(P14-118-20) Comparison of Serum Retinol and Retinol-Binding Protein with the Modified Relative Dose-Response Test in Estimating Vitamin A Deficiency Among Ghanaian Women and Children

Objectives: To evaluate the sensitivity and specificity of serum retinol (SR) and retinol-binding protein (RBP) in determining vitamin A deficiency (VAD) using the modified relative dose-response (MRDR) test as the reference.

Methods: Subjects included a subset of women and children participating in the Ghana Micronutrient Survey 2017. VAD was determined by the following cut-offs: SR or RBP < 0.7 mmol/L; MRDR ratio of 3,4-didehydroretinol to SR ³0.060. Sensitivity, specificity and the area-under-the-receiver operating characteristic (ROC) curve were calculated for unadjusted and inflammation-adjusted VAD cut-offs (based on C-reactive protein (CRP) and a1-acid-glycoprotein (AGP) for SR and RBP using the MRDR test as the reference.

Results: In 167 children and 178 women, inflammation (elevated CRP and/or elevated AGP) was present in 41% and 16%, respectively. Prevalence of VAD ranged, depending on the indicator used, from 7% (MRDR) to 40% (unadjusted SR) in children and 1% (RBP) to 4% (SR and MRDR) in women. Among children, sensitivity and specificity of unadjusted and adjusted SR and RBP were highly variable among the children. Highest sensitivity was achieved by unadjusted SR (80% of children with VAD correctly identified), while highest specificity was achieved by adjusted RBP (86% of children without VAD correctly identified). The best predictor of VAD in children compared with MRDR was adjusted SR, with a sensitivity of 78%, specificity of 73%, and an area under the ROC curve of 0.76. Among women, specificity was 97% for unadjusted and adjusted SR with an area under the ROC curve of 0.48; additional values could not be calculated due to lack of VAD cases.

Conclusions: SR and RBP were only moderately sensitive and specific for identifying VAD in children in Ghana. Low specificity—falsely identifying VAD—is especially problematic when populations are covered by one or more vitamin A interventions. Overlapping sources of preformed vitamin A can lead to hypervitaminosis, which may affect bone metabolism and growth. More accurate measures should be used in conjunction with SR and RBP for evaluating VAD especially in vulnerable populations.

Funding Sources: UNICEF, Canada’s Ministry of Foreign Affairs, Trade and Development.