Current or recent malaria infection is associated with elevated inflammation-adjusted ferritin concentrations in pre-school children: a secondary analysis of the BRINDA database

Abstract

Inflammation and infections such as malaria affect micronutrient biomarker concentrations and hence estimates of nutritional status. It is unknown whether correction for C-reactive protein (CRP) and α1-acid glycoprotein (AGP) fully captures the modification in ferritin concentrations during a malaria infection, or whether environmental and sociodemographic factors modify this association. Cross-sectional data from eight surveys in children aged 6-59 months (Cameroon, Cote d'Ivoire, Kenya, Liberia, Malawi, Nigeria and Zambia; n 6653) from the Biomarkers Reflecting Inflammation and Nutritional Determinants of Anaemia (BRINDA) project were pooled. Ferritin was adjusted using the BRINDA adjustment method, with values < 12 μg/l indicating iron deficiency. The association between current or recent malaria infection, detected by microscopy or rapid test kit, and inflammation-adjusted ferritin was estimated using pooled multivariable linear regression. Age, sex, malaria endemicity profile (defined by the Plasmodium falciparum infection prevalence) and malaria diagnostic methods were examined as effect modifiers. Unweighted pooled malaria prevalence was 26·0 % (95 % CI 25·0, 27·1) and unweighted pooled iron deficiency was 41·9 % (95 % CI 40·7, 43·1). Current or recent malaria infection was associated with a 44 % (95 % CI 39·0, 52·0; P < 0·001) increase in inflammation-adjusted ferritin after adjusting for age and study identifier. In children, ferritin increased less with malaria infection as age and malaria endemicity increased. Adjustment for malaria increased the prevalence of iron deficiency, but the effect was small. Additional information would help elucidate the underlying mechanisms of the role of endemicity and age in the association between malaria and ferritin.

Keywords: Ferritin; Inflammation; Malaria infection.

Fig. 1.

Mean inflammation-adjusted ferritin concentration (log scale, µg/l) by age group among children with malaria infection (n 1733) and children not infected (n 4920) among children aged 6–59 months from eight datasets from the BRINDA database in malaria endemic countries in Africa. Inflammation adjustment was done with the BRINDA method⁽¹⁸⁾. The model is also adjusted for the interaction between malaria and endemicity level (model M). Effect estimate of age on inflammation-adjusted ferritin concentration in children with malaria infection (> 2 years v. < 2 years): −15 % (–23 %, −6 %). P for interaction malaria * age: 0·002.

Fig. 2.

Mean inflammation-adjusted ferritin concentration (log scale, µg/l) per endemicity profile in children with malaria infection (n 1733) and children not infected (n 4920) – among children aged 6–59 months from eight datasets from the BRINDA database in malaria endemic countries in Africa. Inflammation adjustment was done with the BRINDA method⁽¹⁸⁾. The model is also adjusted for the interaction between malaria and age (model M). Effect estimate of malaria endemicity profile on inflammation-adjusted ferritin concentration in children with malaria infection (high endemicity v. moderate endemicity): −17 % (–25 %, −8 %). P for interaction malaria * endemicity profile: < 0·001.

Fig. 3.

Inflammation-adjusted ferritin concentration (µg/l, geometric mean and 95 % CI) per CRP decile in children with malaria infection and children not infected in (a) moderate (n 4486) and (b) high (n 2167) endemicity profile among children aged 6–59 months from eight datasets from the BRINDA database in malaria endemic countries in Africa. Inflammation adjustment was done with the BRINDA method⁽¹⁸⁾. The CRP deciles were derived from the entire dataset. The maximum value of each decile is: 1st decile: 0·2 mg/l, 2nd decile: 0·4 mg/l, 3rd decile: 0·7 mg/l, 4th decile: 1·2 mg/l, 5th decile: 2 mg/l, 6th decile: 3·4 mg/l, 7th decile: 5·9 mg/l, 8th decile: 13 mg/l, 9th decile: 24 mg/l, 10th decile: 864 mg/l. The difference between inflammation-adjusted ferritin in children with malaria infection and children without malaria infection is positive and statistically significant (P < 0·05) in (a) at every decile except the 10th decile and in (b) at deciles 6, 8, 9 and 10. Statistical significance was assessed with a linear model that included the main effect of malaria and the interaction between malaria and age. CRP, C-reactive protein.