What Food Actually Does

TL;DR. Food does four things: supplies energy, supplies building material, supplies the small parts (vitamins and minerals) that let the machinery run, and feeds the bacteria in your colon. Most nutrition arguments are about the first, most nutrition failures are about the third, and the fourth was ignored for a century. The honest state of the evidence is that dietary patterns beat individual nutrients, that the strongest findings are unglamorous (more fibre, more plants, less ultra-processed food, less sugary drink, less alcohol), and that almost every confident, specific claim you have heard about a single food is weaker than it sounds. This chapter explains what actually happens inside you when you eat, so you can judge the claims yourself.

Key takeaways

• Your resting metabolism is 60 to 70 percent of your daily energy use. Exercise is a smaller share than people assume, which is why exercise is excellent for health and a weak tool for weight loss on its own.
• Protein costs more to process than it delivers in appetite, keeps you full longer, and preserves muscle during weight loss. It is the macronutrient with the clearest practical case.
• Fibre is the most consistently under-eaten thing in modern diets, at roughly half the recommended intake in most countries, and it is among the most consistently protective.
• The same calories behave differently depending on what they arrive with. Fibre, protein, and fat all flatten the glucose response to the same amount of starch.
• In a tightly controlled trial, people ate about 500 more calories a day on an ultra-processed diet than on a minimally processed one matched for sugar, fat, fibre, and macronutrients.
• Alcohol has no safe threshold for cancer risk, and the apparent heart benefit of moderate drinking has largely dissolved under better methods.

Where your energy actually goes

In short: Most of your daily calorie burn happens while you are doing nothing, which is why diet dominates weight and exercise dominates health.

Two consequences follow immediately.

Exercise is a poor weight-loss tool and an excellent health tool. A 45-minute run might burn 400 calories, which is a large muffin. That is why exercise trials for weight loss consistently disappoint, and why the same trials show substantial improvements in blood pressure, insulin sensitivity, cardiorespiratory fitness, and mortality. Those are different outcomes and they should not be conflated.

The thermic effect differs by macronutrient, which is a genuine, if modest, reason protein helps with weight management:

So 100 calories of protein delivers roughly 75 usable calories, while 100 calories of fat delivers about 98. It is not a large enough effect to build a diet on, and it is real.

Protein

In short: The macronutrient with the clearest practical case, and the one most people over 60 do not get enough of.

Chapter 2 explains what a protein and an amino acid are. What matters practically:

How much. The official minimum is 0.8 g per kg of body weight per day, set to prevent deficiency rather than to optimise anything.

Why more with age. Older muscle shows anabolic resistance: it needs a bigger protein dose to trigger the same amount of rebuilding. Combined with typically lower intake and less activity, this is a substantial contributor to the sarcopenia in Chapter 7.

Distribution matters. Roughly 25 to 40 g per meal, three or four times a day, stimulates muscle protein synthesis better than the same total concentrated at dinner, which is how most people actually eat.

What 30 g of protein looks like: 120 g chicken breast, 150 g salmon, 4 large eggs, 170 g Greek yoghurt plus a handful of nuts, 200 g firm tofu, 250 g cooked lentils, 85 g hard cheese, or a standard scoop and a half of whey powder.

Plant versus animal. Animal proteins contain all nine essential amino acids in useful proportions. Individual plant foods are usually low in one or two, and eating a variety across the day resolves it completely. Plant-based eaters need slightly more total protein, since digestibility is somewhat lower, and need a reliable B12 source, which is not optional.

Carbohydrate

In short: The fastest fuel, the one with no minimum requirement, and the one whose effect depends far more on what it arrives with than on how much of it there is.

Carbohydrate is sugars in chains. What determines its effect on you is how quickly it becomes glucose in your blood.

What flattens the glucose response to the same amount of starch:

This is most of why "whole foods" and "refined foods" produce different responses. It is not a mystical property; it is physical structure and what else is on the plate.

Fibre deserves its own paragraph because it is the single most under-eaten component of modern diets. Recommended intake is 25 to 30 g a day; typical intake in most high-income countries is 12 to 18 g. A large meta-analysis found that people in the highest fibre intake groups had roughly 15 to 30 percent lower all-cause and cardiovascular mortality and lower rates of colorectal cancer, type 2 diabetes, and stroke, with a dose-response relationship. Almost no other single dietary change has that breadth of association.

Practically: legumes, whole grains, vegetables, fruit with skins, nuts and seeds. Increase it gradually, because a sudden jump causes bloating while gut bacteria adjust.

Sugar. The specific problems are that liquid sugar produces almost no satiety, that added sugar displaces more useful food, and that it causes dental caries directly (Chapter 54). Sugar-sweetened drinks are the most consistently implicated single dietary item across the obesity, diabetes, and dental literature.

Fat

In short: Essential for membranes, hormones, and four vitamins, and the type matters more than the total.

Total fat intake turned out to matter far less than the low-fat era assumed. What has held up is the type, and specifically what a given fat replaces in the diet:

Two fatty acids are essential: linoleic acid (omega-6) and alpha-linolenic acid (omega-3). The long-chain omega-3s EPA and DHA are made from the latter inefficiently, which is why oily fish (salmon, mackerel, sardines, herring) is the reliable source, and algae-derived supplements are the plant-based equivalent.

Vitamins and minerals: the deficiencies that actually happen

In short: Five deficiencies account for most of the real-world burden, and none of them are what supplement marketing focuses on.

Everything else is mostly noise. Large randomised trials of multivitamins, antioxidants, vitamin D in unselected populations, and fish oil in primary prevention have repeatedly found no benefit, and several found harm (Chapter 63). Correcting a real deficiency works; topping up someone who is replete does not.

Ultra-processed food

In short: The most interesting recent finding in nutrition, with a genuine causal experiment behind it and an unresolved mechanism.

The NOVA classification sorts food by extent of processing rather than by nutrients:

The experiment that matters. Kevin Hall's group at the US National Institutes of Health admitted volunteers to a metabolic ward and fed them either ultra-processed or minimally processed diets, matched for calories offered, sugar, fat, fibre, sodium, and macronutrients, with unlimited eating. On the ultra-processed diet people ate about 500 more calories a day and gained weight; on the minimally processed diet they lost it. Same people, crossed over, same nutrients on paper, opposite outcomes.

That is a genuine causal demonstration, and the mechanism is not settled. Candidates include energy density, softer texture allowing faster eating, low satiety per calorie, additives affecting appetite signalling, and effects on the microbiome. Large observational studies associate ultra-processed intake with obesity, cardiovascular disease, type 2 diabetes, and mortality, with the usual confounding caveats.

The honest position: the finding is real and the mechanism is unknown, so the practical advice ("reduce the share of your diet that is ultra-processed") is well founded while any specific claim about which additive is responsible is not.

Alcohol

In short: A toxin your liver processes at about one drink an hour, with a group 1 carcinogen classification and no safe threshold for cancer.

What happens mechanically. Alcohol is absorbed rapidly, largely from the small intestine. The liver converts it with alcohol dehydrogenase into acetaldehyde, which is toxic and carcinogenic, then with aldehyde dehydrogenase into harmless acetate. The first step is fast and the second is the bottleneck, so acetaldehyde accumulates. Your liver handles roughly one standard drink per hour, and nothing speeds this up: not coffee, not cold showers, not food, though food slows absorption.

Why the East Asian flushing response matters. Roughly a third of people of East Asian descent carry an ALDH2 variant that slows the second step, so acetaldehyde builds up rapidly, causing flushing, nausea, and palpitations. Those who drink despite it have substantially raised oesophageal cancer risk, because they are marinating their oesophagus in a carcinogen (Chapter 62).

What it does over time: liver disease (Chapter 32), seven cancers (mouth, throat, oesophagus, liver, colorectum, breast), hypertension, atrial fibrillation, cardiomyopathy, pancreatitis, brain volume loss, disrupted sleep architecture, and dependence (Chapter 43).

The "moderate drinking is protective" story. Observational studies for decades showed lower mortality in moderate drinkers than in non-drinkers. The comparison group was the problem: it included former heavy drinkers who quit because they were ill, and people too unwell to drink. Studies separating lifelong abstainers, and genetic (Mendelian randomisation) analyses using alcohol-metabolism variants as a natural experiment, have largely dissolved the protective effect. Cancer risk rises from low levels of intake with no threshold, which is why several countries have revised their guidance sharply downward.

Meal timing and fasting

In short: Time-restricted eating and intermittent fasting work mainly by reducing total intake, and trials comparing them to ordinary calorie restriction generally find them equivalent.

Various patterns have been tested: 16:8 time-restricted eating, alternate-day fasting, 5:2. The consistent finding across randomised trials is that when calories are matched, the fasting pattern produces about the same weight loss and metabolic improvement as continuous restriction.

That is not a criticism. If a structured eating window makes it easier for you to eat less, it is a useful tool, and adherence is the binding constraint in every dietary intervention. It is a criticism of the claim that fasting has a unique metabolic magic independent of intake, which the trials do not support.

Two genuine caveats: very late eating is associated with worse glucose handling, since insulin sensitivity follows a daily rhythm and is lower at night. And prolonged fasting is not appropriate for people on insulin or sulfonylureas, in pregnancy, with a history of eating disorder, or who are underweight.

Hydration

In short: Drink to thirst plus a margin in heat and illness, and judge by urine colour rather than by counting glasses.

You lose roughly 2 to 2.5 litres a day and replace it from drinks, food (fruit and vegetables are mostly water), and metabolic water. The "eight glasses a day" rule has no clear origin in evidence and ignores that food supplies perhaps 20 to 30 percent of intake.

Practical markers: pale straw urine is about right; consistently dark suggests too little; consistently colourless suggests more than needed. Thirst is a reliable signal in healthy adults and becomes less reliable with age, which is why dehydration is common in older people and a frequent contributor to hospital admissions, falls, and acute kidney injury.

Drinking far too much matters too. Excessive intake in a short time can dilute blood sodium dangerously, which has killed endurance athletes and participants in drinking contests.

What a good diet actually looks like

In short: The evidence converges on a pattern rather than a rule set, and the pattern is unexciting.

Stripped of everything contested, the areas of genuine agreement across dietary guidelines, cohort studies, and the few long randomised trials:

And the meta-point: almost every diet that works does so by increasing satiety per calorie and reducing intake, whether it calls itself low-carbohydrate, Mediterranean, plant-based, or high-protein. Trials comparing named diets against each other consistently find small differences between them and large differences between people who adhere and people who do not. Choose the one you can sustain.

Six food claims worth dismissing

In short: These recur constantly and none survives contact with the evidence.

1. "Detox" diets and cleanses. Your liver and kidneys do this continuously and no commercial product improves on them. No detox product has ever identified which toxin it removes.
2. "Eating frequently stokes your metabolism." The thermic effect depends on total intake, not how it is divided. Meal frequency trials show no metabolic advantage.
3. "Alkaline diets change your blood pH." Blood pH is held between 7.35 and 7.45 by mechanisms that will dissolve your bones before letting it drift. Food changes urine pH, not blood pH.
4. "Superfoods." No single food has properties that matter against the background of a whole diet. The term is marketing, and it is banned in advertising in the EU without specific authorised health claims.
5. "Food intolerance test panels" (IgG testing). These measure exposure, not intolerance. Every major allergy body advises against them, and they generate long unnecessary exclusion lists (Chapter 46).
6. "Sugar feeds cancer." All cells use glucose, cancer cells included; starving yourself of carbohydrate does not selectively starve a tumour and malnutrition worsens cancer outcomes.

Don't be confused: "processed" is not a synonym for unhealthy. Freezing, pasteurising, canning, fermenting, and milling are all processing, and several of them make food safer and more nutritious. Frozen vegetables are often higher in vitamins than "fresh" produce that has spent a week in transit. The NOVA group that matters is group 4, industrial formulations, not processing in general.

Sources and notes

Energy expenditure components: standard nutrition texts and doubly-labelled-water studies. Thermic effect by macronutrient: Westerterp, Nutrition & Metabolism, 2004. Protein requirement figures: WHO/FAO/UNU; PROT-AGE recommendations (Bauer et al., JAMDA, 2013); Morton et al., BJSM, 2018. Anabolic resistance in ageing muscle: Wall, Gorissen, and van Loon. Fibre intake and outcomes: Reynolds et al., The Lancet, 2019, a series of meta-analyses. Ultra-processed food trial: Hall et al., Cell Metabolism, 2019. NOVA classification: Monteiro et al. PREDIMED: Estruch et al., NEJM, 2018. Trans fat elimination: national policy evaluations. Alcohol and Mendelian randomisation: Millwood et al., The Lancet, 2019; IARC classification of alcohol as a group 1 carcinogen. ALDH2 and oesophageal cancer: Brooks et al., PLoS Medicine, 2009. Intermittent fasting equivalence: Liu et al., NEJM, 2022, and multiple meta-analyses. IgG food testing: position statements from EAACI, AAAAI, and equivalents. Iron deficiency anaemia prevalence: WHO and Global Burden of Disease estimates.

Open questions. The mechanism by which ultra-processed food increases intake is unresolved and is the most important open question in nutrition. Optimal protein intake across the lifespan remains debated. Whether specific saturated fatty acids differ in effect, and how much the food matrix matters relative to nutrient content, is unsettled.

Next: what happens inside you when you move. 👉