ABSTRACT
Objectives: To assess the diagnostic performance of serum concentrations of retinol-binding protein (RBP), transthyretin, retinol concentration measured by fluorometry and RBP:transthyretin molar ratio, either alone or in combination, to estimate the prevalence of vitamin A deficiency (serum retinol concentration <0.70 μmol/L measured by high-performance liquid chromatography(HPLC)). Methods: A cross-sectional study was conducted in 15 primary schools in Kibwezi and Makindu districts in Eastern province, Kenya in June 2010 with 375 schoolchildren (6-12 years), 25 randomly selected from each school by lot quality assurance sampling. Results: Complete data were collected for 372 children. Mean serum concentration of retinol (HPLC), RBP and transthyretin were 0.87 (SD 0.19) µmol/L, 0.67 (SD 0.17) µmol/L and 3.0 (SD 0.62) µmol/L. The mean RBP: Transthyretin molar ratio was 0.23. The prevalence of vitamin A deficiency measured with HPLC was 18%. Transthyretin and RBP showed the largest area under the curve (AUCs 0.96 and 0.93, respectively). Logistic regression resulted in a model predicting vitamin A deficiency based on RBP, transthyretin and C-reactive protein (AUC: 0.98) and prevalence depending cutpoints for the linear predictor were calculated. Conclusions: Combination of transthyretin, RBP and C-reactive protein in a linear predictor showed excellent diagnostic performance in assessing vitamin A status, and has great potential to eventually replace serum retinol concentration measured by HPLC as the preferred method to assess the population burden of vitamin A deficiency. Further research is needed to confirm whether this linear predictor yields similar results in different populations and laboratories. Our methodology can be widely applied for other diagnostic aims.