Methods & Issues seminar: A bit of a DAG: a rationale for variable selection in assessing the causal effect of serum urate on cardiovascular disease incidence

Thursday, 24 May 2012
12:30 p.m. - 1:30 p.m.

Location: At Room 730-373, Building 730, Tāmaki Innovation Campus

Speaker: Dr Simon Thornley
Department: School of Population Health

In the era of electronic health records, epidemiologists may need to make decisions about which of an encyclopaedic range of exposure variables to add into a statistical model. Serum urate has attracted some controversy in recent years as an indicator of CVD risk, with varying estimates of effect and a range of approaches to model building. With South Auckland described as the ‘gout capital of the world’, Simon re-examines this issue. He also introduces DAG: directed acyclic graphs, a concept first proposed by computer scientist Judea Pearl, and how they informed decisions about variable selection.

Simon’s PhD focuses on estimating the effect of a range of exposures on the incidence of cardiovascular disease (CVD), using data from primary care, laboratory and national healthcare and mortality records.

Dr Simon Thornley is a PhD student, Professional Teaching Fellow and Public Health Physician in the Section of Epidemiology and Biostatistics. He has varied research interests which include applied biostatistics, addiction and chronic disease epidemiology.