Causal effects of education on cognition : how do we generate evidence?

Abstract

Background: Education is a key institution in our societies, and should prepare us for future by improving cognition and teaching us needed life skills. Education is associated with many aspects of life, including health. For example, there is hope that improved education may help to reduce the burden of dementia, which is a large public health challenge for which treatment is missing. However, such hopes assume causality of the relationship between education and dementia.

Aim: The primary aim of this doctoral thesis was to examine the relationship between formal education and cognition (i.e. early-life cognition, cognitive decline and neuropathological disturbances to cognition in form of dementia) during the life-course. The secondary aim was to discuss how we can generate evidence on causal relationships and infer causation in epidemiology.

Methods: This thesis adopted a plurality and triangulation of evidence approach with regards to arriving at causal conclusions. The individual studies employed diverse designs in the exploration of links between education, cognition and dementia. Study I described thirty-year trends in the burden of dementia in inpatient records according to educational level. In Study II, we exploited a quasi-experimental comprehensive school reform in order to ascertain causal relationship between education and intelligence in men at military conscription. Further, we explored the heterogeneity of the effect according to childhood socioeconomic position. Study III focused on dementia. In order to investigate the causal effect of education on dementia diagnosis in Swedish registers, we used a primary schooling reform as a natural experiment. The reform had minimal spill-over effects on adult socioeconomic factors. The last study (Study IV) was a systematic review and meta-analysis summarizing the evidence from population-based studies of healthy adults. It examined the association between education and change in episodic memory, a cognitive domain with strong links to dementia.

Results: The burden of dementia in Swedish inpatient records began to decline during the last half a decade. Educational inequalities in dementia incidence remained stable and those with the highest educational attainment had the lowest dementia incidence rates. The comprehensive school reform increased intelligence and reduced socioeconomic disparities in cognition. However, in Study III we did not discover any substantial effect of the primary schooling reform on dementia risk. Similarly, the meta-analytic estimate indicated that the association between education and age-related decline in episodic memory is negligible.

Conclusions: Education is associated with level of cognition, but not decline - at least not in the episodic memory domain. Further, prolonged education cannot be uncritically assumed to reduce dementia burden, especially in absence of spillover effects to adult socioeconomic factors. However, education fulfils one of its many aims by increasing early-life cognition, and also has the potential to reduce socioeconomic inequalities in cognitive ability.