MSPH Events

12noon – 1pm, Thursday 4th December 2008
Centre for MEGA Epidemiology, Seminar Room 139, Level 1, 723 Swanston Street, Carlton

Lyle Gurrin from the Centre for MEGA Epidemiology will present ‘Synthesising the evidence for the association between Transferrin gene G277S and serum iron levels by combining sample means of continuous, binary and truncated data‘ at the Melbourne School of Population Health’s Work in Progress (WIP) Seminar.

Meta-analysis is the statistical tool of choice for summarising the research literature.  One goal of more broad forms of quantitative data synthesis is to reconcile estimates of a single parameter with information on a possibly complex function of several parameters including the target parameter. We encountered just such a situation in the HealthIron study of genetic and environmental modifiers of hereditary haemochromatosis when trying to combine our data with existing results on the association between mean serum iron levels and the G277S substitution in the Transferrin gene, which is thought to be a risk factor for iron deficiency anaemia in women.  Data are available from three other published studies on genotype-specific mean serum iron levels.  In one study these means are not presented separately for pre- and post-menopausal women, but the proportion of women with measurements below a fixed threshold is available for pre-menopausal women.  In another study data are presented only for women with measurements below a similar fixed threshold.  I will outline our method for combining these summary statistics with individual participant data from HealthIron that accommodates the longitudinal nature of the design.  Models were fitted using a full probability specification in the WinBUGS software.  Our results suggest that women with one copy of the G277S substitution have serum iron levels that are on average 10% lower (s.e. 6%) than women who are wild-type for the Transferrin gene, but the effect is likely to be much greater in pre-menopausal women. 

BIOGRAPHY:  Lyle Gurrin is a Senior Lecturer in Biostatistics at the Centre for MEGA Epidemiology.  This is his first foray into evidence synthesis, although he has always had a statistical soft spot for Bayesian methods.

 All welcome. No RSVP required.

All enquiries to Kellie Vizard, Tel: 03 8344 0671