by Eric Helms, Ph.D.

Study Reviewed: Ultra-Processed Diets Cause Excess Calorie Intake and Weight Gain: An Inpatient Randomized Controlled Trial of Ad Libitum Food Intake. Hall et al. (2019)

The diet wars have intensified. Blame for obesity shifts across a gamut from fat, carbs, sugar, processed foods, animal products, and back again. As discussed (article here), the battleground outcome of carbs versus fat is the least promising at the population level in terms of providing a solution. Differences in low fat versus low carb interventions are clinically insignificant at the group level (though they often matter for individuals). This is just one example of the failed attempts to find a “smoking gun” in the obesity epidemic. Weight management is better accomplished with multi- versus single-variable interventions. Rather than a smoking gun, there is a firing squad of mechanisms driving weight gain. In this causal analysis of processed foods as a contributor to weight gain (1), researchers examined a multitude of mechanisms related to energy consumption and body composition change. Researchers used a tightly controlled metabolic ward trial that mimicked real-life using a free-eating model. Specifically, 20 adults were presented diets dominated by processed or unprocessed foods for two weeks apiece, and the diets as presented were matched for energy, macronutrients, sugar, fat, and fiber. However, the subjects were able to consume as much as they desired of the food presented. During the processed diet, participants ate faster and consumed more energy, consuming greater amounts of carbs and fat but similar protein compared to the unprocessed diet; therefore, they gained weight and fat mass when consuming the processed diet, but lost weight and fat mass while consuming the unprocessed diet. Additionally, hormone markers of hunger were lower compared to baseline during the unprocessed diet, and satiety was higher during the unprocessed compared to the processed diet. In this article, we’ll discuss which aspects of the diets could be controlled by the researchers, and which they could only control to some degree. This discussion hints at the inherent issues of processed diets, and why they may be among the deadliest of the firing squad. 

Purpose: The purpose of this study was to determine if processed foods play a causal role in ad libitum (at will or, as desired, “free eating”) energy intake, and subsequent changes in body mass. 

Subjects: 20 (10 male, 10 female) weight-stable adults (31.2 ± 1.6 years; BMI = 27 ± 1.5 kg/m2) participated in this randomized, controlled, two-week crossover trial. 

Methods: Participants lived in a metabolic ward research center continuously for 28 days, during which time they were randomly allocated to either the processed or unprocessed diet condition for two weeks, immediately followed by the opposite condition. Hormone and metabolic data from blood samples, body composition measurements via DXA, 24-hour energy expenditure changes in a metabolic chamber, average energy expenditure from doubly labeled water, and visceral fat changes from MRI were all collected.

Additionally, participants assessed aspects of their eating and food experience via visual analog scales. They rated hunger, fullness, satisfaction, eating capacity, familiarity, and pleasantness, and the researchers tracked how much total food was consumed, which foods were consumed at which meals, and the rate that the participants consumed their food. 

Participants were given three meals daily and told to eat as much as they wanted, with each meal period lasting up to 60 minutes. Subjects also had access to snacks. The subjects were presented food for each meal; if the subjects ate all of the food they were presented, they would have consumed approximately 3900kcal/day, and the diets would have been matched for energy, macronutrients, fiber, total sugar, sodium, and energy density, differing only by the percentage of energy from processed or unprocessed foods as defined by the NOVA classification (2). The subjects were free to eat as much or as little of each presented food as they pleased, meaning the actual diets consumed were free to differ substantially. Briefly, the NOVA system is a peer-reviewed published system using a checklist to group foods according to the extent and purpose of industrial processing. This includes processes and ingredients used to manufacture ultra-processed foods which are designed to create low-cost, long shelf-life,  ready-to-consume, hyper-palatable products likely to displace unprocessed or minimally processed foods. How much of the presented diets were consumed, and whether additional food was consumed via snacks, was ad libitum. Diet specifics are shown in Table 1.

Findings

Energy, energy macronutrient composition, energy distribution across meals, hunger and satiety scores, and eating speed are shown in Figure 2. In short, during the processed diet, participants ate ~500kcal more on average per day (also shown in Table 2) and ate at a faster rate, while rating hunger and satiety similarly on both diets. Additional calories were consumed during the processed diet via carbohydrate and fat, while reaching the same protein intake. The subjects primarily consumed more at breakfast and lunch during the processed diet, while energy intake at dinner was similar in both diets.

Body weight increased during the processed diet but decreased during the unprocessed diet (Figure 3). There was a strong, significant association between energy intake relative to baseline and increases in body weight. Body fat was the only significant body composition change, and changes in lean mass approached significance. However, as I’ll discuss in the interpretation, the nonsignificant “lean mass” gains were likely dominated by changes in body water and were pretty variable between individuals, while changes in body fat were significant and consistent across individuals.

Blood measurements followed the changes in body mass (see Table 3), such that markers clearly indicated or trended toward showing a shift into a calorie deficit during the unprocessed diet. During the unprocessed diet, insulin and leptin were lower versus baseline and the processed diet (while not reaching the cut-off for significance), and T3 (thyroid hormone) was significantly lower, while free fatty acids were significantly higher (indicating body fat was metabolized).

Interpretation

The broad interpretation of this study is pretty straightforward. A diet largely dominated by processed foods is less filling, easier to eat quickly, and more likely to result in greater consumption of total calories, largely driven by a lower protein content per calorie and higher energy density. However, to know what this means in terms of practice, and to understand the nuanced “whys” behind the outcomes, there is a lot more to examine.

The difference between factually correct information and useful information

While less common these days, there was a time in the “evidence-based IIFYM (if it fits your macros) community” that the common response to any questioning of certain foods being “acceptable” for health, fitness, fat loss, or bodybuilding would be met with a statement along the lines of: “diet quality doesn’t matter for body composition change if macros (mainly protein), calories, fiber, and (sometimes) sugar (depending on who was talking) are matched (with the occasional caveat of also taking a multivitamin).” Now, this is factually true, at least mostly, and at least in the short term. I can think of strange diet setups where certain essential fatty acids, amino acids, and micronutrients are lacking and eventually cause problems, but I digress. Despite being a factual statement, I don’t think this is a very useful statement. This factually true statement only has applicability if all these variables are controlled, meaning you are tracking energy and/or macros.

However, most people don’t, and at least in the long term, shouldn’t be in a constant state of tracking macros or calories. As discussed in my video series on flexible dieting (video one, two, three), tracking can be used as an instructive tool, but it should be temporary, as there are potential pitfalls from primarily relying on external cues for regulating energy intake. In the real world, general population gym-goers largely find tracking tedious and unsustainable, in my experience. Thus, unless your goal is not intended to be sustained (for example, a contest preparation resulting in stage leanness), the behaviors utilized shouldn’t be unsustainable either, and 99% of people aren’t going to track their macros for the rest of their lives. 

So again, is it useful to know that a processed diet can be just as effective as an unprocessed one so long as all nutritional variables like calories and macros are controlled? Not really, because in reality, those variables generally won’t be controlled. The only utility of that statement I can see is if you’re talking to someone who is actually afraid of specific processed foods and thinks for some insidious magical reason they are harmful at any frequency or dose. In that case, I could see the value in letting someone know that occasionally having a Snickers bar in the context of a healthy diet is absolutely fine. Moreso, even if someone does track habitually, the knowledge that a processed diet could work isn’t really helpful. Indeed, it could cause problems, as following such a diet would make the qualitative difficulty of maintaining a deficit via a processed diet higher. If people find the same level of satiety and hunger occurs when eating an additional 500kcals (as was shown in this study), it’s logical to assume they’d feel hungrier on a processed diet when eating the same amount of calories they’d otherwise consume on an unprocessed diet.

Why processed diets result in “spontaneous” weight gain, and conversely, why unprocessed diets result in “spontaneous” weight loss

In the real world, energy density is higher on processed diets. If you’re thinking “what is energy density?”, watch my video series on it that I just concluded in this issue (Part 1 here). While the researchers attempted to control for energy density, it probably couldn’t be properly accounted for because processed foods are simply so much higher in energy density. Thus, an attempt was made by adding fiber to the beverages of the processed diet, but unfortunately, fluids aren’t as well “recognized” by the body for inducing satiety (3).

Another reason processed diets lead to food overconsumption in the real world is due to a phenomenon known as protein leverage theory. Simply put, satiety is lower and hunger is higher until a certain threshold of protein is consumed in an ad libitum setting. Thus, because processed foods are so densely packed with carbohydrates and fat, you end up consuming more total calories to reach the same level of protein. Indeed, some research indicates a part of the reason processed diets are fattening is because they have proportionally less protein per calorie (4). In this study, it seems the ad libitum intake of the participants during the processed diet was primarily via carbohydrate- and fat-dominant foods, but there was a remarkably tight ad libitum intake of protein during both diets; because the participants ate more carbs and fats while eating the processed diet to achieve this same protein intake, more total calories had to be eaten. 

If anything, I think the energy density, fiber, and protein differences in the real world would make this effect more pronounced. Yes, I think the 500kcal difference observed in this study is probably a conservative estimation of the true difference in the real world, because in the real world, there aren’t researchers trying to match the energy density, protein, and fiber intakes of what’s presented to you. 

On top of these mechanisms, while the processed diet wasn’t rated as more palatable or tasty or satisfying, it was eaten more quickly and the hormonal responses indicated less satiation and more hunger during the processed diet. This could be due to mouthfeel and texture of processed foods making it easier to eat them more quickly, and thus, more calories can be consumed before satiety “sets in.” It’s also possible that the questions related to hunger and satiety weren’t sensitive or targeted enough to represent the qualitative eating experience differences between diets. Another perspective to consider is that the goal of a food manufacturer is to encourage the consumption of the food they manufacture. This is actually not accomplished by making something super hedonically satisfying, but rather tasty, but not so tasty to cross the threshold into satisfaction, which discourages further consumption (an example would be high fructose corn syrup, which is only slightly sweeter than regular sugar; as most have a 55/45 breakdown of fructose to glucose versus the 50/50 split of standard sugar). Thus, the qualitative descriptors in this study of “pleasantness” and “familiarity” and “satisfaction” might not have picked this up. Nonetheless, based on hormonal data, the participants were more full and less hungry eating the unprocessed diet, and they also ate slower, which might have allowed more time for these hormonal signals to be received, resulting in earlier meal cessation and lower calorie consumption.

According to the authors, protein could only explain, at most, 50% of the difference in energy intake between groups. Thus, all in all, energy density, eating speed, and hormonal differences explain the remaining differences (and perhaps other unmeasured factors). The good news is that unprocessed diets rock. They result in effortless fat loss; eating a whole food diet for the average person will result in higher levels of satiety, less hunger, and subsequently a lower energy intake, thus leading to fat loss. This will of course only last so long, but your “settling point” of adiposity will almost certainly be lower when consuming an unprocessed versus a processed diet. If you aren’t sure what that looks like, think lots of fruits, vegetables, lean proteins, and (in general) a diet dominated by single ingredient foods. I know, groundbreaking stuff.

Interesting nuances of the findings

There were some outcomes which, at first glance, seem like head scratchers. For example, in a 2010 study, energy expenditure was shown to be lower acutely after consuming a processed versus an unprocessed sandwich. Yep, another knock against processed diets is they might reduce the energy-out side of the equation, likely via a reduction in the thermic effect of food and the cost to get the metabolizable energy content of the food (5). Because they’ve undergone substantial industrial processing, our body doesn’t need to expend as many calories to actually digest and metabolize highly processed food. However, in the present study, energy expenditure was higher during the processed diet, almost reaching the cut off for significance (p = 0.06). While this appears contradictory, I don’t think it is. In the prior study, the response to a calorie-matched meal was assessed, but in the present study, the researchers measured the energy expenditure response to an entire diet for 14 days, compared to another diet that was 500kcals higher. Thus, in the present study, we are likely just observing the ramping up of energy expenditure in response to a calorie surplus, which has been shown as a normal response to overfeeding (6), and likely an attempt by the body to maintain metabolic homeostasis (something that is easily overcome at a population level by our obesogenic environment in the modern world). So unfortunately, processed diets may still very well reduce energy expenditure when calorie matched compared to unprocessed diets.

Another interesting finding was that while the only significant body composition changes were changes in body fat, changes in lean mass were almost significant.  There was a nonsignificant, small drop in lean mass during the unprocessed diet and a small increase during the processed diet. However, for one, there is an obligatory loss or gain of lean mass when fat mass changes, as adipose tissue is not completely composed of fat mass and has some lean tissue as part of its structure (7). Further, these changes in lean mass were significantly associated with changes in sodium intake (r = 0.63; p = 0.004), likely indicating that changes in extracellular fluid shifts appeared as lean tissue changes, contributing to this outcome.

As a final, unfortunate, nuanced note on this study, the cost of the processed diet per kcal was ~50% less than the unprocessed diet. Meaning, it’s cheaper to get your energy from processed foods, which doesn’t bode well for society-level behavior change.

Next Steps

I would love to see this field of research expanded into a more applicable phase. Much of Hall’s research is groundbreaking, but stems from the research battleground between carbs versus fat and carbs versus calories, which are – in my opinion – largely unproductive. No, the insulin model of obesity isn’t correct (8); yes, energy balance ultimately dictates changes in body mass (9). But, unfortunately, elucidating these points rarely provides anything actionable. Qualitative interviews around barriers to adopting unprocessed diets, assessments of how to intervene to change behavior, and developing diagnostic tools to assess which behaviors are the largest contributors to an individual’s adiposity to then provide individualized counseling are the directions I’d like to see future research go. Essentially, if we know that processed diets – even in a controlled research setting that likely dampens their effect on overeating – result in substantially more ad libitum food intake, what can we do to make it more likely that people will eat unprocessed diets?

Application and Takeaways

• Processed diets resulted in significant increases in ad libitum consumption of energy. In spite of eating ~500kcal more when consuming a processed diet, participants in this study felt the same level of hunger and satiation on both diets, and subsequently gained body fat on the processed diet. 
• This is primarily due to the following combined aspects of processed diets: they have a higher energy density, less protein per calorie, less fiber per calorie, they are more likely to be eaten quickly, and subsequently, they are less satiating. Also, it’s cheaper to get the same amount of calories from processed foods versus unprocessed, though that didn’t factor into the results of this study. 
• On the flip side, unprocessed, whole food diets result in significant decreases in energy consumption and subsequently, fat loss. This is due to lower energy density, more protein per calorie, more fiber per calorie, slower eating times, and subsequently, more satiation. The downside is that whole food is more expensive.
• As a final note, to help with application, check out the table with the NOVA classification system so you can get a better idea of how processed and unprocessed foods are categorized.

References

1. Hall KD, Ayuketah A, Brychta R, Cai H, Cassimatis T, Chen KY, Chung ST, Costa E, Courville A, Darcey V, Fletcher LA. Ultra-Processed Diets Cause Excess Calorie Intake and Weight Gain: An Inpatient Randomized Controlled Trial of Ad Libitum Food Intake. Cell metabolism. 2019 May 16.
2. Monteiro CA, Cannon G, Levy RB, Moubarac JC, Louzada ML, Rauber F, Khandpur N, Cediel G, Neri D, Martinez-Steele E, Baraldi LG. Ultra-processed foods: what they are and how to identify them. Public health nutrition. 2019 Apr;22(5):936-41.
3. DellaValle DM, Roe LS, Rolls BJ. Does the consumption of caloric and non-caloric beverages with a meal affect energy intake? Appetite. 2005 Apr 1;44(2):187-93.
4. Steele EM, Raubenheimer D, Simpson SJ, Baraldi LG, Monteiro CA. Ultra-processed foods, protein leverage and energy intake in the USA. Public health nutrition. 2018 Jan;21(1):114-24.
5. Barr S, Wright J. Postprandial energy expenditure in whole-food and processed-food meals: implications for daily energy expenditure. Food & nutrition research. 2010 Jan 1;54(1):5144.
6. Levine JA, Eberhardt NL, Jensen MD. Role of nonexercise activity thermogenesis in resistance to fat gain in humans. Science. 1999 Jan 8;283(5399):212-4.
7. Abe T, Dankel SJ, Loenneke JP. Body Fat Loss Automatically Reduces Lean Mass by Changing the Fat-Free Component of Adipose Tissue. Obesity (Silver Spring, Md.). 2019 Mar;27(3):357.
8. Hall KD, Guyenet SJ, Leibel RL. The carbohydrate-insulin model of obesity is difficult to reconcile with current evidence. JAMA internal medicine. 2018 Aug 1;178(8):1103-5.
9. Hall KD, Guo J. Obesity energetics: body weight regulation and the effects of diet composition. Gastroenterology. 2017 May 1;152(7):1718-27.