Introduction Government spending on public goods (eg, education) and social assistance (eg, cash transfers) provides plausible investments in the social determinants of health. Among rich nations, countries with higher social spending and lower income inequality show longer life expectancies. However, studies of both factors have been limited by bias from residual confounding and reverse causation.

Methods This study examined data from the National Longitudinal Mortality Study on 431 637 adults aged 30–74 in 48 USA states followed for 11 years. State per capita social spending (total, welfare, education, health) and income inequality (Gini coefficient) were explored as predictors of individual mortality (all-cause, cardiovascular, cancer) using linear probability models. To reduce bias, models incorporated state and time fixed effects and instrumental variables, and controlled for state- and individual-level covariates.

Results Total public spending and spending on welfare and education, but not healthcare, predicted lower probabilities of death from coronary heart disease (CHD) (per $250 per capita spent on welfare: β=−0.016, p=0.03) and all causes combined (per $250 per capita spent on welfare: β=−0.031, p=0.03). There were weaker associations for colon cancer and no associations for stroke. A higher Gini coefficient predicted a higher probability of CHD mortality but not all-cause mortality. Stratified models suggested stronger effects among those aged 45–59 and with incomes <$25 000/year.

Conclusion Higher state social spending outside the healthcare sector may reduce one's chances of dying from heart disease and all causes combined, particularly for low-income, middle-aged adults. Policies promoting economic equality may further lessen CHD disparities.