Within-individual differences in plasma ferritin, retinol-binding protein, and zinc concentrations in relation to inflammation observed during a short-term longitudinal study are similar to between-individual differences observed cross-sectionally

BACKGROUND: Cross-sectional (CS) surveys indicate that individuals with acute inflammation have higher plasma ferritin (pF), and lower retinol-binding protein (RBP) and zinc (pZn) concentrations than those without. In populations with a high burden of infection, correction factors (CFs) or regression corrections (RCs) are applied to biomarkers to estimate the prevalence of micronutrient (MN) deficiencies adjusted for inflammation. This assumes that individuals with and without inflammation have the same nutritional status, which may not be the case. OBJECTIVES: The aim of this study was to investigate relations between short-term, longitudinal within-individual changes in acute phase proteins (C-reactive protein [CRP], α-1-acid glycoprotein [AGP]) and biomarkers of MN status (pF, soluble transferrin receptor [sTfR], RBP, and pZn), and compare them to CS differences. METHODS: Two blood samples were obtained 21 d apart from 451 asymptomatic Burkinabe children aged 6-23 mo. To calculate CFs, inflammation was defined as CRP >5 mg/L or AGP >1 g/L, or both. The RC approach adjusted MN biomarkers to a presumably healthy reference point within the study population (10th percentile CRP or AGP concentration). CS CFs and RCs were estimated from a naive regression model, treating observations from the same children as independent. Longitudinal CFs and RCs, to estimate effects of within-individual changes in CRP and/or AGP, were estimated from general linear models, accounting for repeated measures. RESULTS: In CS models, geometric mean pF and sTfR concentrations were 8-340% greater, and RBP and pZn 2-18% lower, in children with inflammation than those without. Except for sTfR, biomarker concentrations differed in the same direction and by similar magnitude within individuals whose inflammation status changed during the observation period. Although geometric mean MN concentrations differed significantly when adjusted with CS compared with longitudinal models, the estimated prevalence of MN deficiencies in CS and longitudinally adjusted models was similar. CONCLUSIONS: The CF and RC approaches to adjust MN biomarkers for inflammation between individuals in CS surveys are valid approaches for data collection and programmatic decisions in comparable populations. This study was registered at clinicaltrials.gov as NCT00944853.