Maternal haemoglobin in pregnancy: trajectories, risk factors and consequences for offspring in ALSPAC

Rationale
Maternal haemoglobin (Hb) concentrations are known to change across pregnancy.(1) Women who have Hb concentrations at either the bottom or the top end of the distribution are at increased risk of delivering a low birthweight and premature baby.(2-3) However, a greater decrease in Hb between trimesters is associated with a reduced risk of preterm birth.(4) Risk factors such as older age, overweight, and multiparity are associated with maternal anaemia,(5) but it is unclear if they are also associated with the pattern of change in Hb across pregnancy.
Iron is essential for cell differentiation growth and has a role in the functioning of the immune system(6). It has generally been assumed that infants born at term and with an adequate birthweight have adequate iron stores for the first months of life, but studies (e.g. 7,8) have shown an association between maternal pregnancy Hb and iron status in infants suggesting that maternal Hb concentrations may affect offspring health outcomes.

Aims & Objectives
To describe the average pattern of change in pregnancy haemoglobin concentrations.
To study associations of potential risk factors with maternal haemoglobin concentrations.
To investigate associations with offspring growth, infections, cardiovascular health, and other outcomes of interest.
To assess whether relationships are causal using a Mendelian randomization approach.
To determine whether differential methylation in offspring cord blood mediate any observed associations.

Methods
ALSPAC is a longitudinal, population-based pregnancy cohort based in Bristol (http://www.alspac.bris.ac.uk.)(9) All antenatal haemoglobin measurements have been abstracted from obstetric records (a median of 3 per woman). Trajectories of change in haemoglobin across pregnancy can be modelled using multilevel models. In addition, there is genome wide data on both mothers and offspring and DNA methylation is available for a subgroup.
A number of potential PhD projects could be undertaken based on this proposal, from the very statistical to the much more applied. The student will be encouraged to develop a work plan based around their interests. Students may choose to combine a classic epidemiologic approach with a focus on one or more of statistical modelling of repeat measures (in Stata and/or MLwiN) using multilevel models, strengthening causal inference using a Mendelian randomization approach, investigating the role of epigenetics in explaining observed associations.