Methodology, sources, and limitations

Where the numbers come from

Heart disease, stroke, cancer, respiratory, diabetes, and kidney mortality rates are drawn from the World Health Organization Global Health Estimates 2019/2022 release, reported as age-standardised deaths per 100,000 population. Life expectancy values come from the World Bank Development Indicators. Alzheimer's prevalence is from the Global Burden of Disease 2019 study by the Institute for Health Metrics and Evaluation (IHME).

For the 41 countries marked tier-1 ("rich evidence"), the cited study text is the actual landmark cohort or trial that established the dietary link — Adventist Health Study-2, PREDIMED, UK Biobank, the Lyon Diet Heart Study, the China Study, and others.

Blue Zone designations follow the National Geographic / Dan Buettner framework: Okinawa (Japan), Sardinia (Italy), Ikaria (Greece), Nicoya (Costa Rica), and Loma Linda (California).

The plant-based score

The "plant-based score" is an estimate of the share of total calories drawn from whole plant foods — grains, legumes, vegetables, fruit, nuts, and seeds — in a country's typical diet. It is not a measure of vegetarianism. A score of 70% means roughly seven of every ten calories the average resident eats come from whole plants; the remaining 30% can include animal products, refined foods, and added fats.

The score is anchored on FAOSTAT food supply data, the Alternative Healthy Eating Index where available, and adjusted for nationally-representative dietary surveys (NHANES for the US, EPIC for Europe, KoGES for South Korea, etc.). Where those datasets disagree, I use the median.

This score correlates strongly with downstream outcomes — heart disease, life expectancy, dementia — but it is not a causal lever. It is a proxy for an entire dietary pattern.

Tier-1 vs tier-2 countries

Tier-1 countries (~41) have all four headline values — heart, plant-based score, life expectancy, and Alzheimer's — drawn from a named, citable study. The panel shows that study and the year. Use these confidently.

Tier-2 countries (~138) have heart, plant, life expectancy, and Alzheimer's from WHO Global Health Estimates aggregate data. The panel labels these "WHO baseline data". Use them as broad-strokes context, not as evidence for a specific dietary claim.

Tier-2 secondary causes (stroke, cancer, respiratory, diabetes, kidney disease) are derived from formulas anchored to global correlations between heart disease and other diet-driven outcomes. They carry the "est." badge in the UI to flag this. The 35 most-populous tier-2 countries have these values overridden with actual IHME numbers; the rest are formula-derived approximations.

The formulas behind tier-2 secondary causes

For tier-2 countries without direct IHME data, the secondary cause-of-death rates are computed as:

stroke ≈ heart × 0.55 + (lifeExp − 70) × 0.5 cancer ≈ 82 + (lifeExp − 65) × 3.2 − (plant − 50) × 0.35 diabetes ≈ 8 + (100 − plant) × 0.42 + max(0, heart − 100) × 0.05 respiratory ≈ 35 + heart × 0.18 kidney ≈ 12 + heart × 0.07 + (100 − plant) × 0.1

These are descriptive correlations fitted to known WHO/IHME relationships, not predictions. They produce values in the right order of magnitude — never as precise as direct measurement. If a tier-2 country's secondary-cause numbers look strange, the formula is the most likely culprit, not the underlying biology.

Limitations and honest caveats

This is correlational data. Countries differ in healthcare access, smoking, alcohol, age structure, genetics, and economic development. Diet is one variable among many. The map's claim is not that diet causes these patterns, but that diet correlates with them strongly enough — and the mechanism is biologically well-established enough — that dietary change is the highest-leverage intervention available to most individuals.

The "1-in-N" risk figure in the panel is calculated as (heart × life expectancy) ÷ 100,000 inverted to a 1-in-N format. It is a rough lifetime cumulative risk and should be treated as illustrative.

Single-year snapshot. Most values represent a single recent year (typically 2019–2022). Trends over decades matter and aren't shown here.

Ecological fallacy. National averages don't predict individual outcomes. Living in a country with a 25% plant-based score doesn't mean your diet is 25% plant-based.

Translation of country narratives. Czech versions of the deep-dive content and prevention advice were translated by the author and have not yet been reviewed by a native medical writer. Spot inaccuracies should be reported.

The decades.plus study score

Every study in the Research hub carries a 0–100 decades.plus score and a derived tier (Landmark ≥80, Strong 60–79, Emerging <60). The score is a transparent weighted sum across six components, calculated deterministically at build time:

Study design 25 pts — RCT or meta-analysis 25, systematic review 22, prospective cohort 20, case-control 12, cross-sectional 6, ecological 5, mechanistic 4 Sample size 20 pts — n ≥ 100,000: 20; n ≥ 10,000: 15; n ≥ 1,000: 10; n ≥ 500: 7; below: 4 Funding independence 20 pts — fully public/academic 20, mixed 12, industry 4, undisclosed 8 Journal + peer review 15 pts — top-tier journal (NEJM/Lancet/BMJ/JAMA) 15, mid-tier 10, other peer-reviewed 7, preprint 2 Institution tier 10 pts — top research university or national-cohort body 10, mid 6, other 3 Replication 10 pts — ≥3 independent replications 10, 1–2 replications 6, none 2

The formula favours large, publicly-funded, peer-reviewed, replicated work — the kind of study that should anchor a public-health argument. It deliberately penalises industry-funded or unreplicated work, regardless of how impressive any single result looks. Two honest caveats:

The score reflects evidence quality, not the *size* of the dietary effect or its relevance to any individual. A landmark study on processed meat and colorectal cancer is graded "landmark" because the evidence is solid — not because everyone needs to act on it identically.

The weights are an editorial choice, not a derived constant. We will tune them as the catalogue grows and as readers push back. The scoring code in src/data/studies-score.ts is the source of truth — if you disagree, the disagreement is concrete and addressable.

Re-use and citation

All data on this map is publicly sourced from WHO, IHME, the World Bank, and peer-reviewed literature. The visualisation, code, and editorial framing are released under Creative Commons Attribution 4.0 — you may share, adapt, and republish with attribution to Štěpán Živnůstka.

For corrections, missing studies, or data updates, contact me on Instagram or by email (linked in the footer).