Gene Resilience: Why Your Ethnicity Changes the Health Rulebook
This is Part 2 of a data analysis covering 159 countries and 7,853 individual Americans. Part 1 showed that your waist circumference predicts health 12x better than your diet. This post asks a harder question.
When I started breaking the CDC’s National Health and Nutrition Examination Survey (NHANES) data down by ethnicity, patterns emerged that made me uncomfortable. Not because they were racist, but because they were real enough to deserve a careful look.
In this NHANES sample, 68% of “thin” Asian-Americans still showed at least one bad metabolic marker. Black Americans at the top and bottom of the income split had the same average HbA1c (the marker doctors use for average blood sugar over the past 2–3 months). The body mass index (BMI) chart looked late for Asian populations. And Asian-Americans in this sample developed diabetes far faster than peers in Japan. Those are observations from this dataset, not universal biological laws, and they hint at what I’ve started calling Gene Resilience: each ethnic group has its own metabolic strengths and weaknesses, and the same advice rarely fits two of them.
These aren’t stereotypes. They’re patterns from roughly 8,000 real medical exams, and they need to be read with humility. The same n=1 humility applies to any extreme biohacking protocol, for example the growth hormone cost of polyphasic sleep varies wildly by individual and remains almost entirely unstudied at the population level. My goal here is not to claim final answers. It’s to show where the data suggests different groups may need different screening priorities.
I’m a behavioral designer, not a geneticist. But as a Taiwanese-American whose own body mass index (BMI) of 29.8 and waist of 100 cm put me squarely in the danger zone for my ethnicity, these findings are personal.
Important note: All findings are observational correlations from cross-sectional data. Ethnicity in NHANES is self-reported and captures both genetic and cultural factors. Full methodology in Part 1.
Evidence and limits
This post combines my own NHANES cuts with public country-level datasets. If you want to inspect the backbone, start with the NHANES 2017-March 2020 pre-pandemic release, the full methodology in Part 1, and the WHO expert consultation on BMI in Asian populations.
Whenever I use phrases like “suggests,” “points toward,” or “likely,” I mean correlation inside this dataset, not proof of biological cause.
Speed Run Notes: 8 Findings in 30 Seconds
From NHANES data on 7,853 Americans and 159-country health databases:
Your ethnicity changes the rules. Asians get diabetes at BMI 23, not 25. The BMI chart is wrong by 6 full points for Asian populations. 68% of “thin” Asian-Americans already have at least one bad health marker. America gives Asian-Americans 4x the obesity rate of their ancestral countries (and ~1.5x the diabetes rate).
Money doesn’t fix everything. Wealthy Black Americans have identical blood sugar to poor Black Americans (Cohen’s d = 0.00). Income helps Whites and Mexicans, but not Black Americans.
Each ethnicity has its own “kryptonite.” Asians: metabolic disease at low BMI. Whites: cancer, heart attacks, arthritis. Blacks: pre-diabetes and stroke regardless of income. Mexicans: liver damage. No group is globally healthier — risk routes through different channels.
Same evidence-driven approach, applied to behavior: The Octalysis Framework maps the 8 Core Drives behind why people actually change. Used by Google, Tesla, and LEGO across 1,000+ projects.
Table of Contents
Author Credibility: Yu-kai Chou
Yu-kai Chou created the Octalysis Framework after studying gamification since 2003 — years before the term entered mainstream vocabulary. As a Human-Systems Architect & Behavioral Designer, his framework has been applied by LEGO, Microsoft, Porsche, Coca-Cola, Salesforce, and MrBeast, impacting over 1.5 Billion Users.
Chou has taught the Octalysis methodology at Harvard, Stanford, Yale, Tesla, Google, BCG, and IDEO.
His work has been cited by Harvard, Stanford, MIT, Forbes, Wall Street Journal, Wired, US Department of Energy, NIST, NSF, NCBI, US Department of Education, ClinicalTrials.gov, and Google Scholar — with 3,700+ more academic publications. Explore his books here.
1. Korea Eats 3,375 Calories a Day and Lives to 84
Among the 32 wealthiest countries (gross domestic product, or GDP, over $30K per capita), there is zero correlation between total daily calories and life expectancy.
| Country | Daily Calories | Life Expectancy |
|---|---|---|
| Japan | 2,652 | 84.7 |
| South Korea | 3,375 | 83.7 |
| Ireland | 3,918 | 82.4 |
| United States | 3,847 | 77.0 |
| Denmark | 3,815 | 81.5 |
Korea eats 725 more calories per day than Japan and lives nearly as long. Ireland eats 3,918 calories and outlives America by 5.4 years.
The difference isn’t how much you eat. It’s what you eat. Korea consumes 220 kg of vegetables per person per year. The US: 122 kg.
“Eat less” is the wrong advice. “Eat differently” is closer to the truth.
2. The US Is Dead Last Among 32 Rich Countries in Life Expectancy
I ranked all 32 countries with GDP over $30,000 per capita by life expectancy.
The United States came in 32nd out of 32. Behind Oman, Saudi Arabia, Kuwait, and Bahrain.
It’s not close. The US (77.0 years) trails the #1 spot (Hong Kong/Japan, 84.7) by nearly 8 years. That’s despite having the highest GDP per capita of any large country in the dataset.
Taiwan, with less than half America’s GDP per capita, outlives the US by 4 years. The US has the highest calorie intake (3,847/day), the highest sugar percentage (15.2% of calories), and among the lowest vegetable consumption of any wealthy country.
3. Japan Sleeps the Least and Lives the Longest
I merged our 159-country health dataset with sleep data from 50 countries (World Population Review, based on app tracking data). The results contradicted what I expected.
Japan averages 5 hours and 52 minutes of sleep per night, dead last out of 50 countries. South Korea: 6 hours flat. These are the two shortest-sleeping countries in the dataset.
They also rank 2nd and 3rd for life expectancy.
| Country | Avg Sleep | Life Expectancy | Diabetes Rate |
|---|---|---|---|
| Japan | 5h 52m | 84.7 | 8.1% |
| South Korea | 6h 02m | 84.3 | 6.9% |
| Kuwait | 6h 15m | 80.4 | 25.6% |
| United States | 7h 06m | 79.3 | 10.7% |
| Netherlands | 7h 24m | 82.2 | 5.0% |
| New Zealand | 7h 27m | 82.1 | 6.7% |
Across all 50 countries, sleep hours do not significantly predict life expectancy (r = 0.25, p = 0.09). After controlling for calorie intake, the correlation drops to near zero (r = 0.11, p = 0.45). A 2025 PNAS study analyzing 71 countries found the same thing: national sleep duration doesn’t predict life expectancy, heart disease, or diabetes in the expected direction.
Now flip the question. Sleep does predict diabetes at the country level, and strongly (r = -0.65, R² = 0.43). Countries that sleep less have more diabetes, even after controlling for GDP, calories, and obesity rates. That’s a bigger effect than anything else I measured.
Except Japan and Korea break the pattern entirely. They sleep the least and have among the lowest diabetes rates. Kuwait sleeps almost as little and has 3x the diabetes. One plausible explanation is diet, but this dataset can’t isolate cause. Japan and Korea consume 215-220 kg of vegetables per person per year, while Kuwait and Mexico consume far less.
Within the US (NHANES data), short sleep does correlate with worse health: people sleeping under 6 hours have an HbA1c of 6.04 vs 5.84 for 7-8 hour sleepers (Cohen’s d = 0.17). But oversleeping shows no penalty (d = 0.00).
The way I read this: sleeping less than your culture’s norm may signal individual stress or health problems. But a culture that simply sleeps less isn’t necessarily worse off, especially if their diet compensates. The same pattern of “great pitch, clunky first run” shows up in my Brave Browser onboarding UX audit.
4. Money Fixes Health for Some Races, Not Others
We compared the poorest Americans (below poverty line) with the wealthiest (over 4.5x poverty line) within each ethnic group:
| Ethnicity | Poor HbA1c | Rich HbA1c | Gap | Cohen’s d |
|---|---|---|---|---|
| Asian | 5.98 | 5.82 | 0.16 | 0.14 |
| White | 5.75 | 5.59 | 0.15 | 0.16 |
| Mexican | 6.18 | 5.82 | 0.35 | 0.25 |
| Black | 6.01 | 6.01 | 0.00 | 0.00 |
For Black Americans in this NHANES slice, being wealthy did not correspond to lower HbA1c. Rich Black Americans and poor Black Americans both averaged 6.01.
Money helps Asians and Whites a little (d ≈ 0.15). Mexicans somewhat more (d = 0.25). Black Americans: nothing.
Income alone doesn’t explain the gap in this sample. This dataset can’t cleanly separate income from care quality, chronic stress, neighborhood conditions, or other structural factors. Larger cross-sectional work has also found glycemic-control disparities persisting even after adjusting for access-to-care variables.
5. 68% of “Thin” Asian-Americans Are Already Metabolically Sick
When we looked at Asian-Americans with “normal” BMI (under 25), a full 68% had at least one bad health marker: pre-diabetic blood sugar, high blood pressure, or high cholesterol. Only 54% of normal-weight White Americans had the same problem in the same sample.
Then we compared “metabolically healthy obese” White Americans (BMI 30+, but with perfect blood sugar, blood pressure, AND cholesterol) against “metabolically unhealthy normal weight” Asian-Americans:
| Group | HbA1c | BP | Chol | n |
|---|---|---|---|---|
| Obese but healthy Whites | 5.26 | 114 | 168 | 206 |
| Thin but sick Asians | 5.82 | 125 | 210 | 194 |
The “thin” Asians were worse across every marker. Every single one.
If you’re Asian and your BMI is “normal,” don’t assume you’re healthy. Get your HbA1c, blood pressure, and cholesterol checked. The scale is lying to you.
Effect size: Cohen’s d = 0.68 (large).
6. The BMI Chart Is Wrong by 6 Full Points
The World Health Organization expert consultation kept the global BMI chart, but it also flagged lower public-health action points for many Asian populations. The standard BMI 25 “overweight” line was built around European datasets.
Here’s what our data shows:
| BMI | % Asian Pre-Diab | % White Pre-Diab |
|---|---|---|
| 20-23 | 32% | 15% |
| 23-25 | 46% | 20% |
| 25-27 | 52% | 30% |
| 27-29 | 57% | 34% |
| 29-31 | 56% | 31% |
At BMI 25, the official “overweight” line, nearly half of Asians are already pre-diabetic. Whites don’t hit that level until BMI 34. The threshold is 6 BMI points too late for Asians.
That’s a lot of people walking around thinking they’re fine because a chart designed for Europeans told them so.
In practice, the consultation added 23 as an important action point for increased risk in many Asian populations, even though the global categories were retained. The American Diabetes Association adopted the lower threshold in its 2015 Standards of Medical Care, citing Hsu et al. on optimum body mass index cut points for Asian-American screening. Most doctors and health apps still show the universal number first.
7. America Gives Asians Diabetes 4x Faster Than Their Home Countries
Asian-Americans (age 30-60) in our data: 16.2% obese (BMI ≥ 30), 42.5% pre-diabetic, 10.9% diabetic.
Japan’s benchmarks (per World Obesity Federation and OECD Health Statistics): ~4% obese, ~7% diabetic. Broadly similar ancestral background, very different food environment. In this comparison, the US sample shows roughly 4x the obesity and about 1.5x the diabetes.
White Americans roughly double their European peers’ obesity rates (46% vs 20-25%). Asians quadruple theirs, though part of this larger multiplier may reflect the lower baseline in Asian countries, where calorie intake and portion sizes start much smaller.
One plausible explanation, consistent with Finding 6, is that Asians develop metabolic problems at much lower BMI thresholds. Even modest weight gain, which the American food environment makes likely, may push Asians past metabolic thresholds faster than it does for other groups. This dataset supports the pattern, not a single definitive mechanism.
8. “Gene Resilience”: Each Ethnicity Has Its Own Kryptonite
With 9 different health conditions measured across 4 ethnic groups, a pattern emerged that I didn’t expect. Each genetic background seems to have a specific vulnerability profile, and each has conditions it’s unusually resistant to.
| Condition | Asian | White | Black | Mexican |
|---|---|---|---|---|
| Pre-diabetes | 40.5% | 33.4% | 47.1% | 39.3% |
| Depression | 3.0% | 8.4% | 7.8% | 7.6% |
| Systemic inflammation | 17.8% | 31.7% | 35.1% | 34.3% |
| Liver damage (ALT) | 7.5% | 7.0% | 4.8% | 11.6% |
| Gout risk | 11.2% | 11.3% | 13.5% | 8.6% |
| Cancer (ever) | 4.4% | 18.7% | 7.6% | 5.8% |
| Heart attack | 2.1% | 6.9% | 3.9% | 2.5% |
| Stroke | 1.9% | 6.4% | 6.8% | 2.6% |
| Arthritis | 14.3% | 39.0% | 31.4% | 20.6% |
(Bold = highest rate for that condition. Italic = lowest.)
Each group’s vulnerability profile:
Asian Americans: The metabolic vulnerability is clear (40.5% pre-diabetic, and still the weakest profile once you compare the lower-BMI bands in Finding 6). But they have the lowest rates of depression (3%), cancer (4.4%), heart attack (2.1%), stroke (1.9%), arthritis (14.3%), and systemic inflammation (17.8%). Their bodies look metabolically fragile in this sample, but more resilient against inflammatory, cardiovascular, and mental-health conditions.
White Americans: The inverse. They handle metabolic stress better than any other group (lowest pre-diabetes at the same BMI). But they carry the highest burden of cancer (18.7%), heart attacks (6.9%), arthritis (39%), and depression (8.4%). Something about either European genetics or the lifestyle patterns associated with White Americans makes them vulnerable to chronic diseases of aging and inflammation even while their blood sugar stays relatively controlled.
Black Americans: The data paints a concerning picture. Highest pre-diabetes (47.1%), highest inflammation (35.1%), highest stroke (6.8%). And as Finding 4 showed, income alone didn’t shift HbA1c in this slice. The consistency of these elevated rates across metabolic, inflammatory, and cerebrovascular categories suggests income isn’t the only layer in play.
Mexican Americans: A mixed profile. Moderate pre-diabetes, but the highest liver damage rate (11.6%, using ALT, a common liver-enzyme marker, and more than double Black Americans). Low heart attacks and strokes. The liver vulnerability is noteworthy because it’s consistent with published literature on nonalcoholic fatty liver disease (NAFLD) showing higher burden in Mexican-origin and broader Hispanic populations.
I started calling this framework “Gene Resilience” because each group seems to have built-in strengths and weaknesses that the data exposes when you look across enough conditions. It’s not that any group is globally “healthier.” It’s that each group’s genetics (combined with cultural and environmental factors I can’t fully separate) routes health risk into different channels: the same vulnerability-keyed lens I apply to motivation in my other frameworks.
The practical implication: the health screening you need depends on your genetic background. For Asian Americans, waist and blood sugar deserve more attention than cardiac risk. Cancer screening and joint health may matter more for White Americans than their HbA1c suggests. And for Black Americans, inflammation and stroke markers deserve monitoring regardless of income level. These aren’t stereotypes. They’re statistical patterns in roughly 8,000 real medical exams.
How We Did This
This analysis uses NHANES 2017-March 2020 pre-pandemic data (7,853 Americans), Global Burden of Disease data (159 countries), and multiple supplementary datasets. All effect sizes are reported as Cohen’s d or R². We tested every finding across all 4 ethnicities, but the data is still observational rather than causal.
Read the full methodology, data sources, and limitations in Part 1 →
Use these patterns in your own life and your own design work
If you found these patterns useful for your health, the same kind of data-driven analysis underpins Part 1 on why your waist beats your food diary 12-to-1. And if you design products, services, or programs that try to change human behavior, my Octalysis Framework is the same kind of “look at the actual evidence, not the inherited assumption” treatment, applied to motivation instead of metabolism.
These patterns aren’t comfortable. But pretending they don’t exist means giving the same health advice to every group, advice that was designed for and tested on White Europeans.
The health screening you need depends on your genetic background. For Asian-Americans, waist circumference and blood sugar deserve more attention than cardiac risk. Cancer screening and joint health may matter more for White Americans than their HbA1c suggests. And for Black Americans, inflammation and stroke markers deserve monitoring regardless of income level.
If you haven’t read Part 1 yet: Your Waist Knows More Than MyFitnessPal covers why your waist circumference predicts health 12x better than your food diary.
— Yu-kai Chou