-protein diet (sense 1) is associated with a particular disease, then we might be tempted to conclude that it is a disease of protein deficiency. But using nutritional geometry we are able to see that protein is not deficient in the diet at all – it is spot-on the level needed. By thinking in terms of mixtures, and putting the animal's own appetite at the center of models, nutritional geometry therefore greatly strengthens our ability to link diets with disease.
Question: What have you found, regarding the correlation between certain diets and disease, so far?
Professor David Raubenheimer: We've found that diets too low in protein relative to fats and carbs lead to energy over-consumption, obesity, and physiological signs associated with such diseases as diabetes and cardiovascular disease. On the other hand, diets that are too high in protein, especially animal proteins, can lead to the early onset of symptoms of ageing, including cancers. We are now extending this research in several directions, including the roles of dietary fiber and comparisons of animal and plant proteins in driving these responses.
Question: Do you have anything else you want to add?
Professor David Raubenheimer: At the start of this interview, I said that the challenge for human nutrition is not a lack of knowledge or information, but the ability to make sense of it all. Many of the particular points that we've made in relation to nutritional geometry and the Protein Leverage Hypothesis are not new to us – they have already been established in the huge sea of existing data. For example, any good nutritionist would avoid falling into the trap of assuming that a diet (sense 1) with a low proportion or protein/carbs will necessarily lead to a diet (sense 2) deficient in protein, by comparing actual intakes to recommended intakes.
Our point, though, is that we need ways to bring these separate pieces of the puzzle together into a bigger picture to understand how they fit together. Nutritional geometry provides this. For example, as we have seen above the Protein Leverage Hypothesis brings together a lot of information to produce a new understanding of obesity.
And in this model obesity is not treated in isolation, but considered in the context of other important health issues, such as accelerated ageing. In this way, nutritional geometry can help to identify new priorities for research. It needs, for example, to be established how prevalent the role of protein dilution is in the global patterns of obesity, and how it interacts with other factors to drive the crisis. This will help to identify the most effective ways to take charge of the hugely complex food systems that we've created and reduce the suffering from disease.
Interview by Brooke Hunter