Abstract
BACKGROUND: Identifying modifiable risk factors for congenital heart disease (CHD) is important for prevention. Previous studies have reported associations between 25-hydroxyvitamin D (25(OH)D) in pregnancy and CHD in the offspring. However, these studies contain important methodological limitations. We aimed to investigate 25(OH)D levels during pregnancy in relation to offspring risk of CHD using both multivariable regression (MVR) analysis and Mendelian randomization (MR) analysis. METHODS: Data from the ALSPAC (Avon Longitudinal Study of Parents and Children), BiB (Born in Bradford), and MoBa (Norwegian Mother, Father and Child Cohort) studies (N=8722 (77 cases) for MVR analysis, N=74 953 (646 cases) for MR) were used. MVR analysis was adjusted for offspring sex, maternal age, education, body mass index, smoking, alcohol consumption, and parity. One sample MR was performed with weighted genetic risk score for 25(OH)D based on published genome-wide significant genetic variants. Sensitivity analyses explored the relevance and validity of the genetic risk score. Results were pooled across the 3 cohorts in a meta-analysis with random effects and consistency between results from the MVR and MR was examined. RESULTS: Pooled results from the adjusted MVR suggested higher maternal pregnancy 25(OH)D associated with lower CHD risk, though the CIs were wide and included the null (odds ratio [OR], 0.79 [95% CI, 0.53, 1.06] per 1 SD higher 25(OH)D). By contrast precise MR results did not support a causal relationship (OR, 0.99 [95% CI, 0.91, 1.07] per 1 SD higher 25(OH)D genetic risk score). CONCLUSIONS: We did not find robust evidence supporting an effect of maternal pregnancy 25(OH)D levels on offspring CHD.