In addition, preliminary Pearson correlations were computed among the variables under study to assess criteria for mediation and potential mediator models.
Model 3 included adjustments for confounders as in model 2, and additional adjustments for potential mediation by gestational hypertension, pre-eclampsia, gestational diabetes mellitus or glycosuria during pregnancy, and birth weight.
In model 3 we adjusted for potential mediation of any association by gestational age at birth and birthweight, and in model 4 for potential mediation by BMI of the child at the time of vascular assessment.
In model 2, we additionally adjusted for potential mediation of associations by offspring 25(OH D measured in childhood.
When adjusted for potential mediation and confounding by adiposity, the association was attenuated, with the incidence increased by 13% per serving/day (5.8%to21%1%; I=79%).
In both cohorts, adjustment for potential mediation by fasting glucose levels reduced the ethnic difference by approximately 50% and statistically the difference was consistent with the null, whereas adjustment for postload glucose levels resulted in very little mediation.
In regression analyses for the association between ethnicity and cord-blood leptin levels, we additionally adjusted for potential mediation by fasting/postload glucose levels and cord-blood insulin levels.
Because the DAN had no significant direct relationships to cognition, these models did not meet the criteria for potential mediation and are therefore not reported in detail; we note, however, that the indirect path from the DAN to FPCN to cognition was significant for all cognitive domains.
In order to test for potential mediation between depression and any effects on immune function by the cortisol:DHEAS ratio or cytokines, linear regressions were run predicting neutrophil function from hip fracture group entered at step 1 in the model, and the cortisol:DHEAS ratio or cytokine variable entered at step 2. Differences in degrees of freedom reflect occasional missing data.
To test for mediation, we added potential mediators to the fully-adjusted models and assessed the magnitude of attenuation of the estimated effect of exposure to maternal depression.
