A new analysis of nationwide health records from South Korea offers a detailed look at how income is tied to health in midlife—and why those health differences may not fully account for later gaps in survival.
Researchers Liran Einav, Amy Finkelstein, and Sehyun Hong studied “prime-age” adults by linking individual cardiovascular screening results to income tax data and later mortality outcomes. Their dataset drew on a representative sample of roughly 350,000 Koreans between the ages of 47 and 57 who underwent population-wide health examinations, allowing the team to compare objective clinical measures, self-reported behaviors, and subsequent deaths across the income distribution.
Across both men and women, the authors found a clear pattern for several major risk factors: higher income was associated with lower rates of elevated blood pressure, high blood glucose, and high body mass index (BMI). In other words, these conditions showed a negative income gradient—health improved as income rose. High cholesterol, however, did not follow the same income-linked pattern, standing out as an exception among the clinical measures examined.
The strength of the relationship between income and disease also differed by sex. For the health conditions that did vary by income, the gradients were steeper among women than men, indicating larger health differences across income levels for women in this midlife group. Importantly, these disparities were not evenly spread throughout the income range. Instead, the differences tended to be concentrated in the upper half of the income distribution, suggesting that much of the action occurs among middle- and higher-income groups rather than being driven solely by the poorest households. The same broad shape appeared when the researchers examined health behaviors, with behavior-income gradients also stronger at the top half of incomes.
When the team turned to mortality, the income pattern again appeared concentrated among higher earners, but with a different gender profile: the mortality-income gradient was steeper for men than for women. That is, men’s later death rates varied more strongly with income than women’s did, even though women showed sharper income gradients for several midlife disease measures.
A central takeaway of the study is that these midlife clinical indicators and behaviors—while strongly predictive of who dies later—do not explain much of the income gap in mortality. Even for cardiovascular deaths, where blood pressure, glucose, and BMI would be expected to matter a great deal, differences in measured health and behaviors by income accounted for only a small share of the subsequent mortality-income relationship.
The findings point to a broader conclusion: whatever drives the well-documented link between income and longevity may not primarily be differences in disease burden already visible in the late 40s and 50s. Instead, the authors suggest, other factors beyond measured midlife health—potentially including later-life disease development, access to or quality of care, cumulative exposures, social conditions, or other unmeasured risks—appear to play a larger role in shaping mortality differences at older ages.
The researchers thanked the Korean National Health Insurance Service for access to the data (approval no. NHIS-2023-1-821) and acknowledged collaborators and conference participants for feedback. Hong reported support from France’s Agence Nationale de la Recherche (ANR-17-EURE-001). The authors also noted standard disclosures, including that the views expressed do not necessarily reflect those of the National Bureau of Economic Research, and disclosed advisory and equity ties involving Nuna Health.
Why it matters
Income-related gaps in life expectancy are a major public policy concern, and this study suggests the roots of those gaps may not be captured by standard midlife risk-factor screenings alone. For health systems and governments, the results imply that reducing mortality inequality may require more than improving midlife detection of hypertension or diabetes. The next challenge is identifying which later-life mechanisms—care access, treatment differences, environment, or other social determinants—actually translate income into longer lives.