Learning Through Chain Event Graphs: The Role of Maternal Factors in Childhood Type I Diabetes

Chain event graphs are a graphical representation of a statistical model derived from event trees, previously applied to cohort studies but not to case-control studies. We apply the chain event graph framework to a Yorkshire case-control study of childhood type I diabetes, to examine four exposure variables associated with the mother, three of which are fully observed (her school-leaving-age, amniocenteses during pregnancy and delivery type) and one with missing values (her rhesus factor), while incorporating previous type I diabetes knowledge. We conclude that the unknown rhesus factor values are likely to be missing not at random, and are mainly rhesus positive. The mother's school-leaving-age and rhesus factor are not associated with the diabetes status of the child, whereas having at least one amniocentesis and, to a lesser extent, delivering by cesarean are, with the combination of both procedures further increasing the probability of diabetes. This application of chain event graphs to case-control data allows for the inclusion of missing data and prior knowledge, while investigating associations in the data. Communication of the analysis with the clinical expert is more straightforward than with traditional modelling, and this approach can be applied retrospectively or when assumptions for traditional analyses are not held.