Effect of iron or multiple micronutrient supplements on the prevalence of anaemia among anaemic young children of a malaria-endemic area: a randomized double-blind trial

Authors


Corresponding Author Hermann Zosé Ouédraogo, Département Biomédical et de Santé publique, Institut de Recherche en Sciences de la Santé (IRSS), 03 BP 7192 Ouagadougou. Tel.: +226 50 33 35 94; Fax: +226 50 36 03 94; E-mail: ouedher68@yahoo.fr

Summary

Objective  To assess the effect of supplementation with iron or multiple micronutrients (MM) on the prevalence of anaemia in a malaria-endemic area.

Methods  A community-based randomized double-blind trial was conducted in rural Burkina Faso, including children aged 6–23 months with haemoglobin (Hb) concentrations of 70–109 g/l who were randomized into an iron group (Fe, n = 96), an iron and zinc group (IZ, n = 100) or an MM group (MM, n = 100), 5 days/week for 6 months. All children were provided with insecticide-treated bednets; those who had a Plasmodium falciparum (PF) positive-smear at baseline and/or at each monthly checking received antimalarial therapy.

Results  The mean (SD) endpoint Hb concentration was higher in the MM group [113.2 (13.6) g/l] than in the IZ group [106.3 (15.6) g/l] and the Fe group [107.1 (12.9) g/l] (P = 0.001). Children in the MM group were more likely to recover from anaemia than those in the Fe group [prevalence rate ratios, PRR (95% confidence interval, CI) = 1.62 (1.22–2.15), P < 0.001]. The IZ group did not differ from the Fe group [PRR (95% CI) = 0.94 (0.65–1.35), P = 0.72]. None of the interactions on the effect of supplementation of baseline age (0.13), or baseline height-for-age z-score (P = 0.33), or incident PF parasitemia (P = 0.99), was significant.

Conclusion  In this malaria-endemic area, in combination with malaria management, the MM supplement was more efficacious than the Fe supplement and the IZ supplement for reducing anaemia. Further investigation into limiting factors and amounts of micronutrients that would be more efficacious for reducing anaemia is recommended.

Abstract

Objectif:  Evaluer l’effet de la supplémentation en fer ou multiples micronutriments (MM) sur la prévalence de l’anémie dans une région endémique pour la malaria.

Méthodes:  Une étude randomisée en double aveugle basée sur la communauté a été menée en zone rurale au Burkina-Faso, chez des enfants âgés de 6 à 23 mois avec des concentrations d’hémoglobine (Hb) de 70 à 109 g/L qui ont été randomisés dans un groupe recevant du fer (Fe, n = 96), un groupe recevant du fer et du zinc (FZ, n = 100) ou un groupe recevant MM (MM, n = 100), 5 jours/semaine pendant 6 mois. Tous les enfants ont reçu des moustiquaires imprégnées d’insecticide, ceux qui avaient un frottis positif pour Plasmodium falciparum au départ et/ou à chaque contrôle mensuel, ont reçu un traitement antimalarique.

Résultats:  La moyenne (±SD) de la concentration d’Hb, utilisée comme critère limite, était plus élevée dans le groupe MM [113,2 (13,6) g/L] que dans le groupe FZ [106,3 (15,6) g/L] et le groupe Fe [107,1 (12,9) g/L] (P = 0,001). Les enfants dans le groupe MM étaient plus susceptibles de se remettre de l’anémie que ceux dans le groupe Fe (PRR (IC95%) = 1,62 (1,22 - 2,15), p < 0,001). Il n’avait pas de différence entre le groupe FZ et le groupe Fe (PRR (IC95%) = 0,94 (0,65 - 1,35), p = 0,72). Il n’y avait aucune interaction significative entre l’effet de la supplémentation et l’âge de départ (0,13), la taille pour l’âge z-score de départ (p = 0,33), ou l’incidence de parasitémie P. falciparum (p = 0,99).

Conclusion:  Dans cette région endémique pour la malaria, en plus de la prise en charge de la malaria, la supplémentation de MM était plus efficace que celle du Fe et celle du FZ pour la réduction de l’anémie. Des investigations supplémentaires sur les facteurs limitant et les quantités de micronutriments qui seraient plus efficaces pour réduire l’anémie sont recommandées.

Abstract

Objetivo:  Evaluar el efecto de la suplementación con hierro o múltiples micronutrientes (MM) sobre la prevalencia de anemia en un área endémica para malaria.

Métodos:  Ensayo aleatorizado, doble ciego, basado en la comunidad, en un área rural de Burkina Faso, incluyendo niños con edades entre los 6 y 23 meses, con concentraciones de hemoglobina (Hb) de 70-109 g/L que fueron aleatorizados para el grupo de hierro (Fe, n = 96), un grupo de hierro y zinc (HZ, n = 100) o un grupo de MM (MM, n = 100), 5 días/semana durante 6 meses. Todos los niños recibieron mosquiteras impregnadas; aquellos que tenían una lámina positiva para Plasmodium falciparum (PF) al comienzo del estudio y/o en cualquiera de los revisiones mensuales recibieron terapia antimalárica.

Resultados:  El resultado de la media (SD) de la concentración Hb era mayor en el grupo MM [113.2 (13.6) g/L] que en el grupo de IZ [106.3 (15.6) g/L] y el grupo de Fe [107.1 (12.9) g/L] (p = 0.001). Los niños en el grupo MM tenían una mayor probabilidad de recuperarse de la anemia que aquellos en el grupo de Fe (PRR (95% CI) = 1.62 (1.22 – 2.15), p < 0.001). El grupo de IZ no difería del grupo Fe (PRR (95% CI) = 0.94 (0.65 – 1.35), p = 0.72). Ninguna de las interacciones era significativa sobre el efecto de la suplementación mirado por edad al momento de comenzar el estudio (0.13), sobre el z-score altura-por-edad al comienzo del estudio (p = 0.33), o sobre una parasitemia por P. falciparum incidente (p = 0.99).

Conclusión:  En esta área endémica para malaria, y en combinación con el manejo de la malaria, la suplementación con MM fue más eficaz que la suplementación con Fe o con HZ para reducir la anemia. Se recomienda una mayor investigación sobre los factores limitantes y las cantidades de micronutrientes que serían más eficaces para reducir la anemia.

Introduction

Anaemia is a widespread public health problem with major consequences for human health as well as social and economic development (WHO & UNICEF 2004). One of the most dramatic health effects of anaemia is the higher risk of child mortality (Brabin et al. 2001). Iron deficiency is the main cause of anaemia in many developing countries. The negative consequences of iron deficiency anaemia on cognitive and physical development of children are also of major concern (MI & UNICEF 2004). Other nutritional deficiencies besides iron, such as vitamin B12, folate and vitamin A, can also cause anaemia, although the magnitude of their contribution is unclear (WHO & UNICEF 2004). Anaemia is an indicator of poor nutrition; but infectious diseases, in particular malaria and helminth infections, are important factors contributing to the high prevalence of anaemia in many populations (Crawley 2004). An integrated approach is thus required to combat this public health problem (WHO & UNICEF 2004).

Iron supplementation is internationally recommended as a strategy for controlling iron deficiency in pregnant women and young children in developing countries (WHO 2001). WHO recommends that where the diet does not include fortified foods, or where prevalence of anaemia in children approximately 1 year of age exceeds 40%, supplements of iron at a dosage of 2 mg/kg of body weight/day should be given to all children between 6 and 23 months of age (WHO 2001). Recently, this policy was questioned, since a study reported that routine supplementation with iron and folic acid in preschool children in a population with high prevalence of malaria resulted in an increased risk of severe illness and death (Sazawal et al. 2006). An Expert Consultation was convened by the WHO to examine this issue in Lyon (France) in June 2006. The conclusions stated that in areas where malaria is endemic, iron deficiency in children aged 6–23 months should be controlled in a context of comprehensive and effective health care, including the provision of insecticide-treated nets and vector control for the prevention of malaria, and prompt recognition and treatment of malaria with effective antimalarial drugs (WHO 2007).

The coexistence of iron deficiency with other micronutrients deficiencies makes combined micronutrient supplementation attractive for the prevention of anaemia. At the behest of UNICEF, the International Research on Infant Supplementation (IRIS I) was implemented to test multiple micronutrient (MM) supplements made of either iron and zinc, or iron, zinc and 13 vitamins and other minerals (IRIS I Study Group 2003). The daily MM supplement was the most efficacious of the treatments tested (weekly MM supplement, daily iron supplement and daily iron and zinc supplement) for improving both infant anaemia and micronutrient status (Allen & Shrimpton 2005). Therefore, WHO and UNICEF were advised to make programme recommendations on the use of daily MM supplements instead of iron supplements for preventing iron deficiency anaemia during infancy (Allen & Shrimpton 2005). Little is known about the effect of MM supplementation on anaemia in malaria-endemic areas. Some micronutrients are likely to interact with the malaria parasite, leading to either synergistic or antagonist effects on malaria morbidity, and consequently on haemoglobin response. Iron supplementation is suspected to worsen some malariometric indices (Shankar 2000; Sazawal et al. 2006). The role of vitamin A and zinc supplementation in malaria morbidity is conflicting, showing beneficial effect (Shankar et al. 1999, 2000), or not (Binka et al. 1995; Müller et al. 2001). Other micronutrient deficiencies showed various effects in different setting with respect to malaria morbidity: exacerbation (vitamin B1), protection (vitamin E) or both exacerbation (riboflavin) and protection (vitamin C) (Shankar 2000). Thus, malaria morbidity might influence the effect of MM supplementation on the haemoglobin (Hb) response and anaemia. A study in Tanzanian children reported the efficacy of a MM supplement combined with malaria management on anaemia (Ekvall et al. 2000). However, the tested MM supplement did not contain some micronutrient such as zinc or riboflavin, and the study did not allow comparison to supplementation with iron alone.

Anaemia is common in Burkina Faso, affecting almost all children (93.6%) in rural areas (Measuredhs 2004). This study was undertaken in this malaria-endemic area, to assess the effect of treatment with iron, iron and zinc or MM supplements, combined with malaria management, on the prevalence of anaemia among children aged 6–23 months with mild-to-moderate anaemia.

Subjects and methods

Study site

In rural Burkina Faso, the under-five children’s death rate reaches 202 per 1000 living births; stunting affects 41.6% and wasting 19.6% (Measuredhs 2004). Vitamin A deficiency (serum retinol <0.70 μmol/l) was found in 85% of children aged 12–36 months and in 64% of their mothers (Zagréet al. 2000). The main activity aimed at controlling vitamin A deficiency is vitamin A supplementation through the National Immunization Days or the National Micronutrient Days. Malaria is endemic with stable transmission and seasonal peaks. The entomological inoculation rate reaches 131 per person and year, with 102 occurring around September (Traore 2003). Among children aged 0–3 years, the parasite prevalence varies from 68% to 83% depending on the season (Traore 2003). The national recommendations for malaria control in children are mainly based on primary prevention through use of insecticide-treated nets and effective treatment with arthemeter-lumefantrine (Burkina Faso 2005). However, due to frequent lack of this drug in health services, routine treatment of most uncomplicated cases is still based on chloroquine, sulfadoxine-pyrimethamine or amodiaquine. Soil-transmitted helminth infections affect 21% of the school-aged population of Burkina Faso, hookworm being involved in 19% (Brooker et al. 2006). The prevalence is 31.2% in children aged 2–14 years at Kongoussi (Ouedraogo et al. 2004), a poor rural district of Burkina Faso with 256 205 inhabitants, 12% of whom were younger than 2 years.

Study design and subjects

This community-based randomized double-blind trial was performed from August 2006 to February 2007, including the highest malaria transmission period. To detect a difference of 25% of anaemia between the iron group and one of the two other groups at the end of the intervention with 90% power and a two-side type error of 5%, we needed to include 85 children in each group (expecting a decrease in anaemia from 100% to 50% in the iron group and from 100% to 25% in the MM group). IRIS I reported 66% and 53% reductions in the prevalence of anaemia in the MM and iron groups, respectively (Smuts et al. 2005). Characteristics of the children, who were older, and the character of the intervention that involved larger iron dosages and malaria prompt recognition and treatment, allowed us to expect at least the same effect in the iron group and a higher effect in the MM group. Indeed, a higher iron dose (in comparison to the IRIS I study) should correspond to more absorbed iron due to enhancers such as ascorbic acid and riboflavin in the MM group. The sample size was raised to 102 by group accounting for 20% loss to follow-up.

Criteria for enrolment were age 6–23 months, Hb concentration range of 70–109 g/l and absence of severe wasting (weight-for-height z-score ≤3). Samples of 17 children were drawn from 30 villages (total 510 children) selected through the ‘probability proportionate to size’ cluster sampling adopted from the expanded programme on immunization survey methods (Lemeshow & Robinson 1985), and invited to attend the village health post at a fixed day for screening. Eligible children were randomized to receive either 15 mg iron (Fe group), 15 mg iron and 10 mg zinc (IZ group) or MM (MM group: 15 mg iron, 10 mg zinc, 375 μg vitamin A, 5 μg vitamin D, 6 mg vitamin E, 0.5 mg vitamin B1, 0.5 mg riboflavin, 6 mg niacin, 150 μg folic acid, 0.5 mg vitamin B6, 0.9 mg vitamin B12, 35 mg vitamin C, 10 μg vitamin K, 50 μg iodine, 0.7 mg copper). Iron was included as ferrous fumarate. Table 1 presents the composition of a daily dosage of the MM supplement. This composition is similar to that of the foodlet used in the IRIS I study (International Research on Infant Supplementation (IRIS I) Study Group 2003), with the exception of the iron and zinc dosages that were 10 mg and 5 mg, respectively. For this individual-based randomization, the mother was asked to pick one number from an urn containing 306 random numbers (102 numbers corresponded to each group).

Table 1.   Composition of the daily dose (18 g) of the multiple micronutrient supplement
NutrientsUnitQuantity
Added micronutrients
 Vitamin Aμg RE375
 Vitamin Cmg35
 Vitamin Dμg5
 Vitamin Emg6
 Vitamin Kμg10
 Vitamin B1mg0.5
 Riboflavinmg0.5
 Vitamin B6mg0.5
 Vitamin B12μg0.9
 Folic acidμg150
 Niacinmg6
 Ironmg15.1
 Zincmg10.2
 Coppermg0.7
 Iodineμg59
From the vehicle
 EnergyKcal97
 Proteing2.4
 Calciummg40.3
 Phosphorusmg48.6
 Potassiummg89.9
 Magnesiummg14.6
 Sodiummg16.0

Micronutrient supplementation

Micronutrient supplements were manufactured by Nutriset (Malaunay, France) as a specifically fortified, plumpynut-like, semi-solid product presented in 90 ml opaque boxes coded A, B and C each lot of boxes contained in white packing labelled A, B and C, respectively. The products had the same consistency, taste, colour and flavour. The codes were kept unknown to the investigators until data analysis was completed. From Monday through Friday, between 6 and 8 a.m., each mother took her child to the village health post. There she received 18 g (i.e. two spoonfuls, by using standards spoons provided by the manufacturer) of the appropriate supplement and fed it to her child under supervision of a local surveyor. Mothers who did not present in the morning were met later at home by the local surveyor, for supplement administration. Each day, in each village, in each group, one box was opened to serve all children of the group; the next box was opened only after all the content in the previous box had been served. An opened box was not used the next day.

Malaria and helminthiasis prevention and treatment

Malaria management consisted of prevention and cases detection and treatment. All mothers received a long-lasting, deltamethrin-impregnated mosquito net, made of 100% polyester (PermaNet® 2.0; Vestergaard Frandsen Disease Control Textiles, Lausanne, Switzerland) and instruction for effective use for children. The local surveyor was asked to help with installing bednets, to regularly check their quality during the study and to replace damaged ones. All children were screened for malaria at baseline, then monthly. Children with a positive smear for malaria either at baseline or a monthly screening were artemether + lumefantrine-treated (Coartem®; Novartis Pharma S.A.S., France) regardless of clinical status. Children with illness (including fever or a history of fever) during the study were referred to the health post for routine treatment. Albendazole was given to children at baseline, as a way to control for soil-transmitted infections (Hall et al. 2008). Children aged 12–23 months received 200 mg albendazole; children aged 6–11 months did not (WHO 2001, 2002).

Data collection and processing

Data collection involved: (1) an interview of the mothers at baseline using a questionnaire, (2) a baseline medical examination of mothers and children, (3) a baseline and monthly malaria detection, (4) recording of the supplement administration.

The questionnaire elicited baseline demographic and socio-economic data, as well as child feeding and caring practices and child morbidity (diarrhoea, fever and cough) in the fortnight preceding the interview. A general practitioner examined mothers for goiter according to the definition and classification of the International Council for the Control of Iodine Deficiency Disorders (Dunn & Van Der Haar 1990), and children for splenomegaly as classified according to Hackett (Gentilini 1993). Anthropometrical measurements were performed on children and mothers by a nutritionist according to WHO recommendations (WHO 1995). Children’s capillary blood was obtained by lab technicians through a finger prick for Hb concentration measurement and blood smear preparation. Hb concentration was measured using a HemoCue® machine (Hemocue HB 201+, Angelholm, Sweden) to the nearest 1 g/l. Blood smears to detect malaria infection at baseline (prevalent malaria) and then monthly (incident malaria) were stained with Giemsa and read in duplicate at the local hospital laboratory. Discordant results underwent a third reading at the Institut de Recherche en Sciences de la Santé in Ouagadougou. The mean parasite density of the two concordant results was considered. Reading was performed as recommended by Trape (1985).

Data were double entered and validated using Epi-info version 6.04d (CDC, Atlanta, GA, USA), then analysed under SPSS 12.0 for Windows (SPSS Inc, Chicago IL, USA). Height-for-age, weight-for-age and weight-for-height z-scores (HAZ, WAZ and WHZ) were computed using the Anthro software (WHO Anthro 2006, Geneva, Switzerland), considering the new WHO reference population. We quoted household assets as follows: housing (wall made with cement = 1, roof made with sheet metal = 1, electricity available = 1, tap water available = 1), transportation (bicycle = 1, moped = 2, motorcycle = 3, car = 4) and domestic equipment (radio = 1, television = 2, refrigerator = 3). These quotes were added to create the household equipment index that varied from 0 to 17. We defined the equipment index as high (9–17), middle (5–8) and low (0–4). These cut-offs were chosen to get the three groups that were the most approximately equal. Anaemia was defined as Hb concentration <110 g/l (WHO 2001). With respect to incident malaria parasitemia, children were regrouped into three categories: no positive smear, one positive smear and two positive smears out of the six monthly screening (no child experienced more than two positive smears).

Ethical considerations

The study received written ethical approval from the ‘Comité d’Ethique pour la Recherche en Santé’ of the Ministry of health of Burkina Faso (report no. 2005-012), and from the ethics board of the ‘Fond pour la recherché Scientifique Médicale, FRSM’, of Belgium. Informed written consent was obtained from caregivers before children’s inclusion in the study. Children with severe wasting detected during the screening were referred to the local hospital for treatment and those with severe anaemia (Hb concentration <70 g/l) underwent a 3-month iron therapy (Fercefol®; Exphar S.A., Brussels, Belgium) free of charge.

Statistical analysis

Normal distribution of quantitative data was visually checked using histogram, box-plots and normality plots, and tested using the Kolmogorov–Smirnov test of normality. Baseline characteristics (including Hb concentration) and incident PF parasitemia were compared between groups by using one-way analysis of variance (anova) to compare means, and chi-squared test to compare proportions.

The endpoint values considered for analysis were the Hb concentration and the recovery from anaemia. Analysis was by intent to treat. The strategy of data analysis was set in two steps. First, univariate analysis was applied, and second, multiple regression analysis was performed to correct for imbalances at baseline and to test for the interaction on the effect of supplementation of incident PF parasitemia.

The paired t-test was used to compare baseline and endpoint Hb concentration within groups. The between-groups comparison of the endpoint Hb concentration was performed by using anova and the comparisons between the iron group and each other groups were planned; a post hoc comparison of the IZ or the MM group to the Fe group was performed by using the Dunnett’s t-test. A multiple linear regression was established including variables that were associated with endpoint Hb concentration, that is to say household equipment index, mother’s age, mother income-generating activity, baseline HAZ, incident PF positive smear and supplementation group. In the final model, only covariates that were significant at the 0.1 level were kept; interactions of the supplementation group with all other covariates were tested.

We compared the proportions of children who recovered from anaemia at the end of intervention using chi-squared tests. Variables that were included in the linear regression were considered for logistic regression analysis as they were associated with the recovery from anaemia. To categorize baseline Hb concentration into two equal categories, the median value of 90 g/l was used as cut-off. Adjusted odd ratio [OR (95% confidence interval, CI)] were computed from the final model and tested with Wald’s chi square. Goodness of fit was assessed by Hosmer and Lemeshow’s test. OR (95% CI) were converted into prevalence rate ratios [PRR (95% CI)] according to Osborn and Cattaruzza (Osborn & Cattaruzza 2002).

Results

Of the 456 children who were screened, the 299 (65.6%) eligible children were randomized into three supplementation groups (97, 101 and 101 children in the Fe, IZ and MM groups, respectively, Figure 1). The numbers between groups differed, as the expected number of 306 eligible children was not reached. One child died in each group. Mean (SD) compliance with the supplementation (%) was 90.4 (12.3), 89.3 (11.5) and 86.3 (13.4) for Fe, IZ and MM groups, respectively. Based on children’s weights, the mean (SD) iron dose (mg/kg/day) was 1.9 (0.3), [2.1 (0.3) for infants aged 6–11 months, and 1.7 (0.3) for children aged 12–23 months at baseline]. Baseline characteristics are shown in Table 2. Although non-significant, there were between-group differences, those of children’s age, fever in the last fortnight, incident PF parasitemia, WAZ and WHZ showing P values <0.20.

Figure 1.

 Trial profile.

Table 2.   Baseline characteristics
 Fe (n = 96)IZ (n = 100)MM (n = 100)P
  1. PF, Plasmodium falciparum; HAZ, height-for-age z-score; WAZ, weight-for-age z-score; WHZ, weight-for-height z-score.

Household
 Low equipment index, %40.631.031.00.21
Mothers
 Age (years), mean (SD)27.4 (6.2)26.9 (6.8)27.1 (6.9)0.86
 >25 years, %55.246.053.00.40
 Uneducated, %81.386.078.00.34
 Unemployed, %63.761.153.90.58
 Body mass index (kg/m2), mean (SD)20.9 (2.0)21.2 (2.2)21.1 (2.6)0.67
 <18.5 kg/m2, %9.47.08.00.83
 Presence of visible goiter, %32.328.031.00.80
Children
 Age (months), mean (SD)12.7 (4.9)13.5 (4.8)14.0 (5.5)0.20
 6–11 months, %51.046.044.00.60
 Female, %46.956.053.00.43
 No longer breastfed, %3.12.02.00.84
 Without complementary feeding, %26.028.024.00.56
 Incompletely vaccinated, %11.510.06.00.39
 Diarrhoea in the last fortnight, %16.718.014.00.74
 Fever in the last fortnight, %21.925.033.00.19
 Cough in the last fortnight, %2.12.05.00.37
 Presence of splenomegaly, %5.25.06.00.95
 PF parasitemia of 1–4999/μl, %29.240.042.00.20
 PF parasitemia ≥5000/μl, %7.38.011.0
 HAZ, mean (SD)−1.11 (1.24)−0.92 (1.12)−1.18 (1.14)0.27
 ≤2, %22.916.022.00.42
 WAZ, mean (SD)−1.58 (1.20)−1.22 (1.05)−1.39 (1.02)0.07
 ≤2, %34.424.027.30.26
 WHZ, mean (SD)−1.37 (1.16)−1.02 (1.03)−1.08 (1.09)0.06
 ≤2, %29.219.017.20.09

There were 22/96 (22.9%), 26/100 (26.0%) and 27/100 (27.0%) children who had one or two smears positive for PF parasitemia in the Fe, IZ and MM groups, respectively (P = 0.79). Table 3 presents Hb concentration at both baseline and end of intervention. Mean (SD) baseline Hb concentration was 90.8 (9.1) g/l, 89.6 (10.7) g/l and 90.4 (11.4) g/l in the Fe, IZ and MM groups, respectively (P = 0.71). It increased to 107.1 (12.9) g/l in the Fe group (P < 0.001), 106.3 (15.6) g/l in the IZ group (P < 0.001) and 113.2 (13.6) g/l in the MM group (P < 0.001). This corresponded to an increment of 16.3 (13.6) g/l [15.4 (13.7) for children aged 6–11 months and 17.7 (13.5) for children aged 12–23 months, P = 0.55], 16.6 (15.0) g/l [16.9 (14.7) for children aged 6–11 months and 16.4 (15.3) for children aged 12–23 months, P = 0.88], and 22.8 (14.6) g/l [22.0 (13.8) for children aged 6–11 months and 23.4 (15.3) for children aged 12–23 months, P = 0.65] in the Fe, IZ and MM groups, respectively.

Table 3.   Baseline and endpoint values and increment in haemoglobin concentration (g/l)
 Supplementation groups
Fe (n = 96)IZ (n = 100)MM (n = 100)P
Mean (SD)Mean (SD)Difference (95% CI)PMean (SD)Difference (95% CI)P
  1. †ANOVA for between-group comparison.

  2. ‡Dunnett’s t-test for post hoc multiple comparisons, IZ and MM were compared to Fe.

  3. §Paired t-test for within group comparison.

Baseline haemoglobin concentration90.8 (9.1)89.6 (10.7)−1.2 (−4.5 to 2.1)90.4 (11.4)−0.4 (−3.7 to 2.9)0.71
Endpoint haemoglobin concentration107.1 (12.9)106.3 (15.6)−0.8 (−5.3 to 3.6)0.88113.2 (13.6)6.1 (1.6–10.6)0.0050.001
Increment in haemoglobin concentration16.3 (13.6)16.6 (15.0)0.4 (−3.2 to 5.0)0.9822.8 (14.6)6.5 (2.0–11.1)0.0030.002
P§<0.001<0.001  <0.001   

The between-group difference in mean (SD) endpoint Hb concentration was significant (P = 0.001). The difference (95% CI) of endpoint Hb concentration was 6.1 (1.6–10.6) g/l between the MM and Fe groups (P = 0.005), and −0.8 (−5.3 to 3.6) between the IZ and Fe groups (P = 0.88). In multiple linear regression, the coefficients of regression were 6.5 (2.7–10.2) g/l (P < 0.001) and −0.6 (−4.4 to 3.1) g/l (P = 0.74) for the MM and IZ groups, respectively, the iron group being the reference category (Table 4). The interaction terms of either baseline age (P = 0.58), baseline HAZ (P = 0.07), or incident PF parasitemia (P = 0.54), and supplementation were not significant.

Table 4.   Coefficients of regression of the endpoint haemoglobin concentration
 b (g/l)95% CI (g/l)P
  1. Results of multiple linear regression (n = 296).

Supplementation groups
 Multiple micronutrient compared to iron6.52.7–10.2<0.001
 Iron and zinc compared to iron−0.6−4.4 to 3.10.74
Baseline haemoglobin concentration (g/l)0.50.3–0.6<0.001
Baseline HAZ1.1−0.2 to 2.40.10
Incident PF parasitemia
 One compared to no positive smear−0.5−4.2 to 3.20.78
 Two compared to no positive smear−7.3−15.3 to 0.80.08

There were 39/96 (40.6%), 39/100 (39.0%) and 62/100 (62.0%) children who recovered from anaemia in the Fe, IZ and MM groups, respectively (P = 0.001). Table 5 presents the final model of logistic regression. Children in the MM group were more likely to recover from anaemia than those in the Fe group: PRR (95% CI) = 1.62 (1.22–2.15), P < 0.001. The IZ group did not differ from the Fe group: PRR (95% CI) = 0.94 (0.65–1.35), P = 0.72. None of the interactions on the effect of supplementation of baseline age (0.13), or baseline HAZ (P = 0.33), or incident PF parasitemia (P = 0.99) was significant.

Table 5.   Unadjusted and adjusted prevalence rate ratio (95% CI) for recovery from anaemia after the 6-month intervention
 Unadjusted†Adjusted‡
n% of recoveryPRR (95% CI)PPRR (95% CI)P
  1. PF, Plasmodium falciparum; HAZ, height-for-age z-score.

  2. †Results of bivariate analysis.

  3. ‡Results of logistic regression (odds-ratios were converted into PRR).

Supplementation groups
 Multiple micronutrient10062.01.53 (1.16–2.01) 0.0011.62 (1.22–2.15)<0.001
 Iron and zinc10039.00.96 (0.65–1.40)0.94 (0.65–1.35)0.72
 Iron9640.611
Incident PF parasitemia
 No positive smear22148.42.66 (1.00–7.07) 0.143.36 (1.22–9.23)0.019
 One positive smear6448.42.66 (0.95–7.45)3.41 (1.20–9.66)0.021
 Two positive smears1118.211
Baseline HAZ
 ≥223650.41.44 (1.03–2.01)0.0331.54 (1.09–2.19)0.015
 ≤26035.011
Baseline haemoglobin concentration
 ≥90 g/l15656.41.52 (1.19–1.95)0.001 <0.001
 <90 g/l14037.111.60 (1.24–2.07)

Discussion

Haemoglobin concentration increased and anaemia prevalence decreased in the three supplementation groups. The increase of Hb concentration and reduction of anaemia prevalence in the MM group were superior to those in the Fe group. This finding was expected, as a result of the additional effects of the other components of the MM that can enhance hematopoiesis, indirectly by favouring iron absorption (vitamin C, riboflavin) or metabolism (cooper, vitamin A, vitamin B12, riboflavin, vitamin C) and/or independently of iron (Allen 1998; Powers 1998; Schultink & Gross 1998; Fishman et al. 2000; Semba & Bloem 2002).

In this study, the increase of Hb concentration reached 22.8 g/l in the MM group; it was 11 g/l in the IRIS study that used a MM with the same vitamins and minerals composition (Smuts et al. 2005). This is due to differences in the study population (older children and lower baseline Hb concentration in the present study) and in the intervention (larger dosage of iron, anthelminthic treatment, and malaria prevention and treatment in the present study). Indeed, other studies reported that anthelminthic treatment and malaria management are associated with a better response of Hb concentration or with anaemia reduction. In the study by Stoltzfus et al., mebendazole treatment reduced anaemia in children aged <24 months with light infection (Stoltzfus et al. 2004). Korenromp et al. reported that malaria prevention through insecticide-treated bednets, antimalarial chemoprophylaxis and insecticide residual spraying led to the increase of Hb concentration by, an average, 7.6 g/l (95% CI: 6.1–9.1) in African children, and the impact on anaemia was larger in sites with a higher baseline parasite prevalence (Korenromp et al. 2004). The study by Ekvall et al. demonstrated that MM supplementation combined with malaria treatment synergistically contributed to the increase of Hb concentration by 22 g/l (95% CI: 13–30) compared with 7 g/l (95% CI: 3–10) without combined malaria treatment (Ekvall et al. 2000).

The interaction of incident PF parasitemia with supplementation was not significant. This could be explained either by absence of a significant additional micronutrient influence on incident PF parasitemia (the between-group difference was not significant, P = 0.79), or the active and precocious cases finding and treatment we performed.

The actual iron dosage that was calculated a posteriori appeared close to the recommended 2 mg/kg/day for children aged 6–23 months with mild-to-moderate anaemia; however, it appeared below this recommended amount for children aged 12–23 months. This is one factor that may limit the effect of the intervention on Hb concentration. The iron dosage was chosen to be about 1.5 times the recommended dietary allowance (RDA), instead of 1 RDA used in the IRIS study (International Research on Infant Supplementation (IRIS I) Study Group 2003), to take into account the anaemic status of the children. It is likely that higher iron dosage would have more improved the rates of recovery from anaemia. Indeed, a study by Desai et al. (2003) of children aged 2–36 months with Hb concentrations of 70–110 g/l using bednets and receiving antimalarial drugs and iron supplements (3–6 mg/kg body weight/day) for 12 weeks showed 75% of recovery from anaemia. The higher rate of recovery from anaemia may be explained by the iron dosages that were 1–4 mg/kg body weight/day (i.e. about 8–32 mg/day) higher than those of the present study. A second reason that may limit the efficacy of the supplement is the presence of calcium that was brought by the food vehicle. This may contribute to reduce iron bioavailability (Kamp et al. 2003).

There were some non-significant group differences with respect to baseline characteristics. This may be due to the randomization procedure. Picking a card from an urn is not the ideal although practical. However, the multiple regressions controlled for this baseline imbalance.

The IZ group was not different of the iron group with respect to the increase of Hb concentration and the reduction of anaemia prevalence. This suggests that the iron–zinc absorption interaction described in other studies did not occur in this study (Dijkhuizen et al. 2001; Lind et al. 2003; Wieringa et al. 2007), perhaps because the vehicle used in this study was food-like; indeed, there is no interaction between iron and zinc when they are given in a meal (Lonnerdal 1998). Although this trial focused on micronutrients, the use of a food-like vehicle provided children with substantial energy. Indeed, the daily dosage provided 97 kcal representing 16%, 14% and 11% of the recommended daily energy requirements for children aged 6–8 months, 9–11 months and 12–23 months, respectively (615 kcal, 686 kcal and 894 kcal for respective age categories) (Dewey & Brown 2003). This may be helpful in this population with high prevalence of energy malnutrition.

In this study, the use of insecticide-treated bednets did not fully prevent malaria incidence, as in a previous insecticide-treated bednet trial in Burkina Faso (Müller et al. 2006). For ethical reasons, our study did not allow comparison of malaria incidence in the supplementation groups to that in a group without provision of iron; but we can range the observed malaria frequency below the high burden of malaria reported by observational studies in rural Burkina Faso (Müller et al. 2003; Traore 2003; Stich et al. 2006). For financial reasons, we did not use the polymerase chain reaction-based malaria detection that is more sensitive and specific (WHO 2000), and that would have improved the detection of malaria parasitemia and therefore the effect of the intervention on anaemia.

In conclusion, in this malaria-endemic area, the MM supplement was more efficacious than iron supplement for reducing anaemia, when combined with malaria management. The used MM regimen, integrated with malaria management and anthelminthic treatment in this intervention remained insufficient to totally control anaemia. Further investigation into limiting factors and amounts of micronutrients that would be more efficacious for reducing anaemia is recommended.

Acknowledgements

The study was supported by a grant from the ‘Fonds pour la Recherche Scientifique et Médicale’ (FRSM) of Belgium (convention no. 3.4534.03), and by the David and Alice Van Buuren Foundation.

Ancillary