Inflammation Adjustment by Two Methods Decreases the Estimated Prevalence of Zinc Deficiency in Malawi

Abstract

Serum zinc concentration (SZC) is used widely to assess population-level zinc status. Its concentration decreases during inflammatory responses, which can affect the interpretation of the results. This study aimed to re-estimate the prevalence of zinc deficiency in Malawi based on the 2015-2016 Malawi Micronutrient Survey (MNS) data, by adjusting SZC measures with markers of inflammation. SZC and inflammation data from 2760 participants were analysed. Adjustments were made using: (1) The Internal Correction Factor (ICF) method which used geometric means, and (2) The Biomarkers Reflecting Inflammation and Nutritional Determinants of Anemia (BRINDA) method, which used linear regression. Mean SZC values increased significantly when adjustments were made by either ICF or BRINDA (p < 0.001). The national prevalence of zinc deficiency decreased from 62% to 59%, after ICF adjustment, and to 52% after BRINDA adjustment. ICF and BRINDA values of SZC were highly correlated (p < 0.001, r = 0.99), but a Bland-Altman plot showed a lack of agreement between the two methods (bias of 2.07 µg/dL). There was no association between the adjusted SZC and stunting, which is a proxy indicator for zinc deficiency. Inflammation adjustment of SZC, using ICF or BRINDA, produces lower estimates of zinc deficiency prevalence, but the lack of agreement between the adjustment methods warrants further research. Furthermore, the lack of association between SZC and stunting highlights the need to explore other biomarkers and proxies of population zinc assessment. This study demonstrates the importance of considering inflammatory confounders when reporting SZC, to ensure accuracy and to support policy decision making.

Keywords: C-reactive protein (CRP); alpha 1-acid glycoprotein (AGP); biomarkers; children; inflammation; women of reproductive age; zinc.

Figure 1

Density distribution of SZC values (a) before adjustment, (b) after adjusting with the ICF method, and (c) after adjusting with BRINDA method. Dashed lines correspond to mean values.

Figure 2

Method comparisons between ICF and BRINDA approaches, presented as a linear correlation in the left panel and Bland–Altman plot in the right panel.