Socioeconomic differences in pregnancy metabolic profiles: evidence from the multi-ethnic Born in Bradford cohort study

Background Socioeconomic position (SEP) is related to many aspects of health. Studies have found differences in both SEP and metabolic profiles between White European (WE) and South Asian (SA) pregnant women. However, socioeconomic differences in metabolic profiles of these groups have not been described. Our aim was to examine and compare associations between a composite measure of SEP that reflects its multidimensional nature and pregnancy metabolic profiles in WE and SA women.

Methods Data were from the Born in Bradford study; a population-based cohort of pregnant women booked for delivery at Bradford Royal Infirmary. Ethnicity was either self-reported or abstracted from medical records (~50% were WE and 50% SA). Nineteen individual-, household-, and area-based indicators of SEP (collected by recruitment questionnaires or linkage to residential address) were combined using latent class analysis separately in WE and SA women to derive ethnicity-specific SEP latent variables. A targeted nuclear magnetic resonance platform measured 148 metabolic traits in mid-pregnancy (~28 weeks gestation) fasting serum samples. Linear regression was used to examine associations between SEP latent sub-groups and each metabolic trait. Models were weighted by latent class probabilities to allow for uncertainty in latent class membership assignment, and adjusted for gestational age to control for gestational age-related differences in metabolic traits. Study methods were pre-specified in a publicly available analysis plan (https://osf.io/xrf8s/).

Results Five and three SEP sub-groups (ranging from lowest to highest SEP) were identified in WE (n=4,496) and SA women (n=4,944), respectively. Socioeconomic differences were found in over 80 metabolic traits, with most associations relating to HDL particles, and indicating trends to lower levels of these traits with lower SEP. For instance, when compared with the highest SEP sub-groups, mean% differences (95%CI) in apolipoprotein-A1 were -1.0% (-1.9 to -0.1) [High-Medium SEP], -2.5% (-3.5 to -1.5) [Medium SEP], -4.1% (-5.1 to -3.2) [Low-Medium SEP], and -5.4% (-6.4 to -4.5) [Lowest SEP] in WE women (P=2.5 x 10^-38), and -2.4% (-3.1 to -1.6) [Medium SEP], and -3.1% (-4.0 to -2.3) [Lowest SEP] in SA women (P=2.0 x 10^-13). Associations with metabolic traits were generally similar in WE and SA women. Results were similar when analyses were restricted to the two largest ethnicity groups (White British and Pakistani women).

Conclusion Our findings suggest widespread socioeconomic differences in metabolic profiles of pregnant WE and SA women and might help inform public health interventions. An interactive web application will be developed to present all association results.