Comparison of plasma ferritin concentration with the ratio of plasma transferrin receptor to ferritin in estimating body iron stores: results of 4 intervention trials.

BACKGROUND Efforts to develop global programs for the control of iron deficiency require simple, low-cost, and accurate indicators of iron status. OBJECTIVE We aimed to compare estimates of body iron (BI) stores, as calculated from either plasma ferritin concentration alone (BI-ferritin) or the ratio of plasma transferrin receptor (TfR) to ferritin (BI-TfR/ferritin). DESIGN Data were analyzed from 4 previously completed, randomized intervention trials that enrolled infants, schoolchildren, or pregnant women (total n = 1189, after excluding subjects with elevated C-reactive protein). RESULTS The correlation coefficients between BI-ferritin and BI-TfR/ferritin were >0.95 for all studies. The kappa index ranged from 0.5 to 1.0. All of the sensitivities of BI-ferritin for identifying persons with low iron stores (defined as BI-TfR/ferritin < 0 mg/kg body wt) were >0.90. All of the specificities were >0.90 except the study of pregnant women (specificity = 0.66). The effect sizes of iron intervention trials were significantly greater for change in iron reserves estimated by BI-TfR/ferritin than by BI-ferritin in 2 studies with larger effect sizes (1.11 compared with 1.00 and 1.56 compared with 1.44, respectively; P < 0.05) and 1 study with medium effect size (0.70 compared with 0.57; P < 0.05). However, there were no significant differences between estimates of these effect sizes for 1 study with a medium effect size and 1 study with a smaller effect size (0.78 compared with 0.83 and 0.37 compared with 0.35, respectively; P > 0.2). CONCLUSION Plasma ferritin concentration alone provides a good approximation of total BI reserves, as estimated by BI-TfR/ferritin, on the basis of high correlation, sensitivity, and specificity among nonpregnant persons with unelevated C-reactive protein.