(P10-021-24) Assessment of Vitamin A Status in Pregnant Women in South Africa

Objectives: In pregnancy, vitamin A (VA) status markers are influenced by haemodilution, inflammation and adiposity. We aimed to assess VA status of pregnant women ( < 18 weeks gestation) in South Africa, at the beginning of pregnancy (conception) by accounting for possible effects of pregnancy duration, inflammation and adiposity.

Methods: In 250 pregnant women aged 18-39 years and < 18 weeks of gestation (by ultrasonography) living in Johannesburg, South Africa, we determined serum concentrations of retinol, and of retinol-binding protein 4 (RBP4), C-reactive protein (CRP) and α₁-acid glycoprotein (AGP). Using multiple linear regression analysis, we modelled serum concentrations of retinol and RBP4 during early pregnancy and extrapolated conception status, as a function of gestational age (proxy for haemodilution), inflammatory markers and mid-upper arm circumference (MUAC, proxy for adiposity). Various transformations of independent variables were employed to address potential non-linear relationships with dependent variables. Low VA status (retinol or RBP4 < 1.05 μmol/L) was determined in models where MUAC was held at the mean value, while gestational age and inflammatory markers were set to zero.

Results: Cube root and multiple fractional polynomial transformation of independent variables yielded better model fit than log transformation as advocated by the Biomarkers Reflecting Inflammation and Nutritional Determinants of Anemia (BRINDA). With MUAC held at the mean, model-based  prevalence estimates of low VA status at conception were 0% (97.5%CI: 0%-1.5%) and 4% (97.5%CI: 2.0%-8.0%), based on retinol and RBP4, respectively. Both retinol and RBP4 concentrations declined with gestational age independently of inflammation and MUAC. At the start of the second pregnancy trimester, mean serum retinol concentrations approached the cut-off value indicating low VA status.

Conclusions: In conclusion, the proposed adjustment approach can be used can be used as an alternative to the BRINDA method. Furthermore, our analysis indicates that in this population, VA status was probably adequate at conception, but declined substantially with pregnancy progression.

Funding Sources: This project was supported by PepsiCo, the National Research Foundation of South Africa, South African Medical Research Council and University of South Africa’s Research Department.