Principal Investigator: Professor Nishi Chaturvedi
Department: Institute of Cardiovascular Sciences
Institution: University College London (UCL)
University College London, Institute of Cardiovascular Sciences, Gower Street,
London, W1E 6BT
Type 2 diabetes is associated with increases in cardiovascular disease (myocardial infarction,
stroke and heart failure), cognitive decline and cancer. HbA1c, a blood test which indicates
blood glucose levels (glycaemia) over the preceding 3 months, regardless of diabetes status,
is available on all Biobank participants at baseline. We will examine risks of cardiovascular
disease, cognitive decline and cancer across the glycaemic spectrum, including in people
without diabetes. We will investigate how risks differ by gender and ethnicity. Additionally,
we will explore the impact of anti-diabetic, anti-hypertensive and lipid-lowering medication
on cardiovascular and cognitive outcomes throughout the glycaemic range.
Since HbA1c measures chronic hyperglycaemia more reliably than random/ fasting blood
glucose testing alone, the availability of this measure in the Biobank cohort provides a new
opportunity to investigate the impact of glycaemia on health outcomes. UK Biobank has
large numbers of participants with baseline glycaemic status and incident measures of
disease, thus providing a unique and powerful instrument to address these questions. By
defining the role of sub-clinical hyperglycaemia in cardiovascular disease, cancer and cognitive decline, we aim to improve the prevention and diagnosis of these diseases, in line
with UK Biobank’s core goals.
To address these aims, we will require access to data only, to include the following health
outcomes records: HES, GP data, cancer and death registers. Measures of HbA1c, sociodemographics,
lifestyle factors, anthropometry, cardiovascular risk factors (e.g. blood
pressure and cholesterol) and medical and drug history from baseline will all be required.
We will use statistical models to show how baseline HbA1c relates to the health outcomes
of interest, what other factors influence these associations and whether associations are
different in population sub-groups.
We intend to use the full dataset of approximately 500,000 people. Outcomes are often
compared in populations with and without diabetes, but the validity of this binary
classification and widely used cut-off points is not established. Including people with no
diabetes as well those with treated and untreated diabetes offers an opportunity to
disentangle effects of diabetes treatments on disease.