Factors Modifying the Energy Value of the Diet: Why a Calorie Is Not Always a Calorie
Introduction
For decades, nutritional science and food labeling have relied on a seemingly simple premise: the energy content of food can be calculated by summing the calories contributed by proteins, fats, and carbohydrates using universal conversion factors. These factors, primarily derived from the work of Atwater in the late 19th century, assign a fixed energy value to each macronutrient (4 kcal/g for protein and carbohydrate, 9 kcal/g for fat) regardless of the food source (Capuano et al., 2018). This approach has become the global standard, allowing consumers to compare products and dietitians to plan menus. However, a growing body of evidence suggests that this premise is fundamentally flawed. The amount of energy actually available to the human body — the metabolizable energy —can differ significantly from the theoretical value calculated by Atwater factors.
The discrepancy between a bomb calorimeter, which measures the total chemical energy released upon complete combustion of food, and the human digestive system is considerable. Unlike a calorimeter, our gastrointestinal tract is not a perfectly efficient furnace. We lack the enzymes to digest certain structural carbohydrates, our digestive secretions are variable, and the physical form of food can drastically limit enzyme access. As Capuano et al. (2018) highlight, the food matrix—the natural structural barriers within plant tissues—plays a paramount role. For instance, the cell walls of almonds, composed of indigestible dietary fiber, can encapsulate intracellular lipids, rendering them inaccessible to pancreatic lipase. Consequently, a significant portion of the fat in whole nuts passes through the small intestine undigested, leading to a measured metabolizable energy value that is up to 30% lower than the label claim (Capuano et al., 2018; Baer et al., 2016).
This phenomenon is not limited to nuts. The evolutionary significance of food processing underscores how thermal treatment fundamentally alters nutrient bioavailability. Carmody & Wrangham (2009) emphasize that cooking, while often increasing the digestibility of starch and protein through gelatinization and denaturation, can also reduce energy gain through mechanisms such as the Maillard reaction (which blocks lysine residues) and the loss of fat during roasting. Similarly, the way food is physically structured—from the intact cells of a cooked legume to the aerated network of a bread crumb—determines the kinetics of digestion and the extent of nutrient absorption (Grundy et al., 2016).
In this article, we will move beyond the simplistic concept of energy balance to explore the specific factors that modify the energy value of the human diet. We will not be discussing the thermodynamics of energy equilibrium or the homeostatic regulation of body weight. Instead, we will focus on the following key modifiers: (1) the structural integrity of the food matrix (cell walls and biopolymer networks), (2) the effects of industrial and domestic processing (thermal and non-thermal), (3) the role of macronutrient interactions and anti-nutritional factors, and (4) the influence of the gut microbiota on colonic fermentation. By understanding these factors, we can appreciate why a diet based on whole, minimally processed foods may provide significantly less usable energy than a diet of equivalent chemical composition consisting of refined and ultra-processed products.
1. The Food Matrix and Macrostructure: Nutrient Capsules
The journey of food from the plate to the bloodstream is not a simple dissolution of nutrients into a reactive soup. Instead, it is a physical battle of disassembly, where the body must break down intricate biological structures to access the energy stored within. The food matrix—the physical and chemical organization of a food—is the first major modifier of dietary energy value. In plant-based foods, this matrix is primarily defined by the cell wall, a rigid, indigestible barrier that acts as a natural capsule for intracellular nutrients.
1.1 Plant Cell Walls as the Primary Barrier
The cell walls of edible plants are supramolecular composites of cellulose, hemicelluloses, pectins, and often lignin (Grundy et al., 2016). Crucially, humans lack the endogenous enzymes (cellulases, hemicellulases) required to hydrolyze these polysaccharides. Consequently, cell walls remain largely intact throughout the upper gastrointestinal tract, passing through the stomach and small intestine without significant degradation. This has profound implications for the nutrients trapped inside.
For an intracellular nutrient—whether it is a lipid droplet in an almond, a starch granule in a wheat endosperm, or a protein body in a legume—to be digested, it must first encounter pancreatic enzymes. This encounter can only happen if the cell wall is either physically ruptured (releasing the contents) or sufficiently porous to allow enzyme diffusion. As Grundy et al. (2016) explain, the degree of cell wall encapsulation at the moment of swallowing is the single most important factor determining the digestibility of macronutrients in plant foods. If cells remain intact, the encapsulated nutrients are effectively shielded from enzymatic attack.
Direct evidence for this mechanism comes from human ileostomy studies, where effluent can be collected directly from the terminal ileum. Capuano et al. (2018) cite research showing that after consumption of white beans, intact cells filled with starch can be recovered in ileal effluent. Similarly, large particles of almond tissue recovered from ileostomy volunteers show a core of intact cells where lipids remain untouched, while only the ruptured cells at the surface are digested. This encapsulation mechanism explains why the metabolizable energy of whole nuts is consistently lower than predicted: a substantial fraction of lipids literally never come into contact with lipase.
1.2 Rupture vs. Separation: The Fate of Cells during Mastication
Mastication (chewing) is the first physical process that determines the extent of cell wall disruption. The mechanical behavior of plant tissue varies dramatically between food types, leading to two distinct outcomes: cell rupture or cell separation.
Hard, brittle plant tissues, such as raw nuts and seeds, tend to fracture along cellular boundaries and directly through cells. Grundy et al. (2016) describe how mastication of almonds produces particles that contain a mixture of ruptured cells (where the contents are exposed) and fully intact cells (where contents remain encapsulated). The proportion of ruptured cells is directly related to the particle size distribution achieved during chewing. Smaller particles have a higher surface area-to-volume ratio and thus a greater fraction of fractured cells. However, even particles as small as 1-2 mm can contain thousands of intact cells. Conversely, softer tissues, such as cooked legumes (chickpeas, beans), exhibit cell separation rather than rupture. During cooking, the pectic polysaccharides that glue adjacent cells together are solubilized, allowing cells to slide apart without breaking their walls. In this case, whole, intact cells are released into the digestive stream. These separated cells remain entirely resistant to digestion because their cell walls are impermeable to amylase and lipase (Grundy et al., 2016).
This biophysical distinction has a quantifiable impact on energy availability. For a 2 mm particle of wheat endosperm, it is estimated that approximately 62% of starch remains encapsulated within intact cells, while only 38% is exposed on fractured surfaces (Capuano et al., 2018). For a 2 mm almond particle, containing hundreds of thousands of parenchyma cells, the vast majority of lipid remains inaccessible.
1.3 Macrostructural Integrity in Grains and Legumes
The concept of encapsulation extends beyond nuts to staple crops. In cereals and legumes, the cell wall not only acts as a physical barrier to enzymes but also limits the physical transformation of the starch during cooking. For starch to be efficiently digested, it must undergo gelatinization—a process where water and heat disrupt the semi-crystalline granule structure. However, if starch granules are encapsulated within intact cell walls, water ingress is restricted, and the granules cannot fully swell and gelatinize (Capuano et al., 2018; Grundy et al., 2016). This results in partially gelatinized, distorted starch granules that are inherently less susceptible to α-amylase.
The degree of milling or processing dramatically alters this effect. Whole grains, where the endosperm cells remain largely intact, produce a slower and lower glycemic response compared to refined flours, where cell walls are completely destroyed. Louis et al. (2021) review how the physical form of dietary fiber influences microbial access, noting that intact cell wall fragments provide a different fermentation substrate than purified soluble fibers. This structural integrity is the basis for the classification of resistant starch type 1 (RS1)—starch that is physically inaccessible because it is trapped within intact cell walls.
1.4 Texture, Eating Rate, and Passive Overconsumption
Energy intake is positively correlated with energy intake rate. Higher meal energy intake rates (kcal/min) are associated with greater ad libitum energy consumption across all study conditions. Data represent each main meal and dessert per subject.
Adapted from Lasschuijt et al. (2023), Figure 5. CC BY 4.0
Food texture, a direct consequence of food matrix structure, plays a unique role in modifying energy intake not through digestion, but through its effect on eating behavior. Hard, chewy, or viscous foods require more oral processing (chewing) per gram consumed. This slower eating rate prolongs oro-sensory exposure—the time during which taste receptors are stimulated.
As Lasschuijt et al. (2023) demonstrate in a controlled trial, consuming a diet of hard-textured foods compared to soft-textured foods of identical energy density and macronutrient composition led to a 33% reduction in daily energy intake. The mechanism is straightforward: softer foods are consumed faster, allowing for a greater number of calories to be ingested before satiation signals (fullness) have time to develop. This phenomenon, often termed "passive overconsumption," is particularly relevant to many modern ultra-processed foods, which are engineered to have soft textures and high eating rates.
In contrast, hard-textured foods, even when industrially processed, slow down the rate of energy intake. Lasschuijt et al. (2023) found that the energy intake rate (kcal/min) was 82% higher for soft-textured meals compared to hard-textured meals. This effect is independent of the processing level (minimally processed vs. ultra-processed). Therefore, texture manipulation—choosing whole apples over applesauce, intact nuts over nut butters—represents a powerful behavioral strategy to reduce energy intake without altering the chemical composition of the diet.
1.5 Summary: The Capsule Principle
The food matrix, particularly the plant cell wall, acts as a natural capsule that determines the bioaccessibility of macronutrients. The degree to which these capsules are broken depends on the food's mechanical properties, masticatory efficiency, and cooking method. Intact cells pass through the small intestine undigested, delivering encapsulated nutrients to the colon for fermentation (with lower energy yield) or to excretion. This principle explains why whole almonds, walnuts, pistachios, and intact grains consistently provide less metabolizable energy than their processed counterparts (Baer et al., 2016; Capuano et al., 2018). Understanding the capsule principle is the first step in appreciating that a calorie is not a calorie: the same chemical energy, locked inside an intact cell wall, is fundamentally less available to the human body than energy released from a ruptured cell or a refined flour.
2. Technological and Culinary Processing: The Dual Role of Heat
The application of heat is arguably the most transformative intervention in the history of human food preparation. From the discovery of fire to the modern extrusion cooker, thermal processing has been used to render food safe, palatable, and digestible. However, heat is a double-edged sword. While it can unlock nutrients and drastically increase metabolizable energy, it can just as readily degrade essential compounds, generate indigestible complexes, and alter food structure in ways that reduce energy availability. Understanding this duality is essential for evaluating the true energy value of cooked and manufactured foods.
2.1 The Positive Effects of Heat: Unlocking Energy
For much of human evolution, the primary benefit of cooking was a net gain in energy. Raw foods, particularly plants, are notoriously inefficient energy sources. Carmody & Wrangham (2009) present compelling physiological evidence that humans cannot maintain body weight or reproductive health on wholly raw diets, even when those diets are varied and energy-dense. Cooking externalizes part of the digestive process, effectively pre-digesting foods in ways that our own enzymes cannot.
Starch Gelatinization
The most dramatic effect of moist heat on plant foods is the gelatinization of starch. Raw starch granules are semi-crystalline structures that resist hydrolysis by α-amylase. When heated in the presence of sufficient water (typically above 60-70°C), the hydrogen bonds stabilizing the crystalline structure break, allowing water to penetrate the granule. The granule swells irreversibly, loses its crystallinity, and becomes amorphous. In this gelatinized state, starch is rapidly and almost completely hydrolyzed by pancreatic α-amylase into glucose (Capuano et al., 2018; Carmody & Wrangham, 2009).
The nutritional significance of gelatinization is enormous. Capuano et al. (2018) report that cooking increases ileal starch digestibility from 71% to 96% in wheat, and from 47% to 98% in green bananas. This transformation turns an indigestible or slowly digested energy source into a rapidly available one. However, the extent of gelatinization depends critically on water availability and the integrity of the encapsulating matrix. Starch locked within intact plant cell walls may only partially gelatinize, remaining significantly less digestible than fully gelatinized, free starch.
Protein Denaturation and Inactivation of Anti-nutrients
Heat also improves protein digestibility. Native proteins are folded into complex three-dimensional structures (tertiary and quaternary conformations) that can shield peptide bonds from proteolytic enzymes (trypsin, pepsin, chymotrypsin). Thermal denaturation unfolds these structures, exposing the peptide backbone and rendering proteins more susceptible to hydrolysis (Khalid et al., 2023).
Furthermore, many plant foods contain heat-labile anti-nutritional factors. Trypsin inhibitors, abundant in legumes (soybeans, lentils) and some cereals, are proteins that specifically block the active site of trypsin. Similarly, amylase inhibitors interfere with starch digestion. Moist heat effectively denatures and inactivates these inhibitors, thereby increasing the overall digestibility of both protein and starch (Capuano et al., 2018). Conversely, dry heat (roasting, toasting) may be less effective at inactivating these compounds.
2.2 The Negative Effects of Heat: Losses and Indigestible Complexes
While moderate heat enhances energy availability, excessive or inappropriate thermal processing can have the opposite effect, reducing the metabolizable energy of foods and even generating harmful compounds.
The Maillard Reaction and Protein Cross-linking
The Maillard reaction, a non-enzymatic condensation between reducing sugars and free amino groups (primarily from lysine and arginine), is responsible for the desirable flavors, aromas, and brown colors of baked, roasted, and fried foods. However, from an energy perspective, the Maillard reaction reduces protein digestibility. When lysine or arginine residues are bound to sugar moieties, they become unavailable to trypsin, which requires free ε-amino groups for cleavage (Capuano et al., 2018; Khalid et al., 2023).
Moreover, intensive heating promotes the formation of high-molecular-weight protein aggregates called melanoproteins. These aggregates are resistant to proteolysis. Khalid et al. (2023) note that the protein digestibility of bread crust, which is exposed to temperatures exceeding 180°C, is significantly lower than that of bread crumb. Similarly, prolonged high-temperature cooking of meat can lead to the formation of cross-linked "limit peptides" that escape complete hydrolysis (Khalid et al., 2023). The net effect is a reduction in the proportion of ingested protein that is absorbed as amino acids.
Lipid Oxidation and Reduced Fat Digestibility
Fats, while generally stable during gentle heating, are susceptible to oxidation and polymerization at high temperatures (deep-frying, prolonged roasting). The formation of polar oxidation products (aldehydes, ketones, dimers, polymers) alters the chemical structure of triglycerides. Capuano et al. (2018) cite studies in rats showing that the digestibility of thermally treated sunflower oil can be up to 30% lower than that of fresh oil. It is unclear whether this is due to poor absorption of the oxidized products themselves or to their inhibitory effect on pancreatic lipase. Additionally, cooking can cause physical damage to the structural barriers (cell walls) that encapsulate lipids, but paradoxically, it can also denature surface proteins (oleosins) on oil bodies, potentially facilitating lipase access (Grundy et al., 2016).
Thermal Degradation of Vitamins
Heat-labile vitamins, particularly water-soluble ones (vitamin C, thiamin, folate, B6) and certain fat-soluble vitamins (vitamin A, vitamin E), are significantly degraded during thermal processing. Boiling can cause leaching into cooking water, while high-temperature frying and baking can cause direct thermal destruction. Arowora et al. (2025) note that steaming and microwaving, which use less water and shorter heating times, retain significantly more vitamins compared to boiling and frying. This loss of micronutrients does not directly affect macronutrient energy value but is critical for overall metabolic health.
2.3 Cooling and Retrogradation: The Case of Leftover Starches
The effect of heat on starch is not always permanent. When cooked, gelatinized starch is allowed to cool, the amylose and amylopectin chains begin to re-associate into ordered, crystalline structures. This process, known as retrogradation, is essentially a partial reversal of gelatinization. Retrograded starch is resistant to α-amylase and is classified as resistant starch type III (RS3) (Capuano et al., 2018; Louis, 2021).
The practical implication is profound: cooked and cooled potatoes, rice, or pasta provide significantly less metabolizable energy than the same food consumed immediately after cooking. Capuano et al. (2018) cite an ileostomy study where only 3% of starch in freshly cooked potatoes was not hydrolyzed, whereas 12% of starch in cooled potatoes was recovered in the ileal effluent. This means that a serving of cold potato salad or reheated pasta effectively has a lower caloric value than a freshly boiled potato, even though the chemical composition is identical. The retrograded starch passes to the colon, where it is fermented by the microbiota into short-chain fatty acids, yielding less energy for the host (approximately 2 kcal/g instead of 4 kcal/g).
2.4 Processing Paradigms: From Traditional to Industrial
While the principles of heat transfer apply universally, the scale and conditions of industrial processing differ markedly from domestic cooking. Extrusion, used to produce breakfast cereals and snack foods, combines high heat, pressure, and mechanical shearing. This not only gelatinizes starch but also disrupts molecular interactions, often increasing digestibility beyond that achieved by simple boiling. However, the addition of lipids during extrusion can form amylose-lipid complexes (resistant starch type V), which may reduce digestibility (Capuano et al., 2018). Frying, as noted, introduces large amounts of exogenous fat that is absorbed into the food matrix, simultaneously increasing energy density while potentially reducing fat quality through oxidation.
In summary, heat is a powerful but context-dependent modifier of food energy value. It can unlock energy (gelatinization, denaturation, anti-nutrient inactivation) or lock it away (Maillard reaction, protein aggregation, lipid oxidation, retrogradation). The net effect depends on the temperature, duration, water activity, and the specific food matrix. Recognizing this duality allows for informed choices in food preparation, from selecting gentle steaming over deep-frying to understanding the caloric consequence of letting a pot of pasta cool overnight.
3. Chemical Composition and Nutrient Interactions
Beyond physical structure and thermal history, the intrinsic chemical composition of food and the interactions between different nutrients profoundly modify energy availability. The body does not treat an isocaloric mix of glucose, fructose, starch, and fiber equally. The specific type of carbohydrate, the conformation of protein molecules, the degree of fatty acid saturation, and the presence of anti-nutritional factors all influence how much energy is ultimately extracted from a meal. Furthermore, nutrients do not act in isolation; they interact with each other and with the gut environment in ways that can either facilitate or hinder digestion and absorption.
3.1 Carbohydrates: Beyond the Simple/Complex Dichotomy
The conventional classification of carbohydrates as “simple” (sugars) and “complex” (starches and fibers) is insufficient to explain their differential impact on metabolizable energy. The key variables are digestibility, absorption rate, and fermentability.
Resistant Starch (RS): The Fraction That Escapes
Starch is not a single entity. A significant fraction of starch in typical diets resists digestion in the small intestine and passes to the colon. This is termed resistant starch (RS), and it yields substantially less energy (approximately 2 kcal/g) than fully digestible starch (4 kcal/g) because it is fermented by colonic bacteria rather than absorbed as glucose.
Capuano et al. (2018) and Louis et al. (2021) describe five types of RS, each arising from different mechanisms:
- RS1 (Physically inaccessible): Starch trapped within intact cell walls (e.g., whole grains, seeds, legumes). The cell wall acts as a physical barrier to α-amylase.
- RS2 (Native granular): Raw starch granules with a compact, crystalline structure that resists gelatinization (e.g., raw potatoes, green bananas).
- RS3 (Retrograded): Starch that has been cooked and then cooled, causing the amylose and amylopectin chains to recrystallize (e.g., cooked and cooled potatoes, pasta, rice).
- RS4 (Chemically modified): Starches that have been chemically cross-linked or esterified during industrial processing.
- RS5 (Amylose-lipid complexes): Helical complexes formed between amylose and fatty acids or other lipids.
The nutritional consequence is profound. A diet rich in RS provides less metabolizable energy than a diet with the same total starch content but in a fully digestible form. Moreover, RS is a prebiotic substrate that feeds beneficial gut bacteria, leading to the production of short-chain fatty acids (SCFAs), which have systemic health benefits but lower caloric value for the host (Louis et al., 2021).
Dietary Fiber: Not Just Ballast
Non-starch polysaccharides (NSP) and lignin—collectively known as dietary fiber—are the principal components of plant cell walls. Humans lack endogenous enzymes to hydrolyze β-linkages (e.g., β(1→4) in cellulose). Therefore, fiber contributes negligible direct energy in the small intestine. However, soluble fibers (pectins, β-glucans, guar gum) can increase digesta viscosity, slowing down enzyme diffusion and nutrient absorption (Grundy et al., 2016). Insoluble fibers (cellulose, many hemicelluloses) primarily act as bulking agents and physical barriers.
When fiber (and RS) reaches the colon, it is fermented by the gut microbiota. The resulting SCFAs (acetate, propionate, butyrate) are absorbed and can provide approximately 2 kcal per gram of fiber, compared to the 4 kcal/g for digestible carbohydrate (Capuano et al., 2018). This is why the Atwater factors were later adjusted to include a lower energy value for fiber (2 kcal/g in the EU, 4 kcal/g in the US for labeling purposes, despite the lower actual yield). The discrepancy between chemical energy (bomb calorimeter) and metabolizable energy is largest for high-fiber foods.
3.2 Proteins: Structure, Anti-nutrients, and Amino Acid Blocking
The digestibility of dietary protein varies widely depending on its source and the presence of interfering compounds.
Protein Conformation and Aggregation
Native globular proteins (e.g., in raw eggs) have tightly folded structures that partially protect peptide bonds from proteases. Thermal denaturation unfolds these structures, dramatically increasing digestibility (Carmody & Wrangham, 2009). Conversely, excessive heating can cause protein aggregation and cross-linking (e.g., via disulfide bonds or Maillard reaction products), which renders proteins less digestible (Khalid et al., 2023; Capuano et al., 2018).
In plant foods, proteins are often sequestered within protein bodies—intracellular organelles that may have their own membranes and structural proteins. The integrity of these bodies upon cooking and mastication influences protein bioaccessibility. Capuano et al. (2018) note that in raw almonds, a large portion of protein remains undigested, and digestibility increases significantly when the matrix is finely ground.
Anti-nutritional Factors
Many plant foods contain compounds that directly inhibit protein digestion:
- Trypsin inhibitors: Found in legumes (soybeans, lentils, chickpeas) and some cereals. These small proteins bind to the active site of trypsin, reducing proteolysis. Heat treatment effectively denatures and inactivates them (Capuano et al., 2018).
- Phytates (phytic acid): Present in cereal bran, nuts, and seeds. While primarily known for chelating minerals (iron, zinc, calcium), phytates can also form complexes with proteins, reducing their solubility and susceptibility to proteases.
- Tannins and polyphenols: Found in tea, coffee, berries, and some legumes. These compounds can bind to dietary proteins (and digestive enzymes) through hydrophobic interactions and hydrogen bonding, forming indigestible complexes and reducing overall protein digestibility (Capuano et al., 2018).
- Glucosinolates and isothiocyanates: Present in Brassica vegetables (broccoli, cabbage). Isothiocyanates, the breakdown products, are electrophilic and can react with amino acids and proteins, potentially decreasing their digestibility.
3.3 Lipids: Emulsification, Encapsulation, and the Calcium Effect
Fat digestion is generally efficient (≥95% absorption), but certain factors can substantially reduce its metabolizable energy.
Physical Barriers to Lipase
As discussed in Section 1, the primary mechanism limiting lipid digestibility is physical encapsulation within intact plant cell walls. This has been demonstrated repeatedly for almonds, walnuts, and pistachios, where metabolizable energy is 5-30% lower than predicted (Baer et al., 2016; Capuano et al., 2018). Furthermore, in processed foods like cheese, the protein matrix (casein network) can encapsulate fat globules, restricting access to gastric and pancreatic lipases (Capuano et al., 2018).
The Role of Bile Salts and Emulsification
Lipid digestion requires emulsification into small droplets to increase the surface area for lipase. Bile salts are amphipathic molecules that facilitate this process. Some soluble dietary fibers (e.g., pectins, β-glucans, guar gum) can bind bile salts or increase digesta viscosity, thereby reducing the efficiency of emulsification and micelle formation. This can lead to increased fecal fat excretion (Grundy et al., 2016).
Calcium and Fatty Acid Soaps
Dietary calcium, particularly from dairy products, can reduce fat absorption by forming insoluble calcium soaps (calcium salts of long-chain fatty acids) in the intestinal lumen. Capuano et al. (2018) cite a meta-analysis showing that an additional 1241 mg/d of dairy calcium increased fecal fat excretion by approximately 5.2 g/d, equivalent to about 47 kcal/d. This mechanism partly explains why high-calcium diets (e.g., from cheese) may be associated with lower weight gain over time.
Fatty Acid Chain Length and Saturation
Medium-chain triglycerides (MCTs) are absorbed directly into the portal blood without requiring chylomicron formation, and they may increase diet-induced thermogenesis (DIT) or TEF slightly more than long-chain triglycerides (LCTs) (Tzeravini, 2024). However, the practical effect on total metabolizable energy is small. There is also evidence that polyunsaturated and monounsaturated fatty acids (PUFA, MUFA) are more readily oxidized and may produce a higher DIT compared to saturated fatty acids (SFA), although results are inconsistent (Tzeravini, 2024).
3.4 Nutrient Interactions: Synergy and Antagonism
Nutrients do not act independently. Their interactions create emergent effects on energy metabolism.
- Fiber–Enzyme Interactions: Soluble fibers (guar gum, pectins) can inhibit α-amylase and lipase activity directly through non-competitive or mixed-type inhibition, in addition to their viscosity effects (Grundy et al., 2016).
- Protein–Lipid Complexes: Amylose-lipid complexes (RS5) reduce starch digestibility. Similarly, proteins can interact with polyphenols, and lipids can be trapped within protein networks (e.g., in cheese).
- Phytate–Mineral–Protein: Phytic acid binds divalent cations (Fe2+, Zn2+, Ca2+) and also forms insoluble complexes with proteins at low pH, reducing both mineral and protein bioavailability.
- Carbohydrate–Protein (Maillard): Advanced glycation end-products (AGEs) reduce protein digestibility.
3.5 The Thermic Effect of Food (TEF) as a Modifier
While not strictly a factor of food composition alone, the thermic effect of food—the energy expended during digestion, absorption, and assimilation—is directly influenced by macronutrient composition. TEF accounts for approximately 10% of total energy expenditure but varies widely:
- Protein: 20-30% of energy consumed is lost as heat.
- Carbohydrate: 5-10%.
- Fat: 0-3%.
Thus, a high-protein meal will have a lower net metabolizable energy than a high-fat meal of equal gross caloric value because more energy is used to process the protein (Westerterp et al., 1999; Tzeravini, 2024). This is one reason why protein-rich diets can be effective for weight management: they increase satiety and postprandial thermogenesis, reducing net energy available for storage.
3.6 Summary: The Chemical Complexity of Energy Extraction
The chemical composition of food is not a static predictor of energy value. The specific molecular forms of carbohydrates (RS vs. digestible starch), the presence of anti-nutrients (trypsin inhibitors, phytates, tannins), the degree of unsaturation of fatty acids, and the interactions between nutrients all modulate digestibility and metabolizable energy. Furthermore, the obligatory energy cost of processing different macronutrients (TEF) further reduces net energy gain. A refined, high-fat, low-fiber diet will generally provide a higher net energy yield than a whole-food, plant-based diet with identical gross chemical composition, simply because more of the energy is immediately available and less is lost to fermentation, fecal excretion, or thermogenesis.
4. Bioaccessibility and Digestive Effects
The journey of food through the gastrointestinal tract is not a passive conduit; it is a series of dynamic, physico-chemical transformations. Bioaccessibility—the fraction of a nutrient that is released from the food matrix and becomes available for absorption—is determined by the interplay of enzymatic activity, gut motility, pH, and the resident microbial community. Even when nutrients are liberated from the food matrix, their ultimate absorption and metabolic fate depend on the efficiency of the small intestine and, for resistant fractions, on the fermentative activity of the colonic microbiota. This section explores these digestive-level factors that further modify the energy value of the diet.
4.1 The Role of the Gut Microbiota in Colonic Fermentation
The human colon harbors a dense and diverse community of microorganisms—the gut microbiota—that possess enzymes absent from the human genome. Chief among these are glycoside hydrolases and polysaccharide lyases capable of breaking down dietary fiber, resistant starch, and other non-digestible carbohydrates (Louis et al., 2021). This microbial metabolism has two major consequences for energy availability.
Short-Chain Fatty Acids (SCFAs) as Energy Substrates
When the microbiota ferments complex carbohydrates, the primary end-products are the short-chain fatty acids acetate, propionate, and butyrate. These SCFAs are absorbed across the colonic epithelium and enter host metabolism. However, the energy yield from fermentation is substantially lower than from small intestinal digestion. Whereas digestible carbohydrate yields approximately 4 kcal/g (as glucose oxidized to CO2 and water), fermentable carbohydrate yields approximately 2 kcal/g, primarily as SCFAs that are used locally by colonocytes (butyrate) or delivered to the liver (acetate, propionate) for further oxidation (Capuano et al., 2018; Louis et al., 2021). This discrepancy explains why the Atwater factors assign a lower energy value to dietary fiber.
Microbial Cross-Feeding and Metabolic Cooperation
Main routes of metabolic cross-feeding of dietary fibre by the human gut microbiota. Primary degraders break down complex fibres into oligosaccharides and monosaccharides, which are then fermented to short-chain fatty acids (SCFAs). Cross-feeding occurs through metabolites such as lactate, formate, and hydrogen.
Adapted from Louis et al. (2021), Figure 1. CC BY 4.0
The degradation of complex dietary fiber is rarely accomplished by a single microbial species. Instead, it involves a consortium of primary degraders, secondary fermenters, and cross-feeders (Louis et al., 2021). For example, Ruminococcus bromii is a keystone species for the degradation of resistant starch (RS). It partially hydrolyzes RS into maltose and glucose, which it does not fully utilize itself, thereby releasing these simpler sugars for other bacteria. Similarly, butyrate-producing Firmicutes (e.g., Eubacterium rectale, Roseburia spp.) rely on cross-feeding of lactate or acetate generated by Bifidobacteria or other primary fermenters. This metabolic network ensures that energy is extracted from fiber at the community level, but the efficiency of this process—and the proportion of ingested energy that is lost as heat, methane, or hydrogen—varies depending on the composition and functional capacity of an individual’s microbiota (Louis et al., 2021).
Impact of Food Processing on Microbiota Composition
The physical and chemical modifications induced by food processing directly influence which microbial species thrive. Diets rich in ultra-processed foods, which are typically low in fermentable fiber and high in emulsifiers and artificial additives, have been associated with reduced microbial diversity, loss of SCFA-producing bacteria, and increased gut permeability (dysbiosis). Conversely, minimally processed, fiber-rich diets promote the growth of beneficial species associated with anti-inflammatory and metabolic health benefits (Arowora et al., 2025).
4.2 Ileal Brake and Nutrient Sensing
The presence of undigested nutrients in the terminal ileum triggers a feedback mechanism known as the ileal brake. This response, mediated by peptide hormones such as glucagon-like peptide-1 (GLP-1) and peptide YY (PYY), slows gastric emptying and reduces intestinal motility, thereby prolonging the exposure of chyme to digestive enzymes and increasing absorption efficiency (Grundy et al., 2016). Foods with high levels of resistant starch or encapsulated lipids, which deliver nutrients to the ileum, can activate this brake. This may reduce appetite at subsequent meals, indirectly affecting energy intake. Conversely, highly processed foods that are rapidly and completely digested in the proximal small intestine may fail to trigger the ileal brake, allowing for higher overall energy absorption and faster return of hunger.
4.3 Individual Variability in Digestive Efficiency
The factors discussed above are not uniform across individuals. Genetic, developmental, and environmental factors create substantial inter-individual variability in how efficiently energy is extracted from food.
Microbiota Composition and Function
As noted, the specific composition of the gut microbiota determines the capacity to ferment different types of fiber and resistant starch. An individual lacking keystone species like R. bromii may excrete a large fraction of ingested RS, obtaining significantly less energy from starchy meals than an individual with a robust amylolytic community (Louis et al., 2021). This has been directly demonstrated in human studies where participants with low levels of R. bromii excreted undigested starch in their feces.
Gut Transit Time
The rate at which digesta moves through the gastrointestinal tract—orocaecal transit time and colonic transit time—affects the duration of enzyme exposure and the extent of fermentation. Longer transit times generally increase digestibility and absorption, but may also increase colonic fermentation of resistant fractions. Transit time itself is influenced by diet (fiber increases bulk and accelerates transit), physical activity, and age.
Digestive Enzyme Secretion
There is considerable variation in the secretion of pancreatic enzymes (amylase, lipase, proteases) and bile acids. Genetic polymorphisms, pancreatic health, and previous dietary habits can all influence enzymatic capacity. For example, individuals with higher salivary amylase gene copy numbers digest starch more rapidly, which may influence postprandial glucose responses and potentially energy extraction.
4.4 Diet-Induced Thermogenesis (TEF) as an Energy Cost
Although not strictly a digestive process, the metabolic cost of processing food—diet-induced thermogenesis (DIT), also known as the thermic effect of food (TEF)—represents energy that is consumed (oxidized) during the digestion, absorption, and assimilation of nutrients, and therefore is not available for ATP synthesis or storage. TEF accounts for approximately 10% of total daily energy expenditure but varies with meal composition (Tzeravini, 2024).
- Protein has the highest thermogenic effect (20–30% of its energy content), due to the costs of deamination, urea synthesis, and protein turnover.
- Carbohydrate has an intermediate effect (5–10%).
- Fat has the lowest thermogenic effect (0–3%).
- Alcohol has a high thermogenic effect (10–30%).
Thus, two meals with identical gross energy content but different macronutrient proportions will result in different net metabolizable energy availability. A high-protein meal yields less net energy than a high-fat meal. Furthermore, the thermic effect is influenced by meal frequency, palatability (cephalic phase), and the individual’s insulin sensitivity and thyroid status (Tzeravini, 2024). This obligatory energy expenditure is a genuine modifier of the diet’s energy value, as it represents energy that is consumed but cannot be used for metabolic work or stored as fat.
4.5 Summary: From Matrix to Metabolome
The bioaccessibility and ultimate absorption of food energy are the products of a complex series of steps involving physical disruption, enzymatic hydrolysis, microbial fermentation, and hormonal feedback. The gut microbiota plays a critical role in extracting energy from otherwise indigestible fractions, but this process yields less energy (approximately 2 kcal/g) than small intestinal digestion (4 kcal/g). Individual variability in microbiota composition, transit time, and enzyme secretion means that the same food can provide different amounts of metabolizable energy to different people. Finally, the obligatory thermic effect of food processing reduces net energy gain, with protein requiring the greatest energy cost. Understanding these digestive-level modifiers is essential for a complete picture of dietary energy value.
5. Practical Implications: Why This Matters
The preceding sections have demonstrated that the metabolizable energy of a food is not a fixed number dictated solely by its chemical composition. Instead, it is a dynamic outcome of the food’s physical structure, its thermal history, the chemical form of its nutrients, and the complex interplay with the consumer’s digestive physiology. These insights are not merely academic curiosities; they have profound practical implications for how we assess dietary energy, formulate public health recommendations, and interpret food labels.
5.1 The Inaccuracy of Current Calorie Labeling
Most countries require food labels to display energy content calculated using the Atwater general factors (4-4-9 system) or a modified version thereof. However, as we have seen, these factors systematically overestimate the metabolizable energy of many whole plant foods, particularly nuts, seeds, legumes, and intact grains.
- Almonds: Measured energy is approximately 30% lower than Atwater-predicted values (Capuano et al., 2018).
- Walnuts: Measured energy is 21% lower (Baer et al., 2016).
- Pistachios: Measured energy is 5% lower, demonstrating nut-to-nut variability (Capuano et al., 2018).
A consumer choosing between 100 g of dry-roasted almonds (labeled as 629 kcal) and 100 g of chocolate-coated cookies (labeled as 547 kcal) might logically select the cookies for weight management. Yet the actual metabolizable energy of the almonds is only about 460 kcal—significantly lower than both the label claim and the cookies. This systematic overestimation can mislead consumers away from nutrient-dense, whole foods toward more processed, energy-dense options (Capuano et al., 2018).
5.2 Diet-Specific Discrepancies
The error introduced by using general Atwater factors is not uniform across all dietary patterns. It is most pronounced in diets rich in intact plant cells, dietary fiber, and resistant starch—precisely the diets recommended for chronic disease prevention.
Capuano et al. (2018) modeled the energy content of three diets—Western, Mediterranean, and vegan—using three different calculation systems: (1) general Atwater factors, (2) specific factors (Merrill & Watt, 1973), and (3) corrected factors based on empirical digestibility data. Key findings include:
- Western diet (highly processed, low fiber): General Atwater factors provided a slight underestimation of actual energy content (by approximately 2–3%).
- Mediterranean diet (moderate fiber, intact nuts, legumes): General factors overestimated energy by 4.2%.
- Vegan diet (high fiber, whole grains, legumes, nuts): General factors overestimated energy by 6.9%, and specific factors (which do not distinguish available carbohydrate from fiber) overestimated by 8.3%.
The overestimation was primarily driven by the low digestibility of macronutrients encapsulated within intact cell walls. For individuals following a high-fiber, plant-based diet, the actual calorie intake may be significantly lower than predicted. This may partly explain why such diets are effective for weight management without requiring conscious calorie restriction.
5.3 Dietary Strategies for Modulating Energy Intake
Understanding the factors that modify energy value allows for evidence-based strategies to reduce net calorie intake without reducing food volume or increasing hunger.
Choose Whole Over Refined
Consuming intact grains (e.g., steel-cut oats, brown rice, whole wheat berries) rather than refined flours preserves the encapsulating cell walls. This increases resistant starch (RS1 content), reduces starch gelatinization during cooking, and slows eating rate due to greater chewiness. The result is lower metabolizable energy and improved satiety per gram of carbohydrate.
Eat Nuts and Seeds in Their Intact Form
Whole almonds, walnuts, pistachios, and seeds provide significantly less metabolizable energy than their processed forms (nut butters, flours, oils). The cell walls remain largely intact during mastication and transit, encapsulating lipids and proteins. Grinding or roasting can increase digestibility by rupturing cells and denaturing surface proteins (oleosins) on oil bodies (Grundy et al., 2016; Capuano et al., 2018). Thus, for weight management, whole nuts are superior to nut pastes or butters.
Leverage Retrogradation (Cooling)
Cooking and then cooling starchy foods (potatoes, rice, pasta, legumes) increases resistant starch type III (RS3) content. This fraction escapes small intestinal digestion and is fermented in the colon, yielding approximately 2 kcal/g rather than 4 kcal/g (Capuano et al., 2018). A practical strategy: prepare potato salad, sushi rice, or cold pasta salads. Reheating may partially reverse retrogradation, but some RS3 may persist.
Favor Whole Fruits Over Juices and Smoothies
Blending or juicing disrupts cell walls, releasing sugars and making them rapidly absorbable. Whole fruits require mastication, have intact cell walls that encapsulate sugars, and provide satiety through their physical structure. Apple juice causes a rapid glycemic spike; a whole apple provides the same sugar content but with slower absorption and lower net insulin demand.
Use Fermentation to Enhance Nutrient Availability
Fermentation (e.g., sourdough bread, yogurt, kimchi, tempeh) partially hydrolyzes anti-nutritional factors (phytates, trypsin inhibitors) and pre-digests proteins and carbohydrates. This can increase the bioaccessibility of minerals and reduce the energy cost of digestion (lower TEF), but it also increases net metabolizable energy compared to the unfermented raw material. For populations requiring high energy density (e.g., athletes, malnourished individuals), fermentation is beneficial; for those seeking weight loss, the effect is less desirable.
5.4 Implications for Food Labeling and Policy
Current food labeling systems are based on outdated assumptions. Several reforms could improve accuracy and consumer trust:
- Adopt specific or corrected Atwater factors for food categories (e.g., nuts, legumes, high-fiber cereals) rather than applying general factors universally. The Merrill & Watt (1973) specific factors are available but underutilized (Capuano et al., 2018).
- Mandate digestibility-corrected energy values for products where encapsulation or resistant starch significantly alters energy availability. This would require validated in vitro or in vivo assays.
- Distinguish between available and fermentable carbohydrate on labels. Current labels often lump dietary fiber into total carbohydrate but assign it a lower energy value inconsistently between jurisdictions (2 kcal/g in EU, effectively 4 kcal/g in US labeling practice).
- Incorporate eating rate or texture information into dietary guidance. Soft, highly processed foods promote passive overconsumption; hard, fibrous foods slow intake and increase satiety (Lasschuijt et al., 2023). Simple metrics (e.g., “chews per gram” or “eating rate category”) could supplement calorie counts.
5.5 Clinical and Personal Nutrition Relevance
For individuals with specific metabolic conditions, the modifiers discussed here have direct clinical relevance:
- Type 2 diabetes and insulin resistance: Slow-release carbohydrate sources (intact grains, legumes, cooled starches) reduce postprandial glucose excursions. The lower available energy from RS and fiber may also assist with weight management.
- Obesity and weight loss: Strategies that increase fecal energy loss (encapsulated fats, RS, fiber) and increase TEF (higher protein intake) can create an energy deficit without caloric restriction. Whole nuts, seeds, and legumes are particularly valuable.
- Malabsorption and undernutrition: For individuals who require high energy density (elderly, cancer patients, athletes), processing that increases digestibility (grinding, cooking, fermentation) is beneficial. Refined flours, oils, and nut butters provide more metabolizable energy per gram than their whole counterparts.
- Inflammatory bowel diseases (IBD): Fermentable fibers can exacerbate symptoms in some patients due to gas production and osmotic effects. Individualized fiber guidance is necessary.
5.6 Summary: Moving Beyond the Calorie Myth
The notion that “a calorie is a calorie” is a convenient simplification for energy balance equations, but it fails as a biological reality. The metabolizable energy of food is modified by the food matrix, processing history, nutrient interactions, digestive efficiency, and gut microbiota composition. Public health messaging, food labeling, and clinical dietary advice must evolve to reflect this complexity. Practical strategies—choosing whole over refined, intact nuts over butters, cooled starches over freshly boiled, and whole fruits over juices—can meaningfully reduce net energy intake without requiring portion control or inducing hunger. Recognizing the factors that modify energy value empowers consumers to make evidence-based food choices that align with their metabolic health goals.
6. Conclusion
The energy value of a food is not a fixed, intrinsic property that can be deduced solely from its chemical composition. This review has shown that metabolizable energy—the energy actually available to the human body—is a dynamic outcome of multiple interacting factors. A calorie measured in a bomb calorimeter is not equivalent to a calorie processed by the human digestive system.
We have identified four major classes of modifiers that systematically alter the energy value of the diet.
First, the food matrix and macrostructure act as natural capsules for nutrients. The intact cell walls of plants, composed of indigestible dietary fiber, physically encapsulate starch, lipids, and proteins. If cells remain intact at the moment of swallowing, encapsulated macronutrients pass through the small intestine without being hydrolyzed, reaching the colon where they are fermented (with lower energy yield) or excreted. Hard, brittle tissues (nuts, seeds, raw vegetables) fracture during mastication, but even small particles contain thousands of intact cells. Soft tissues (cooked legumes) often undergo cell separation without rupture, preserving the encapsulation barrier. Food texture also governs eating rate: hard-textured foods slow consumption and reduce energy intake by up to 33% compared to soft-textured foods of identical energy density (Lasschuijt et al., 2023).
Second, thermal and industrial processing have a dual role. Moderate heating (boiling, steaming) increases energy availability by gelatinizing starch, denaturing proteins, and inactivating anti-nutrients (trypsin inhibitors, amylase inhibitors). However, excessive heat (roasting, frying, extrusion) can reduce digestibility through the Maillard reaction (blocking lysine and arginine), protein cross-linking, and lipid oxidation. Cooling cooked starchy foods promotes retrogradation, forming resistant starch type III (RS3), which provides only 2 kcal/g compared to 4 kcal/g for digestible starch (Capuano et al., 2018). The net effect of heat depends on temperature, time, water activity, and the food matrix.
Third, the chemical form of nutrients and their interactions determine the efficiency of extraction. Resistant starch (types RS1–RS5) and non-starch polysaccharides escape small intestinal digestion, yielding approximately 2 kcal/g upon colonic fermentation. Anti-nutritional factors (phytates, tannins, trypsin inhibitors) reduce protein and mineral bioavailability unless inactivated by heat. The thermic effect of food (TEF) varies dramatically: protein requires 20–30% of its gross energy for processing, carbohydrate 5–10%, fat 0–3% (Westerterp et al., 1999; Tzeravini, 2024). Thus, a high-protein meal provides significantly less net metabolizable energy than a high-fat meal of identical gross caloric value. Calcium can form insoluble fatty acid soaps, increasing fecal fat excretion (Capuano et al., 2018).
Fourth, bioaccessibility and digestive physiology introduce individual variability. The gut microbiota harbors enzymes that ferment resistant carbohydrates, but the efficiency of this process varies with microbial composition. Keystone species like Ruminococcus bromii are essential for degrading resistant starch; individuals lacking such species excrete undigested starch (Louis et al., 2021). Gut transit time, enzyme secretion, and activation of the ileal brake further modulate absorption. The obligatory energy cost of digestion (TEF) is influenced by meal composition, palatability, and individual metabolic health.
Practical implications are substantial. Current food labels, which use general Atwater factors, systematically overestimate the metabolizable energy of whole plant foods (nuts, legumes, intact grains) by 5–30% (Capuano et al., 2018; Baer et al., 2016). For a vegan or high-fiber Mediterranean diet, the overestimation amounts to 4–8% of total daily energy—a difference sufficient to influence weight management over months to years. Consumers can use this knowledge strategically: choose whole nuts over nut butters, intact grains over refined flours, cooled starches over freshly boiled, and whole fruits over juices. These choices reduce net energy intake without reducing food volume or increasing hunger.
Limitations and future directions. The evidence reviewed comes primarily from in vitro digestion models, ileostomy studies, and short-term controlled feeding trials. Long-term studies are needed to assess whether adaptations in gut microbiota or digestive efficiency occur with sustained consumption of low-bioaccessibility diets. Furthermore, individual variability in response to these modifiers suggests a role for personalized nutrition approaches.
Final conclusion. The concept that “a calorie is a calorie” is a convenient simplification for energy balance equations but fails as a biological reality. The metabolizable energy of the diet is modified by the food matrix, processing history, nutrient–nutrient interactions, and the consumer’s digestive physiology. Recognizing these factors empowers evidence-based food choices, improves the accuracy of dietary assessment, and informs more effective public health strategies. A shift from nutrient-centric to food-centric thinking—acknowledging that how a food is structured and prepared is as important as what it contains—is essential for advancing the science of nutrition and its application to human health.
References
- Anih, D.C., Arowora, K.A., Boyi, R.-H.N., Ugwuoke, K.C., Albukhari, A.F., Abah, M.A., Timothy, M. and Muhammad, Z.I. (2025) ‘Effects of Food Processing on Nutrient Bioavailability and Metabolic Health’, Asian Journal of Biological Sciences, vol 18(4): 885-897, 2025. DOI: 10.3923/ajbs.2025.885.897
- Baer, D.J., Gebauer, S.K. and Novotny, J.A. (2016) ‘Walnuts consumed by healthy adults provide less available energy than predicted by the Atwater factors’, The Journal of Nutrition, 146(1), pp. 9-13. DOI: 10.3945/jn.115.217372 PubMed
- Capuano, E., Oliviero, T., Fogliano, V. and Pellegrini, N. (2018) ‘Role of the food matrix and digestion on calculation of the actual energy content of food’, Nutrition Reviews, 76(4), pp. 274-289. DOI: 10.1093/nutrit/nux072 PubMed
- Carmody, R.N. and Wrangham, R.W. (2009) ‘The energetic significance of cooking’, Journal of Human Evolution, 57(4), pp. 379-391. DOI: 10.1016/j.jhevol.2009.02.011 PubMed
- Grundy, M.M.-L., Edwards, C.H., Mackie, A.R., Gidley, M.J., Butterworth, P.J. and Ellis, P.R. (2016) ‘Re-evaluation of the mechanisms of dietary fibre and implications for macronutrient bioaccessibility, digestion and postprandial metabolism’, British Journal of Nutrition, 116(5), pp. 816-833. DOI: 10.1017/S0007114516002610 PubMed
- Khalid, W., Maggiolino, A., Kour, J., Arshad, M.S., Aslam, N., Afzal, M.F., Meghwar, P., Zafar, K.-u.-W., De Palo, P. and Korma, S.A. (2023) ‘Dynamic alterations in protein, sensory, chemical, and oxidative properties occurring in meat during thermal and non-thermal processing techniques: A comprehensive review’, Frontiers in Nutrition, 9, 1057457. DOI: 10.3389/fnut.2022.1057457 PubMed
- Lasschuijt, M., Camps, G., Mars, M., Siebelink, E., de Graaf, K. and Bolhuis, D. (2023) ‘Speed limits: the effects of industrial food processing and food texture on daily energy intake and eating behaviour in healthy adults’, European Journal of Nutrition, 62, pp. 2949-2962. DOI: 10.1007/s00394-023-03202-z PubMed
- Louis, P., Solvang, M., Duncan, S.H., Walker, A.W. and Mukhopadhya, I. (2021) ‘Dietary fibre complexity and its influence on functional groups of the human gut microbiota’, Proceedings of the Nutrition Society, 80(4), pp. 386-397. DOI: 10.1017/S0029665121003694
- National Academies of Sciences, Engineering, and Medicine (2023) Dietary Reference Intakes for Energy. Washington, DC: The National Academies Press. DOI: 10.17226/26818 PubMed
- Tzeravini, E., Tentolouris, A., Kokkinos, A., Tentolouris, N. and Katsilambros, N. (2024) ‘Diet induced thermogenesis, older and newer data with emphasis on obesity and diabetes mellitus - A narrative review’, Metabolism Open, 22, 100291. DOI: 10.1016/j.metop.2024.100291 PubMed
- Westerterp, K.R., Wilson, S.A.J. and Rolland, V. (1999) ‘Diet induced thermogenesis measured over 24h in a respiration chamber: effect of diet composition’, International Journal of Obesity, 23(3), pp. 287-292. DOI: 10.1038/sj.ijo.0800810 PubMed
