UCSF

More schooling predicts better cardiovascular disease (CVD) outcomes, but a causal relationship between education and later-life blood pressure (BP) outcomes is less clear. Most prior work also evaluates the effect of education at the mean of BP, giving little insight into effects across the whole outcome distribution. We leveraged natural experiments, variation in compulsory schooling laws (CSLs) affecting K-12 education in the US as well as variation in the availability of colleges by state in the US, to evaluate how increased education impacted later-life BP. We also evaluate the intention-to-treat (ITT) association of college availability with later-life BP. We used quantile regression to evaluate the effects of education along the distribution of BP. We performed two-sample instrumental variable (IV) analyses using data from the US Census and from the Health and Retirement Study (HRS) to evaluate how variation in CSLs and college availability impacted later-life BP. In these analyses, we used linear regression in the first stage and both linear and quantile regression in the second stage. In the ITT analysis, we used linear and quantile regression. All models were adjusted for individual- and state-level covariates. To evaluate whether the effects of education varied by race and sex, we also generated results with race by sex interaction terms. We found no evidence of an effect of education on BP in the overall study population for the CSL IV analysis. In the college IV analysis, we found that increasing the number of colleges in a state decreased SBP, although confidence intervals included the null in CQR models. In the overall ITT analysis, we found no evidence of an association for 2-year colleges, but a small decrease in SBP associated with each additional 4-year college. In subgroup analyses, however, we found important differences in the effect of education. In the CSL IV analysis, among White women, each additional year of K-12 education consistently lowered SBP across the SBP distribution, while among Black men, education consistently elevated SBP across the SBP distribution. In the college IV analysis, among White women, each additional 2- or 4-year college consistently lowered SBP across the SBP distribution while effect estimates in other subgroups all included the null. In ITT analyses, we found that women had decreased SBP at the upper end of the SBP distribution while men had increased SBP over the same range. Black women consistently had decreased SBP associated with additional 2- and 4-year colleges, although estimates were imprecise. Quantile regression identified differential relationships between education and blood pressure distribution that mean regression is unable to estimate. Race by sex subgroup analyses showed differential impacts of education on later-life BP. Future research should determine if these results are consistent and to understand why the expected education and blood pressure gradient is reversed for Black men.