products typically formulated from many ingredients. Their ingredients include those also used in group 3 foods, such as salt, sugar, and fats, and substances rarely used in cooking, such as additives that mimic the sensory qualities of group 1 foods.
The NOVA system has been criticized for being “rather simple and crude” compared to other classifications that are based primarily on nutritional composition [84]. In our view, however, the fact that it does not focus specifically on nutrients but on processing is a significant strength of the NOVA system, which enables it to integrate diet and health with the food environment and nutrition transitions frameworks to the extent that nutrient‐based classification could not. One reason for this is that the healthfulness of foods is determined not only by their nutritional content but the physicochemical matrix within which the nutrients occur [85]. Processing fundamentally alters the food matrix, and the NOVA system captures this in a way that nutrient‐based classifications do not. Another more fundamental reason is that the composition of a particular category of food does not in itself influence health, but it does so only in relation to the proportion it contributes to diets. The probability that a food category will contribute excessively to diets is determined not only by its nutrient composition but also by social factors including price, convenience, marketing, and hedonic manipulations such as the addition of flavorings [86]. Since these factors are strongly associated with the corporate strategies and commercial success of the ultra‐processing of foods [1], the NOVA system is particularly well suited for examining the ecological potential for foods to influence diets and health.
Protein leverage and nutrition transitions
An urgent priority, increasingly gaining policy attention, is to understand better how people interact with transitioning food environments to shape their diets [75]. This is a complex issue for many reasons. Food environments can be characterized across different scales, including the home, work, school, and neighborhood food environments, and there are challenges in defining and delineating them at the different scales [87]. Alternatively, they can be framed according to constructs likely to influence eating outcomes, including the community environment (e.g. type and location of food outlets, opening hours), the organizational environment (home, work, school, etc.), the consumer environment (nutrition information, available healthy options, food pricing, etc.) and information environment (media, advertising, dietary guidelines), which themselves can be challenging to define and measure [88]. Understanding the routes via which food environments influence consumer behavior adds an additional level of complexity. There are multiple factors involved, including both direct influences of food environments on diets as well as influences that act indirectly via psychological mediators such as perceived control [89,90], and various factors can interact synergistically to influence behavior [91].
Recent research suggests that the protein leverage hypothesis might provide a new approach for integrating with existing public health frameworks to understand how human biology interacts with transitioning food environments to generate epidemics of obesity and associated disease. It has become apparent that a significant dimension of the global nutrition transition to obesogenic food environments is the displacement of traditional diets by a dietary pattern in which NOVA category 4, ultra‐processed foods, are a staple source of calories [92]. These products already dominate the food supplies of high‐income countries, for example comprising 61% of food energy purchased by US households [93], and their consumption is rapidly increasing in middle and lower‐income countries [94,95]. Additionally, several studies have established links between ultra‐processed foods and obesity [92,96–99].
To test whether protein leverage is a biological mediator that links ultra‐processed foods to obesity, Martínez Steele et al. [68] used nutritional geometry to analyze NHANES data (2009–2010) in relation to their contribution to the US diet (Fig. 6.7). There was a strong inverse relationship between consumption of ultra‐processed foods and the overall proportion of dietary protein, with mean protein content falling from 18.2 to 13.3% between the lowest and highest quintiles of the dietary contribution of ultra‐processed foods. This occurred because ultra‐processed foods tend to be lower in percent protein and higher in energy from fats and carbohydrates than less highly processed foods. As predicted by protein leverage, a rise in total energy intake was associated with the fall in percent protein as the dietary contribution of ultra‐processed foods increased, whereas absolute protein intake remained near‐constant. Recently, Hall et al. [60] showed similar evidence for prioritization of protein intake in an inpatient randomized controlled trial.
Although more research is needed, existing evidence thus suggests that the strong human protein appetite might interact with the proliferation of ultra‐processed foods in industrialized food environments to generate excess energy intake and obesity. This is a fundamental paradigm shift – it suggests that humans over‐eat fats and carbohydrates (and total energy) not because they have particularly strong appetites for those macronutrients, but because their appetite for protein is stronger than the appetites for fats and carbohydrates. This shift in framing could have a significant impact on obesity research, prevention, and management.
For example, much has been written on the fact that the reduction in fat intake associated with US dietary guidelines failed to stem the rise of obesity. This fact is significant not only because it reflects a failed public health initiative, but it might well have exacerbated the US dietary crisis by ostensibly vindicating dietary fats and demonizing carbohydrates [100–102]. The resulting “macronutrient wars” [103] have further polarized scientific and public debate, diverting attention away from the rational question of “what diet is healthy overall?” towards extreme dietary philosophies focussed around minimizing or excluding a particular macronutrient. Viewed from the perspective of protein leverage, however, a likely reason that reducing dietary fat did not solve the obesity problem is that human appetite systems ensured it was replaced by carbohydrate calories to maintain protein near the target ratio (as in Fig. 6.4). Indeed, amid the ensuing debate around fat vs. carbohydrates, the percentage of energy contributed by protein in the US diet decreased marginally (by 1%), which as noted above, is sufficient to drive an obesity epidemic, and as predicted by the protein leverage hypothesis obesity continued to rise [104].
Figure 6.7 Relationship between ultra‐processed food consumption and protein leverage in the United States. The symbols represent protein and non‐protein energy intakes for the lowest (green) to highest (red) quintiles of ultra‐processed food (UPF) consumption reported in the National Health and Nutrition Examination Survey 2009–2010. Numbers in the legend show the mean dietary contribution of UPF (% of total energy intake) for each quintile. The negatively sloped diagonals represent daily total energy intakes (calculated as the sum of X + Y), and the positive radials represent the dietary protein: non‐protein energy ratio (X/Y). The dark vertical, horizontal, and diagonal lines represent alternative models of macronutrient regulation (as in Fig. 6.1c). As predicted by the protein leverage hypothesis, increasing UPF in the diet was associated with decreased percent dietary protein (18.3–13.3%) and increased total energy intake (8.2–8.9 MJ), while absolute protein intake varied little.
Source: From Martínez Steele et al. [68].
Another example where protein leverage can potentially resolve unexplained phenomena, concerns the fact that the higher the target level for protein intake the more susceptible an individual would be to energy over‐consumption on an obesogenic diet [53] (Fig. 6.6). This could explain the high susceptibility to obesity of populations or individuals with a history of high protein diets. For example, the islandic populations of Oceania, which have until recently remained on marine‐based diets rather than having shifted to carbohydrate‐rich