Skip to content

Independent, primary-source health reference.

Data updated

VitalDecades Evidence-first healthy aging

Search

Type a word like "colorectal" or "hypertension". Search runs on the published site.

The Big Levers

The big levers: what the evidence says about diet, exercise, sleep, and more

Every brief states the bottom line first, grades the strength of the evidence behind it, and cites its primary source. Nothing here is personalized advice.

Strong evidence

  • Physical activity

    Meeting the federal minimum of 150 to 300 minutes of moderate aerobic activity a week, plus muscle-strengthening activity twice a week, is associated with a substantially lower risk of death across large pooled observational studies; going well beyond that minimum adds only modest further benefit on top of it.

    Strong

Moderate evidence

  • Sleep

    Large observational cohorts consistently link sleeping less than about seven hours a night to higher risk of cardiometabolic disease and, in one long-running cohort, to higher dementia incidence; almost none of this evidence is randomized, and the one study designed to test causation for dementia did not confirm it.

    Moderate

  • Social connection

    Large meta-analyses find people with stronger social relationships have roughly 50% greater odds of survival over a given follow-up period than people with weaker ones, and loneliness, social isolation, and living alone are each associated with roughly a quarter to a third higher mortality risk, a set of associations the 2023 U.S. Surgeon General's advisory judged comparable in size to other well-established mortality risk factors; this evidence is overwhelmingly observational, and randomized trials of interventions meant to reduce loneliness show smaller, less consistent effects than the observational association would suggest.

    Moderate

  • Strength and muscle

    Randomized trials, several rated high-certainty by Cochrane, show resistance training reliably improves strength and physical function in older adults, while exercise programs emphasizing balance and functional training substantially lower fall rates; how much any specific protein target adds on its own, and how far these gains extend into longer independent living, rests on thinner and more mixed evidence.

    Moderate

Mixed evidence

  • Alcohol

    The long-standing claim that moderate drinking protects the heart has weakened substantially once studies correct for the bias of comparing drinkers with people who abstain because they are already sick; genetic evidence finds the apparent protection against stroke is largely non-causal, all-cause mortality risk climbs with intake with no confirmed protective floor, and cancer risk rises from the first drink with no threshold identified below which risk is zero.

    Mixed

  • Weight

    BMI is a weight-for-height ratio that cannot distinguish fat from muscle or bone, and in adults over 65 several large meta-analyses find higher mortality risk at the low end of the "normal" BMI range rather than at the high end; separately, randomized trials of intentional weight loss show real benefit for preventing diabetes, no significant cardiovascular-event benefit in one large trial of adults who already had diabetes, and better preservation of muscle and bone when weight loss is combined with exercise rather than diet alone.

    Mixed

  • Cognitive health

    The 2024 Lancet Commission estimates that 14 modifiable factors collectively account for about 45% of dementia cases worldwide, a population-attributable-fraction calculation built from observational epidemiology rather than a randomized-trial-proven number; one large randomized multidomain trial found a statistically significant but modest benefit on a composite cognitive test score, and randomized evidence for commercial brain-training programs shows gains largely confined to the trained task rather than reliably transferring to broader cognitive function.

    Mixed

  • Nutrition patterns

    Randomized trials support specific dietary patterns for specific outcomes, DASH for blood pressure and PREDIMED for cardiovascular events, though PREDIMED's original publication was retracted and republished after a randomization irregularity; a broader claim that any single named pattern reliably slows aging or preserves cognition rests on much weaker, largely observational or null evidence.

    Mixed