AbstractDespite reducing rates of cardiovascular disease in high income countries, individuals who are the most socioeconomically deprived remain at the highest risk of disease. The mechanisms by which the inequalities arise are still unknown. In this thesis I use causal inference methods, including Mendelian randomisation (MR), mediation analysis and polygenic scores, to understand the aetiology of educational inequalities in cardiovascular disease, using UK Biobank.
Establishing causality in epidemiology can be challenging, due to unmeasured (or mis-measured) confounding, measurement error and reverse causality. One method to overcome these sources of bias is MR. In this thesis I demonstrate using simulations and applied examples how MR can be applied to mediation analysis, identifying sources of bias and methodological limitations.
Using MR mediation methods and non-genetic (phenotypic) mediation methods I demonstrate that body mass index, systolic blood pressure and lifetime smoking behaviour mediate up to 40% of the association between education and cardiovascular disease. Intervening on these intermediate risk factors would likely reduce cases of cardiovascular disease attributable to low educational attainment.
I then investigate inequalities in prescribing of statins as a primary cardiovascular preventative medication. I identified clear inequalities, where for a given level of underlying cardiovascular risk (assessed via QRISK3 score) individuals with lower educational attainment were less likely to receive statins.
Finally, explore the role of education as an effect modifier of genetic suscepbility to cardiovascular disease.I demonstrate that on the additive scale, higher education protects against genetic susceptibility to body mass index and smoking but accentuates genetic susceptibility to low-density lipoprotein cholesterol and systolic blood pressure. On the multiplicative scale, higher education accentuates genetic susceptibility to atrial fibrillation and coronary heart disease.
This thesis demonstrates that body mass index, systolic blood pressure, smoking and statin use all likely contribute to educational inequalities in cardiovascular disease, whilst contributing to the development of methods to improve causal inference in social epidemiology.
|Date of Award||21 Jan 2021|
|Supervisor||Laura D Howe (Supervisor), Neil M Davies (Supervisor), Amy E Taylor (Supervisor) & George Davey Smith (Supervisor)|