Use of Modified Relative Dose‐Response to Examine Sensitivity and Specificity of Retinol Binding Protein as An Indicator of Vitamin A Deficiency among Nepali Children 6–23 Mo of Age

Abstract

A population‐based survey in two districts of Nepal assessed retinol binding protein (RBP) and inflammation indicators among 2347 children 6–23 mo, and the modified relative dose‐response (MRDR) and serum retinol among 151 randomly selected children in a sub‐sample. In previous analyses, vitamin A deficiency(VAD) for RBP was defined as <0.84 μmol/L, calculated by examining the serum retinol‐RBP relationship. MRDR assesses VAD as a qualitative measure of vitamin A liver reserves. MRDR defines VAD as ≥0.060. Retinol isotope dilution is the most sensitive biomarker of vitamin A status to vitamin A liver stores, but is not feasible to assess in cross‐sectional, population‐based surveys. Using MRDR as the standard, we examined RBP sensitivity and specificity to detect VAD among the sub‐sample, and also excluding those with any inflammation. RBP sensitivity was 34.6% and specificity was 75.0%. RBP was, and MRDR was not, influenced by inflammation. Excluding those with inflammation, RBP sensitivity was 23.1% and specificity was 81.2%. RBP showed poor sensitivity to identify VAD both with and without inflammation, while specificity of RBP was good. When feasible, including a sub sample of MRDR is useful in surveys because it provides more information about VAD status in children.