Oral iron supplementation for preventing or treating anaemia among children in malaria-endemic areas

  • Review
  • Intervention

Authors


Abstract

Background

Iron-deficiency anaemia is common during childhood. Iron supplementation has been claimed to increase the risk of malaria.

Objectives

To assess the effect of iron on malaria and deaths.

Search strategy

We searched The Cochrane Library (2009, issue 1); MEDLINE; EMBASE; LILACS and metaRegister of Controlled Trials, all up to March 2009. We scanned references of included trials.

Selection criteria

Individually and cluster-randomized controlled trials conducted in hypoendemic to holoendemic malaria regions and including children < 18 years. We included trials comparing orally administered iron with or without folic acid vs. placebo or no treatment. Iron fortification was excluded. Antimalarials and/or antiparasitics could be administered to either group. Additional micronutrients could only be administered equally to both groups.

Data collection and analysis

The primary outcomes were malaria-related events and deaths. Secondary outcomes included haemoglobin, anaemia, other infections, growth, hospitalizations, and clinic visits. We assessed risk of bias using domain-based evaluation. Two authors independently selected studies and extracted data. We contacted authors for missing data. We assessed heterogeneity. We performed fixed-effect meta-analysis and presented random-effects results when heterogeneity was present. We present pooled risk ratios (RR) with 95% confidence intervals (CIs). We used adjusted analyses for cluster-randomized trials.

Main results

Sixty-eight trials (42,981 children) fulfilled the inclusion criteria. Iron supplementation did not increase the risk of clinical malaria (RR 1.00, 95% CI 0.88 to 1.13; 22,724 children, 14 trials, random-effects model). The risk was similar among children who were non-anaemic at baseline (RR 0.96, 95% CI 0.85 to 1.09). An increased risk of malaria with iron was observed in trials that did not provide malaria surveillance and treatment. The risk of malaria parasitaemia was higher with iron (RR 1.13, 95% CI 1.01 to 1.26), but there was no difference in adequately concealed trials. Iron + antimalarial was protective for malaria (four trials). Iron did not increase the risk of parasitological failure when given during malaria (three trials). There was no increased risk of death across all trials comparing iron versus placebo (RR 1.11, 95% CI 0.91 to 1.36; 21,272 children, 12 trials). Iron supplementation increased haemoglobin, with significant heterogeneity, and malaria endemicity did not affect this effect. Growth and other infections were mostly not affected by iron supplementation.

Authors' conclusions

Iron does not increase the risk of clinical malaria or death, when regular malaria surveillance and treatment services are provided. There is no need to screen for anaemia prior to iron supplementation.

Plain language summary

Iron supplements for children living in malaria-endemic countries

Children commonly develop anaemia (low haemoglobin) after birth. Anaemia is associated with several ill-effects, including hindering motor development and learning skills, and it may have an adverse effect on immunity. Babies and children are therefore commonly given iron supplements to prevent or treat anaemia. In countries where malaria is prevalent, it has been suggested that iron supplementation increases the risk of malaria and deaths. The high dose of iron which is given as medicine may result in free iron circulating in the blood and available to the malaria parasite, which promotes its growth. We therefore aimed to assess the effects of iron administered to children living in countries where malaria is prevalent. We included only randomized controlled trials that compared iron given orally as a medicinal product (and not as food or drink fortification) with placebo or no treatment.

Iron did not increase the risk of malaria disease, indicated by fever and presence of parasites in the blood. The presence of parasites in the blood was slightly higher with iron overall, but not in trials with adequate randomization methods. There was no increased risk of death among iron-treated children. Although more than 70 trials were identified for this review, malaria-related outcomes and deaths were reported in only 16 and 11 trials, respectively. Iron supplementation increased haemoglobin by about 1 g/dL in areas where malaria is highly prevalent. At the end of follow up, which varied between two weeks and six months after the end of iron supplementation, the gain was smaller but still present at 0.4 g/dL. Iron did not increase the risk of respiratory infections, but episodes of diarrhoea were more frequent with iron when it was administered with zinc. Children given iron visited medical clinics less than children given placebo, but the rate of hospitalization was similar. Weight and height at the end of treatment were similar. Iron did not adversely affect rates of cure when given together with antimalarial treatment in three trials that examined this issue.

Our conclusions are that iron supplementation does not adversely affect children living in malaria-endemic areas. The evidence shown in our review is limited by the lack of trials examining the relevant outcomes and the limited information allowing us to analyse factors that can affect our results, such as the children's baseline level of haemoglobin. Based on our review, routine iron supplementation should not be withheld from children living in countries where malaria is prevalent.

Background

Description of the condition

Childhood anaemia

Childhood anaemia is perhaps one of the most widespread and important public health problems in sub-Saharan Africa (WHO 2008). Anaemia is defined as a deficiency of red blood cells that can lead to lack of oxygen-carrying ability. The deficiency occurs either through reduced production or an increased loss of red blood cells. Symptoms of anaemia are usually not specific and include tiredness, shortness of breath, dizziness, and palpitations (awareness of heartbeat). Diagnosis is by examining the level of haemoglobin (the protein in the red blood cells responsible for carrying oxygen). Haemoglobin levels defining anaemia in children vary according to the age group: between 6 and 59 months the cut-off value is 11 g/dL, for children 5 to 11 years 11.5 g/dL, for children 12 to 14 years and non-pregnant women older than 15 years 12 g/dL, and for adolescent men older than 15 years it is 13 g/dL (WHO 2001).

The causes of anaemia in developing countries are numerous and often multifactorial, and include micronutrient deficiencies (iron, vitamin A, folate, etc.), infectious diseases (malaria, HIV, intestinal helminths) and haemoglobinopathies (WHO 2001).

Iron and iron deficiency

Iron is an important mineral that the body needs to produce haemoglobin. Iron is also a component of many enzymes essential for proper cell development and cell growth of the brain, muscle, and the immune system (Beard 2001). Iron is a component of the peroxidase and nitrous oxide-generating enzymes that participate in the immune response to infections and is probably involved in the regulation of production and action of cytokines (mediators of immune function released during early stages of infection). A relatively large amount of iron is required for red blood cell production (erythropoiesis) in the first few months after birth. This is usually derived from iron stored by the fetus in the last months of pregnancy. However, by the age of four to six months of life, these stores become marginal or depleted. A child whose diet does not provide enough iron risks developing iron-deficiency anaemia. Infants with a low total body iron at birth are particularly prone to iron deficiency. This is often exacerbated by the early introduction of cereal-based weaning foods from which iron absorption can be as low as 5% (FAO/WHO 2005). Iron deficiency may be worsened by chronic blood loss from the intestines as a result of intestinal parasitic infections (Stoltzfus 1997). Iron-deficiency anaemia is characterized by pallor, fatigue, and weakness. Loss of appetite, strange food cravings like eating dirt (pica), hair loss, and light-headedness among others can also occur. Because iron-deficiency anaemia tends to develop slowly, adaptation occurs and the disease could go unrecognized for long periods. The diagnosis of iron-deficiency anaemia will be suggested by these features and by blood tests indicating low haemoglobin, low ferritin, and low iron levels. In developing countries interpretation of these and other biochemical tests is limited by the confounding effects of infection, inflammation, and malnutrition (Nyakeriga 2004; Zimmermann 2005). Treatment involves iron supplementation and dietary modifications such as increasing the amount of iron-rich foods.

Iron deficiency is a common nutrient deficiency that affects approximately two billion people worldwide, resulting in over 500 million cases of anaemia (WHO 2004). Globally, the most significant contributor to the onset of anaemia is iron deficiency (WHO 2008). In sub-Saharan Africa, the prevalence of iron-deficiency anaemia is estimated around 60% (WHO 2004; WHO 2008), with 40% to 50% of children under five years in developing countries being iron deficient (UNICEF 1998). Community prevalence figures in a single study in Kenya in children less than 15 years showed that up to 80% of the children had anaemia by 11 months of age (Bloland 1999).

Based on estimates of iron-deficiency anaemia as a risk factor for death, iron deficiency has been estimated to cause 726,000 deaths in the perinatal and childhood periods globally, with the greatest toll in South-East Asia and Africa (WHO 2004; FAO/WHO 2005). Experimental and observational studies have linked iron deficiency with several adverse consequences of child development, including impairments in cognitive functions and motor development (Pollitt 1993; Grantham-McGregor 2001; Gewa 2008), growth (Lawless 1994), immune function, and increased risk of infection (Berger 2000; Beard 2001). The relative risk for mental retardation associated with a 1 g/dL increase in population mean haemoglobin has been estimated at 0.78 (95% CI 0.70 to 0.86) (WHO 2004). These studies have been criticized for their ability to adjust for confounders and because they cannot establish causality (Oppenheimer 2001). Conversely, iron is required by many pathogens for their survival and pathogenesis (killing ability) (Beard 2001). Removal of free, circulating iron seems to be an important part of the host (human) response to infection. Thus, a theory that iron deficiency may be an important defence mechanism led to the term "nutritional immunity" (Kochan 1973).

Malaria and anaemia

Malaria is a leading cause of morbidity and mortality in children in sub-Saharan Africa (Breman 2001; WHO malaria 2008). Most infections are caused by the most virulent parasite species, Plasmodium falciparum ( WHO malaria 2008), which is transmitted to humans by the bite of an infected female anopheles mosquito. Trends and general patterns of malaria transmission vary greatly geographically, and children are vulnerable to malaria from the age of approximately three months or earlier when immunity acquired from the mother starts waning.

Malaria causes anaemia through destruction of red blood cells (haemolysis), increased clearance of infected and uninfected red blood cells by the spleen, and cytokine-induced dyserythropoiesis (abnormal formation of red blood cells) (Menendez 2000; Ekvall 2003). A single overwhelming episode of malaria, repeated episodes due to reinfection or failure to clear parasitaemia adequately as a result of inadequate treatment (no treatment, antimalarial drug resistance, or poor compliance) may result in life-threatening anaemia and death (Crawley 2004). Studies show that in areas of intense transmission, most cases of severe malarial anaemia, blood transfusion, and death occurred in infants (Slutsker 1994; Schellenberg 1999; Kahigwa 2002) and in children less than five years of age (Newton 1997; Biemba 2000) with case-fatality rates in hospitals between 8% and 18% (Slutsker 1994; Biemba 2000). Whether malaria infection contributes to iron deficiency is unclear. The observation of an increased haematologic recovery when iron has been administered after a malaria episode (Bojang 1997) suggests that malaria infection has a role in iron deficiency. However, bone marrow studies in Gambian children with acute attacks of malaria showed adequate stores of iron (Abdalla 1990). Placental malaria is associated with fetal anaemia and low birthweight (Desai 2007).

The sickle cell trait and other haemoglobinopathies are considered protective of malaria, due to reduced avidity of dysmorphic, microcytic erythrocytes to the malaria parasite (Ifediba 1985; Luzzi 1990). Whether iron-deficiency anaemia is protective is debatable. The erythrocytic form of the malaria parasite requires free iron (lacking in an iron-deficient individual). One observational study investigated the association between iron status and malaria and concluded that iron deficiency was associated with a small but significant degree of protection from episodes of clinical malaria in a cohort of young children living on the Kenyan coast (Nyakeriga 2004).

Description of the intervention

While observational studies associate iron-deficiency with several adverse consequences among children, interventional studies assessing the effect of iron supplementation have generally not shown significant improvement in these outcomes. Systematic reviews of randomized controlled trials on iron supplementation in children demonstrated that iron supplementation improved mental development and intelligence scores, but not motor development or growth (Ramakrishnan 2004; Sachdev 2005; Sachdev 2006). A comprehensive review recently summarized the health benefits and risks of iron supplementation in early childhood (Iannotti 2006). Anaemia and iron status indicators were most commonly improved, developmental outcomes were variable, a positive effect on weight but not height was reported, and effects on infectious disease morbidity and mortality were mixed. It should be noted that the time frame of randomized controlled trials may have been inadequate to evaluate developmental outcomes fully.

Further meta-analyses have assessed the effects of iron supplementation on infectious disease morbidity. Gera 2002 compiled randomized controlled trials assessing any form of iron supplementation (including food fortification). Overall there was no difference in the incidence rate ratio for all recorded infections. Diarrhoea was more frequent in the iron-supplemented group, as was the rate of malaria parasitaemia at end of treatment (RR 1.43, 95% CI 1.08 to 1.91; 2390 children, nine trials). The latter was positively associated with the baseline rate of parasitaemia. Oppenheimer 2001 reviewed the effects of iron on infections from in vivo animal studies to clinical trials. Pooling of clinical trials was deemed inappropriate due to the large qualitative heterogeneity between existing studies. A qualitative summary noted that malaria was more frequent with iron in five of nine trials and none of the trials conducted in malarial regions favoured iron with regard to infectious outcomes. A review by the International Nutritional Anemia Consultative Group pooled 13 randomized controlled trials in children or adults and showed a small, non-statistically significant increase in the risk of clinical malaria (RR 1.01, 95% CI 0.9 to 1.3, eight trials) and a higher risk of being slide positive for P. falciparum at the end of the iron-supplementation period (RR 1.17, 95% CI 1.08 to 1.25, 13 trials) (INACG 1999).

Recently, a large randomized controlled trial evaluating the effect of iron and folate supplements in a malaria-intense area of Zanzibar was terminated prematurely on the recommendation of the study Data Safety and Management Board, due to a higher proportion of hospitalizations or death in the iron and folic acid-containing groups (Sazawal 2006 (C)a). Subgroup analysis revealed that the risk was limited to children who were iron-replete when iron supplementation was started. This trial heightened global concern about routine, non-selective iron supplementation policy in areas where malaria is highly prevalent, with regard to global childhood outcomes: mortality, overall infectious morbidity, and hospital admissions.

Iron supplementation guidelines

In areas where anaemia prevalence is 40% or more in children aged six months to 24 months, guidelines generally recommend that children of normal birthweight receive oral iron (12.5 mg, based on 2 mg/kg/day of elemental iron) daily between the ages of six months and 24 months, and children with low birthweight receive the same amount of iron between the ages two months and 24 months (Stoltzfus 1998; INACG 1999). Iron-deficiency anaemia is treated with iron supplement either continuously (daily) or intermittently (at intervals) with an oral preparation of elemental iron (3 mg/kg/day). Until recently, the World Health Organization (WHO) guidelines for children living in malaria-endemic areas were no different than the general recommendations (WHO 2003). However, following the above-described trial showing adverse effects of iron supplementation in Zanzibar (Sazawal 2006 (C)a), a new consensus statement has been issued (WHO 2007). The consultation concluded that while in malaria-endemic areas "there is evidence that iron supplements reduce serious morbidity in iron-deficient children if given with good health care, including treatment for malaria, other parasitic infestations, and infections..., where there is limited malaria prevention and clinical care, universal iron supplementation is associated with an increased risk of adverse events." Consequently, the consensus recommended against universal iron supplementation for children under the age of two years living in malaria-endemic areas. Screening to identify iron-deficient children was recommended with directed treatment of iron-deficient children only, accompanied by early identification and appropriate treatment of malaria. Specifically, during the first six months, iron supplementation was recommended only for preterm and low-birthweight infants who sleep under insecticide-treated bed nets and where all episodes of malaria illness can be promptly treated with effective antimalarials. For children above six months, only fortified complementary foods according to WHO guidelines were universally recommended, since these provide physiological doses of iron and not higher doses. Other fortification strategies were not recommended. Oral iron was recommended only for iron-deficient children (or those with clinical symptoms of severe anaemia in locations where testing is not available) with early diagnosis and appropriate management of infections. Folate supplementation was not recommended due to the potential interaction with antifolate antimalarials.

Why it is important to do this review

Observational studies show an association between anaemia, iron deficiency and impaired child development. Interventional studies and their meta-analyses to date did not show an improvement in most clinical outcomes with iron supplementation. In this situation, it is imperative to ensure that iron supplementation does not cause harm. This is especially important when considering routine iron supplementation for young children living in malarial areas, since the harms have been linked to malaria. Current WHO guidelines that recommend screening for iron-deficiency prior to iron supplementation are probably impractical. Assessment might be logistically not feasible in most malaria-endemic areas and ascertainment of iron deficiency in areas where infections are prevalent is difficult.

We looked at all randomized controlled trials assessing iron supplementation for children in malaria-endemic areas. We specifically searched for outcomes related to malaria and data for all-cause mortality, which ultimately combines the benefit and harm of the intervention. We also attempted to assess the effect of iron supplementation on haemoglobin; although it is well-recognised that iron supplementation increases haemoglobin, we expected that if an interaction between iron supplementation and malaria exists, the effects of iron supplementation on haemoglobin will be smaller in areas where malaria is highly prevalent. Since iron has been claimed to harm mainly iron replete children, we aimed to explore the differential effects of iron for children anaemic/iron deficient at baseline and those who were not.

Objectives

To evaluate the effects and safety of iron supplementation, with or without folic acid, in children living in malaria-endemic areas.

Methods

Criteria for considering studies for this review

Types of studies

Randomized controlled trials (RCTs) that randomized individuals or clusters. Cluster-randomized trials were considered eligible only if including at least two units per study arm.

Types of participants

Children less than 18 years living in a malaria-endemic area, regardless of the presence of any other medical condition or disease. We included children with or without anaemia, iron deficiency and/or malaria at baseline. For the purpose of this review we used the Hay definitions for malaria endemicity (Hay 2004; Table 1) and included children living in countries defined as hypoendemic, mesoendemic, hyperendemic, or holoendemic for malaria, unless specifically stated in the publication that the trial was conducted in an area without malaria. Studies conducted in countries not listed as malaria-endemic, but stating that the study was conducted in an area endemic for malaria, were included. We included multicentre trials involving countries where malaria is not endemic and planned to extract or obtain data for those children from malaria-endemic countries only. As this was not possible, we decided to exclude studies if > 30% of children in the study came from non-endemic regions.

Table 1. Description and location of malaria-endemic areas
  1. AFRO: WHO African Regional Office
    AMRO: WHO Americas Regional Office
    EMRO: WHO Eastern Mediterranean Regional Office
    EURO: WHO Europe Regional Office
    SEARO: WHO South East Asian Regional Office
    WPRO: WHO Western Pacific Regional Office

Area definitionParasite ratesDescriptionGeographical location
Hypoendemicity (also called designated unstable malaria)10% or less of children aged 2 to 9 years, but may be higher for part of the yearAreas where little transmission and during the average year the effects upon the general population are unimportantAFRO: Chad
AMRO: Belize, Bolivia, El Salvador, Guatemala, Mexico, Nicaragua, Costa Rica, Paraguay
EMRO: Afghanistan, Iraq, Oman
EURO: Armenia, Azerbaijan, Georgia, Kyrgyzstan, Tajikistan
SEARO: Nepal
WPRO: China
Mesoendemicity (also called unstable and stable malaria)11% to 50% of children aged 2 to 9 yearsTypically found among rural communities in subtropical zones where wide geographical variations in transmission existAFRO: Angola, Botswana, Cape Verde, Chad, Eritrea, Ethiopia, Kenya, Mauritania, Namibia, Niger, Zambia, Zimbabwe
AMRO: Brazil, Colombia, Ecuador, Guyana, Panama, Peru, Venezuela
EMRO: Iran, Pakistan, Saudi Arabia
SEARO: Bangladesh, Bhutan, India, Indonesia, Sri Lanka, Thailand
WPRO: Malaysia
Hyperendemicity (also called stable malaria)Consistently > 50% among children aged 2 to 9 yearsAreas where transmission is intense but seasonal; immunity is insufficient in all age groupsAFRO: Angola, Benin, Burkina Faso, Cameroon, Central African Republic, Chad, Congo, Cote d'Ivoire, Equatorial Guinea, Gabon, Gambia, Ghana, Guinea, Guinea-Bissau, Liberia, Madagascar, Malawi, Mali, Mozambique, Nigeria, Sao Tome and Principe, Senegal, Sierra Leone, Togo, Uganda,Tanzania, Zambia
SEARO: Timor-Leste
WPRO: Papua New Guinea, Philippines, Solomon Islands, Vanuatu, Vietnam
Holoendemicity (also called stable malaria)Consistently > 75% among infants 0 to 11 monthsIntense transmission resulting in a considerable degree of immunity outside early childhoodAFRO: Central African Republic, Democratic Republic of Congo, Tanzania, Uganda, Burundi, Madagascar, Malawi, Mozambique
AMRO: Dominican Republic, Suriname
EMRO: Djibouti, Somalia, Sudan, Yemen
SEARO: Myanmar
WPRO: Cambodia, Lao People's Democratic Republic

Types of interventions

Intervention

Orally administered iron tablets or elixir, irrespective of dose, duration, or interval of administration. Co-administration of an antimalarial drug, antihelminthic drug and/or folic acid was permitted.

Control

Placebo, no treatment, antimalarial, or antihelminthic drugs.

Trials administering the same micronutrient/s to both study arms were included (e.g. zinc, vitamin A, vitamin C in the same dose).

Thus, the following comparisons were included.

  • Iron alone versus placebo/no treatment.

  • Iron + folic acid versus placebo/no treatment.

  • Iron with anti-malarial drug versus anti-malarial drug.

  • Iron with anti-malarial drug versus placebo/no treatment.

Excluded
  • Iron administered as fortification of food or water.

  • Iron given parenterally.

  • Dose comparisons of iron.

  • The comparisons of:

    • iron versus antimalarial drug;

    • iron versus antihelminthic drug;

    • iron versus folic acid;

    • iron versus micronutrients.

Types of outcome measures

We extracted the following outcomes at end of treatment. If further follow up was given, we extracted the outcomes at end of follow up, up to 12 months after the end of therapy.

Primary outcomes
  • Malaria-related outcomes (all outcomes refer to all malaria species):

    • clinical malaria (defined by symptoms and laboratory diagnosis);

    • severe malaria (clinical malaria with evidence of vital organ dysfunction);

    • malaria parasite prevalence and density.

  • Death from any cause.

Secondary outcomes
  • Prevalence of anaemia, as defined in study.

  • Haemoglobin levels (end of treatment values and change from baseline to end of treatment).

  • Hospitalizations from any cause.

  • Morbid episodes caused by other infections than malaria (including diarrhoea, pneumonia, sepsis, meningitis, measles, pertussis).

  • Adherence (proportion of days supplement taken).

  • Weight and height measures, preferably using change from baseline to end of treatment values.

Search methods for identification of studies

We attempted to identify all relevant trials regardless of language or publication status (published, unpublished, in press, and in progress).

Electronic searches

Databases

The search specialist at the Cochrane Infectious Diseases Group editorial base searched the following databases using the search terms and strategy described in Table 2: the Cochrane Infectious Diseases Group Specialized Register; Cochrane Central Register of Controlled Trials (CENTRAL), published in The Cochrane Library (2009, issue 1); MEDLINE (1966 to March 2009); EMBASE (1980 to March 2009); and LILACS (1982 to March 2009). We also searched the metaRegister of Controlled Trials (mRCT) using 'iron' and 'malaria' as search terms.

Table 2. Detailed search strategies
  1. ^Cochrane Infectious Diseases Group Specialized Register.
    ^^Search terms used in combination with the search strategy for retrieving trials developed by The Cochrane Collaboration (Higgins 2006); upper case: MeSH or EMTREE heading; lower case: free text term.

Search setCIDG SR^CENTRALMEDLINE^^EMBASE^^LILACS^^
1ironironironironiron
2ferrousferrousferrousFERROUS-SULPHATEferrous
31 or 2IRON COMPOUNDSIRON COMPOUNDS1 or 21 or 2
4malaria1 or 2 or 31 or 2 or 3supplem$malaria
5anemiasupplem*supplem*3 and 4anemia
6anaemia4 and 54 and 5malariaanaemia
74 or 5 or 6malariamalariaanemia4 or 5 or 6
83 and 7anemiaanemia6 or 73 and 7
9anaemiaanaemia5 and 8
107 or 8 or 97 or 8 or 9child$
116 and 106 and 10infant$
12child*10 or 11
13infant*9 and 12
1412 or 13
1511 and 14

Searching other resources

Researchers, organizations, and pharmaceutical companies

We contacted the primary investigators of all included trials, those ongoing or those awaiting assessment to ask for unpublished data and further trials.

Conference proceedings

We searched the proceedings of the Fourth Multilateral Initiative on Malaria Pan-African Conference (13 to 18 November 2005, Cameroon) for relevant abstracts.

Reference lists

We scanned the bibliographies of all included trials, pertinent reviews, and previous meta-analyses for additional references.

Data collection and analysis

Selection of studies

Two authors (JUO and JO) inspected the abstract of each reference identified and obtained the full text of relevant articles. Both authors independently reviewed the articles and applied inclusion criteria. If a potentially relevant trial was eligible for the review but the information was unclear, we contacted the trial authors for clarification. Areas of disagreement were resolved by discussion with a third author (MP or DY). Each trial was scrutinized to identify multiple publications from the same data set. We documented the justification for excluding trials from the review. Studies are named by the first author and year of publication. The addition of a, b, c... is given for different studies from the same author and year of publication; the addition of I, II, III... is given to the same study entered more than once to Review Manager 5 (RevMan) (RevMan 2008), since results were reported only by stratification (age, haemoglobin, etc.). The addition of (C) to the trial's identification denotes that the trial was cluster-randomized.

Data extraction and management

Two authors independently extracted data into a pre-piloted data extraction spreadsheet. DY extracted data from all trials and double extraction was allotted between JUO, JO, and MP. Differences in the data extracted were resolved by discussion. Data were entered into RevMan 5 by one author (MP, JUO, or JO).

Individually randomized trials

For dichotomous outcome measures, we recorded the number of participants experiencing the event and the number analysed in each treatment group. For count data, we recorded the number of events and the number of person-months of follow up in each group. If the number of child-months was not reported, the product of the duration of follow up and the sample sizes available at the beginning and at the end of the study were extracted to estimate this figure. For continuous data, we extracted means (arithmetic or geometric) and a measure of variance (standard deviation (SD), standard error (SE), or confidence interval (CI)) together with the numbers analysed in each group. SDs were computed from SEs (SD=SE*SQRT(N)) or 95% CIs (SD=SQRT(N)*(upperCI-lowerCI)/3.92), assuming a normal distribution of the values. Haemoglobin values in g/dL were approximated by multiplying hematocrit or packed cell volume (PCV) values by 0.34, when haemoglobin was not reported. For count data we calculated the rate ratio and SE for each study. Zero events were replaced by 0.5. If adjusted incidence rate ratios were reported in the original studies we used these data with SEs. The SE was calculated from 95% CIs as SE=(upper-lower)/3.92.

Cluster-randomized trials

We recorded the unit of randomization (e.g. household, compound, sector, village), the number of clusters in the trial and the average cluster size. The statistical methods used to analyse the trial were documented along with details describing whether these methods adjusted for clustering or other covariates. We planned to extract estimates of the intracluster correlation (ICC) coefficient for each outcome, but none of the trials provided these data. Missing information was requested from trial authors.

Where results have been adjusted for clustering, we extracted the treatment effect estimate and the standard deviation or confidence interval. If the results were not adjusted for clustering, we extracted the same data as for the individually randomized trials (see below, 'Unit of analysis issues', regarding the use of these data).

Assessment of risk of bias in included studies

Two authors independently assessed the risk of bias. We graded the generation of the allocation sequence and allocation concealment as adequate, unclear, inadequate, or not described according to Juni 2001, as recommended in the Cochrane Handbook for Systematic Reviews of Interventions (Cochrane Handbook 2008 chapter 8). We contacted the authors if information was missing. We recorded whether analysis was by intention-to-treat and, if not, the number of patients excluded from analysis for the primary outcomes. We assessed blinding by recording whether participants, investigators, or outcome assessors were aware of the treatment group allocation. We present a summary of the quality assessment in a table and detail the actual methods in the 'Characteristics of included studies'.

We assessed selective reporting for all-cause mortality, since we expected all trials to report all-cause mortality. We did not expect all other outcomes to be reported in all trials, since included trials targeted primarily a range of outcomes (haematinic, infectious, behavioural, cognitive, etc.)

Measures of treatment effect

For dichotomous data we calculated risk ratios (RR) and for continuous data we calculated absolute mean differences (MD), with 95% CIs. Standardized mean differences (SMD) were calculated for the outcomes of weight and height. For count data we computed rate ratios with SEs for each trial. Infectious episodes, hospitalizations, and clinic visits were analysed as count data, and rate ratios per child-months are reported.

Unit of analysis issues

Two issues arose in the present review: the analysis of cluster-randomized trials and outcomes occurring more than once per individual randomized.

1. When cluster-RCTs reported results as if they were individually randomized, we extracted the data reported in the trial and used an estimated design effect (DE) to adjust standard errors or sample size (Cochrane Handbook 2008 chapter 16). The equations recommended in the Handbook used for all analyses were: unadjusted standard error of the log risk ratio [SE(lnRR)] * DE0.5 = adjusted SE(lnRR) and DE = 1 + [(average cluster size - 1) * intracluster correlation coefficient]. Based on other trials included in the review we assumed an average cluster size of 1.5 for households and 32 for classes, when the average cluster size or number of clusters and individuals were not reported (see tables of included studies, Table 3 and Table 4 for reported and assumed cluster sizes). The DEs or intracluster correlation coefficients (ICCs) used for the different outcomes were the following:

Table 3. Analysis of cluster-randomized trials adjusting standard errors
  1. In bold - results provided in publication/from authors adjusted for clustering.
    n - number of outcomes; N - number evaluated; Int - intervention; cont - control; DE - design effect used for adjustment (see methods for derivation of design effect and ICC used per outcome).

Study IDOutcomen Int reportedN Int reportedn Cont reportedN Cont reportedAverage cluster sizeDEUnadjusted RR (95% CI)ln(RR)Unadjusted SE(lnRR)Adjusted SE(lnRR)/ sample size
Adam 1997 (C)Clinical malaria7236649372Household (used 1.5)1.341.49 (1.07 to 2.08)0.400.170.20
Adam 1997 (C)Parasitaemia127368101372Household (used 1.5)1.341.27 (1.02 to 1.58)0.240.110.13
Adam 1997 (C)Severe malaria (necessitated hospitalization)4140532382Household (used 1.5)1.341.21 (0.78 to 1.88)0.190.220.26
Mebrahtu 2004 (C)Parasitaemia3073071.51.34OR 0.9 (0.72 to 1.19) Converted to RR 0.980.470.280.32
Mebrahtu 2004 (C)High-grade parasitaemia3073071.51.34OR 1.04 (0.82 to 1.34) Converted to RR 1.030.030.120.14
Sazawal 2006 (C)aClinical malaria467795041180061.41.16 (1.00 to 1.34)0.150.07
Sazawal 2006 (C)aSevere malaria (cerebral)795080061.41.32 (1.02 to 1.70)0.280.13
Sazawal 2006 (C)bClinical malaria14815308041.20.46 (0.24-0.88)-0.780.33
Sazawal 2006 (C)bSevere malaria (cerebral)4815158041.20.26 (0.09 to 0.81)-1.350.56
Smith 1989 (C)Clinical malaria1497889Household (used 1.5)1.341.60 (0.42-0.71)0.470.420.48
Smith 1989 (C)Parasitaemia28971689Household (used 1.5)1.341.61 (0.93 to 2.76)0.470.280.32
Smith 1989 (C)High-grade parasitaemia17971189Household (used 1.5)1.341.42 (0.70 to 2.86)0.350.130.15
Table 4. Analysis of cluster-randomized trials adjusting sample size
  1. None of the trials provided results adjusted for clustering for the outcomes reported in the table.
    n - number of outcomes; N - number evaluated; Int - intervention; Cont - control; DE - design effect used for adjustment (see methods for derivation of design effect and ICC used per outcome).

Study IDOutcomen Int reportedN Int reportedn Cont reportedN Cont reportedAverage cluster sizeDEn Int adjustedN Int adjustedn Cont adjustedN Cont adjusted
Adam 1997 (C)Anaemia364368357374Household (used 1.5)1.4260263255267
Hall 2002 (C)Anaemia273551356562202.7799199129203
Hettiarachchi 2008 (C)Anaemia2819378181323.887502047
Mebrahtu 2004 (C)All-cause mortality034023441.51.00103402344
Mebrahtu 2004 (C)Anaemia1802321722721.51.4129166123194
Roschnik 2004 (C)Anaemia125708205802293.603519657223
Roschnik 2003 (C)Anaemia133224110203303.7036613055
Sarma 1977 (C)Anaemia9412545253.23313814
Sazawal 2006 (C)aAll-cause mortality149795013080061.41.00114979411307996
Sazawal 2006 (C)bAll-cause mortality881598041.21.000488159804
Sazawal 2006 (C)bAnaemia430873271.21.432205234
Seshadri 1984b (C)Anaemia7422447222.95214816
  • Malaria (Sazawal 2006 (C)a): unadjusted RR 1.14 (CI 1.01 to 1.30) (using number of events reported and number evaluated for clinical malaria), SE(lnRR) = 0.064391; Adjusted RR 1.16 (CI 1.00 to 1.34) (reported in the publication for the same outcome), SE(lnRR) = 0.074661. DE = (0.074661/0.064391)2 = 1.3444 for an average cluster size of 1.4 (households). All trials reporting on malaria-related outcomes used households as the unit of randomization and the same DE of 1.34 was used for all trials and all malaria-related outcomes.

  • Deaths (Tielsch 2006): unadjusted RR 1.04 (0.80 to 1.34), SE(lnRR) = 0.131585; Adjusted RR 1.03 (0.78 to 1.37), SE(lnRR) = 0.143692. DE = (0.143692/0.131585)2 = 1.192481 for an average cluster size of 82 (sectors), ICC = 0.002. The DE adjusted for a cluster size of 1.5 (household) was 1.001.

  • Anaemia (Kaiser 2006): we expected significant intracluster correlation between children in the same household (given similar nutritional status and infection incidences) and a lower degree of clustering at the community level. We did not find, however, ICC estimates in the literature for these units and the cluster-RCTs included in the review did not provide data allowing us to calculate DE/ ICC. Ngnie-Teta 2007 reported that the degree of community-level clustering with regard to moderate-to-severe anaemia among Beninese and Malian children was 0.14 to 0.19. Assuncao 2007 reported an ICC of 0.07 and DE of 2.5 in a cluster survey of all children under six years of age, where clusters of about 30 children comprised several households each. Kaiser 2006 reported the design effects between 1.4 and 2.4 for anaemia < 11 g/dL in three cluster surveys in Afghanistan and Mongolia (DEs between 1.4 and 2.4), allowing us to calculate ICCs between 0.093 and 0.100. We used a DE of 1.4 for trials using households as the unit of randomization and an ICC value of 0.093 to calculate the DE of trials using larger units of randomization (DE range 2.8 to 3.9). The ICC for haemoglobin, though measuring the same thing, is much smaller because it is a more precise measure. We used an ICC of 0.000 for households and 0.00271 for large clusters (school/class) basing on values reported for households and the district health authority level, respectively, although these values refer to adults in England (Gulliford 1999).

  • Diarrhoea (Kaiser 2006): the pooled design effect from five observational studies in Kaiser 2006 was 3.1, for an average cluster size of 17, ICC = 0.131. The DE adjusted for a cluster size of 1.5 was calculated as 1.065.

  • Infectious episodes: a DE of 1.36 was calculated from Sazawal 2006 (C)a that reported both raw numbers and adjusted RR/SE.

We used the design effect to adjust SEs or sample sizes of cluster-RCTs. When one or more of the cluster-RCTs reported RRs adjusted for clustering we used these data and pooled results using the generic inverse variance method in RevMan. This was used for all the malaria-related outcomes and diarrhoea. For the other outcomes, where none of the cluster-RCTs provided cluster-adjusted RRs, we divided the number of reported events and number of individuals in each arm by the design effect, thus reducing the reported sample size to its effective sample size given clustering (only the number evaluated for continuous outcomes). In this case, standard methods for pooling of trials' dichotomous and continuous outcomes were used.

2. Count data (such as diarrhoea, respiratory infections, fever) were re-expressed as episodes/child-month. This analysis may be imprecise since it assumes an equal distribution of the episodes throughout the period of observation; multiple events were common for these outcomes in individual children (up to eight events/child-year), the intervention (iron) may affect the likelihood of children experiencing multiple events, and the person-months of follow up was usually not reported but estimated as the product of the number of children evaluated and the trial's follow up. When rate ratios and SEs were reported in the publications we used these, preferably using the adjusted values.

3. Trials with four study arms were included twice in the meta-analysis for different comparisons. For three-armed trials we added the iron groups (dichotomous outcomes) or used the arm with the higher or more frequent (i.e. daily) dosing of iron (continuous outcomes). We did not include a study arm more than once in the same meta-analysis and did not divide study arms.

Dealing with missing data

We contacted the trial authors if the available data were unclear, missing, or reported in a format that is different to the format that we required.

We aimed to perform an intention-to-treat analysis where the trial authors accounted for all randomized participants; otherwise we performed a complete-case analysis.

Assessment of heterogeneity

Heterogeneity in the results of the trials was assessed by visually examining the forest plot to detect non-overlapping confidence intervals and using the Chi2 test of heterogeneity (P < 0.1 indicating statistical significance) and the I2 measure of inconsistency (with a value > 50% denoting moderate levels of heterogeneity). When statistical heterogeneity was present we investigated the reasons using subgroup analysis and meta-regression and report results using both the fixed and random-effects model.

Assessment of reporting biases

A funnel plot was constructed to assess the effect of small studies for the main outcome (when including more than 10 trials).

Data synthesis

We compiled dichotomous data using the Mantel Haenszel model and used the generic inverse variance method for continuous data. We compiled risk ratios and SEs using the generic inverse variance method. For the outcome of parasitaemia an analysis of odds ratios was performed in addition to the analysis of risk ratios, to permit the inclusion of one cluster-randomized trial that reported results as odds ratios (ORs) (Mebrahtu 2004 (C)).

We performed meta-regression to assess the effect of continuous variables (iron dose and duration of supplementation, mean children's age, and mean baseline haemoglobin) on effect estimates.

We used RevMan 2008 version 5.0.18 for data analysis. We performed meta-regression and subgroup comparisons using Comprehensive Meta-Analysis version 2.2, using random-effects regression, unrestricted maximum likelihood method.

Results for the Mantel Haenszel fixed-effect model are presented with 95% confidence intervals (CI) for the distribution of the effect estimate among studies. Results for the random-effects model are presented with 95% CI for the confidence in the mean effect estimate obtained from the meta-analysis. The dispersion of the effect estimate among the studies was calculated as 2´tau below the random-effects pooled estimate to 2´tau above it for absolute measures and EXP(2´tau below) to EXP(2´tau above) the LN(effect estimate) for relative measures.

Subgroup analysis and investigation of heterogeneity

We conducted subgroup analyses by treatment versus prevention of anaemia (> 50% or < 50% of children in trial anaemic at baseline, respectively), age (included children < five years versus > five years), and malaria endemicity (hypoendemic or mesoendemic areas, referred to as hypo/mesoendemic, versus hyperendemic or holoendemic areas, referred to as hyper/holoendemic). Studies performed during the rainy season and stating specifically that malaria was highly prevalent at the time of the trial in the region were classified in the hyper/holoendemic subgroup regardless of the endemicity of the country in which the trial was performed. We performed subgroup analyses only on complete trials (e.g. separating between trials recruiting children below or above five years of age), and not on subgroups of children within trials unless children were stratified at baseline according to the variable tested (age, anaemia).

Sensitivity analysis

We carried out sensitivity analyses by exclusion of studies with the pre-specified characteristics, as necessary, to assess the effects of: studies reporting on episodes/events rather that patients with one or more events for dichotomous outcomes; the addition of other micronutrients and/or anti-parasitic drugs to both study arms; iron dose; duration of treatment and follow up; and baseline haemoglobin.

In addition, we performed sensitivity analyses to assess the effect of study methodological risk assessment. We primarily assessed the effect of allocation concealment by separating the analysis for studies reporting adequate allocation concealment versus others.

Results

Description of studies

See: Characteristics of included studies; Characteristics of excluded studies; Characteristics of studies awaiting classification.

Results of the search

Three searches were conducted: in July 2007 (201 citations), June 2008 (50 additional citations), and March 2009 (25 additional citations). We inspected the references of all relevant trials for further titles, which yielded many references not identified in the original search. Overall, 303 publications were potentially eligible for inclusion and we reviewed their full text: 105 citations, representing 73 individual trials fulfilled the inclusion criteria; 104 publications were excluded (representing 95 different studies) and four are awaiting assessment (see 'Characteristics of studies awaiting classification'). All four trials awaiting assessment assessed malaria-related outcomes in hyperendemic or holoendemic areas. Two trials were identified in trial registries and are not yet published (Gomes 2001; Denno 2006). One was identified as a conference proceeding (Browne 2005) and might refer to the same trial as (Gomes 2001). Finally, the results of the comparison between iron + folic acid + zinc + vitamin A versus zinc + vitamin A (comprising 16,199 children), from a large cluster-randomized trial included in the review (Sazawal 2006 (C)c), are not yet published.

We made an attempt to contact the authors of all included and potentially relevant trials where the full publication did not provide enough information. We primarily requested data on malaria and all-cause mortality. Correspondence was established with 22 authors, of which 19 could supply further information, most commonly related to trial methods only.

Included studies

A description of the 73 included trials is provided in the 'Characteristics of included studies' tables. Five trials did not report outcomes relevant to our review and are not discussed further pending availability of information from the trial authors. All description and results refer to the 68 trials reporting one or more of the review-defined outcomes.

The trials were published between the years 1973 and 2007. Forty trials reported the years in which the study was performed (also between 1973 and 2007). Overall 42,981 children were recruited in the included trials: 15,873 (36.9%) in 54 individually-randomized trials and 27,144 in 14 cluster-RCTs. The largest cluster-RCT included two separate, independent cohorts: the main trial (Sazawal 2006 (C)a, 15,956 children) and an independent substudy (Sazawal 2006 (C)b, 1619 children). Previously unpublished data on the outcome of malaria from the substudy are included in this review. Adherence was reported in 35 studies and the average overall adherence to all study drugs was good (85%) and lower in cluster-RCTs (72%) compared to individually randomized trials (89.4%). We extracted data and outcomes from all publications listed for each included study (one to four publications per study).

We performed three main comparisons: 1) iron given for prevention or treatment of anaemia (65 trials) 2) iron with an anti-malarial drug given for prevention or treatment of anaemia (four trials) 3) iron given together with antimalarial drugs in the treatment of clinical malaria (three trials). These comparisons are described separately.

1. Iron given for prevention or treatment of anaemia (42,259 children; 65 trials)

The following interventions were assessed in these trials:

  • iron alone versus placebo/no treatment: 50 trial arms;

  • iron + folic acid versus placebo/no treatment: 16 trial arms;

  • iron with anti-malarial drug versus anti-malarial drug: five trial arms.

Iron was administered for a mean duration of 4.2 months (SD 2.7), median 3.5 months (range 1.5 to 12 months). A follow-up period longer than treatment duration was reported in 25 trials, with follow ups after treatment ranging between one week and 10 months.

Antihelminthic drugs directed at intestinal parasites were added to both arms in 17 trials and to the iron arm alone in one additional trial. Additional micronutrients were added similarly to both trial arms in 13 trials, as described in the 'Characteristics of included studies'.

The trials were performed in areas of varying malaria transmission rates:

  • holoendemic areas: six trials, all of which reported on malaria-related outcomes;

  • hyperendemic areas: 16 trials, of which three specified that the trial was performed during the rainy period with malaria and three reported that the trial was performed in the dry season. Five of these trials reported on malaria-related outcomes, including two of the trials performed during the rainy season;

  • mesoendemic areas: 36 trials, of which five specified that the trial was performed during the rainy period. Six trials reported on malaria-related outcomes, including four trials conducted during the rainy season

  • hypoendemic areas: four trials, none of which reported on malaria;

  • combined: one trial was multicentred (Smuts 2005) and the majority of children were from mesoendemic or hyperendemic countries and did not assess malaria (included together with hypo/mesoendemic trials for sensitivity analyses).

Thirty-three trials recruited mostly children < five years (all children < five years in 28). Twenty-eight trials recruited mostly children > five years (all > five years in 23). Four trials did not target a specific age group. The mean haemoglobin at baseline was highly variable (median 10.9, range 8.5 to 15.7 g/dL), as was the baseline prevalence of anaemia (range 0% to 100%). Thirty trials excluded children based on a minimal haemoglobin value, which was most commonly set at 8 g/dL. Ten studies excluded children above a threshold of haemoglobin that was set at 12 g/dL. In most trials (58%) more than 50% of recruited children were anaemic at baseline. All children were anaemic in 10 trials (anaemia defined most commonly as Hb < 11 g/dL), in four studies all were non-anaemic, and eight other trials performed randomized stratification by anaemia status at baseline and reported results separately for anaemic and non-anaemic children. Since most trials did not separate between the interventions of prevention (iron given for non-anaemic children) and treatment (iron given for children with anaemia), our primary analysis refers to both. Separate analyses for prevention and treatment are represented in the subgroup analyses of anaemic versus non-anaemic children at baseline.

2. Iron with anti-malarial drug versus placebo/no treatment (1915 children; four trials, also included in comparison #1)

The multi-armed design of four trials permitted the assessment of the comparison of iron combined with an antimalaria drug versus placebo for the treatment or prevention of anaemia (Greisen 1986 (C); Menendez 1997; Verhoef 2002; Massaga 2003). All were conducted in malaria hyper/holoendemic settings, as expected.

3. Iron given to children with clinical malaria (722 children; three trials)

Three trials assessed the addition of iron to the treatment of children with proven clinical malaria. The mean baseline haemoglobin among children in these trials was 8.9 (Nwanyanwu 1996), 4.1 (van den Hombergh 1996) and 9.6 g/dL (van Hensbroek 1995). The trials reported primarily on rates of parasitaemia and parasite density at end of treatment.

Excluded studies

The specific reasons for exclusion are detailed in the 'Characteristics of excluded studies' tables. The major reasons for exclusion were:

  • RCT conducted in non-malaria-endemic areas: 29 publications. Ten of these studies were performed in hypoendemic or mesoendemic areas but the publication or contact with the authors established that there was no malaria activity in the region of the trial;

  • incompatible intervention, such as administration of iron together with other micronutrients: 10 publications;

  • iron administered as fortification of food or drink: 20 publications;

  • parenteral administration of iron: seven publications;

  • both study arms were given iron (dose, schedule or other comparisons): 16 publications;

  • cluster-RCT with less than two clusters per arm: one publication;

  • non-randomized design: 19 publications;

  • study included adults or did not separate adults and children: two publications.

Risk of bias in included studies

The risk assessment pertains to the 68 trials with outcomes relevant to our review and is presented graphically overall in the review in Figure 1 and per study in Figure 2.

Figure 1.

Methodological quality graph: review authors' judgements about each methodological quality item presented as percentages across all included studies. The unclear category for incomplete outcome data represents studies that did not report this outcome, thus the relevant percentages are only those of yes or no.

Figure 2.

Methodological quality summary: review authors' judgements about each methodological quality item for each included study. The unclear category for incomplete outcome data represents studies that did not report this outcome, thus the relevant percentages are only those of yes or no.

Allocation

Allocation concealment was described as adequate in 28 trials. The concealment was judged inadequate in one trial that used alternation for randomization of households (Smith 1989 (C)). All other trials did not describe their methods clearly or did not provide a description. The generation of the randomization sequence was described as adequate in 27, inadequate in the one trial using alternation, and unclear or undescribed in all others. Both allocation concealment and generation were described as adequate in 18 trials.

Several studies (Seshadri 1982a; Seshadri 1982b; Kashyap 1987; Latham 1990) described the randomization as follows: all children were divided into pairs or groups of three. One child in each group was randomly allocated to the intervention and the others were assigned to control. These studies were classified as unclear for generation of the randomization sequence and allocation concealment, since it was not clear whether this was done openly in front of the children (inadequate concealment) or on a list prior to seeing the children (adequate concealment).

Blinding

Forty-six trials described double blinding or stated that the trial was double blind with no description of the blinding techniques (see 'Risk of bias' tables). In one further trial the field workers that distributed the treatment were blinded (Vaughan 1977). The remaining 21 studies were open-labelled.

Incomplete outcome data

Incomplete outcome data reporting was assessed for the outcomes of mortality, malaria, and haemoglobin/anaemia. Overall, only for mortality were there more studies with high risk of bias than studies with no risk of bias, due to unwarranted exclusions of patients from analysis (Figure 1). Overall the number of patients dropped from outcome reporting was equally distributed in both study arms (RR for drop-outs 1.00, 95% CI 0.94 to 1.07, Analysis 1.1).

Selective reporting

As described above, the trials were designed to investigate different primary outcomes. Thus, we expected inconsistent reporting of the outcomes pre-specified in our review. However, we did expect mortality to be reported in all trials, as a safety measure and as global measure of infection-related benefit and harm. We found mortality reported only in 16/73 included trials, among which no deaths were reported in six trials. We obtained further data from the authors of nine trials; all responded in general that "there was no mortality in the study". These data may not be reliable and the denominator, assumed to be all included participants, might have been different.

Other potential sources of bias

Fourteen of the included trials were cluster-randomized, using households (five trials) or schools/classes (nine trials) as the unit of randomization. Only Sazawal 2006 (C)a and its substudy (Sazawal 2006 (C)b) adjusted the comparisons for clustering (for the main outcomes). The other trials reported results per individual only and did not provide data regarding the ICC. The average cluster size could usually be calculated from the number of clusters and individuals included in the trial. In the analysis these trials are described under 'Unit of analysis issues'. Crude results reported in the publication, the design effect used for adjustment, and the adjusted results used in the meta-analyses for the main outcomes are provided in Table 3 (for outcomes pooled using adjusted standard errors) and in Table 4 (for outcomes pooled using the effective sample size). All data entered into the meta-analyses were adjusted for clustering using the calculations and assumptions detailed in our methods.

Effects of interventions

1. Iron versus placebo for treatment or prevention of anaemia (65 trials)

Malaria-related outcomes

Sixteen trials provided published outcome data on malaria. Outcomes were obtained through author correspondence in two additional trials (Sazawal 2006 (C)b; Fahmida 2007). Malaria-related outcomes reported in the trials and their definitions are described in Table 5. Five trials restricted outcome assessment to P. falciparum species, which was also the mostly frequently found malaria species in most other trials that reported on the type of malaria (Table 5). The authors of one study replied that malaria was collected, but the data were on paper forms and can no longer be obtained (Powers 1983). Taylor 2001 reported on malaria without separating results according to the study arms and we did not succeed in contacting the authors. The authors of 14 trials replied that malaria-related outcomes were not collected in their trials (Sarma 1977 (C); Greisen 1986 (C); Latham 1990; Idjradinata 1993; Dossa 2001a; Dossa 2001b; Hall 2002 (C); Hess 2002; Shah 2002; Baqui 2003; Zlotkin 2003; Nagpal 2004; Aggarwal 2005; Hettiarachchi 2008 (C)). Nearly all trials reporting malaria-related outcomes performed regular surveillance for malaria using blood smears at baseline and during treatment (either at regular intervals or whenever febrile) and offered trial participants treatment when symptomatic (Table 5). Notably, no surveillance or treatment outside the hospital were offered in the main Sazawal 2006 (C)a trial, unlike its substudy (Sazawal 2006 (C)b) where monitoring was performed and treatment was offered to children at their home. The baseline rate of malaria parasitaemia (reported in 11/16 trials) ranged from 0% to 70% of children (mean 45%).

Table 5. Studies reporting malaria as an outcome: malaria definitions, types of outcomes and methods of surveillance and treatment within the trial
  1. Time of assessment: refers to time from randomization; m - month, Tx - treatment, FU - follow up

Study IDClinical definitionLaboratory definitionMalaria-related outcomes reportedTime of assessmentMalaria surveillance and treatment
Adam 1997 (C)Physician's diagnosis of malariaAny parasitaemia (all malaria species)Clinical malaria; any parasitaemia; malaria necessitating hospitalization (used as 'severe malaria'; parasite density (all N events/N individuals, unadjusted for clustering)3 m - end of TxBlood smears for malaria obtained before, during and after treatment. Children with clinical malaria referred to local hospital and treated
Berger 2000Isolated feverParasite density > 3000 (P. falciparum, P. malariae and P. ovale assessed. > 97% were P. falciparum)Parasite index (%, used as 'parasitaemia'); parasitaemia above 3000 (used as severe malaria) and 10,000 (%); parasite density

3 m - end of Tx

9m - end of FU

Blood smears for malaria obtained at baseline, end of Tx (3 m) and end of FU (6 m). Chloroquine treatment given for all isolated fevers
Desai 2003Fever >= 37.5Any parasitaemia (P. falciparum) with fever or parasitaemia > 5000/mm3 aloneClinical malaria; any parasitaemia; hazard ratios for these; parasite density3 m - end of TxBlood smears at baseline and every 4 weeks. Oral quinine given for any fever with parasitaemia and cases of severe malaria referred for further treatment
Fahmida 2007Not statedNot statedPatients with "malaria" (used primarily as 'clinical malaria')6 m - end of TxNot stated
Gebresellassie 1996Fever >= 37.5°C with signs and symptoms suggestive of malaria and other diagnoses ruled outPresence of parasites in blood (all species, P. falciparum 88.9%)Children with at least 1 episode of clinical malaria; cumulative incidence of parasitaemia; parasite density > 5000 (used as 'severe malaria'); parasite density

3 m - end of Tx

6 m - end of FU

Blood smears negative at baseline and repeated weekly. Chloroquine +/- primaquine given for any positive smear
Harvey 1989Fever and headache at the same timeAny parasitaemia (P. falciparum 67%, P. vivax 26.4%, P. malariae 6.6%)First episodes of clinically-suspected malaria (used primarily as 'clinical malaria'); any parasitaemia

4 m - end of Tx

6 m - end of FU

Blood smears for malaria obtained at 0, 6, 16 and 24 weeks. Chloroquine given for any illness reported as fever and/or headache
Latham 1990Not assessedAny positive smear (malaria species not stated)Any positive smear; parasite density8 m - end of FUBlood smears for malaria obtained at baseline and end of treatment. Treatment not stated.
Lawless 1994Child recall of clinical illnessAny positive blood smear (malaria species not stated)'Malaria' is not defined (used as 'clinical malaria')3.5 m - end of TxNo blood smears at baseline or during the trial (only at end of treatment). Treatment not stated.
Leenstra 2009Fever >= 37.5Positive blood smear (malaria species not stated)Episodes of clinical malaria and RRs adjusted for school; episodes of malaria parasitaemia and parasitaemia > 500 parasites/mm3 (used as 'severe malaria') and RRs adjusted for school, age, and baseline parasitaemia5 m - end of TxBlood smears for malaria at baseline (1/4 of participants positive) and monthly during the trial. No treatment offered for positive smears; symptomatic cases referred to physician.
Massaga 2003History of fever within the last 24 to 72 hours or measured temperature of >= 37.5Any level of parasitaemia (P. falciparum)Clinical malaria as first or only episode per patient (used as ';clinical malaria') and episodes of clinical malaria; episodes of clinical malaria associated with parasitaemia > 5000 parasites/microL (used as 'severe malaria')6 m - end of TxBlood smears for malaria at baseline and every 2 weeks. Sulfadoxine/ pyrimethamine treatment given for uncomplicated cases; complicated and severe malaria referred to the hospital.
Mebrahtu 2004 (C)Not assessedAny positive smear (P. falciparum)Parasitaemia as OR (95% CI) adjusted for repeated measurements in each child12 m - end of TxBlood smears for malaria at baseline and end of treatment. In addition, monthly smears from a random sample (50% of randomized). Treatment not stated.
Menendez 1997Fever >= 37.5Parasitaemia of any density (P. falciparum)First or only episode of clinical malaria1 yr (6 months after end of Tx)Blood smears for malaria at baseline, week 8 and for any fever. Chloroquine treatment given for clinical malaria.
Nwanyanwu 1996Fever > 37.5> 500 asexual parasites/microL thick smear (P. falciparum)Parasitaemia, parasite density1 m - end of Tx (all children treated for malaria)All children smear positive at baseline per study design and treated throughout the trial (trial's intervention)
Richard 2006Any fever within the past 72 hoursP. falciparum (29%) or P. vivax (71%), any densityEpisodes of falciparum and/or vivax malaria (used primarily as 'clinical malaria')7 m - end of TxBlood smears for malaria at baseline and whenever febrile. Treatment given for all clinical cases.
Sazawal 2006 (C)aFever > 38 °C and parasitaemia>1000 OR history of fever and parasitaemia > 3000 or parasitaemia >10,000 parasites/mm3 regardless of fever (mostly P. falciparum)Not assessedMalaria-related adverse events, defined as hospital admission or death due to malaria (used primarily as 'clinical malaria'). RRs with 95% CI adjusted for multiple events per child and clustering; cerebral malaria (used as 'severe malaria')Not fixed. End of Tx about 1 yr and end of FU about 18 mNo baseline or routine surveillance for malaria during the trial. Treatment given only if admitted to the hospital and malaria diagnosed
Sazawal 2006 (C)bFever > 38 °C and parasitaemia > 1000 OR history of fever and parasitaemia > 3000 or parasitaemia > 10,000 parasites/mm3 regardless of fever (mostly P. falciparum)Not assessedMalaria-related adverse events, defined as hospital admission or death due to malaria (used primarily as 'clinical malaria'). RRs with 95% CI adjusted for multiple events per child and clustering; cerebral malaria (used as 'severe malaria')Not fixed. End of Tx about 1 yr and end of FU about 18 mBlood smear for malaria at baseline, 6 and 12 months. Sulfadoxine/ pyrimethamine treatment delivered to home to all slide-confirmed malaria cases or clinical disease presenting during the study
Smith 1989 (C)Fever > 37.5> 500 parasites/mm3 (mostly P. falciparum)Visits for clinical malaria; parasitaemia > 500; fever with parasitaemia > 5000 parasites/mm3 (used as 'severe malaria' (all N events/N individuals, unadjusted for clustering)3 m - end of TxBlood smear for malaria at baseline, 2 weeks and end of treatment. No treatment at baseline; clinical malaria referred to local healthcare services
van den Hombergh 1996Not usedAny parasitaemia (P. falciparum)Parasitaemia; parasite density3 m - end of Tx (all children treated for malaria)All children smear positive at baseline per study design and treated (trial's intervention). Repeat blood smears at 2, 4, 8, 12 weeks and recurrent clinical malaria re-treated
van Hensbroek 1995Fever or history of fever in the past 48 hours with parasitaemia or parasitaemia > 5000 parasites/mm3 regardless of feverAny parasitaemia (P. falciparum)Parasitological failure1 m - end of Tx (all children treated for malaria)All children smear positive at baseline per study design and treated (trial's intervention). Repeat blood smears at 1, 4 weeks and clinical failures treated with quinine
Verhoef 2002Axillary temperature >= 37.5 °CDipstick test for P. falciparumNumber of children with malaria infection (used primarily as 'clinical malaria')3 m - end of TxDipstick for P. falciparum tested at baseline, 4, 8 and 12 weeks. Confirmed with blood smear if febrile and treated with sulfadoxine/ pyrimethamine, amodiaquine or halofantrine

Trials reporting on malaria were conducted mostly in holoendemic settings or in a malaria-endemic region during the rainy season, in countries where malaria does not occur perennially (12 trials, Smith 1989 (C); Latham 1990; Gebresellassie 1996; Adam 1997 (C); Lawless 1994; Berger 2000; Verhoef 2002; Massaga 2003; Mebrahtu 2004 (C); Sazawal 2006 (C)a; Sazawal 2006 (C)b; Leenstra 2009). Three trials were conducted in hyperendemic (Menendez 1997; Harvey 1989; Desai 2003) and two in mesoendemic countries (Fahmida 2007; Richard 2006) with no specification of trial dates/season. Five were cluster-randomized trials (Smith 1989 (C); Adam 1997 (C); Mebrahtu 2004 (C); Sazawal 2006 (C)a; Sazawal 2006 (C)b; and included the bulk of patients in this comparison (19,313 individuals in cluster-RCTs and 4688 in individually randomized trials). Two cluster-RCTs reported results per individuals only (Adam 1997 (C); Smith 1989 (C)), one reported ORs with 95% CIs adjusted for repeated measurements in children but not for clustering (Mebrahtu 2004 (C)) and both Sazawal 2006 (C)a and its substudy (Sazawal 2006 (C)b, unpublished data) reported RRs with 95% CIs adjusted for clustering. Baseline haemoglobin was reported in all but two trials (Menendez 1997 and Sazawal 2006 (C)a main study) and none of the trials stratified randomization by anaemia status at baseline. The median treatment duration was 3.9 months (range 2 to 12).

Clinical malaria

Clinical malaria was defined in the studies as fever (usually > 37.5 °C) and parasitaemia.

Individually-randomized trials: trials comparing iron versus placebo yielded a RR of 1.01 (95% CI 0.90, 1.12; 2725 children, eight trials), with a control event rate of 28.1% (unadjusted). There was no significant heterogeneity in this comparison (P = 0.63, I2 = 0%). Trials comparing iron with an antimalarial drug versus an antimalarial drug alone yielded a RR of 0.83 (95% CI 0.67 to 1.03) (RRs < 1 favour iron). This comparison included four trials with 1221 children, a control event rate of 22.7%, and was significantly heterogenous (P = 0.07, I2 = 58%). Pooling all individually randomized trials resulted in a RR of 0.96 (95% CI 0.87 to 1.06; 3946 participants, nine trials), without statistically significant heterogeneity (Analysis 1.2). This analysis does not include one individually randomized trial (Leenstra 2009) that reported only incidence rate ratios with 95% CIs adjusted for confounding by school, without the number of events/evaluated children.

All trials: a compilation including all trials reporting on clinical malaria (including the cluster-RCTs and Leenstra 2009 where reported rate ratios were used as risk ratios) is presented in Analysis 1.3. Results were mostly unchanged for the comparisons of iron versus placebo/no treatment (RR 1.03, 95% CI 0.92 to 1.15; 3928 children, 11 trials), with the cluster-RCTs contributing 7.2% (Adam 1997 (C) and 1.2% (Smith 1989 (C)) of the weight in the meta-analysis for this comparison (compared to their unadjusted weights of 10.9% and 1.9%, respectively). No new trials were added to the comparison of iron + antimalarial versus antimalarial. Sazawal 2006 (C)a and Sazawal 2006 (C)b were the only trials comparing iron + folic acid versus placebo (vitamin A added to both arms). Their results were contradictory, as viewed in Analysis 1.3: Sazawal 2006 (C)a showed a borderline significant advantage to placebo (RR 1.16, 95% CI 1.00 to 1.34), while Sazawal 2006 (C)b showed a significant advantage to iron + folic acid (RR 0.26, 95% CI 0.09 to 0.81).

Overall, there was no statistically significant difference between iron versus placebo (RR 1.02, 95% CI 0.94 to 1.11, fixed-effect model; 22,724 individuals, 14 trials, 17 comparisons), with statistically significant heterogeneity (P = 0.02, I2 = 46%). The random-effects RR was 1.00 (95% CI 0.88 to 1.13) and the 95% CIs for malaria were -0.41 to 2.41 (Tau2 = 0.03).

Subgroup analyses

In 10 trials more than 50% of children were anaemic at baseline (100% of children in five trials). In the other four trials, most children did not have anaemia at baseline. No significant differences between iron and placebo were seen both in the trials recruiting children with anaemia ("treatment" RR 1.01, 95% CI 0.79 to 1.30, Analysis 1.4, random-effects model with significant heterogeneity) and trials recruiting non-anaemic children ("prevention" RR 0.96, 95% CI 0.85 to 1.09, Analysis 1.5, without heterogeneity). The lower RRs (favouring iron) among non-anaemic children point against the possibility that iron supplementation adversely affects iron-replete children. Six of the trials assessing iron alone included only or mostly children < five years (RR 1.01, 95% CI 0.86 to 1.18), as did all the trials assessing iron + antimalarials, and Sazawal 2006 (C)a (combined RR 1.01, 95% CI 0.86 to 1.19, random-effects model due to significant heterogeneity, I2 = 50%). The other five trials assessing iron alone included mainly children > five years (RR 1.03, 95% CI 0.90 to 1.19, without heterogeneity, I2 = 0%). All trials were conducted in hyperendemic or holoendemic countries, except for two trials in the comparison of iron versus placebo/no treatment (Richard 2006; Fahmida 2007). Their exclusion did not affect results (overall RR 1.03, 95% CI 0.95 to 1.12). A post-hoc analysis was conducted based on the facilities available within the trial for diagnosis and treatment of malaria. Trials in which regular surveillance for malaria at baseline and during iron supplementation was performed, and treatment for malaria was provided to children with malaria as part of the trial's design, showed no significant difference with iron supplementation, with heterogeneity (RR 0.93, 95% CI 0.84 to 1.04, I2 = 45%; eight trials, Analysis 1.8). Trials in which no surveillance was performed or treatment was not provided outside the local healthcare system showed a significantly increased rate of clinical malaria with iron supplementation, without heterogeneity (RR 1.16, 95% CI 1.03 to 1.31, I2 = 0%; six trials, Analysis 1.9).

By meta-regression none of the associations between effect-estimates and baseline haemoglobin, treatment duration, or baseline parasitaemia rates were statistically significant. The effect of iron dose on the relative risks approached statistical significance with higher doses associated with RRs favouring iron (0.96, 95% CI 0.94 to 1.00) and a decrease in RR for every 1 mg increase in elemental iron (14 trials, P = 0.057).

Sensitivity analyses

Three trials reported on episodes (Richard 2006; Leenstra 2009) or visits (Smith 1989 (C)) of clinical malaria rather than patients with first or only episode. Leenstra 2009 and Richard 2006 also administered vitamin A to both study arms. These trials were included in the comparison of iron versus control; their exclusion did not affect the pooled RR for this comparison (RR 1.03, 95% CI 0.92 to 1.15) with RR 1.00 (95% CI 0.88 to 1.13) without these three trials. One trial reported on children with clinical malaria only at end of follow up, six months after completing iron supplementation (Menendez 1997). Its exclusion did not affect the results.

All the trials included in this analysis were double blinded. The RRs for trials reporting adequate allocation concealment (Analysis 1.6) were similar to those that did not report or describe concealment methods clearly (Analysis 1.7). In the comparison of iron + antimalarial versus antimalarial one trial with unclear allocation concealment (Desai 2003) was associated with a low RR (0.59, 95% CI 0.40 to 0.86) compared to the other trials with adequate concealment (RR 1.01, 95% CI 0.77 to 1.32, three trials). Similar trends were observed for adequate versus unclear generation of the randomization sequence. The funnel plot was asymmetrical, indicating that small studies favouring iron could be missing (Figure 3).

Figure 3.

Funnel plot of comparison: 1 Iron +/- folic acid vs. placebo or no treatment for prevention or treatment of anaemia, outcome: 1.3 Clinical malaria (all trials).

In summary, for the outcome of clinical malaria, no significant difference was observed between iron versus placebo. The pooled random-effects RR was 1.00 (95% CI 0.88 to 1.13), based on 14 trials. Statistically significant heterogeneity could not be explained by different effects in trials administering iron for prevention of anaemia and those administering iron for treatment of anaemia. Rather, trials including mainly non-anaemic children had lower RRs (favouring iron) than trials including mostly anaemic children. Two additional secondary analyses pointed against a harmful effect of iron on clinical malaria: higher doses of iron were protective rather than harmful and funnel plot asymmetry indicated that, if at all, missing studies favoured iron. The largest trial contributing 30% of the weight for this meta-analysis showed surprising results: the main study including 15,956 children showed a borderline disadvantage to iron (event rate of 5.9% with iron + folic acid versus 5.1% with placebo, Sazawal 2006 (C)a), while a separate substudy including 1619 children showed a significant advantage to iron (event rate of 3.7% with iron + folic acid versus 1.7% with placebo, Sazawal 2006 (C)b). In the substudy, children with severe anaemia at baseline were excluded (Hb < 7 g/dL) , malaria parasitaemia was monitored at baseline and during the trial, and children with parasitaemia or other clinical symptoms of malaria or other intercurrent infections were provided with free treatment physically delivered to their homes. The intervention and all other trial characteristics were similar for the main and substudy. A consequent post-hoc analysis showed similar distribution of results in the other trials, with effect estimates favouring iron with close surveillance and treatment of malaria and an increased risk for malaria with iron in trials that did not perform surveillance and did not provide children with treatment for malaria.

Parasitaemia

Individually-randomized trials: in the individually-randomized trials reporting specifically on parasitaemia at end of treatment there was no significant difference between iron and placebo (RR 1.04, 95% CI 0.88 to 1.22; 1365 children, four trials, without heterogeneity (I2 = 0%), Analysis 1.10). This analysis does not include one individually-randomized trial that reported incidence rate ratios (Leenstra 2009).

All trials: the compilation of all trials, including three cluster-RCTs (Smith 1989 (C); Adam 1997 (C); Mebrahtu 2004 (C); and Leenstra 2009) is shown in Analysis 1.11. Seven trials compared iron versus placebo and showed a statistically significant advantage to placebo (RR 1.15, 95% CI 1.02 to 1.29; 2760 participants, seven trials, without heterogeneity (I2 = 0%)). One trial compared iron + antimalarial versus antimalarial (Desai 2003). The overall result was a small advantage to placebo (RR 1.13, 95% CI 1.01 to 1.26; 3184 children, eight trials), without heterogeneity (I2 = 0%). Four additional trials reported on 'malaria' without defining the outcome (Lawless 1994; Verhoef 2002; Richard 2006; Fahmida 2007). These trials were included primarily in the meta-analysis for clinical malaria, since child assessment in the trial included both clinical and parasitological follow up. When including these trials in the meta-analysis for parasitaemia, the advantage to placebo was no longer statistically significant (RR 1.09, 95% CI 0.99 to 1.19; 4584 children, 12 trials, Analysis 1.12).

Three trials reported on parasitaemia also at end of follow up (Harvey 1989; Gebresellassie 1996; Berger 2000) and one trial reported only end of follow up results (Latham 1990). The end of follow up ranged between two to six months after end of treatment. All four trials were individually randomized and assessed iron versus placebo. Their compilation demonstrated an advantage to the control arm (RR 1.18, 95% CI 1.03 to 1.35, with some heterogeneity (P = 0.24, I2 = 29%), Analysis 1.13). Analysis by the random-effects model gave a RR of 1.17 (95% CI 1.00 to 1.37; 941 participants, four trials), with a dispersion range of 0.96 to 1.43 (Tau2 = 0.01).

Subgroup analyses

Subgroup and sensitivity analyses were performed in Analysis 1.12, that included most trials (overall RR 1.08, 95% CI 0.98 to 1.19). Two trials included mostly non-anaemic children ("prevention" RR 1.11, 95% CI 0.93 to 1.32). All other trials recruited mostly children with anaemia ("treatment" RR 1.08, 95% CI 0.97 to 1.20). In seven trials that included mostly children < five years, the RR was 1.13 (95% CI 0.98 to 1.30) compared to RR 1.06 (95% CI 0.94 to 1.20) for children > five years (five trials). Restricting the analysis to the subgroup of trials conducted in hyper/holoendemic countries (excluding Fahmida 2007 and Richard 2006) did not change the result (RR 1.10, 95% CI 0.99 to 1.22).

By meta-regression none of the associations between effect-estimates and baseline haemoglobin, treatment duration, iron dose, or baseline parasitaemia rates were statistically significant.

Sensitivity analyses

Most trials reporting an advantage to placebo had unclear or inadequate allocation concealment methods, and their pooled RR was 1.17 (95% CI 1.03 to 1.32; six trials, Analysis 1.15). In trials reporting adequate allocation concealment methods the pooled RR was 1.01 (95% CI 0.88 to 1.15, six trials, Analysis 1.14). Three of the four trials included in the analysis at end of follow up had no description of the allocation concealment. All the trials included in the comparison of parasitaemia at end of treatment were double blinded. One trial reporting on parasitaemia at end of follow up only was open and its exclusion from the analysis at end of follow up (Analysis 1.13) widened the pooled confidence intervals without altering the RR (1.16, 95% CI 0.93 to 1.43).

It was difficult to establish whether the trials reported on children with parasitaemia or parasitaemia episodes. Two trials clearly reported on 'cumulative incidence' (Gebresellassie 1996) or included repeated episodes (Leenstra 2009). Without these two trials, the pooled RR was 1.08 (95% CI 0.98 to 1.19).

Leenstra 2009 reported incidence rate ratios with 95% CIs adjusted for age, baseline parasitaemia, and school. These were used in our analysis as relative risks. Exclusion of Leenstra 2009 lowered RRs and results were no longer statistically significant: RR 1.11 (95% CI 0.98 to 1.25) for any parasitaemia and RR 1.07 (95% CI 0.98 to 1.18) for any parasitaemia or 'malaria'. One cluster-RCT (Mebrahtu 2004 (C)) reported only ORs adjusted for repeated measurements. For this trial, the RR was calculated from the OR using the baseline parasitaemia rate reported in the publication as the assumed control risk, and the SE(OR) were used and adjusted for clustering, assuming that the dispersion surrounding ORs is similar to that surrounding RRs. Analyses were repeated using ORs, to enable the use of the data as reported in Mebrahtu 2004 (C). In the OR analyses Leenstra 2009 is not included since ORs could not be calculated from the given data. Both for parasitaemia at end of treatment (OR 1.09, 95% CI 0.93 to 1.28) and for parasitaemia or 'malaria' (OR 1.08, 95% CI 0.94 to 1.24), there was no significant difference between iron versus placebo.

In summary, for the outcome of parasitaemia there was a trend in favour of placebo at end of treatment (RR 1.13, 95% CI 1.01 to 1.26; 3184 participants, eight trials) and end of follow up (RR 1.17, 95% CI 1.00 to 1.37; 941 participants, four trials, Analysis 1.13). The only variables found to affect these effect estimates related to the methodological quality of the trials and the methods of results' reporting (patients or episodes). Exclusion of trials that did not report number of events/assessed clearly or those with unclear methods of allocation concealment shifted the pooled relative risk towards no effect. We found no association between clinical variables such as baseline haemoglobin status, baseline parasitaemia, or children's age on the effect of iron on malaria parasitaemia.

Severe malaria

Four trials reported severe malaria using varying definitions: cerebral malaria (Sazawal 2006 (C)a and Sazawal 2006 (C)b), clinical episodes of malaria associated with parasitaemia > 5000 parasites/µl (Massaga 2003), and malaria that necessitated hospitalization (Adam 1997 (C)). Analysis 1.16 presents separate trial results. Two trials favoured iron (Sazawal 2006 (C)b and Massaga 2003 in the comparison of iron versus placebo), Sazawal 2006 (C)a significantly favoured placebo, and there was no significant difference in Adam 1997 (C) and the comparison of iron + antimalarial versus antimalarial drug in Massaga 2003.

Five trials comparing iron versus placebo reported on high-grade parasitaemia (Analysis 1.17) using varying parasite density thresholds (Table 5). There was an advantage to placebo, without statistical significance (RR 1.15, 95% CI 0.93 to 1.43), without heterogeneity (I2 = 0%). For two trials reporting adequate allocation concealment the RR was 1.08 (95% CI 0.84 to 1.38) and for the three with unclear/inadequate methods the RR was 1.45 (95% CI 0.93 to 2.28).

Parasite density

Parasite density was reported differently in the studies, with differences referring both to the unit of measurement and the denominator (Table 6). Meta-analysis was therefore not possible and results are shown in Table 6 for each study. Qualitatively, parasite density was higher in the iron supplemented group in four, lower in one and similar in one of the six trials that reported on parasite density at end of treatment.

Table 6. Comparative malaria parasitaemia rates
  1. RBC = red blood cell
    WBC = white blood cell

Study IDInterventionUnit of measurementIronControlNo. iron

No.

Control

Favours
For prevention or treatment of anaemia
Adam 1997 (C)Iron vs. placeboGeometric mean, parasites/ microL15,0598225368 slides372 slidesControl
Berger 2000Iron vs. placeboGeometric mean, RBC/mm361.225.749 children with malarial index39 children with malarial indexControl
Desai 2003

Iron + antimalaria vs. antimalaria

Iron vs. placebo (with single-dose antimalarial Tx)

Geometric mean, parasites/mm3

1705

2569

2485

3778

129 children

127 children

127 children

108 children

Iron
Gebresellassie 1996Iron vs. placeboAverage parasite density class (parasite density classified in ascending order from 1 to 10)5.25.0239 children241 childrenControl/similar
Latham 1990Iron vs. placeboGeometric mean, infected RBCs/100 WBC4.81.928 children26 childrenControl
Mebrahtu 2004 (C)Iron vs. placeboGeometric mean, parasites/microL (counting against 200 to 500 WBC, assuming 8000 WBC/microL

Age < 30 m 3402

Age > 30 m 2188

Age < 30 m 3422

Age > 30 m 2046

273 children (225 households)265 children (225 households)Similar
For treatment of malaria
Nwanyanwu 1996Iron daily + antimalarial vs. iron weekly + antimalarial vs. antimalarialMean, parasites/microL (counting against 300 WBC, assuming 6000 WBC/microL

4927 (daily)

2207 (weekly)

1812

77 (daily)

63 (weekly)

children

75 childrenControl
van den Hombergh 1996Iron + antimalarial + folic acid vs. antimalarial + folic acidGeometric mean, parasites/microL5308930248 children47 childrenIron (at baseline groups unbalanced favouring placebo)
Deaths

The data on mortality at end of therapy or end of follow up were sparse. Mortality data were available in 25/65 trials and in 15 of them no deaths were reported among all children in the study. The unadjusted control event rate among studies conducted in hypo/mesoendemic areas was 0.3% (eight studies) and in hyper/holoendemic settings was 1.01% (16 studies). Thus, the meta-analysis for this outcome relies only on 10 trials, where at least one death was reported, and a low event rate. The median follow up for mortality in these trials was 3.75 months (range 1 to 12 months). Three trials contributing to the meta-analysis (both in holoendemic settings) were cluster-randomized (Mebrahtu 2004 (C); Sazawal 2006 (C)a; Sazawal 2006 (C)b). To adjust for clustering we divided the reported number of events and children evaluated by the design effect for mortality (Table 4). Results are reported including all trials together.

The comparison of iron alone versus placebo favoured iron (RR 0.87, 95% CI 0.48 to 1.57; 8209 participants, 19 trials), while that of iron with antimalarial versus antimalarial alone favoured antimalarial alone (RR 1.30, 95% CI 0.67 to 2.50; 1261 participants, five trials). Only Sazawal 2006 (C)a and its substudy Sazawal 2006 (C)b contributed to the comparison of iron + folic acid versus folic acid (RR 1.13, 95% CI 0.90 to 1.42). All differences were without statistical significance, as was the pooled RR for all trials (1.11, 95% CI 0.91 to 1.36, without heterogeneity, Analysis 1.18).

Three trials were conducted in hypo/mesoendemic areas and the pooled RR for mortality was 0.50 (95% CI 0.12 to 1.98; 1717 participants, Analysis 1.19). Six trials were conducted in hyper/holoendemic areas, or in settings of intense malaria transmission, and their pooled RR was 1.13 (95% CI 0.92 to 1.39), in agreement with the largest trial, Sazawal 2006 (C)a. Excluding Sazawal 2006 (C)a which dominated this comparison, yielded a pooled RR of 0.98 (95% CI 0.65 to 1.47) overall and a RR of 1.05 (95% CI 0.68 to 1.61) in hyper/holoendemic areas. Including two trials that compared iron + antimalarial versus antimalarial for proven malaria did not change the results for this comparison (pooled RR 1.11, 95% CI 0.91 to 1.36, without heterogeneity, Analysis 1.20).

No trends were observed when separating trials by allocation concealment methods (Analysis 1.21). All trials contributing to this meta-analysis were double blinded. Three trials did not report results clearly by intention-to-treat. Their exclusion yielded an overall RR of 0.97 (95% CI 0.61 to 1.53) for mortality, RR 0.50 (95% CI 0.12 to 1.98) in hypo/mesoendemic areas, and RR 1.06 (95% CI 0.65 to 1.73) in hyper/holoendemic areas. Given the unexpected rate of studies reporting no deaths, we conducted a post-hoc sensitivity analysis including these trials in the analysis, imputing 1 for null values in studies with no outcome (Analysis 1.22). Results did not change overall (RR 1.10, 95% CI 0.90 to 1.33; 26,329 children, 25 trials, I2 = 0%, imputing 1 for no deaths, and RR 1.11 (95% CI 0.91 to 1.36) in the 10 trials and 20,903 children reporting at least one death. Due to the paucity of data, no further subgroup or sensitivity analyses were conducted.

In summary, only few trials provided data regarding deaths following iron supplementation. Based on 25/65 trials comparing iron given for prevention or treatment of anaemia, at a median follow up of about four months, there is no significant difference in mortality between iron versus placebo, RR 1.11 (95% CI 0.91 to 1.36).

Haemoglobin and anaemia

Haemoglobin was reported comparatively as absolute values at end of treatment in 41 comparisons (Analysis 1.23); absolute values at end of follow up in 16 comparisons (Analysis 1.26); change from baseline at end of treatment in seven comparisons (Analysis 1.27); and change from baseline at end of follow up in seven comparisons (Analysis 1.28). End of follow up in these trials varied between two weeks and six months after end of treatment. Three trials reporting mean haemoglobin with no dispersion measure (Greisen 1986 (C); Zavaleta 2000; Majumdar 2003) and one trial reporting t value, df, and P value for the difference in means (Smith 1989 (C)) are not included in these analyses. Anaemia as a dichotomous outcome was reported at end of treatment in 28 comparisons (Analysis 1.29) and at end of follow up in 10 (Analysis 1.30). The haemoglobin threshold defining 'anaemia' differed in the primary studies, varying between 7 to 12 g/dL. The most commonly used threshold was Hb < 11 g/dL, and when more than one threshold for anaemia was reported, we used 11 g/dL for uniformity. Significant heterogeneity was observed in all the comparisons for haemoglobin and anaemia, thus the random-effects model is reported in addition to the fixed-effect model for all comparisons and the reasons for heterogeneity were investigated. Analyses are presented including the cluster-RCTs, adjusted by calculation of the effective sample size (Table 4).

Overall, iron treatment increased haemoglobin at end of treatment by 0.87 g/dL (95% CI 0.82 to 0.91; 9482 participants, 41 trials, fixed-effect model), with an inconsistency measure for this meta-analysis of 95% (P for heterogeneity < 0.00001). The random-effects model MD was 0.90 g/dL (95% CI 0.68 to 1.11) and the 95% CIs for haemoglobin were -0.67 to 2.36 (Tau2 = 0.62). Heterogeneity was not explained by a difference between hypo/mesoendemic versus hyper/holoendemic settings or the co-administration of folic acid (Analysis 1.23); the random-effects haemoglobin MD was 0.86 (95% CI 0.56 to 1.15) in hypo/mesoendemic settings versus 0.99 (95% CI 0.59 to 1.38) in hyper/holoendemic settings, both without folic acid, and 0.99 (95% CI 0.34 to 1.64) versus 0.80 (95% CI 0.54 to 1.06), with folic acid, respectively.

Further variables that might affect the effect of iron were assessed, including haemoglobin and anaemia status at baseline, iron dose, duration of treatment, co-administration of zinc, eradication of intestinal parasites before or during the trial, co-administration of antimalaria treatment, children's age and trial methods, including cluster-randomization. Of all these variables only co-administration of zinc or antiparasite treatment and baseline haemoglobin status significantly affected the combined results (Table 7). Significant heterogeneity persisted in all subgroup analyses, including explorative post-hoc analyses combining different variables (eg dose*duration, co-administration of zinc among anaemic children, etc). As expected, the effect of iron was more pronounced among children with anaemia at baseline (MD +1.13 (95% CI 0.80 to 1.45; 4767 participants, 28 trials) than for non-anaemic children (MD +0.60, 95% CI 0.39 to 0.80; 4715 participants, 19 trials, Analysis 1.24). The respective meta-regression showed a decrease of the MD between iron and placebo of 0.23 (95% CI - 0.38 to -0.08) for every 1 g/dL unit increase in the placebo group's baseline haemoglobin (59 comparisons, P = 0.003, random-effects regression). Interestingly, the direction of the effect of treatment duration was negative, with longer treatment durations associated with smaller effect estimates, but this regression was not significant when the trials were pooled using the random-effects model. Iron dose (elemental iron equivalent in mg/kd/day) was positively correlated with effect estimates, again without statistical significance. Co-administration of zinc and antiparasite treatment reduced effect estimates: MD in haemoglobin of +1.00 (95% CI 0.74 to 1.25; 6580 participants, 36 trials) g/dL without both, reduced to +0.47 (95% CI 0.14 to 0.81; 2042 participants, eight trials) when zinc was co-administered to both trial arms and to +0.62 (95% CI 0.31 to 0.92; 860 participants, five trials) with antiparasite treatment (Analysis 1.25). The effects of iron were smaller in adequately concealed trials compared to those with unclear methods, and in double blinded compared to other trials, without a statistically significant difference between these subgroups (Table 7).

Table 7. Sensitivity analyses: haemoglobin at end of treatment
  1. 1Differences in means (95% confidence intervals) using the random-effects model and number of study arms included in the subgroup.
    2P value for the heterogeneity between subgroups shown. Analyses were performed using Comprehensive Meta-analysis.

Variable assessedWith variable1Without variable1Difference between subgroups2
Hyper- or holoendemic regions for malaria0.947 (0.620-1.273), 160.880 (0.616-1.144, 430.756
Age < 5 years0.831 (0.608-1.054), 371.012 (0.603-1.421), 220.447
Co-administration of folic acid with iron0.978 (0.631-1.326), 120.872 (0.630-1.115), 470.624
Co-administration of zinc with iron0.486 (0.161-0.811), 80.961 (0.729-1.194), 510.020
Co-administration of an antiparasitic drug with iron0.576 (0.292-0.860), 70.938 (0.707-1.169), 520.052
Anaemia at baseline, defined as Hb in placebo group < 11 g/dL1.126 (0.805-1.447), 330.598 (0.398-0.798), 260.006
Treatment of anaemia, defined as all children in study with anaemia (as defined in study)1.914 (1.157-2.672), 90.713 (0.543-0.883), 500.002
Iron treatment duration > 6 months0.753 (0.510-0.995), 240.994 (0.684-1.304), 350.228
Cluster RCTs0.922 (0.491-1.352), 80.892 (0.655-1.130), 510.907
Adequate allocation concealment0.716 (0.460-0.972), 261.044 (0.738-1.350), 330.107
Adequate generation of the randomization sequence0.939 (0.683-1.195), 240.863 (0.562-1.164), 350.707
Double blinding0.861 (0.622-1.100), 481.060 (0.738-1.381), 110.330

At end of follow up, the effects of iron supplementation were smaller compared to those obtained at end of treatment in malaria hyper/holoendemic settings. Compared to an increase of more than 1 g/dL haemoglobin at end of treatment, the difference at end of follow up was 0.33 (95% CI 0.25 to 0.40) using the fixed-effect model, with significant heterogeneity, and 0.46 (95% CI 0.24 to 0.67), 95% CI for haemoglobin dispersion -0.14 to 1.06, Tau2 = 0.09, using the random-effects model (Analysis 1.26). Few trials conducted in hypo/mesoendemic settings reported end of follow up results, where the increase obtained at end of treatment was maintained at end of follow up (Analysis 1.26).

The respective analyses of the change from baseline in haemoglobin values at end of treatment and end of follow up comprised fewer studies. Overall, results agreed with the analysis of end values, with similar heterogeneity. The MD in the change in haemoglobin was 0.91 (95% CI 0.56 to 1.26; 1610 participants, eight trials) at end of treatment (Analysis 1.27) and 0.41 (95% CI -0.01 to 0.84; 2293 participants, four trials) at end of follow up (Analysis 1.28) for hyper/holoendemic settings without folic acid (random-effects model). In an analysis comparing the same set of studies that reported both end and change from baseline values, the MDs observed were 0.72 (95% CI 0.48 to 0.94) for end values and 0.66 (95% CI 0.43 to 0.89) for change from baseline values (random-effects model), 18 studies, with no significant difference (analysis not shown).

Iron supplementation reduced the prevalence of anaemia significantly in most individual studies. The meta-analyses for anaemia were as heterogenous as those for haemoglobin. The pooled RRs were similar in malaria hyper/holoendemic areas and hypo/mesoendemic areas at end treatment (Analysis 1.29). In hyper/holoendemic setting the fixed-effect RR was 0.69 (95% CI 0.66 to 0.73) and the random-effects RR was 0.49 (95% CI 0.31 to 0.78; 3660 participants, 14 trials), with 95% CIs for the relative risk for anaemia of 0.09 to 2.58, Tau2 = 0.69. In hypo/mesoendemic settings the fixed-effect RR was 0.67 (95% CI 0.61 to 0.73) and the random-effects RR 0.56 (95% CI 0.42 to 0.76; 2668 participants, 14 trials), with relative risks for anaemia between 0.22 to 1.43, Tau2 = 0.22. At end of follow up the relative decrease in the prevalence of anaemia was smaller in hyper/holoendemic settings compared to that observed at end of treatment, random-effects RR 0.73 (95% CI 0.59 to 0.88; 2213 participants, eight trials, 95% CI for dispersion 0.47 to 1.14, Tau2 = 0.05). Assuming similar reasons for heterogeneity as for haemoglobin, further subgroup and sensitivity analyses were not pursued.

In summary, most individual studies have shown that iron treatment increases haemoglobin and decreases the risk of anaemia. The studies varied in the absolute effect estimate and all analyses were highly heterogenous. Thus, the pooled estimates of effect shown may be imprecise. Many reasons for heterogeneity exist, only some of which were examined. Variables underlying heterogeneity that were identified included baseline children's haemoglobin and co-administration of zinc or antiparasitics. The effects of iron were similar in areas where malaria is highly prevalent and areas with lower transmission rates. During follow up, after cessation of treatment, the gain afforded by iron decreased in hyper/holoendemic settings for malaria, but a benefit was still apparent.

Other outcomes

Eleven studies provided data on respiratory infections: three reported on upper respiratory infections (de Silva 2003; Nagpal 2004; Aggarwal 2005), three on lower respiratory infections (Berger 2000; Lind 2004; Berger 2006), one on pneumonia (Fahmida 2007), and four trials did not define the type of respiratory infection (Angeles 1993; Rosado 1997; Baqui 2003; Richard 2006). All were individually randomized. There was no difference between iron and placebo overall (rate ratio 0.97, 95% CI 0.91 to 1.04; 22,577 child-months, 11 trials, I2 = 0%), or in the subgroups of children in hypo/mesoendemic versus hyper/holoendemic areas and with or without zinc co-administration (Analysis 1.31).

Diarrhoeal episodes were reported altogether in 14 studies, including two cluster-RCTs (Adam 1997 (C); Sazawal 2006 (C)a). Diarrhoea was usually defined as three or more loose stools per day with a symptom-free interval of one to three days separating individual episodes. Although this outcome was usually reported as 'infectious diarrhoea', the symptoms could not have been well-differentiated from diarrhoea related to iron or iron/zinc supplementation and these outcomes were not reported separately. Overall, there was a small higher risk of diarrhoea with iron (rate ratio 1.08, 95% CI 1.02 to 1.14, I2 = 39%, Analysis 1.32). However, this disadvantage was limited to the subgroup of studies where zinc was co-administered with iron (rate ratio 1.15, 95% 1.07 to 1.24, I2 = 63%) for iron + zinc versus zinc, random-effects 1.13 (95% CI 0.98 to 1.29; 11,328 child-months, six trials). Without zinc there was no significant difference between iron and placebo in hypo/mesoendemic areas (rate ratio 1.03, 95% CI 0.95 to 1.12; 8254 child-months, seven trials) and in hyper/holoendemic settings (rate ratio 0.92, 95% CI 0.77 to 1.10; 199,625 child-months, six trials), both with no significant heterogeneity.

Eight trials reported on febrile episodes separately from respiratory and diarrhoeal episodes; three trials reported on 'other disease episodes' and two trials reported results for all infectious episodes combined (Sazawal 2006 (C)a; Leenstra 2009). Definitions and reporting methods were highly variable. Results are shown per outcome (Analysis 1.33). We limited the analysis to the occurrence of fever in hyper/holoendemic areas for malaria, where febrile episodes are likely to reflect malaria. There was no significant difference between iron and placebo for the number of febrile episodes (rate ratio 1.05, 95% CI 0.94 to 1.17, I2 = 3%; 13,799 child-months, five trials from Analysis 1.33 "Fever episodes" excluding Fahmida 2007), but Dossa 2001b reported significantly more fever days with iron (15 versus two days) in a small study.

All five trials reporting on the need for hospitalization or the number of clinic visits were conducted in malaria hyper/ holoendemic areas. Rates of hospitalization and clinic visits were not significantly different for iron versus placebo (Analysis 1.34).

Comparative children's weight and height after treatment were reported in 30 and 26 trials, respectively, as end values or change from baseline. Results were inconsistently reported as absolute values or Z scores matched for age or height/weight. The analyses shown are based on absolute weight in kg and height in cm, but for three trials that reported only weight and height for age Z scores (Berger 1997; Richard 2006 and Fahmida 2007). Overall, there was no difference in children's weight (SMD at end of treatment 0.02 kg, 95% CI -0.05 to 0.09; 6334 participants, 18 trials, I2 = 40%, Analysis 1.35) and the SMD in the mean change from baseline was +0.21 (95% CI -0.05 to 0.46, 1688 participants, 12 trials) with significant heterogeneity, I2 = 84% (Analysis 1.36). At end of treatment, there was a difference between studies conducted in hypo/mesoendemic settings that tended in favour of iron (SMD +0.05, 95% CI -0.04 to 0.13 I2 = 45%) and studies conducted in hyper/holoendemic settings that tended in favour of placebo (SMD -0.08, 95% CI -0.17 to 0.01, I2 = 0%). This was not reflected in different studies that reported on change from baseline. There were no significant differences in height at end of treatment (Analysis 1.37) or change from baseline (Analysis 1.38). In hypo/mesoendemic settings the SMD was 0.02 (95% CI -0.04 to 0.07; 4746 participants, 14 trials) for end of treatment and +0.22 (95% CI -0.01 to 0.45; 1350 participants, nine trials) for the change from baseline. In hyper/holoendemic settings it was -0.05 (95% CI -0.14 to 0.04; 1895 participants, four trials) and +0.06 (95% CI -0.42 to 0.54; 338 participants, three trials), respectively. Significant heterogeneity was present only in the change from baseline analyses.

In summary, iron treatment did not affect the incidences of pneumonia, diarrhoea (without zinc co-supplementation), hospitalization rates, clinic visits or febrile episodes in malaria-endemic settings. Within the timeframe of these trials, iron was not associated with a nutritional gain. Rather, there was a trend toward lower weight and height at end of treatment with iron in malaria hyper/holoendemic areas.

2. Iron + antimalarial versus placebo for treatment or prevention of anaemia (four trials)

Malaria-related outcomes

Three trials reported on malaria-related outcomes and all were individually randomized. The trials showed uniformly that the intervention was protective for clinical malaria (pooled RR 0.54, 95% CI 0.43, 0.67; 728 children, three trials, with no heterogeneity (I2 = 0%), Analysis 2.1). The degree of protection for the comparison of an antimalarial drug alone versus placebo in these trials was RR 0.53 (95% CI 0.43 to 0.67).

Deaths

There was no difference in the risk of death for the three trials combined (RR 1.05, 95% CI 0.52 to 2.11, Analysis 2.2). The risk ratio for antimalarial alone versus placebo in these trials was 0.92 (95% CI 0.45 to 1.91).

Haemoglobin and anaemia

Haemoglobin at end of treatment was reported in one trial and was significantly improved with iron + antimalarial drug (MD 0.91, 95% CI 0.47 to 1.35; 151 participants), but not with an antimalarial drug alone (Analysis 2.3). The addition of iron to an antimalarial drug resulted in a reduction in the number of anaemic children (RR 0.44, 95% CI 0.33 to 0.60; 295 participants, two trials, Analysis 2.4). In one trial, at end of follow up, there was still an improvement in anaemia (RR 0.37, 95% CI 0.26 to 0.54) for iron + antimalarial versus placebo and 0.48 (95% CI 0.34 to 0.67) for antimalarial alone versus placebo (Analysis 2.4).

Respiratory infections, diarrhoea, and other infections were not reported in these studies. In two trials, both the number of hospitalizations and the number of clinic visits was reduced, both with iron + antimalarial and antimalarial drug alone (Analysis 2.5).

3. Iron versus placebo for treatment of proven malaria (three trials)

All three trials included in this comparison (van Hensbroek 1995; Nwanyanwu 1996; van den Hombergh 1996) were individually-randomized, open labelled trials. Generation of the randomization sequence and concealment methods were unclear in all three trials. The same antimalarial treatment was administered to both study arms. The studies were conducted among children who attended the MRC clinic in The Gambia (van Hensbroek 1995), the outpatient and paediatric ward of a hospital in Tanzania (van den Hombergh 1996), and in an outpatient clinic in Malawi (Nwanyanwu 1996).

Malaria-related outcomes

All three trials reported on rates of parasitological failure at end of treatment. Their pooled RR showed no significant difference between iron and placebo (RR 0.96, 95% CI 0.74 to 1.24, 583 patients, three trials, with some heterogeneity (P = 0.21, I2 = 36%). The result for the random-effects model was 0.98 (95% CI 0.69 to 1.39).

Two trials reported on parasite density at end of treatment. One favoured iron and the other favoured placebo (Table 6).

Deaths

All three trials reported on mortality, which was null in one study. No significant difference was noted in the other two (Analysis 3.2).

Results for haemoglobin were reported in all three trials, but in a format that could not be combined. van den Hombergh 1996 reported means with SDs at end of treatment from which a mean difference of +0.36 (95% CI -0.28 to 1.01) g/dL haemoglobin could be calculated, without a statistically significant difference between iron and placebo. van Hensbroek 1995 reported an adjusted difference in mean haemoglobin of +0.7 (95% CI 0.21 to 1.2) at end of treatment and +0.81 (95% CI 0.32 to 1.3) at end of follow up, but that a positive effect of iron on the recovery from severe anaemia (all children with haemoglobin < 5 g/dL at baseline) could not be demonstrated. Nwanyanwu 1996 reported the mean change in haemoglobin at end of treatment without SDs or statistical significance: +4.2 versus +3.5 g /dL for children with baseline haemoglobin < 8 g/dL and +2.2 versus +2.2 for children with baseline haemoglobin > 8 g/dL.

Pneumonia was reported in one study (van den Hombergh 1996) and the risk was significantly higher with iron compared to placebo (rate ratio 5.20, 95% CI 2.05 to 13.16; 288 child-months, Analysis 3.3). Diarrhoea or other infectious complications were not reported. There was no significant difference in the number of hospitalizations in one trial (van Hensbroek 1995), nor in the number of extra attendances for clinical care in another trial (van den Hombergh 1996) (Analysis 3.4). In one study there was no significant difference in the nutritional status at end of treatment: weight for age Z-score difference of -0.12 +/- 0.65 with iron versus +0.02 +/- 0.55 with placebo.

Discussion

Results of a recent, large randomized controlled trial led to modification of the recommended practice guidelines regarding iron supplementation for children living in malaria-endemic countries. The trial showed a significantly higher rate of death or hospitalization due to adverse events, mainly malaria and other infection-related adverse events, with iron supplementation (Sazawal 2006 (C)a). The World Health Organization (WHO) recommended that in malaria-endemic areas iron be given only to iron-deficient children following screening tests for iron status and only where malaria can be adequately prevented and promptly diagnosed and treated (WHO 2007). The practical difficulties of such a strategy in developing countries, where malaria is endemic, are obvious. We therefore set out to compile the full evidence and examine these results in the light of all other studies conducted to date. We aimed to assess primarily the effect of iron supplementation on malaria, for children living in malaria-endemic areas. We defined malaria-related outcomes and all-cause mortality as primary outcomes.

Summary of main results

Overall, iron supplementation did not increase the risk of clinical malaria (RR 1.00, 95% CI 0.88 to 1.13; 22,724 participants, 14 trials, 17 comparisons, random-effects model), with similar risk in studies that included children that were not anaemic at baseline (RR 0.96, 95% CI 0.85 to 1.09; four trials, six comparisons, fixed-effect model). The rate of parasitaemia after treatment was higher among iron supplemented children (RR 1.13, 95% 1.01 to 1.26; 3184 participants, eight trials), but this analysis was sensitive to trials' risk of bias; in trials reporting adequate allocation concealment there was no disadvantage to iron supplementation (RR 1.01, 95% CI 0.88 to 1.15, six trials). Similar results were observed for high-grade parasitaemia in fewer studies, while varying definitions of severe malaria were reported only in three trials, one of which showed a higher rate of cerebral malaria with iron (Sazawal 2006 (C)a). There was no significant difference in all-cause mortality at end of follow up (RR 1.11, 95% CI 0.91 to 1.36, 10 trials, 12 comparisons contributing to this meta-analysis).

The compilation of all trials permitted the investigation of several open debates. We found no evidence to support the recommendation for screening children prior to iron supplementation. There was no increased risk of malaria with iron in trials where children were non-anaemic at baseline. The WHO recommendation is based on a post-hoc, and thus non-randomized, subgroup analysis of the Sazawal 2006 (C)b substudy. The results of the substudy differed significantly from the results of the main trial (Analysis 1.3). Moreover, severely anaemic children (Hb < 7 g/dL) were excluded form the substudy, but included in the main trial. Yet, iron was beneficial in the substudy where children were presumably less anaemic than in the main study, which showed a harmful effect of iron. The only other difference between the main trial and the substudy consisted of the child's care during the trial. The substudy's children were tested for malaria parasitaemia and treated in their homes for malaria or other infectious illnesses. Analysing the data from all trials confirmed a trend whereby iron supplementation may be harmful without organized surveillance and treatment services for malaria. Finally, we can probably conclude that folic acid is not the reason for the increased rate of adverse events in Sazawal 2006 (C)a, since the same intervention in the substudy was protective for malaria.

Iron increased haemoglobin heterogeneously (by about 1 g/dL in malaria hyperendemic settings) and prevented anaemia (by approximately 50%). An interaction was found with zinc, where co-administration of zinc decreased the effect of iron; with haemoglobin status at baseline, where anaemic children gained more than non-anaemic children; and with antiparasite treatment, where co-administration of antiparasitics decreased the effects of iron. Co-administration of antimalarial treatment did not affect results. The effect of iron was similar in locations with high rates of malaria transmission and locations with no or low rates of transmission, although at end of follow up the gain achieved in highly endemic settings was lower than that achieved at end of treatment, while the same gain was maintained in areas with low transmission rates. However, there was a significant advantage to iron supplementation also at end of follow up in both settings.

Iron treatment did not affect the incidence of respiratory infections or pneumonia (rate ratio 0.97, 95% CI 0.91 to 1.04; 22,577 child-months, 11 trials, 16 comparisons). Diarrhoeal episodes were increased when iron was given with zinc (rate ratio 1.13, 95% CI 0.98 to 1.29; 11,328 child-months, six trials), but not without (rate ratio 1.01, 95% CI 0.94 to 1.09; 207,879 patient-days, 13 trials). There was no increased risk of febrile episodes in high malaria transmission locations with iron (rate ratio 1.05, 95% CI 0.94 to 1.17; 13,799 child-months, five trials), nor a difference in the number of child visits to the clinic or need for hospitalization. Iron did not affect children's weight or height after supplementation.

We also included trials comparing iron with an antimalarial treatment versus placebo. Four trials assessed this comparison and all included additional trial arms assessing antimalarial alone and iron alone. No significant benefit and no increased risk were observed with iron. The combination of iron + antimalarial prevented clinical malaria and the rate of hospitalizations to a similar degree as antimalarial treatment alone. All-cause mortality was not affected in either comparison.

Finally, we included trials assessing the administration of iron to children with proven malaria, together with treatment for malaria. Iron supplementation did not increase the risk of parasitological failure at end of treatment (RR 0.98, 95% CI 0.69 to 1.39, three trials). Mortality was rare (3/331 versus 2/253 for iron versus control, non-significant).

In summary, we did not find an increased risk of malaria or death with iron supplementation, whether given for prevention or treatment of anaemia. An increased risk exists when there are no regular malaria surveillance and treatment facilities available. Haemoglobin and anaemia improve to a similar degree in malarial and non-malarial regions. No effects of iron on other infections were demonstrated. Zinc co-supplementation with iron increased the risk for diarrhoea .

Overall completeness and applicability of evidence

We included trials that were conducted in countries defined from hypoendemic to holoendemic for malaria. Not all trials were relevant for the assessment of the effect of iron supplementation on malaria, since malaria was rare or non-existent in hypo/mesoendemic countries. However, malaria was assessed and reported only in locations where the question is relevant. We used a broad endemicity definition to avoid missing studies. Thus, the assessment of malaria applies to settings where malaria was prevalent during the time of the trial. Six additional trials/trial arms may add data to this comparison: four ongoing or recently completed trials (Gomes 2001; Browne 2005; Denno 2006; Sazawal 2006 (C)c for the comparison of iron + zinc versus zinc), and two trials from which we could not obtain data that were collected during the trial (Powers 1983; Taylor 2001).

Mortality data were poorly reported. Data were available only for 25 out of 65 trials that assessed iron supplementation for treatment or prevention of anaemia. In 15 of them no deaths were reported and the overall mortality rate in these 25 trials was 1.7%. Some trials reporting on mortality referred only to children available for analysis at end of treatment/follow up and in others the number of children evaluated for this outcome was not clear. Deaths should obviously be assessed among all children randomized, mainly those lost to follow up. Thus, we consider this comparison incomplete and accordingly its results may not be applicable.

We included only oral iron supplied as a medicinal product, providing higher than physiological doses of iron. We did not include trials that assessed iron fortified foods or drinks that provide lower, physiological doses of iron. Similarly, we did not include trials assessing intravenous iron, since this is not the intervention of interest when examining the risks of routine iron supplementation in childhood. These trials are listed among excluded trials and may complement the evidence on malaria and other infectious complications, although they probably address a different question than that of our review.

Quality of the evidence

We encountered several problems when attempting to extract and use the data reported in the primary studies. We highlight these points to direct future studies.

  1. Much of the evidence relies on cluster-randomized trials, such as the most recent trial by Sazawal et al (Sazawal 2006 (C)a and Sazawal 2006 (C)b). Naturally, these were the largest trials and thus carried a large weight in the meta-analysis. However, there is a correlation between individuals for all the major outcomes assessed in these trials. Within households it to be expected that anaemia, iron status, malaria, and other contagious or non-contagious infectious complications will be highly correlated. In classes or schools, the correlation between individuals may be smaller, but the large cluster size increases the cluster effect. These trials should be planned and analysed accordingly. For complete transparency and to permit a full assessment of the trial, the report should provide intracluster correlation coefficients (ICCs) for the different outcomes assessed and describe the clusters and their size; the trials may report individual patient results but results adjusted for clustering must also be provided. In fact, it was difficult to deduct that randomization was performed by clusters in some of the cluster-RCTs and out of 13 cluster-RCTs included in our review only Sazawal 2006 (C)a/Sazawal 2006 (C)b reported adjusted analyses. Moreover, even in this recently published trial, ICCs were not provided to permit re-calculation of the effects, and adjustment was performed only for the primary outcomes. In our analyses we used estimated ICCs to adjust the weight of the cluster-RCTs in the meta-analysis. We opted to include these trials despite this major limitation to obtain some view of the full evidence. However, we cannot be sure that the contribution of these large trials to the compiled analysis is correct. In general, the cluster-RCTs shifted pooled effect estimated towards no effect or in favour of placebo. Their exclusion did not significantly alter the results in all comparisons.

  2. Randomization methods were described as appropriate, carrying a low risk for bias, in about 40% of trials. In others there was no description. We noted that allocation concealment affected the results for malaria parasitaemia, with an exaggerated effect estimate favouring placebo in trials with no description of allocation concealment.

  3. Most of the trials were double blinded. However, there was heterogeneity with regard to the management of children identified as anaemic during the trial and after treatment. Some trials supplied iron to all children identified as anaemic below a certain threshold during the trial. After treatment, during follow up, children that remained anaemic were all given iron per protocol, the possibility of treatment, or were not addressed specifically. We could not assess the effects of this variable due to the large heterogeneity in trial protocols and poor reporting. This factor could underlie some of the unexplained heterogeneity observed in our analyses for haemoglobin and anaemia.

Potential biases in the review process

The limitations of the primary studies, detailed above, are reflected in our analysis. In addition, there are limitations related primarily to our analysis.

The paucity of trials contributing to our primary outcomes, malaria and mortality, precluded adequate assessment of the many variables that could affect the results observed. A partial list of these variables would include co-administration or consumption outside the trial protocol of other micronutrients known to affect these outcomes, including zinc, vitamin A and others; nutritional status of individual children; baseline iron and haemoglobin status; baseline parasitaemia, more precise malaria transmission rates at the time of the trial and malaria surveillance and treatment services accessible to children; and methodological issues including comparability of the study groups and more precise effects of clustering. Despite the paucity of trials, the number of included children in these trials was large; more than 21,000 for clinical malaria and more than 25,000 for mortality, each child contributing 1 to 12 months of follow up. Individual patient analysis might yield more information.

We did not extract data regarding nutrition of children at baseline and during the trial. It is possible that iron and other micronutrients confer less benefit among undernourished children in developing countries. Thus, this variable might explain heterogeneity and should be added perhaps in future updates of the review.

The management of outcomes recurring more than once, and frequently many times, during the time of the trial may not have been optimal in our review. These include malaria parasitaemia, perhaps clinical malaria, respiratory infections, and diarrhoeal episodes. Firstly, the trial reports were not clear as to whether results apply to individuals with first/ only episode or to episodes of infection. Secondly, we compiled rate ratios per child-month and SEs from individual trials for these outcomes. This assumes a constant distribution of the outcome over time, which is probably incorrect given the seasonal and outbreak nature of these diseases. We assume that no major effect was missed in these analyses, since sensitivity analyses excluding trials reporting on episodes did not significantly alter results and since the trials' durations were usually long enough to compensate for the episodic nature of these infections.

We did not assess all consequences of iron supplementation. HIV and TB were not addressed in our review and were not reported in the primary studies. We did not attempt to address psychomotor and cognitive outcomes assessed in another Cochrane Review (Martins 2001) and our assessment of the nutritional effects of iron supplementation was limited to a single outcome of absolute weight and height at end of follow up. A more detailed assessment can include change from baseline values, age/weight/height adjusted Z scores, and subgroup analyses according to baseline iron and haemoglobin status.

Finally, despite our attempts to identify and include all studies, it is likely that more unpublished RCTs exist, as those identified in PhD format (Gebresellassie 1996; Adam 1997 (C)) or others (Roschnik 2003 (C)).

Agreements and disagreements with other studies or reviews

The questions addressed in our review were addressed by other meta-analyses focusing on individual outcomes, but not limited to malaria-endemic areas.

INACG 1999 included 13 trials (nine trials in children, four trials in adults including pregnant women) and assessed the evidence on the risks and benefits of iron supplementation in malarious areas (INACG 1999). Iron supplementation was associated with a small, non-statistically significant increase in the risk of malarial attack RR 1.1 (CI 0.9 to 1.3). Iron supplementation increased the odds of being slide positive for Plasmodium falciparum at the end of supplementation period (RR 1.17, CI 1.08 to 1.25) with a non-significant absolute increase in prevalence of infection of 5.7% (95% CI -1.2 to 12.6). The malarial outcome across all trials was not related to age or change in level of haemoglobin.

Gera and Sachdev assessed the effect of iron supplementation on infections in children (Gera 2002). For malaria, the outcome of parasitaemia at end of treatment was analysed. Their meta-analysis showed a significantly higher risk of parasitaemia with iron supplementation, but the authors concluded that there is no significant effect of iron on malaria due to the association of the effect with baseline parasitaemia. Although not limited to malaria-endemic areas, their analysis of malaria naturally included trials conducted in malaria endemic areas. Unlike our analysis, this meta-analysis compiled one trial that included children with malaria (van den Hombergh 1996) together with other trials on prevention or treatment of anaemia; included two publications that were considered as a single trial in our analysis (Berger 2000 and Chippaux 1991 listed under this reference); included one trial that administered parenteral iron, excluded from our review (Oppenheimer 1986); analysed two cluster-RCTs as if individually randomized (Smith 1989 (C); Mebrahtu 2004 (C)); and did not include six trials that were included in our analysis on parasitaemia at end of treatment (Lawless 1994; Adam 1997 (C); Verhoef 2002; Desai 2003; Richard 2006; Fahmida 2007). This meta-analysis concurred with ours on the lack of effect of iron supplementation on respiratory and overall infection rates. An increased risk of diarrhoea of small magnitude was observed in both meta-analyses, which we showed to be due to the effect of co-administration of zinc with iron. Oppenheimer 2001 summarized semi-qualitatively that iron may enhance the risk of malaria and non-malarial infections in malarial regions. Iannotti 2006 examined infectious complications in developing countries and based their conclusion of increased risk for malaria and other infection-related adverse events in malaria-endemic countries mainly on the largest trial by Sazawal et al (Sazawal 2006 (C)a), noting the lack of information on TB and HIV.

Previous meta-analyses have shown that iron treatment increases haemoglobin and prevents anaemia. The meta-analyses differed in those variables examined and found to explain heterogeneity. Gera et al pointed at mean baseline haemoglobin and malaria hyperendemicity as significant variables (Gera 2007). In their analysis the weighted mean difference in haemoglobin was 0.65 (CI 0.35 to 0.94) g/dL in 11 malaria hyperendemic countries compared to 0.75 (CI 0.61 to 0.90) in 80 non-endemic countries. In our analysis the absolute mean difference in haemoglobin between iron and control were larger in hyper/holoendemic countries compared to lower transmission rate settings, without statistical significance (1.06 versus 0.80 g/dL for end values and 0.91 versus 0.41 g/dL for change from baseline, respectively, both without folic acid at end of treatment). Their analysis was based on computed change from baseline values for all studies with calculated or computed SDs, while we used only the reported effect estimates (end values or change from baseline), separated between end of treatment and end of follow-up results and stratified the analysis by co-administered micronutrients. Iannotti et al summarized qualitatively that haemoglobin improvements appeared to be related to baseline status and to residence in malarial endemic regions, from a set of 26 RCTs performed in developing countries (Iannotti 2006).

Most meta-analyses have concurred with our results that iron supplementation does not significantly improve growth indicators within the timeframe of RCTs (Bhandari 2001; Ramakrishnan 2004; Iannotti 2006; Sachdev 2006). Sachdev 2006 found that positive effects were present in malaria hyperendemic regions and among children above five years of age and negative effects were observed in developed countries and with supplementation for six months or longer, but these analyses were considered exploratory and requiring confirmation. Sachdev et al also examined the effects of iron supplementation on mental and motor development in children, finding an improvement in mental development score with iron, particularly apparent for intelligence tests above seven years of age and in initially anaemic or iron-deficient anaemic subjects (Sachdev 2005). These outcomes were not examined in our review.

Current WHO guidelines rely strongly on the most recent RCT showing an increased risk for adverse events, including deaths, with iron supplementation (Sazawal 2006 (C)a, Sazawal 2006 (C)b; WHO 2007). This was by far the largest trial to date and its advantages include adequate randomization methods and double blinding, inclusion of all children aged 1 to 35 months in the community, mimicking as closely as possible the real-life circumstances of a community intervention, and the assessment of clinically-relevant outcomes. This was probably the only trial to date powered to assess the effect of iron supplementation on clinical malaria. Several unique features of the trial should be noted. Malaria assessment was based on hospital admissions due to malaria, unlike other trials that assessed all cases of clinical malaria and/or parasitaemia. The trial's conclusions are based on the results at the time the data monitoring committee stopped the trial for harm, according to the trial's protocol. At the time of the trial, the primary health care system in Zanzibar was weak (prior to the Zanzibar Malaria Control Program) and the main trial's protocol did not offer children special malaria prevention, diagnosis or treatment services, unlike the design of other, smaller scale, trials (see Table 5). In the substudy, as noted above, surveillance for parasitaemia was performed and children received treatment according to study protocol in their homes. The substudy showed that iron supplementation is protective for malaria, while the main trial showed that iron is harmful. This was substantiated by a subgroup meta-analysis including other trials. It is difficult to understand why closer surveillance for malaria affects the consequences of iron supplementation on malaria. In other trials, it is possible that without surveillance and treatment the more severe cases of malaria are manifested and counted as outcomes, while organized surveillance results in cases of mild disease counted as outcomes. However, this cannot explain the differences between Sazawal 2006 (C)a and Sazawal 2006 (C)b, since only admissions due to malaria were counted in both trials. The comparison between iron, folic acid, zinc, and vitamin A versus zinc and vitamin A, at the time the iron arm was stopped, can also contribute to our understanding of the effects of iron and folic acid. This comparison includes 16,199 more children with more than 200,000 child-months follow up.

Authors' conclusions

Implications for practice

We did not find an increased risk of malaria, other infectious outcomes or all-cause mortality with iron supplementation for children living in malaria-endemic areas. Careful analyses separating between specific interventions, specific outcomes, and adjusting for trial design did not point at increased risk for these outcomes in hyperendemic regions nor in children that were non-anaemic at baseline. Iron may be associated with an increased risk of malaria in settings with no access to malaria prevention and treatment services. Iron supplementation significantly improves haemoglobin and reduces the prevalence of anaemia, mainly among anaemic children. Iron deficiency and anaemia in the long-term have been shown to impair cognitive and motor development (Pollitt 1993; Grantham-McGregor 2001), growth (Lawless 1994), and immune function (Oppenheimer 2001), and contribute to childhood mortality in developing countries (FAO/WHO 2005; WHO 2004). Based on our review, iron supplementation should not be withheld from children living in malaria-endemic countries. A higher gain is achieved among anaemic children, thus baseline testing may assist decision making. However, in settings where iron deficiency and anaemia affects most children, as in some parts of sub-Saharan Africa (WHO 2004), iron supplementation should not be restricted for fear of infections or death. Prevention and treatment of malaria should be offered to children regardless of iron supplementation, since these interventions reduce malaria, mortality, and anaemia (Lengeler 2004; Meremikwu 2008).

Implications for research

  1. An individual patient data meta-analysis of the trials that were conducted in areas with significant rates of malaria transmission and assessed malaria-related outcomes can probably resolve the remaining questions and adjust for many covariates, including baseline comparability of the study groups, children's age, dose and duration of iron supplementation, co-supplementation with other micronutrients and nutritional status, baseline iron levels, malaria transmission rates, and many other variables that cannot be adjusted for simultaneously in a meta-analysis based on published data. Such an analysis should include the 18 trials reporting on malaria identified in this review and attempt to retrieve data from the few additional trials where malaria-related outcomes were collected during the trial (Powers 1983; Taylor 2001) and the ongoing/completed trials (Gomes 2001; Browne 2005; Denno 2006 and the substudy and the zinc arm of Sazawal et al Sazawal 2006 (C)b,Sazawal 2006 (C)c). Since malaria is just one cause of morbidity and mortality in malaria-endemic locations, and since iron supplementation and prevention of anaemia have significant health benefits, global outcomes should be assessed, mainly mortality, among all randomized children.

  2. A large trial, mimicking a real-life routine supplementation intervention, similar to that reported in Sazawal 2006 (C)a, assessing iron with/without folic acid, in which all children are offered currently recommended malaria prevention and treatment services is warranted. Trials assessing iron supplementation should be large enough and of sufficient duration to assess rare outcomes, such as mortality.

  3. An intervention of iron given to children after an attack of malaria to boost haemoglobin should be assessed in randomized controlled trials.

  4. Well-conducted observational studies assessing the effects of iron supplementation are important given that randomized controlled trials measure only a limited duration of iron supplementation and may not represent the child population in need of iron supplementation, given the low risk of death observed in this review. Growth and developmental outcomes would probably be better assessed in such long-term studies.

Acknowledgements

Dr Juliana U Ojukwu was awarded a Reviews for Africa Programme Fellowship (www.mrc.ac.za/cochrane/rap.htm), funded by a grant from the Nuffield Commonwealth Programme, through The Nuffield Foundation.

The editorial base for the Cochrane Infectious Diseases Group is funded by the UK Department for International Development (DFID) for the benefit of developing countries. Dafna Yahav received funding from the editorial base.

We acknowledge the assistance and contribution to the the review of Sarah Donegan (advised and assisted with data analysis); Harriet G. McLehose (assisted in writing and final drafting of the protocol and review); Paul Garner (assisted with study design, analysis, and co-ordination) and Leonard Leibovici (data analysis and interpretation).

We thank the Cochrane Infectious Diseases Group for their support of this review, without which this work could not have been performed. Special thanks to Vittoria Lutje who designed and performed the searches. We thank Prof. Jimmy Volmink, Dr. Taryn Young, and Karen Essex of the South African Cochrane Center for inviting us to meet and work on this review.

We would like to thank Profs. H.P. Sachdev and Tarun Gera for their help obtaining unpublished data for trials included in this review. We thank all the authors who responded to requests for further data and provided the data where available (see under 'Characteristics of included studies'). We thank the reviewers of our manuscript who raised important issues related to the analyses presented. We urge all the authors of the primary studies to correct the data used from their studies if necessary, add data if available mainly for the primary outcomes of malaria and mortality, and to point out any other inaccuracies in our analysis.

Data and analyses

Download statistical data

Comparison 1. Iron +/- folic acid vs. placebo or no treatment for prevention or treatment of anaemia
Outcome or subgroup titleNo. of studiesNo. of participantsStatistical methodEffect size
1 Drop-outs by location4734704Risk Ratio (M-H, Fixed, 95% CI)1.00 [0.94, 1.07]
1.1 Hypo or mesoendemic3111754Risk Ratio (M-H, Fixed, 95% CI)1.00 [0.94, 1.08]
1.2 Hyper or holoendemic1722950Risk Ratio (M-H, Fixed, 95% CI)1.00 [0.86, 1.17]
2 Clinical malaria (individually-randomized trials)93946Risk Ratio (M-H, Fixed, 95% CI)0.96 [0.87, 1.06]
2.1 Iron vs. placebo/no treatment82725Risk Ratio (M-H, Fixed, 95% CI)1.01 [0.90, 1.12]
2.2 Iron + folic acid vs. placebo/no treatment00Risk Ratio (M-H, Fixed, 95% CI)Not estimable
2.3 Iron + antimalarial vs. antimalarial41221Risk Ratio (M-H, Fixed, 95% CI)0.83 [0.67, 1.03]
3 Clinical malaria (all trials)1422724Risk Ratio (Random, 95% CI)1.00 [0.88, 1.13]
3.1 Iron vs. placebo/no treatment113928Risk Ratio (Random, 95% CI)1.03 [0.92, 1.15]
3.2 Iron + folic acid vs. placebo/no treatment217575Risk Ratio (Random, 95% CI)0.77 [0.31, 1.90]
3.3 Iron + antimalarial vs. antimalarial41221Risk Ratio (Random, 95% CI)0.89 [0.63, 1.26]
4 Clinical malaria (subgroup anaemia at baseline)9 Risk Ratio (Random, 95% CI)1.01 [0.79, 1.30]
4.1 Iron vs. placebo/no treatment7 Risk Ratio (Random, 95% CI)1.12 [0.93, 1.35]
4.2 Iron + folic acid vs. placebo/no treatment1 Risk Ratio (Random, 95% CI)0.46 [0.24, 0.88]
4.3 Iron + antimalarial vs. antimalarial2 Risk Ratio (Random, 95% CI)0.89 [0.37, 2.11]
5 Clinical malaria (subgroup non-anaemic at baseline)4 Risk Ratio (Fixed, 95% CI)0.96 [0.85, 1.09]
5.1 Iron vs. placebo/ no treatment4 Risk Ratio (Fixed, 95% CI)0.97 [0.85, 1.11]
5.2 Iron + folic acid vs. placebo/ no treatment0 Risk Ratio (Fixed, 95% CI)Not estimable
5.3 Iron + antimalarial vs. antimalarial2 Risk Ratio (Fixed, 95% CI)0.93 [0.69, 1.25]
6 Clinical malaria (sensitivity adequate concealment)8 Risk Ratio (Fixed, 95% CI)1.04 [0.95, 1.14]
6.1 Iron vs. placebo/no treatment6 Risk Ratio (Fixed, 95% CI)1.00 [0.88, 1.13]
6.2 Iron + folic acid vs. placebo/no treatment2 Risk Ratio (Fixed, 95% CI)1.11 [0.96, 1.28]
6.3 Iron + antimalarial vs. antimalarial3 Risk Ratio (Fixed, 95% CI)1.01 [0.77, 1.32]
7 Clinical malaria (sensitivity unclear concealment)6 Risk Ratio (Fixed, 95% CI)0.96 [0.81, 1.13]
7.1 Iron vs. placebo/no treatment5 Risk Ratio (Fixed, 95% CI)1.07 [0.89, 1.29]
7.2 Iron + folic acid vs. placebo/no treatment0 Risk Ratio (Fixed, 95% CI)Not estimable
7.3 Iron + antimalarial vs. antimalarial1 Risk Ratio (Fixed, 95% CI)0.59 [0.40, 0.86]
8 Clinical malaria (subgroup malaria surveillance and treatment)8 Risk Ratio (Fixed, 95% CI)0.93 [0.84, 1.04]
8.1 Iron vs. placebo/no treatment6 Risk Ratio (Fixed, 95% CI)0.98 [0.88, 1.11]
8.2 Iron + folic acid vs. placebo/no treatment1 Risk Ratio (Fixed, 95% CI)0.46 [0.24, 0.88]
8.3 Iron + antimalarial vs. antimalarial4 Risk Ratio (Fixed, 95% CI)0.84 [0.68, 1.05]
9 Clinical malaria (subgroup no routine surveillance for malaria)6 Risk Ratio (Fixed, 95% CI)1.16 [1.03, 1.31]
9.1 Iron vs. placebo/ no treatment5 Risk Ratio (Fixed, 95% CI)1.16 [0.93, 1.45]
9.2 Iron + folic acid vs. placebo/no treatment1 Risk Ratio (Fixed, 95% CI)1.16 [1.00, 1.34]
10 Any parasitaemia, end of treatment (individually-randomized trials)41365Risk Ratio (M-H, Fixed, 95% CI)1.04 [0.88, 1.22]
10.1 Iron vs. placebo/ no treatment3941Risk Ratio (M-H, Fixed, 95% CI)1.08 [0.91, 1.28]
10.2 Iron + antimalarial vs. antimalarial1424Risk Ratio (M-H, Fixed, 95% CI)0.87 [0.56, 1.33]
11 Any parasitaemia, end of treatment (all trials)83184Risk Ratio (Fixed, 95% CI)1.13 [1.01, 1.26]
11.1 Iron vs. placebo/no treatment72760Risk Ratio (Fixed, 95% CI)1.15 [1.02, 1.29]
11.2 Iron + antimalarial vs. antimalarial1424Risk Ratio (Fixed, 95% CI)0.87 [0.56, 1.33]
12 Any parasitaemia or 'malaria', end of treatment (all trials)124584Risk Ratio (Fixed, 95% CI)1.09 [0.99, 1.19]
12.1 Iron vs. placebo/no treatment114160Risk Ratio (Fixed, 95% CI)1.10 [1.00, 1.21]
12.2 Iron + antimalarial vs. antimalarial1424Risk Ratio (Fixed, 95% CI)0.87 [0.56, 1.33]
13 Any parasitaemia, end of follow up4941Risk Ratio (M-H, Random, 95% CI)1.17 [1.00, 1.37]
13.1 Iron vs. placebo/no treatment4941Risk Ratio (M-H, Random, 95% CI)1.17 [1.00, 1.37]
14 Any parasitaemia or 'malaria', end of treatment (all trials), adequate allocation concealment6 Risk Ratio (Fixed, 95% CI)1.01 [0.88, 1.15]
14.1 Iron vs. placebo/no treatment5 Risk Ratio (Fixed, 95% CI)1.03 [0.89, 1.18]
14.2 Iron + antimalarial vs. antimalarial1 Risk Ratio (Fixed, 95% CI)0.87 [0.56, 1.33]
15 Any parasitaemia or 'malaria', end of treatment (all trials), unclear allocation concealment6 Risk Ratio (Fixed, 95% CI)1.17 [1.03, 1.32]
15.1 Iron vs. placebo/no treatment6 Risk Ratio (Fixed, 95% CI)1.17 [1.03, 1.32]
15.2 Iron + antimalarial vs. antimalarial0 Risk Ratio (Fixed, 95% CI)Not estimable
16 Severe malaria (clinical definition)4 Risk Ratio (Random, 95% CI)Totals not selected
16.1 Iron vs. placebo/no treatment2 Risk Ratio (Random, 95% CI)Not estimable
16.2 Iron + folic acid vs. placebo no treatment2 Risk Ratio (Random, 95% CI)Not estimable
16.3 Iron + antimalarial vs. antimalarial1 Risk Ratio (Random, 95% CI)Not estimable
17 High-grade parasitaemia5 Risk Ratio (Fixed, 95% CI)1.15 [0.93, 1.43]
17.1 Iron vs. placebo/ no treatment5 Risk Ratio (Fixed, 95% CI)1.15 [0.93, 1.43]
18 All-cause mortality by intervention2526329Risk Ratio (M-H, Fixed, 95% CI)1.11 [0.91, 1.36]
18.1 Iron vs. placebo/no treatment198209Risk Ratio (M-H, Fixed, 95% CI)0.87 [0.48, 1.57]
18.2 Iron + folic acid vs. placebo/no treatment416859Risk Ratio (M-H, Fixed, 95% CI)1.13 [0.90, 1.42]
18.3 Iron + antimalarial vs. antimalarial51261Risk Ratio (M-H, Fixed, 95% CI)1.30 [0.67, 2.50]
19 All-cause mortality by location1020903Risk Ratio (M-H, Fixed, 95% CI)1.11 [0.90, 1.36]
19.1 Hypo or mesoendemic31717Risk Ratio (M-H, Fixed, 95% CI)0.50 [0.12, 1.98]
19.2 Hyper or holoendemic719186Risk Ratio (M-H, Fixed, 95% CI)1.13 [0.92, 1.39]
20 All-cause mortality by intervention (including treatment for malaria trials)1221272Risk Ratio (M-H, Fixed, 95% CI)1.11 [0.91, 1.36]
20.1 Iron vs. placebo/no treatment73122Risk Ratio (M-H, Fixed, 95% CI)0.87 [0.48, 1.57]
20.2 Iron + folic acid vs. placebo/no treatment216560Risk Ratio (M-H, Fixed, 95% CI)1.13 [0.90, 1.42]
20.3 Iron + antimalarial vs. antimalarial61590Risk Ratio (M-H, Fixed, 95% CI)1.31 [0.71, 2.43]
21 All-cause mortality (sensitivity concealment)1020903Risk Ratio (M-H, Fixed, 95% CI)1.11 [0.90, 1.36]
21.1 Adequate allocation concealment819767Risk Ratio (M-H, Fixed, 95% CI)1.10 [0.90, 1.35]
21.2 Unclear allocation concealment21136Risk Ratio (M-H, Fixed, 95% CI)1.46 [0.38, 5.59]
22 All-cause mortality by location (sensitivity 0 = 1)2526329Risk Ratio (M-H, Fixed, 95% CI)1.10 [0.90, 1.33]
22.1 Hypo or mesoendemic83040Risk Ratio (M-H, Fixed, 95% CI)0.68 [0.27, 1.67]
22.2 Hyper or holoendemic1723289Risk Ratio (M-H, Fixed, 95% CI)1.12 [0.92, 1.37]
23 Haemoglobin, end of treatment, by location419482Mean Difference (IV, Fixed, 95% CI)0.87 [0.82, 0.91]
23.1 Iron vs. placebo, hypo or mesoendemic234684Mean Difference (IV, Fixed, 95% CI)0.80 [0.74, 0.86]
23.2 Iron vs. placebo, hyper or holoendemic103426Mean Difference (IV, Fixed, 95% CI)1.06 [0.97, 1.15]
23.3 Iron + folic acid vs. placebo, hypo or mesoendemic5907Mean Difference (IV, Fixed, 95% CI)0.72 [0.54, 0.91]
23.4 Iron + folic acid vs. placebo, hyper or holoendemic3465Mean Difference (IV, Fixed, 95% CI)0.80 [0.54, 1.06]
24 Haemoglobin, end of treatment, by anaemia at baseline419482Mean Difference (IV, Random, 95% CI)0.90 [0.68, 1.11]
24.1 Baseline Hb < 11 g/dL284767Mean Difference (IV, Random, 95% CI)1.13 [0.80, 1.45]
24.2 Baseline Hb > 11 g/dL194715Mean Difference (IV, Random, 95% CI)0.60 [0.39, 0.80]
25 Haemoglobin, end of treatment, by zinc and antiparasitic treatment419482Mean Difference (IV, Random, 95% CI)0.90 [0.68, 1.11]
25.1 Iron alone vs. placebo366580Mean Difference (IV, Random, 95% CI)1.00 [0.74, 1.25]
25.2 Iron + zinc vs. zinc82042Mean Difference (IV, Random, 95% CI)0.47 [0.14, 0.81]
25.3 Iron + antiparasitic vs. antiparasitic5860Mean Difference (IV, Random, 95% CI)0.62 [0.31, 0.92]
26 Haemoglobin, end of follow up, by location164909Mean Difference (IV, Random, 95% CI)0.69 [0.26, 1.12]
26.1 Iron vs. placebo, hypo or mesoendemic4454Mean Difference (IV, Random, 95% CI)0.92 [-0.48, 2.33]
26.2 Iron vs. placebo, hyper or holoendemic73113Mean Difference (IV, Random, 95% CI)0.52 [0.23, 0.80]
26.3 Iron + folic acid vs. placebo, hypo or mesoendemic3210Mean Difference (IV, Random, 95% CI)1.03 [-0.55, 2.62]
26.4 Iron + folic acid vs. placebo, hyper or holoendemic21132Mean Difference (IV, Random, 95% CI)0.34 [0.12, 0.56]
27 Haemoglobin, change from baseline, end of treatment, by location226018Mean Difference (IV, Random, 95% CI)0.51 [0.34, 0.68]
27.1 Iron vs. placebo, hypo or mesoendemic122595Mean Difference (IV, Random, 95% CI)0.40 [0.22, 0.58]
27.2 Iron vs. placebo, hyper or holoendemic81610Mean Difference (IV, Random, 95% CI)0.91 [0.56, 1.26]
27.3 Iron + folic acid vs. placebo, hypo or mesoendemic11784Mean Difference (IV, Random, 95% CI)0.07 [-0.05, 0.18]
27.4 Iron + folic acid vs. placebo, hyper or holoendemic129Mean Difference (IV, Random, 95% CI)1.33 [0.90, 1.75]
28 Haemoglobin, change from baseline, end of follow up, by location73768Mean Difference (IV, Random, 95% CI)0.76 [0.29, 1.22]
28.1 Iron vs. placebo, hypo or mesoendemic00Mean Difference (IV, Random, 95% CI)Not estimable
28.2 Iron vs. placebo, hyper or holoendemic42293Mean Difference (IV, Random, 95% CI)0.41 [-0.01, 0.84]
28.3 Iron + folic acid vs. placebo, hypo or mesoendemic1142Mean Difference (IV, Random, 95% CI)4.18 [3.65, 4.71]
28.4 Iron + folic acid vs. placebo, hyper or holoendemic21333Mean Difference (IV, Random, 95% CI)0.39 [0.15, 0.62]
29 Anaemia by location, end of treatment286328Risk Ratio (M-H, Random, 95% CI)0.51 [0.40, 0.67]
29.1 Hypo or mesoendemic142668Risk Ratio (M-H, Random, 95% CI)0.56 [0.42, 0.76]
29.2 Hyper or holoendemic143660Risk Ratio (M-H, Random, 95% CI)0.49 [0.31, 0.78]
30 Anaemia by location, end of follow up102589Risk Ratio (M-H, Random, 95% CI)0.67 [0.53, 0.83]
30.1 Hypo or mesoendemic2376Risk Ratio (M-H, Random, 95% CI)0.38 [0.07, 1.91]
30.2 Hyper or holoendemic82213Risk Ratio (M-H, Random, 95% CI)0.73 [0.59, 0.88]
31 URTI/pneumonia episodes per patient-month, by location1122577Risk Ratio (Fixed, 95% CI)0.97 [0.91, 1.04]
31.1 Hypo or mesoendemic without zinc88378Risk Ratio (Fixed, 95% CI)0.98 [0.90, 1.08]
31.2 Hypo or mesoendemic with zinc59024Risk Ratio (Fixed, 95% CI)0.94 [0.84, 1.04]
31.3 Hyper or holoendemic without zinc22871Risk Ratio (Fixed, 95% CI)1.02 [0.79, 1.32]
31.4 Hyper or holoendemic with zinc12304Risk Ratio (Fixed, 95% CI)1.05 [0.80, 1.38]
32 Diarrhoeal episodes per patient-month, by location14219207Risk Ratio (Random, 95% CI)1.06 [0.98, 1.15]
32.1 Hypo or mesoendemic without zinc78254Risk Ratio (Random, 95% CI)1.03 [0.95, 1.12]
32.2 Hypo or mesoendemic with zinc59024Risk Ratio (Random, 95% CI)1.14 [0.98, 1.32]
32.3 Hyper or holoendemic without zinc6199625Risk Ratio (Random, 95% CI)0.92 [0.77, 1.10]
32.4 Hyper or holoendemic with zinc12304Risk Ratio (Random, 95% CI)0.99 [0.67, 1.46]
33 Infections per patient-month, by location, individually-randomized trials13 Risk Ratio (Fixed, 95% CI)Subtotals only
33.1 Fever episodes715683Risk Ratio (Fixed, 95% CI)1.03 [0.93, 1.14]
33.2 Days with fever1110Risk Ratio (Fixed, 95% CI)8.37 [1.91, 36.58]
33.3 Disease episodes other than diarrhoea or respiratory infections33096Risk Ratio (Fixed, 95% CI)1.03 [0.80, 1.33]
33.4 All disease episodes2208082Risk Ratio (Fixed, 95% CI)1.20 [1.00, 1.42]
34 Hospitalizations and clinic visits5 Risk Ratio (Fixed, 95% CI)Subtotals only
34.1 Hospitalization, iron vs. placebo4199641Risk Ratio (Fixed, 95% CI)0.99 [0.90, 1.09]
34.2 Hospitalization, iron + antimalarial vs. antimalarial25922Risk Ratio (Fixed, 95% CI)1.22 [0.96, 1.57]
34.3 Clinic visit, iron vs. placebo25808Risk Ratio (Fixed, 95% CI)0.95 [0.88, 1.02]
34.4 Clinic visit, iron + antimalarial vs. antimalarial37395Risk Ratio (Fixed, 95% CI)1.02 [0.95, 1.09]
35 Weight, end value186334Std. Mean Difference (IV, Random, 95% CI)0.02 [-0.05, 0.09]
35.1 Iron vs. placebo, hypo or mesoendemic144439Std. Mean Difference (IV, Random, 95% CI)0.05 [-0.04, 0.13]
35.2 Iron vs. placebo, hyper or holoendemic41895Std. Mean Difference (IV, Random, 95% CI)-0.08 [-0.17, 0.01]
36 Weight, change from baseline121688Std. Mean Difference (IV, Random, 95% CI)0.21 [-0.05, 0.46]
36.1 Iron vs. placebo, hypo or mesoendemic91350Std. Mean Difference (IV, Random, 95% CI)0.09 [-0.19, 0.37]
36.2 Iron vs. placebo, hyper or holoendemic3338Std. Mean Difference (IV, Random, 95% CI)0.48 [-0.10, 1.06]
37 Height, end value186641Std. Mean Difference (IV, Random, 95% CI)-0.00 [-0.05, 0.05]
37.1 Iron vs. placebo, hypo or mesoendemic144746Std. Mean Difference (IV, Random, 95% CI)0.02 [-0.04, 0.07]
37.2 Iron vs. placebo, hyper or holoendemic41895Std. Mean Difference (IV, Random, 95% CI)-0.05 [-0.14, 0.04]
38 Height, change from baseline121688Std. Mean Difference (IV, Random, 95% CI)0.18 [-0.03, 0.38]
38.1 Iron vs. placebo, hypo or mesoendemic91350Std. Mean Difference (IV, Random, 95% CI)0.22 [-0.01, 0.45]
38.2 Iron vs. placebo, hyper or holoendemic3338Std. Mean Difference (IV, Random, 95% CI)0.06 [-0.42, 0.54]
Analysis 1.1.

Comparison 1 Iron +/- folic acid vs. placebo or no treatment for prevention or treatment of anaemia, Outcome 1 Drop-outs by location.

Analysis 1.2.

Comparison 1 Iron +/- folic acid vs. placebo or no treatment for prevention or treatment of anaemia, Outcome 2 Clinical malaria (individually-randomized trials).

Analysis 1.3.

Comparison 1 Iron +/- folic acid vs. placebo or no treatment for prevention or treatment of anaemia, Outcome 3 Clinical malaria (all trials).

Analysis 1.4.

Comparison 1 Iron +/- folic acid vs. placebo or no treatment for prevention or treatment of anaemia, Outcome 4 Clinical malaria (subgroup anaemia at baseline).

Analysis 1.5.

Comparison 1 Iron +/- folic acid vs. placebo or no treatment for prevention or treatment of anaemia, Outcome 5 Clinical malaria (subgroup non-anaemic at baseline).

Analysis 1.6.

Comparison 1 Iron +/- folic acid vs. placebo or no treatment for prevention or treatment of anaemia, Outcome 6 Clinical malaria (sensitivity adequate concealment).

Analysis 1.7.

Comparison 1 Iron +/- folic acid vs. placebo or no treatment for prevention or treatment of anaemia, Outcome 7 Clinical malaria (sensitivity unclear concealment).

Analysis 1.8.

Comparison 1 Iron +/- folic acid vs. placebo or no treatment for prevention or treatment of anaemia, Outcome 8 Clinical malaria (subgroup malaria surveillance and treatment).

Analysis 1.9.

Comparison 1 Iron +/- folic acid vs. placebo or no treatment for prevention or treatment of anaemia, Outcome 9 Clinical malaria (subgroup no routine surveillance for malaria).

Analysis 1.10.

Comparison 1 Iron +/- folic acid vs. placebo or no treatment for prevention or treatment of anaemia, Outcome 10 Any parasitaemia, end of treatment (individually-randomized trials).

Analysis 1.11.

Comparison 1 Iron +/- folic acid vs. placebo or no treatment for prevention or treatment of anaemia, Outcome 11 Any parasitaemia, end of treatment (all trials).

Analysis 1.12.

Comparison 1 Iron +/- folic acid vs. placebo or no treatment for prevention or treatment of anaemia, Outcome 12 Any parasitaemia or 'malaria', end of treatment (all trials).

Analysis 1.13.

Comparison 1 Iron +/- folic acid vs. placebo or no treatment for prevention or treatment of anaemia, Outcome 13 Any parasitaemia, end of follow up.

Analysis 1.14.

Comparison 1 Iron +/- folic acid vs. placebo or no treatment for prevention or treatment of anaemia, Outcome 14 Any parasitaemia or 'malaria', end of treatment (all trials), adequate allocation concealment.

Analysis 1.15.

Comparison 1 Iron +/- folic acid vs. placebo or no treatment for prevention or treatment of anaemia, Outcome 15 Any parasitaemia or 'malaria', end of treatment (all trials), unclear allocation concealment.

Analysis 1.16.

Comparison 1 Iron +/- folic acid vs. placebo or no treatment for prevention or treatment of anaemia, Outcome 16 Severe malaria (clinical definition).

Analysis 1.17.

Comparison 1 Iron +/- folic acid vs. placebo or no treatment for prevention or treatment of anaemia, Outcome 17 High-grade parasitaemia.

Analysis 1.18.

Comparison 1 Iron +/- folic acid vs. placebo or no treatment for prevention or treatment of anaemia, Outcome 18 All-cause mortality by intervention.

Analysis 1.19.

Comparison 1 Iron +/- folic acid vs. placebo or no treatment for prevention or treatment of anaemia, Outcome 19 All-cause mortality by location.

Analysis 1.20.

Comparison 1 Iron +/- folic acid vs. placebo or no treatment for prevention or treatment of anaemia, Outcome 20 All-cause mortality by intervention (including treatment for malaria trials).

Analysis 1.21.

Comparison 1 Iron +/- folic acid vs. placebo or no treatment for prevention or treatment of anaemia, Outcome 21 All-cause mortality (sensitivity concealment).

Analysis 1.22.

Comparison 1 Iron +/- folic acid vs. placebo or no treatment for prevention or treatment of anaemia, Outcome 22 All-cause mortality by location (sensitivity 0 = 1).

Analysis 1.23.

Comparison 1 Iron +/- folic acid vs. placebo or no treatment for prevention or treatment of anaemia, Outcome 23 Haemoglobin, end of treatment, by location.

Analysis 1.24.

Comparison 1 Iron +/- folic acid vs. placebo or no treatment for prevention or treatment of anaemia, Outcome 24 Haemoglobin, end of treatment, by anaemia at baseline.

Analysis 1.25.

Comparison 1 Iron +/- folic acid vs. placebo or no treatment for prevention or treatment of anaemia, Outcome 25 Haemoglobin, end of treatment, by zinc and antiparasitic treatment.

Analysis 1.26.

Comparison 1 Iron +/- folic acid vs. placebo or no treatment for prevention or treatment of anaemia, Outcome 26 Haemoglobin, end of follow up, by location.

Analysis 1.27.

Comparison 1 Iron +/- folic acid vs. placebo or no treatment for prevention or treatment of anaemia, Outcome 27 Haemoglobin, change from baseline, end of treatment, by location.

Analysis 1.28.

Comparison 1 Iron +/- folic acid vs. placebo or no treatment for prevention or treatment of anaemia, Outcome 28 Haemoglobin, change from baseline, end of follow up, by location.

Analysis 1.29.

Comparison 1 Iron +/- folic acid vs. placebo or no treatment for prevention or treatment of anaemia, Outcome 29 Anaemia by location, end of treatment.

Analysis 1.30.

Comparison 1 Iron +/- folic acid vs. placebo or no treatment for prevention or treatment of anaemia, Outcome 30 Anaemia by location, end of follow up.

Analysis 1.31.

Comparison 1 Iron +/- folic acid vs. placebo or no treatment for prevention or treatment of anaemia, Outcome 31 URTI/pneumonia episodes per patient-month, by location.

Analysis 1.32.

Comparison 1 Iron +/- folic acid vs. placebo or no treatment for prevention or treatment of anaemia, Outcome 32 Diarrhoeal episodes per patient-month, by location.

Analysis 1.33.

Comparison 1 Iron +/- folic acid vs. placebo or no treatment for prevention or treatment of anaemia, Outcome 33 Infections per patient-month, by location, individually-randomized trials.

Analysis 1.34.

Comparison 1 Iron +/- folic acid vs. placebo or no treatment for prevention or treatment of anaemia, Outcome 34 Hospitalizations and clinic visits.

Analysis 1.35.

Comparison 1 Iron +/- folic acid vs. placebo or no treatment for prevention or treatment of anaemia, Outcome 35 Weight, end value.

Analysis 1.36.

Comparison 1 Iron +/- folic acid vs. placebo or no treatment for prevention or treatment of anaemia, Outcome 36 Weight, change from baseline.

Analysis 1.37.

Comparison 1 Iron +/- folic acid vs. placebo or no treatment for prevention or treatment of anaemia, Outcome 37 Height, end value.

Analysis 1.38.

Comparison 1 Iron +/- folic acid vs. placebo or no treatment for prevention or treatment of anaemia, Outcome 38 Height, change from baseline.

Comparison 2. Iron + antimalarial vs. placebo
Outcome or subgroup titleNo. of studiesNo. of participantsStatistical methodEffect size
1 Clinical malaria3 Risk Ratio (M-H, Fixed, 95% CI)Subtotals only
1.1 Iron + antimalarial vs. placebo3728Risk Ratio (M-H, Fixed, 95% CI)0.54 [0.43, 0.67]
1.2 Antimalarial alone vs. placebo3724Risk Ratio (M-H, Fixed, 95% CI)0.53 [0.43, 0.67]
2 All-cause mortality3 Risk Ratio (M-H, Fixed, 95% CI)Subtotals only
2.1 Iron + antimalarial vs. placebo3728Risk Ratio (M-H, Fixed, 95% CI)1.05 [0.52, 2.11]
2.2 Antimalarial alone vs. placebo3724Risk Ratio (M-H, Fixed, 95% CI)0.92 [0.45, 1.91]
3 Haemoglobin at end of treatment1304Mean Difference (IV, Fixed, 95% CI)0.53 [0.22, 0.84]
3.1 Iron + antimalarial vs. placebo1151Mean Difference (IV, Fixed, 95% CI)0.91 [0.47, 1.35]
3.2 Antimalarial alone vs. placebo1153Mean Difference (IV, Fixed, 95% CI)0.15 [-0.29, 0.59]
4 Anaemia3 Risk Ratio (M-H, Fixed, 95% CI)Subtotals only
4.1 Iron + antimalarial vs. placebo, end of treatment2295Risk Ratio (M-H, Fixed, 95% CI)0.44 [0.33, 0.60]
4.2 Iron + antimalarial vs. placebo, end of follow up1420Risk Ratio (M-H, Fixed, 95% CI)0.37 [0.26, 0.54]
4.3 Antimalarial alone vs. placebo, end of treatment2298Risk Ratio (M-H, Fixed, 95% CI)0.81 [0.67, 0.98]
4.4 Antimalarial alone vs. placebo, end of follow up1415Risk Ratio (M-H, Fixed, 95% CI)0.48 [0.34, 0.67]
5 Hospitalizations and clinic visits2 Risk Ratio (Fixed, 95% CI)Subtotals only
5.1 Hospitalization, iron + antimalarial vs. placebo25904Risk Ratio (Fixed, 95% CI)0.59 [0.48, 0.73]
5.2 Hospitalization, antimalarial alone vs. placebo25850Risk Ratio (Fixed, 95% CI)0.48 [0.39, 0.60]
5.3 Clinic visit, iron + antimalarial vs. placebo25904Risk Ratio (Fixed, 95% CI)0.88 [0.82, 0.95]
5.4 Clinic visit, antimalarial alone vs. placebo25850Risk Ratio (Fixed, 95% CI)0.85 [0.79, 0.92]
Analysis 2.1.

Comparison 2 Iron + antimalarial vs. placebo, Outcome 1 Clinical malaria.

Analysis 2.2.

Comparison 2 Iron + antimalarial vs. placebo, Outcome 2 All-cause mortality.

Analysis 2.3.

Comparison 2 Iron + antimalarial vs. placebo, Outcome 3 Haemoglobin at end of treatment.

Analysis 2.4.

Comparison 2 Iron + antimalarial vs. placebo, Outcome 4 Anaemia.

Analysis 2.5.

Comparison 2 Iron + antimalarial vs. placebo, Outcome 5 Hospitalizations and clinic visits.

Comparison 3. Iron + antimalarial vs. antimalarial alone for proven malaria
Outcome or subgroup titleNo. of studiesNo. of participantsStatistical methodEffect size
1 Parasitological failure3583Risk Ratio (M-H, Random, 95% CI)0.98 [0.69, 1.39]
2 All-cause mortality3584Risk Ratio (M-H, Fixed, 95% CI)1.42 [0.24, 8.44]
3 Pneumonia1288Risk Ratio (M-H, Fixed, 95% CI)5.2 [2.05, 13.16]
4 Hospitalizations and clinic visits2 Risk Ratio (Fixed, 95% CI)Subtotals only
4.1 Hospitalization1288Risk Ratio (Fixed, 95% CI)2.33 [0.90, 6.07]
4.2 Clinic visit1273Risk Ratio (Fixed, 95% CI)0.65 [0.29, 1.46]
5 Haemoglobin at end of treatment (for calculation)196Mean Difference (IV, Fixed, 95% CI)0.36 [-0.28, 1.01]
Analysis 3.1.

Comparison 3 Iron + antimalarial vs. antimalarial alone for proven malaria, Outcome 1 Parasitological failure.

Analysis 3.2.

Comparison 3 Iron + antimalarial vs. antimalarial alone for proven malaria, Outcome 2 All-cause mortality.

Analysis 3.3.

Comparison 3 Iron + antimalarial vs. antimalarial alone for proven malaria, Outcome 3 Pneumonia.

Analysis 3.4.

Comparison 3 Iron + antimalarial vs. antimalarial alone for proven malaria, Outcome 4 Hospitalizations and clinic visits.

Analysis 3.5.

Comparison 3 Iron + antimalarial vs. antimalarial alone for proven malaria, Outcome 5 Haemoglobin at end of treatment (for calculation).

What's new

DateEventDescription
6 August 2009AmendedError in data for graph 1.34 corrected for Issue 4, 2009: Thanks to an observant reader, we identified a log conversion error for two graphs 1.34.3 and 1.34.4. The data were correct, but we had not carried out the appropriate anti-logging and this gave graphs suggesting unrealistically large effects which should have been picked up by the Editor. This has now been corrected.

History

Protocol first published: Issue 3, 2007
Review first published: Issue 3, 2009

Contributions of authors

Juliana U Ojukwu: conceived the idea for the review, wrote the protocol, identified studies for inclusion/exclusion, extracted the data, entered data to RevMan, participated in data analysis, and reviewed all drafts and the final review.

Joseph Okebe: wrote the protocol, identified studies for inclusion/exclusion, extracted the data, entered data to RevMan, participated in data analysis, and reviewed all drafts and the final review.

Dafna Yahav: extracted the data from all included studies, entered data to RevMan, participated in data analysis, and reviewed all drafts and the final review.

Mical Paul: planned data extraction, extracted the data, entered data to RevMan, data analysis, and wrote the review.

Declarations of interest

All authors - none declared.

Sources of support

Internal sources

  • UK Department for International Development (DFID), UK.

    The editorial base for the Cochrane Infectious Diseases Group is funded by the UK Department for International Development (DFID) for the benefit of developing countries. Dafna Yahav received funding from the editorial base.

External sources

  • The Nuffield Foundation, Afghanistan.

    Dr Juliana U Ojukwu was awarded a Reviews for Africa Programme Fellowship (www.mrc.ac.za/cochrane/rap.htm), funded by a grant from the Nuffield Commonwealth Programme, through The Nuffield Foundation.

Differences between protocol and review

Anaemia was removed from the list of primary outcomes to the list of secondary outcomes.

The secondary outcome: 'Iron levels (e.g. serum iron, ferritin levels, total iron binding capacity, zinc protoporphyrin concentration, zinc level)' is not included in the current version of the review and will be added to the updated versions of the review. These outcomes were extracted but reporting was sparse.

Characteristics of studies

Characteristics of included studies [ordered by study ID]

Adam 1997 (C)

Methods

Cluster-randomized

Trial years: May 1993 to October 1995

Unit of randomization: household

Number of units randomized: not stated

Average cluster size: not stated

Adjustment for clustering: none

Methods of adjustment: not stated

Participants

Number of children: 841 randomized, 738 evaluated

Age: mean 45.2 months (range 6 to 84 months)

Setting: school, rural

% anaemic at baseline: 100% (anaemia definition: Hb < 10.9 g/dL), mean haemoglobin (SD) at baseline: iron arm: 8.27 (1.2) g/dL; placebo: 8.27 (1.3) g/dL

% malaria at baseline: 12.35%

Interventions

Ferrous sulfate elixir, about 3 mg/kg/day elemental iron vs. placebo elixir

Duration of treatment: 12 weeks

Duration of follow up: 12 months

Outcomes

Main objective/outcome: Effect of iron supplementation on malaria

Review outcomes reported in the trial:

1. Clinical malaria, parasitaemia, severe malaria, parasite density

2. Anaemia

3. Hospitalization

4. Haemoglobin (end and change)

5. All infections, diarrhoea

Notes

Trial location: north-western Ethiopia, Shehdi town, and Aftit village

Malaria endemicity: mesoendemic (trial included the rainy season)

Language of publication: English

Exclusion criteria: Hb < 6 and Hb > 11, debilitating chronic disease or acute infection, new residents or about to leave the region

PHD dissertation

Risk of bias
ItemAuthors' judgementDescription
Adequate sequence generation?YesRandom number tables. Done in random permuted blocks of 4 households
Allocation concealment?UnclearNo description
Blinding?
All outcomes
YesSame bottles as intervention used for placebo elixir. Participants and those who supplied the medications were blinded
Incomplete outcome data addressed?
Mortality
UnclearNot evaluated
Incomplete outcome data addressed?
Malaria
No738/841 evaluated
Incomplete outcome data addressed?
Haemoglobin or anaemia
No742/841 evaluated

Aggarwal 2005

Methods

Individual randomized

Trial years: April 1998 to February 1999

Participants

73 randomized, 62 evaluated

Age: mean 57.4 days (range 50 to 80 days), all were predominantly breast fed

Setting: community, urban

% anaemic at baseline: not given, baseline definition of anaemia not stated, mean haemoglobin (SD) at baseline: iron arm: 11.5 (1.3) g/dL; placebo: 11.7 (1.2) g/dL

% malaria at baseline: not stated

Interventions

Ferric ammonium citrate oral drops, about 3 mg/kg/day elemental iron vs. daily oral placebo drops

Duration of treatment: 8 weeks

Duration of follow up: 4 weeks

Outcomes

Main objective/outcome: The haematological utility of iron supplementation in predominantly breast fed term low birth weight young infants

Review outcomes reported in the trial:

1. Haemoglobin (change)

2. Weight and height

Notes

Trial location: New Delhi, India

Malaria endemicity: mesoendemic

Language of publication: English

Exclusion criteria: twinning, congenital malformations, past blood transfusion, adverse neonatal events requiring hospitalization, past blood sampling (> 10 ml), receiving iron supplementation, significant current morbidity, and maternal antepartum haemorrhage

Risk of bias
ItemAuthors' judgementDescription
Adequate sequence generation?YesComputer generated random numbers
Allocation concealment?YesCentral, sealed and opaque envelopes
Blinding?
All outcomes
YesStated as double blind
Incomplete outcome data addressed?
Mortality
UnclearNot evaluated
Incomplete outcome data addressed?
Malaria
UnclearNot evaluated
Incomplete outcome data addressed?
Haemoglobin or anaemia
No62/73 evaluated

Aguayo 2000

Methods

Individual randomized

Trial years: not stated

Participants

73 randomized, 64 evaluated

Age: mean 110.9 months

Setting: school, urban

% anaemic at baseline: 0% (anaemia definition: Hb < 14.4 g/dL; cut-off value recommended for pre-school children living 4000m above sea level), mean haemoglobin (SD) at baseline: iron arm: 15.66 (0.60) g/dL; placebo: 15.74 (0.63) g/dL

% malaria at baseline: not stated

Interventions

Study arms:

1. Iron: ferrous sulfate tablet, about 0.43 mg/kg/d elemental iron (3 mg/kg/week)

2. Placebo: 1 tablet x 1/week

Duration of treatment: 18 weeks

Duration of follow up: 18 weeks

Outcomes

Main objective/outcome: Effect of weekly iron on growth and haemoglobin status in non-anaemic children

Review outcomes reported in the trial:

1. Anaemia

2. Haemoglobin (change)

3. Weight and height

Notes

Trial location: outskirts of La-Paz, Bolivia

Malaria endemicity: hypoendemic

Language of publication: English

Exclusion criteria: haemoglobin < 14.4 g/dL

Risk of bias
ItemAuthors' judgementDescription
Adequate sequence generation?YesTable with randomly assorted digits
Allocation concealment?UnclearNo description
Blinding?
All outcomes
YesStated as double blind
Incomplete outcome data addressed?
Mortality
UnclearNot evaluated
Incomplete outcome data addressed?
Malaria
UnclearNot evaluated
Incomplete outcome data addressed?
Haemoglobin or anaemia
No64/73 evaluated

Akenzua 1985

Methods

Individual randomized

Trial years: not stated

Participants

112 randomized, 97 evaluated

Age: range 1 to 14 years

Setting: community, rural

% anaemic at baseline: 100% (anaemia definition: packed cell volume (PCV) < 33%), mean haemoglobin 10 g/dL

% malaria at baseline: not stated

Interventions

Study arms:

1. Unsupervised administration of: ferrous fumarate tablets, about 2 mg/kg/d + folic acid tablets 5 mg/d + antimalaria (single dose of 5 mg/kg chloroquine orally)

2. Unsupervised administration of: ferrous fumarate syrup, about 1.5 mg/kg/d + folic acid tablets 5 mg/d + antimalaria (single dose of 5 mg/kg chloroquine orally)

3. Supervised administration of: ferrous fumarate tablets, about 2 mg/kg/d + folic acid tablets 5 mg/d + antimalaria (single dose of 5 mg/kg chloroquine orally)

4. Proguanil hydrochloride tablets, 50 mg daily

5. Folic acid + chloroquine

6. Iron IM + chloroquine

7. Iron IM + chloroquine + folic acid

Duration of treatment: 6 weeks

Duration of follow up: 6 weeks

Outcomes

Main objective/outcome:

To determine more accurately the extent to which folate deficiency contributes to the anaemia of childhood in the community; to find out how the prevalence of anaemia in children can be reduced by 50 % or more; to decide on a cheap and effective supplementation programme as a public health measure applicable in the community

Review outcomes reported in the trial:

1. Anaemia

2. Haemoglobin (PCV) change

Notes

Trial location: Nigeria

Malaria endemicity: hyperendemic

Language of publication: English

Exclusion criteria: haemoglobinopathies (SC, SS); refusal of consent

Risk of bias
ItemAuthors' judgementDescription
Adequate sequence generation?UnclearPrepared set of random numbers
Allocation concealment?UnclearNot described
Blinding?
All outcomes
NoOpen
Incomplete outcome data addressed?
Mortality
UnclearNot evaluated
Incomplete outcome data addressed?
Malaria
UnclearNot evaluated
Incomplete outcome data addressed?
Haemoglobin or anaemia
No97/112 evaluated

Angeles 1993

Methods

Individual randomized

Trial years: not stated

Participants

80 randomized, 76 evaluated

Age: mean 37.3 months (range 2 to 5 years)

Setting: school, urban

% anaemic at baseline: 100% (anaemia definition: Hb < 11 g/dL), mean haemoglobin (SD) at baseline: iron arm: 10.2 (0.9) g/dL; placebo arm: 10.3 (0.8) g/dL

% malaria at baseline: not stated

Interventions

Study arms:

1. Iron: hydrated ferrous sulfate powder, about 2.8 mg/kg/d elemental iron plus vitamin C 20 mg daily

2. Vitamin C 20 mg daily

Duration of treatment: 8 weeks

Duration of follow up: 8 weeks

Outcomes

Main objective/outcome:

Effect of iron supplementation on growth and haematological status

Review outcomes reported in the trial:

1. Fever, respiratory infections, diarrhoea

2. Haemoglobin status (end and change)

3. Ferritin level

4. Weight and height (end and change)

Notes

Trial location: Jakarta, Indonesia

Malaria endemicity: mesoendemic

Language of publication: English

Exclusion criteria: acute infections

Risk of bias
ItemAuthors' judgementDescription
Adequate sequence generation?UnclearNo description
Allocation concealment?UnclearNo description
Blinding?
All outcomes
YesStated as double blind
Incomplete outcome data addressed?
Mortality
UnclearNot evaluated
Incomplete outcome data addressed?
Malaria
UnclearNot evaluated
Incomplete outcome data addressed?
Haemoglobin or anaemia
No76/80 evaluated

Baqui 2003

Methods

Individual randomized

Trial years: not stated

Participants

645 randomized, 249 evaluated

Age (per study): 5 to 6 months at enrolment

Setting: community

None were anaemic at baseline (anaemia: Hb < 9 g/dL), mean haemoglobin 10.5 g/dL

% malaria at baseline not described

Interventions

Type of iron: ferrous sulphate capsules orally about 0.43 mg/kg/day (20 mg elemental iron weekly) + riboflavin 1 mg/week vs. zinc + riboflavin vs. ferrous sulphate + zinc vs. micronutrient mix (not used in review) vs. riboflavin alone. Duration of treatment duration: 6 months

Duration of treatment: 6 months

Duration of follow up: 6 months

Outcomes

Main objective/outcome:

The effect of iron and/or zinc supplementation on rates of growth during 6 months of supplementation

Review outcomes reported in the trial:

1. Deaths

2. Any infection

3. Change in haemoglobin

4. Ferritin

5. Weight and height

6. Adverse events

Notes

Trial location: Bangladesh

Malaria endemicity: mesoendemic

Language of publication: English

Exclusion criteria: low weight-for-age, Hb < 9 g/dL, any signs of neurological disorder, physical handicap, or chronic illness affecting feeding, activity or cognitive development

Risk of bias
ItemAuthors' judgementDescription
Adequate sequence generation?UnclearBlock randomization was reportedly done
Allocation concealment?UnclearNot described
Blinding?
All outcomes
YesStated as double blind
Incomplete outcome data addressed?
Mortality
YesAll randomized patients included in analysis
Incomplete outcome data addressed?
Malaria
UnclearNot relevant
Incomplete outcome data addressed?
Haemoglobin or anaemia
NoData available for 249/645 participants

Berger 1997

Methods

Individual randomized

Trial years: March 1995 to November 1995

Participants

176 randomized, 173 evaluated

Age: 3.3 to 8.3 years, 96.3% were 4 to 6.9 years. Mean age per study arm: weekly iron: 67.8 SD 7.6 months, daily iron:69.6 SD 12.2 months, placebo: 67.2 SD 9.3 months

Setting: school

100% anaemia (Hb <11 g/dL) at baseline

% malaria at baseline: not stated

Interventions

Ferrous sulfate tablet 3 to 4 mg/kg/dose weekly vs. ferrous sulfate 3 to 4 mg/kg/daily 5 days a week vs. placebo once weekly

Duration of treatment duration: 16 weeks

Duration of follow up: 16 weeks 

Outcomes

Main objective/outcome:

To compare the efficacy of weekly and daily iron on haemoglobin status of anaemic children

Review outcomes reported in the trial:

1. Anaemia

2. Haemoglobin (end and change)

3. Protoporphyrin level

4. Weight and height (end)

Notes

Trial location: La-Paz, Bolivia

Malaria endemicity: hypoendemicity

Language of publication: English

Exclusion criteria: Hb > 14.4 g/dL

Risk of bias
ItemAuthors' judgementDescription
Adequate sequence generation?UnclearChildren were randomized into 3 groups
Allocation concealment?UnclearNo description
Blinding?
All outcomes
YesStated as double blind
Incomplete outcome data addressed?
Mortality
UnclearNot evaluated
Incomplete outcome data addressed?
Malaria
UnclearNot evaluated
Incomplete outcome data addressed?
Haemoglobin or anaemia
NoData available for 173/176 participants

Berger 2000

Methods

Individual randomized

Trial years: not stated

Participants

197 randomized, 163 evaluated

Age: 6 to 36 months. Mean age per study arm: intervention: 22.8 SD 8.42 months, placebo: 24.9 SD 8.3 months

Setting: community

% Anaemia (Hb < 11 g/dL) at baseline: iron arm: 84.5% placebo arm: 79.8%, mean haemoglobin: iron arm: 9.89 SD 1.16 g/dL, placebo arm:10.04 SD 1.06 g/dL

% malaria at baseline: iron arm: 59.3, placebo arm: 63.6

Interventions

Iron betainate tablet 2 to 3 mg/kg/day elemental iron vs. placebo

Duration of treatment: 3 months

Duration of follow up: 9 months

Outcomes

Main objective/outcome:

Impact of iron supplementation on haematological status, cell-mediated immunity and susceptibility to infections

Review outcomes reported in the trial:

1. Parasitaemia (% plasmodic indice), parasitaemia > 3000, malaria density

2. Anaemia

3. Diarrhoea

4. Respiratory infections

5. Haemoglobin (end and change)

6. Ferritin, total iron binding capacity (TIBC), protoporphyrin

Notes

Trial location: sea region, Togo

Malaria endemicity: hyperendemicity

Language of publication: English

Exclusion criteria: Hb < 8 g/dL

Risk of bias
ItemAuthors' judgementDescription
Adequate sequence generation?UnclearRandomized assignment of children into an intervention and placebo groups
Allocation concealment?UnclearNo description
Blinding?
All outcomes
YesStated as double blind
Incomplete outcome data addressed?
Mortality
No163 out of 197 participants were evaluated
Incomplete outcome data addressed?
Malaria
No163 out of 197 participants were evaluated
Incomplete outcome data addressed?
Haemoglobin or anaemia
No163 out of 197 participants were analysed

Berger 2006

Methods

Individual randomized

Trial years: March 1998 to November 1998

Participants

988 randomized. 760 to 780 (depending on outcome assessed) evaluated

Age: mean 5.9 months (range: 4 to 7 months)

Setting: community, rural

% anaemia at baseline: 54.1% (defined Hb < 11 g/dL), mean haemoglobin 10.9 g/dL

% malaria at baseline: not stated

Interventions

Ferrous sulphate syrup 10 mg/day (about 1.5 mg/kg/d elemental iron) vs. zinc vs. ferrous sulphate + zinc vs. placebo. 100,000 IU of vitamin A was given to all infants at the start of the study.

Duration of treatment: 6 months

Duration of follow up: 6 months

Outcomes

Main objective/outcome:

To evaluate the effect of combined iron–zinc supplementation on micronutrient status, growth and morbidity

Review outcomes reported in the trial:

1. Anaemia

2. Any infection

3. Respiratory infections

4. Diarrhoea

5. Haemoglobin (end and change)

6. Ferritin, zinc, TIBC

7. Weight and height

Notes

Location: district of Que Vo, 50 km northwest of Hanoi in the Red River Delta in Vietnam

Malaria endemicity: hyperendemic

Language of publication: English

Exclusion criteria: chronic or acute illness, severe malnutrition or congenital abnormality, Hb < 7 g/dL

Risk of bias
ItemAuthors' judgementDescription
Adequate sequence generation?YesComputer-generated
Allocation concealment?UnclearNo description
Blinding?
All outcomes
YesStated as double blind
Incomplete outcome data addressed?
Mortality
UnclearNot reported
Incomplete outcome data addressed?
Malaria
UnclearNot reported
Incomplete outcome data addressed?
Haemoglobin or anaemia
No864/ 988 participants were evaluated

Bhatia 1993 (C)

Methods

Cluster-randomized

Trial years: not stated

Unit of randomization: 4 preschools

Average cluster size: 65.5

Adjustment for clustering: none

Methods of adjustment: not stated

Participants

4 pre-schools randomized (262 participants), 156 participants evaluated

Age: 3 to 5 years. Mean age per study arm:

Iron anaemic: 46.8 months

Iron non-anaemia: 58.3 months

Placebo anaemic: 46.2 months

Placebo non-anaemic: 49.6 months

Malaria endemicity: mesoendemicity

Setting: preschool

% Anaemia (Hb < 11 g/dL; Hb 7 to 10 moderate anaemia) at baseline: iron anaemic:100%, iron non-anaemia: 0.0%, placebo anaemic:100.0%, placebo non-anaemic: 0.0%, mean haemoglobin: iron anaemic: 9.2 SE 0.1, iron non-anaemia: 11.6 SE 0.1, placebo anaemic: 9.4 SE 8.1, placebo non-anaemic: 11.5 SE -0.2

% malaria at baseline: not stated

Interventions

Intervention arm:

Iron 40 mg elemental iron per day (3 mg/kg/day) for anaemic children vs. iron for non-anaemic children vs. placebo for anaemic children vs. placebo for non-anaemic children

Duration of treatment: 6 months

Duration of follow up: 6months

Outcomes

Main objective/outcome:

1. To determine the growth status of moderately anaemic and non-anaemic (normal) young children living under similar environmental conditions

2. To evaluate the growth status of anaemic and normal children before and after supplementation with iron

Review outcomes reported in the trial:

1. Haemoglobin level (end and change)

2. Weight and height (end)

Notes

Location: Baroda City, India

Malaria endemicity: mesoendemic

Language of publication: English

Exclusion criteria: mild anaemia Hb <10.1 to 10.9 g/dL

Risk of bias
ItemAuthors' judgementDescription
Adequate sequence generation?Unclear2 out of 4 classes were randomly selected for treatment
Allocation concealment?UnclearNo description
Blinding?
All outcomes
NoControl different from intervention
Incomplete outcome data addressed?
Mortality
UnclearNot evaluated
Incomplete outcome data addressed?
Malaria
UnclearNo description
Incomplete outcome data addressed?
Haemoglobin or anaemia
No156 out of 262 participants were evaluated

Boivin 1993

Methods

Individual randomized

Trial years: not reported

Participants

50 randomized, 47 evaluated

Age (per study): means for boys and girls were defined separately as 7 and 8 years respectively

Setting: school

Available mean Hb at baseline: 10.3 g/dL (SD 0.86)

% malaria at baseline: all malaria cases were excluded

Interventions

Iron syrup (formulation not stated) 20 ml/day (about 0.8 mg/kd/day elemental iron) vs. placebo. Children were also randomized to tetramisole /levamisole vs placebo

Duration of treatment: 4 weeks

Duration of follow up: 4 weeks

Outcomes

Main objective/outcome:

To evaluate the effects of both deworming and iron supplements separately and in combination against the backdrop of the health, socioeconomic, and nutritional characteristics

Review outcomes reported in the trial:

None reported

Notes

Trial location: Zaire

Malaria endemicity: hyperendemic

Language of publication: English

Exclusion criteria: presence of trophozoites of malaria

Risk of bias
ItemAuthors' judgementDescription
Adequate sequence generation?UnclearRandomly selected
Allocation concealment?UnclearNot described
Blinding?
All outcomes
NoOpen study
Incomplete outcome data addressed?
Mortality
UnclearNot reported
Incomplete outcome data addressed?
Malaria
UnclearNot reported
Incomplete outcome data addressed?
Haemoglobin or anaemia
UnclearNot reported

Charoenlarp 1973

Methods

Individual randomized

Trial years: not reported

Participants

460 randomized, 437 evaluated

Age: 5 to 14 years

Setting: school, rural

Mean HCT at baseline: placebo: 36.3 SD 2.7%, iron: 36.4 SD 2.6%, folic acid: 36.2 SD 2.6%, iron + folic acid: 36.8 SD 2.4% months

% anaemia at baseline: 52%

% malaria at baseline: not stated

Interventions

Ferrous gluconate tablet once daily 5 days a week (about 1.1 mg/kd/d) vs. placebo vs. ferrous gluconate + folic acid 10 mg once daily 5 days a week vs. folic acid

Duration of treatment: 3 months

Duration of follow up: 3 months

Outcomes

Main objective/outcome:

To evaluate the haemoglobin values after iron and folic acid supplementation

Review outcomes reported in the trial:

1. Haemoglobin (change)

Notes

Trial location: Thailand

Malaria endemicity: mesoendemic

Language of publication: English

Exclusion criteria: haematocrit levels within 3 SD of the initial mean values

Risk of bias
ItemAuthors' judgementDescription
Adequate sequence generation?UnclearNo description
Allocation concealment?UnclearNo description
Blinding?
All outcomes
NoOpen
Incomplete outcome data addressed?
Mortality
UnclearNot evaluated
Incomplete outcome data addressed?
Malaria
UnclearNot evaluated
Incomplete outcome data addressed?
Haemoglobin or anaemia
No437 out of 460 children evaluated

Chwang 1988

Methods

Individual randomized

Trial years: not stated

Participants

241 participants randomized and evaluated

Age range 8.2 to 13.5 years

Setting: community

Participants were stratified by anaemia status into 2 intervention groups, mean Hb for anaemic group was 9.7 g/dL and the non-anaemic group was 13.25 g/dL

% malaria at baseline: not stated

Interventions

Ferrous sulphate tablets 10 mg/day (2 mg/kg/day elemental iron) vs. saccharin + tapioca (placebo)

Duration of treatment: 12 weeks

Duration of follow up: 12 weeks

Outcomes

Main objective/outcome:

Effect of oral iron on blood iron levels and growth

Review outcomes reported in the trial:

1. Haemoglobin (end)

2. Iron levels (end), TIBC

3. Weight and height (change, absolute values)

Notes

Trial location: 3 villages in subdistrict of Kalibawang in central Java, Indonesia

Malaria endemicity: mesoendemic

Language of publication: English

Exclusion criteria: Hb 11.1 to 11.9 g/dL

Risk of bias
ItemAuthors' judgementDescription
Adequate sequence generation?UnclearNot described
Allocation concealment?UnclearNot described
Blinding?
All outcomes
YesStated as double blind but method not described
Incomplete outcome data addressed?
Mortality
UnclearNot evaluated
Incomplete outcome data addressed?
Malaria
UnclearNot evaluated
Incomplete outcome data addressed?
Haemoglobin or anaemia
YesAll participants evaluated

de Silva 2003

Methods

Individual randomized

Trial years: not stated 

Participants

453 randomized, 363 evaluated

Age: 5 to 10 years

Setting: hospital

% anaemia (Hb < 11.5 g/dL) at baseline: 52.6%, mean haemoglobin 11.4 g/dL

% malaria at baseline: not reported

Interventions

Ferrous sulfate 60 mg/day elemental iron (3 mg/kg/day) vs. placebo

Duration of treatment: 8 weeks

Duration of follow up: 8 weeks

Outcomes

Main objective/outcome:

To evaluate the effects of iron supplementation on iron status and morbidity in children with or without infection

Review outcomes reported in the trial:

Anaemia (end)

1. Diarrhoea, respiratory infections

2. Haemoglobin (change)

3. Ferritin

4. Adverse events

Notes

Trial location: 1 village in Sri Lanka

Malaria endemicity: mesoendemic

Language of publication: English

Exclusion criteria: Hb < 7 g/dL, severe malnutrition, asthma, chronic diarrhoea 

Risk of bias
ItemAuthors' judgementDescription
Adequate sequence generation?UnclearChildren were stratified by age and sex and were randomly assigned to receive iron or placebo on a 3:1 basis within each stratum
Allocation concealment?UnclearNot described
Blinding?
All outcomes
YesStated as double blind
Incomplete outcome data addressed?
Mortality
UnclearNot evaluated
Incomplete outcome data addressed?
Malaria
UnclearNot evaluated
Incomplete outcome data addressed?
Haemoglobin or anaemia
No363 out of 453 participants evaluated

Desai 2003

Methods

Individual randomized

Trial years: April to November 1999

Participants

546 randomized, 491 evaluated

Age range 2 to 36 months, average mean for all groups is 11.6 months

Setting: community

100% anaemic at baseline (defined as Hb < 11 g/dL), mean haemoglobin 9.5 g/dL

20% to 28% malaria prevalence at baseline

Interventions

Ferrous sulfate suspension (40 mg/ml) 3 to 6mg/kg/day elemental iron + sulfadoxine-pyrimethamine 25/2.25 mg as a single dose at baseline, week 4 and 8 (IPT) vs. IPT vs. ferrous sulfate + sulfadoxine-pyrimethamine 25/2.25 mg as a single dose at baseline vs. placebo + sulfadoxine-pyrimethamine 25/2.25 mg as a single dose at baseline

Duration of treatment: 8 weeks

Duration of follow up: 24 weeks

Outcomes

Main objective/outcome: 

The efficacy of single and combined therapy with iron supplementation and IPT with SP in improving haemoglobin concentrations among anaemic preschool children

Review outcomes reported in the trial:

1. Deaths

2. Clinical malaria, parasitaemia, malaria density

3. Anaemia

4. Haemoglobin (end)

5. Clinic visits

Notes

Trial location: 15 villages in Asembo, Bondo district, Western Kenya

Malaria endemicity: hyperendemic

Language of publication: English

Exclusion criteria: parasite count > 20,000, Hb SS phenotype

Risk of bias
ItemAuthors' judgementDescription
Adequate sequence generation?YesRandom number listing generated independently before the study
Allocation concealment?UnclearNot described
Blinding?
All outcomes
YesStated as double blind
Incomplete outcome data addressed?
Mortality
No491/554 participants evaluated
Incomplete outcome data addressed?
Malaria
No8 were excluded before the first dose of interventions, 55 were lost to follow up, 13 died by week 24. 491/554 participants evaluated.
Incomplete outcome data addressed?
Haemoglobin or anaemia
No428/554 participants evaluated

Devaki 2007

Methods

Individual randomized

Trial years: not stated

Participants

120 randomized, 115 evaluated

Age range: 15 to 18 years

Setting: school, urban

Four study groups. Haemoglobin at baseline: 1) placebo mean 13.6 SD 0.2 2) iron: mean 13.5 SD 0.2 3) iron supplementation for iron deficient children: mean 12.5 SD 0.2 4) iron supplementation for children with iron-deficiency anaemia (100% anaemic at baseline) mean 10.0 SD 0.2. Anaemia defined as Hb < 11 g/dL for boys, < 10.5 g/dL for girls and transferrin saturation < 16% for both sexes.

% malaria at baseline: not stated

Interventions

Iron (III)-hydroxide polymaltose complex 100 mg/day, 6 days a week (about 2.2 mg/kd/day elemental iron) vs. placebo vs. iron for iron deficient children vs. iron for iron deficient anaemia children

Duration of treatment: 8 months

Duration of follow up: 8 months

Outcomes

Main objective/outcome:

To evaluate the effects of iron supplementation on immunological parameters

Review outcomes reported in the trial:

1. Haemoglobin (end)

2. Ferritin, TIBC

3. Adverse effects

Notes

Trial location: Tirupati (Andhra Pradesh State), South India

Malaria endemicity: mesoendemic

Language of publication: English

Exclusion criteria: haemoglobin per group

Risk of bias
ItemAuthors' judgementDescription
Adequate sequence generation?UnclearNo description
Allocation concealment?UnclearNo description
Blinding?
All outcomes
NoOpen
Incomplete outcome data addressed?
Mortality
UnclearNot reported
Incomplete outcome data addressed?
Malaria
UnclearNot reported
Incomplete outcome data addressed?
Haemoglobin or anaemia
No115 out of 120 children evaluated (2 and 3 drop-outs in the ID and IDA groups)

Dossa 2001a

Methods

Individual randomized

Trial years: not stated

Participants

177 participants randomized

Age range 3 to 5 years. Mean 46 months.

Setting: community, rural

76% anaemic at baseline, mean haemoglobin 10.5 g/dL

% malaria at baseline: not stated

Interventions

Ferrous sulphate 60 mg/day elemental iron (about 4.6 mg/kg/d) + albendazole 200 mg/day for 3 days, 1 month later same dose vs. ferrous sulphate + placebo + albendazole + placebo vs. placebo + placebo

Duration of treatment: 3 months

Duration of follow: 10 months

Outcomes

Main objective/outcome:

The effects of iron and deworming treatments on appetite and physical growth performance in preschool children

Review outcomes reported in the trial:

1. Deaths

2. Haemoglobin (end and change)

3. Weight and height

Notes

Trial location: Agblangandan, South Benin 10 km from Cotonou, Benin

Malaria endemicity: hyperendemic

Language of publication: English

Exclusion criteria: not stated

Risk of bias
ItemAuthors' judgementDescription
Adequate sequence generation?UnclearChildren were selected and randomly assigned to 4 treatment groups
Allocation concealment?UnclearNo description
Blinding?
All outcomes
YesDouble blind
Incomplete outcome data addressed?
Mortality
No138/177 participants evaluated
Incomplete outcome data addressed?
Malaria
UnclearNot reported
Incomplete outcome data addressed?
Haemoglobin or anaemia
No138/177 participants evaluated

Dossa 2001b

Methods

Individual randomized

Trial years: not stated

Participants

154 participants randomized, but only 76 in the relevant intervention groups, 74 were evaluated

Age range 3 to 30 months, mean 22 months

Setting: community

100% anaemic at baseline, mean haemoglobin 9.5 g/dL

% malaria at baseline: not stated

Interventions

Ferrous fumarate 66 mg/day elemental iron (about 7.3 mg/kg/d) vs. placebo (Seresta forte). Both arms received mebendazole 200 mg/d for 3 days

Duration of treatment duration: 6 weeks

Duration of follow up: 5.5 months

Outcomes

Main objective/outcome:

The effects of iron and deworming treatments on physical growth performance, haemoglobin level, and intestinal helminth egg loads in preschool children

Review outcomes reported in the trial:

1. Deaths

2. Haemoglobin (end and change)

3. Fever, diarrhoea

4. Weight and height change

Notes

Trial location: Ze, south Benin 50 km from Cotonou, Benin

Malaria endemicity: hyperendemic

Language of publication: English

Exclusion criteria: not stated

Risk of bias
ItemAuthors' judgementDescription
Adequate sequence generation?YesTable of random numbers
Allocation concealment?YesA researcher not involved in the trial allocated children by the randomization code
Blinding?
All outcomes
YesDouble blind, placebo used
Incomplete outcome data addressed?
Mortality
No74/76 participants evaluated
Incomplete outcome data addressed?
Malaria
UnclearNot reported
Incomplete outcome data addressed?
Haemoglobin or anaemia
No74/76 participants evaluated

Fahmida 2007

Methods

Individual randomized

Trial years: July 1998 until March 1999

Participants

800 participants randomized, but only 392 in the relevant intervention groups. All were evaluated

Age range 3 to 6months, mean 5.1 ± 1.1 months

Setting: community

83% anaemic at baseline, mean haemoglobin 9.6 g/dL

% malaria at baseline: not stated

Interventions

Iron sulfate syrup 10 mg/day (about 2 mg/kg/day elemental iron) + zinc sulfate vs. zinc sulfate vs. iron + zinc + vitamin (not used in review) vs. placebo (not used in review)

Duration of treatment duration: 6 months

Duration of follow up: 12 months

Outcomes

Main objective/outcome:

To investigate the effect of supplementation on improving infants' micronutrient status and linear growth

Review outcomes reported in the trial:

1. Clinical malaria

2. Deaths

3. Anaemia

4. Fever, diarrhoea, pneumonia

5. Haemoglobin (end and change)

6. Ferritin, TIBC

7. Weight and height (end)

Notes

Trial location: East Lombok, West Nusa Tenggara, Indonesia

Malaria endemicity: mesoendemic

Language of publication: English

Exclusion criteria: congenital abnormalities, Hb < 6 g/dL

Risk of bias
ItemAuthors' judgementDescription
Adequate sequence generation?YesAllocation to supplementation groups was conducted using systematic random sampling in each sex group. The randomization of the subjects in the study was done (1) firstly, by assigning for each intervention group A-D codes (the A, B, C, D codes were randomly assigned to placebo, Zn, Zn + Fe and Zn + Fe + Vit A groups), then (2) secondly, each child was randomly assigned to each A-D using systematic random sampling
Allocation concealment?YesCentral
Blinding?
All outcomes
YesStated as double blind
Incomplete outcome data addressed?
Mortality
YesAll participants evaluated
Incomplete outcome data addressed?
Malaria
No314/392 participants evaluated
Incomplete outcome data addressed?
Haemoglobin or anaemia
No308/392 participants evaluated

Gebresellassie 1996

Methods

Individual randomized

Trial years: February 1994 until July 1994

Participants

500 participants randomized, 480 evaluated

Age range 5 to 14 years, mean 10.3 years

Setting: school

91% anaemic (Hb < 12 g/dL) at baseline, mean haemoglobin 9.5 g/dL

% malaria at baseline: 98% with >= 1 episodes) of malaria attack in the past 14 days; negative malaria smears on initial screening for all

Interventions

Ferrous sulphate 60 mg/day elemental iron (about 2.5 mg/kg/day) vs. placebo

Duration of treatment duration: 3 months

Duration of follow up: 6 months

Outcomes

Main objective/outcome:

To assess the effect of oral iron on host susceptibility to malaria infection in children with mild-moderate IDA

Review outcomes reported in the trial:

1. Clinical malaria, cumulative incidence of parasitaemia, parasite density, parasitaemia > 5000/ul

2. Deaths

3. Anaemia

4. Haemoglobin (end)

5. Ferritin

Notes

Trial location: Northwest Ethiopia, Beles Valley (Pawe), Ethiopia

Malaria endemicity: mesoendemic

Language of publication: English

Exclusion criteria: Hb > 12 or < 5, SF > 12, positive malaria smears on initial screening, concurrent major illnesses; on iron supplementation past 6 m, < 12 m residence in the area

Risk of bias
ItemAuthors' judgementDescription
Adequate sequence generation?YesComputer generated code
Allocation concealment?YesCentral list of children merged with the computer generated random list
Blinding?
All outcomes
YesField workers, technicians, parents, and children blinded. Placebo used in coded bottles
Incomplete outcome data addressed?
Mortality
No480/500 participants evaluated
Incomplete outcome data addressed?
Malaria
No480/500 participants evaluated
Incomplete outcome data addressed?
Haemoglobin or anaemia
No480/500 participants evaluated

Gopaldas 1983I

Methods

Individual randomized

Trial years: not stated

Participants

170 participants randomized (90 participants for age group 5 to 9 years)

Age range: 5 to 9 years

Setting: school, urban

Anaemic at baseline: 93% (Indian), 95% (WHO) for 5 to 9 years; (WHO: Hb < 12 g/dL, Indian: Hb < 11 g/dL), mean haemoglobin 9.6 g/dL

Malaria at baseline: not stated

Interventions

Iron 20 mg/day elemental iron (about 0.8 mg/kg/day) + folic acid 100 mcg/ day vs. placebo vs. iron + folic acid + mebendazole 100 mg x 2/day for 3 days (not used in review) vs. iron + folic acid + mebendazole 100 mg x 2/day for 3 days (not used in review) + vitamin A vs. mebendazole + tinidazole 50 mg/kg for 3 days (not used in review)

Duration of treatment: 4 months

Duration of follow up: 12 months

Outcomes

Main objective/outcome:

To evaluate the feasibility, efficiency, nutritional impact, and cost of delivering differential packages of health and nutrient inputs for 2 school terms

Review outcomes reported in the trial:

1. Haemoglobin (end)

Notes

Trial location: Baroda, India

Malaria endemicity: mesoendemic

Language of publication: English

Exclusion criteria: boys, income, no consent, participation in another nutritional programme, age unknown

Risk of bias
ItemAuthors' judgementDescription
Adequate sequence generation?UnclearNot described
Allocation concealment?UnclearNot described
Blinding?
All outcomes
NoOpen
Incomplete outcome data addressed?
Mortality
UnclearNot reported
Incomplete outcome data addressed?
Malaria
UnclearNot reported
Incomplete outcome data addressed?
Haemoglobin or anaemia
YesIntention-to-treat analysis

Gopaldas 1983II

Methods

Individual randomized

Trial years: not stated

Participants

170 participants randomized (80 participants for age group 10 to 13 years)

Age range: 10 to 13 years

Setting: school, urban

92% (Indian), 93% (WHO) for 10 to 13 years anaemic at baseline (WHO: Hb < 12 g/dL, Indian: Hb < 11 g/dL) at baseline, mean haemoglobin 9.6 g/dL

% malaria at baseline: not stated

Interventions

Iron 20 mg/day elemental iron (about 0.5 mg/kg/day) + folic acid 100 mcg/ day vs. placebo vs. iron + folic acid + mebendazole 100 mg x 2/day for 3 days (not used in review) vs. iron + folic acid + mebendazole 100 mg x 2/day for 3 days (not used in review) + vitamin A vs. mebendazole + tinidazole 50 mg/kg for 3 days (not used in review)

Duration of treatment: 4 months

Duration of follow up: 12 months

Outcomes

Main objective/outcome:

To evaluate the feasibility, efficiency, nutritional impact, and cost of delivering differential packages of health and nutrient inputs for 2 school terms

Review outcomes reported in the trial:

1. Haemoglobin (end)

Notes

Trial location: Baroda, India

Malaria endemicity: mesoendemic

Language of publication: English

Exclusion criteria: boys, income, no consent, participation in another nutritional programme, age unknown

Same trial as Gopaldas 1983I

Risk of bias
ItemAuthors' judgementDescription
Adequate sequence generation?UnclearNot described
Allocation concealment?UnclearNot described
Blinding?
All outcomes
NoOpen
Incomplete outcome data addressed?
Mortality
UnclearNot reported
Incomplete outcome data addressed?
Malaria
UnclearNot reported
Incomplete outcome data addressed?
Haemoglobin or anaemia
YesIntention-to-treat

Gopaldas 1985

Methods

Individual randomized

Trial years: not stated

Participants

210 participants randomized

Age range: 8 to 15 years

Setting: school, rural

Hb at baseline: 10.78 SD 0.15 g/dL

% malaria at baseline: not stated

Interventions

Ferrous sulphate 30 mg/day elemental iron vs. ferrous sulfate 40 mg/day elemental iron (1 to 1.3 mg/kd/day) vs. placebo

Duration of treatment: 2 months

Duration of follow up: 4 months

Outcomes

Main objective/outcome:

The effects of iron supplementation on haemoglobin level

Review outcomes reported in the trial:

1. Anaemia

2. Haemoglobin (end and change)

Notes

Trial location: Baroda, Balwadi children, India

Malaria endemicity: mesoendemic

Language of publication: English

Exclusion criteria: not stated

Risk of bias
ItemAuthors' judgementDescription
Adequate sequence generation?UnclearStratified by age and randomized
Allocation concealment?UnclearNot described
Blinding?
All outcomes
NoPatient blinded ("the study was blind. Investigators recognized drugs only by their colour and the children did not know what they are receiving". Intervention tablets were sugar coated).
Incomplete outcome data addressed?
Mortality
UnclearNot reported
Incomplete outcome data addressed?
Malaria
UnclearNot reported
Incomplete outcome data addressed?
Haemoglobin or anaemia
YesIntention-to-treat analysis

Greisen 1986 (C)

Methods

Cluster randomized

Trial years: May to June 1981

Unit of randomization: 12 school classes

Average cluster size: 38.7

Adjustment for clustering: none

Methods of adjustment: none

Participants

12 school classes were divided in 2 equal groups according to their listing on the class registers yielding 24 groups, overall 464 children

Age range: 5 to 15 years

Setting: school, rural

28% anaemic at baseline (Hb < 12 g/dL), mean haemoglobin 12.4 g/dL

% malaria at baseline: not stated

Interventions

Iron-fumarate 66 mg/day on school days (about 2 mg/kd/day elemental iron) + placebo vs. iron-fumarate + chloroquine 300 mg at baseline and 28 days + tetrachlorethylene liquid 2.5 ml at baseline vs. iron-fumarate + chloroquine vs. iron-fumarate + tetrachloroethylene

Duration of treatment: 6 weeks

Duration of follow up: 6 weeks

Outcomes

Main objective/outcome:

To evaluate association between anaemia and running distance

Review outcomes reported in the trial:

1. Deaths

2. Haemoglobin (end and change)

Notes

Trial location: Namwala township in the great plains of the Kafue river, Zambia

Malaria endemicity: hyperendemic

Language of publication: English

Exclusion criteria: acute illness, increased reticulocyte count

Risk of bias
ItemAuthors' judgementDescription
Adequate sequence generation?YesTable of random numbers (12 school classes were divided in 2 equal groups according to their listing on the class registers yielding 24 groups)
Allocation concealment?YesPharmacy
Blinding?
All outcomes
YesDouble blind, placebo used
Incomplete outcome data addressed?
Mortality
No225 out of 464 were evaluated
Incomplete outcome data addressed?
Malaria
UnclearNot reported
Incomplete outcome data addressed?
Haemoglobin or anaemia
No225 out of 464 were evaluated

Hall 2002 (C)

Methods

Cluster-randomized

Trial years: started January 2000

Unit of randomization: school

Number of units randomized: 60 schools

Average cluster size: authors statement "We did not look at size of school or sub-district. But since they were all community schools, they were all small rural schools"

Adjustment for clustering: not mentioned

Methods of adjustment: no adjustment method was used

Participants

Number of children:1201 randomized, 1113 evaluated

Age range: mean 11.4 years range (6 to 19 years)

Setting: school; rural

% anaemic at baseline: 55.8% (anaemia definition: age 5 to 11.99 y Hb <11.5 g/dL, age 12 to 14.99 y Hb < 12 g/dL, > 15 y - boys Hb < 13 g/dL, girls Hb < 12 g/dL), mean haemoglobin 10.5 g/dL

% malaria at baseline: not stated

Interventions

Study arms:

1. Iron: ferrous sulphate tablets, about 0.25 mg/kg/d elemental iron + folic acid + albendazole

2. Control: albendazole only

All children received vitamin A before intervention

Duration of treatment: 10 weeks

Duration of follow up: 2 weeks after end treatment, 14 to 16 weeks from baseline survey weeks

Outcomes

Main objective/outcome:

To assess the effect of weekly iron on haemoglobin status

Review outcomes reported in the trial:

1. Deaths

2. Prevalence of anaemia

3. Haemoglobin (end and change)

4. Growth parameters

5. Adverse events

Notes

Trial location: Kolondieba district in Sikasso region of south eastern Mali

Malaria endemicity: hyperendemic

Language of publication: English

Exclusion criteria: severe anaemia (haemoglobin < 8 g/dL)

Risk of bias
ItemAuthors' judgementDescription
Adequate sequence generation?YesRandom number table
Allocation concealment?UnclearNo description
Blinding?
All outcomes
NoOpen
Incomplete outcome data addressed?
Mortality
No1113/1201 evaluated
Incomplete outcome data addressed?
Malaria
UnclearNot evaluated
Incomplete outcome data addressed?
Haemoglobin or anaemia
No1113/1201 evaluated

Harvey 1989

Methods

Individual randomized

Trial years: started June 1985

Participants

318 randomized, up to 298 evaluated for malaria outcomes, 318 evaluated for haemoglobin

Age: mean 9.7 years (range 8 to 12 years)

Setting: school, rural

% anaemic at baseline: 92% (anaemia definition: Hb < 12 g/dL), mean haemoglobin 10.7 g/dL

% malaria at baseline: 70.5%

Interventions

Study arms:

1. Iron: ferrous sulphate tablets, about 3.8 mg/kg/d elemental iron

2. Placebo: 75% cellulose, 25% lactose tablets

Duration of treatment:16 weeks

Duration of follow up: 24 weeks

Outcomes

Main objective/outcome:

To investigate the effects of iron therapy and changes in iron status on malarial infection in children with mild to moderate iron deficiency and some immunity to malaria

Review outcomes reported in the trial:

1. Malaria (clinical and uncomplicated)

2. Haemoglobin (end and change)

3. Adherence

Notes

Trial location: north coast/Madang, Papua New Guinea

Malaria endemicity: hyperendemic

Language of publication: English

Exclusion criteria: Hb < 8 g/dL or > 12 g/dL, signs of puberty

Risk of bias
ItemAuthors' judgementDescription
Adequate sequence generation?UnclearOf 318 patients authors formed 156 matched pairs based on Hb, age and oval-shaped RBC. Members of each pair were randomized to either iron or placebo
Allocation concealment?UnclearNot described
Blinding?
All outcomes
YesDouble blind
Incomplete outcome data addressed?
Mortality
UnclearNot evaluated
Incomplete outcome data addressed?
Malaria
No298 and 279 children analysed at 16 weeks and 24 weeks respectively
Incomplete outcome data addressed?
Haemoglobin or anaemia
YesIntention-to-treat

Hess 2002

Methods

Individual randomized

Trial years: 1999 to 2000

Participants

169 randomized, 166 evaluated

Age: mean 8.5 years (range 5 to 14 years)

Setting: school, rural

% anaemic at baseline:

Iron arm: 84%; placebo arm: 85% (anaemia definition: Hb < 12 g/dL in age 12 years and above and Hb < 11.5 g/dL in age 5 to 11 years), mean haemoglobin 10.9 g/dL

% malaria at baseline: not stated

Interventions

Study arms:

1. Iron: ferrous sulphate tablets, about 1 mg/kg/d elemental iron + albendazole single dose (400 mg) at baseline

2. Placebo: identical looking tablets + albendazole single dose (400 mg) at baseline

Half received a single dose of iodinized poppy seed oil containing 200 mg

Duration of treatment: 16 weeks

Duration of follow up: 20 weeks

Outcomes

Main objective/outcome:

To investigate change in response to iodine after iron supplementation

Review outcomes reported in the trial:

1. Prevalence of anaemia

2. Haemoglobin (end and change)

3. Ferritin (end)

4. Zinc (end)

5. Total iron binding capacity (TIBC)

6. Growth parameters

Notes

Trial location: Danané health district, an area of endemic goiter in the mountains of western Côte d'Ivoire

Malaria endemicity: hyperendemic

Language of publication: English

Exclusion criteria: Hb < 8 g/dL

Risk of bias
ItemAuthors' judgementDescription
Adequate sequence generation?YesComputer generated
Allocation concealment?UnclearNo description
Blinding?
All outcomes
YesDouble blind
Incomplete outcome data addressed?
Mortality
UnclearNot evaluated
Incomplete outcome data addressed?
Malaria
UnclearNot evaluated
Incomplete outcome data addressed?
Haemoglobin or anaemia
No166/169 evaluated

Hettiarachchi 2008 (C)

Methods

Cluster randomized

Trial years: not stated

Unit of randomization: class

Number of units randomized: not stated

Average cluster size: not stated

Adjustment for clustering: not mentioned

Methods of adjustment: no adjustment method was stated

Participants

Number of children: 821 randomized, 774 evaluated

Age: mean 13.5 years (range 12 to 16 years)

Setting: school, urban and rural

% anaemic at baseline: 57.1% (anaemia definition: Hb < 12 g/dL), mean haemoglobin 11.6 g/dL

% malaria at baseline: not stated

Interventions

Study arms:

1. Iron: ferrous fumarate capsules, about 1.3 mg/kg/d elemental iron

2. Zinc: zinc sulphate, 14 mg per day

3. Iron + zinc: same doses as above

4. Placebo: anhydrous lactose

All arms received mebendazole tablets 500 mg single dose 2 weeks before study

Duration of treatment: 6 months

Duration of follow up: 6 months

Outcomes

Main objective/outcome:

To assess efficacy of iron and zinc in improving anthropometry, Hb, Zinc and ferritin

Review outcomes reported in the trial:

1. Prevalence of anaemia

2. Haemoglobin (end and change)

3. Ferritin (end)

4. Growth parameters (end)

Notes

Trial location: Galle district, Sri Lanka

Malaria endemicity: mesoendemic

Language of publication: English

Exclusion criteria: haemoglobin < 8 g/dL, acute or chronic disease, inflammatory conditions, drug consumption other than paracetamol or antihistamines, currently on nutritional supplementation, donated or received blood the last 4 months

Risk of bias
ItemAuthors' judgementDescription
Adequate sequence generation?YesTable of random numbers
Allocation concealment?UnclearNo description
Blinding?
All outcomes
YesDouble blind
Incomplete outcome data addressed?
Mortality
UnclearNot evaluated
Incomplete outcome data addressed?
Malaria
UnclearNot evaluated
Incomplete outcome data addressed?
Haemoglobin or anaemia
No774/821 evaluated

Idjradinata 1993

Methods

Individual randomized

Trial years: not stated

Participants

126 randomized, 119 evaluated

Age: mean 14.4 months (range 12 to 18 months)

Setting: community, urban

% anaemic at baseline: 40% iron-deficiency anaemia (IDA) (IDA definition: Hb 10.5 g/dL or less, transferrin saturation 10% or less and ferritin 12 microgram/l or less), mean haemoglobin 11.5 g/dL

% malaria at baseline: not stated

Interventions

Study arms:

1. Iron for IDA: ferrous sulphate syrup, about 3 mg/kg/d elemental iron

2. Iron for iron-deficient without anaemia: same as above

3. Iron for iron sufficient: same as above

4. Placebo for IDA: similar in appearance and taste syrup

5. Placebo for iron-deficient without anaemia: same as above

6. Placebo for iron sufficient: same as above

Duration of treatment: 4 months

Duration of follow up: 4 months

Outcomes

Main objective/outcome:

To investigate effects of iron supplementation on mental and motor development of iron deficient infants

Review outcomes reported in the trial:

1. Haemoglobin (end)

2. Prevalence of anaemia

3. Infection episodes

4. Ferritin (end)

5. Total iron binding capacity (TIBC) (end)

6. Growth parameters (end and change)

Notes

Trial location: Bandung, Indonesia

Malaria endemicity: mesoendemic

Language of publication: English

Exclusion criteria:

Included: birthweight > 2500 g, singleton, no major congenital anomalies or perinatal complications, no jaundice treated with phototherapy, no hospital admission or supplementation with micronutrients during the 6 months before trial, no neuromotor delay, no chronic illness or folic acid deficiency, Hb 8 g/dL or more, no signs of abnormal Hb or thalassaemia, weight, height, and head circumference within 2SD

Excluded: Hb between 10.5 to 12 g/dL

Risk of bias
ItemAuthors' judgementDescription
Adequate sequence generation?YesTable of random numbers separately for each iron status class (IDA, ID, iron sufficiency)
Allocation concealment?UnclearNo description
Blinding?
All outcomes
YesDouble blind
Incomplete outcome data addressed?
Mortality
UnclearNot evaluated
Incomplete outcome data addressed?
Malaria
UnclearNot evaluated
Incomplete outcome data addressed?
Haemoglobin or anaemia
No119/126 evaluated

Kapur 2003

Methods

Individual randomized

Trial years: not stated

Participants

545 randomized, 451 consented and actually included, 232 evaluated

Age: mean 20.6 months (range 9 to 36 months)

Setting: community, urban

% anaemic at baseline: 57.3% (anaemia definition: Hb < 11 g/dL), mean haemoglobin 10.6 g/dL

% malaria at baseline: not stated

Interventions

Study arms:

1. Nutrition education only: formal meetings with mothers

2. Iron: ferium, about 0.35 mg/kg/d elemental iron

3. Nutrition education plus iron: ferium, about 0.35 mg/kg/d

4. Placebo: sugar syrup weekly

Duration of treatment: 8 weeks

Duration of follow up:16 weeks

Outcomes

Main objective/outcome:

To compare the effect of nutrition education and/or iron on iron status

Review outcomes reported in the trial:

1. Haemoglobin (end and change)

2. Ferritin (end)

3. Adverse events

Notes

Trial location: Delhi, India

Malaria endemicity: mesoendemic

Language of publication: English

Exclusion criteria: not stated

Risk of bias
ItemAuthors' judgementDescription
Adequate sequence generation?YesRandom number table
Allocation concealment?YesSealed envelopes
Blinding?
All outcomes
NoOpen
Incomplete outcome data addressed?
Mortality
UnclearNot evaluated
Incomplete outcome data addressed?
Malaria
UnclearNot evaluated
Incomplete outcome data addressed?
Haemoglobin or anaemia
No232/545 evaluated

Kashyap 1987

Methods

Individual randomized

Trial years: not stated

Participants

166 randomized, 166 evaluated

Age: range 8 to 15 years

Setting: school, urban

% anaemic at baseline: 70% (anaemia definition: Hb < 11 g/dL), mean haemoglobin 10.3 g/dL

% malaria at baseline: not stated

Interventions

Study arms:

1. Iron: ferrous sulphate tablets, about 1.7 mg/kg/d elemental iron

2. Placebo: sugar tablets

Duration of treatment: 120 days of active supplementation during 8-month period

Duration of follow up: end of treatment, 4 months after the end of treatment

Outcomes

Main objective/outcome:

To evaluate the effect of iron supplementation on cognitive function

Review outcomes reported in the trial:

1. Anaemia

2. Haemoglobin (end)

3. Iron (end)

4. Ferritin (end)

Notes

Trial location: Baroda, India

Malaria endemicity: mesoendemic

Language of publication: English

Exclusion criteria: boys, a level of family income exceeding a certain cut-off

Risk of bias
ItemAuthors' judgementDescription
Adequate sequence generation?Unclear83 matched pairs, 1 subject from each pair was randomly assigned to either iron or placebo
Allocation concealment?UnclearNo description
Blinding?
All outcomes
YesDouble blind
Incomplete outcome data addressed?
Mortality
UnclearNot evaluated
Incomplete outcome data addressed?
Malaria
UnclearNot evaluated
Incomplete outcome data addressed?
Haemoglobin or anaemia
YesIntention-to-treat

Kianfar 1999

Methods

Individual randomized

Trial years: winter 1996 to spring 1997

Participants

523 randomized, 523 evaluated

Age: mean 16.3 years

Setting: school, urban

% anaemic at baseline: 50% (anaemia definition: Hb < 12 g/dL in Rashat and < 12.7 g/dL in Zahedan (adjusted for altitude)), mean haemoglobin 12.5 g/dL

% malaria at baseline: not stated

Interventions

Study arms:

1. Daily iron for anaemic: ferrous sulphate tablets, about 1 mg/kg/d elemental iron

2. Daily iron for non anaemic: same as above

3. Twice weekly iron for anaemic: ferrous sulphate tablets, about 0.3 mg/kg/d

4. Twice weekly iron for non anaemic: same as above

5. Once weekly iron for anaemic: ferrous sulphate tablets, about 0.15 mg/kg/d

6. Once weekly iron for non anaemic: same as above

7. Control anaemic: no treatment

8. Control non anaemic: no treatment

Duration of treatment: 3 months

Duration of follow up: 3 months

Outcomes

Main objective/outcome:

To determine effects of daily and intermittent iron on haemoglobin

Review outcomes reported in the trial:

1. Prevalence of anaemia

2. Adherence

3. Haemoglobin (change)

4. Ferritin (end)

5. Adverse events

Notes

Trial location: Zahedan and Rashat (capitals of Sistan-Baluchestan and Gilan provinces), Iran)

Malaria endemicity: mesoendemic

Language of publication: English

Exclusion criteria: not stated

Risk of bias
ItemAuthors' judgementDescription
Adequate sequence generation?UnclearNo description
Allocation concealment?UnclearNo description
Blinding?
All outcomes
NoOpen
Incomplete outcome data addressed?
Mortality
UnclearNot evaluated
Incomplete outcome data addressed?
Malaria
UnclearNot evaluated
Incomplete outcome data addressed?
Haemoglobin or anaemia
YesAnalysis by intention-to-treat

Latham 1990

Methods

Individual randomized

Trial years: April 1986 to November 1986

Participants

55 randomized, 54 evaluated

Age: mean 8 years

Setting: school

% anaemic at baseline: not stated (anaemia definition: Hb < 12 g/dL), mean haemoglobin (SE): iron arm: 11.6 (0.18) g/dL; placebo arm: 11.5 (0.18) g/dL

% malaria at baseline: iron arm: 76%, placebo arm: 46%

Interventions

Study arms:

1. Iron: ferrous sulphate tablets, about 2.85 mg/kg/d elemental iron

2. Placebo: saccharin tablets

All groups received albendazole tablets 400 mg single dose once after 32 weeks

Duration of treatment: 15 weeks

Duration of follow up: 32 weeks

Outcomes

Main objective/outcome:

To determine whether iron given to school children in Kenya improves growth

Review outcomes reported in the trial:

1. Uncomplicated malaria

2. Death

3. Malaria density

4. Haemoglobin (end and change)

5. Growth parameters (end and change)

Notes

Trial location: Kwale district, Coast Province, south of Mombasa, Kenya

Malaria endemicity: holoendemic, undertaken during rainy season

Language of publication: English

Exclusion criteria: haematuria and proteinuria (indicative of Schistosoma haematobium), absence on the day of first examination, serious disease or malnutrition, Hb < 8 g/dL, heavy infections with hookworms (> 10,000 eggs per gram stool), and refusal to participate

Risk of bias
ItemAuthors' judgementDescription
Adequate sequence generation?UnclearChildren were paired by gender within the Hb rankings, from each pair one was randomly assigned to placebo and the other to iron
Allocation concealment?UnclearNo description
Blinding?
All outcomes
YesDouble blind
Incomplete outcome data addressed?
Mortality
No54/55 evaluated
Incomplete outcome data addressed?
Malaria
No54/55 evaluated
Incomplete outcome data addressed?
Haemoglobin or anaemia
No54/55 evaluated

Lawless 1994

Methods

Individual randomized

Trial years: March 1990 to July 1990

Participants

87 randomized, 86 evaluated

Age: mean 8.7 years (range 6 to 11 years)

Setting: school, rural

% anaemia at baseline: 75.5% (anaemia definition: Hb < 12 g/dL), mean haemoglobin 11.1 g/dL

% malaria at baseline: not stated

Interventions

Study arms:

1. Iron: ferrous sulphate sustained release capsules, about 1.4 mg/kg/d elemental iron

2. Placebo: identical placebo capsules

Duration of treatment: 14 weeks

Duration of follow up: 14 weeks

Outcomes

Main objective/outcome:

To determine effects of iron given to school children in Kenya on motor, weight, and height

Review outcomes reported in the trial:

1. Clinical malaria

2. Diarrhoea

3. Haemoglobin (end and change)

4. Ferritin (end)

5. Growth parameters (change)

Notes

Trial location: Coast Province, Shamu village, Kenya

Malaria endemicity: holoendemic

Language of publication: English

Exclusion criteria: Hb < 8 g/dL, heavy hookworm infection (>10,000 eggs/gram faeces), hematuria

Risk of bias
ItemAuthors' judgementDescription
Adequate sequence generation?YesRandom number table
Allocation concealment?UnclearNot described
Blinding?
All outcomes
YesDouble blind
Incomplete outcome data addressed?
Mortality
UnclearNot evaluated
Incomplete outcome data addressed?
Malaria
No86/87 evaluated
Incomplete outcome data addressed?
Haemoglobin or anaemia
No86/87 evaluated

Leenstra 2009

Methods

Individual randomized

Trial years: April 1998 to November 1998

Participants

279 randomized, 279 evaluated

Age: mean 13.8 years (range 12 to 18 years)

Setting: school, urban

% anaemia at baseline: 30.5% (anaemia definition: Hb < 12 g/dL), mean haemoglobin 12.8 g/dL

% malaria at baseline: 25.4%

Interventions

Study arms:

1. Iron plus vitamin A: ferrous sulphate tablets weekly, about 0.4 mg/kg/d elemental iron + vitamin A capsule 25,000U per week

2. Iron only: same as above

3. Vitamin A only: same dosage as above

4.Placebo

Duration of treatment: 5 months

Duration of follow up: 5 months

Outcomes

Main objective/outcome:

To determine effects of iron and vitamin A on haemoglobin, iron status, malaria, and other morbidities in schoolgirls

Review outcomes reported in the trial:

1. Clinical malaria

2. Severe malaria

3. Infections

4. Adverse events

Notes

Trial location: Kisumu city, on shores of lake Victoria, Nyanza province, western Kenya

Malaria endemicity: mesoendemic, undertaken during rainy season

Language of publication: English

Exclusion criteria: Hb < 7 g/dL, severe vitamin A deficiency (xerophthalmia), pregnancy, concomitant disease requiring hospitalization

Risk of bias
ItemAuthors' judgementDescription
Adequate sequence generation?UnclearNo description
Allocation concealment?UnclearNo description
Blinding?
All outcomes
YesDouble blind
Incomplete outcome data addressed?
Mortality
UnclearNot evaluated
Incomplete outcome data addressed?
Malaria
YesIntention-to-treat
Incomplete outcome data addressed?
Haemoglobin or anaemia
UnclearNot evaluated

Lind 2004

Methods

Individual randomized

Trial years: July 1997 to May 1999

Participants

680 randomized, 680 evaluated for mortality, 549 evaluated for anaemia

Age: mean 6.2 months

Setting: community, rural

% anaemia at baseline: 41% (anaemia definition: Hb < 11 g/dL), mean haemoglobin 11.4 g/dL

% malaria at baseline: not stated

Interventions

Study arms:

1. Placebo: syrup

2. Iron: ferrous sulphate syrup daily, about 1.7 mg/kg/d elemental iron

3. Zinc: zinc syrup 10 mg once a day

4. Iron plus zinc: same doses as above

Each dose of all supplements included 30 mg ascorbic acid

Duration of treatment: 6 months

Duration of follow up: 6 months

Outcomes

Main objective/outcome:

To determine effect of iron, zinc or both on growth, psychomotor development and incidence of infectious diseases

Review outcomes reported in the trial:

1. Anaemia prevalence

2. Death

3. Total infections

4. Diarrhoea

5. Pneumonia

6. Haemoglobin (end)

7. Ferritin (end)

8. Growth parameters (end)

9. Adverse effects

8. TIBC (end)

Notes

Trial location: Purworejo, central Java, Indonesia

Malaria endemicity: mesoendemic

Language of publication: English

Exclusion criteria: metabolic or neurologic disorders; physical handicaps affecting development, feeding, or activity;severe or protracted illness; Hb < 9 g/dL

Risk of bias
ItemAuthors' judgementDescription
Adequate sequence generation?YesPlanned and generated by an independent statistician and was performed in blocks of 20; randomization list
Allocation concealment?YesPlanned and generated by an independent statistician and was performed in blocks of 20; randomization list
Blinding?
All outcomes
YesDouble blind (researchers and participants blinded)
Incomplete outcome data addressed?
Mortality
YesAnalysis by intention-to-treat
Incomplete outcome data addressed?
Malaria
UnclearNot evaluated
Incomplete outcome data addressed?
Haemoglobin or anaemia
No549/680 evaluated

Lozoff 1996

Methods

Individual randomized

Trial years: 1986 to 1990

Participants

54 randomized

Age: mean 17.0 months

Setting: community, periurban

% anaemia at baseline: 0% (anaemia definition: Hb < 10 g/dL), mean haemoglobin 13.2 g/dL

% malaria at baseline: not stated

Interventions

Study arms:

1. Iron: iron syrup, about 6 mg/kg/d elemental iron

2. Placebo: syrup

Duration of treatment: 6 months

Duration of follow up: 6 months

Outcomes

Main objective/outcome:

To determine effect of iron on mental development

Review outcomes reported in the trial:

1. Anaemia prevalence

2. Serum iron (end)

3. Ferritin (end)

4. Zinc protoporphyrin

5. Growth parameters

Notes

Trial location: Desamparados, near San Jose, Costa Rica

Malaria endemicity: hypoendemic

Language of publication: English

Exclusion criteria: Hb < 12.5 g/dL, birth weight < 2.5 kg , birth complications, acute or chronic medical conditions

Risk of bias
ItemAuthors' judgementDescription
Adequate sequence generation?UnclearNo description
Allocation concealment?UnclearNo description
Blinding?
All outcomes
YesDouble blind (neither study families nor project personnel were aware of a child's group or treatment until the end of study participation)
Incomplete outcome data addressed?
Mortality
UnclearNot evaluated
Incomplete outcome data addressed?
Malaria
UnclearNot evaluated
Incomplete outcome data addressed?
Haemoglobin or anaemia
UnclearNot evaluated

Majumdar 2003

Methods

Individual randomized

Trial years: not stated

Participants

126 randomized, 100 evaluated

Age: range 6 to 24 months

Setting: community, urban

% anaemia at baseline: 0% (anaemia definition: Hb < 11 g/dL), mean haemoglobin 13.9 g/dL

% malaria at baseline: not stated

Interventions

Study arms:

1. Iron: iron syrup daily, about 2 mg/kg/d elemental iron

2. Placebo: identical placebo

Duration of treatment: 4 months

Duration of follow up: 4 months

Outcomes

Main objective/outcome:

Effect of iron therapy on growth and Hb status

Review outcomes reported in the trial:

1. Haemoglobin (end and change)

2. Ferritin (end)

3. Weight and height (change)

Notes

Trial location: New Delhi, India

Malaria endemicity: mesoendemic

Language of publication: English

Exclusion criteria: major congenital anomaly, prenatal complication, hospital admission or iron supplementation during the months before enrolment, chronic illness, anaemia other than iron deficiency, recent blood transfusion

Risk of bias
ItemAuthors' judgementDescription
Adequate sequence generation?UnclearNot described
Allocation concealment?YesConsecutively numbered bottles with code known only to the nurse
Blinding?
All outcomes
YesStated as double blind
Incomplete outcome data addressed?
Mortality
UnclearNot evaluated
Incomplete outcome data addressed?
Malaria
UnclearNot evaluated
Incomplete outcome data addressed?
Haemoglobin or anaemia
No100/126 evaluated

Massaga 2003

Methods

Individual randomized

Trial years: June 1999 to May 2000

Participants

291 randomized, 291 evaluated

Age: mean 14.3 weeks

Setting: community, rural

% anaemia at baseline: 0% (anaemia definition: PCV < 24%), mean haemoglobin 9.9 g/dL

% malaria at baseline: mean 31.5%

Interventions

Study arms:

1. Iron: ferric ammonium citrate suspension daily, about 7.5 mg/kg/d elemental iron

2. Placebo oral suspension

3. Iron as described above + amodiaquine oral suspension 25 mg/kg once every 2 m (overall 3 doses)

4. Amodiaquine only as described above

Duration of treatment: 6 months

Duration of follow up: 10 months

Outcomes

Main objective/outcome: Infections

Review outcomes reported in the trial:

1. Malaria

2. Anaemia

3. Death

Notes

Trial location: Muheza district, north-eastern Tanzania

Malaria endemicity: holoendemic

Language of publication: English

Exclusion criteria: infants with congenital malformation, conditions that needed hospital treatment, fever within preceding 2 weeks, PCV < 24%, participants on chemoprophylaxis

Risk of bias
ItemAuthors' judgementDescription
Adequate sequence generation?YesComputer generated
Allocation concealment?YesCentral
Blinding?
All outcomes
YesDouble blind
Incomplete outcome data addressed?
Mortality
YesIntention-to-treat
Incomplete outcome data addressed?
Malaria
YesIntention-to-treat
Incomplete outcome data addressed?
Haemoglobin or anaemia
YesIntention-to-treat

Mebrahtu 2004 (C)

Methods

Cluster-randomized

Trial years: 1996 to 1997

Unit of randomization: household

Number of units randomized: 451 housholds

Average cluster size: 1.5 children per household

Adjustment for clustering: yes

Methods of adjustment: generalized estimating equation approach was used to account for repeated measurements in children

Participants

684 children randomized, 684 evaluated for mortality, 614 evaluated for malaria, 459 evaluated for anemia

Age: mean 33.4 months (range 4 to 71 months)

Setting: community, rural

% anaemia at baseline: 94.40% (anaemia definition: Hb < 11 g/dL), mean haemoglobin 8.7 g/dL

% malaria at baseline: not stated

Interventions

Study arms:

1. Iron: ferrous sulphate syrup daily, about 1 mg/kg/d elemental iron

2. Placebo syrup

Randomization was also done by child to oral mebendazole 500 mg every 3 m vs. placebo

Duration of treatment: 12 months

Duration of follow up: 12 months

Outcomes

Main objective/outcome:

To assess the effect of low-dose, long-term iron supplementation on malaria infection

Review outcomes reported in the trial:

1. Malaria (any malaria, severe malaria)

2. Mortality

3. Haemoglobin (end)

4. Ferritin (end)

Notes

Trial location: Pemba island, Tanzania

Malaria endemicity: holoendemic

Language of publication: English

Exclusion criteria: severe anaemia (Hb < 7 g/dL)

Risk of bias
ItemAuthors' judgementDescription
Adequate sequence generation?UnclearNo description
Allocation concealment?YesPharmacy, sealed envelopes
Blinding?
All outcomes
YesDouble blind
Incomplete outcome data addressed?
Mortality
YesIntention-to-treat
Incomplete outcome data addressed?
Malaria
No614/684 evaluated
Incomplete outcome data addressed?
Haemoglobin or anaemia
No459/684 evaluated

Mejia 1988

Methods

Individual randomized

Trial years: not stated

Participants

115 randomized, 99 evaluated

Age: range 1 to 8 years

Setting: community, rural and urban

% anaemia at baseline: 100% (anaemia definition: HCT < 1.5 SD below the value for age and place), mean haemoglobin 10.4 g/dL

% malaria at baseline: not stated

Interventions

Study arms:

1. Iron: ferrous sulphate syrup, about 3 mg/kg/d elemental iron

2. Vitamin A syrup 10,000 IU/d

3. Iron + vitamin A as described above

4. Placebo syrup

Duration of treatment: 2 months

Duration of follow up: 2 months

Outcomes

Main objective/outcome:

Effect of vitamin A ± iron on haematological status

Review outcomes reported in the trial:

1. Anaemia

2. Haemoglobin (end and change)

3. Ferritin (end)

Notes

Trial location: Guatemala city and smaller cities, Guatemala

Malaria endemicity: hypoendemic

Language of publication: English

Exclusion criteria: not stated

Risk of bias
ItemAuthors' judgementDescription
Adequate sequence generation?YesThe children names were randomly drawn as in a raffle and assigned sequentially to groups I-IV
Allocation concealment?UnclearNot described
Blinding?
All outcomes
YesStated as double blind
Incomplete outcome data addressed?
Mortality
UnclearNot evaluated
Incomplete outcome data addressed?
Malaria
UnclearNot evaluated
Incomplete outcome data addressed?
Haemoglobin or anaemia
No99/115 evaluated

Menendez 1997

Methods

Individual randomized

Trial years: 1995 to 1995

Participants

832 randomized, 832 evaluated

Age: range 8 to 48 weeks

Setting: community, rural

% anaemia at baseline: not stated

% malaria at baseline: not stated

Interventions

Study arms:

1. Iron: ferrous glycine sulphate syrup daily, about 2 mg/kg/d elemental iron

2. Placebo syrup

3. Iron (same as above) + pyrimethamine + dapsone (Deltaprim) syrup 3.125 mg + 25 mg once weekly

4. Pyrimethamine + dapsone (Deltaprim) alone as described above

Duration of treatment: iron - 16 w, antimalaria - 40w

Duration of follow up: 1 year

Outcomes

Main objective/outcome:

Haemoglobin, anaemia and iron-related outcomes

Review outcomes reported in the trial:

1. Malaria

2. Mortality

3. Anaemia

4. Hospitalizations

Notes

Trial location: Ifakara, Kilombero District, Morogoro Region, south-eastern Tanzania

Malaria endemicity: hyperendemic

Language of publication: English

Exclusion criteria: packed cell volume < 25%

Risk of bias
ItemAuthors' judgementDescription
Adequate sequence generation?YesSequential numbers of a randomization code
Allocation concealment?YesRandomization code kept by an independent monitor - central
Blinding?
All outcomes
YesStated as double blind
Incomplete outcome data addressed?
Mortality
YesIntention-to-treat
Incomplete outcome data addressed?
Malaria
YesIntention-to-treat
Incomplete outcome data addressed?
Haemoglobin or anaemia
YesIntention-to-treat

Mwanri 2000

Methods

Individual randomized

Trial years: not stated

Participants

136 randomized, 135 evaluated

Age: mean 10.8 (range 9 to 12 years)

Setting: school, rural     

% anaemia: at baseline: 100% (anaemia definition: Hb < 12 g/dL), mean haemoglobin 10.5 g/dL

% malaria at baseline: not stated

Interventions

Study arms:

1. Iron: ferrous sulphate tablets thrice weekly, about 0.65mg/kg/d elemental iron

2. Vitamin A (retinyl acetate) 5000 IU thrice weekly

3. Iron + vitamin A (both as described above)

4. Placebo tablets

All subjects were dewormed for helminthiasis 2 weeks before baseline survey

Duration of treatment: 3 months

Duration of follow up: 3 months

Outcomes

Main objective/outcome:

Effects of dietary supplements on anaemia and growth

Review outcomes reported in the trial:

1. Anaemia

2. Haemoglobin (change)

3. Weight and height changes

Notes

Trial location: Bagamoyo district, coastal area of Tanzania

Malaria endemicity: hyperendemic

Language of publication: English

Exclusion criteria: chronic illnesses, physical impairments, severe anaemia (Hb < 8 g/dL)

Risk of bias
ItemAuthors' judgementDescription
Adequate sequence generation?YesThe RAND function of Excel was used to implement randomization
Allocation concealment?YesPharmacy
Blinding?
All outcomes
YesDouble blind
Incomplete outcome data addressed?
Mortality
UnclearNot evaluated
Incomplete outcome data addressed?
Malaria
UnclearNot evaluated
Incomplete outcome data addressed?
Haemoglobin or anaemia
No135/136 evaluated

Nagpal 2004

Methods

Individual randomized

Trial years: April 1999 to March 2000

Participants

100 randomized, 71 evaluated

Age: mean 5.25 months (range 4 to 6 months)

Setting: community, urban

% anaemia at baseline: not stated (anaemia definition Hb < 11 g/dL), mean haemoglobin 11.2 g/dL

% malaria at baseline: not stated

Interventions

Study arms:

1. Iron: ferric ammonium citrate drops daily, about 2 mg/kg/day elemental iron

2. Placebo: identical solution

Duration of treatment: 8 weeks

Duration of follow up: 8 weeks

Outcomes

Main objective/outcome:

Haematological utility of iron supplementation in predominantly breast fed young infants

Review outcomes reported in the trial:

1. Haemoglobin (end and change)

2. Ferritin (end)

3. Weight and height changes

Notes

Trial location: New Delhi, India

Malaria endemicity: mesoendemic

Language of publication: English

Exclusion criteria: birth weight < 2500 g, gestational age < 37 weeks, twins, congenital malformation, history of blood transfusion, blood sampling (> 10 ml) prior to recruitment, infants already receiving iron supplementation, adverse neonatal events requiring admission to the special newborn care nursery, and those with significant current morbidity.

Risk of bias
ItemAuthors' judgementDescription
Adequate sequence generation?YesComputer generated random numbers
Allocation concealment?YesRandomization sequence was sealed in an opaque envelope at a central place
Blinding?
All outcomes
YesDouble blind
Incomplete outcome data addressed?
Mortality
UnclearNot evaluated
Incomplete outcome data addressed?
Malaria
UnclearNot evaluated
Incomplete outcome data addressed?
Haemoglobin or anaemia
Unclear71/100 evaluated

Nwanyanwu 1996

Methods

Individual randomized

Trial years: March to May 1995

Participants

222 randomized, 215 evaluated for mortality and malaria, 143 evaluated for haemoglobin change

Age: mean 26 months

Setting: community, urban

% anaemia: at baseline: not stated (anaemia definition: Hb < 8 g/dL), mean haemoglobin 8.9 g/dL

% malaria at baseline: 100% with clinical malaria

Interventions

Study arms:

1. Sulphadoxine-pyrimethamine (SP) - 0.5 tablet once daily for age < 4 years, 1 tablet once daily for ages 4 to 5 years. Each tablet contains: sulphadoxine 500 mg and pyrimethamine 25 mg.

1. Daily iron: ferrous sulphate syrup daily, about 6 mg/kg/d elemental iron + sulphadoxine-pyrimethamine (SP) tablets as described above

2. Weekly iron: ferrous sulphate syrup weekly, about 0.85 mg/kg/d elemental iron + sulphadoxine-pyrimethamine (SP) tablets as described above

Duration of treatment: 4 weeks

Duration of follow up: 4 weeks

Outcomes

Main objective/outcome:

To determine whether oral iron supplementation enhances haematological recovery in young children with malaria treated with SP

Review outcomes reported in the trial:

1. Malaria

2. Death

3. Haemoglobin change

Notes

Trial location: city of Matiki, Malawi

Malaria endemicity: hyperendemic

Language of publication: English

Exclusion criteria: hospitalization, Hb < 5 g/dL, refused consent, urine positive for 4-aminoquinolines or sulphonamides

Risk of bias
ItemAuthors' judgementDescription
Adequate sequence generation?UnclearNo description
Allocation concealment?UnclearNo description
Blinding?
All outcomes
NoOpen
Incomplete outcome data addressed?
Mortality
No215/222 evaluated
Incomplete outcome data addressed?
Malaria
No215/222 evaluated
Incomplete outcome data addressed?
Haemoglobin or anaemia
No143/222 evaluated

Olsen 2006

Methods

Individual randomized

Trial years: November 1994 to January 1996

Participants

231 children randomized, 231 evaluated for mortality, 200 for haemoglobin end and change

Age: mean 8.7 years

Setting: community

% anaemia at baseline: 47.8% (anaemia definition: Hb < 11 g/dL for age < 5 years, < 11.5 g/dL for age 5 to 11 years, < 12 g/dL for 12 to 13 years and for females > 13 years, and < 13 g/dL for males > 13 years), mean haemoglobin 11.5 g/dL

% malaria at baseline: 60.6%

Interventions

Study arms:

1. Iron: ferrous dextran tablets twice weekly, about 0.7 mg/kg/d elemental iron

2. Placebo tablets twice weekly

Duration of treatment: 12 months

Duration of follow up: 12 months

Outcomes

Main objective/outcome:

Effect of 12 months of twice weekly iron supplementation on haemoglobin and ferritin

Review outcomes reported in the trial:

1. Death

2. End and change in Hb

Notes

Trial location: Kisumu district of Nyanza province, Kenya

Malaria endemicity: mesoendemic

Language of publication: English

Exclusion criteria: Hb<8g/dL, pregnancy and refusal to participate

Risk of bias
ItemAuthors' judgementDescription
Adequate sequence generation?YesComputer generated
Allocation concealment?YesSealed envelopes kept in a central location
Blinding?
All outcomes
YesDouble blind
Incomplete outcome data addressed?
Mortality
YesIntention-to-treat analysis
Incomplete outcome data addressed?
Malaria
UnclearNot evaluated
Incomplete outcome data addressed?
Haemoglobin or anaemia
Unclear200/231 evaluated

Palupi 1997

Methods

Individual randomized

Trial years: not stated

Participants

299 randomized

Age: mean 3.5 years (range 2 to 5 years)

Setting: community, rural

% anaemia at baseline: 36.7% (anaemia definition: Hb < 11 g/dL), mean haemoglobin 11.3 g/dL

% malaria at baseline: not stated

Interventions

Study arms:

1. Iron: ferrous sulphate syrup weekly, about 0.35 mg/kg/d elemental iron + a single dose of albendazole (400 mg) a week before commencing supplements

2. Iron only (as described above)

3. Placebo syrup

Duration of treatment: 9 weeks

Duration of follow up: 9 weeks

Outcomes

Main objective/outcome:

Effect of iron ± deworming on Hb

Review outcomes reported in the trial:

1. Anaemia

2. End and change in Hb

Notes

Trial location: Indonesia

Malaria endemicity: mesoendemic

Language of publication: English

Exclusion criteria: not stated

Risk of bias
ItemAuthors' judgementDescription
Adequate sequence generation?UnclearNot described
Allocation concealment?UnclearNot described
Blinding?
All outcomes
YesDouble blind
Incomplete outcome data addressed?
Mortality
UnclearNot evaluated
Incomplete outcome data addressed?
Malaria
UnclearNot evaluated
Incomplete outcome data addressed?
Haemoglobin or anaemia
No289/299 evaluated

Powers 1983

Methods

Individual randomized

Trial years: not stated

Participants

80 randomized, 40 evaluated

Age: range 4 to 12 years

Setting: community, rural

% anaemia: at baseline: not stated, mean haemoglobin 11.1 g/dL

% malaria at baseline: not stated

Interventions

Study arms:

1. Iron: ferrous sulphate syrup daily, about 2 mg/kg/d elemental iron+ chloroquine tablets 6 days before the supplementation and thereafter weekly

2. Iron (as described above) + riboflavin

3. Placebo (lactose tablets) + chloroquine tablets 6 days before the supplementation and thereafter weekly

Duration of treatment: 6 weeks

Duration of follow up: 6 weeks

Outcomes

Main objective/outcome:

Haematological status

Review outcomes reported in the trial:

1. Mortality

2. Haemoglobin end and change

3. End iron level

Notes

Trial location: Keneba village, Gambia

Malaria endemicity: hyperendemic

Language of publication: English

Exclusion criteria: not stated

Risk of bias
ItemAuthors' judgementDescription
Adequate sequence generation?UnclearNot described
Allocation concealment?YesSealed envelopes
Blinding?
All outcomes
YesDouble blind
Incomplete outcome data addressed?
Mortality
NoAuthors only stated that there were 'no deaths' in the study, denominator derived from the Hb outcome - for which 40 out of 80 children were evaluated
Incomplete outcome data addressed?
Malaria
UnclearNot evaluated
Incomplete outcome data addressed?
Haemoglobin or anaemia
No40/80 evaluated

Richard 2006

Methods

Individual randomized

Trial years: February to September 1998

Participants

855 randomized, 836 evaluated for malaria, 748 evaluated for mortality and haemoglobin

Age: range 0.5 to 15 years

Setting: school, rural

% anaemia at baseline: 46.4% (anaemia definition: Hb < 11 g/dL for age < 5 years, Hb < 11.5 g/dL for age 5 to 11 years, and Hb < 12 g/dL for age > 11 years), mean haemoglobin 11.4 g/dL

% malaria at baseline: 5%

Interventions

Study arms:

1. Iron: iron sulphate syrup daily, about 0.75 mg/kg/d elemental iron

2. Iron (as described above) + zinc 20 mg/d

3. Zinc only (20 mg/d)

4. Placebo syrup

Duration of treatment: 7 months

Duration of follow up: 7 months

Outcomes

Main objective/outcome:

Effect of daily iron and/or zinc on morbidity - malaria, diarrhoea, and respiratory infections

Review outcomes reported in the trial:

1. Mortality

2. Malaria

3. End haemoglobin

Notes

Trial location: Santa Clara village, Peru

Malaria endemicity: mesoendemic

Language of publication: English

Exclusion criteria: chronic illness (congenital diseases or major illness requiring medical care and/or medication determined by the physician at baseline evaluation) or severe malnutrition

Risk of bias
ItemAuthors' judgementDescription
Adequate sequence generation?YesComputer generated
Allocation concealment?UnclearNot described
Blinding?
All outcomes
YesTriple blinded - participants, study personnel, and data analyst all blinded
Incomplete outcome data addressed?
Mortality
No748/855 evaluated
Incomplete outcome data addressed?
Malaria
No836/855 evaluated
Incomplete outcome data addressed?
Haemoglobin or anaemia
No748/855 evaluated

Rico 2006

Methods

Individual randomized

Trial years: January 2001 to June 002

Participants

602 randomized

Age: mean 7 years (range 6 to 8 years)

Setting: school, urban

% anaemia at baseline: 21.7% iron deficiency, defined as serum ferritin < 15 mcg/l

Interventions

Study arms:

1. Iron: ferrous fumarate tablets daily, about 1.3 mg/kg/d elemental iron

2. Zinc: zinc oxide tablets, 30 mg/d

3. Iron + zinc (administration of both as described above)

4. Placebo: sugar tablets daily

Duration of treatment: 6 months

Duration of follow up: 12 months

Outcomes

Main objective/outcome:

Effect of iron and zinc supplementation on lowering lead concentrations and improving cognitive performance

Review outcomes reported in the trial: none

Notes

Trial location: City of Torreon, northern Mexico

Malaria endemicity: hypoendemic

Language of publication: English

Exclusion criteria: blood lead concentration (PbB) > 45 mcg/dL, Hb < 9 g/dL

Risk of bias
ItemAuthors' judgementDescription
Adequate sequence generation?YesList of random numbers
Allocation concealment?UnclearNot described
Blinding?
All outcomes
YesDouble blind
Incomplete outcome data addressed?
Mortality
UnclearNot evaluated
Incomplete outcome data addressed?
Malaria
UnclearNot evaluated
Incomplete outcome data addressed?
Haemoglobin or anaemia
UnclearNot evaluated

Rosado 1997

Methods

Individual randomized

Trial years: not stated

Participants

219 randomized, 194 evaluated

Age: mean 28.4 months (range 18 to 36 months)

Setting: community, rural

% anaemia at baseline: not stated (anaemia definition: Hb < 11.5 g/dL), mean haemoglobin 10.8 g/dL

% malaria at baseline: not stated

Interventions

Study arms:

1. Iron: ferrous sulphate syrup 5 days a week, about 1.8 mg/kg/d elemental iron

2. Zinc: zinc methionine, 20 mg elemental zinc daily 5 days per week

3. Iron + zinc (administration as described above)

4. Placebo

Duration of treatment: 12 months

Duration of follow up: 12 months

Outcomes

Main objective/outcome:

To assess the extent to which growth stunting could be reversed and the number of infectious episodes reduced by zinc and/or iron

Review outcomes reported in the trial:

1. Anaemia

2. End haemoglobin

3. End iron levels, end ferritin

4. Weight and height

5. Infections - diarrhoea and respiratory tract infections

Notes

Trial location: Valley of Solis, Mexico

Malaria endemicity: hypoendemic

Language of publication: English

Exclusion criteria: not stated, included only growth stunted children

Risk of bias
ItemAuthors' judgementDescription
Adequate sequence generation?UnclearNo description
Allocation concealment?UnclearNo description
Blinding?
All outcomes
YesDouble blind
Incomplete outcome data addressed?
Mortality
UnclearNot evaluated
Incomplete outcome data addressed?
Malaria
UnclearNot evaluated
Incomplete outcome data addressed?
Haemoglobin or anaemia
No194/219 evaluated

Roschnik 2003 (C)

Methods

Cluster-randomized

Trial years: February to September 2002

Unit of randomisation: schools

Number of units randomized: 40 schools

Average cluster size: 29

Adjustment for clustering: none

Methods of adjustment: not stated

Participants

Number of children: 40 schools, 1160 were tested for haemoglobin at baseline. Number randomized not stated

Age: 7 to 8 years and 10 to 12 years

Setting: school, rural

% anaemic at baseline: 54%, mean Hb 11.8 g/dL

% malaria at baseline: no or little malaria, not reported further

Interventions

Ferrous sulfate tablets 65 mg/week elemental iron (about 0.3 mg/kd/d) + folic acid 0.25 mg/ week vs. no treatment. In addition all children received praziquantel 600 mg once, 1 week before the beginning of the trial

Duration of treatment: 3.5 months

Duration of follow up: 4.5 months

Outcomes

Main objective/outcome: to evaluate the effectiveness of weekly school-based iron supplementation: its impact on mean haemoglobin concentration and anaemia prevalence, on school attendance, performance, drop-out, and repetition rates

Review outcomes reported in the trial:

1. Anaemia

2. Haemoglobin (end)

Notes

Trial location: Mangochi District in Malawi, upland and coastal areas

Malaria endemicity: hyperendemic

Language of publication: English

Exclusion criteria: not stated

Risk of bias
ItemAuthors' judgementDescription
Adequate sequence generation?YesRandom number table (inside each class 33% of children were selected for the trial - started from a random number and taking every third trial from this number on)
Allocation concealment?UnclearNot described
Blinding?
All outcomes
NoOpen
Incomplete outcome data addressed?
Mortality
YesAll participants evaluated
Incomplete outcome data addressed?
Malaria
UnclearNot reported
Incomplete outcome data addressed?
Haemoglobin or anaemia
YesAll participants evaluated

Roschnik 2004 (C)

Methods

Cluster-randomized

Trial years: July 2001

Unit of randomization: schools

Number of units randomized: 51 schools

Average cluster size: 29

Adjustment for clustering: none

Methods of adjustment: not stated

Participants

Number of children: 51 schools, 1510 individuals randomized

Age: 7 to 8 years and 10 to 12 years, mean 9.2 years

Setting: school, rural

% anaemic at baseline: 17% (anaemia defined by age 5 to11.9 Hb<11.5g/dL, age 12 to 14.9 Hb < 12 g/dL), mean Hb 12.5 g/dL

% malaria at baseline: no or little malaria, not reported further

Interventions

Ferrous sulfate tablets 108 mg/week elemental iron (about 0.57 mg/kd/d) vs. no treatment

Duration of treatment: 2.5 months

Duration of follow up: about 5.3 months

Outcomes

Main objective/outcome: Effect of weekly iron supplementation on haemoglobin levels

Review outcomes reported in the trial:

1. Anaemia

2. Death

3. Haemoglobin (end)

Notes

Trial location: Islands of Iloilo and Guimaras in the west Visayas region of the Philippines

Malaria endemicity: hypoendemic (originally hyperendemic, but author wrote that little malaria in the specific area)

Language of publication: English

Exclusion criteria: not stated

Risk of bias
ItemAuthors' judgementDescription
Adequate sequence generation?YesRandom number table (inside each class 33% of children were selected for the trial - started from a random number and taking every third trial from this number on)
Allocation concealment?UnclearNot described
Blinding?
All outcomes
NoOpen
Incomplete outcome data addressed?
Mortality
YesAll participants evaluated
Incomplete outcome data addressed?
Malaria
UnclearNot reported
Incomplete outcome data addressed?
Haemoglobin or anaemia
YesAll participants evaluated

Sarma 1977 (C)

Methods

Cluster-randomized

Trial years: not stated

Unit of randomization: preschools

Number of units randomized: 10

Average cluster size: 25

Adjustment for clustering: none

Methods of adjustment: none

Participants

10 preschools, 255 individuals randomized

Age: 2 to 6 years

Setting: school

% anaemia at baseline: 54% (anaemia defined as Hb < 11 g/dL). Mean haemoglobin 110.8 g/dL

% malaria at baseline: not stated

Interventions

Iron tablets (formulation not stated) 20 mg/day elemental iron (about 1.3 mg/kg/d) + folic acid 100 mcg/day vs. no treatment

Duration of treatment: 12 months on school days (the school year included only 265 days)

Duration of follow up: 12 months (school year)

Outcomes

Main objective/outcome: effectiveness of iron and folate given by teachers

Review outcomes reported in the trial:

1. Anaemia

2. Haemoglobin (end and change)

Notes

Trial location: India

Malaria endemicity: mesoendemic

Language of publication: English

Exclusion criteria: not stated

Risk of bias
ItemAuthors' judgementDescription
Adequate sequence generation?YesRandom number table
Allocation concealment?UnclearNo description
Blinding?
All outcomes
NoOpen study
Incomplete outcome data addressed?
Mortality
UnclearNot reported
Incomplete outcome data addressed?
Malaria
UnclearNot reported
Incomplete outcome data addressed?
Haemoglobin or anaemia
No89/ 255 individuals evaluated

Sazawal 2006 (C)a

Methods

Cluster-randomized trial

Unit of randomization: households

Number of units randomized: 22,959

Average cluster size: 1.4

Adjustment for clustering: was performed for adverse events (episodes of infection) and admissions. For mortality and cause-specific mortality adjustment for clustering is not reported.

Methods of adjustment: for analysis of adverse events and admissions, Anderson Gill time-to-event survival methods in Cox regression with robust estimation of standard error to account for multiple events per child or within household were used (SAS version 9.0, STATA version 8.2). For total mortality and cause-specific mortality, Cox regression with exact handling for ties was used.

Trial years: January 2002 to August 2003

Participants

22,959 units and 32,155 individuals; 15,956 in the 2 arms relevant for this review

Age: 1 to 35 months, mean about 18 months

Setting: community

% anaemic at baseline: not stated

% malaria at baseline: not stated

Interventions

Iron tablets (preparation not stated) dissolved in water or breast milk 12.5 mg/day + folic acid 50 mcg/day + vitamin A vs. placebo + vitamin A vs. iron + folic acid + zinc 10 mg/day + vitamin A (not used in review) vs. zinc + vitamin A. Children aged 1 to 11 months received a half dose of iron.

Duration of treatment: not fixed from < 3 months to max 18 months (until age 48 m or study discontinuation). Most received the intervention for about 12 months.

Duration of follow up: not fixed. Maximum 18 months (until age 48 months or study discontinuation)

Outcomes

Main objective/outcome: Composite of death or hospital admission (looking very specifically at malaria)

Review outcomes reported in the trial:

1. Clinical malaria, severe malaria

2. Deaths

3. Hospitalization

4. Any infection, diarrhoea

Notes

Trial location: Tanzania

Malaria endemicity: holoendemic

Language of publication: English

Exclusion criteria: none

Comparison relevant to this review (iron + folic) stopped at interim analysis based on recommendation from the data and safety monitoring board. The board received data from the main trial every month and established at the beginning of the trial that it would do further analysis of the data when the difference in mortality between any 2 groups reached a P value of 0.2 or less. Stopping rules not defined in publication. No statement on sample size and analysis adjustment for interim monthly monitoring and truncation.

Risk of bias
ItemAuthors' judgementDescription
Adequate sequence generation?YesAllocation sequence generated at the WHO controlled by computer (page 136). Permuted in blocks of 16
Allocation concealment?YesLabelled the strips of supplements with 16 letter codes - 4 for each of the groups. This letter code was hidden in the batch number on each strip of tablets
Blinding?
All outcomes
YesDouble blind. Strips of supplements coded with 16 letter codes
Incomplete outcome data addressed?
Mortality
UnclearNumber evaluated not specifically stated in article; assumed to be all participants randomized
Incomplete outcome data addressed?
Malaria
UnclearNumber evaluated not specifically stated in article; assumed to be all participants randomized
Incomplete outcome data addressed?
Haemoglobin or anaemia
NoEvaluated only children in a sub-study: 635/15,956 children in the 2 arms relevant for this review

Sazawal 2006 (C)b

Methods

Cluster-randomized trial (independent substudy of Sazawal 2006 (C)a

Unit of randomization: households

Number of units randomized: 2818 before exclusion of anaemic children

Average cluster size: 1.2

Adjustment for clustering: was performed for adverse events (episodes of infection) and admissions. For mortality and cause-specific mortality adjustment for clustering is not reported.

Methods of adjustment: for analysis of adverse events and admissions, Anderson Gill time-to-event survival methods in Cox regression with robust estimation of standard error to account for multiple events per child or within household were used (SAS version 9.0, STATA version 8.2). For total mortality and cause-specific mortality, Cox regression with exact handling for ties was used.

Trial years: March 2002 to November 2002

Participants

3171 individuals; 1619 in the 2 arms relevant for this review

Age: 1 to 35 months, mean about 22.5 months

Setting: community

% anaemic at baseline: 57% (mean Hb 9.7 g/dL)

% malaria at baseline: not stated

Interventions

Iron tablets (preparation not stated) dissolved in water or breast milk 12.5 mg/day + folic acid 50 mcg/day + vitamin A vs. placebo + vitamin A vs. iron + folic acid + zinc 10 mg/day + vitamin A (not used in review) vs. zinc + vitamin A. Children aged 1 to 11 months received a half dose of iron.

Duration of treatment: not fixed from < 3 months to max 18 months (until age 48 m or study discontinuation). Most received the intervention for about 12 months.

Duration of follow up: not fixed. Maximum 18 months (until age 48 months or study discontinuation).

Outcomes

Main objective/outcome: Composite of death or hospital admission (looking very specifically at malaria)

Review outcomes reported in the trial:

1. Clinical malaria, severe malaria

2. Deaths

3. Anaemia

Notes

Trial location: Tanzania

Malaria endemicity: holoendemic

Language of publication: English

Exclusion criteria: Hb <7 g/dL

This was a separate, independent, substudy of the bigger Sazawal 2006 (C)a trial. Separate households were randomized to the substudy, where children had baseline blood samples, anaemic children excluded (Hb < 7 g/dL), half-yearly surveillance for malaria and clinical infections performed, and treatment for malaria offered throughout the trial.

Comparison relevant to this review (iron + folic) stopped at interim analysis based on recommendation from the data and safety monitoring board. The board received data from the main trial every month and established at the beginning of the trial that it would do further analysis of the data when the difference in mortality between any 2 groups reached a P value of 0.2 or less. Stopping rules not defined in publication. No statement on sample size and analysis adjustment for interim monthly monitoring and truncation.

Risk of bias
ItemAuthors' judgementDescription
Adequate sequence generation?YesAllocation sequence generated at the WHO controlled by computer (page 136). Permuted in blocks of 16
Allocation concealment?YesLabelled the strips of supplements with 16 letter codes - 4 for each of the groups. This letter code was hidden in the batch number on each strip of tablets
Blinding?
All outcomes
YesDouble blind. Strips of supplements coded with 16 letter codes
Incomplete outcome data addressed?
Mortality
YesAll participants evaluated
Incomplete outcome data addressed?
Malaria
YesAll participants evaluated
Incomplete outcome data addressed?
Haemoglobin or anaemia
YesAll participants evaluated

Seshadri 1982a

Methods

Individual randomized

Trial years: not stated

Participants

94 randomized

Age: 5 to 8 years (stratified into three groups: 5 to 6, 6 to 7, and 7 to 8 years)

Setting: school

% anaemia at baseline: not stated (defined as Hb < 11 g/dL), mean Hb 10.5 g/dL

% malaria at baseline: not stated

Interventions

Iron tablets (preparation not stated) 20 mg/day elemental iron (about 1 mg/kg/d) + folic acid 0.1 mg/day vs. no treatment

Duration for treatment: 2 months

Duration of follow up: 2 months

Outcomes

Main objective/outcome: evaluate the effect of iron + folate on cognitive score

Review outcomes reported in the trial:

1. Haemoglobin (end), however the number of children evaluated was not given (not used in review)

Notes

Trial location: India

Malaria endemicity: mesoendemic

Language of publication: English

Exclusion criteria: malnourished children (weight for age<60% of standard)

Risk of bias
ItemAuthors' judgementDescription
Adequate sequence generation?UnclearChildren were stratified by age, every 3rd child was randomly assigned to the control group and the other 2 to the experimental group 
Allocation concealment?UnclearChildren were stratified by age, every 3rd child was randomly assigned to the control group and the other 2 to the experimental group 
Blinding?
All outcomes
NoOpen trial
Incomplete outcome data addressed?
Mortality
UnclearNot reported
Incomplete outcome data addressed?
Malaria
UnclearNot reported
Incomplete outcome data addressed?
Haemoglobin or anaemia
UnclearNumber evaluated not stated

Seshadri 1982b

Methods

Individual randomized

Trial years: not stated

Participants

28 randomized (14 pairs of boys)

Age (per study): 5 to 6 years

Setting: school

% anaemia at baseline: 100% (defined as Hb <10.5 g/dL). Mean haemoglobin about 9.7 g/dL.

% malaria at baseline: not stated

Interventions

Iron tablets (preparation not stated) 40 mg/day elemental iron (about 1 mg/kg/d) + folic acid 0.2 mg/day vs. placebo. All received a 3-day course of mebendazole at start of treatment

Duration of treatment: 2 months

Duration of follow up: 2 months

Outcomes

Main objective/outcome: to evaluate the effect of iron + folate on cognitive test performance

Review outcomes reported in the trial:

1. Haemoglobin (end and change)

Notes

Trial location: India

Malaria endemicity: mesoendemic

Language of publication: English

Exclusion criteria: malnourished children (weight for age < 60% of standard), Hb > 10.5 g/dL or < 8 g/dL, RBC morphology other than hypochromic-microcytic, IQ test ( < 70 or > 100

Risk of bias
ItemAuthors' judgementDescription
Adequate sequence generation?Yes14 pairs of boys matched for growth measures, Hb, IQ and parents income and education. One child from each pair was randomly assigned to a group by coin toss
Allocation concealment?UnclearOne child from each pair was randomly assigned to the intervention group by coin toss and the other to placebo
Blinding?
All outcomes
YesStated as double blind, placebo used
Incomplete outcome data addressed?
Mortality
UnclearNot reported
Incomplete outcome data addressed?
Malaria
UnclearNot reported
Incomplete outcome data addressed?
Haemoglobin or anaemia
YesAll participants evaluated

Seshadri 1984a

Methods

Individual randomized

Trial years: not stated

Participants

149 randomized

Age: 4 to 8 years (stratified to 2 groups: 4 to 6 years and 6 to 8 years)

Setting: school

% anaemia at baseline: 61% of iron group and 59% of control group (Hb < 11 g/dL), mean Hb 10.3 g/dL

% malaria at baseline: not stated

Interventions

Iron tablets, preparation not stated, 20 mg/day elemental iron (about 1 mg/kg/d) + folic acid 0.1 mg/day vs. no treatment

Duration of treatment: 2 months

Duration of follow up: 2 months

Outcomes

Main objective/outcome: to evaluate haemoglobin status after iron + folate supplementation

Review outcomes reported in the trial:

1. Anaemia

2. Haemoglobin (end)

Notes

Trial location: India

Malaria endemicity: mesoendemic

Language of publication: English

Exclusion criteria: income category

Risk of bias
ItemAuthors' judgementDescription
Adequate sequence generation?Unclear10 to 12 children within each of the year age groups were randomly allotted to the control group and the rest to the experimental group
Allocation concealment?UnclearNo description
Blinding?
All outcomes
NoOpen trial
Incomplete outcome data addressed?
Mortality
UnclearNot reported
Incomplete outcome data addressed?
Malaria
UnclearNot reported
Incomplete outcome data addressed?
Haemoglobin or anaemia
YesAll participants evaluated

Seshadri 1984b (C)

Methods

Cluster-randomized

Unit of randomization: preschools

Number of units randomized: 22,959

Average cluster size: 22

Adjustment for clustering: none

Methods of adjustment: none

Trial years: not stated

Participants

4 preschools, 89 individuals randomized

Age: not stated

Setting: school

% anaemic at baseline: 73%, mean Hb 10.3 g/dL

% malaria at baseline: not stated

Interventions

Iron tablets (preparation not stated) 20 mg/day elemental (about 1 mg/kg/d) + folic acid 0.1 mg/day vs. placebo

Duration of treatment: 4 months (2 different periods of 60 days in 1 school year)

Duration of follow up: 12 months follow up after net 120 days of treatment (3 months after end of treatment only in intervention group)

Outcomes

Main objective/outcome: Anaemia

Review outcomes reported in the trial:

1. Anaemia

2. Haemoglobin (end and change)

Notes

Trial location: India

Malaria endemicity: mesoendemic

Language of publication: English

Risk of bias
ItemAuthors' judgementDescription
Adequate sequence generation?UnclearNo description
Allocation concealment?UnclearNo description
Blinding?
All outcomes
NoOpen, but placebo used
Incomplete outcome data addressed?
Mortality
UnclearNot reported
Incomplete outcome data addressed?
Malaria
UnclearNot reported
Incomplete outcome data addressed?
Haemoglobin or anaemia
YesAll participants evaluated

Shah 2002

Methods

Individual randomized

Trial years: March 1998 to March 1999

Participants

209 randomized

Age: 113-day18 years, mean of about 15 years

Setting: school

% anaemia at baseline: 68.8% (haematocrit < 36%), mean baseline haematocrit about 33%

% malaria at baseline: not stated

Interventions

Ferrous sulphate tablets 350 mg/day (about 1 mg/kg/d elemental iron) + folic acid 1.5 mg/day vs. iron 350 mg/week + folic acid vs. no treatment

Treatment duration: about 3.5 months

Duration of follow up: about 4 months

Outcomes

Main objective/outcome: to compare daily vs. weekly iron + folate for control of anaemia and Hb status in girls

Review outcomes reported in the trial:

1. Deaths

2. Anaemia

3. Haemoglobin (end and change) 

4. Adverse events

Notes

Trial location: Nepal

Malaria endemicity: hypoendemic

Language of publication: English

Exclusion criteria: male sex, chronic illness, long-term allopathic or indigenous drug treatments, hospitalization for severe illness in the past 2 weeks

Risk of bias
ItemAuthors' judgementDescription
Adequate sequence generation?YesComputer generated numbers
Allocation concealment?YesSealed envelopes
Blinding?
All outcomes
NoOpen
Incomplete outcome data addressed?
Mortality
UnclearNot reported
Incomplete outcome data addressed?
Malaria
UnclearNot reported
Incomplete outcome data addressed?
Haemoglobin or anaemia
YesAll participants evaluated

Smith 1989 (C)

Methods

Cluster-randomized

Trial years: July to August 1983

Unit of randomisation: household

Number of units randomized: not stated

Average cluster size: not stated

Adjustment for clustering: none

Methods of adjustment: not stated

Participants

Number of children: 213 children

Age: 6 months to 5 years, mean about 2.7 years

Setting: community

% anaemic at baseline: 100% (defined as Hb < 11.1 g/dL and mean corpuscular volume < 70 (less than the third percentile by age)), mean Hb 9.3 g/dL

% malaria at baseline: not stated

Interventions

Ferrous sulphate elixir of crushed tablets in orange juice 3 to 6 mg/kg/d elemental iron vs. orange juice (placebo)

Duration of treatment: 12 weeks

Duration of follow up: 13 weeks

Outcomes

Main objective/outcome: Hb/iron + malaria status

Review outcomes reported in the trial:

1. Clinical malaria, parasitaemia, parasitaemia > 5000

2. Deaths

3. Febrile disease

Notes

Trial location: Gambia

Malaria endemicity: hyperendemic

Language of publication: English

Exclusion criteria: Hb < 5g/dL

Risk of bias
ItemAuthors' judgementDescription
Adequate sequence generation?NoThe first compound on the compound list for each village was randomly assigned and compounds were assigned alternately thereafter 
Allocation concealment?NoAlternation
Blinding?
All outcomes
YesParents, field workers, and study investigator blinded
Incomplete outcome data addressed?
Mortality
YesAll participants evaluated
Incomplete outcome data addressed?
Malaria
No186/213 participants evaluated
Incomplete outcome data addressed?
Haemoglobin or anaemia
UnclearNot reported

Smuts 2005

Methods

Individual randomized

Trial years: not stated

Participants

1134 randomized

Age: 6 to 11 months, mean 8.7 months

Setting: community

% anaemia at baseline: 64.9% and 58.6% (defined Hb < 11 g/dL) in iron and placebo groups, respectively. Mean Hb 10.7 g/dL

% malaria at baseline: not stated

Interventions

Iron as chewable tablets/foodlets 10 mg/day elemental iron (about 1 mg/kg/d)

Duration of treatment duration: 6 months

Duration of follow up: 6 months

Outcomes

Main objective/outcome: to test the hypothesis that improving micronutrient status would improve growth of infants at high risk for anaemia

Review outcomes reported in the trial:

1. Anaemia

2. Haemoglobin (end and change)

3. Weight and height

Notes

Trial location: Vietnam, South-Africa, Peru, Indonesia

Malaria endemicity: meso- and hyperendemic

Language of publication: English

Exclusion criteria: birth < 37 w or < 2500 g, severe wasting (>-3 Z-score), Hb < 8, fever > 39

Risk of bias
ItemAuthors' judgementDescription
Adequate sequence generation?UnclearNot described
Allocation concealment?YesCentrally
Blinding?
All outcomes
YesDouble-blind
Incomplete outcome data addressed?
Mortality
UnclearNot reported
Incomplete outcome data addressed?
Malaria
UnclearNot reported
Incomplete outcome data addressed?
Haemoglobin or anaemia
Yes481/571 participants evaluated

Soekarjo 2004 (C)

Methods

Cluster-randomized

Unit of randomization: classes

Average cluster size: 48

Adjustment for clustering: none

Methods of adjustment: not stated

Trial years: October 1996 to May 1997

Participants

45 classes, 2163 children

Age (per study): average 12 to 15 years

Setting: school

% anaemia at baseline: not stated

% malaria at baseline: not stated

Interventions

Ferrous sulphate tablets 60 mg elemental iron weekly (about 0.2 mg/kg/d) + folic acid 250 mcg x 1/week + vitamin A 10,000 U x 1/week vs. vitamin A

Duration of treatment: 14 weeks

Duration of follow up: 14 week

Outcomes

Main objective/outcome: anaemia

Review outcomes reported in the trial:

Haemoglobin (change)

Notes

Trial location: Indonesia

Malaria endemicity: mesoendemic

Language of publication: English

Exclusion criteria: not stated

Risk of bias
ItemAuthors' judgementDescription
Adequate sequence generation?UnclearNo description
Allocation concealment?UnclearNo description
Blinding?
All outcomes
NoOpen
Incomplete outcome data addressed?
Mortality
UnclearNot reported
Incomplete outcome data addressed?
Malaria
UnclearNot reported
Incomplete outcome data addressed?
Haemoglobin or anaemia
No2012/2163 participants evaluated

Soemantri 1989

Methods

Individual randomized

Trial years: not stated

Participants

130 randomized

Age: mean 10.4 ± 1.6 (anaemic children) and mean 10.5 ± 1.5 (non-anaemic children)

Setting: school

% anaemia: at baseline: study stratified by anaemia, mean haemoglobin about 9.7 in anaemic group and 13.3 in non-anaemic group (Hb < 11 g/dL)

% malaria at baseline: not stated

Interventions

Ferrous sulphate tablets 2 mg/kg/d elemental iron vs. saccharin + tapioca (control). Pyrantel pamoate was given to all parasite-positive children before randomization.

Duration of treatment: 3 months

Duration of follow up: 6 months

Outcomes

Main objective/outcome: Effect of iron on learning achievement of iron deficient anaemic children

Review outcomes reported in the trial:

1. Haemoglobin (end and change)

2. TIBC

Notes

Trial location: Indonesia

Malaria endemicity: mesoendemic

Language of publication: English

Exclusion criteria: < 80th percentile of weight and height or mid-arm circumference < 85th percentile of Indonesian growth standards, positive parasite egg count by stool examination after deworming treatment, acute or chronic illness, clinical signs of malnutrition, physical handicaps, mental retardation, neurological dysfunction, or haematological disorders, IQ < 75, acute or severe morbidity, Hb 11.1 to 11.9 g/dL and saturation 13% to 19%

Risk of bias
ItemAuthors' judgementDescription
Adequate sequence generation?UnclearNot described
Allocation concealment?UnclearNot described
Blinding?
All outcomes
YesDouble blind (placebo had same size and colour as study drug)
Incomplete outcome data addressed?
Mortality
UnclearNot reported
Incomplete outcome data addressed?
Malaria
UnclearNot reported
Incomplete outcome data addressed?
Haemoglobin or anaemia
YesAll participants evaluated

Soewondo 1989

Methods

Individual randomized

Trial years: 1983 to 1984

Participants

176 randomized

Age: preschool children

Setting: school

% anaemia: at baseline: stratified by haemoglobin and iron status. Mean haemoglobin 10.6 g/dL for iron deficient and anaemic children, 11.7 for iron deficient and 12.3 for iron replete children

% malaria at baseline: not stated

Interventions

Ferrous sulphate syrup 50 mg/day elemental iron (about 3 mg/kg/d) vs. placebo

Duration of treatment: 2 months

Duration of follow up: 2 months

Outcomes

Main objective/outcome: effects of iron supplementation on performance in learning tasks

Review outcomes reported in the trial:

1. Haemoglobin (end)

2. Ferritin, TIBC, protoporphyrin

Notes

Trial location: Indonesia

Malaria endemicity: mesoendemic

Language of publication: English

Exclusion criteria: not stated

Risk of bias
ItemAuthors' judgementDescription
Adequate sequence generation?UnclearNot described
Allocation concealment?UnclearNot described
Blinding?
All outcomes
UnclearDouble blind, placebo used
Incomplete outcome data addressed?
Mortality
UnclearNot reported
Incomplete outcome data addressed?
Malaria
UnclearNot reported
Incomplete outcome data addressed?
Haemoglobin or anaemia
YesAll participants evaluated

Taylor 2001

Methods

Individual randomized

Trial years: 1996 to 1997

Participants

428 randomized

Age: 6-15 years

Setting: school

% anaemia at baseline: 33.5% (defined as Hb < 12 g/dL), mean haemoglobin 12.3 g/dL

% malaria at baseline: 5%

Interventions

Ferrous fumarate 200 mg tabs containing 65 mg elemental iron (about 0.3 mg/kg/d) + 100 mcg folate once weekly + albendazole 400 mg/day + praziquantel 40 mg/kg/day for 3 days vs. placebo + albendazole + praziquantel for 3 days vs. placebo + albendazole + praziquantel single dose vs. ferrous fumarate + antihelminthic placebo vs. placebo all

Duration of treatment: iron for 10 weeks, anthelminthics at 6 and 12 months

Duration of follow up: 12 months

Outcomes

Main objective/outcome: to determine whether different combinations of antihelminthics and iron would improve nutritional and health status

Review outcomes reported in the trial:

1. Haemoglobin (end and change)

Notes

Trial location: South Africa

Malaria endemicity: mesoendemic

Language of publication: English

Exclusion criteria: females post-puberty

Risk of bias
ItemAuthors' judgementDescription
Adequate sequence generation?UnclearNot described
Allocation concealment?UnclearNot described
Blinding?
All outcomes
YesDouble blind, placebo used
Incomplete outcome data addressed?
Mortality
UnclearNot reported
Incomplete outcome data addressed?
Malaria
UnclearNot reported
Incomplete outcome data addressed?
Haemoglobin or anaemia
No275/428 participants evaluated

van den Hombergh 1996

Methods

Individual randomized

Trial years: April to June 1993

Participants

100 randomized

Age: < 30 months

Setting: hospital/community

% anaemia at baseline: 100% (defined as Hb < 5 g/dL), randomization stratified by receipt of blood products, mean Hb 4.1 g/dL

% malaria at baseline: 100% with clinical malaria

Interventions

Ferrous fumarate tablets 200 mg/day (65 mg elemental iron, about 4 mg/kg/d + folic acid 100 mcg) vs. folic acid. In addition all children received quinine sulphate treatment for malaria

Duration of treatment: 3 months

Duration of follow up: 3 months

Outcomes

Main objective/outcome: adverse effect of iron treatment on malaria infection

Review outcomes reported in the trial:

1. Malaria parasitaemia, parasite density

2. Deaths

3. Anaemia

4. Haemoglobin (end)

5. Clinic visits

6. Pneumonia

7. Weight

Notes

Trial location: Tanzania

Malaria endemicity: hyperendemic

Language of publication: English

Exclusion criteria: cerebral malaria, non-falciparum malaria, sickle cell anaemia, and children fulfilling the criteria in whom malarial anaemia was not the main medical problem (e.g. meningitis, measles)

Risk of bias
ItemAuthors' judgementDescription
Adequate sequence generation?UnclearNo description
Allocation concealment?UnclearNo description
Blinding?
All outcomes
NoOpen
Incomplete outcome data addressed?
Mortality
No96/100 participants evaluated
Incomplete outcome data addressed?
Malaria
No94/100 participants evaluated
Incomplete outcome data addressed?
Haemoglobin or anaemia
YesAll participants evaluated

van Hensbroek 1995

Methods

Individual randomized

Trial years: July to December 1992

Participants

600 randomized

Age: 6 months to 9 years

Setting: hospital/community

% anaemia at baseline: 74% (defined as Hb < 11 g/dL), mean Hb 9.6 g/dL

% malaria at baseline: 100% with clinical malaria

Interventions

Sodium iron edentate syrup, 27.5 mg x 3/d elemental iron for children < 20 kg, 41.25 mg x 3/d elemental iron for children > 20 kg (about 6 mg/kg/d elemental iron) + pyrimethamine+sulfadoxine (SP) single dose vs. placebo + SP vs. placebo + chloroquine vs. folic acid + chloroquine (not used for this review) vs. folic acid + SP (not used for this review)

Duration of treatment: 1 month

Duration of follow up: 4 months after end of rainy season

Outcomes

Main objective/outcome: The effect of iron or folic acid plus antimalarial on malarial anaemia

Review outcomes reported in the trial:

1. Malaria parasitaemia

2. Deaths

3. Haemoglobin (change)

Notes

Trial location: Gambia

Malaria endemicity: hyperendemic

Language of publication: English

Exclusion criteria: severe underlying disease or complicated malaria that required hospital admission, non-falciparum malaria or less than 5 parasites per high power field

Risk of bias
ItemAuthors' judgementDescription
Adequate sequence generation?UnclearChildren were allocated at random to receive either chloroquine or Fansidar as antimalarial treatment and iron, folic acid or placebo as supplementation
Allocation concealment?UnclearNo description
Blinding?
All outcomes
NoOpen (iron syrup vs. sugar tablets)
Incomplete outcome data addressed?
Mortality
UnclearNumber randomized not reported per group
Incomplete outcome data addressed?
Malaria
UnclearNumber randomized not reported per group
Incomplete outcome data addressed?
Haemoglobin or anaemia
UnclearNumber randomized not reported per group

Vaughan 1977

Methods

Individual randomized

Trial years: not stated

Participants

189 randomized

Age: 7 to 14 years

Setting: school

% anaemia at baseline: 100% (defined as Hb < 11 g/dL)

% malaria at baseline: not stated

Interventions

Ferrous sulphate tablets 200 mg x 2/day (about 2.3 mg/kg/d elemental iron) + tetrachloroethylene (TCE) 2 to 3 cc single dose vs. ferrous sulphate + placebo vs. TCE vs. placebo

Duration of treatment: 1 month

Duration of follow up: 1 month

Outcomes

Main objective/outcome: to evaluate effectiveness of using local dispensaries and primary schools in the ambulant treatment of 'healthy' anaemic villagers

Review outcomes reported in the trial: none (haemoglobin results reported only qualitatively)

Notes

Trial location: Tanzania

Malaria endemicity: hyperendemic

Language of publication: English

Exclusion criteria: not stated

Risk of bias
ItemAuthors' judgementDescription
Adequate sequence generation?UnclearChildren were randomly allocated to one of the treatment groups
Allocation concealment?UnclearNo description
Blinding?
All outcomes
UnclearBlinded, probably not optimally
Incomplete outcome data addressed?
Mortality
UnclearNot reported
Incomplete outcome data addressed?
Malaria
UnclearNot reported
Incomplete outcome data addressed?
Haemoglobin or anaemia
UnclearNot reported

Verhoef 2002

Methods

Individual randomized

Trial years: 1998 to 2000

Participants

328 randomized

Age: 2 to 36 months, mean about 18 months  

Setting: community

% anaemia at baseline: 72% in this age group from an earlier survey (defined as Hb < 11 g/dL), mean Hb 9.6 g/dL

% malaria at baseline: as indicated by a dipstick test result, 31% in this age group from an earlier survey

Interventions

Ferrous fumarate suspension 6 mg/kg/week elemental iron (about 0.86 mg/kg/d) given in 2 doses (twice a week) + sulfadoxine/pyrimethamine (SP) 25/1.25 mg/kg once every 4 weeks vs. ferrous fumarate + placebo vs. SP + placebo vs. placebo + placebo

Duration of treatment: 3 months

Duration of follow up: 3 months

Outcomes

Main objective/outcome: effect of intermittent iron and SP on Hb in symptom-free children

Review outcomes reported in the trial:

1. Clinical malaria

2. Anaemia

3. Haemoglobin (end)

Notes

Trial location: Kenya

Malaria endemicity: mesoendemic

Language of publication: English

Exclusion criteria: Hb < 6 or >11 g/dL, axillary temp > 37.5 °C, symptoms suggestive of malaria or anaemia, or any systemic illness occurring in combination with a blood dipstick test result indicating current or recent malaria infection

Risk of bias
ItemAuthors' judgementDescription
Adequate sequence generation?YesTables with randomized permutations
Allocation concealment?YesThe order of children listed was concealed from the person generating the allocation schedule
Blinding?
All outcomes
YesDouble blind: field investigators, participants
Incomplete outcome data addressed?
Mortality
YesAll participants evaluated
Incomplete outcome data addressed?
Malaria
YesAll participants evaluated
Incomplete outcome data addressed?
Haemoglobin or anaemia
No307/328 participants evaluated

Wasantwisut 2006

Methods

Individual randomized 

Trial years: not stated

Participants

674 randomized, 256 evaluated for review outcomes

Age: 4 to 6 months, mean 4.5 months

Setting: community

% anaemia at baseline: 30% (defined as Hb < 11 g/dL, mean haemoglobin 11.5 g/dL

% malaria at baseline: not reported

Interventions

Ferrous sulphate syrup 10 mg/day (about 1.5 mg/kg/d) vs. placebo vs. ferrous sulphate + zinc sulphate 10 mg/day vs. zinc. In addition all supplements contained vitamin C and each subject received vitamin A at the beginning of trial.

Duration of treatment: 6 months

Duration of follow up: 6 months

Outcomes

Main objective/outcome: to test the hypothesis that supplementation of iron or zinc alone, or iron and zinc combined, can improve iron and zinc status and growth of infants

Review outcomes reported in the trial:

1. Anaemia

2. Deaths

3. Haemoglobin (end)

4. Ferritin, zinc

5. Weight and height

6. Adverse events

Notes

Trial location: Thailand

Malaria endemicity:  mesoendemic

Language of publication: English

Exclusion criteria: congenital abnormalities, Hb < 8.0 g/dL, chronic illnesses, or children who were bottle fed

Risk of bias
ItemAuthors' judgementDescription
Adequate sequence generation?YesRandom numbers
Allocation concealment?YesThe randomization was done by a statistician who was not involved in the study
Blinding?
All outcomes
YesDouble blind, placebo used
Incomplete outcome data addressed?
Mortality
No256/674 participants evaluated
Incomplete outcome data addressed?
Malaria
UnclearNot reported
Incomplete outcome data addressed?
Haemoglobin or anaemia
No256/674 participants evaluated

Zavaleta 2000

Methods

Individual randomized

Trial years: August to December 1996

Participants

312 randomized

Age: 12 to 18 years

Setting: school

% anaemia at baseline: 15.4%, 18.5% and 19.8% across groups (defined as Hb < 12 g/dL), mean haemoglobin 12.7 g/dL

% malaria at baseline: not stated

Interventions

Ferrous sulphate tablets 60 mg/day elemental iron (about 0.63 mg/kg/d) on school days vs. ferrous sulphate tablets 60 mg twice weekly + placebo other school days vs. placebo

Duration of treatment: 17 weeks

Duration of follow up: 17 weeks

Outcomes

Main objective/outcome: to assess the feasibility, efficacy, and acceptability of reducing anaemia in adolescent girls attending public school using daily or intermittent iron supplementation

Review outcomes reported in the trial:

1. Anaemia

2. Haemoglobin (end)

Notes

Trial location: Peru

Malaria endemicity: mesoendemic

Language of publication: English

Exclusion criteria: irregular menstruation in the preceding 3 months, any multivitamin-mineral supplement in the last 6 months, Hb < 8 g/dL

Risk of bias
ItemAuthors' judgementDescription
Adequate sequence generation?UnclearAssigned at random
Allocation concealment?UnclearDistributed in coded blister packages
Blinding?
All outcomes
YesDouble blind, placebo used
Incomplete outcome data addressed?
Mortality
UnclearNot reported
Incomplete outcome data addressed?
Malaria
UnclearNot reported
Incomplete outcome data addressed?
Haemoglobin or anaemia
Unclear296/312 participants evaluated

Zlotkin 2003

  1. a

    d = days
    Fe = iron
    Hb = haemoglobin
    HCT - haematocrit
    ID = iron deficiency
    IDA = iron-deficiency anaemia
    IM = intramuscular
    IPT = iron + sulfadoxine-pyrimethamine
    m = months
    PCV = packed cell volume
    RBC = red blood cell
    SP = sulfadoxine-pyrimethamine
    TCE = tetrachloroethylene
    TIBC = total iron binding capacity
    Vit A = vitamin A
    w = weeks
    y = years
    Zn = zinc

Methods

Individual randomized

Trial years: Oct 1999 to March 2000

Participants

437 randomized, 165 evaluated

Age: mean 16.5+/-3.9 months and 15.2 +/- 4.1 months for iron vs. placebo

Setting: community

% anaemia at baseline: 0% (defined as Hb < 10 g/dL), mean haemoglobin 12.7 g/dL

% malaria at baseline: 62.3% (202/324 children who completed intervention)

Interventions

Ferrous sulphate drops 12.5 mg/day elemental iron (about 1.25 mg/kd/d) vs. placebo sachets sprinkled on food vs. iron fumarate sprinkles (not used in review) vs. iron fumarate sprinkles + vitamin A (not used in review)

Duration of treatment: 6 months

Duration of follow up: 18 months (only children who were not anaemic at the end of supplementation were followed for the additional period of time)

Outcomes

Main objective/outcome: to compare the efficacy of microencapsulated iron fumarate sprinkles ± Vit A with iron sulphate drops with placebo in preventing recurrent anaemia and to determine the long-term haematological outcome

Review outcomes reported in the trial:

1. Anaemia

2. Deaths

3. Haemoglobin (end and change)

4. Ferritin

Notes

Trial location: Ghana

Malaria endemicity: hyperendemic

Language of publication: English

Exclusion criteria: Hb < 10 g/dL, age 8 to 20 m, only breast feeding children

Risk of bias
ItemAuthors' judgementDescription
Adequate sequence generation?YesComputer generated
Allocation concealment?YesSealed opaque envelopes
Blinding?
All outcomes
NoOpen trial, intervention and control arms different
Incomplete outcome data addressed?
Mortality
YesAll participants evaluated
Incomplete outcome data addressed?
Malaria
UnclearNot reported
Incomplete outcome data addressed?
Haemoglobin or anaemia
No165/220 participants evaluated

Characteristics of excluded studies [ordered by study ID]

StudyReason for exclusion
  1. a

    Fe = iron
    RCT = randomized controlled trial
    Vit C = vitamin C

Abdelrazik 2007Non-RCT
Adu-Afarwuah 2008Fortification of food or drink
Agarwal 2003All groups given iron (dose, schedule, or other comparisons)
Ahmed 2001Study not in children
Anand 2007Fortification of food or drink
Angeles-Agdeppa 1997Incompatible intervention (iron + other micronutrients)
Anonymous 2006Editorial (non-RCT)
Arcanjo 2008Fortification of food or drink
Asibey-Berko 2007Fortification of food or drink
Assunçăo 2007Fortification of food or drink
Aukett 1986Non-endemic areas: England
Baird 1997Non-RCT
Barclay 1991Non-endemic areas
Bates 1987Incompatible intervention (iron + other micronutrients) iron + vit C + riboflavin vs. placebo
Beasley 2000Incompatible intervention (iron + other micronutrients: iron vs. B12)
Bender-Gotze 1980RCT conducted in non-endemic area: Germany
Berger 1992Non-RCT
Bojang 1997RCT, blood transfusion vs. iron (parenteral administration of iron)
Bradfield 1968Non-RCT
Bruner 1996Non-endemic areas
Brunser 1993Non-endemic area (Chile), iron administered as fortification of milk
Carter 2005RCT, all groups received iron
Chandramohan 2005RCT, all groups received iron
Cusick 2005RCT, all groups received iron
Deinard 1986 Non-endemic area: Minnesota, USA
Desai 2004Dose comparison, all groups given iron
Dewey 2002Non-endermic area; Sweden, Honduras
Dijkhuizen 2001Stated specifically in study that the area is malaria-free
Diouf 2002Non-RCT (correspondence)
Domelloff 2001Neither Honduras nor Sweden mentioned in our malaria-endemic areas table
Ekvall 2000Incompatible intervention (iron + other micronutrients: multivitamins vs. promethazine hydrochloride)
Engstrom 2008Cluster-randomized trial. Inclusion criteria of children in control clusters (no iron for 6 months) differed systematically from those of iron-supplementation clusters (no iron for 1 month)
Fuerth 1972Non-endemic area: California
Giovannini 2006Fortification of food or drink
Gomber 1998All children were given iron supplementation
Gomes 2000Randomization to antimalaria treatment. All children received iron
Greisen 1986Non-RCT
Hathirat 1992Stated specifically in study that the area is malaria-free
Heywood 1989RCT, parenteral iron
Hirve 2007Fortification of food or drink
Honig 1978RCT with intramuscular iron
Hussen 1985Non-endemic area: Egypt. Not known whether RCT
Hyder 2007Fortification of food or drink
Ip 2009Fortification of food or drink
Isager 1974Non RCT (review article)
Jacobi 1972Non-RCT
Kanani 2000Cluster-RCT with less than 2 units per arm
Kleinschmidt 1965Non-RCT
Kurz 1985Non-RCT
le Cessie 2002Longitudinal study
Le Huong 2007Fortification of food or drink
Lima 2006Non-RCT
Liu 1995Comparison of different iron administration schedules. No placebo group
Liu 1996Dose comparison, all groups given iron
Lozoff 1982Incompatible intervention (iron + other micronutrients)
Lozoff 2003Non-endemic areas
Lutter 2008Fortification of food or drink
Maldonado 2007Fortification of food or drink
Mamiro 2001Non RCT (cross-sectional survey)
Migasena 1972Stated specifically in study that the area is malaria-free
Mitra 1997Stated specifically in study that the area is malaria-free
Morales 2008Fortification of food or drink
Morley 1999Non-endemic areas: England
Murray 1978RCT, adults
Muñoz 2000The supplements used were a part of a beverage
Naghii 2007Fortification of food or drink
Nguyen 2002Incompatible interventions: group 1 placebo, group 2 iron, group 3 daily iron, group 4 weekly iron. Only groups 3 and 4 were assigned randomly
Oppenheimer 1986RCT, parenteral iron
Oski 1978RCT, parenteral iron
Oski 1983Non-randomized trial
Parks 1989Non endemic area: Birmingham
Pereira 1978Non-RCT
Perrone 1999Non endemic area: Italy
Rahimy 2007Non-RCT
Rahman 1999Stated specifically in study that the area is malaria-free
Sarma 2006Fortification of food or drink
Schellenberg 2001RCT, all groups received iron
Schellenberg 2004Dose comparison, all groups given iron
Schultink 1995All groups given iron (dose, schedule, or other comparisons)
Schumann 2009Fortification of food or drink
Sharma 2000All groups given iron (dose, schedule, or other comparisons)
Singla 1982Incompatible intervention (iron + other micronutrients: iron + FA + B12 vs. placebo)
Sungthong 2002Stated specifically in study that the area is malaria-free
Tee 1999Stated specifically in study that the area is malaria-free
Thibault 1993Non-endemic area: France
Thu 1999All groups given iron (dose, schedule, or other comparisons)
Tielsch 2006Non-endemic area, according to correspondence with the author
Tomashek 2001RCT, all groups received iron
van Stuijvenberg 2008Fortification of food or drink
Walter 1986Non-RCT
Wegmüller 2006Fortification of food: iodized salt fortified with 3 mg Fe/g vs. iodized salt
Williams 1999Non-endemic areas: England
Yalcin 2000Non-endemic area
Yang 2004Non-endemic area
Zimmermann 2007According to correspondence with authors, the trial used iron fortified biscuits
Zlotkin 2005Fortification of food or drink

Characteristics of studies awaiting assessment [ordered by study ID]

Browne 2005

Methods

Individual randomized controlled trial, double blind, placebo controlled

Trial years: May 2001 to August 2004

ParticipantsInfants in forest belt of Ghana (1791 children randomized)
Interventions

(Data unavailable on iron preparation), oral route, daily for 12 months. Antimalarial given at 10, 14 weeks and at 9 month of age

Treatment duration: 9 months; overall follow up duration: 9 months

Outcomes

Anaemia

Clinical malaria

Side effects

NotesConference abstract: letter sent to first and last authors 6 July 2008. May be the same trial as Gomes 2001 (conducted in same region, similar number of children randomized).

Denno 2006

Methods

Randomized controlled trial, 2 x 2 factorial design, double blind, placebo controlled

Trial years: June 2005 to April 2006

ParticipantsStudy completion date: 2006 
Interventions
  1. Weekly iron supplementation iron (II) sulphate tablet at dose of 2 mg/kg body weight and monthly placebo resembling sulphadoxine-pyrimethamine tablet.

  2. Weekly placebo resembling iron (II) sulphate tablet and monthly treatment with sulphadoxine-pyrimethamine according to body weight (¼ tablet for those < 5 kg, ½ tablet for those 5 to 10 kg and 1 tablet for those 10 to 15 kg.

  3. Weekly iron supplementation iron (II) sulphate tablet at dose of 2 mg/kg body weight and monthly sulphadoxine-pyrimethamine tablet

  4. Weekly placebo resembling iron (II) sulphate tablet and monthly placebo resembling sulphadoxine-pyrimethamine tablet

Outcomes

Haemoglobin change

Weight to age and length to age ratio b

Plasmodium falciparum in thick peripheral blood film

NotesNCT00301054 completed. Author contacted and stated that the primary investigator of the study, Dr. Anthony Ofosu, has conducted a preliminary analysis. However, the pharmaceutical company which supplied the drugs and placebos for the study and helped with the drug blinding codes has been unable to provide the authors with the codes.

Gomes 2001

Methods

Randomized controlled trial, double blind, placebo controlled

Trial years: June 1999 to June 2001

ParticipantsAll infants aged at least 10 weeks attending a Mother and Child Health (MCH) Clinic or Child Welfare Clinic (CWC) in the Afigya-Sekyere District, Ashanti Region, in the Forest belt of rural Ghana (1800 infants in total randomized)
Interventions1. Iron supplementation and placebo
2. Placebo and sulphadoxine-pyrimethamine
3. Iron supplementation and sulphadoxine-pyrimethamine
Outcomes

Control of malaria

Severe anaemia

NotesISRCTN85737357 completed; letter sent to author 7 July 2008. No response yet.

Sazawal 2006 (C)c

MethodsCluster-randomized controlled trial, double blind, placebo controlled
ParticipantsChildren aged 1 to 35 months living in Pemba, Zanzibar
Interventions

1. Iron + folic acid + vitamin A

2. Iron + folic acid + vitamin A + zinc

3. Vitamin A

4. Zinc + vitamin A

Outcomes

Clinical malaria

Cerebral malaria

Other infections

Hospital admissions

Mortality

NotesThe comparison of iron + folic acid + vitamin A vs. vitamin A is included in this review. The comparison of iron + folic acid + vitamin A + zinc vs. zinc + vitamin A is also eligible for inclusion but results will have to be published before they will be available for this review.