Five things move an adult’s health more than everything else put together: not smoking, moving a little every day, not carrying excess belly fat, keeping alcohol low, and getting enough sleep. Read them as a stack, biggest at the base — and a lever only counts if it applies to you: you have nothing to quit if you don’t smoke, nothing to lose if you are already lean and metabolically healthy. Pull these five and you have captured most of what a lifestyle can buy your health; the rest is real but small.
And one regimen reaches many diseases at once: the same few levers — above all moving your body, not smoking, and clearing excess fat — lower risk across heart disease, dementia, diabetes and cancer simultaneously, so there is no separate anti-cancer or anti-dementia diet to run. That shared breadth is a large part of why these levers rank first. -> Shared Modifiable Levers Across Age-Related Diseases
The five levers, biggest first
Don’t smoke — the largest lever, wherever it is present. There is no good-enough dose and no safe cigarette here: this is quit-or-not. Smokers die at roughly three times the rate of never-smokers and lose at least a decade of life; quitting before 40 erases about 90% of the added risk, and quitting later still helps a great deal. For a non-smoker this lever is simply absent — already pulled. -> Smoking and Mortality
Move your body a little every day — the first minutes matter most. The steepest drop in risk comes from leaving the least-active group at all, and it is mostly light, everyday movement — walking, chores, being on your feet — not workouts. Going from barely moving to modestly active — roughly an added hour a day of light movement, or about five extra minutes of something brisk, or a couple of thousand more steps than a sedentary day — is linked to roughly halving your risk of dying. Then the curve flattens: past a modest daily amount, more buys little.
Long sitting is a separate risk, but activity largely cancels it. If a desk job or commute keeps you sitting eight hours or more, an hour or so of moderate movement a day offsets it — and sitting less is not a substitute for moving more. Strength follows the same shape — one or two short sessions a week, a small weekly total rather than four trips to the gym, captures most of the gain. -> Physical Activity Dose and Mortality, Muscle-Strengthening Activity and Mortality
Don’t carry excess belly fat — clear it, don’t chase a number. The risk is not weight on the scale but fat spilling into your liver and around your organs once you pass your own personal fat threshold. The target is the self-verifiable endpoint — waist coming down, blood sugar and liver markers back in range — which for many people is a modest loss, not a fixed percentage. Keep muscle while you do it, with strength work and enough protein. Pull this lever for blood sugar, liver and function — not on a promise it prevents heart attacks (the trials have not shown that). If you are already lean and metabolically healthy, you have nothing here to do. -> Ectopic Fat and Depot-Specific Risk, Fatty Liver MASLD and Weight Loss, Does Weight Loss Reduce Cardiovascular Events
Keep alcohol low — less is better, with no safe threshold. There is no good-enough dose to aim for. The old finding that moderate drinkers outlive abstainers is largely an artifact — correct for how the comparison group is built, or read the genetic evidence, and the protection disappears. Risk of cancer and vascular disease rises from low intake with no safe floor, so this lever is only ever a harm to cut, never a rock to add. -> Alcohol and Mortality and Vascular Disease
Get enough sleep — the fix is not being chronically short. Aim for more than about six hours a night. The mortality-and-diabetes penalty is small until sleep drops well below that, and it bites hardest under about five hours; six-to-seven hours sits barely above the healthiest point. The actionable arm is short sleep — sleeping long is a marker of underlying illness, not a habit to trim. -> Sleep Duration and Mortality
The smaller levers add a little
Everything below the five is genuinely small — worth a sentence, not a campaign.
Saturated fat — no need to give up red meat. Keep saturated fat modest and swap some of it for unsaturated fat (olive oil, nuts, fish rather than butter). It might help a little, mostly against heart attacks rather than length of life — and the “a little” is the honest size of it. The swap is what matters: trading saturated fat for unsaturated helps, trading it for refined carbohydrate does not. -> Saturated Fat Intake and Replacement
Salt — only if your blood pressure is drifting up. If it is, cutting some salt brings pressure down a little, and the higher your pressure already sits, the more the cut is worth. If your pressure is fine, this is not your lever. -> Blood Pressure Lowering and Cardiovascular Events
Protein — for older adults and anyone losing weight. If you are ageing or in a deliberate calorie deficit, keeping protein up while you keep lifting protects muscle and bone. For a younger person eating normally, it is not a lever worth chasing. -> Protein and Resistance Training for Muscle and Strength
Fibre — eat whole plant foods; the famous gram target is softer than it sounds. People who eat the most fibre die less often, but that comes from observational data on self-reported diets and is tangled up with eating more vegetables and living healthier generally — so much of it is the whole pattern, not fibre itself. And fibre is not one thing: only the viscous kind (oats, psyllium) has a clean trial effect, and it is small — a little lower cholesterol. Eat plenty from whole plant foods; a rough 25-29 g/day is a reasonable floor, not a precise or proven target, and a supplement is no substitute for the food pattern. -> Dietary Fibre and Health
Things to look out for
A handful of readings tell you whether the levers are working and when to escalate. These are self-monitoring signals, not diagnoses — a reading that stays out of range means pull the owning lever harder and talk to a clinician. This page never names or doses a drug.
- Blood pressure — the clearest signal that the weight, activity, alcohol and salt levers are working. If it stays up despite them, that is the conversation about medication (off this page).
- apoB, or LDL cholesterol — the one blood marker that is measurable and even-handed across very different diets. Years of exposure matter, not a single reading; a level that has run high for years is a clinician conversation.
- Fasting glucose or HbA1c — your metabolic-status readout, moved by the adiposity and activity levers. A number creeping up is the early warning; it is a marker to act on, not a target to chase with medication on your own.
- Waist — the at-home proxy for your personal fat threshold; steer the belly-fat lever by it, not the bathroom scale.
- (65+) grip strength, or how hard it is to rise from a chair — the functional readout of how the muscle levers are holding.
None of these is a sixth lever — they are the dials on the five you already have.
Where it stops being worth the effort
Once the five big levers are handled — you don’t smoke, you move daily, you are not carrying ectopic fat, you drink little, you sleep enough — the gains that remain are small by construction, and often uncertain. That is a finding, not a hole in this page: for someone already lean, active, non-smoking and sleeping well, the large levers are pulled, and “there is little left to optimize” is a real result that licenses you to stop.
Whole-diet patterns — Mediterranean, DASH and the like — are not treated here as a single lever, and that is not merely a coverage choice. The whole-pattern evidence on hard outcomes is genuinely thin: mostly surrogate markers or confounded observation, with no clean hard-outcome contrast to anchor a claim (the one Mediterranean trial with hard endpoints, PREDIMED, is methodology-discounted). So this is a gap in the literature, not a body of evidence merely waiting to be pulled in. The even-handed handle the evidence does support is component-level and measurable — apoB and metabolic markers — not the pattern label. -> Dietary Patterns
After about 65, the stack shifts
The levers do not change on your birthday, but their order does as you age.
- Balance and footwork become the surest lever — against falls. Training balance and gait cuts the fall rate substantially, and it is one of the best-evidenced moves in the whole guide. Resistance training on its own is not proven to prevent falls — the evidence there is uncertain, not a demonstrated failure — so train balance specifically rather than assuming that lifting covers it. Pillay’s network meta-analysis for the Canadian Task Force (219 RCTs, 167,864 participants) ranked the whole fall-prevention menu and kept balance on top, pinning the dose Sherrington could not: an effective programme is supervised — «> two sessions, not including ini- tial instruction» — and long-duration — «> 3 months», with group tai chi a top-tier option; a one-off class does not do it (Pillay et al., 2024). Pillay «included 125 of 283 studies included in the previous review», so this refines Sherrington, not independent corroboration. -> Exercise for Preventing Falls in Older Adults
- Protein and strength work move up the stack. Ageing blunts the muscle’s response to protein, and muscle and bone get harder to hold, so keeping both protein and resistance training up matters more than it did at 40. -> Protein and Resistance Training for Muscle and Strength, Anabolic Resistance
- Blood pressure matters more, but the big effect is mostly medication. The large gain from lowering an older adult’s blood pressure comes chiefly from drugs, which sit off this page; the lifestyle piece — salt, weight, activity — is real but modest. -> Blood Pressure Lowering and Cardiovascular Events
- The sign-flippers — leave these to a clinician. Some things that look protective can flip to harm in older adults: testosterone (which raised fractures in trial), calcium-plus-vitamin-D when you are not deficient, the tightest blood-pressure targets, and edge-of-evidence statins. These are individual calls a doctor makes with your labs, not general advice. -> Testosterone Adiposity and Muscle, Grip Strength and Mortality
The bottom line
In priority order, for most adults:
- If you smoke, quit — nothing else on this page comes close.
- Move a little every day — the first minutes matter most; you do not need a training plan.
- Keep belly and visceral fat down — steer by waist and metabolic markers, not the scale.
- Keep alcohol low — less is better; no amount helps.
- Protect your sleep — don’t run chronically short.
Everything past these is small. Do the five, then stop optimizing — the ceiling is real, and reaching it is the goal.
What this adds to “eat well and exercise”
If a good doctor applied today’s guidelines to your situation and said the same thing, this page has added nothing. And you can gauge how small the leftover levers are by how loudly they are argued: the big levers are settled and boring, so the noise — the supplement, the superfood, the perfect macro split — gathers around the little ones. Here is where this page does more than repeat “eat well and exercise”:
- It ranks, rather than lists. Guidance tends to present a flat menu — diet, exercise, sleep, alcohol, all worth doing. This page says which lever is largest for you and in what order to spend effort, because effort put into a small lever while a big one stands is not a smaller win, it is the wrong move.
- It says what each lever does not buy. Losing weight reliably improves blood sugar and liver fat, but the trials have not shown it prevents heart attacks — so it is pulled for metabolic health, not on a cardiac promise. The apparent benefit of “moderate” drinking is largely an artifact of who ends up in the non-drinking comparison group, not a reason to drink. And 10,000 steps is a marketing number, not a threshold — most of the step benefit banks well below it.
- It gives permission to stop. Once the big levers are handled, the gains that remain are small by construction. “Your remaining levers are small and uncertain” is a finding, not a failure — it licenses you to stop optimizing.
- Only your health is on the scale. This weighs health effects and nothing else — not cost, not the environment, not what is fashionable. Where another consideration exists, it is yours to weigh, not folded in here.
What this is, and is not
- Health only. This weighs effects on your physical health — the length and quality of your life. It does not price cost, convenience, the environment or anything else; those weightings are yours.
- No drugs prescribed. Where medication is the realistic alternative, this page names it for comparison, but selecting, dosing or combining any drug is a clinician’s job, not this page’s.
- A starting point, not personal advice. This page is the population-level picture. Your own situation — your history, your labs, your goals — moves the order, and that tailoring happens with a person, not a page.
- The loop is open. This wiki grades whether its claims are internally coherent and faithful to their sources — not whether following them makes you better off. A clean audit is not a proven outcome. Treat it as well-sourced reasoning, not a guarantee.
The evidence behind each directive
The front of this guideline is kept to rough magnitudes; the precise figures, certainties and studied ranges live here, in a backing layer that does not clutter the directives. Every figure carries its source; a number read off a wide interval or the edge of the studied range is not a hard target.
Why one regimen reaches many diseases (the shared-lever backbone)
The big levers are not disease-specific. Assemble the modifiable-risk-factor lists the fabric holds for dementia, cardiovascular disease, cancer and Parkinson’s, and a small set recurs in the same protective direction — physical activity, not smoking, low adiposity, low alcohol — so pulling the big rocks once buys risk reduction on several patient-important outcomes at the same time. That strengthens the ranking rather than adding competing tasks: a lever that moves three outcomes outranks one that moves one, at equal per-outcome effect. -> Shared Modifiable Levers Across Age-Related Diseases
Physical activity is the most cross-cutting lever. It is protective for dementia (Livingston et al., 2024), for Parkinson’s (RR 0.79, 0.68-0.91) (Chen et al., 2021), and a protective cause in the cancer evidence (World Cancer Research Fund & American Institute for Cancer Research, 2018) — and the same act also preserves physical function and cuts frailty and joint pain, not only prevents disease -> Shared Modifiable Levers Across Age-Related Diseases. Body fatness is a cause of 12 of the 17 cancers reviewed (World Cancer Research Fund & American Institute for Cancer Research, 2018) on top of being a midlife dementia risk factor (Livingston et al., 2024).
The activity-and-dementia link is now held first-hand: across 58 cohorts, being active carried a relative risk of 0.80 (0.77-0.84) for all-cause dementia, and the signal survived a reverse-causation check — holding at RR 0.79 (0.71-0.87) in cohorts followed 20 years or more — and was not modified by ApoE genotype (Iso-Markku et al., 2022). That de-secondhands the dementia protection cited above (a type-F refinement of the same estimate, not an independent second witness — Iso-Markku is the study that borrowed cell rests on) (inferred from Iso-Markku et al., 2022). It adds a second patient-important outcome to an already top-ranked, non-substitutable lever, so it strengthens activity’s place rather than adding a competing task.
Two cautions bind. The per-disease numbers are co-membership evidence, not one commensurable effect size — each risk ratio is for a different outcome in a different population, so a shared lever is not a same-size effect across diseases. And the pattern is disease-specific at the edges: smoking’s association flips to apparently protective in Parkinson’s, which the source attributes to reverse causation over the long pre-diagnosis prodrome, not to any protective effect (Chen et al., 2021) — the one place the shared-lever logic would license a harmful act (smoke to protect the brain) is exactly where the signal is an artifact -> The U-Shaped Association Artifact.
Don’t smoke
Smokers die at roughly three times the rate of never-smokers and lose at least a decade of life; quitting before 40 removes about 90% of the excess risk, and cessation still helps substantially at older ages (Jha et al., 2013). This is the largest single modifiable lever wherever it is present — and entirely absent for a non-smoker, which is why a universal list gates it on “if you smoke.” -> Smoking and Mortality
Move your body
Across eight accelerometer cohorts, moving from the least-active quarter to the second roughly halved all-cause mortality (total-activity HR 0.48, 95% CI 0.43-0.54), with the most-active quarter at HR 0.27 (0.23-0.32); the increment that bought the first jump was overwhelmingly light, incidental movement, and most of the benefit banks in that first step — about 5 min/day of moderate-to-vigorous activity, or ~60 min/day of light movement, the least-to-second-quarter increment. Returns continue above it but shrink (HR 0.48 -> 0.34 -> 0.27 across the upper quarters); the curve flattens only near the top of the studied range, where «No further risk reductions occurred with higher levels of activity except for low light intensity physical activity where the risk appeared to decrease further» (Ekelund et al., 2019) -> Physical Activity Dose and Mortality.
In everyday terms that first step is small: the least-active-to-second-quarter gap was «broadly equal to 60 min/day of light intensity physical activity» — or, at higher intensity, only about 5 min/day of moderate-to-vigorous activity, against roughly 70 min/day less sitting (Ekelund et al., 2019) — so “a little” is an added hour of light movement, or five minutes of something brisk, above a sedentary day.
In steps, the same lowest-to-second quartile step runs from a median 3553 to 5801 steps/day (about +2,300 steps), and the plateau sits near 6,000-8,000/day over age 60 and 8,000-10,000 for younger adults; 10,000 is a marketing number (Paluch et al., 2022).
Resistance training adds an independent signal — about 15% lower all-cause mortality (RR 0.85) and 17% lower type-2 diabetes (RR 0.83), on an L-shaped curve with no harmful upper arm (Momma et al., 2022); the decision that carries the effect is train-versus-not, and roughly two sets twice a week captures most of the gain (Currier et al., 2023). Meeting both aerobic and strengthening guidelines beats either alone (HR 0.71) (World Health Organization, 2020).
At older ages, balance and functional training is the active ingredient for falls — it cuts the fall rate about 24% (rate ratio 0.76, 0.70-0.81, high certainty), while resistance training alone is not proven to (RaR 1.14, 0.67-1.97, very low certainty — the evidence is uncertain, not a demonstrated null) (Sherrington et al., 2019).
Sitting is a separate exposure, and activity offsets it. Sedentary time and activity are only weakly correlated — a person can be high on both — so the sedentary-mortality gradient survives adjustment for activity, and “move more” does not by itself address sitting (Ekelund et al., 2019).
But activity largely cancels the sitting risk. In a harmonised meta-analysis of over a million adults, those who sat «more than 8h/day but who also reported >35.5 MET-hour/week of activity (HR=1.04; 95% CI, 0.99, 1.10)» carried no excess mortality, while the least-active, least-sitting group sat at «HR=1.27, 95% CI, 1.22, 1.31» — the most-active heavy sitters did better than the least-active light sitters (Ekelund et al., 2016). The offsetting dose is higher than the mortality plateau: «High levels of moderate intensity physical activity (i.e. about 60 to 75 minutes per day) appear to eliminate the increased risk of death associated with high sitting time» (Ekelund et al., 2016).
Two residuals matter. Television is the exception — «this high activity level attenuates, but does not eliminate the increased risk associated with high TV viewing time», with «>3 hours/day … associated with higher mortality rates regardless of physical activity, except in the most active quartile, where mortality rates were significantly higher only in the >5 hours/day group» (Ekelund et al., 2016). And the sitting threshold is measurement-dependent: device-measured risk rose «more pronounced at greater than 9.5 hours» (Ekelund et al., 2019), above the >8 h self-report figure. All of this is observational with self-reported sitting — a robust, plausible association, not a proven causal cancellation. -> Physical Activity Dose and Mortality, Exercise for Preventing Falls in Older Adults
Don’t carry excess belly fat
Cardiometabolic risk tracks where fat is stored — spilling into liver and pancreas once a person passes their own personal fat threshold — more faithfully than BMI does, so steer by depot and metabolic status, not the scale (Taylor & Holman, 2014). At a matched weekly energy deficit, exercise removes more visceral fat than eating less and is the only dose-dependent arm (-0.15 per 1000 kcal/week; diet 0.03, not significant), so weight the deficit toward activity and keep protein and lifting to protect lean mass (Recchia et al., 2023).
Be honest about the outcome: weight-loss diets lower all-cause mortality (RR 0.82, ~6 fewer deaths per 1000) but not cardiovascular mortality (RR 0.93) (Ma et al., 2017), and the large lifestyle trial in type-2 diabetes was null for cardiovascular events (Look AHEAD, HR 0.95) (Look AHEAD Research Group, 2013) — so pull this for glycaemia, liver, weight and function, not on a promise about heart attacks.
Where the driver is ectopic fat the reversal is real: a total-diet-replacement programme put 46% of short-duration type-2 diabetes into remission at 12 months, rising with kilograms lost (0% among those who gained, up to 86% at >=15 kg lost) (Lean et al., 2018), and liver fat responds on a graded ladder (~5% loss cuts steatosis, 7-10% inflammation, >=10% fibrosis) (European Association for the Study of the Liver, 2024).
“Metabolically healthy obesity” is real but not durably benign — excess risk emerges only after ~10 years and is small (RR 1.24, 1.02-1.55) (Kramer et al., 2013). This lever is biggest for the centrally-adipose and near-empty for the lean-and-healthy. -> Ectopic Fat and Depot-Specific Risk, Does Weight Loss Reduce Cardiovascular Events
Keep alcohol low
The apparent protection of moderate drinking is largely an artifact of how abstainers are defined and of reverse causation: a bias-corrected meta-analysis erases the protective arm (low-volume RR 0.97, 0.88-1.07, not significant) (Stockwell et al., 2016), as does Mendelian randomization, a genetic natural experiment (Millwood et al., 2019). All-cause mortality becomes statistically unmissable around 45 g/day in men and 25 g/day in women, but risk is already elevated below that (25-44 g/day in men, RR 1.05) — a significance point, not a safe ceiling (Zhao et al., 2023).
For cancer and vascular disease the risk rises from low intake with no safe threshold: a global estimate attributed some 41,300 cancer cases in 2020 to light drinking of up to ~10 g/day alone (about one drink) (Rumgay et al., 2021). There is no rock to add here, only a harm to cut at the top end. -> Alcohol and Mortality and Vascular Disease, The U-Shaped Association Artifact
The dementia picture repeats the same J-curve. A dose-response meta-analysis found all-cause dementia risk elevated once intake passes «23 drinks/week or 38 g/day» (Xu et al., 2017). The apparent protection at low intake is unadjudicated: Xu ran no referent-correction and no genetic check, and concedes former drinkers who quit because they were already ill sit in the abstainer reference group (Xu et al., 2017) — the same sick-quitter artifact that dissolved the mortality J-curve. So there is no protective dose to chase for the brain either; only the harm arm above the knee is causally coherent (inferred from Xu et al., 2017).
Get enough sleep
Yin’s per-hour dose-response meta-analysis (67 articles, 3.58 million participants, restricted cubic splines, 7 h reference) resolves the short arm hour by hour: 6 h sits barely above the 7 h nadir (RR 1.01), 5 h is 1.04, and the arm steepens only below ~5 h (3 h RR 1.12) (Yin et al., 2017). The pooled short-sleep penalty below 7 h is «1.06 (95% CI, 1.04-1.07) per 1-hour reduction», against «1.13 (95% CI, 1.11-1.15) per 1-hour increment» above it (Yin et al., 2017) — so the short-arm risk is driven by the chronic-short (sub-5 h) end, not the 6-7 h band.
That is why Cappuccio’s single pooled short-sleep RR of 1.12 (1.06-1.18), which bundles 6-7 h with chronic <5 h, over-states the 6-7 h risk (Cappuccio et al., 2010): the actionable floor is more than about 6 h, with the real penalty below ~5-6 h.
Short sleep also raises incident type-2 diabetes about 9% per hour of sleep lost (RR 1.09, 1.04-1.15) (Shan et al., 2015). The actionable arm is short sleep; long sleep is a marker of illness, not a target to trim. -> Sleep Duration and Mortality
The cognition evidence points the same way, keeping the actionable arm on the short side. Short sleep (under 7 h) tracks faster cognitive decline (RR 1.27, 1.12-1.42) but «was not associated with future risk of all-cause dementia and AD» (Zhang et al., 2025). The long-sleep arm is the larger one for dementia, but it reads as a preclinical marker rather than a cause — Zhang calls prolonged sleep «a preclinical marker driven by the APOE ε4 carrier gene» (Zhang et al., 2025), age-dependent and with no genetic or referent check run against it. So the brain evidence does not license trimming a long sleeper’s hours; the lever stays the same — don’t run chronically short (inferred from Zhang et al., 2025).
The smaller levers
Protein and muscle (rises with age and during weight loss). Added protein augments the fat-free (lean) mass gained from resistance training only up to a soft, non-significant break-point near 1.6 g/kg of body weight per day (95% CI 1.03-2.20, p=0.079 — a statistically soft knee, not a sharp optimum) (Morton et al., 2017) — so aim for roughly 1.6-2.2 g/kg as a floor, not a ceiling to chase past. Two caveats keep the number honest: it is per total body weight, measured in healthy trained adults (people with obesity were not studied), so it does not transport to obesity (where total-vs-lean scaling diverges ~twofold — an untested gap, adjudicated at Protein Intake). ageing blunts the muscle-protein response, so an expert consensus (a recommendation, not a trial-derived target) advises older adults ~25-30 g per meal and ~1.0-1.2 g/kg/day, with ~1.6 g/kg under training (Bauer et al., 2013), and higher protein preserves fat-free mass during a deficit (Refalo et al., 2025). -> Protein and Resistance Training for Muscle and Strength, Anabolic Resistance
Fat quality. Replacing saturated fat with polyunsaturated fat is WHO’s one strong replacement recommendation, but its reach is narrow: it cuts cardiovascular events (RR 0.83, 0.70-0.98, ~15 fewer per 1000, moderate certainty) while all-cause mortality stays null (RR 0.96, 0.90-1.03) (World Health Organization, 2023) — a number-needed-to-treat near 56 in the primary-prevention trials (Hooper et al., 2020). The swap sets the sign: saturated-for-unsaturated helps, saturated-for-refined-carbohydrate does not. -> Saturated Fat Intake and Replacement
Sodium and blood pressure. Cutting sodium lowers systolic pressure about 3.39 mmHg overall (high certainty), roughly threefold more in the already-hypertensive (4.06 mmHg) than the normotensive (1.38 mmHg) (World Health Organization, 2012); blood-pressure lowering itself cuts major events about 10% per 5 mmHg, with the relative benefit broadly constant across baseline pressure (Blood Pressure Lowering Treatment Trialists Collaboration, 2021). Western intake runs ~9-10 g salt against a ~5 g target (European Society of Cardiology, 2021). Decide on overall risk, not the number. -> Blood Pressure Lowering and Cardiovascular Events
Fibre — eat whole plant foods; there is no robust single gram-target. Total dietary fibre is not one exposure: it lumps viscous (oats, psyllium), fermentable, and bulking fractions that act by different mechanisms, and the clean evidence covers only one of them (Brown et al., 1999; inferred from Reynolds et al., 2019). Highest-vs-lowest total fibre tracks ~15-30% lower all-cause mortality (RR 0.85), CHD (RR 0.76) and type 2 diabetes — about 13 fewer deaths per 1000 (CI 8-18) (Reynolds et al., 2019) — but that rests on self-reported intake and is heavily confounded with overall vegetable and whole-food eating: much of the signal is the healthy-eating pattern, not fibre itself (Measurement Error in Dietary Assessment). The one fraction with a clean, mechanism-backed trial effect is viscous/soluble fibre, and only on a surrogate: it lowers LDL ~-0.057 mmol/L per gram over the practical 2-10 g/day range, «only a small contribution to dietary therapy to lower cholesterol» (Brown et al., 1999). So treat Reynolds’ «no less than 25-29 g per day with additional benefits likely to accrue» (Reynolds et al., 2019) as a guidance floor from whole plant foods — an observational, confounded figure, not a proven or type-agnostic target, and no licence to swap a psyllium supplement for the whole-food pattern the mortality data actually described. -> Dietary Fibre and Health
Diet and the brain — a small, mediated signal, not a lever to chase. Two observational meta-analyses attach a cognition signal to eating pattern: higher fruit and vegetable intake tracks lower dementia risk (OR 0.84, 0.78-0.91) (Zhou et al., 2022), and higher flavonoid intake tracks lower risk of adverse cognitive events (OR 0.90, 0.83-0.98) (Peng et al., 2026). Both rest on self-reported (FFQ) diet, both are confounded with healthy-user eating, and both are null for Alzheimer’s disease specifically (fruit-and-veg AD OR 0.88, 0.76-1.01; flavonoid AD OR 0.90, 0.69-1.17) (Zhou et al., 2022) (Peng et al., 2026).
The two are not independent witnesses — flavonoids are a component of fruit and vegetables, so the flavonoid result re-expresses part of the same FFQ signal rather than confirming it by a second route — and the shared Alzheimer’s-null suggests the signal rides the cardiometabolic route the big rocks already cover, not a separate anti-neurodegenerative effect. This is one mediated signal that might help a little: eat the whole plant foods for their better-evidenced reasons, not for a flavonoid cognition target this evidence cannot isolate. (Peng et al., 2026; inferred from Zhou et al., 2022) -> Flavonoid Intake and Cognitive Function, Fruit and Vegetable Intake and Health
Things to look out for (the readouts)
These are self-monitoring signals, not diagnoses. Each points back to a lever and, when it stays out of range, to a clinician — the guideline itself never names or doses a drug.
- Blood pressure is the best-evidenced downstream readout: lowering it cuts major cardiovascular events about 10% per 5 mmHg, the relative benefit broadly constant across baseline pressure (Blood Pressure Lowering Treatment Trialists Collaboration, 2021), with the lifestyle handles being weight, activity, alcohol and sodium (figures under Sodium and blood pressure above). A pressure that stays high despite the levers is the medication conversation. Lowering blood pressure also lowers dementia risk: pooling five double-blind placebo-controlled antihypertensive trials, treatment cut incident dementia (OR 0.87, 0.75-0.99, at a mean BP lowering of 10/4 mmHg) (Peters et al., 2022) — but that proven effect runs through medication, which stays off this page, so read it as one more reason to watch the reading and take a persistently high one to a clinician, not as a sixth lever (inferred from Peters et al., 2022). -> Blood Pressure Lowering and Cardiovascular Events
- apoB (or LDL cholesterol) is the one blood marker the fabric treats as an even-handed, individually-measurable discriminator across very different diets. The lipoprotein-ASCVD dose is cumulative — the log-linear association with risk holds across >2 million participants, and «this effect appears to increase with increasing duration of exposure to LDL-C» (Ference et al., 2017) — so a level that has run high for years matters more than a single reading. In the insulin-resistant / hypertriglyceridemic person LDL-C can under-state the particle burden, where «direct measurement of LDL particle number or apoB concentration … may more accurately reflect the causal effect of LDL on ASCVD» (Ference et al., 2017). Lifestyle moves it only partly; a persistently high apoB routes to a clinician. -> LDL ApoB and Cumulative Exposure
- Fasting glucose / HbA1c is a metabolic-status readout, moved by the adiposity and activity levers (the DiRECT remission ladder above is the same axis). Read it as a marker to stratify with, not a lever to pull: cheap insulin-resistance proxies such as the triglyceride-glucose (TyG) index are «considered a reliable surrogate marker of insulin resistance» (Liu et al., 2022) and predict cardiovascular incidence, but «there was no association between the TyG index and mortality» (Liu et al., 2022) — a surrogate for a surrogate, useful for placing a person, not a number to steer by. -> Insulin Resistance Surrogates and Cardiovascular Risk
- Waist / central adiposity is the at-home proxy for the personal fat threshold — steer the adiposity lever by it rather than the scale (Taylor, above). -> Ectopic Fat and Depot-Specific Risk
- (65+) grip strength is the functional readout of the elderly stack: grip predicts mortality (HR ~1.16-1.20 per 5 kg lower) but is a marker to track, not a target a supplement moves (Celis-Morales, below). -> Grip Strength and Mortality
The sign-flippers to leave to a clinician (older adults)
Testosterone raised fractures in the TRAVERSE trial (HR 1.43, 1.04-1.97) rather than lowering them, with no mortality benefit (Snyder et al., 2024). Calcium-plus-vitamin-D did not reduce fractures in non-deficient older adults and raised kidney-stone risk (RR ~1.18) (Kahwati et al., 2018). The overall primary-prevention statin benefit is small (all-cause RR ~0.92, pooled across ages, not an age-specific figure) and the US Preventive Services Task Force reaches no recommendation at age >=76, judging the evidence insufficient there (US Preventive Services Task Force, 2022).
The most intensive blood-pressure target helps (SPRINT all-cause mortality HR 0.73; number-needed-to-treat ~61 to prevent one primary cardiovascular event, a different endpoint) but causes more hypotension, syncope and acute kidney injury (SPRINT Research Group, 2015). And grip strength and muscle mass predict mortality (grip HR ~1.16-1.20 per 5 kg lower) but are markers to track, not targets a supplement can move (Celis-Morales et al., 2018). -> Testosterone Adiposity and Muscle, Grip Strength and Mortality
Evidence box
Question ’For any adult across the whole population — including metabolic impairment, overweight and obesity, not only the healthy — which modifiable lifestyle exposures carry the largest expected effect on patient-important outcomes, ranked by effect size x certainty with per-stratum modifiers, stated as short directives, where health effect is the only axis weighed, pharmacotherapy is out of scope, and the whole is assembled bottom-up from held evidence rather than anchored to existing guidance?‘ Evidence included 49 sources — 30 gold, 18 high, 1 moderate Overall certainty Medium (see Rating Certainty of Evidence) Source-selection note 1 source(s) below the gold evidence bar feed this page: Taylor (mechanism, moderate). Each labelled by tier; none load-bearing for the core claims. Last updated 2026-09-04 · Independently reviewed: No · Full edit history