We develop data analysis techniques for researchers across the BRC to use and apply to their research projects. We also help assess the quality of all statistical analysis across the BRC to provide assurance to each project.
We ensure career development of junior staff through training in advanced analytical methods, mentoring by senior staff and encouraging participation in professional groups including the NIHR Statistics Group.
Some of our core activities include:
Methods to make causal inferences:
Researchers across the BRC seek to make causal inferences to provide evidence on the impact of interventions. This means they want to know if one thing (such as an inactive lifestyle) causes a particular outcome (such as diabetes).
To help make these causal links, researchers use various statistical methods and models, such as instrumental variable methods (particularly Mendelian randomization), and marginal structural models. These can help researchers assess whether a new treatment, for example, is helping patients to get better, or whether the patients just happened to get better at the same time as taking the new treatment.
Prognostic modelling and risk stratification (led by Professor Jonathan Sterne).
Several BRC research themes analyse large datasets to predict clinically important outcomes. Statistical models that help predict outcomes can directly affect patient management, for example by helping clinicians focus treatments or interventions on those at higher risk of illness or death. They can also inform the design of studies comparing different strategies for patient management, helping to provide more personalised patient care, by identifying groups most likely to benefit or at high risk of adverse outcomes.
Once we have evidence of a potentially modifiable causal mechanism, and a corresponding intervention, we will test this in early phase (e.g. using biomarker or surrogate outcomes), feasibility or pilot studies (these are planned by all Research Themes). We ensure high quality design, analysis and reporting of such studies.