The direct and indirect effects of low education on Covid-19 mortality: an observational study of one million Covid-19 deaths in the first two years of the pandemic in the USA

Background: While many studies have found a high correlation between socio-economic inequalities and risk of Covid-19 deaths, there is reason to believe that much of this association is the product of differing levels of education. This thesis aims to, first, estimate the association between education and Covid-19 deaths, and second, identify the mechanisms through which low education gives rise to higher risk of Covid-19 death.

Aims: The core of this thesis is two empirical papers already submitted for review to academic journals. The aim of the first paper is to measure the direct effects of low education on Covid-19 deaths, after controlling for both other risk factors and potential mediators. The aim of the second paper is to measure the indirect effects of low education through four mediators: vaccination, distrust in science, poor health, and lack of preventative health behavior.

Methods: The primary analytical tool of the thesis is multivariate models of fortnightly (n=60) Covid-19 deaths in 3108 US counties for the period 20 January 2020 to 10 May 2022. In paper 1, I model the direct (unmediated) effect of education, controlling for economy, race, geography, lack of vaccination, anti-science views (vote Republican), poor health, and lack of preventative health behavior. In paper 2, I model the indirect effect of education through four mediators: vaccination, distrust in science (Republican vote), poor health, and preventative health behavior.

Findings: Paper 1 finds a strong direct (unmediated) effect of low education on Covid-19 mortality (IRR [incidence rate ratios] = 1.17; 95% CI [confidence interval]: 1.15, 1.20). Paper 1 theorizes that this strong direct effect reflects education's impact on (1) collective cultures, such as norms of mask wearing, and (2) individual literacy, such as ability to engage with scientific communication. Paper 2 finds that all four potential mediators are significant mediators of the impact of education on Covid-19 deaths: vaccination (IRR [incident rate ratio] = 1.02; 95% CI: 1.012, 1.028); distrust in science (Republican vote) (IRR: 1.05; 95% CI: 1.039, 1.063); poor health (IRR: 1.01; 95% CI: 1.003, 1.014); and preventative health behavior (IRR: 1.01, 95% CI: 1.001, 1.018). Paper 2 argues that these strong indirect effects are evidence that low levels of education reduce an individual’s likelihood of having an agentic engaged approach to their own health.

Interpretation: Low education is strongly predictive of Covid-19 deaths, with an effect size of a university degree comparable to that of being over 65 in age. Low education likely accounts for between 1 in 10 and 1 in 7 deaths in low education counties. Low education appears to reduce an individual’s proactive and literate engagement with their own health, increasing their likelihood of death from Covid-19. Education needs to be conceptualized as a high-risk factor for Covid-19 death and be considered when attempting to combat Covid-19 in disadvantaged communities. Measures of low education, combined with measures of the four mediators identified in this thesis provide policy makers with characteristics of communities that are at highest risk, and in most need of interventions. The role of low education in generating both lack of agency and distrust in medical and scientific institutions presents serious challenges to addressing the vulnerability of low education communities.