Agricultural and nutritional educational interventions for reducing aflatoxin exposure to improve infant and child growth in low‐ and middle‐income countries

Abstract

This is a protocol for a Cochrane Review (Intervention). The objectives are as follows:

To assess the effects of agricultural and nutritional educational interventions that aim to reduce the intake of aflatoxins in infants, children, and pregnant and lactating women at the household or community level in low‐ and middle‐income countries (LMIC), on prenatal and postnatal growth.

Background

Aflatoxins are a type of mycotoxin that affect global food security by causing the contamination of food crops, mainly in low and middle‐income countries (LMICs), such as those in Africa. Human exposure to aflatoxins is considered a major public health concern due to its potential harmful effects on human health. Aflatoxins are compounds that are produced during secondary metabolism by some members of the fungal genus Aspergillus (notably species in the sectionflavi) (Frisvad 2019). Four main forms of aflatoxin are commonly found in food crops, namely aflatoxin B1 (AFB1), B2 (AFB2), G1 (AFG1), and G2 (AFG2). AFB1 is the most potent and prevalent form of aflatoxin, accounting for an average of 70% of the total aflatoxin content in food. Another important type of aflatoxin, especially to infant and child health, is aflatoxin M1 (AFM1), which is a product of AFB1 hydroxylation during the metabolism of AFB1. AFM1 is a frequent contaminant of milk in lactating animals, resulting from the consumption of AFB1‐contaminated feed. AFM1 has also been reported in human breast milk (Watson 2017; WHO 2018).

A variety of food crops, such as cereals, ground and tree nuts, legumes, oilseeds, spices, milk, meat, and dried fruit are susceptible to colonisation with aflatoxin‐producing fungi. Factors influencing fungal and aflatoxin contamination of food crops include the climate of the region, the genotype of crop planted, soil type, stress or damage to the crop due to drought before harvesting, insect activity, poor timing of harvesting, heavy rains at and after harvesting, inadequate drying of the crop before storage, and poor storage conditions (Strosnider 2006). Maize and groundnuts are the major dietary sources of human aflatoxin exposure, due to high consumption rates of these foods as single staple foods by communities in LMICs. It is also likely that animal milk may be contaminated with AFM1 in some of these communities, depending on factors, such as the quantities of aflatoxin‐contaminated feed ingested by animals, genetics of the animals, seasonal variation, the milking process, and environmental conditions. Contaminated milk may also subsequently contaminate milk products, such as yoghourt and cheese (Iqbal 2015). Thus, many LMIC households are plagued with adverse health consequences of aflatoxin exposure. The negative impacts on human health include an increased risk of acute toxicity, as recorded in several outbreaks of aflatoxicosis in Africa (IARC 2015; Kamala 2018), and chronic effects, such as an increased risk of liver cancer (IARC 2015).

To protect people from the harmful effects of aflatoxin exposures, international food safety standards stipulate maximum levels for aflatoxins in various foods (Codex Alimentarius). These standards are operational in industrialised nations, but have little effect in LMICs. Typically, food consumed from subsistence farming rarely enters any sort of regulatory inspection for aflatoxin. Even if contamination levels were below the maximum levels, many people in LMICs consume such high levels of maize and groundnut products that their daily aflatoxin exposure would still render them vulnerable to disease (Klangwiset 2010).

Global dietary intake estimates for aflatoxin, based on estimates of typical maize and groundnut consumption, contamination levels, and body weight, indicate a much higher burden of exposure in LMICs in Sub‐Saharan Africa, China, Southeast Asia, and Latin America, compared with Western Europe and North America. Compounding the risk of exposure is the low level of surveillance and monitoring of these staple foods in LMICs. These countries often lack resources and analytical capacity, and as a result, there is a widening gap between the quality and quantity of prevalence data generated by laboratories in high‐income countries compared to LMICs (IARC 2015). Sampling procedures of foods for analytical purposes may also be problematic, since aflatoxin contamination is often not evenly distributed throughout batches of food, for example in stored grain (WHO 2018).

Biomarkers are objective markers of aflatoxin exposure and are more accurate in assessing the degree of individual exposure than food‐based exposure assessments. There are three validated biomarkers of aflatoxin exposure: urinary biomarkers (AFM1 and aflatoxin‐N7‐guanine), reflecting exposure in the prior 24 to 48 hours; and serum levels of aflatoxin‐albumin (AF‐alb), reflecting cumulative exposure over the prior two to three months. The application of the AF‐alb biomarker has confirmed a high prevalence of aflatoxin exposure in several locations in East and West Africa (IARC 2015), as well as parts of Asia, such as Malaysia (Leong 2012). This biomarker correlates with the dietary intake estimates of aflatoxin in children who consume maize‐based diets (Routledge 2014). Other aflatoxin metabolites (for example, serum AFM1 or AFG1, urinary AFB1 or AFG1, and milk AFM1 or AFG2) are indicative of exposure, but since their levels do not correlate with aflatoxin intake, they are not considered accurate biomarkers of exposure (Smith 2018). However, it should be noted that there is no validated ranking system available to categorise individuals with different exposure levels to aflatoxin.

Description of the condition

Child undernutrition is a major public health burden in LMICs. Stunting, defined as a height‐for‐age z‐score of more than two standard deviations below the child growth standard median is the most prevalent form of child undernutrition (WHO 1986; WHO 2006). Globally, stunting affects an estimated 151 million or 22.2% of children below the age of five years. Two‐thirds of all stunted children live in LMICs. Africa is the only region in the world where the number of stunted children has increased from 50.6 million in 2000 to 58.7 million children in 2017 (UNICEF‐WHO‐The World Bank 2018).

Growth faltering typically begins in utero, followed by subsequent growth faltering after birth, especially in the first two years of life. Foetal growth restriction is an important contributor to stunted linear growth in the postnatal period. Longitudinal data over ten years from LMICs estimate that 20% of childhood stunting can be attributed to foetal growth restriction (Christian 2013). Dietary exposure to high levels of aflatoxins during pregnancy has been widely documented in several LMICs, and may be a major contributor to foetal growth restriction, and in turn, to childhood stunting (Castelino 2014; Piekkola 2012; Shuaib 2010; Turner 2007). Cohort studies show that this exposure may contribute to adverse birth outcomes, such as low birth weight (Shuaib 2010), and stunted growth in the first year of life (Turner 2007). However, others found no association between exposure to aflatoxins in pregnancy and post‐natal growth outcomes. Smith 2018 suggests that the lack of associations in these studies may be partly explained by the use of transient exposure indicators (for example, AFM1, AFG1), instead of accurate biomarkers of exposure (for example, AF‐alb). Longitudinal data from West Africa on the association of aflatoxins and growth of young children suggest that serum AF‐alb levels are associated with stunted growth (Gong 2004). However, longitudinal studies from Tanzania and Nepal found no association between aflatoxin exposure and growth faltering (Mitchell 2017; Shirima 2015), possibly due to lower exposure levels to aflatoxins in these studies, when compared to Gong 2004. Aflatoxin exposure may affect child growth by three potential mechanisms: damage to the intestinal epithelium (enteropathy), liver toxicity, and reduced immune function. Enteropathy is associated with a reduced uptake of nutrients; liver damage may result in less production of insulin‐like growth factors; while immune suppression may enhance the susceptibility of aflatoxin‐exposed children to infections, such as diarrhoea (IARC 2015; Watson 2017).

Numerous factors contribute to growth faltering in the postnatal period. For example, infants in many households in LMICs are introduced to weaning foods (complementary foods) of low nutritional quality and at high risk of microbial contamination, resulting in high rates of diarrhoea and other infectious diseases. In addition, complementary foods fed to children are often made from locally grown, low quality, and often contaminated cereals and nuts, such that these infants are exposed to a range of mycotoxins. Several studies from Africa have reported the contamination of cereal‐ and nut‐based complementary foods with aflatoxins (Kimanya 2014; Ojuri 2018; Ojuri 2019). A household surveillance study in Nigeria reported AFB1 levels in cereal‐based complementary foods of about 100 times higher than the maximum limits. These levels remained high throughout the year, suggesting a chronic, high level of exposure (Ojuri 2018). Although aflatoxins are found in human breast milk in high risk regions of LMICs, the level of exposure of these infants to aflatoxins in breast milk is likely to be lower than those consuming complementary foods; thus the exclusive breastfeeding period is a window of lower exposure, and is critical to child health (Braun 2018; IARC 2015). However, the consumption of contaminated fresh cow's milk by some young children in LMICs, who mainly consume cereal‐based diets, may contribute significantly to their level of aflatoxin exposure, as milk from small‐scale dairy farms in Kenya and Brazil is frequently contaminated (Goncalves 2017; Kagera 2018).

Limited biomarker data in children suggest that the consumption of other mycotoxins that are present in staple foods, such as fumonisin in maize or deoxynivalenol (DON) in wheat, may have interactive effects on child growth. Chronic exposure to both mycotoxins, in addition to aflatoxins, was documented in children aged between 6 and 24 months from Tanzania (Kimanya 2014; Shirima 2015; Srey 2014). Fumonisin alone, or in combination with aflatoxins, was associated with stunting in one of these studies (Shirima 2015).

In summary, aflatoxin exposure during pregnancy and the first 1000 days of life may be an important contributing factor to growth faltering in infants and young children, especially in LMICs. Infants and young children from households consuming diets lacking diversity, i.e. consisting mainly of a single staple food, such as maize, are likely to be most at risk of exposure to aflatoxins. It was previously suggested that exposure to aflatoxins may be related to the development of severe acute malnutrition in children, but no prospective data are available (Watson 2017).

Description of the intervention

This review will examine agricultural education and nutritional education as interventions that are aimed at reducing aflatoxin intake at the household or community level in LMIC settings. We will exclude intervention studies that are primarily aimed at reducing the accumulation of aflatoxins in food crops, such as agricultural technologies (for example genetically resistant, or modified food crops, biocontrol methods) or practices during the planting stage. We will also exclude studies investigating the supplementation of human diets with dietary enterosorbents (specific foods or compounds to reduce the bio‐availability of aflatoxin from contaminated foods), since these interventions are primarily aimed at reducing human exposure to aflatoxins during outbreaks of acute aflatoxicosis.

High risk households often rely on subsistence farming and the production or consumption of a single staple food, such as maize. Agricultural education to reduce human consumption of aflatoxins may include interventions that train household and community members involved in agricultural production at the household or community level, regarding the use of optimal agricultural practices after harvesting (for example, sorting, storage, etc). Nutritional education to reduce the consumption of aflatoxins by young children may include interventions such as the education of mothers (during pregnancy or post‐partum) or caregivers regarding increasing the variety of their diet (household dietary diversity), rather choosing foods at low risk of aflatoxin contamination instead of those known to be at higher risk, and optimal food preparation practices, including traditional practices. These interventions may also involve replacing contaminated foods, by providing uncontaminated foods to households, the households' food supplies, or the community food supply (for example, local food shops).

How the intervention might work

Agricultural education

Agricultural educational interventions that promote optimal agricultural practices during the post‐harvesting period, aimed at household or community members, may reduce the accumulation of aflatoxins in staple food crops of LMIC households or communities; therefore lowering the exposure of infants, children, and pregnant and lactating women from these households or communities to food crops that contain high levels of aflatoxins.

Interventions to reduce the accumulation of aflatoxins in food crops include the promotion of practices such as early harvesting; sorting of freshly harvested cereals or nuts to remove those that are physically damaged or have visible moulds using methods such as hand sorting or flotation; ensuring that crops are dried properly as soon as possible after harvesting; storing crops in dry, well‐vented structures; or administering appropriate amounts of registered insecticides to minimise insects in storage facilities. Educational interventions may also promote the use of processing practices, such as washing, crushing, wet and dry milling, and dehulling, as recommended by international food safety standards to reduce the aflatoxin contamination of cereals (CAC 2003;Matumba 2015; Karlovsky 2016; Okeke 2018).

Nutritional education, with or without replacement of contaminated foods

Nutritional educational interventions that promote a greater variety of food, and are aimed at mothers or caregivers from LMIC communities that consume diets consisting mostly of a single staple food, may lead to lower intake of aflatoxins for infants and young children in their households. Mothers or caregivers could also be educated about food choices that are likely to reduce the exposure of their households to aflatoxin, for example, consuming sorghum and millet instead of maize (Bandyopadhyay 2007; Klangwiset 2010). Education may also include an explanation of various food preparation practices that can reduce household consumption of aflatoxins.

Aflatoxins are largely unaffected by routine cooking temperatures, since they decompose at temperatures of 237 ˚C to 306 ˚C. Boiling maize grids reduces aflatoxins by only 28% (Kabak 2008). However, the amount of water used during cooking may be a critical factor in the amount of aflatoxins left for human consumption, due to the dilution effect (Ezekiel 2019). In regions where water is scarce, the grain is more likely to be cooked in only enough water to be soaked, whereas in others, grains are washed or cooked in plenty of water (or both), some of which is then discarded (Edwards 2018). Boiling rice in excess water or with pressure cooking methods, results in a reduction of aflatoxin content from 88% to 89% (Kabak 2008). Traditional methods of cooking maize meal in Central and South America (called nixtamalisation) involves the addition of alkaline compounds, such as lime, which is then washed out. It has been shown that when there is sufficient washing of the lime‐treated product before consumption, the aflatoxin levels are reduced (IARC 2015). However, the efficacy of this method has been questioned, since the chemical reaction that temporarily inactivates aflatoxins may reverse in the gastric acid of the stomach (Strosnider 2006). Other food preparation methods that reduce aflatoxins include roasting (especially in peanuts), baking, or frying (Afolabi 2014; CAC 2003).

Access of households to industrially‐processed foods, such as infant cereals, or spreads and pastes containing nuts, can contribute to aflatoxin exposure, especially of infants and young children (Ojuri 2019).

Extrusion processing is used by the food industry to manufacture processed foods, such as breakfast cereals and snacks, from maize or peanut flour, and involves high temperatures (> 150 ˚C), high pressure, and severe shear forces. The destruction of aflatoxins during this process is dependent on factors, such as extruder temperature, screw speed, and moisture content of the extrusion mixture. Such destruction processes usually only remove a proportion of the aflatoxins present (Kabak 2008; Karlovsky 2016).

Costs and cost‐effectiveness

The costs and estimated cost‐effectiveness (in terms of reductions in aflatoxin exposure) of some of the agricultural and nutritional educational intervention strategies, as described above, have been reviewed by Klangwiset 2010. They described the promotion of good post‐harvest agricultural practices, as a result of educational intervention strategies, as highly cost‐effective (Klangwiset 2010). Although the costs of replacing contaminated food crops have not been reported, strategies like these are likely to be costly.

Why it is important to do this review

Undernourished children in LMICs are at an increased risk of death from infectious diseases (Black 2013). More specifically, linear growth faltering in the period between conception and the first two years of life is associated with poorer cognitive function in early childhood; sustained effects on cognition, executive function, and school attainment throughout childhood; and ultimately, reduced economic productivity in adulthood, due to lost cognitive potential (Black 2017; Sudfeld 2015). The costs of childhood stunting have been estimated as a reduction in the current per capita income of the workforce in countries in Sub‐Saharan Africa and South Asia, of up to 10% (Galasso 2017).

A recent literature review on child undernutrition stated that current interventions aimed at reducing child undernutrition may be undermined by high levels of exposure to aflatoxins in populations of vulnerable mothers and children (Watson 2017). There is a need to systematically evaluate the effectiveness of different preventive strategies on the exposure by pregnant women and children to aflatoxins, and also to assess whether these strategies are successful in promoting childhood growth and health. The findings from this review will assist decision makers in local, regional, and national governments of affected areas to formulate effective public health interventions for local, affected communities. Public health strategies, such as the targeted application of effective agricultural practices, or dietary interventions (or both) to reduce aflatoxin exposure in communities, may be able to assist in the prevention of childhood growth faltering, and promote the long‐term well‐being of communities in LMICs. Understanding the costs and feasibility of different aflatoxin control interventions can also help decision makers to optimally allocate resources in LMICs.

Objectives

To assess the effects of agricultural and nutritional educational interventions that aim to reduce the intake of aflatoxins in infants, children, and pregnant and lactating women at the household or community level in low‐ and middle‐income countries (LMIC), on prenatal and postnatal growth.

Methods

Criteria for considering studies for this review

Types of studies

We will include randomised controlled trials (RCTs) and cluster‐randomised controlled trials with at least two intervention and two control sites (EPOC 2017a). We are including cluster‐randomised trials, as it is likely that eligible studies will randomise different communities, instead of individual households, or people within communities.

Types of participants

Infants and children (aged < 18 years at the start of the study), and pregnant and lactating women, from low‐ and middle‐income countries (LMIC), as defined by the World Bank (World Bank 2019).

Types of interventions

We will include agricultural and nutritional educational interventions, of any duration, which aim to reduce aflatoxin intake of infants, children, and pregnant and lactating women, at household and community levels in LMICs. The comparator must be an inactive control intervention, i.e. no intervention, or usual support (as reported by the study authors). For example, if a community is already supported by agricultural aid workers, these workers will provide routine guidance to households on the prevention of mycotoxins in the control group, whereas households in the intervention community could receive education and support relating to specific intervention components (such as those described above). We will also include studies in which households or communities are randomly assigned to a waiting list, and receive the intervention after the intervention group (a waiting list control group (Higgins 2011)). We will exclude any study that randomises two different educational interventions to reduce household or community aflatoxin consumption without including an inactive control group.

Interventions will be broadly categorised as:

• Agricultural education

• Nutritional education, with or without replacement of contaminated foods

Types of outcome measures

We will analyse the following outcomes from our included studies. However, the outcomes reported by individual studies will not form part of the eligibility criteria for this review.

Primary outcomes

Prenatal growth outcomes measured in infants at birth

• Birth length for gestational age z‐score

• Birth weight for gestational age z‐score

• Low birth weight (less than 2500 g)

Postnatal growth outcomes measured during infancy, childhood, and adolescence (up to the age of 18 years)

• Length‐ or height‐for‐age z‐score (LAZ or HAZ)

• Stunting (defined as LAZ or HAZ more than two standard deviations below the reference median value)

Secondary outcomes

Prenatal growth outcomes measured in infants at birth

• Length at birth (cm)

• Weight at birth (g)

• Gestational age (weeks)

Postnatal growth outcomes measured during infancy, childhood, and adolescence (up to the age of 18 years)

• Weight‐for‐height z‐score (WHZ)

• Wasting (defined as WLZ or WHZ more than two standard deviations below the reference median value)

• Weight‐for‐age z‐score (WAZ)

Other secondary outcomes

• Morbidity from infectious diseases in children (for example, diarrhoea, malaria, HBV and HIV infection diagnosed by a medical doctor)

• Unintended effects of agricultural and nutritional educational interventions to reduce the aflatoxin intake of infants, children, pregnant and lactating women (for example, an increase in household food expenditure).

Timing of outcome assessment

We will exclude any study with a length of follow‐up (i.e. time from baseline to first outcome measurement) of less than four weeks.

We plan to group the analyses of changes in outcomes in the short‐term (1 to 3 months), medium‐term (from more than 3 months to 6 months, from more than 6 months to 12 months), and long‐term (> 1 year, > 2 years, > 3 years, etc).

Search methods for identification of studies

We will attempt to identify all relevant studies, regardless of language or publication status (published, unpublished, in press, ongoing).

Electronic searches

Our detailed search strategy for MEDLINE PubMed is in Appendix 1. This strategy will be adapted for the other electronic sources and reported in full in the review.

International databases

• Cochrane Central Register for Controlled Trials (CENTRAL)

• MEDLINE PubMed

• Embase Ovid

• CINAHL EBSCO

• Web of Science – core collection

• Africa‐Wide EBSCO

• LILACS (Latin American and Caribbean Health Science Information database) virtual health library

• CAB direct

• Agricola

We will search the following registries to identify unpublished or ongoing studies:

• ClinicalTrials.gov (www.clinicaltrials.gov);

• WHO International Clinical Trials Registry Platform (apps.who.int/trialsearch/)

We will translate search records written in a language other than English, as possible. Otherwise, we will place these in the ’Studies awaiting classification’ section of the review until we can obtain a translation.

Searching other resources

We will check the reference lists of all the included studies, and contact relevant organisations and agencies (such as the Partnership for Aflatoxin Control In Africa (PACA); the Consultative Group on International Agricultural Research (CGIAR) Programme on Agriculture for Nutrition and Health; the UK Department for International development (DFID); the International Food Policy Research Institute (IFPRI); the Food Safety and Codex Unit, Food and Agriculture Organization (FAO); and the Department of Food Safety and Zoonoses, World Health Organization (WHO)). We will also contact experts in the field and the authors of relevant ongoing studies to obtain any additional or unpublished data, if available.

Data collection and analysis

Selection of studies

After removing duplicate search records, using Covidence software (Covidence), two review authors (MV, CNE) will independently screen the remaining titles and abstracts of the retrieved records, to assess eligibility for inclusion. We will obtain the full‐text articles of records identified as potentially eligible; two review authors will then independently screen the full‐text articles to determine final eligibility. If full‐text articles cannot be obtained, or if we have the full‐text article but need more information about the study to determine eligibility, we will attempt to contact the study authors to obtain further study details. If we are unsuccessful in obtaining further information, we will classify such studies as ’Studies awaiting classification’ until further detail is published or made available to us. We will resolve any disagreements at any stage of the eligibility assessment process through discussion and consultation with a third review author (AS), where necessary.

Data extraction and management

Two review authors (MV, CNE) will independently extract information from the included studies into Covidence, after piloting the data extraction template. We will extract information on study design, funding source, study setting, types of participants, a description of the interventions examined, and costs of the intervention, if reported. We will use the PROGRESS framework (Cochrane‐Campbell Methods Group Equity checklist) to record the relevant baseline characteristics of the studies' participants (O'Neill 2014).

We will extract data for primary and secondary outcomes at all time points, using Covidence. Where a study reports outcome data for multiple time points, i.e. more than one time point per analysis period, we will extract and use the data from the longest time point (for example, where results are available at 6 and 9 months, we will only use the 9‐month data for the analysis period 'from more than 6 months to 12 months'). For individual studies that do not use the metric system to report outcomes, we will convert the outcomes to SI units, where possible.

We will resolve any disagreements at any stage of the data extraction and management process through discussion and consultation with a third review author (AS), where necessary.

Assessment of risk of bias in included studies

We will use the Cochrane Effective Practice and Organisation of Care Group ’Risk of bias’ tool as a framework to assess the risk of bias of all included studies, in Covidence (Covidence; EPOC 2017b). Two review authors (MV, CNE) will independently assess the risk of bias for each study; any disagreement will be resolved by discussion and reaching consensus, or by involving a third review author (AS).

Assessing risk of bias in randomised controlled trials (RCTs) and cluster‐RCTs

The 'Risk of bias' assessment at study level will consider study design and reporting characteristics relevant to our pre‐specified growth outcomes at birth or postnatally, where available. All growth outcomes are observer‐reported and do not involve judgements (Higgins 2019). We will assess the following nine domains (Higgins 2017):

(1) Random sequence generation

We will assess studies as:

• low risk of bias if there is a random component in the sequence generation process (for example, random number table; computer random number generator);

• high risk of bias if a non‐random approach has been used (for example, odd or even date of birth); and

• unclear risk of bias if not specified in the paper.

(2) Allocation concealment

We will assess studies as:

• low risk of bias if the unit of allocation was by individual household, and there was some form of centralised randomisation scheme;

• high risk of bias if investigators enrolling households could possibly foresee assignments and potentially introduce selection bias (e.g. open random allocation); and

• unclear risk of bias if not specified in the paper.

Cluster‐randomised trials often randomise all clusters (i.e. communities) at once, therefore, the lack of concealment of an allocation sequence should not usually be an issue (Higgins 2011).

(3) Baseline outcome measurements similar

Studies will be assessed as:

• low risk of bias if baseline outcome characteristics of participants from the study and control groups are reported and are similar;

• high risk of bias if important differences in outcomes between groups were present prior to intervention and were not adjusted for in the analysis; and

• unclear risk of bias if there was no baseline measure of outcome (note: if assessed as high or unclear risk, but there is sufficient information to do an adjusted analysis, the assessment should be low).

(4) Baseline characteristics similar

We will assess studies as:

• low risk of bias if individual participant or household characteristics were measured prior to the intervention, and no important differences were present across study groups. In RCTs, score low risk if imbalanced but appropriate adjusted analysis was performed (for example, analysis of covariance);

• high risk of bias if there is no report of characteristics in text or tables, or if there are differences between control and intervention groups; and

• unclear risk of bias if it is not clear in the paper (for example, when characteristics are mentioned in text but no data were presented).

A small number of clusters are often randomised in cluster‐randomised controlled trials, therefore, there is a possibility of chance baseline imbalance between the randomised groups of either the clusters (e.g. communities) or individual households within the clusters (Higgins 2011).

(5) Incomplete outcome data

We will assess outcomes in each included study as:

• low risk of bias if missing outcome measures were unlikely to bias the results (for example, the proportion of missing data for individual pregnant women or children was similar in the intervention and control groups, or the proportion of missing data was less than the effect size, i.e. unlikely to overturn the study result);

• high risk of bias if missing outcome data were likely to bias the results; and

• unclear risk of bias if not specified in the paper.

(6) Knowledge of the allocated interventions adequately prevented during the study

We will assess the risk of performance bias associated with blinding as:

• low risk if the study authors state explicitly that the primary outcome variables were assessed blindly, or the outcomes are objective, for example, anthropometric measurements of infants or children;

• high risk if the outcomes were not assessed blindly; and

• unclear risk if not specified in the paper.

(7) Protection against contamination

We will assess studies as:

• low risk of bias if allocation was by community and it is unlikely that the control group received the intervention;

• high risk of bias if it is likely that the control group received the intervention; and

• unclear risk of bias if community aid workers were allocated within a specific geographical area and it is possible that communication between intervention and control aid workers could have occurred.

(8) Selective outcome reporting

We will assess studies as:

• low risk of bias if there is no evidence that outcomes were selectively reported (for example, the study protocol is available and all of the study’s prespecified (primary and secondary) outcomes that are of interest in the review have been reported in the prespecified way; if the study protocol is not available, it is clear that the published reports include all expected outcomes, including those that were prespecified);

• high risk of bias if some important outcomes were subsequently omitted from the results (for example, not all prespecified primary outcomes were reported); one or more primary outcomes were reported using measurements, analysis methods, or subsets of the data that were not prespecified; one or more primary outcomes were not prespecified (for example, one or more reported primary outcomes were not prespecified); or the study report failed to include results for a key outcome that would be expected to have been reported for such a study; and

• unclear risk of bias if outcomes were not prespecified in the study protocol or published report.

(9) Other risks of bias

We will detail other possible sources of bias (if any) for each included study and give a rating of low, high, or unclear risk of bias for this item.

For cluster‐RCTs, we will also assess the risk of bias for the following domains (Higgins 2011):

• recruitment bias (for example, when individuals are recruited to the trial after the clusters have been randomised);

• incorrect analysis (when clustering is not taken into account in the analysis); and

• comparability with individually randomised trials.

Overall risk of bias assessment

We will assess the overall risk of bias of an included study as follows:

• low risk (low risk of bias for all key domains);

• unclear risk (unclear risk of bias for one or more key domains); and

• high risk (high risk of bias for one or more key domains.

Included studies at high risk of bias will be those with a high or unclear risk of bias in the following domains: similarity of baseline outcome measurements (selection bias) and incomplete outcome data (attrition bias). These 'Risk of bias' summary assessments will inform our sensitivity analysis (see Sensitivity analysis).

We will assess the overall risk of bias of outcomes included in the Summary of Findings table across included studies as:

• low risk (most information is from studies at low risk of bias);

• unclear risk (most information is from studies at low or unclear risk of bias), and

• high risk (the proportion of information from studies at high risk of bias is sufficient to affect the interpretation of results (EPOC 2017c)).

These 'Risk of bias' summary assessments will inform our judgements regarding the quality of the evidence for each outcome, as part of the GRADE process, in our ‘Summary of findings’ tables (see Data synthesis).

Measures of treatment effect

For dichotomous outcomes (for example, prevalence of stunting), we will present proportions; for two‐group comparisons, we will present results as risk ratios (RRs) or odds ratios (ORs) with 95% confidence intervals (CIs).

For continuous outcomes (for example, weight at birth), we will use the mean differences (MDs) with 95% CIs if outcomes are measured in the same way between trials. Where continuous data have been reported using different units across included studies, we will calculate and present the standardised mean difference (SMD). In the case of some studies reporting endpoint data and others reporting changes from baseline data, we will enter both types of data in the same meta‐analysis if the outcomes have been reported using the same unit or scale.

Unit of analysis issues

Studies with more than two intervention groups:

If we identify studies with more than two intervention groups, where possible, we will combine groups to create a single pair‐wise comparison, or use the methods set out in the Cochrane Handbook for Systematic Reviews of Interventions to avoid double counting of study participants (Higgins 2011). If the control group is shared by two or more study arms in a meta‐analysis, we will divide the control group over the number of relevant subgroup categories to avoid double counting the participants (i.e. for dichotomous data, we will divide the events and the total population, while for continuous data, we will assume the same mean and standard deviation, but will divide the total population).

Cluster‐randomised controlled trials:

If the study authors of cluster‐randomised trials have not appropriately accounted for the cluster design in their analyses, we will request the intra cluster correlation coefficient (ICC) from the study authors. If the ICC is not available, we will calculate the trial's effective sample size to account for the effect of clustering the data. In order to do so, we will use the ICC derived from the trial (if available), or from another source (for example, using the ICC derived from other, similar trials); or we will estimate the ICC, giving reasons for our choice, and then calculate the design effect, which is 1 + (c − 1) ICC, where c is the average cluster size. Estimated values are arbitrary, but we prefer to use them to adjust the effect estimates due to the implausibility that the ICC is actually zero.

For continuous data, we only need to adjust for the sample size; we will not change the means and standard deviations (SDs). For dichotomous outcomes, we will divide the sample size and the number of people who experienced the event by the design effect. Then, we will combine the estimates from the cluster RCTs with trials that have parallel group designs (Higgins 2011).

Dealing with missing data

We will contact study authors to request any missing or unreported data, such as group means, standard deviations, details of attrition, or details of interventions received by the control groups. If outcome data are only reported for participants completing the trial, we will contact the study authors for additional information to enable an intention‐to‐treat analysis as far as possible. We will assess the extent and impact of missing data and attrition for each included study during the ’Risk of bias’ assessment.

Assessment of heterogeneity

For each meta‐analysis, we will examine the forest plots visually to determine whether heterogeneity of the size and direction of treatment effect is present between studies. We will use the I² statistic, Tau², and the Chi² test to quantify the level of heterogeneity among the studies in each analysis. We will define substantial heterogeneity as Tau² > 0, and either I² > 50% or a low P value (< 0.10) in the Chi² test. We will note this in the text, and explore it by conducting the prespecified subgroup analyses to account for potential sources of clinical heterogeneity (see Subgroup analysis and investigation of heterogeneity). We will also consider other potential sources of clinical heterogeneity, for example, differences in the nature of the interventions delivered or the presence of co‐contamination with other mycotoxins. We will also examine methodological sources of heterogeneity by examining studies with different levels of risk of bias in a sensitivity analysis (see Sensitivity analysis). We will use caution in the interpretation of results with high levels of unexplained heterogeneity. We will not perform a meta‐analysis if the I² statistic is higher than 90%.

Assessment of reporting biases

Where we suspect reporting bias (see selective outcome reporting), we will attempt to contact study authors, asking them to provide missing outcome data. Where this is not possible, and the missing data are thought to introduce serious bias, we will explore the impact of including such studies in the overall assessment of results using a sensitivity analysis. We do not anticipate that there will be sufficient studies contributing data for any particular outcome for us to examine possible publication bias; however, for future updates of the review, if more than 10 studies reporting the same outcome of interest are available, we will generate funnel plots and visually examine them for asymmetry (Review Manager 2014).

Data synthesis

We will use a random‐effects meta‐analysis to combine data across more than one study, as we anticipate that there may be natural heterogeneity between studies, attributable to the different study settings, intervention strategies, or both. For meta‐analysis of RCTs, we intend to use the crude or unadjusted effect estimates (Review Manager 2014).

Should data not be suitable for pooling in meta‐analyses, we will present the data in forest plots without the pooled estimate, or in tables, as appropriate. If needed, we will conduct a narrative synthesis, by adopting a systematic approach to presentation, as per the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011), and the reporting guideline Improving Conduct and Reporting of Narrative Synthesis of Quantitative Data (ICONS‐Quant (Campbell 2018)).

We will present the main results as ’Summary of findings’ tables for each comparison, by summarising the following primary and secondary outcomes at medium‐term follow‐up time points (where applicable).

We will include the following primary and secondary outcomes in our 'Summary of Findings' table:

• Birth length for gestational age z‐score;

• Birth weight for gestational age z‐score;

• Low birth weight (defined as less than 2500 g);

• Length‐ or height‐for‐age z‐score (LAZ or HAZ);

• Stunting (defined as LAZ or HAZ more than two standard deviations below the reference median value);

• Weight‐for‐height z‐score (WHZ); and

• Unintended effects of agricultural and nutritional educational interventions to reduce the aflatoxin intake of infants, children, pregnant and lactating women.

For each individual outcome in the 'Summary of findings' tables, we will assess the quality of the evidence using the GRADE approach and GRADEpro GDT software (Balshem 2011; Gradepro GDT).

Subgroup analysis and investigation of heterogeneity

If we detect substantial heterogeneity in our primary outcomes, we plan to carry out the following subgroup analyses, where data allow, i.e. where we have three or more included studies in a meta‐analysis.

• Baseline age of infant and child participants: up to 6 months; from older than 6 months to 12 months; from older than 12 months to 24 months; > 24 months (preschool); primary school; secondary school.

• Length of follow‐up; short‐term (1 to 3 months), medium‐term (from more than 3 months to 6 months, from more than 6 months to 12 months), and long‐term (> 1 year, > 2 years, > 3 years, etc).

• Baseline dietary diversity (defined as a qualitative measure of food consumption reflecting household access to a variety of foods; this can also be a proxy for the nutrient adequacy of the diet of individuals (FAO 2011));

  • for child participants (child dietary diversity score (CDDS), calculated by counting the number of food groups consumed within a reference period; maximum score of 7 (WHO 2007));

  • for pregnant and lactating mothers (women's dietary score (WDDS), calculated by counting the number of food groups consumed within a reference period; maximum score of 9 (FAO and FHI 360));

  • on household level (household dietary diversity score (HDDS), calculated by counting the number of food groups consumed within a reference period; maximum score of 12 (Swindale 2006)).

Sensitivity analysis

We plan to conduct sensitivity analyses for primary outcomes if we have three or more studies per meta‐analysis, assessing the effect of:

• Risk of bias: removing studies with a high risk of bias ‐ see Assessment of risk of bias in included studies; and

• Clustering effect: assessing the strength of the clustering effect in our analysis for studies that have not been adjusted for clustering. In studies where the ICC was estimated from similar studies, sensitivity analysis will be conducted by using higher and lower assumptions of the strength of the clustering effect.

Appendices

Appendix 1. MEDLINE PubMed search strategy

Search	Query
#23	Search #17 AND #22
#22	Search #18 OR #19 OR #20 OR #21
#21	Search ("developing country"[tw] OR "developing countries"[tw] OR "developing nation"[tw] OR "developing nations"[tw] OR "developing population"[tw] OR "developing populations"[tw] OR "developing world"[tw] OR "less developed country"[tw] OR "less developed countries"[tw] OR "less developed nation"[tw] OR "less developed nations"[tw] OR "less developed population"[tw] OR "less developed populations"[tw] OR "less developed world"[tw] OR "lesser developed country"[tw] OR "lesser developed countries"[tw] OR "lesser developed nation"[tw] OR "lesser developed nations"[tw] OR "lesser developed population"[tw] OR "lesser developed populations"[tw] OR "lesser developed world"[tw] OR "under developed country"[tw] OR "under developed countries"[tw] OR "under developed nation"[tw] OR "under developed nations"[tw] OR "under developed population"[tw] OR "under developed populations"[tw] OR "under developed world"[tw] OR "underdeveloped country"[tw] OR "underdeveloped countries"[tw] OR "underdeveloped nation"[tw] OR "underdeveloped nations"[tw] OR "underdeveloped population"[tw] OR "underdeveloped populations"[tw] OR "underdeveloped world"[tw] OR "middle income country"[tw] OR "middle income countries"[tw] OR "middle income nation"[tw] OR "middle income nations"[tw] OR "middle income population"[tw] OR "middle income populations"[tw] OR "low income country"[tw] OR "low income countries"[tw] OR "low income nation"[tw] OR "low income nations"[tw] OR "low income population"[tw] OR "low income populations"[tw] OR "lower income country"[tw] OR "lower income countries"[tw] OR "lower income nation"[tw] OR "lower income nations"[tw] OR "lower income population"[tw] OR "lower income populations"[tw] OR "underserved country"[tw] OR "underserved countries"[tw] OR "underserved nation"[tw] OR "underserved nations"[tw] OR "underserved population"[tw] OR "underserved populations"[tw] OR "underserved world"[tw] OR "under served country"[tw] OR "under served countries"[tw] OR "under served nation"[tw] OR "under served nations"[tw] OR "under served population"[tw] OR "under served populations"[tw] OR "under served world"[tw] OR "deprived country"[tw] OR "deprived countries"[tw] OR "deprived nation"[tw] OR "deprived nations"[tw] OR "deprived population"[tw] OR "deprived populations"[tw] OR "deprived world"[tw] OR "poor country"[tw] OR "poor countries"[tw] OR "poor nation"[tw] OR "poor nations"[tw] OR "poor population"[tw] OR "poor populations"[tw] OR "poor world"[tw] OR "poorer country"[tw] OR "poorer countries"[tw] OR "poorer nation"[tw] OR "poorer nations"[tw] OR "poorer population"[tw] OR "poorer populations"[tw] OR "poorer world"[tw] OR "developing economy"[tw] OR "developing economies"[tw] OR "less developed economy"[tw] OR "less developed economies"[tw] OR "lesser developed economy"[tw] OR "lesser developed economies"[tw] OR "under developed economy"[tw] OR "under developed economies"[tw] OR "underdeveloped economy"[tw] OR "underdeveloped economies"[tw] OR "middle income economy"[tw] OR "middle income economies"[tw] OR "low income economy"[tw] OR "low income economies"[tw] OR "lower income economy"[tw] OR "lower income economies"[tw] OR "low gdp"[tw] OR "low gnp"[tw] OR "low gross domestic"[tw] OR "low gross national"[tw] OR "lower gdp"[tw] OR "lower gnp"[tw] OR "lower gross domestic"[tw] OR "lower gross national"[tw] OR lmic[tw] OR lmics[tw] OR "third world"[tw] OR "lami country"[tw] OR "lami countries"[tw] OR "transitional country"[tw] OR "transitional countries"[tw])
#20	Search (Africa[tw] OR Asia[tw] OR Caribbean[tw] OR West Indies[tw] OR South America[tw] OR Latin America[tw] OR Central America[tw] OR Afghanistan[tw] OR Albania[tw] OR Algeria[tw] OR Angola[tw] OR Antigua[tw] OR Barbuda[tw] OR Argentina[tw] OR Armenia[tw] OR Armenian[tw] OR Aruba[tw] OR Azerbaijan[tw] OR Bahrain[tw] OR Bangladesh[tw] OR Barbados[tw] OR Benin[tw] OR Byelarus[tw] OR Byelorussian[tw] OR Belarus[tw] OR Belorussian[tw] OR Belorussia[tw] OR Belize[tw] OR Bhutan[tw] OR Bolivia[tw] OR Bosnia[tw] OR Herzegovina[tw] OR Hercegovina[tw] OR Botswana[tw] OR Brasil[tw] OR Brazil[tw] OR Bulgaria[tw] OR Burkina Faso[tw] OR Burkina Fasso[tw] OR Upper Volta[tw] OR Burundi[tw] OR Urundi[tw] OR Cambodia[tw] OR Khmer Republic[tw] OR Kampuchea[tw] OR Cameroon[tw] OR Cameroons[tw] OR Cameron[tw] OR Camerons[tw] OR Cape Verde[tw] OR Central African Republic[tw] OR Chad[tw] OR Chile[tw] OR China[tw] OR Colombia[tw] OR Comoros[tw] OR Comoro Islands[tw] OR Comores[tw] OR Mayotte[tw] OR Congo[tw] OR Zaire[tw] OR Costa Rica[tw] OR Cote d'Ivoire[tw] OR Ivory Coast[tw] OR Croatia[tw] OR Cuba[tw] OR Cyprus[tw] OR Czechoslovakia[tw] OR Czech Republic[tw] OR Slovakia[tw] OR Slovak Republic[tw] OR Djibouti[tw] OR French Somaliland[tw] OR Dominica[tw] OR Dominican Republic[tw] OR East Timor[tw] OR East Timur[tw] OR Timor Leste[tw] OR Ecuador[tw] OR Egypt[tw] OR United Arab Republic[tw] OR El Salvador[tw] OR Eritrea[tw] OR Estonia[tw] OR Ethiopia[tw] OR Fiji[tw] OR Gabon[tw] OR Gabonese Republic[tw] OR Gambia[tw] OR Gaza[tw] OR Georgia Republic[tw] OR Georgian Republic[tw] OR Ghana[tw] OR Gold Coast[tw] OR Greece[tw] OR Grenada[tw] OR Guatemala[tw] OR Guinea[tw] OR Guam[tw] OR Guiana[tw] OR Guyana[tw] OR Haiti[tw] OR Honduras[tw] OR Hungary[tw] OR India[tw] OR Maldives[tw] OR Indonesia[tw] OR Iran[tw] OR Iraq[tw] OR Isle of Man[tw] OR Jamaica[tw] OR Jordan[tw] OR Kazakhstan[tw] OR Kazakh[tw] OR Kenya[tw] OR Kiribati[tw] OR Korea[tw] OR Kosovo[tw] OR Kyrgyzstan[tw] OR Kirghizia[tw] OR Kyrgyz Republic[tw] OR Kirghiz[tw] OR Kirgizstan[tw] OR "Lao PDR"[tw] OR Laos[tw] OR Latvia[tw] OR Lebanon[tw] OR Lesotho[tw] OR Basutoland[tw] OR Liberia[tw] OR Libya[tw] OR Lithuania[tw])
#19	Search (Macedonia[tw] OR Madagascar[tw] OR Malagasy Republic[tw] OR Malaysia[tw] OR Malaya[tw] OR Malay[tw] OR Sabah[tw] OR Sarawak[tw] OR Malawi[tw] OR Nyasaland[tw] OR Mali[tw] OR Malta[tw] OR Marshall Islands[tw] OR Mauritania[tw] OR Mauritius[tw] OR Agalega Islands[tw] OR Mexico[tw] OR Micronesia[tw] OR Middle East[tw] OR Moldova[tw] OR Moldovia[tw] OR Moldovian[tw] OR Mongolia[tw] OR Montenegro[tw] OR Morocco[tw] OR Ifni[tw] OR Mozambique[tw] OR Myanmar[tw] OR Myanma[tw] OR Burma[tw] OR Namibia[tw] OR Nepal[tw] OR Netherlands Antilles[tw] OR New Caledonia[tw] OR Nicaragua[tw] OR Niger[tw] OR Nigeria[tw] OR Northern Mariana Islands[tw] OR Oman[tw] OR Muscat[tw] OR Pakistan[tw] OR Palau[tw] OR Palestine[tw] OR Panama[tw] OR Paraguay[tw] OR Peru[tw] OR Philippines[tw] OR Philipines[tw] OR Phillipines[tw] OR Phillippines[tw] OR Poland[tw] OR Portugal[tw] OR Puerto Rico[tw] OR Romania[tw] OR Rumania[tw] OR Roumania[tw] OR Russia[tw] OR Russian[tw] OR Rwanda[tw] OR Ruanda[tw] OR Saint Kitts[tw] OR St Kitts[tw] OR Nevis[tw] OR Saint Lucia[tw] OR St Lucia[tw] OR Saint Vincent[tw] OR St Vincent[tw] OR Grenadines[tw] OR Samoa[tw] OR Samoan Islands[tw] OR Navigator Island[tw] OR Navigator Islands[tw] OR Sao Tome[tw] OR Saudi Arabia[tw] OR Senegal[tw] OR Serbia[tw] OR Montenegro[tw] OR Seychelles[tw] OR Sierra Leone[tw] OR Slovenia[tw] OR Sri Lanka[tw] OR Ceylon[tw] OR Solomon Islands[tw] OR Somalia[tw] OR Sudan[tw] OR Suriname[tw] OR Surinam[tw] OR Swaziland[tw] OR Syria[tw] OR Tajikistan[tw] OR Tadzhikistan[tw] OR Tadjikistan[tw] OR Tadzhik[tw] OR Tanzania[tw] OR Thailand[tw] OR Togo[tw] OR Togolese Republic[tw] OR Tonga[tw] OR Trinidad[tw] OR Tobago[tw] OR Tunisia[tw] OR Turkey[tw] OR Turkmenistan[tw] OR Turkmen[tw] OR Uganda[tw] OR Ukraine[tw] OR Uruguay[tw] OR USSR[tw] OR Soviet Union[tw] OR Union of Soviet Socialist Republics[tw] OR Uzbekistan[tw] OR Uzbek OR Vanuatu[tw] OR New Hebrides[tw] OR Venezuela[tw] OR Vietnam[tw] OR Viet Nam[tw] OR West Bank[tw] OR Yemen[tw] OR Yugoslavia[tw] OR Zambia[tw] OR Zimbabwe[tw] OR Rhodesia[tw])
#18	Search (Developing Countries[Mesh:noexp] OR Africa[Mesh:noexp] OR Africa, Northern[Mesh:noexp] OR Africa South of the Sahara[Mesh:noexp] OR Africa, Central[Mesh:noexp] OR Africa, Eastern[Mesh:noexp] OR Africa, Southern[Mesh:noexp] OR Africa, Western[Mesh:noexp] OR Asia[Mesh:noexp] OR Asia, Central[Mesh:noexp] OR Asia, Southeastern[Mesh:noexp] OR Asia, Western[Mesh:noexp] OR Caribbean Region[Mesh:noexp] OR West Indies[Mesh:noexp] OR South America[Mesh:noexp] OR Latin America[Mesh:noexp] OR Central America[Mesh:noexp] OR Afghanistan[Mesh:noexp] OR Albania[Mesh:noexp] OR Algeria[Mesh:noexp] OR American Samoa[Mesh:noexp] OR Angola[Mesh:noexp] OR "Antigua and Barbuda"[Mesh:noexp] OR Argentina[Mesh:noexp] OR Armenia[Mesh:noexp] OR Azerbaijan[Mesh:noexp] OR Bahrain[Mesh:noexp] OR Bangladesh[Mesh:noexp] OR Barbados[Mesh:noexp] OR Benin[Mesh:noexp] OR Byelarus[Mesh:noexp] OR Belize[Mesh:noexp] OR Bhutan[Mesh:noexp] OR Bolivia[Mesh:noexp] OR Bosnia‐Herzegovina[Mesh:noexp] OR Botswana[Mesh:noexp] OR Brazil[Mesh:noexp] OR Bulgaria[Mesh:noexp] OR Burkina Faso[Mesh:noexp] OR Burundi[Mesh:noexp] OR Cambodia[Mesh:noexp] OR Cameroon[Mesh:noexp] OR Cape Verde[Mesh:noexp] OR Central African Republic[Mesh:noexp] OR Chad[Mesh:noexp] OR Chile[Mesh:noexp] OR China[Mesh:noexp] OR Colombia[Mesh:noexp] OR Comoros[Mesh:noexp] OR Congo[Mesh:noexp] OR Costa Rica[Mesh:noexp] OR Cote d'Ivoire[Mesh:noexp] OR Croatia[Mesh:noexp] OR Cuba[Mesh:noexp] OR Cyprus[Mesh:noexp] OR Czechoslovakia[Mesh:noexp] OR Czech Republic[Mesh:noexp] OR Slovakia[Mesh:noexp] OR Djibouti[Mesh:noexp] OR "Democratic Republic of the Congo"[Mesh:noexp] OR Dominica[Mesh:noexp] OR Dominican Republic[Mesh:noexp] OR East Timor[Mesh:noexp] OR Ecuador[Mesh:noexp] OR Egypt[Mesh:noexp] OR El Salvador[Mesh:noexp] OR Eritrea[Mesh:noexp] OR Estonia[Mesh:noexp] OR Ethiopia[Mesh:noexp] OR Fiji[Mesh:noexp] OR Gabon[Mesh:noexp] OR Gambia[Mesh:noexp] OR "Georgia (Republic)"[Mesh:noexp] OR Ghana[Mesh:noexp] OR Greece[Mesh:noexp] OR Grenada[Mesh:noexp] OR Guatemala[Mesh:noexp] OR Guinea[Mesh:noexp] OR Guinea‐Bissau[Mesh:noexp] OR Guam[Mesh:noexp] OR Guyana[Mesh:noexp] OR Haiti[Mesh:noexp] OR Honduras[Mesh:noexp] OR Hungary[Mesh:noexp] OR India[Mesh:noexp] OR Indonesia[Mesh:noexp] OR Iran[Mesh:noexp] OR Iraq[Mesh:noexp] OR Jamaica[Mesh:noexp] OR Jordan[Mesh:noexp] OR Kazakhstan[Mesh:noexp] OR Kenya[Mesh:noexp] OR Korea[Mesh:noexp] OR Kosovo[Mesh:noexp] OR Kyrgyzstan[Mesh:noexp] OR Laos[Mesh:noexp] OR Latvia[Mesh:noexp] OR Lebanon[Mesh:noexp] OR Lesotho[Mesh:noexp] OR Liberia[Mesh:noexp] OR Libya[Mesh:noexp] OR Lithuania[Mesh:noexp] OR Macedonia[Mesh:noexp] OR Madagascar[Mesh:noexp] OR Malaysia[Mesh:noexp] OR Malawi[Mesh:noexp] OR Mali[Mesh:noexp] OR Malta[Mesh:noexp] OR Mauritania[Mesh:noexp] OR Mauritius[Mesh:noexp] OR Mexico[Mesh:noexp] OR Micronesia[Mesh:noexp] OR Middle East[Mesh:noexp] OR Moldova[Mesh:noexp] OR Mongolia[Mesh:noexp] OR Montenegro[Mesh:noexp] OR Morocco[Mesh:noexp] OR Mozambique[Mesh:noexp] OR Myanmar[Mesh:noexp] OR Namibia[Mesh:noexp] OR Nepal[Mesh:noexp] OR Netherlands Antilles[Mesh:noexp] OR New Caledonia[Mesh:noexp] OR Nicaragua[Mesh:noexp] OR Niger[Mesh:noexp] OR Nigeria[Mesh:noexp] OR Oman[Mesh:noexp] OR Pakistan[Mesh:noexp] OR Palau[Mesh:noexp] OR Panama[Mesh:noexp] OR Papua New Guinea[Mesh:noexp] OR Paraguay[Mesh:noexp] OR Peru[Mesh:noexp] OR Philippines[Mesh:noexp] OR Poland[Mesh:noexp] OR Portugal[Mesh:noexp] OR Puerto Rico[Mesh:noexp] OR Romania[Mesh:noexp] OR Russia[Mesh:noexp] OR "Russia (Pre‐1917)"[Mesh:noexp] OR Rwanda[Mesh:noexp] OR "Saint Kitts and Nevis"[Mesh:noexp] OR Saint Lucia[Mesh:noexp] OR "Saint Vincent and the Grenadines"[Mesh:noexp] OR Samoa[Mesh:noexp] OR Saudi Arabia[Mesh:noexp] OR Senegal[Mesh:noexp] OR Serbia[Mesh:noexp] OR Montenegro[Mesh:noexp] OR Seychelles[Mesh:noexp] OR Sierra Leone[Mesh:noexp] OR Slovenia[Mesh:noexp] OR Sri Lanka[Mesh:noexp] OR Somalia[Mesh:noexp] OR South Africa[Mesh:noexp] OR Sudan[Mesh:noexp] OR Suriname[Mesh:noexp] OR Swaziland[Mesh:noexp] OR Syria[Mesh:noexp] OR Tajikistan[Mesh:noexp] OR Tanzania[Mesh:noexp] OR Thailand[Mesh:noexp] OR Togo[Mesh:noexp] OR Tonga[Mesh:noexp] OR "Trinidad and Tobago"[Mesh:noexp] OR Tunisia[Mesh:noexp] OR Turkey[Mesh:noexp] OR Turkmenistan[Mesh:noexp] OR Uganda[Mesh:noexp] OR Ukraine[Mesh:noexp] OR Uruguay[Mesh:noexp] OR USSR[Mesh:noexp] OR Uzbekistan[Mesh:noexp] OR Vanuatu[Mesh:noexp] OR Venezuela[Mesh:noexp] OR Vietnam[Mesh:noexp] OR Yemen[Mesh:noexp] OR Yugoslavia[Mesh:noexp] OR Zambia[Mesh:noexp] OR Zimbabwe[Mesh:noexp])
#17	Search #13 AND #16
#16	Search #14 OR #15
#15	Search "Mycotoxins"[Mesh:NoExp] OR "Aflatoxins"[Mesh]
#14	Search aflatoxin* OR mycotoxin*
#13	Search (#1 OR #2 OR #3 OR #4 OR #5 OR #6 OR #7 OR #8 OR #9 OR #10 OR #11 OR #12 OR #13)
#12	Search "Child"[Mesh]
#11	Search "Infant"[Mesh]
#10	Search "Adolescent"[Mesh]
#9	Search "Lactation"[Mesh]
#8	Search "Breast Feeding"[Mesh]
#7	Search "Pregnancy"[Mesh]
#6	Search "Pregnant Women"[Mesh]
#5	Search (child* OR toddler* OR adolescent*)[tiab]
#4	Search (newborn* OR neonate* OR infant* OR baby OR babies OR neonatal OR prenatal)[tiab]
#3	Search (lactating OR lactation)[tiab]
#2	Search breastfeed* OR "breast feed" OR "breast fed" OR "breast feeding" OR "breast milk" OR breastmilk
#1	Search (pregnan* OR mother* OR maternal)[tiab]

Contributions of authors

All review authors contributed to the development of the protocol. Two authors (AS, NR) developed the search strategy. Two authors (MV, CNE), with the assistance of other authors (AS, CN), will screen potential studies, extract data, and assess risk of bias. Three authors (MV, CNE, TE), with the assistance of other authors (NR, AS, CN), will plan and conduct the data analysis. All authors will be involved in drafting the final review for submission.

Sources of support

Internal sources

• Stellenbosch University, South Africa.

External sources

• Department for International Development, UK.

  Project number 300324‐104

Declarations of interest

Marianne E. Visser ‐ none known

Chibundu N. Ezekiel ‐ none known

Anel Schoonees ‐ none known

Tonya M. Esterhuizen ‐ none known

Nicola Randall ‐ none known

Celeste E. Naude ‐ none known

New