Differential ferritin interpretation methods that adjust for inflammation yield discrepant iron deficiency prevalence

Abstract

We reassessed the iron deficiency (ID) prevalence in a South African trial that formed part of the International Research on Infant Supplementation study by comparing four methods that account for the high prevalence of acute (28.6%) and chronic (41.8%) inflammation observed in the study. Serum ferritin (SF) was measured as marker of iron status in 192 apparently healthy, 4-13-month-old infants. Alpha-1 glycoprotein and C-reactive protein concentrations were determined to indicate chronic and acute inflammation, respectively. The ID prevalence was obtained by four methods that adjust for inflammation: (1) excluding infants with inflammation; (2) using a higher cut-off (SF < 30 μg L(-1) ); (3) using different cut-offs for infants with vs. without inflammation (SF < 30 μg L(-1) vs. SF < 12 μg L(-1) ); and (4) adjusting SF concentrations with correction factors (CFs) were compared with a reference method (SF < 12 μg L(-1) ) not accounting for inflammation. Using the higher SF cut-off method resulted in the highest ID prevalence (52.1%), followed by using two different cut-offs (31.8%), using CFs (21.9%) and excluding subjects with inflammation (17.6%). The CF method showed the best agreement with the reference method. Disregarding inflammation resulted in a significantly lower ID prevalence (17.2%). ID anaemia (IDA) prevalence ranged from 13.2% to 24.5%, with the lowest prevalence (12.0%) for the reference method. Our analysis highlights the challenge of assessing ID and IDA using only SF as marker of iron status in the presence of inflammation. We demonstrate the importance of measuring inflammation markers to account for their elevating effect on SF.

Keywords: IRIS study; South Africa; acute phase proteins; anaemia; infants; infection; serum ferritin.

Figure 1

Differences in the prevalence of ID, IDA and anaemia not associated with ID for the five methods employed to interpret SF concentrations. ¹Higher cut‐off method; ² AGP ≥ 1 g L⁻¹ and/or CRP > 5 mg L⁻¹; ³Different cut‐offs for SF for subjects with inflammation (30 μg L⁻¹) vs. no inflammation (12 μg L⁻¹). *ID prevalence differed significantly from the reference method (P < 0.005), McNemar test.