Table 2 Associations between maternal preconception circulating biomarker levels and Child Behaviour Checklist (CBCL) scores (Model 2).

From: Maternal preconception circulating blood biomarker mixtures, child behavioural symptom scores and the potential mediating role of neonatal brain microstructure: the S-PRESTO cohort

Cluster

Biomarker

CBCL outcome

BKMRa (N = 109)

Linear regression (complete data, N = 117)

   

Cluster PIP

Conditional PIP

Estimateb

SE

Beta

SE

P-value

95% CI

8

Flavin monophosphate

Internalizing

0.828

0.624

0.189

0.243

0.177

0.102

0.084

−0.024 to 0.377

8

Riboflavin

Internalizing

0.828

0.376

0.115

0.217

0.182

0.093

0.049

0.001 to 0.364

9

Thiamine

Internalizing

0.768

0.775

0.220

0.231

0.312

0.089

5.0 × 10–4

0.136 to 0.487

9

Thiamine monophosphate

Internalizing

0.768

0.110

0.025

0.101

0.241

0.095

0.011

0.054 to 0.428

9

Pyridoxal phosphate

Internalizing

0.768

0.049

0.004

0.033

0.149

0.095

0.118

−0.038 to 0.336

9

Pyridoxic acid

Internalizing

0.768

0.037

0.003

0.028

0.146

0.090

0.105

−0.030 to 0.323

9

Pyridoxal

Internalizing

0.768

0.030

0.000

0.015

0.096

0.084

0.256

−0.070 to 0.261

  1. BKMR Bayesian kernel machine regression, CBCL Child Behaviour Checklist, CI confidence interval, PIP posterior inclusion probability, Rhat potential scale reduction factor, SE standard error.
  2. aBKMR analysis for each CBCL outcome was performed by simultaneously accounting for 67 biomarkers (11 clusters). This table presents the associations that meet the following criteria: (1) Cluster PIP is greater than 0.75; (2) Conditional PIP for one biomarker in the corresponding cluster is greater than 0.5; (3) MCMC effective sample sizes are greater than 100 for all biomarkers in the cluster; and (4) Rhat (potential scale reduction factor) are smaller than 1.1 for all biomarkers in the cluster. Model 2 adjusted for child sex, age at CBCL assessment, maternal ethnicity, mother’s highest education, household income, maternal age at preconception visit, nulliparity, maternal preconception body-mass index, gestational age at birth, and maternal preconception mental health score.
  3. bEffect estimates from BKMR indicate the difference in the mean outcome when a single exposure is fixed at its 75th percentile as compared to when it is fixed at its 25th percentile when all of the other exposures are fixed at their median value.