A Source-Grounded Guide to Nutrition & Lifestyle Evidence

This is a source-grounded appraisal of the nutrition, exercise, and lifestyle evidence — what an exposure does, for whom, how much, and how confident we should be, with every claim traced back to the study or guideline it came from.

A playful Guide to Health Research”

From Evidence to Action

Turning imperfect evidence into a real decision — what to weigh, and what the evidence can’t tell you.

Better than What?Every choice is 'better than what, for whom?' — weighing adherence, competing goals and real-world context when the evidence is partial, without defaulting to 'it depends'.

Acting before Evidence?When it's reasonable to act ahead of a settled meta-analysis — keeping 'how strong is the evidence?' apart from 'should I act?', reaching a direction by triangulating trials, cohorts and genetic experiments, and letting reversibility, cost and the size of the lever move the threshold, not the certainty grade.

Limits of EvidenceWhat nutrition evidence can and can't show — why 'no evidence' usually means unstudied, not disproven, and how to choose when the evidence is silent, without falling back on anecdote.

Baseline Risk & Absolute BenefitThe same '25% lower risk' is a big deal or a rounding error depending on where you start. Why a treatment's benefit scales with your absolute risk, how to tell a real subgroup effect from baseline-risk arithmetic, and when a moved marker does — and doesn't — carry through to an outcome you care about.

Evidenced GuidelineThe five lifestyle levers that move health the most, ranked on one page — what each one actually buys (and what it doesn't), where the gains run out, and how the list shifts after 65. Health evidence only; no feasibility, cost or palatability on the scale.

Big Rocks

The levers that actually move health, sorted by how much they move. If you only change a few things, change these.

Start with the maps — these rank all the levers for a whole person, so you can see what is worth your attention before diving into any one of them:

Health Levers — Typical AdultThe largest levers for a typical drifting-into-middle-age adult, sorted to help the majority without harming the leaner minority.

Health Levers — After 70How the priorities invert with age: muscle, function, bone and falls move to the front.

Dementia PreventionCan anything actually prevent dementia? A dedicated brain-health bundle (diet, exercise, cognitive training) moves test scores but hasn't been shown to prevent the disease — and targeting the highest-risk people doesn't rescue it. The one lever proven in a randomized trial is lowering blood pressure: small in absolute terms, on a pill already taken for the heart. The diet and sleep levers mostly work through the same cardiometabolic rocks.

Then the individual big rocks — the states and behaviours that carry most of the achievable effect:

SmokingStopping is the single largest health lever there is — and it stays the largest at every age. A lifelong smoker dies at about three times the rate and loses more than a decade of life; quitting recovers most of it (roughly 90% if before \~40), and it is never too late. For a non-smoker, this lever is already pulled.

Body FatWhere fat sits matters more than the scale — the risk is the fat around and inside the organs, and the 'overweight is protective' paradox is a measurement artifact. Whether a given person's fat is raising their risk, and whether losing it reaches hard outcomes or only surrogates; after 70 the advice flips to protecting the body, not shrinking it. How to lose it — and keep it off — is its own guide.

Losing Fat & Keeping It OffThe body defends against a deficit — metabolism drops and hunger rises — so keeping the loss off is the hard part, not taking it off. What each method actually moves (diet, exercise, total-diet-replacement, carb-restriction, GLP-1), how to hold on to muscle, and why the source and speed of the deficit matter less than whether you can sustain it.

Type 2 DiabetesPrevention and remission by lifestyle: losing enough weight can send early type-2 diabetes into remission — roughly half in the first year or two, fading as weight returns — and intensive lifestyle beats metformin at stopping prediabetes progressing. Which route works for whom, and what remission does and doesn't buy on hard outcomes.

Blood PressureWhich exposures lower blood pressure and by how much — and, separately, which lowering is actually proven to prevent strokes and heart attacks. Ranking diet, weight, sodium, potassium, alcohol and drugs by effect, certainty and what your absolute risk makes them worth. A big rock when your pressure — or your CV risk — is already up; much smaller if it isn't.

Blood Lipids & ApoBThe apoB-containing particles (LDL and its cousins) cause atherosclerosis — one of the most settled facts in cardiovascular medicine — and lowering them cuts heart attacks and strokes in proportion to how much, and how long. Measure LDL-C for most; add apoB when metabolism drifts (diabetes, high triglycerides). Whether to act turns on absolute risk, not the number alone — a zero calcium score can defer a statin. Dietary cholesterol and eggs are a near-non-issue for most.

Physical ActivityCardio, resistance and walking — what each buys on which outcome, and how to split a fixed weekly budget across them for the most benefit per minute.

ProteinHow much, from what source, and for whom — the RDA is a deficiency floor, not a muscle target, and \~1.6 g/kg is a region, not a point. A small lever for most, a big one for older adults, lifters and dieters; the plant-ward shift is the modest longevity signal, and the kidney worry holds only for existing disease.

AlcoholNo safe level, and no real small-dose benefit — the protective 'J-curve' is a study artifact. Where each outcome's harm starts, and whether pattern (don't binge) or beverage type make drinking less harmful.

Medium Stones

Real levers, but second-order: meaningful decisions where the effect is moderate, nested inside a big rock, or turns on your specifics. Worth getting right — after the big rocks, not before them.

SleepHow much, how regular, how consolidated — which lever actually moves health, the sleep–obesity loop (and what the leptin/ghrelin evidence actually shows), and whether sleep aids treat the underlying problem (CBT-I vs pills). A real lever on mortality and metabolic risk, but — unlike the big rocks — not carried by the multi-factor risk rankings, so held here at medium.

GLP-1 DrugsOzempic/Wegovy and the newer agents — the proven benefits against the real risks, and which stratum the trade-off favours. A powerful lever into body fat and glycaemia for the stratum it fits, not a universal big rock.

Dietary PatternsMediterranean, DASH, vegetarian, keto, carnivore — no single pattern wins. You can refute a diet (an adequacy or energy-balance floor) but not rank the survivors: above the floor the head-to-head evidence is near-null, and the evidenced benefit traces to shared components — the big rocks — not to the name on the diet.

SodiumSodium, potassium and the potassium-enriched salt swap — how far cutting salt lowers blood pressure, how well that carries through to strokes and heart attacks, who gains most, and what the low-intake question does and doesn't settle.

Added SugarWhich sugar, in what form, on which outcome — and whether the effect survives once you hold the calories equal. Dental caries is the one harm the sugar molecule itself drives; for weight, liver and metabolic risk the harm tracks the calories, delivered most easily by the drink.

Dietary FatSaturated, unsaturated and trans fats — which swaps help which outcome, how strong the LDL-to-events chain is, and where the food matrix and metabolic state change the answer.

Processed Food'Processed' is a manufacturing description, not a health category: the label spans the clearest dietary harms (sugary soda, processed meat) and neutral-to-beneficial foods (wholegrain bread, fibre isolates). Decomposed by aspect and category — and why 'eat less processed' does not survive as its own lever.

MeatRed, processed, and poultry as distinct exposures — heart vs lifespan vs cancer, where the harm concentrates and where it doesn't, and how much each meat moves the needle.

FishOily vs lean, apex vs prey, wild vs farmed, and the mercury trade-off — why the choice is usually a species swap, not eat-or-avoid.

FibreCholesterol, gut transit, blood sugar and satiety — how much each moves, and how the effect size compares to the big rocks.

Plant FoodsWhich plant food is 'best' is mostly not answerable — the outcome evidence resolves at the dietary pattern or the shared component (fibre), rarely at the single food. Fibre (\~25-29 g/day) is the one modest lever; antinutrients are defused by ordinary cooking; whole-fruit sugar isn't free sugar. A moderate, mostly-observational lever: the large-looking numbers are associations with a small proven core, so it sits below the big rocks — worth getting right after them, not before, and choosing among plant foods is refinement.

Sun ExposureA genuine two-sided trade-off. Intermittent sun and sunburn raise melanoma risk, and sunscreen cut it in a randomized trial; against that, a large cohort ties sun avoidance to higher all-cause mortality — an association that may reflect who avoids the sun rather than the sun itself. Vitamin D pills don't reproduce the mortality signal. Low-to-moderate confidence, and the balance shifts with skin type and latitude.

Small Pebbles

Popular worries and ‘which is best’ debates — mostly scares and low-stakes swaps — held to the same evidence bar, where the honest answer is usually the effect is small. Worth reading mostly so you can stop thinking about them.

Soy ProductsDoes soy feminize men, raise breast-cancer risk, or harm the thyroid? The three concerns under the same standard — plus what soy does measurably (a modest LDL drop) and why tofu, tempeh and isolate aren't one thing.

Artificial SweetenersCancer, appetite, insulin, weight — and why the answer depends on what the sweetener replaces.

CoffeeMortality, heart, type-2 diabetes, liver — how much, filtered vs unfiltered, and why the benefits are the bean, not the caffeine.

Meal Timing & FastingTime-restricted eating, one-meal-a-day, alternate-day and prolonged fasting — why when and whether you eat mostly act through the calorie deficit, and where a long fast turns harmful.

Organic FoodPesticides, nutrients and outcomes — and why pasture-vs-grain often matters more than the label.

DairyMilk, cheese, butter and yoghurt behave like different foods — read apart, dairy is a small or neutral lever. The milk-shortens-life scare is a single-cohort artifact; more dairy doesn't prevent fractures; prefer fermented, treat butter as a saturated-fat question, and low-fat vs full-fat is a wash. The cancer question stays an open gap.

EggsFor a healthy person eggs are close to a non-lever for the heart — the cholesterol scare is a category error (how much an egg contains is not how much it raises the cholesterol in your blood), and there's no protective case either, so stop counting them. Two low-certainty caveats survive at high habitual intake (a heart-failure and a small mortality signal), and in people with diabetes the direction may reverse. Judge the breakfast the egg replaces, not the egg.

FluorideDrink the tap water, use the toothpaste, buy the filter? The caries benefit is well established. The child-cognition worry that drives the fear sits at exposures higher than fluoridated water, not at it, and the bone data are reassuring. The clearest anti-hype case in this set — read it once and stop worrying.