Food fortification with multiple micronutrients: impact on health outcomes in general population

Abstract

Background

Vitamins and minerals are essential for growth and maintenance of a healthy body, and have a role in the functioning of almost every organ. Multiple interventions have been designed to improve micronutrient deficiency, and food fortification is one of them.

Objectives

To assess the impact of food fortification with multiple micronutrients on health outcomes in the general population, including men, women and children.

Search methods

We searched electronic databases up to 29 August 2018, including the Cochrane Central Register of Controlled Trial (CENTRAL), the Cochrane Effective Practice and Organisation of Care (EPOC) Group Specialised Register and Cochrane Public Health Specialised Register; MEDLINE; Embase, and 20 other databases, including clinical trial registries. There were no date or language restrictions. We checked reference lists of included studies and relevant systematic reviews for additional papers to be considered for inclusion.

Selection criteria

We included randomised controlled trials (RCTs), cluster‐RCTs, quasi‐randomised trials, controlled before‐after (CBA) studies and interrupted time series (ITS) studies that assessed the impact of food fortification with multiple micronutrients (MMNs). Primary outcomes included anaemia, micronutrient deficiencies, anthropometric measures, morbidity, all‐cause mortality and cause‐specific mortality. Secondary outcomes included potential adverse outcomes, serum concentration of specific micronutrients, serum haemoglobin levels and neurodevelopmental and cognitive outcomes. We included food fortification studies from both high‐income and low‐ and middle‐income countries (LMICs).

Data collection and analysis

Two review authors independently screened, extracted and quality‐appraised the data from eligible studies. We carried out statistical analysis using Review Manager 5 software. We used random‐effects meta‐analysis for combining data, as the characteristics of study participants and interventions differed significantly. We set out the main findings of the review in 'Summary of findings' tables, using the GRADE approach.

Main results

We identified 127 studies as relevant through title/abstract screening, and included 43 studies (48 papers) with 19,585 participants (17,878 children) in the review. All the included studies except three compared MMN fortification with placebo/no intervention. Two studies compared MMN fortification versus iodised salt and one study compared MMN fortification versus calcium fortification alone. Thirty‐six studies targeted children; 20 studies were conducted in LMICs. Food vehicles used included staple foods, such as rice and flour; dairy products, including milk and yogurt; non‐dairy beverages; biscuits; spreads; and salt. Fourteen of the studies were fully commercially funded, 13 had partial‐commercial funding, 14 had non‐commercial funding and two studies did not specify the source of funding. We rated all the evidence as of low to very low quality due to study limitations, imprecision, high heterogeneity and small sample size.

When compared with placebo/no intervention, MMN fortification may reduce anaemia by 32% (risk ratio (RR) 0.68, 95% confidence interval (CI) 0.56 to 0.84; 11 studies, 3746 participants; low‐quality evidence), iron deficiency anaemia by 72% (RR 0.28, 95% CI 0.19 to 0.39; 6 studies, 2189 participants; low‐quality evidence), iron deficiency by 56% (RR 0.44, 95% CI 0.32 to 0.60; 11 studies, 3289 participants; low‐quality evidence); vitamin A deficiency by 58% (RR 0.42, 95% CI 0.28 to 0.62; 6 studies, 1482 participants; low‐quality evidence), vitamin B2 deficiency by 64% (RR 0.36, 95% CI 0.19 to 0.68; 1 study, 296 participants; low‐quality evidence), vitamin B6 deficiency by 91% (RR 0.09, 95% CI 0.02 to 0.38; 2 studies, 301 participants; low‐quality evidence), vitamin B12 deficiency by 58% (RR 0.42, 95% CI 0.25 to 0.71; 3 studies, 728 participants; low‐quality evidence), weight‐for‐age z‐scores (WAZ) (mean difference (MD) 0.1, 95% CI 0.02 to 0.17; 8 studies, 2889 participants; low‐quality evidence) and weight‐for‐height/length z‐score (WHZ/WLZ) (MD 0.1, 95% CI 0.02 to 0.18; 6 studies, 1758 participants; low‐quality evidence). We are uncertain about the effect of MMN fortification on zinc deficiency (RR 0.84, 95% CI 0.65 to 1.08; 5 studies, 1490 participants; low‐quality evidence) and height/length‐for‐age z‐score (HAZ/LAZ) (MD 0.09, 95% CI 0.01 to 0.18; 8 studies, 2889 participants; low‐quality evidence). Most of the studies in this comparison were conducted in children.

Subgroup analyses of funding sources (commercial versus non‐commercial) and duration of intervention did not demonstrate any difference in effects, although this was a relatively small number of studies and the possible association between commercial funding and increased effect estimates has been demonstrated in the wider health literature. We could not conduct subgroup analysis by food vehicle and funding; since there were too few studies in each subgroup to draw any meaningful conclusions.

When we compared MMNs versus iodised salt, we are uncertain about the effect of MMN fortification on anaemia (R 0.86, 95% CI 0.37 to 2.01; 1 study, 88 participants; very low‐quality evidence), iron deficiency anaemia (RR 0.40, 95% CI 0.09 to 1.83; 2 studies, 245 participants; very low‐quality evidence), iron deficiency (RR 0.98, 95% CI 0.82 to 1.17; 1 study, 88 participants; very low‐quality evidence) and vitamin A deficiency (RR 0.19, 95% CI 0.07 to 0.55; 2 studies, 363 participants; very low‐quality evidence). Both of the studies were conducted in children.

Only one study conducted in children compared MMN fortification versus calcium fortification. None of the primary outcomes were reported in the study.

None of the included studies reported on morbidity, adverse events, all‐cause or cause‐specific mortality.

Authors' conclusions

The evidence from this review suggests that MMN fortification when compared to placebo/no intervention may reduce anaemia, iron deficiency anaemia and micronutrient deficiencies (iron, vitamin A, vitamin B2 and vitamin B6). We are uncertain of the effect of MMN fortification on anthropometric measures (HAZ/LAZ, WAZ and WHZ/WLZ). There are no data to suggest possible adverse effects of MMN fortification, and we could not draw reliable conclusions from various subgroup analyses due to a limited number of studies in each subgroup. We remain cautious about the level of commercial funding in this field, and the possibility that this may be associated with higher effect estimates, although subgroup analysis in this review did not demonstrate any impact of commercial funding. These findings are subject to study limitations, imprecision, high heterogeneity and small sample sizes, and we rated most of the evidence low to very low quality. and hence no concrete conclusions could be drawn from the findings of this review.

Plain language summary

Impact of food fortification with multiple micronutrients on health

Review question
 Does multiple micronutrient fortification improve health?

Background
 Vitamins and minerals are important for growth and body functioning. Micronutrient deficiencies are common in many populations, and food fortification is one of the interventions to reduce the burden of micronutrient deficiencies and improve health in the general population. Food fortification involves adding micronutrients to processed foods. There have been studies with various single micronutrient fortification, dual micronutrient fortification and multiple micronutrient fortification, including zinc, iron, selenium, vitamin A, vitamin B complexes, vitamin C and vitamin E. We reviewed the evidence about the impact of food fortification with multiple micronutrients (MMNs) on health in the general population.

Study characteristics
 We included 43 studies (48 papers) in 19,585 participants (17,878 children) in this review. The evidence is current to August 2018. Most of the included studies assessed the impact of food fortification with MMN compared to placebo or to no intervention; two studies compared food fortification with MMN to iodised salt and one study compared food fortification with MMN to food fortification with calcium alone. Most of the studies (36 out of 43) targeted children. Twenty studies were conducted in developing countries. Food used for fortification included staple foods, such as rice and flour; dairy products, including milk and yogurt; non‐dairy beverages; biscuits; spreads; and salt. A high proportion of studies were funded by commercial sources (e.g. manufacturers of micronutrients), which can be associated with finding more beneficial effects than independently‐funded studies.

Key results
 Food fortification with MMN may reduce anaemia by 32%, iron deficiency anaemia by 72%, micronutrient deficiencies (including iron deficiency by 56%, vitamin A deficiency by 58%, vitamin B2 deficiency by 64%, vitamin B6 deficiency by 91% and vitamin B12 deficiency by 58%). MMN fortification may also improve child growth measured as weight for age and weight for height/length. We are uncertain of the effect of MMN fortification on zinc deficiency and child growth measured as height/length for age. The included studies did not report on any side effects associated with MMN fortification, including deaths and diseases. We are uncertain of the effect of food fortification with MMN compared to iodised salt for iron deficiency anaemia and vitamin A deficiency.

Quality of the evidence
 The quality of the evidence was low to very low, due to limitations in the study methods that could introduce a risk of bias, high heterogeneity (variation in the results from study to study), and small sample sizes.

Although the review suggests some positive effects of MMN fortification compared to no intervention, a number of other factors should also be considered. Firstly, there is no information on possible side effects of the MMN fortification. Secondly, we could not perform various subgroup analyses to identify whether MMN fortification is more effective in different population groups, food vehicles, dosage, duration of intervention and geographical region, due to limited number of studies in each subgroup. We performed a subgroup analysis to compare commercial and non‐commercially‐funded studies and did not find a significant difference between their results, although we remain cautious about these findings. Our results are uncertain, due to the low quality of the evidence.

Background

Description of the condition

Vitamins and minerals are essential for growth and maintenance of a healthy body, and their deficiencies can lead to various diseases. Micronutrient deficiencies account for a substantial global burden of disease, with iron and vitamin A deficiency being among the 15 leading causes of global morbidity and mortality (WHO 2002). Globally about 1.62 billion people are anaemic, with the highest prevalence among preschool children followed by pregnant women (Benoist 2008). Iron, iodine, folate, vitamin A, and zinc deficiencies are the most widespread micronutrient deficiencies, and all of these are common contributors to poor growth, intellectual impairments, perinatal complications, and increased risk of morbidity and mortality (Bailey 2015). In 2014, iron deficiency anaemia was one of the three most common causes of disability‐adjusted life years (DALYs) lost among adolescents along with other micronutrient deficiencies accounting for over 2500 DALYs per 100,000 adolescents (Akseer 2017; WHO 2014). About 190 million preschool children and 19.1 million pregnant women are vitamin A‐deficient (WHO 2009). Iodine and zinc deficiencies are estimated to affect 29% and 17% respectively of the world’s population (Black 2013), with approximately 82% of pregnant women worldwide having inadequate zinc intake. Prevalence of suboptimal body stores of vitamins B6 and B12 have also been reported (McLean 2008). Populations from developing countries are believed to be most affected, with multiple micronutrient (MMN) deficiencies frequently co‐existing among more than two billion people affected (Bailey 2015; Best 2011; Dijkhuizen 2001; Ramakrishnan 2002; Stanger 2009).

Micronutrient deficiencies can result in impairments in mental and physical growth and development, and immune competence. They may also adversely affect reproductive outcomes (Gibson 2002; Haimi 2014; Viteri 2002). MMN deficiencies are associated with increased incidence and severity of infectious illness and mortality from diarrhoea, measles, malaria and pneumonia (Ibrahim 2017). In preschool children, zinc deficiency has been associated with an increased risk of diarrhoea, malaria and pneumonia, while vitamin A deficiency is associated with increased risk of mortality due to diarrhoea (Black 2008; Black 2013). The prevalence of iron deficiency anaemia during pregnancy is a risk factor for maternal mortality (Allen 2008).

Several strategies have been implemented to combat micronutrient deficiencies, including exclusive breastfeeding during the first six months of life, nutrition education, food rationing, control of parasitic infections and nutritional supplementation (Bhutta 2008; Bhutta 2013). Food fortification is one of these strategies, in which a variety of micronutrient combinations can be added to foods, including zinc, iron, selenium, vitamin A, vitamin B complexes, vitamin C, vitamin D and vitamin E (Allen 2006).

Description of the intervention

Food fortification adopts an integrated approach and provides support to reduce micronutrients malnutrition when other existing food supplies fail to do so (Allen 2006). Food fortification is the process by which micronutrients are added to processed foods and has been applied at various levels and directed to different age groups (De Lourdes Samaniego‐Vaesken 2012; Allen 2006). A range of micronutrient combinations has been used to fortify foods. There have been studies with various single micronutrient fortifications, dual micronutrient fortification and up to 20 micronutrient fortifications, including zinc, iron, selenium, vitamin A, vitamin B complexes, vitamin C and vitamin E. The advantage of fortification of food items consumed by the general population is that no or minimal behaviour change is required on the part of the population (Serdula 2010a; Serdula 2010b). This provides an advantage in terms of coverage and efficiency. Food fortification could potentially also be cost‐effective, as it can be targeted at different age groups at a time (Hurrell 1997; Lotfi 1996; Allen 2006). In contrast, supplementation depends upon a viable delivery mechanism and the availability and access to the intended individuals (Harrison 2010). For fortification, however, issues related to safe and effective levels of micronutrients being used, the relevance of the micronutrients and appropriate food vehicle need to be considered (Allen 2006; Allen 2008).

One of the issues concerning food fortification is that many of the these studies are funded by the manufacturers of fortified food products. The source of funding could be one of the biases in studies evaluating nutrition interventions if the researchers have a vested interest in the outcomes of the research. Industry‐funded research might skew the evidence towards solutions that favour industry interests by focusing on food components that can be manipulated and marketed by food companies (Fabbri 2018). These concerns should therefore be explicitly evaluated when considering the evidence on food fortification.

How the intervention might work

Fortification could be mass fortification (that is, adding micronutrients to foods that are commonly consumed, such as flour, salt, sugar and cooking oil), or point‐of‐use fortification, which involves adding single‐dose packets of vitamins and minerals in powder form that can be sprinkled onto any ready‐to‐eat food consumed at home, school, nurseries, refugee camps or any other place where possible (WHO 2014; Zlotkin 2005). For this review, we focus only on mass food fortification. Food fortification can combat micronutrient deficiencies at the population level at reasonable cost, making it a very efficient public health intervention.

Many trials have shown the positive impact of food fortification. Sazawal 2007 showed that milk fortified with multiple micronutrients reduced the odds for days with severe illnesses by 15% (95% confidence interval (CI) 5% to 24%), the incidence of diarrhoea by 18% (95% CI 7% to 27%) and the incidence of acute lower respiratory illness by 26% (95% CI 3% to 43%) in children. Another study by Osei 2010 reported improvement in vitamin A, vitamin B12, folate and total body iron status after fortification of school meals in a village in India. A review by Aaron 2015 reported a reduction in risk of anaemia (relative risk (RR) 0.58, 95% CI 0.29 to 0.88), iron deficiency (RR 0.34, 95%CI 0.21 to 0.55), and iron deficiency anaemia (RR 0.17, 95% CI 0.06 to 0.53) after use of fortified beverages in school‐aged children.

Concerns exist that food fortification may result in unacceptably high micronutrient levels among those consuming higher amounts of fortified foods. However, exceeding the upper intake level has not been shown to increase the risk of adverse effects, according to a review of data from national surveys conducted in European countries (Hennessy 2013).

Why it is important to do this review

Several questions remain about the use of fortified foods. Many reviews to evaluate the use of fortified foods have suggested benefit, but usually focus on a particular food vehicle or a subset of the general population. A review by Eichler 2012 focused on MMN‐fortified dairy and cereal products delivered to pre‐school children in developing countries, and showed it to be effective in reducing anaemia. Das 2013 focused on the use of MMN‐fortified foods for women and children only. There is no evidence to suggest which combinations work best and whether certain combinations may work better with particular food vehicles. It is also unclear which population subgroups may derive the most benefit from these interventions and under what conditions.

This review serves as a comprehensive assessment of the effect of MMN fortification on the population as a whole, without being limited to certain age groups, regions or a particular food vehicle. Reviews of home or point‐of‐use fortification of food through micronutrient powders (De‐Regil 2011; Salam 2013), and ready‐to‐use therapeutic food (RUTF) (Schoonees 2019) already exist, and we have not focused on these areas.

Objectives

To assess the impact of food fortification with MMNs on health outcomes in the general population, including men, women and children.

Methods

Criteria for considering studies for this review

Types of studies

We have included:

• Randomised controlled trials (RCTs);

• Quasi‐randomised trials;

• Cluster‐RCTs (c‐RCTs);

• Controlled before‐after (CBA) studies;

• Interrupted time series (ITS) studies.

We intended to include quasi‐experimental study designs, CBA and ITS, along with RCTs, since we planned to assess the effectiveness of large‐scale programme evaluations that might not have been conducted in a randomised fashion. We applied no language or publication status restrictions.

Types of participants

We included studies that assess the effects of food fortification in the general population, including men, women and children. We also included studies that targeted fortification in specific populations (e.g. older people, pregnant women, women of reproductive age, and children at school through institutions such as schools or care facilities). We included studies from all countries, regardless of their level of income and development.

We excluded studies conducted among some special population groups, including critically‐ill people, anaemic people or people diagnosed with any specific diseases.

Types of interventions

Intervention: MMN fortification (three or more micronutrients) by any food vehicle, compared with a single micronutrient or no fortification.

We have not included studies evaluating point‐of‐use or home fortification of foods, therapeutic blended food or food supplementation.

Types of outcome measures

Primary outcomes

• Anaemia (defined as haemoglobin (Hb) concentration < 11 g/dL)

• Iron‐deficiency anaemia (defined as Hb concentration < 11 g/dLwith serum ferritin < 15 µg/l)

• Deficiency of specific micronutrients (iron, zinc, vitamin A, B vitamins) (as defined by the World Health Organization (WHO) micronutrient deficiency cut‐offs)

• Anthropometric outcomes (e.g. incidence of stunting (defined as below minus two standard deviations from median height for age of reference population), wasting (defined as below minus two standard deviations from median weight for height of reference population) and underweight (defined as below minus two standard deviations from median weight for age of reference population)

• Morbidity (e.g. infectious diseases such as pneumonia, sepsis and diarrhoea)

• All‐cause mortality (defined as death due to any cause)

• Cause‐specific mortality (as defined by study authors) due to pneumonia, diarrhoea or malaria

Secondary outcomes

• Potential adverse outcomes (as defined by study authors)

• Serum haemoglobin levels (measured as g/dL)

• Serum concentration of specific micronutrients (folate, ferritin, vitamin A, B vitamins, zinc)

• Neuro‐developmental and cognitive outcomes

Search methods for identification of studies

Electronic searches

We searched the following electronic databases for primary studies, without date or language restrictions. We conducted the final search on 29 August 2018.

• Cochrane Central Register of Controlled Trials (CENTRAL; 2018), in the Cochrane Library, including the Cochrane Effective Practice and Organisation of Care (EPOC) Group Specialised Register and Cochrane Public Health Specialised Register (searched 29 August 2018);

• MEDLINE and MEDLINE(R) In‐Process; and Other Non‐Indexed Citations Ovid (searched 29 August 2018);

• PubMed for the most recent six months to identify records that are 'Epub ahead of print'; (searched 29 August 2018);

• Embase Ovid (1974 to August 2018) (searched 29 August 2018);

• Cumulative Index to Nursing and Allied Health Literature (CINAHL) Plus EBSCOhost; 1937 to August 2018 (searched 29 August 2018);

• PsycINFO (www.apa.org/pubs/databases/index; searched 30 August 2018);

• Education Resources Information Center (ERIC) (eric.ed.gov/ searched 22 August 2018);

• Latin American and Caribbean Health Sciences Literature (LILACS); (lilacs.bvsalud.org/en; searched 29 August 2018);

• AGRIS (aims.fao.org/search/node searched 30 August 2018);

• Science Citation Index and Social Sciences Citation Index Web of Science (SCI and SSCI; 1970 to August 2018);

• Food Science and Technology Abstracts (www.ebsco.com/products/research‐databases/fsta‐food‐science‐and‐technology‐abstracts searched 31 August 2018);

• AgriCOLA (agricola.nal.usda.gov/; searched 28 August 2018);

• Global Index Medicus ‐ AFRO (indexmedicus.afro.who.int/ searched 31 August 2018);

• EMRO (www.emro.who.int/index.html searched 31 August 2018);

• Pan American Health library (PAHO)/WHO Institutional Repository for Information Sharing (iris.paho.org/xmlui; searched 30 August 2018);

• WHOLIS Global Index Medicus (WHO Library Database; search.bvsalud.org/ghl/?lang=en&submit=Search&where=REGIONAL; searched 29 August 2018);

• WPRO (https://www.who.int/library/databases/wpro/en/ searched 29 August 2018);

• Global Index Medicus (Index Medicus for the South‐East Asian Region; IMSEAR) (search.bvsalud.org/ghl/?lang=en&submit=Search&where=REGIONAL; searched 31 August 2018);

• 3ie Database of Impact studies (www.3ieimpact.org/evidence‐hub/impact‐evaluation‐repository; searched 29 August 2018);

• EPPI centre databases ‐ DoPHER (eppi.ioe.ac.uk/webdatabases4/Intro.aspx?ID=9 ) and TROPHI (eppi.ioe.ac.uk/webdatabases4/Intro.aspx?ID=12) searched 29 August 2018;

• OpenGrey (www.opengrey.eu/ searched 30 August 2018);

• Index to Conference Proceedings (mjl.clarivate.com/scope/scope_cpci‐s/ searched 30 August 2018);

• ClinicalTrials.gov (clinicaltrials.gov/), and WHO International Clinical Trials Registry Platform (ICTRP; www.who.int/ictrp/en/) (searched 30 August 2018).

We adapted the MEDLINE search strategy (Appendix 1) for use in the other databases using the appropriate controlled vocabulary as applicable (Appendix 2; Appendix 3; Appendix 4; Appendix 5; Appendix 6). We handsearched the journals and the proceedings of major relevant conferences relating to food and nutrition, including Micronutrient Forum, Nutrition and Food Sciences and Hidden Hunger. We also handsearched the journals in which the included studies appeared most frequently. We searched the top five journals (according to the number of included studies provided) for the previous 12 months.

Searching other resources

We checked the reference lists of included studies and relevant systematic reviews for additional papers to consider for inclusion.

Data collection and analysis

Selection of studies

Two review authors (SBM and KM) independently assessed all the studies identified potentially for inclusion. We resolved any disagreement through discussion or, if required, consulted a third review author (RAS or JKD).

Data extraction and management

We designed a form for the extraction of data. Two of the review authors (from SBM, AM, ZL, KM and RK) extracted the data from each eligible study using the agreed form. We resolved discrepancies through discussion or, when required, consulted a third review author (JKD or RAS). We entered data into Review Manager 5 (RevMan 2014), and checked them for accuracy by double data entry (SBM, ZL and RAS), having one review author entering data into a separate file and comparing the results. For the studies that reported outcomes at multiple time points, we extracted data for each time point and reported the outcomes at the last reported time period.

We used the PROGRESS checklist (place, race, occupation, gender, religion, education, socioeconomic status, social status) (O’Neill 2014; Welch 2016) to record whether outcome data were reported by socio‐demographic characteristics known to be important from an equity perspective. We also recorded whether studies included specific strategies to address diversity or disadvantage. Where available, we extracted data on costs and process/implementation and source of funding of the primary studies. These details are presented in the Characteristics of included studies tables.

Assessment of risk of bias in included studies

Two review authors (from SBM, RAS, JKD and RK) independently assessed risks of bias (RoB) for each study using the criteria outlined in the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2017), with the exception that we conducted 'Risk of bias' assessments at the level of each study, and not for individual outcomes. We used the Cochrane Effective Practice and Organisation of Care (EPOC 2017) nine‐point criteria for non‐RCTs and CBA studies to determine the quality of all eligible studies. We did not find any eligible studies using an ITS study design. We resolved any disagreement by discussion or by involving a third assessor. We report the risks of bias for each study in the Characteristics of included studies table. We did not exclude studies on the grounds of their quality, but clearly report methodological quality when presenting the results of the studies.

Random sequence generation (checking for possible selection bias)

We have described for each included study the method used to generate the allocation sequence in sufficient detail to allow an assessment of whether it should produce comparable groups. We have assessed the methods as:
 • low risk of bias (any truly random process, e.g. random‐number table, computer random‐number generator);
 • high risk of bias (any non‐random process, e.g. odd or even date of birth, hospital or clinic record number);
 • unclear risk of bias.

Allocation concealment (checking for possible selection bias)

We have described for each included study the method used to conceal allocation to interventions prior to assignment and assessed whether intervention allocation could have been foreseen in advance of or during recruitment, or changed after assignment. We have assessed the methods as:
 • low risk of bias (e.g. telephone or central randomisation, consecutively‐numbered sealed opaque envelopes);
 • high risk of bias (open random allocation, unsealed or non‐opaque envelopes, alternation, date of birth);
 • unclear risk of bias.

Blinding of participants and personnel (checking for possible performance bias)

We have described for each included study the methods used, if any, to blind study participants and personnel from knowledge of which intervention a participant received. We have considered that studies are at low risk of bias if they were blinded, or if we judge that the lack of blinding would be unlikely to affect results. We have assessed blinding separately for different outcomes or classes of outcomes. We have assessed the methods as:
 • low, high or unclear risk of bias for participants;
 • low, high or unclear risk of bias for personnel.

Blinding of outcome assessment (checking for possible detection bias)

We have described for each included study the methods used, if any, to blind outcome assessors from knowledge of which intervention a participant received. We have assessed blinding separately for different outcomes or classes of outcomes. We have assessed methods used to blind outcome assessment as:
 • low, high or unclear risk of bias.

Incomplete outcome data (checking for possible attrition bias due to the amount, nature and handling of incomplete outcome data)

We have described for each included study, and for each outcome or class of outcomes, the completeness of data including attrition and exclusions from the analysis. We have stated whether attrition and exclusions were reported and the numbers included in the analysis at each stage (compared with the total randomised participants), reasons for attrition or exclusion where reported, and whether missing data were balanced across groups or were related to outcomes. We considered studies with more than 20% loss to follow‐up, or with an imbalanced loss to follow‐up in different groups, to have insufficient completeness of outcome data. We also looked at the amount and distribution across intervention groups and the reasons for outcomes being missing. Where sufficient information was reported, or was supplied by the trial authors, we have re‐included missing data in the analyses which we undertook. We assessed methods as:
 • low risk of bias (e.g. no or minimal missing outcome data, missing outcome data balanced across groups);
 • high risk of bias (e.g. numbers or reasons for missing data imbalance across groups, ’as treated’ analysis done with substantial departure of intervention received from that assigned at randomisation);
 • unclear risk of bias.

Selective reporting (checking for reporting bias)

We have described for each included study how we investigated the possibility of selective outcome reporting bias and what we found. We have assessed the methods as:
 • low risk of bias (where it is clear that all of the study’s prespecified outcomes and all expected outcomes of interest to the review have been reported);
 • high risk of bias (where not all the study’s prespecified outcomes have been reported, one or more reported primary outcomes were not prespecified, outcomes of interest are reported incompletely and so cannot be used, study fails to include results of a key outcome that would have been expected to have been reported);
 • unclear risk of bias.

Other bias

We have described for each included study any important concerns we have about other possible sources of bias. We have assessed whether each study was free of other problems that could put it at risk of bias:
 • low risk of other bias;
 • high risk of other bias;
 • unclear whether there is a risk of other bias.

Additional criteria for cluster‐RCTs

We assessed and report five additional criteria for all the included c‐RCTs in the section of 'Other bias' in the RoB table. These are recruitment bias; baseline imbalance; loss of clusters; incorrect analysis; and comparability with individually randomised trials.

Recruitment bias

We have assessed whether the individuals were recruited to the trial after the clusters have been randomised. We have assessed the methods as:

• low, high or unclear risk of bias.

Baseline imbalance

We have assessed the reporting of the baseline comparability of clusters, or statistical adjustment for baseline characteristics. We have assessed the methods as:

• low, high or unclear risk of bias.

Loss of clusters

we have assessed whether there was any loss of complete clusters or omission of complete clusters from the analysis. We have assessed the methods as:

• low, high or unclear risk of bias.

Incorrect analysis

We have assessed whether appropriate analysis has been conducted for adjusting the clustering. We have assessed the methods as:

• low, high or unclear risk of bias.

Comparability with individually randomised trial

We assessed the possible differences between the intervention effects in individually‐randomised and cluster‐randomised trails. We have assessed the methods as:

• low, high or unclear risk of bias.

Overall risk of bias

We have made explicit judgments about whether studies are at high risk of bias, according to the criteria given in the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2017). We assessed the likely magnitude and direction of the bias and whether we considered it likely to have had an impact on the findings. We judged the studies to be at overall high risk of bias if they were at high or unclear risk for any of the four criteria of allocation concealment, blinding of participants and personnel, blinding of outcome assessment or incomplete outcome data. We explored the impact of the level of bias by conducting sensitivity analyses for those studies with high or unclear risk of bias in any of the above four domains i.e. according to the method and adequacy of allocation concealment; blinding status of the participants/personnel and outcome assessor; or percentage lost to follow‐up or attrition of 20% or more, or with an imbalanced loss to follow‐up in different groups.

Measures of treatment effect

Dichotomous data

For dichotomous data, we present results as a summary risk ratio (RR) with a 95% confidence interval (CI).

Continuous data

For continuous data, we have used the mean difference (MD) if outcomes were measured in the same way between trials. We would have used the standardised mean difference (SMD) to combine trials that measured the same outcome but with different methods. Where the studies reported change in continuous outcomes and did not report endline values, we combined these data using the MD.

Unit of analysis issues

Cluster‐randomised trials

We have included cluster‐randomised trials in the analyses along with individually‐randomised trials. We used cluster‐adjusted estimates from c‐RCTs where available. If the studies had not adjusted for clustering, we attempted to adjust their standard errors using the methods described in the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2019), using an estimate of the intra‐cluster correlation coefficient (ICC) derived from the trial. If the trial did not report the cluster‐adjusted estimated or the ICC, we imputed an ICC from a similar study included in the review, adjusting if the nature or size of the clusters was different (e.g. households compared to classrooms). We assessed any imputed ICCs using sensitivity analysis. We considered it reasonable to combine the results from both if there was little heterogeneity between the study designs and if we considered the interaction between the effect of intervention and the choice of randomisation unit to be unlikely. We have also acknowledged heterogeneity in the randomisation unit and performed a subgroup analysis to investigate the effects of the randomisation unit.

Studies with more than two treatment groups

When we identified studies with more than two intervention groups (multi‐arm studies), we combined groups to create a single pair‐wise comparison or used the methods set out in theCochrane Handbook for Systematic Reviews of Interventions (Higgins 2019) to avoid double‐counting study participants. For the subgroup analyses, when the control group was shared by two or more study arms, we divided the control group (events and total population) over the number of relevant subgroups to avoid double‐counting the participants.

Dealing with missing data

We have described missing data, including dropouts. Differential dropout rates can lead to biased estimates of the effect size, and bias may arise if the reasons for dropping out differ across groups. We have reported the reasons for dropout. If data were missing for some cases, or if the reasons for dropping out were not reported, we have contacted the authors. For included studies, we have noted levels of attrition. We have explored the impact of including studies with high levels of missing data in the overall assessment of treatment effect by using sensitivity analysis. For all outcomes, we have carried out analyses, as far as possible, on an intention‐to‐treat basis, i.e. we attempted to include all participants randomised to each group in the analyses, and all participants were analysed in the group to which they were allocated, regardless of whether or not they received the allocated intervention. The denominator for each outcome in each trial was the number randomised minus any participants whose outcomes were known to be missing.

Assessment of heterogeneity

We assessed the included studies for clinical, methodological, and statistical heterogeneity. We assessed clinical heterogeneity by comparing the distribution of important factors, such as the study participants, study setting, dose and duration of the intervention and co‐interventions. We evaluated methodological heterogeneity on the basis of factors such as the method of sequence generation, allocation concealment, blinding of outcome assessment, and losses to follow‐up. We have assessed statistical heterogeneity in each meta‐analysis using the T², I² and Chi² statistics. We regard heterogeneity as substantial if I² was greater than 30% and either T² was greater than zero, or there was a low P value (< 0.10) in the Chi² test for heterogeneity.

Assessment of reporting biases

If there were 10 or more studies in the meta‐analysis we investigated reporting biases (such as publication bias) using funnel plots. We have assessed funnel plot asymmetry visually, and if asymmetry was visually apparent in any of the plots, we attempted to investigate it through sensitivity analysis (where possible) and compared the random‐effects with the fixed‐effect model. We considered non‐reporting bias as one of the possible explanations of the funnel plot asymmetry.

Data synthesis

We carried out the statistical analysis using the Review Manager 5 software. We categorised the studies into the following three comparisons:

• MMN fortification versus placebo or no intervention;

• MMN fortification versus iodised salt;

• MMN fortification versus calcium only fortification.

We used random‐effects meta‐analysis for combining data, as the characteristics of study participants and interventions differed significantly. We present the results as the average treatment effect with a 95% confidence interval, and the estimates of T² and I². We used the Mantel‐Haenszel method for dichotomous data, the inverse variance for continuous data and generic Inverse variance for synthesis including data originating from c‐RCTs. We synthesised the findings from the RCTs and c‐RCTs together but did not pool non‐randomised studies, as we judged them to be too few in number. We have reported the findings from the non‐randomised studies separately.

GRADE and 'Summary of findings' tables

We have set out the findings of the primary outcomes in 'Summary of findings' tables, prepared using the GRADE approach (Guyatt 2008) and using GRADE profiler software (GRADEpro). We have listed the outcomes for each comparison with estimates of relative effects along with the number of participants and studies contributing data for those outcomes. For each individual outcome, we assessed the quality of the evidence using the GRADE approach, which involves consideration of within‐study risk of bias (methodological quality), directness of evidence, heterogeneity, precision of effect estimates and risk of publication bias. We have rated the quality of the body of evidence for each key outcome as high, moderate, low or very low. We present these findings according to the standardised language adapted from Glenton 2010.

Subgroup analysis and investigation of heterogeneity

We have conducted subgroup analyses on the basis of the following, where data permitted (where there were at least three studies in each subgroup):

• population (children, women of reproductive age, adults)

• baseline micronutrient status (malnourished, normal)

• various combination of MMNs (e.g. different numbers and types of micronutrients used)

• low‐to‐middle income countries versus high‐income countries

• duration of intervention (zero to six months, six to 12 months, more than 12 months)

• food vehicle used for fortification

• studies with and without commercial funding

We assessed differences between subgroups by interaction tests, and by inspection of the subgroups' CIs; non‐overlapping CIs indicate a statistically significant difference in treatment effect between the subgroups. Inferences for clinical relevance were based on these subgroup analyses, where possible. This ensured that the review's conclusions considered specific contextual factors in relation to food vehicle, target population and dose, among others.

Sensitivity analysis

We performed sensitivity analyses to examine the effect of removing studies at high overall risk of bias (those with high or unclear risk of bias according to the method and adequacy of allocation concealment, blinding status of the participants, or percentage lost to follow‐up, or attrition of 20% or more, or with an imbalanced loss to follow‐up in different groups).

Results

Description of studies

See Characteristics of included studies; Characteristics of excluded studies; Additional tables.

Results of the search

We identified 5789 records, of which we screened 126 full texts and included 43 studies (from 48 papers) with 19,585 participants (17,878 children) in the review (Figure 1).

1.

Study flow diagram showing results of the literature search.

Included studies

Types of studies
 We include 43 studies which meet the eligibility criteria. Most of the included studies (39) were RCTs, of which six studies were c‐RCTs (DeGier 2016; Liu 1993; Perignon 2016; Rahman 2015; Vinodkumar 2009; Wang 2017). There were four CBA studies (Abrams 2003; Adams 2017; Azlaf 2017; Mardones 2007).

Participants and Settings

Thirty‐six of the studies were conducted among children. Most of these (29) were conducted among pre‐school and school‐aged children (Aaron 2011; Abrams 2003; Adams 2017; Ash 2003; Azlaf 2017; DeGier 2016; Economos 2014; Hieu 2012; Hyder 2007; Jinabhai 2001; Lopriore 2004; Nga 2009; Osendarp 2007; Perignon 2016; Petrova 2019; Pinkaew 2013; Pinkaew 2014; Powers 2016; Rahman 2015; Sazawal 2013; Solon 2003; Taljaard 2013; Thankachan 2012; Thankachan 2013; Van Stuijvenberg 1999; Vaz 2011; Vinodkumar 2009; Wang 2017; Zimmerman 2004). Four studies included infants aged from six months to 12 months (Faber 2005; Gibson 2011; Liu 1993; Oelofse 2003), and three studies included children aged one to three years (Nesamvuni 2005; Sazawal 2007; Villalpando 2006). Three studies targeted pregnant women (Järvenpaa 2007; Mardones 2007; Tatala 2002), three studies targeted adults (Tapola 2004; Tucker 2004; Van het Hof 1998), while one study targeted an elderly population aged over 70 years (Chin A Paw 2000).

Twenty studies were conducted in lower‐middle income‐countries (LMICs) (Aaron 2011; Adams 2017; Ash 2003; Azlaf 2017; DeGier 2016; Gibson 2011; Hieu 2012; Hyder 2007; Nga 2009; Perignon 2016; Rahman 2015; Sazawal 2007; Sazawal 2013; Solon 2003; Tatala 2002; Thankachan 2012; Thankachan 2013; Vaz 2011; Vinodkumar 2009; Zimmerman 2004), 13 in upper‐middle‐income countries (UMICs) (Abrams 2003; Faber 2005; Jinabhai 2001; Liu 1993; Lopriore 2004; Nesamvuni 2005; Oelofse 2003; Pinkaew 2013; Pinkaew 2014; Taljaard 2013; Van Stuijvenberg 1999; Villalpando 2006; Wang 2017) and nine in high‐income countries (HICs) (Chin A Paw 2000; Economos 2014; Järvenpaa 2007; Mardones 2007; Petrova 2019; Powers 2016; Tapola 2004; Tucker 2004; Van het Hof 1998). One trial was conducted in both LMICs and HICs (Osendarp 2007).

We conducted a descriptive analysis of the PROGRESS‐Plus factors reported by included trials. We present an account of this analysis in Table 3. Most trials did not directly report on these factors. The analysis suggests that equity‐related variables and analysis are commonly overlooked by trials, thus affecting the generation of evidence on how inequities are identified and how interventions can contribute to mitigate or reduce them. We present all PROGRESS‐Plus factors reported by the included trials in Table 4.

1. Summary of PROGRESS‐Plus Factors.

PROGRESS‐Plus factors	Summary of reported factors
Place of residence/ setting	No. studies conducted in low‐middle‐income countries (LMICs): 20 No. studies conducted in high‐income countries (HICs): 16 One in both LMIC and HIC All with community‐dwelling individuals
Race/ethnicity/ culture/language	Most race and ethnicity categories were self‐reported. This category was scarcely reported in included studies (published data). In all, 34 studies did not report any data related to race or ethnicity. Three studies vaguely described ethnicity as: "Wagogo and Wakaguru people" (Ash 2003), "Farmers" (Nga 2009) and "mixed Berber and Arab descent" (Zimmerman 2004) The lack of a representative sample by race/ethnicity impacts on the risk of bias in those studies that report this factor, and it is unknown in those that do not None of the studies reported on cultural or language variables
Occupation	This is one of the most under‐reported categories in the studies. None of the included studies reported on this factor, except for three studies indicating that some of their participants worked either as farmers, rice farmers or construction workers, skilled labourers, or garment factory workers (Aaron 2011; Nga 2009; Perignon 2016; Thankachan 2012)
Gender/sex	This category was reported in almost every study, although some did not provide the specific distribution of the sample by sex when participants from both sexes were included. Most studies were conducted with children, but some also included adolescents, pregnant women and adults
Religion	This is one of the most under‐reported categories in the studies. Only two studies mentioned that the participants were Muslims (Pinkaew 2013; Pinkaew 2014)
Education	Two studies mentioned that the participants had primary level of education (Aaron 2011; Abrams 2003); however, many studies were carried out in school settings
Socioeconomic status	This factor was mostly poorly reported or not reported at all in the published data of the included studies. 12 studies did report this factor with various degrees of detail: Abrams 2003 specified that the families were low‐income urban families; Faber 2005; Hyder 2007; Pinkaew 2013; Pinkaew 2014; Thankachan 2012 and Van Stuijvenberg 1999 specified that the participants had low socio‐economic status; Gibson 2011 mentioned that the participants were middle income; Lopriore 2004 specified that the participants were Saharawi refugees; Osendarp 2007 included participants from two different countries (One HIC: Australia, and other LMIC: Indonesia) and specified that the participants form Australia had higher socio‐economic status while those from Indonesia had middle to low socio‐economic status; Villalpando 2006 mentioned that the participants belonged to poor per‐urban communities, while Vinodkumar 2009 specified that the families had a monthly income of less than INR 2000 (USD 50)
Social capital	No study directly reported any measurement of social capital. Indirectly, some studies reported that participants were recruited through primary schools, non‐formal schools, housing complexes and home‐care organisations, thus indicating that participants had at least one social connection or network
Plus (other characteristics)	All studies reported on age, as this factor is essential for their analysis. Many also reported the participants’ Body Mass Index (BMI). Despite this being collected, the studies did not examined BMI from a social determinants of health perspective. Some studies reported other behavioural factors such as smoking, alcohol intake and physical activity. Studies including children also reported parent education, occupation, children's anthropometric status (stunted, wasted, underweight), infections, water source, cooking fuel and toilet use, although very infrequently
Recruitment methods	Most studies recruited their participants through similar strategies: schools, mailings, printed ads and flyers distributed in university campuses, community centres, prenatal clinics or through advertisement on local radio and television. This strategy influences the composition of the samples and explains why many of them are not representative at population level. Most of the studies took place in LMICs and in children, hence the use of schools and community centres; however, this strategy may leave out literate individuals who are less connected to organisations and schools or with less access to newspapers and other written outlets

HIC: high‐income countries; LMIC: low middle income countries

2. PROGRESS‐Plus factors reported in included studies.

Study	PROGRESS‐Plus factors reported in included studies									Recruitment method
	Place of residence/ setting	Race/ ethnicity/ culture/ language	Occupation	Gender/ sex	Religion	Education	Socioeconomic status	Social capital	Plus (other characteristics)
Aaron 2011	LMIC (Nigeria)	‐	agrarian communities	Children (5 to 13 years)	‐	Primary	‐	Government‐operated primary schools	Household size, water source, cooking fuel, toilet type, maternal education and occupation, paternal education and occupation	Government‐operated primary schools
Abrams 2003	UMIC (Botswana)	‐	‐	Children (5 to 11 years	‐	Primary	Lower‐income urban families	Public schools	Age BMI	Public schools
Adams 2017	LMIC (Bangladesh)	‐	‐	Children (6 to 11 years)	‐	Primary	Rural disadvantaged districts	Primary schools	‐	Primary schools
Azlaf 2017	LMIC (Morocco)	‐	Farming	Children (7 to 9 years)	‐	Primary	Rural low‐income communities	Primary school	Water, sanitation and hygiene indicators, parent education, family income and budget	Primary school
Ash 2003	LIC (Tanzania)	Wagogo and Wakaguru people	‐	Children 6 to 11 years	‐	‐	‐	6 rural primary schools	Age BMI	Schools
Chin A Paw 2000	HIC (Netherlands)	‐	‐	Independently living, frail elderly men and women 70 years or older	‐	‐	‐	Housing complexes, home care organisations	‐	By mail from senior housing complexes, meals‐on‐wheels programmes, home‐care organisations, and general practitioners from the surroundings of Wageningen
DeGier 2016	LMIC (Cambodia)	‐	‐	Children aged 9.71 ± 2.42	‐	‐	‐	Primary schools in rural Kampong Speu province	‐	‐
Economos 2014	HIC ( USA)	‐	‐	Children 6 to 10 years	‐	‐	‐	‐	BMI	Children were recruited at Boston University Medical Center and Tufts Medical Center from the hospital paediatric clinics and through local print and online classified advertisements
Faber 2005	UMIC (South Africa)	‐	‐	Infants 6 to 12 months	‐	‐	Low SES	‐	‐	Infants were recruited through the community‐based health programme
Gibson 2011	LMIC (Zambia)	‐	‐	Infants aged 6 months	‐	‐	Middle income	‐	Breastfeeding duration, Weight‐for‐age Z‐score, Length‐for‐age Z‐score, Weight‐for length Z‐score, BMI‐for‐age Z‐score, SES. Maternal education and HIV status	‐
Hieu 2012	LMIC (Vietnam)	‐	‐	Children (6 to 9 years)	‐	‐	‐	‐	‐	School
Hyder 2007	LMIC (Bangladesh)	‐	‐	Adolescent girls	‐	‐	Low SES	Non‐formal primary education (NFPE) 9 schools	SES, menstruation, BMI	‐
Järvenpaa 2007	HIC (Finland)	‐	‐	Pregnant women 19 to 40 years	‐	‐	‐	‐	BMI, BP, present diseases, current medication, alcohol consumption, smoking habits, physical activity, and use of vitamins and other nutrients	Health care units
Jinabhai 2001	UMIC (South Africa)	‐	‐	Children aged 8 to 10 years	‐	‐	Rural community	‐	‐	Primary schools
Liu 1993	UMIC (Beijing, China)	‐	‐	Children aged 6 to 13 yrs	‐	‐	‐	Primary schools	Body weight and length,	‐
Lopriore 2004	UMIC (Algeria)	‐	‐	Children aged 3 to 6 yrs	‐	‐	Saharawi refugees	‐	‐	‐
Mardones 2007	HIC (Chile)	Ethnically mixed families (Amerindian and Hispanic)	‐	Pregnant women	‐	‐	Urban health clinics	‐	‐	Antenatal clinics
Nesamvuni 2005	UMIC (South Africa)	‐	‐	Children aged 1 to 3 yrs	‐	‐	‐	‐	Demographic, socio‐economic and dietary data, height, weight, haemoglobin, hematocrit, serum retinol and retinol‐binding protein(RBP). Anthropometric, blood and serum	Children at the creches and the well‐baby clinic were screened and the first 60 undernourished children who had weight‐for‐age or height‐for‐age below the 5th percentile of the National Center for Health Statistics (NCHS) identified. The parents/guardians of these children were contacted and recruited to voluntarily participate in the study
Nga 2009	LMIC (Vietnam)	‐	Farming	Children aged 6 to 8 yrs	‐	‐	‐	‐	Sociodemographic characteristics of the children (age, sex, illness history, medical supplements), mothers (age, and education, family size, and household socioeconomic status)	Pupils were recruited from 2 schools that had been selected on the basis of a high prevalence of anaemia and parasite infestations among school children during an earlier survey
Oelofse 2003	UMIC (South Africa)	Black community, Kayamandi	Most of the inhabitants work in the industries in the city or as domestic workers in private homes	Children aged 6 to 12 months	‐	‐	Urban disadvantaged communities, low SES	‐	Baseline food intake	Local clinics
Osendarp 2007	HIC (Australia) and LMIC (Indonesia)	‐	‐	Children aged 6 to 10 years	‐	‐	South Australian government metropolitan schools of higher SES in Adelaide, and schools in the central district of Jakarta of middle to low SES	‐	BMI, MUAC, WAZ, HAZ, WHZ, highest education in household	In Australia, the intervention was home‐based, with the children being recruited through invitations distributed either through the schools or through an additional media drive. A general, unpersonalised invitation to the parents of children in the appropriate age range was distributed through the schools
Perignon 2016	LMIC (Cambodia)	‐	Rice farming	Children aged 6 to 16 years	‐	‐	‐	Primary schools	Parasitic infection	All parents of children from the 20 schools were invited to attend a meeting at which the study procedures were explained. Written informed consent was obtained from the parents as was verbal assent from the participating children
Petrova 2019	HIC (Spain)	‐	‐	Children aged 8 to 14 years	‐	‐	‐	‐	‐	School
Pinkaew 2013	UMIC (Thailand)	‐	‐	Children aged 7 to 12 years Male : Female (10:10)	Muslim	‐	Low income	Schools in southern Thailand	‐	‐
Pinkaew 2014	UMIC (Thailand)	‐	‐	Children aged 8 to 12 yrs Male : Female 12 : 13	Muslim	‐	Low income	‐	Weight, height and BMI	‐
Powers 2016	HIC (UK)	‐	‐	Adolescent girls (aged 16 to 19 years)	‐	‐	‐	‐	‐	Schools and colleges
Rahman 2015	LMIC (Bangladesh)	‐	‐	Children aged 6 to 15 yrs	‐	‐	‐	‐	BMI	‐
Sazawal 2007	LMIC (India)	‐	‐	Children aged 1 to 3 yrs	‐	‐	‐	‐	Father/mother literacy, father/mother occupation, SES, weight, height, wasted, stunted	‐
Sazawal 2013	LMIC (Bangladesh)	‐	‐	School‐attending children aged 6 to 9 years	‐	‐	‐	Primary schools	Mother age, mother education, father education, mother employment, father employment, father income	‐
Solon 2003	LMIC (Philippines)	‐	‐	School children from grades 1 to 6	‐	‐	‐	‐	‐	School
Taljaard 2013	UMIC (South Africa)	‐	‐	School‐attending children aged 6 to 11 years	‐	‐	‐	‐	Stunted, wasted, underweight	‐
Tapola 2004	HIC (Finland)	‐	‐	Healthy volunteers aged 26 to 65 yrs; 39 men and 29 women	‐	‐	‐	‐	BMI, BP	‐
Tatala 2002	LIC (Tanzania)	‐	Semi‐arid, agricultural population	Pregnant women	‐	‐	‐	‐	‐	‐
Thankachan 2012	LMIC (India)	‐	Construction workers, skilled labourers, or garment factory workers	Children aged 6 to 12 years	‐	‐	Low SES	‐	Height, weight, BMI, stunting, SES, religion	‐
Thankachan 2013	LMIC (India)	‐	‐	School‐attending children aged 6 to 12 years	‐	‐	‐	‐	Height, weight, BMI, stunting, thinness, household head education, religion	‐
Tucker 2004	HIC (USA)	‐	‐	Healthy volunteers aged 50 to 85 years	‐	‐	‐	‐	Education, ethnicity, smoker, alcohol intake, no. of medications used	Through advertisements in local newspapers, posters, radio, and mailing lists
Van het Hof 1998	HIC (Netherlands)	‐	‐	Non‐smoking participants, healthy, aged 18 to 65 yrs	‐	‐	‐	‐	‐	Volunteers were recruited from employees of the laboratory and from inhabitants of Vlaardingen and the surrounding district
Van Stuijvenberg 1999	UMIC (South Africa)	‐	‐	Children aged 6 to 11 years	‐	‐	Low SES	‐	Stunted, underweight, parasitic infection	‐
Vaz 2011	LMIC (India)	‐	‐	Children aged 7 to 10.5 years	‐	‐	Middle socio‐economic groups	‐	‐	Schools
Villalpando 2006	UMIC (Mexico)	‐	‐	Infants aged 10 to 30 months	‐	‐	Poor peri‐urban communities	‐	Weight, length, SES indicators	Local health facility registry
Vinodkumar 2009	LMIC (India)	‐	‐	Children aged 5 to 18 years	‐	‐	The families of all the children had a monthly income of less than INR 2000 (USD 50)	Schools	‐	‐
Wang 2017	UMIC (China)	‐	‐	Children aged 12 to 14 years	‐	‐	‐	‐	Gender, weight	Schools
Zimmerman 2004	LMIC (Morroco)	Mixed Berber and Arab descent	‐	Children aged 6 to 14 years	‐	‐	‐	‐	‐	‐

BMI: body mass index; BP: blood pressure; HIC: high‐income country; LIC: low‐income country; LMIC: low middle income country; MUAC: mid‐upper arm circumference; SES: socio‐economic status; UMIC: upper middle‐income country

Duration of Intervention

The duration of intervention varied from a minimum of eight weeks to a maximum of one year. The duration of intervention in 29 studies was six months or less (Aaron 2011; Abrams 2003; Ash 2003; Chin A Paw 2000; DeGier 2016; Economos 2014; Faber 2005; Hieu 2012; Järvenpaa 2007; Jinabhai 2001; Liu 1993; Lopriore 2004; Nga 2009; Oelofse 2003; Perignon 2016; Petrova 2019; Pinkaew 2013; Pinkaew 2014; Powers 2016; Rahman 2015; Solon 2003; Tapola 2004; Tatala 2002; Thankachan 2012; Thankachan 2013; Tucker 2004; Vaz 2011; Villalpando 2006; Wang 2017), while in 14 studies the duration of intervention was between six months and one year (Adams 2017; Azlaf 2017; Gibson 2011; Hyder 2007; Mardones 2007; Nesamvuni 2005; Osendarp 2007; Sazawal 2007; Sazawal 2013; Taljaard 2013; Van het Hof 1998; Van Stuijvenberg 1999; Vinodkumar 2009; Zimmerman 2004).

Food vehicles

Food vehicles used included staple food, such as rice and flour (DeGier 2016; Faber 2005; Gibson 2011; Nesamvuni 2005; Oelofse 2003; Perignon 2016; Pinkaew 2013; Pinkaew 2014; Powers 2016; Rahman 2015; Thankachan 2012; Tucker 2004), dairy products, including milk and yogurt (Azlaf 2017; Chin A Paw 2000; Mardones 2007; Petrova 2019; Sazawal 2007; Sazawal 2013; Van het Hof 1998; Villalpando 2006; Wang 2017), non‐dairy beverages (Aaron 2011; Abrams 2003; Ash 2003; Economos 2014; Hyder 2007; Järvenpaa 2007; Osendarp 2007; Solon 2003; Taljaard 2013; Tapola 2004; Tatala 2002; Thankachan 2013; Vaz 2011), biscuits (Adams 2017; Hieu 2012; Jinabhai 2001; Liu 1993; Nga 2009; Van Stuijvenberg 1999), spreads (Lopriore 2004), and salt (Vinodkumar 2009; Zimmerman 2004).
 
 Outcomes

Anaemia, micronutrient deficiencies, anthropometric measures and serum micronutrient levels were the most commonly reported outcomes. Eight studies reported neuro‐cognitive outcomes in children (Faber 2005; Nga 2009; Osendarp 2007; Taljaard 2013; Thankachan 2012; Van Stuijvenberg 1999; DeGier 2016; Petrova 2019). None of the included studies reported on morbidity, adverse events, or all‐cause or cause‐specific mortality.

Funding

Fourteen of the included studies were fully commercially funded (Ash 2003; Economos 2014; Järvenpaa 2007; Mardones 2007; Osendarp 2007; Petrova 2019; Powers 2016; Sazawal 2007; Taljaard 2013; Tapola 2004; Thankachan 2012; Thankachan 2013; Van Stuijvenberg 1999; Vaz 2011); 13 of the included studies had partial commercial funding (Aaron 2011; Abrams 2003; Chin A Paw 2000; Faber 2005; Gibson 2011; Hyder 2007; Jinabhai 2001; Nesamvuni 2005; Pinkaew 2013; Pinkaew 2014; Solon 2003; Tatala 2002; Tucker 2004); 14 studies were non‐commercially funded (Adams 2017; Azlaf 2017; Hieu 2012; Liu 1993; Rahman 2015; Villalpando 2006; DeGier 2016; Lopriore 2004; Nga 2009; Perignon 2016; Vinodkumar 2009; Zimmerman 2004; Sazawal 2013; Wang 2017), while two studies (Van het Hof 1998; Oelofse 2003) did not specify the source of funding.

Excluded studies

We excluded 78 studies at full‐text screening. Common reasons for exclusion included point‐of‐use fortification, a pre‐post design without a control group, no outcomes of interest, and supplementation rather than fortification. See Characteristics of excluded studies for a full list of reasons for exclusion.

Risk of bias in included studies

See Figure 2; Figure 3 for the 'Risk of bias' summary and graph.

2.

Risk of bias summary: review authors' judgements about each risk of bias item for each included study.

3.

Risk of bias graph: review authors' judgements about each risk of bias item presented as percentages across all included studies.

Of the 39 included RCTs, we considered five to be at overall low risk of bias (Perignon 2016; Petrova 2019; Rahman 2015; Sazawal 2007; Vaz 2011) , and the remaining 34 to be at overall high risk of bias, due to concerns around allocation concealment, blinding of participants, or incomplete outcome data.

Allocation

We judged 18 studies to be at a low risk for random sequence generation (DeGier 2016; Economos 2014; Faber 2005; Gibson 2011; Hieu 2012; Lopriore 2004; Nga 2009; Osendarp 2007; Perignon 2016; Petrova 2019; Rahman 2015; Sazawal 2007; Sazawal 2013; Tatala 2002; Thankachan 2012; Thankachan 2013; Vaz 2011; Wang 2017); we judged one study to be at high risk of bias for sequence generation (Zimmerman 2004), while the rest were at unclear risk of bias.

We rated nine studies at a low risk for allocation concealment (Chin A Paw 2000; DeGier 2016; Hieu 2012; Perignon 2016; Petrova 2019; Rahman 2015; Sazawal 2007Sazawal 2013; Vaz 2011), one study at high risk of bias for allocation concealment (Wang 2017), while the rest were at unclear risk of bias.

Blinding

We judged 24 studies to be at low risk of bias for blinding of participants and personnel (Aaron 2011; Ash 2003; DeGier 2016; Economos 2014; Gibson 2011; Hieu 2012; Hyder 2007; Lopriore 2004; Nga 2009; Osendarp 2007; Perignon 2016; Petrova 2019; Powers 2016; Rahman 2015; Sazawal 2007; Sazawal 2013; Solon 2003; Taljaard 2013; Thankachan 2012; Thankachan 2013; Van Stuijvenberg 1999; Vaz 2011; Villalpando 2006; Zimmerman 2004), four studies at high risk of bias for blinding of participants and personnel (Chin A Paw 2000; Oelofse 2003; Van het Hof 1998; Wang 2017), while rest were at unclear risk of bias.

We rated 23 studies were at low risk of bias for blinding of outcome assessment (Aaron 2011; Ash 2003; DeGier 2016; Economos 2014; Gibson 2011; Hieu 2012; Hyder 2007; Lopriore 2004; Nga 2009; Osendarp 2007; Perignon 2016; Petrova 2019; Powers 2016; Rahman 2015; Sazawal 2007; Sazawal 2013; Solon 2003; Thankachan 2012; Thankachan 2013; Van het Hof 1998; Vaz 2011; Villalpando 2006; Zimmerman 2004), four studies at high risk of bias for blinding of outcome assessment (Chin A Paw 2000; Nesamvuni 2005; Oelofse 2003; Wang 2017), while the rest were at unclear risk of bias.

Incomplete outcome data

As the studies involved significant lifestyle changes and were carried out over a period of many weeks and months, dropouts were present, but these were either comparable in the different trial arms, or few and addressed and accounted for. We rated 26 studies at low risk of attrition bias (Aaron 2011; Ash 2003; Faber 2005; Hyder 2007; Järvenpaa 2007; Nga 2009; Perignon 2016; Petrova 2019; Pinkaew 2013; Pinkaew 2014; Powers 2016; Rahman 2015; Sazawal 2007; Solon 2003; Taljaard 2013; Tapola 2004; Thankachan 2012; Thankachan 2013; Tucker 2004; Van het Hof 1998; Van Stuijvenberg 1999; Vaz 2011; Villalpando 2006; Vinodkumar 2009; Wang 2017; Zimmerman 2004), 12 studies at high risk of attrition bias (Chin A Paw 2000; DeGier 2016; Economos 2014; Gibson 2011; Hieu 2012; Liu 1993; Lopriore 2004; Nesamvuni 2005; Oelofse 2003; Osendarp 2007; Sazawal 2013; Tatala 2002), while one study (Jinabhai 2001) was rated at unclear risk of bias.

Selective reporting

Most of the studies did not report trial registration details, but in most cases the outcomes discussed in the paper were reported. We judged only one study (DeGier 2016) to be at high risk for selective reporting, since it was powered to assess the micronutrient status but this outcome was not reported in the paper. There was a very minimal risk of reporting bias in the studies and generally we did not detect selective reporting. None of the included studies mentioned or reported on adverse effects.

Other potential sources of bias

We rated all studies at low risk for other potential bias.

For the six c‐RCTs (DeGier 2016; Liu 1993; Perignon 2016; Rahman 2015; Vinodkumar 2009; Wang 2017), we have assessed and reported additional criteria. We found all six c‐RCTs to be at low risk for recruitment bias, baseline imbalance, loss of clusters, incorrect analysis, and for comparability with individually‐randomised trials.

Risk of bias for CBA

Four studies (Abrams 2003; Adams 2017; Azlaf 2017; Mardones 2007) were assessed on additional criteria based on EPOC 2017, since these were CBA studies.

We judged all four studies to be at high risk for random sequence generation, allocation concealment and knowledge of the allocated interventions adequately prevented during the study; we rated Mardones 2007 at high risk for incomplete outcome data, while all studies were at low risk for all other criteria, including baseline outcome measurements, baseline characteristics, incomplete outcome data, protection against contamination and selective outcome reporting. We found no other sources of bias.

Risk of bias for CBA studies are reported under the 'Other Bias' section of their respective 'Risk of bias' table.

Effects of interventions

See: Table 1; Table 2

for the main comparison.

MMN fortification compared with placebo/no intervention
Patient or population: General population Settings: Community and schools Intervention: MMN fortification Comparison: Placebo/no intervention
Outcomes		Illustrative comparative risks* (95% CI)		Relative effect (95% CI)	No of participants (studies)	Quality of the evidence (GRADE)	Comments
		Assumed risk	Corresponding risk
		Placebo/no intervention	MMN fortification
Anaemia Defined as haemoglobin (Hb) concentration < 11 g/dL Measured at after 6 months and 6 ‐ 12 months of intervention		311 per 1000a	211 per 1000 (174 to 382)	RR 0.68 (0.56 to 0.84)	3746 participants (11 studies)	⊕⊕⊝⊝ lowb, c	‐
Micronutrient deficiencies	Iron deficiency Defined as serum ferritin < 15 µg/l Measured after 6 months and 6 ‐ 12 months of intervention	253 per 1000a	111 per 1000 (81 to 152)	RR 0.44 (0.32 to 0.60)	3289 participants (11 studies)	⊕⊕⊝⊝ lowb, c	‐
	Vitramin A deficiency Defined as serum retinol < 0.70 µmol/l Measured after 6 months of intervention	222 per 1000a	93 per 1000 (62 to 138)	RR 0.42 (0.28 to 0.62)	1482 participants (6 studies)	⊕⊕⊝⊝ lowb, c	‐
	Zinc deficiency Defined as serum zinc level < 0.66 mcg/mL Measured after 6 months and 6 ‐ 12 months of intervention	490 per 1000a	411 per 1000 (319 to 529)	RR 0.84 (0.65 to 1.08)	1490 participants (5 studies)	⊕⊝⊝⊝ very lowb, c, d	‐
Anthropometric	Weight‐for‐age z‐scores (WAZ) Measured as Z‐scores (standard deviation scores) Measured after 6 months and 6 ‐ 12 months of intervention	Mean WAZ score was −0.94 for the control groupa	Mean WAZ score 0.10 higher (0.02 to 0.17 higher)		2889 participants (8 studies)	⊕⊕⊝⊝ lowb, d	‐
	Height‐for‐age z‐scores/length‐for‐age z‐scores (HAZ/LAZ) Measured as Z‐scores (standard deviation scores) Measured after 6 months and 6 ‐ 12 months of intervention	Mean HAZ/LAZ score was −1.18 for the control groupa	Mean HAZ/LAZ score 0.09 higher (0.01 to 0.18 higher)		2889 participants (8 studies)	⊕⊝⊝⊝ very lowb, c, d	‐
	Weight‐for‐height z‐score/weight for length z‐score (WHZ/LHZ) Measured as Z‐scores (standard deviation scores) Measured after 6 months and 6‐12 months of intervention	Mean WHZ/LHZ score was −0.03 in the control groupa	Mean WHZ/WLZ score 0.10 higher (0.02 to 0.18 higher)		1758 participants (6 studies)	⊕⊕⊝⊝ lowb, d	‐
*The basis for the assumed risk (e.g. the median control group risk across studies) is provided in footnotes. The corresponding risk (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). CI: Confidence interval; RR: Risk Ratio; WAZ: Weight‐for‐age z‐score; HAZ/LAZ: Height‐for‐age z‐score/Length‐for‐age z‐score; WHZ/LHZ: Weight‐for‐height z‐score/Length‐for‐height z‐score.
GRADE Working Group grades of evidence High quality: Further research is very unlikely to change our confidence in the estimate of effect. Moderate quality: Further research is likely to have an important impact on our confidence in the estimate of effect and may change the estimate. Low quality: Further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate. Very low quality: We are very uncertain about the estimate.

^aAssumed risk is the mean of the post‐intervention values in the control groups in the included studies.
 ^bDowngraded by one level due to study limitations including lack of randomisation, blinding and attrition.
 ^cDowngraded by one level due to high heterogeneity (I² > 30%).
 ^dDowngraded by one level due to imprecision.

2.

MMN fortification compared with iodised salt
Patient or population: General population Settings: Community Intervention: MMN fortification Comparison: Iodised salt
Outcomes		Illustrative comparative risks* (95% CI)		Relative effect (95% CI)	No of participants (studies)	Quality of the evidence (GRADE)	Comments
		Assumed risk	Corresponding risk
		Iodised salt	MMN fortification
Anaemia Anaemia was defined as a haemoglobin (Hb) concentration < 13 g/dL for boys and < 12 g/dL for girls Measured after 9 months of intervention		698 per 1000a	600 per 1000 (258 to 1403)	RR 0.86 (0.37 to 2.01)	88 participants (1 study)	⊕⊝⊝⊝ very lowb, c	‐
Micronutrient deficiencies	Iron deficiency Defined as serum ferritin < 15 mg/L or serum transferrin concentration > 7.6 mg/L Measured after 9 months of intervention	860 per 1000a	843 per 1000 (705 to 1006)	RR 0.98 (0.82 to 1.17)	88 participants (1 study)	⊕⊝⊝⊝ very lowb, c	‐
	Vitamin A deficiency Defined as serum retinol less than 0.70 µmol/l or less than 20 ug/dL Measured after 9 and 10 months of intervention	388 per 1000d	74 per 1000 (27 to 213)	RR 0.19 (0.07 to 0.55)	363 participants (2 studies)	⊕⊝⊝⊝ very lowb, c	‐
	Zinc deficiency	‐	‐	‐	‐	‐	None of the included studies reported this outcome
Anthropometric outcomes		‐	‐	‐	‐	‐	None of the included studies reported these outcomes
*The basis for the assumed risk (e.g. the median control group risk across studies) is provided in footnotes. The corresponding risk (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). CI: Confidence interval; RR: Risk Ratio
GRADE Working Group grades of evidence High quality: Further research is very unlikely to change our confidence in the estimate of effect. Moderate quality: Further research is likely to have an important impact on our confidence in the estimate of effect and may change the estimate. Low quality: Further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate. Very low quality: We are very uncertain about the estimate.

^aAssumed risk taken from post‐intervention value in the iodised salt group of a single included study.
 ^bDowngraded by one level due to study limitations including lack of randomisation, blinding and attrition.
 ^cDowngraded by two levels due to serious imprecision, including small sample size.
 ^dAssumed risk is the mean of the post‐intervention values in the iodised salt groups in the included studies.

Comparison 1: Multiple micronutrient fortification versus placebo/no intervention

Forty studies compared MMN fortification with placebo or no intervention (Aaron 2011; Abrams 2003; Adams 2017; Ash 2003; Azlaf 2017; Chin A Paw 2000; DeGier 2016; Faber 2005; Gibson 2011; Hieu 2012; Hyder 2007; Järvenpaa 2007; Jinabhai 2001; Liu 1993; Lopriore 2004; Mardones 2007; Nesamvuni 2005; Nga 2009; Oelofse 2003; Osendarp 2007; Perignon 2016; Petrova 2019; Pinkaew 2013; Pinkaew 2014; Powers 2016; Rahman 2015; Sazawal 2007; Sazawal 2013; Solon 2003; Taljaard 2013; Tapola 2004; Tatala 2002; Thankachan 2012; Thankachan 2013; Tucker 2004; Van het Hof 1998; Van Stuijvenberg 1999; Vaz 2011; Villalpando 2006; Wang 2017). Of these, Mardones 2007 did not measure any of the outcomes included in this review.

Primary Outcomes

Among the primary outcomes, included studies reported anaemia, micronutrient deficiencies (iron, zinc, vitamin A, B vitamins) and anthropometric outcomes (weight‐for‐age z‐score (WAZ), height/length‐for‐age z‐score (HAZ/LAZ) and weight‐for‐height/length z‐score (WHZ/WLZ)). None of the included studies reported morbidity, all‐cause mortality or cause‐specific mortality.

Anaemia: Pooled study results

MMN fortification may reduce anaemia by 32% when compared to placebo/no intervention (risk ratio (RR) 0.68, 95% confidence interval (CI) 0.56 to 0.84; I² = 61%; 11 studies, 3746 participants; low‐quality evidence; Analysis 1.1; Figure 4).

1.1. Analysis.

Comparison 1 MMN vs placebo/no intervention, Outcome 1 Anaemia.

4.

Forest plot of comparison: 1 MMN vs Placebo/No intervention, outcome: 1.1 Anaemia.

Findings from Abrams 2003 and Adams 2017 were not pooled with the meta‐analysis since these were non‐randomised studies. Both Adams 2017 and Abrams 2003 reported reduced anaemia in the intervention group compared to the control group (RR 0.67, 95% CI 0.41 to 0.93 and OR 0.48, 95% CI 0.27 to 0.87, respectively).

Iron deficiency anaemia: Pooled study results

MMN fortification may reduce the prevalence of iron deficiency anaemia by 72% (RR 0.28, 95% CI 0.19 to 0.39; I² = 19%; 6 studies, 2189 participants; low‐quality evidence; Analysis 1.2; Figure 5).

1.2. Analysis.

Comparison 1 MMN vs placebo/no intervention, Outcome 2 Iron deficiency anaemia.

5.

Forest plot of comparison: 1 MMN vs Placebo, outcome: 1.2 Iron deficiency anaemia.

Micronutrient deficiencies: Pooled study results

MMN fortification may reduce micronutrient deficiencies including iron deficiency by 56% (RR 0.44; 95% CI 0.32 to 0.60; I² = 54%; 11 studies, 3289 participants; low‐quality evidence; Analysis 1.3; Figure 6); vitamin A deficiency by 58% (RR 0.42, 95% CI 0.28 to 0.62; I² = 31%; 6 studies, 1482 participants; low‐quality evidence; Analysis 1.4; Figure 7); vitamin B2 deficiency by 64% (RR 0.36, 95% CI 0.19 to 0.68; 1 study, 296 participants; low‐quality evidence; Analysis 1.5); vitamin B6 deficiency by 91% (RR 0.09, 95% CI 0.02 to 0.38; I² = 0%; 2 studies, 301 participants; low‐quality evidence; Analysis 1.5) and vitamin B12 deficiency by 58% (RR 0.42, 95% CI 0.25 to 0.71; I² = 0%; 3 studies; n = 728; moderate‐quality evidence; Analysis 1.5). We are uncertain of the effect of MMN fortification on zinc deficiency (RR 0.84, 95% CI 0.65 to 1.08; I² = 74%: 5 studies, 1490 participants; very low‐quality evidence; Analysis 1.6; Figure 8).

1.3. Analysis.

Comparison 1 MMN vs placebo/no intervention, Outcome 3 Micronutrient deficiencies: Iron.

6.

Forest plot of comparison: 1 MMN vs placebo/no intervention, outcome: 1.3 Micronutrient deficiencies: Iron.

1.4. Analysis.

Comparison 1 MMN vs placebo/no intervention, Outcome 4 Micronutrient deficiencies: Vitamin A.

7.

Forest plot of comparison: 1 MMN vs placebo/no intervention, outcome: 1.4 Micronutrient deficiencies: Vitamin A.

1.5. Analysis.

Comparison 1 MMN vs placebo/no intervention, Outcome 5 Micronutrient deficiencies: B Vitamin.

1.6. Analysis.

Comparison 1 MMN vs placebo/no intervention, Outcome 6 Micronutrient deficiencies: Zinc.

8.

Forest plot of comparison: 1 MMN vs placebo/no intervention, outcome: 1.6 Micronutrient deficiencies: Zinc.

Findings from Adams 2017 and Azlaf 2017 were not pooled with the meta‐analysis since these were not RCTs. Adams 2017 reported no effect of MMN fortification on zinc deficiency (RR 0.76, 95% CI 0.5 to 1.02), while Azlaf 2017 reported a significant effect of MMN fortification on vitamin A deficiency (prevalence of vitamin A deficiency of 4.3% in the fortified group compared to 25.2% in the control group (P < 0.001)) which are consistent with the findings of the meta‐analysis.

Anthropometric outcomes: Pooled study results weight for age

MMN fortification may improve WAZ (mean difference (MD) 0.10 z‐scores, 95% CI 0.02 to 0.17; I² = 26%; 8 studies, 2889 participants; low‐quality evidence; Analysis 1.7) and WHZ/WLZ (MD 0.10 z‐scores, 95% CI 0.02 to 0.18; I² = 5%; 6 studies, 1758 participants; low‐quality evidence; Analysis 1.8). We are uncertain about the effect of MMN fortification on HAZ/LAZ (MD 0.09, 95% CI 0.01 to 0.18; I² = 39%; 8 studies, 2889 participants; very low‐quality evidence; Analysis 1.9).

1.7. Analysis.

Comparison 1 MMN vs placebo/no intervention, Outcome 7 Anthropometric: WAZ.

1.8. Analysis.

Comparison 1 MMN vs placebo/no intervention, Outcome 8 Anthropometric: WHZ/WLZ.

1.9. Analysis.

Comparison 1 MMN vs placebo/no intervention, Outcome 9 Anthropometric: HAZ/LAZ.

Almost all the studies included in analyses of the primary outcomes were at high risk of bias. Sensitivity analysis removing all studies at overall high risk of bias left all primary analyses with only one or no included studies, indicating that risks of bias may have an important impact on the reported results.

Secondary outcomes

Among the secondary outcomes, included trials reported on serum haemoglobin, serum micronutrient concentrations (folate, ferritin, vitamin A, B vitamins and zinc) and neuro‐cognitive outcomes. None of the included trials reported any potential adverse effects.

Serum haemoglobin: Pooled study results

MMN fortification may improve serum haemoglobin (MD 3.01 g/L, 95% CI 2.14 to 3.87; I² = 99%; 20 studies, 6985 participants; low‐quality evidence; Analysis 1.10).

1.10. Analysis.

Comparison 1 MMN vs placebo/no intervention, Outcome 10 Biochemical: Serum haemoglobin (g/L).

Serum micronutrient levels: Pooled study results

MMN fortification may improve serum ferritin (MD 8.27 μg/mL, 95% CI 3.26 to 13.27; I² = 88%; 7 studies, 2407 participants; low‐quality evidence; Analysis 1.11), vitamin B6 (MD 35.02 nmol/L, 95% CI 22.95 to 47.09; I² = 82%; 2 studies, 301 participants; low‐quality evidence; Analysis 1.12), vitamin B9 (folate) (MD 12.41 nmol/L, 95% CI 6.55 to 18.28; I² = 100%; 5 studies, 568 participants; low‐quality evidence; Analysis 1.12) and vitamin B12 (MD 61.90 pmol/L, 95% CI 53.56 to 70.23; I² = 100%; 6 studies, 893 participants; low‐quality evidence; Analysis 1.12). We are uncertain of the effect of MMN fortification on serum vitamin A (MD 0.04 μmol/L, 95% CI −0.01 to 0.09; I² = 68%; 13 studies, 2457 participants; low‐quality evidence; Analysis 1.13), on serum zinc (MD 0.25 ug/dL, 95% CI −0.05 to 0.55; I² = 76%; 15 studies, 4428 participants; low‐quality evidence; Analysis 1.14) or on vitamin B1 (MD 4.80 nmol/L, 95% CI −2.77 to 12.37; 1 study, 118 participants; low‐quality evidence; Analysis 1.12).

1.11. Analysis.

Comparison 1 MMN vs placebo/no intervention, Outcome 11 Biochemical: Serum ferritin (μg/mL).

1.12. Analysis.

Comparison 1 MMN vs placebo/no intervention, Outcome 12 Biochemical: B Vitamin.

1.13. Analysis.

Comparison 1 MMN vs placebo/no intervention, Outcome 13 Biochemical: Serum vitamin A (μmol/L).

1.14. Analysis.

Comparison 1 MMN vs placebo/no intervention, Outcome 14 Biochemical: Serum zinc (μg/dL).

Neuro‐cognitive outcome: single‐study results

Eight studies reported various neuro‐cognitive outcomes in children (Faber 2005; Nga 2009; Osendarp 2007; Taljaard 2013; Thankachan 2012; Van Stuijvenberg 1999; DeGier 2016; Petrova 2019). We are uncertain of the effect of MMN fortification on motor development score (MD 1.10, 95% CI 0.17 to 2.03; 1 study, 266 participants; very low‐quality evidence; Analysis 1.15), Raven’s Colored Progressive Matrices test (RCPM) (MD 0.13, 95% CI −0.86 to 1.11; I² = 59%; 2 studies, 1124 participants; very low‐quality evidence; Analysis 1.15), general intelligence (MD −0.07, 95% CI −0.34 to 0.20; 1 study, 251 participants; very low‐quality evidence; Analysis 1.15), verbal learning and memory (MD 0.13, 95% CI −0.10 to 0.37; 1 study, 251 participants; very low‐quality evidence; Analysis 1.15), visual attention (MD 0.09, 95% CI −0.11 to 0.29; 1 study, 251 participants; very low‐quality evidence; Analysis 1.15) and coding (MD −0.53, 95% CI −1.26 to 0.21; 3 studies, 509 participants; I² = 0%; low‐quality evidence; Analysis 1.15).

1.15. Analysis.

Comparison 1 MMN vs placebo/no intervention, Outcome 15 Neuro‐cognitive outcomes.

Taljaard 2013 assessed cognition using the Kaufman Assessment Battery for Children version II (KABC) sub‐tests and the Hopkins Verbal Learning Test (HVLT), suggesting that fortification increased KABC Atlantis (intervention group mean score 5.9 compared to control group mean score 5) and HVLT Discrimination Index scores (intervention group mean score 14.7 compared to control group mean score 13.8); there was no effect on story completion, number recall, rover, triangles, word order, hand movements, recall and recognition. Van Stuijvenberg 1999 used cognitive tests designed to record speed of processing and capacity of working memory in tasks closely related to the intellectual skills required for schoolwork, suggesting improvement in the digit span forward task (short‐term memory) (P < 0.05) only, with no effect on any of the other cognitive functions (verbal fluency, digit copying, writing crosses, counting letters, cancelling letters, reading numbers, digit span backward task and counting backward).

Assessment of reporting bias

We were able to generate funnel plots for five outcomes, as they included 10 or more studies; these include anaemia, iron deficiency, serum haemoglobin, serum vitamin A and serum zinc (Figure 9; Figure 10; Figure 11; Figure 12; Figure 13). The funnel plots for the outcomes of anaemia and serum haemoglobin were visually asymmetrical, suggesting that publication bias could be one of the possible sources of asymmetry. We compared the fixed‐effect model with the random‐effects model for these two anaemia outcomes. We found the estimates to be in a similar direction of effect with overlapping CIs, indicating that smaller studies were not systematically finding more positive results, and that there is no specific indication that reporting bias might be having an important impact on this outcome. For serum haemoglobin, the sensitivity analysis showed a notably smaller result using a fixed‐effect analysis, indicating that smaller studies on average reported more positive results than larger ones. Reporting bias may be one cause of this heterogeneity, although there may be other causes that we were not able to identify (see section on subgroup analysis below).

9.

Funnel plot of comparison: 1 MMN vs Placebo/No intervention, outcome: 1.1 Anaemia.

10.

Funnel plot of comparison: 1 MMN vs placebo/no intervention, outcome: 1.3 Micronutrient deficiencies: Iron.

11.

Funnel plot of comparison: 1 MMN vs placebo/no intervention, outcome: 1.10 Biochemical: Serum haemoglobin (g/L).

12.

Funnel plot of comparison: 1 MMN vs Placebo/No intervention, outcome: 1.13 Biochemical: Serum Vitamin A (umol/L).

13.

Funnel plot of comparison: 1 MMN vs Placebo/No intervention, outcome: 1.14 Biochemical: Serum Zinc (ug/dL).

Subgroup analysis

We could only conduct subgroup analysis by duration of intervention (for the outcomes of WAZ and HAZ/LAZ) and by source of funding (for the outcomes of anaemia and iron deficiency). There were fewer than three studies in each subgroup for all other outcomes.

Subgroup analysis by the duration of intervention suggested no difference between the 'six months or less' intervention duration and 'more than six months to one year' intervention duration for WAZ (Analysis 2.1) and HAZ/LAZ (Analysis 2.2).

2.1. Analysis.

Comparison 2 MMN vs placebo/no intervention (Subgroup analysis by duration of intervention), Outcome 1 Anthropometric: WAZ.

2.2. Analysis.

Comparison 2 MMN vs placebo/no intervention (Subgroup analysis by duration of intervention), Outcome 2 Anthropometric: HAZ/LAZ.

Subgroup analysis by funding suggested no difference between non‐commercial, partial and full commercial funding for anaemia (Analysis 3.1) and iron deficiency (Analysis 3.2).

3.1. Analysis.

Comparison 3 MMN vs placebo/no intervention (Subgroup analysis by funding), Outcome 1 Anaemia.

3.2. Analysis.

Comparison 3 MMN vs placebo/no intervention (Subgroup analysis by funding), Outcome 2 Micronutrient Deficiencies: Iron.

Comparison 2: Multiple micronutrient fortification versus iodised salt

Two studies compared MMN fortification with iodised salt in children (Vinodkumar 2009; Zimmerman 2004). We could not conduct any subgroup analysis for this comparison due to the limited number of studies.

Primary outcomes

Among the primary outcomes, included studies reported anaemia, iron deficiency anaemia and micronutrient deficiencies (iron and vitamin A). None of the included trials in this comparison reported anthropometric outcomes, morbidity, all‐cause mortality or cause‐specific mortality.

Anaemia: SIngle‐study result

We are uncertain of the effect of MMN fortification when compared to iodised salt for anaemia (RR 0.86, 95% CI 0.37 to 2.01; 1 study, 88 participants; very low‐quality evidence; Analysis 4.1).

4.1. Analysis.

Comparison 4 MMN vs iodised salt, Outcome 1 Anaemia.

Iron deficiency anaemia: Pooled study results

We are uncertain of the effect of MMN fortification when compared to iodised salt for iron deficiency anaemia (RR 0.40, 95% CI 0.09 to 1.83; I² = 77%; 2 studies, 245 participants; very low‐quality evidence; Analysis 4.2).

4.2. Analysis.

Comparison 4 MMN vs iodised salt, Outcome 2 Iron deficiency anaemia.

Micronutrient deficiency: Single‐study result

We are uncertain of the effect of MMN fortification compared to iodised salt for iron deficiency (RR 0.98, 95% CI 0.82 to 1.17; 1 study, 88 participants; very low‐quality evidence; Analysis 4.3).

4.3. Analysis.

Comparison 4 MMN vs iodised salt, Outcome 3 Micronutrient deficiencies: Iron.

We are uncertain of the effect of MMN fortification compared to iodised salt on vitamin A deficiency when compared to iodised salt (RR 0.19, 95% CI 0.07 to 0.55; I² = 96%; 2 studies, 363 participants; very low‐quality evidence; Analysis 4.4).

4.4. Analysis.

Comparison 4 MMN vs iodised salt, Outcome 4 Micronutrient deficiencies: Vitamin A.

Secondary outcomes

Among the secondary outcomes, included studies reported serum haemoglobin, serum micronutrient concentrations (ferritin, B vitamins, vitamin A and zinc). None of the included studies reported any potential adverse events or neuro‐cognitive outcomes.

Serum haemoglobin: pooled study results

MMN fortification when compared to iodised salt may improve serum haemoglobin (MD 10.20 g/L, 95% CI 3.06 to 17.35; I² = 96%; 2 trials, 559 participants; low‐quality evidence; Analysis 4.5).

4.5. Analysis.

Comparison 4 MMN vs iodised salt, Outcome 5 Biochemical: Serum haemoglobin (g/L).

Serum micronutrient levels: single‐study results

We are uncertain of the effect of MMN fortification compared to iodised salt on serum ferritin (MD 0.18, 95% CI −36.14 to 36.50; 1 study, 88 participants; very low‐quality evidence; Analysis 4.6), serum vitamin B9 (MD 5.04, 95% CI −0.92 to 11.00; 1 study, 95 participants; very low‐quality evidence; Analysis 4.7), serum vitamin B12 (MD 15,184, 95% CI 6336.35 to 24,031.65; 1 study, 95 participants; very low‐quality evidence; Analysis 4.7), serum vitamin A (MD 2.82, 95% CI −2.88 to 8.51; I² = 87%; 2 studies, 363 participants; very low‐quality evidence; Analysis 4.8) and serum zinc (MD 39.77, 95% CI −86.29 to 165.83; 1 study, 95 participants; very low‐quality evidence; Analysis 4.9).

4.6. Analysis.

Comparison 4 MMN vs iodised salt, Outcome 6 Biochemical: Serum ferritin (ug/L).

4.7. Analysis.

Comparison 4 MMN vs iodised salt, Outcome 7 Biochemical: B Vitamin.

4.8. Analysis.

Comparison 4 MMN vs iodised salt, Outcome 8 Biochemical: Serum vitamin A (umol/L).

4.9. Analysis.

Comparison 4 MMN vs iodised salt, Outcome 9 Biochemical: Serum zinc.

Comparison 3: Multiple micronutrient fortification versus calcium fortification alone

Only one trial (Economos 2014) compared MMN fortification with calcium fortification in children.

Primary outcomes

None of the primary outcomes were reported in the trial.

Secondary outcomes

Among the secondary outcomes, only serum micronutrient levels were reported.

Serum micronutrient levels: SIngle study results

We are uncertain of the effect of MMN fortification on serum vitamin E (MD 5.10, 95% CI 3.49 to 6.71; 1 study, 93 participants; very low‐quality evidence; Analysis 5.1) and serum vitamin D (MD 15.10, 95% CI 3.06 to 27.14; 1 study, 93 participants; very low‐quality evidence; Analysis 5.2), serum calcium (MD 0.00, 95% CI −0.17 to 0.17; 1 study, 88 participants; very low‐quality evidence; Analysis 5.3) and serum vitamin A (MD 0.10, 95% CI −0.03 to 0.23; 1 study, 88 participants; very low‐quality evidence; Analysis 5.4) compared to calcium fortification alone.

5.1. Analysis.

Comparison 5 MMN vs calcium fortification alone, Outcome 1 Biochemical: Serum vitamin E.

5.2. Analysis.

Comparison 5 MMN vs calcium fortification alone, Outcome 2 Biochemical: Serum vitamin D.

5.3. Analysis.

Comparison 5 MMN vs calcium fortification alone, Outcome 3 Biochemical: Serum calcium.

5.4. Analysis.

Comparison 5 MMN vs calcium fortification alone, Outcome 4 Biochemical: Serum vitamin A (umol/L).

Discussion

Summary of main results

This review summarises findings from 43 studies (48 papers) with 19,585 participants (17,878 children). Most of the included studies compared MMN fortification with placebo/no intervention; two studies compared MMN fortification with iodised salt and one study compared it with calcium fortification alone. Most of the included studies targeted children, so the overall evidence generated from this review applies to children. We rated most of the evidence as of low to very low quality, due to study limitations, imprecision, high heterogeneity and small sample size. When compared to placebo/no intervention, MMN fortification may reduce anaemia, iron deficiency anaemia and micronutrient deficiencies (including iron, vitamin A, vitamin B2 and vitamin B6 deficiency). We are uncertain of the effect of MMN fortification on WAZ, WHZ/WLZ, HAZ/LAZ and other micronutrient deficiencies (including zinc and vitamin B12). Among the secondary outcomes, MMN fortification may improve serum haemoglobin, serum folate, serum ferritin, serum vitamin A and serum vitamin B12; but we are uncertain of the effect on serum zinc, serum vitamin B1 and serum vitamin B6 and any of the neuro‐cognitive outcomes. None of the included trials reported morbidity, all‐cause mortality, cause‐specific mortality or adverse events.

Two studies in children compared MMN fortification with iodised salt. We are uncertain of the effect of MMN fortification on anaemia, iron deficiency anaemia, vitamin A deficiency, serum ferritin, vitamin B9, vitamin B12, vitamin A and zinc; MMN fortification compared to iodised salt may improve haemoglobin. One trial compared MMN fortification with calcium fortification alone in children, showing inconclusive results on serum vitamin E and serum vitamin D, serum calcium and serum vitamin A in the MMN fortification group compared to calcium fortification alone.

Most of the included studies did not directly report on PROGRESS‐Plus factors or the equity‐related variables, and none of the studies reported on adverse events.

Overall completeness and applicability of evidence

This review summarises findings from 43 studies conducted between 1998 and 2018. Most of the studies were conducted in low‐ and middle‐income countries, apart from nine studies from high‐income countries, including Australia, Finland, Spain, Netherlands and the USA. Most of the studies compared MMN fortification with placebo, reporting all the primary outcomes except for morbidities, all‐cause mortality and cause‐specific mortality. None of the included trials reported on any adverse events of food fortification with MMN.

Most of the included studies targeted children, and hence the conclusions apply predominantly to children. Twenty‐nine studies were conducted among pre‐school and school‐aged children; four studies included infants aged between six and 12 months; four studies included children aged one to three years; three studies targeted pregnant women; three studies targeted adults, while one study targeted an elderly population aged over 70 years. Trials had interventions of variable duration, ranging from eight weeks to a maximum of one year. Trials used different food vehicles and numbers and concentrations of micronutrients, and the frequency of intake was also not uniform. There is limited information on the baseline nutritional status of the trial participants.

We were unable to conduct planned subgroup analysis by the various study designs, population groups, baseline micronutrient status, various combination of MMNs, low‐to‐middle income countries versus high‐income countries and the food vehicle used for fortification, due to the limited number of trials with various outcomes. Subgroup analyses comparing different durations of intervention did not identify any significant differences in outcomes. Future updates of this review could add data to various subgroups, if available at that time, which could lead to more meaningful conclusions.

Most of the included studies were fully or partially commercially funded, with a few trials being non‐commercially funded. Subgroup analyses comparing commercial versus non‐commercial funding did not find a significant difference in outcomes, although this was not a large number of studies, and the level of commercial funding remains a concern, because of the possibility of conflict of interest, and the possible association between commercial funding and more positive findings (Fabbri 2018). Independent trials and evaluations are needed to truly assess the impact of food fortification with MMN.

None of the included trials reported adverse events, which limits the completeness and applicability of the existing evidence. A descriptive analysis of the PROGRESS‐Plus factors reported by included studies suggests that most studies lack direct reporting on these factors. Equity‐related variables and analyses were commonly missing from the included studies, thus affecting the availability of evidence on how inequities are identified and how food fortification with MMN can contribute to mitigate or reduce them.

Quality of the evidence

We judged most of the outcomes to be of low to very low quality. Outcomes were mainly downgraded due to study limitations, high heterogeneity and imprecision. Study limitations included a lack of blinding of outcome assessment, incomplete outcome data and inconsistency among studies reporting the outcome. There was high heterogeneity for most of the reported outcomes. None of the included studies reported morbidity, all‐cause or cause‐specific mortality. Information on random sequence generation and allocation concealment was unclear in half of the included studies. In more than half of the included studies the methods used to conceal allocation were not described. Blinding of participants and personnel was also not clearly reported in many studies, while some of the studies were judged to be at a high risk of bias for blinding of outcome assessors. We also rated studies at high risk of bias for incomplete outcome data. This represented a major limitation, as most of the studies were fully or partially commercially funded. Lack of information on dietary intake and baseline nutritional status was another limitation of the review.

Potential biases in the review process

We were aware of the possibility of introducing bias at every stage of the reviewing process. We developed a comprehensive search strategy for a list of pre‐identified databases to capture the eligible studies. We tried to minimise bias in a number of ways; two review authors assessed eligibility for inclusion, carried out data extraction and assessed risks of bias. Nevertheless, the process of assessing risk of bias, for example, is not an exact science and includes many personal judgements. While we tried to be as inclusive as possible in our search strategies, the literature identified was predominantly written in English and published in North American and European journals. Although we tried to assess reporting bias, we largely relied on information available in the published trial reports, meaning that reporting bias was not usually apparent.

Agreements and disagreements with other studies or reviews

Our findings agree with another systematic review (Das 2013), which concluded that food fortification with MMN reduced anaemia and improved serum haemoglobin, ferritin and retinol. Another review (Best 2011) evaluating the impact of MMN fortification on micronutrient status, growth, health, and cognitive development of school children also suggested that MMN fortification improved micronutrient status and reduced anaemia prevalence, with some studies reporting positive effects on morbidity, growth, and cognitive outcomes, but the overall effects on these outcomes were equivocal. This review did not conduct any meta‐analyses. A more recent review evaluating the impact of MMN‐fortified non‐dairy beverage interventions in school‐aged children in LMICs, suggested improved serum haemoglobin and ferritin and reduced anaemia and iron deficiency anaemia (Aaron 2015).

Reviews on home fortification with MMN suggests that it is effective in reducing anaemia and iron deficiency in children aged six months to 23 months (De‐Regil 2011). There was very limited evidence on home fortification for pregnant women, suggesting that micronutrient powders for point‐of‐use fortification of foods did not have any clear effect on maternal anaemia and haemoglobin at or near term, compared with multiple micronutrient supplements (Suchdev 2014). Another review of micronutrient powders in women and children also suggests that they are effective in improving anaemia and haemoglobin among children reporting lack of impact on growth; evidence of increased diarrhoea requires careful consideration before recommending the intervention for large‐scale implementation (Salam 2013).

Authors' conclusions

Implications for practice.

The evidence from this review suggests that MMN fortification when compared to placebo may improve anaemia, iron deficiency anaemia, micronutrient deficiencies (including iron, vitamin A, vitamin B2 and vitamin B6 deficiency), serum haemoglobin, serum folate, serum ferritin, serum vitamin A, serum vitamin B12 and some motor and cognitive outcomes. However, there are a number of other factors that should also be considered. Firstly, the quality of the evidence was low to very low. Secondly, there are no reported data to assess possible side effects of the MMN fortification. Thirdly, we could not draw reliable conclusions from various subgroup analyses on population groups, food vehicles, dosage and region, due to a limited number of studies in each subgroup and measuring varying outcomes. Lastly, we remain cautious about the level of commercial funding among the included studies, although a direct effect of commercial funding was not demonstrated in this review.

Implications for research.

The findings of our review provide a number of implications for future research. Future research should focus on generating high‐quality evidence with longer follow‐ups, and assessing the impact in various population groups. The evidence can be consolidated with the use of larger sample sizes and better study designs. It would also be important for study authors to report allocation, randomisation and blinding procedures in detail. There is a need for non‐commercially funded studies and independent evaluations. There are limited data on how fortification affects population groups with variable baseline health status and underlying micronutrient deficiencies and levels of malnutrition. Research should also focus on evaluating the direct health outcomes, including morbidities, mortality and adverse events, especially in LMIC settings. It is also important to report on equity variables for future studies, to assess whether food fortification has any impact on equity.

What's new

Date	Event	Description
19 February 2020	Amended	Typo corrected in Plain language summary (removal of the word 'Title in heading)

Appendices

Appendix 1. MEDLINE search strategy

1. Food

(food or crops or crop or flour or salt or salts or fish or soy foods or sauce* or cereals or carbohydrates or sugar* or Oryza sativa or rice* or milk or bread or oil or oils or beverages or yogurt or margarine or cheese or maize* or condiments or triticum or wheat* or spice or spices or curry powder* or fats or fat or dairy).mp. OR exp Zea mays/

2. Micronutrients

(iron or ferr* or iodine or vitamin a or beta carotene or folic or folate* or micronutrients).mp.

3. Fortification

(enrich* or forti* or enhance* or refine*).mp.

1 AND 2 AND 3

Appendix 2. Embase search strategy

1. (food or 'food supply' or crop or bread or flour or salt or 'fish products' or 'soy food*' or sauce or sugar or wheat or triticum or rice or cereal* or grain* or cheese or dairy or cake* or biscuit* or juice* or yogurt or chocolate or margarine or milk or oil or butter or cream or yogurt or beverages or spice or 'curry powder*' or 'dietary fats' or fat or fats or condiment*)

2. Iron or ferr* or Iodine or Vitamin A or beta Carotene or Folic Acid or folate* or Micronutrients

3. enrich* or forti* or enhance* or refine*

1 AND 2 AND 3

Appendix 3. CINAHL

1. (food or “food supply” or crop or bread or flour or salt or “fish products” or “soy foods” or sauce or sugaror wheat or triticum or rice or cereal* or grain* or cheese or dairy or cake* or biscuit* or juice* or yogurt or chocolate or margarine or milk or maize or oil or butter or cream or yogurt or beverages or Spices or "curry powder*" or fat or fats or condiment*)

2. Iron or ferr* or Iodine or Vitamin A or beta Carotene or Folic Acid or folate* or Micronutrients

3. (enrich* or forti* or enhance* or refine*)

1 AND 2 AND 3

Appendix 4. Cochrane

(Food or "Food Supply" or "Agricultural Crops" or crop or Flour or Salts or "Fish Products" or "Soy Foods" or sauce* or Cereals or "Dietary Carbohydrates" or sugar* or "Oryza sativa" or rice or Milk or Bread or Oils or Beverages or Yogurt or Margarine or Cheese or "Zea mays" or maize* or Condiments or Triticum or wheat* or Spices or "curry powder" or "Dietary Fats" or fat or fats or "Dairy Products") AND (Iron or ferr* or Iodine or "Vitamin A" or "beta Carotene" or "Folic Acid" or folate or Micronutrients) AND (enrich or fortified or enhance or refine)

Appendix 5. WHOLIS

words or phrase “food or food supply or crop or bread or flour or salt or fish product or soy food or sauce or sugar or wheat or triticum or rice or cereal or grain or cheese or dairy or cake or biscuit or juice or yogurt or chocolate or margarine or milk or oil or butter or cream or yogurt or beverages or spices or curry powder or dietary fat or fats or fat” AND words or phrase "fortified or fortification or enriched or enhanced or refined” AND Iron or ferr* or Iodine or Vitamin A or beta Carotene or Folic Acid or folate* or Micronutrients

Appendix 6. Others

LILAC:

Fortification or fortified or enriched (Subject)

FAO/AGRIS:

Fortified (All fields) OR Enrich (All fields) OR fortification (All fields)

African Index Medicus:

fortification or fortified or enriched (titles and keywords)

EMRO

fortification or fortified or enriched (with atleast one word)

PAHO

fortified or fortification or enriched

WPRO:

allintitle: fortified OR fortification OR enriched

IMSEAR:

fortified OR fortification OR enriched

3ie Database of Impact studies

fortification OR enriched OR fortified (Health and Nutrition and Population subcategory

EPPI

fortification OR enriched OR fortified (Free text terms)

Open Grey

fortification OR fortified AND food discipline:(05T ‐ Health services, health administration, community care services) discipline:(06H ‐ Food technology, food microbiology)

Clinical trials.gov:

"food fortification" OR fortified (restricted to interventional studies only)

http://apps.who.int/trialsearch/AdvSearch.aspx

"food fortification" OR fortified or enriched (titles)

Food Science and Technology Abstracts

fortification OR enriched OR fortified

AgriCOLA: https://agricola.nal.usda.gov/

fortification OR enriched OR fortified

Data and analyses

Comparison 1. MMN vs placebo/no intervention.

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 Anaemia	11	3746	Risk Ratio (Random, 95% CI)	0.68 [0.56, 0.84]
2 Iron deficiency anaemia	6	2189	Risk Ratio (M‐H, Random, 95% CI)	0.28 [0.19, 0.39]
3 Micronutrient deficiencies: Iron	11	3289	Risk Ratio (Random, 95% CI)	0.44 [0.32, 0.60]
4 Micronutrient deficiencies: Vitamin A	6	1482	Risk Ratio (Random, 95% CI)	0.42 [0.28, 0.62]
5 Micronutrient deficiencies: B Vitamin	4		Risk Ratio (Random, 95% CI)	Subtotals only
5.1 Vitamin B2	1	296	Risk Ratio (Random, 95% CI)	0.36 [0.19, 0.68]
5.2 Vitamin B6	2	301	Risk Ratio (Random, 95% CI)	0.09 [0.02, 0.38]
5.3 Vitamin B12	3	728	Risk Ratio (Random, 95% CI)	0.42 [0.25, 0.71]
6 Micronutrient deficiencies: Zinc	5	1490	Risk Ratio (M‐H, Random, 95% CI)	0.84 [0.65, 1.08]
7 Anthropometric: WAZ	8	2889	Mean Difference (IV, Random, 95% CI)	0.10 [0.02, 0.17]
8 Anthropometric: WHZ/WLZ	6	1758	Mean Difference (IV, Random, 95% CI)	0.10 [0.02, 0.18]
9 Anthropometric: HAZ/LAZ	8	2889	Mean Difference (IV, Random, 95% CI)	0.09 [0.01, 0.18]
10 Biochemical: Serum haemoglobin (g/L)	20	6985	Mean Difference (Random, 95% CI)	3.01 [2.14, 3.87]
11 Biochemical: Serum ferritin (μg/mL)	7	2407	Mean Difference (Random, 95% CI)	8.27 [3.26, 13.27]
12 Biochemical: B Vitamin	7		Mean Difference (Random, 95% CI)	Subtotals only
12.1 Vitamin B1 (nmol/L)	1	118	Mean Difference (Random, 95% CI)	4.8 [‐2.77, 12.37]
12.2 Vitamin B6 (nmol/L)	2	301	Mean Difference (Random, 95% CI)	35.02 [22.95, 47.09]
12.3 Vitamin B9 (nmol/L)	5	568	Mean Difference (Random, 95% CI)	12.41 [6.55, 18.28]
12.4 Vitamin B12 (pmol/L)	6	893	Mean Difference (Random, 95% CI)	61.90 [53.56, 70.23]
13 Biochemical: Serum vitamin A (μmol/L)	13	2457	Mean Difference (Random, 95% CI)	0.04 [‐0.01, 0.09]
14 Biochemical: Serum zinc (μg/dL)	15	4428	Mean Difference (IV, Random, 95% CI)	0.25 [‐0.05, 0.55]
15 Neuro‐cognitive outcomes	6		Mean Difference (Random, 95% CI)	Subtotals only
15.1 Motor development score	1	266	Mean Difference (Random, 95% CI)	1.1 [0.17, 2.03]
15.2 Raven's coloured matrices	2	1124	Mean Difference (Random, 95% CI)	0.13 [‐0.86, 1.11]
15.3 General intelligence	1	251	Mean Difference (Random, 95% CI)	‐0.07 [‐0.34, 0.20]
15.4 Verbal learning and memory	1	251	Mean Difference (Random, 95% CI)	0.13 [‐0.10, 0.37]
15.5 Visual attention	1	251	Mean Difference (Random, 95% CI)	0.09 [‐0.11, 0.29]
15.6 Coding	3	509	Mean Difference (Random, 95% CI)	‐0.53 [‐1.26, 0.21]

Comparison 2. MMN vs placebo/no intervention (Subgroup analysis by duration of intervention).

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 Anthropometric: WAZ	8	2889	Mean Difference (IV, Random, 95% CI)	0.10 [0.02, 0.17]
1.1 ≤ six months	5	1598	Mean Difference (IV, Random, 95% CI)	0.06 [‐0.02, 0.14]
1.2 > six months to one year	3	1291	Mean Difference (IV, Random, 95% CI)	0.13 [‐0.00, 0.25]
2 Anthropometric: HAZ/LAZ	8	2889	Mean Difference (IV, Random, 95% CI)	0.09 [0.01, 0.18]
2.1 ≤ six months	5	1598	Mean Difference (IV, Random, 95% CI)	0.05 [‐0.04, 0.14]
2.2 > six months to one year	3	1291	Mean Difference (IV, Random, 95% CI)	0.14 [0.01, 0.27]

Comparison 3. MMN vs placebo/no intervention (Subgroup analysis by funding).

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 Anaemia	11		Risk Ratio (Random, 95% CI)	0.68 [0.56, 0.84]
1.1 Non‐commercial funding	4		Risk Ratio (Random, 95% CI)	0.43 [0.17, 1.10]
1.2 Partial commercial funding	3		Risk Ratio (Random, 95% CI)	0.66 [0.54, 0.81]
1.3 Full commercial funding	4		Risk Ratio (Random, 95% CI)	0.77 [0.65, 0.92]
2 Micronutrient Deficiencies: Iron	11		Risk Ratio (Random, 95% CI)	0.43 [0.32, 0.59]
2.1 Non‐commercial funding	4		Risk Ratio (Random, 95% CI)	0.45 [0.18, 1.08]
2.2 Partial commercial funding	3		Risk Ratio (Random, 95% CI)	0.23 [0.10, 0.56]
2.3 Full commercial funding	4		Risk Ratio (Random, 95% CI)	0.45 [0.32, 0.64]

Comparison 4. MMN vs iodised salt.

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 Anaemia	1		Risk Ratio (Random, 95% CI)	Totals not selected
2 Iron deficiency anaemia	2	245	Risk Ratio (Random, 95% CI)	0.40 [0.09, 1.83]
3 Micronutrient deficiencies: Iron	1		Risk Ratio (Random, 95% CI)	Totals not selected
4 Micronutrient deficiencies: Vitamin A	2	363	Risk Ratio (Random, 95% CI)	0.19 [0.07, 0.55]
5 Biochemical: Serum haemoglobin (g/L)	2	559	Mean Difference (Random, 95% CI)	10.20 [3.06, 17.35]
6 Biochemical: Serum ferritin (ug/L)	1	88	Mean Difference (Random, 95% CI)	0.18 [‐36.14, 36.50]
7 Biochemical: B Vitamin	1		Mean Difference (Random, 95% CI)	Totals not selected
7.1 Vitamin B9	1		Mean Difference (Random, 95% CI)	0.0 [0.0, 0.0]
7.2 Vitamin B12	1		Mean Difference (Random, 95% CI)	0.0 [0.0, 0.0]
8 Biochemical: Serum vitamin A (umol/L)	2	363	Std. Mean Difference (Random, 95% CI)	2.82 [‐2.88, 8.51]
9 Biochemical: Serum zinc	1		Mean Difference (Random, 95% CI)	Subtotals only

Comparison 5. MMN vs calcium fortification alone.

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 Biochemical: Serum vitamin E	1		Mean Difference (IV, Random, 95% CI)	Totals not selected
2 Biochemical: Serum vitamin D	1		Mean Difference (IV, Random, 95% CI)	Totals not selected
3 Biochemical: Serum calcium	1		Mean Difference (IV, Random, 95% CI)	Totals not selected
4 Biochemical: Serum vitamin A (umol/L)	1		Mean Difference (IV, Random, 95% CI)	Totals not selected

Characteristics of studies

Characteristics of included studies [ordered by study ID]

Aaron 2011.

Methods	Randomised controlled trial
Participants	Participants in the study were 566 apparently healthy male and female children attending 2 government‐operated primary schools located in Akanga and Akaleku, Nasarawa State, Nigeria, who met the following criteria: 5 – 13 year old and haemoglobin ≥ 70 g/L
Interventions	Intervention (n = 288): Children received a single daily serving of a multi‐micronutrient beverage or a placebo beverage 5 days/week for 6 months. The beverages were isoenergetic and were composed of a proprietary blend of precooked maize and soy protein isolate, providing 689 kJ energy, 4.1 g fat, and 5.2 g protein/serving. Control (n = 278): Placebo Food vehicle: Beverage Dose: Retinol palmitate, 1 mg retinol equivalents, D‐biotin, 47 mg, ascorbic acid, 60 mg, cholecalciferol, 1.2 mg, dl‐a‐tocopherol acetate, 6.5 mg, folic acid, 200 mg, niacinamide, 18 mg, calcium D‐pantothenate, 2.3 mg, pyridoxine HCl, 2 mg, riboflavin, 1.6 mg, thiamine HCl, 1 mg, calcium carbonate and calcium lactate, 84 mg, copper sulfate 5‐hydrate, 1.2 mg, potassium iodate, 150 mg, ferrous bisglycinate chelate and ferrous sulfate, 14 mg, magnesium oxide, 49 mg, manganese glycinate chelate, 4.5 mg, molybdenum amino acid chelate, 29 mg, monosodium phosphate anhydrous, 187 mg, potassium chloride, 276 mg, selenium amino acid complex, 24.8 mg, vanadium nicotinate glycinate chelate, 25 mg, zinc glycinate chelate and zinc oxide, 15 mg, bioflavonoids, 87.5 mg Duration: 6 months Additional Interventions: Per school policy, all children were given a single 200 mg dose of albendazole 1 week prior to the baseline blood draw and 1 month prior to the final blood draw
Outcomes	Haemoglobin, serum ferritin, serum retinol, serum zinc
Notes	Supported by funding from Global Alliance for Improved Nutrition (Geneva, Switzerland). The multi‐micronutrient beverage was developed by International Nutrition and Sport S.A. (Pty) Limited Study duration: January to August 2007
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Quote: "Rosters were obtained from both schools and, using a stratified sampling scheme, children were randomly assigned to groups at the individual level proportionate to the number of male and female students in each school and class level." Comment: Insufficient information to permit judgement
Allocation concealment (selection bias)	Unclear risk	Quote: "Rosters were obtained from both schools and, using a stratified sampling scheme, children were randomly assigned to groups at the individual level proportionate to the number of male and female students in each school and class level." Comment: Insufficient information to permit judgement
Blinding of participants and personnel (performance bias) All outcomes	Low risk	Quote: "Products were masked for taste, colour, aroma, texture, and packaging and were labelled with unique product codes, which were maintained by the manufacturer until after the data were analysed." Comment: Adequately done
Blinding of outcome assessment (detection bias) All outcomes	Low risk	Quote: "Products were masked for taste, colour, aroma, texture, and packaging and were labelled with unique product codes, which were maintained by the manufacturer until after the data were analysed." Comment: Adequately done
Incomplete outcome data (attrition bias) All outcomes	Low risk	Intervention group: 18/288 Control: 14/278 Overall 6% attrition rate Comment: Low attrition rate unlikely to affect results
Selective reporting (reporting bias)	Low risk	Comment: Trial registration or published protocol not identified, but the outcomes specified in the methodology section have been reported in the results
Other bias	Low risk	No additional bias identified

Abrams 2003.

Methods	Controlled before‐after study
Participants	Conducted in Botswana. 311 Children were considered eligible for the study if they were 5 – 11 years old, weighed > 15 kg, had a haemoglobin concentration > 60 g/L and had no known chronic illnesses such as HIV or recent acute illnesses
Interventions	Intervention (n = 164): Children were to receive an average of 240 mL daily of the fortified beverage under direct observation. The treatment group received a fruit‐flavoured beverage containing 419 kJ/240 mL with a proprietary blend of micronutrients. 7 servings were given a week for a total of 1680 ml/week. Because it was not possible to administer drinks on weekends, we gave 2 drinks to each participant on Monday and Friday Control (n = 147): Placebo Food vehicle: Fruit‐flavoured beverage Dose: B‐carotene, 2400 ug, riboflavin 0.4 mg, niacin 2.7 mg, pyridoxine HCl 0.5 mcg, folic acid 140 ug, cyanocobalamin 1 ug, ascorbic acid 60 mg, dl‐a ‐tocopherol acetate 7.5 mg, tricalcium phosphate 120 mg, ferrous bisglycinate chelate 7 mg, potassium iodide 60 g, zinc gluconate 3.75 mg Duration: 8 weeks
Outcomes	Weight, mid‐upper arm circumference, haemoglobin, retinol, ferritin, vitamin B12, folate and riboflavin status
Notes	This project was financed in part with federal funds from the USDA/ARS under Cooperative Agreement number 58–6250‐6–001 and by The Minute Maid Company, Houston, TX Study duration: Study dates not reported.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	High risk	Quote: "All students in one school were provided the experimental beverage, whereas the control beverage was given at the other school. By assigning the subjects to either the fortified beverage or control, using their school for assignment, we hoped to be able to avoid compromising the study by an error in administration, while still obtaining meaningful results." Comment: High risk
Allocation concealment (selection bias)	High risk	Quote: "All students in one school were provided the experimental beverage, whereas the control beverage was given at the other school. By assigning the subjects to either the fortified beverage or control, using their school for assignment, we hoped to be able to avoid compromising the study by an error in administration, while still obtaining meaningful results." Comment: High risk
Blinding of participants and personnel (performance bias) All outcomes	High risk	Quote: "On the basis of input from local medical and educational officials, it was determined that there was a substantial risk of confusing the beverages by attempting to provide more than one drink at each school, and that maintaining blinding would then be nearly impossible." Comment: High risk
Blinding of outcome assessment (detection bias) All outcomes	High risk	Quote: "On the basis of input from local medical and educational officials, it was determined that there was a substantial risk of confusing the beverages by attempting to provide more than one drink at each school, and that maintaining blinding would then be nearly impossible." Comment: High risk
Incomplete outcome data (attrition bias) All outcomes	Low risk	Intervention group: 19/164 Control group: 29/147 Overall 15% attrition rate Comment: Low attrition rate
Selective reporting (reporting bias)	Low risk	Comment: Published protocol not identified, but the outcomes specified in the methodology section have been reported in the results
Other bias	Unclear risk	Baseline outcome measurements: Quote: "The baseline anthropometric and biochemical studies supported the equivalence of the two groups" Comment: Low risk Baseline characteristics: Quote: ""There were no significant differences in baseline characteristics between subjects who completed the study and those who did not." Comment: Low risk Protection against contamination: Quote: "On the basis of input from local medical and educational officials, it was determined that there was a substantial risk of confusing the beverages by attempting to provide more than one drink at each school, and that maintaining blinding would then be nearly impossible. Two schools were required to provide an adequate number of subjects for the trial." Comment: Low risk

Adams 2017.

Methods	Controlled before‐after study conducted among all primary school‐going children aged 6 ‐ 11 years in 10 disadvantaged sub‐districts in Bangladesh from September 2011 to November 2012
Participants	368 primary school children at baseline and 351 children at endline
Interventions	Intervention (n = 191): Daily administration of a packet of fortified biscuit to all primary school‐going children aged 6 to 11 years. Biscuit ingredients were: wheat flour (69% by weight); sugar (12%); vegetable fat (hydrogenated‐75% and liquid‐25% ‐ 13%); soya flour (6%); iodised salt (0.5%); leavening agent (1.0%) and micronutrient premix (1.5 kg premix in 998.5 kg biscuit dough). The fortified biscuit was prepared to provide 300 kcal per single 75 gm packet (approximately 15% of daily calorie requirements), and a range of micronutrients contributing to about 75% of the daily requirements of vitamin A, folate, iron, iodine, zinc and magnesium Energy: 450 kcal Moisture (maximum): 4.5% Protein: 10 ‐ 15 g Fat: 15 g Calcium: 212.5 ‐ 287.5 mg, magnesium: 127.5 ‐ 172.5 mg, vitamin A (retinol): 212.5 ‐ 287.5 mcg, vitamin D: 1.615 ‐ 2.185 mcg, vitamin E: 4.25 ‐ 5.75 mg, Vitamin B1: 0.425 ‐ 0.575 mg, vitamin B2: 0.595 ‐ 0.805 mg, vitamin B3 (niacin): 5.1 ‐ 6.9 mg, vitamin B5 (pantothenic acid): 2.55 ‐ 3.45 mg, vitamin B6: 0.85 ‐ 1.15 mg, vitamin B12: 0.425 ‐ 0.575 mcg, folic acid: 680 ‐ 920 mcg, vitamin C: 17.0 ‐ 23.0 mg, iron: 9.35 ‐ 12.65 mg, iodine: 63.75 ‐ 86.25 mcg zinc: 7.00 ‐ 8.00 mg Control area (n = 177): did not receive any intervention Duration: 12 months
Outcomes	Haemoglobin levels, micronutrient (ferritin, folic acid, vitamin B12, retinol, zinc, iodine, vitamin D) levels, anaemia
Notes	This study was supported by the European Union (EU), through a sub‐contract from the James P. Grant School of Public Health, BRAC University, Dhaka Bangladesh Study duration: September 2011 to November 2012
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	High risk	Quote: "The quantitative component assessed the impact of micronutrient fortification on 351 children aged 6 ‐ 11 years using a cohort pre‐post research design with a control group." Comment: High risk
Allocation concealment (selection bias)	High risk	Quote: "The quantitative component assessed the impact of micronutrient fortification on 351 children aged 6 ‐ 11 years using a cohort pre‐post research design with a control group." Comment: High risk
Blinding of participants and personnel (performance bias) All outcomes	High risk	Comment: The control group did not receive any intervention.
Blinding of outcome assessment (detection bias) All outcomes	High risk	Comment: The control group did not receive any intervention.
Incomplete outcome data (attrition bias) All outcomes	Low risk	Intervention group: 2/191 Control group: 15/177 Overall 4.6% attrition rate Comment: Low risk for attrition bias
Selective reporting (reporting bias)	Low risk	Comment: Published protocol not identified, but the outcomes specified in the methodology section have been reported in the results
Other bias	Unclear risk	Baseline outcome measurements: Quote: "at baseline, the characteristics of primary school students in intervention and control groups were largely similar." Comment: Low risk Baseline characteristics: Quote: "at baseline, the characteristics of primary school students in intervention and control groups were largely similar." Comment: Low risk Protection against contamination: Quote:"Similar measurements were made on a control group of primary school children living in adjacent sub‐districts where the program had not been implemented." Comment: Low risk

Ash 2003.

Methods	Randomised controlled trial
Participants	Conducted in Tanzania. Participants included 830 rural children (aged 6 – 11 years) attending primary schools
Interventions	Intervention (n = 382): Fortified beverage. One serving of the beverage was provided at school during the morning recess Control (n = 392): Unfortified beverage provided 90 kcal in each 25 g individual‐serving sachet Dose: iron 5.4 mg, Vit A 1750 IU, iodine 45 ug, zinc 5.25 mg, ascorbic acid 72 mg, riboflavin 0.6 mg, folic acid 0.14 mg, vit B 12.3 ug, B6 0.7 mg, E 10.5 mg in 25 g sachet Food vehicle: The content of 1 sachet was mixed with 250 mL previously boiled water to make a pleasant‐tasting, orange‐flavoured beverage Duration: 6 months
Outcomes	Serum haemoglobin, ferritin, protoporphyrin, retinol, height, weight , BMI
Notes	The dietary supplement used was developed and produced by food technologists at Procter & Gamble and was made available in the form of a multiple‐micronutrient beverage powder. The beverage was developed to be nutritionally adequate and pleasant‐ tasting without problems of nutrient instability, off colour, or off flavour Study duration: November–December 1995 to July–August 1996
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Quote: "Each child in each stratum was then randomly allocated to receive either the fortified or unfortified beverage in a double‐blind manner." Comment: Insufficient information to permit judgement
Allocation concealment (selection bias)	Unclear risk	Quote: "Each child in each stratum was then randomly allocated to receive either the fortified or unfortified beverage in a double‐blind manner." Comment: Insufficient information to permit judgement
Blinding of participants and personnel (performance bias) All outcomes	Low risk	Quote: "The fortified beverage and the unfortified (placebo) beverage were identical in terms of taste and appearance", "The research team, schoolteachers, and schoolchildren were blinded as to whether the sachets were fortified or unfortified, i.e., the meaning of the label colours was not revealed" Comment: Adequately done
Blinding of outcome assessment (detection bias) All outcomes	Low risk	Quote:"The research team, schoolteachers, and schoolchildren were blinded as to whether the sachets were fortified or unfortified, i.e., the meaning of the label colours was not revealed" Comment: Adequately done
Incomplete outcome data (attrition bias) All outcomes	Low risk	Total loss to follow‐up 7% (56/830), Comment: Low attrition rate
Selective reporting (reporting bias)	Low risk	Comment: Trial registration or published protocol not identified, but the outcomes specified in the methodology section have been reported in the Results
Other bias	Low risk	No additional bias identified

Azlaf 2017.

Methods	A double‐blind CBA study conducted among school‐aged children aged 7 ‐ 9 years living in a rural and mountainous area of Morocco between February and October 2012
Participants	194 school children aged 7 ‐ 9 years were recruited from 3 primary schools
Interventions	Children were divided into 2 groups to receive 200 ml of either: fortified milk (fortified milk group: FMG) or non‐fortified milk (non‐fortified milk group: NFMG) The intervention (n = 79) was delivered to children by the headmaster or by the teachers with consumption of milk during the morning under close supervision Energy (Kcal): 154.8 Fat (%): 5.8 Protein (g): 5.8 Lipids (g): 6 Carbohydrates (g): 19.44 Calcium (mg): 240, iron (mg): 4.2, iodine (g): 45, vitamin A (g): 240, vitamin D3 (g): 3, Control group (n = 115): NFMG Duration: 9 months
Outcomes	Serum vitamin A levels and vitamin A deficiency
Notes	Milk was provided by the Foundation for Child Nutrition. Study duration: February to October 2012
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	High risk	Quote: "This study is a longitudinal interventional, double‐blind (participants and assessors), and controlled one." Comment: High risk
Allocation concealment (selection bias)	High risk	Quote: "This study is a longitudinal interventional, double‐blind (participants and assessors), and controlled one." Comment: High risk
Blinding of participants and personnel (performance bias) All outcomes	Low risk	Quote: "The fortified and unfortified milk were similar in macro‐nutrient composition, taste, aroma, texture and packaging but not in micronutrients content" Comment: Low risk
Blinding of outcome assessment (detection bias) All outcomes	Low risk	Quote: "The fortified and unfortified milk were similar in macro‐nutrient composition, taste, aroma, texture and packaging but not in micronutrients content" Comment: Low risk
Incomplete outcome data (attrition bias) All outcomes	Low risk	Interventiong roup: 10/79 Control group: 4/115 Over 7% attrition rate Comment: Low risk of attrition
Selective reporting (reporting bias)	Low risk	Comment: The study was retrospectively registered in the Pan African Clinical Trial. Registry with the identification number PACTR201410000896410. Outcomes specified in the methodology have been reported in the results section.
Other bias	Unclear risk	Baseline outcome measurements: Comment: Table 2 reports similar baseline outcome measurements between the two groups hence judged to be at low risk of bias for baseline outcome measurements Baseline characteristics: Quote: "The two groups of children were well‐balanced with respect to age and gender.." Comment: Low risk Protection against contamination: Comment: The intervention and control milk were provided in different school hence judged to be at low risk of contamination.

Chin A Paw 2000.

Methods	Randomised controlled trial
Participants	Conducted among 224 independently‐living, frail elderly men and women in the Netherlands The inclusion criteria included: age 70 or older, a need for care services (e.g. home care, meals‐on‐wheels), not participating regularly in physical activities of moderate to high intensity (weekly more than 30 minutes of brisk walking, cycling, gymnastics), BMI (based on self‐reported height and weight) ≤ 25 kg/m2 or involuntary weight loss, non‐institutionalised, not taking multivitamin supplements for the last month, no terminal disease or rapidly deteriorating health status, and the ability to comprehend the procedures of the study
Interventions	Participants were divided into the following 4 groups: Twice‐weekly comprehensive, moderate‐intensity, progressive group exercise (n = 55); Daily enriched foods (n = 57); As 2 and 3 (n = 60); or Neither 2 nor 3 (n = 52). For this review, we have only included data from groups 2 and 4 from the above Intervention: Participants were asked to eat 1 fruit product (100 g portions juice and compote) and 1 dairy product (100 g portions vanilla custard and fruit yogurt, 75 g portions vanilla fruit soft curd cheese) daily for 17 weeks. They were allowed to eat the products either in addition to their daily diet or as a replacement Control: Placebo Food vehicle: A number of fruit and dairy products were enriched with several vitamins and minerals for which elderly people’s intake or status is frequently low Dose: vitamin D (7.5 mg), E (8.9 mg), B1 (1 mg), B2 (1.4 mg), B6 (1.1 mg), folic acid (0.25 mg), B12 (2.5 mg), and C (70 mg), calcium (225 mg), magnesium (75 mg), zinc (4.75 mg), iron (4.25 mg), and iodine (0.24 mg) Duration: 17 weeks
Outcomes	Serum pyridoxine, serum ascorbic acid, serum folate, serum zInc, Serum Iron, mean fitness score, mean performance score
Notes	The Dutch Health Research Council, The Hague, The Netherlands; and Wiebe Visser of the Dutch Dairy Foundation on Nutrition and Health, Maarssen, Roche Nederland B.V., Friesland Coberco Dairy Foods B.V., Campina Melkunie–Mona Division, Bekina Lebensmittel GmbH, subsidiary of Royal Numico NV. Study duration: Enrollment done between January through July 1997 and the intervention continued for 17 weeks.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Quote: "Eligible subjects .. were randomly assigned to: supervised group exercise ...; enriched food products ...; both ..; or a control group .... Group assignment took place before baseline measurements with sealed envelopes." Comment: Insufficient information to permit judgement
Allocation concealment (selection bias)	Low risk	Quote: "Group assignment took place before baseline measurements with sealed envelopes." Comment: Adequately done
Blinding of participants and personnel (performance bias) All outcomes	High risk	Quote: "The nutritional intervention was intended to be double blinded." Comment: Probably not done
Blinding of outcome assessment (detection bias) All outcomes	High risk	Quote: "The nutritional intervention was intended to be double blinded." Comment: Probably not done
Incomplete outcome data (attrition bias) All outcomes	High risk	Total loss to follow‐up of 26% (56/217) Comment: High attrition rate
Selective reporting (reporting bias)	Low risk	Comment: Trial registration or published protocol not identified, but the outcomes specified in the methodology section have been reported in the Results
Other bias	Low risk	No other potential sources of bias identified

DeGier 2016.

Methods	Cluster‐randomised controlled trial
Participants	Conducted in Cambodia. The clusters were 16 primary schools in rural Kampong Speu province, of which 4 were randomly selected for each study group (n = 1977 children). Schools were eligible if they participated in the World Food Program school meal programme and all children were served breakfast daily
Interventions	Intervention: Children received 1 of 3 types of fortified rice or placebo (unfortified white rice) 6 days a week for 6 months: UltraRice_original (n = 479), UltraRice_improved (n = 500), NutriRice (n = 506) Control: Placebo (n = 492) Food vehicle: 3 types of fortified rice Dose: UltraRice original: Iron 10.67 mg, zinc 3 mg, vitamin B1 1.1 mg, folate 0.2 mg; UltraRice improved: retinol 0.64 mg, iron 7.55 mg, zinc 2.0 mg, vitamin B1 1.4 mg, vitamin B3 12 mg, folate 0.3 mg, vitamin B12 0.004; NutriRice: retinol 0.29 mg, iron 7.46 mg, zinc 3.7 mg, vitamin B1 0.7 mg, vitamin B3 8 mg, vitamin B6 0.92 mg, folate 0.1 mg, vitamin B12 0.001 mg Duration: 6 months Additional interventions: After baseline data collection, all children received a single dose of 500 mg mebendazole
Outcomes	Hookworm infection risk; cognitive outcomes
Notes	The research described received funding from United States Department of Agriculture/FAS through a grant (FFE‐442‐2012/038‐ 00, 10.608) to PATH and internal funding from WFP (through the WFP/DSM partnership) and IRD Study duration: November 2012 to June 2013.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Quote: "Three different randomizations, combining different schools to one intervention, were separately generated based on a list number of children per school by iteration to fit the predefined criteria of group size (within 10% of the mean)." Comment: Adequately done
Allocation concealment (selection bias)	Low risk	Quote: "A researcher not involved in the field work (MAD) blindly picked one of the three randomizations, and allocated each group of schools to an intervention arm." Comment: Adequately done
Blinding of participants and personnel (performance bias) All outcomes	Low risk	Quote: "The entire research team and all participants and caregivers were blinded to the allocation. The code was only known to one person with WFP, responsible to allocate the correct type of rice to the right school." Comment: Adequately done
Blinding of outcome assessment (detection bias) All outcomes	Low risk	Quote: "The entire research team and all participants and caregivers were blinded to the allocation. The code was only known to one person with WFP, responsible to allocate the correct type of rice to the right school." Comment: Adequately done
Incomplete outcome data (attrition bias) All outcomes	High risk	Placebo group: 184/492 UltraRice_original: 123/479 UltraRice_improved: 188/500, NutriRice: 281/506 Loss to follow‐up ranges from 26% to 44% per group Comment: High attrition rate
Selective reporting (reporting bias)	High risk	The trial is registered on ClinicalTrials.gov NCT01706419. The study was powered for its primary outcomes (micronutrient status), which are not reported here
Other bias	Low risk	Comment: No additional biases identified

Economos 2014.

Methods	Randomised controlled trial
Participants	176 healthy children (aged 6 ‐ 10 years) were recruited to participate in a 12‐week double‐ blind, randomised controlled trial at Boston University Medical Center and Tufts Medical Center in Boston, MA. In January through June 2005 and 2006, children were recruited from the hospital paediatric clinics and through local print and online classified advertisements Exclusion criteria used to screen potential participants included a history of rickets, diabetes, intestinal malabsorption (i.e. cystic fibrosis, fat malabsorption syndrome, or Crohn’s disease) or severe medical illness, including renal failure; allergies to orange juice; any medical conditions precluding daily consumption of orange juice; currently taking, or having taken < 1 month before start of study, a prescription vitamin D supplement
Interventions	All 3 intervention groups consumed 2 x 240‐mL (16 oz) glasses of juice a day. Total daily intake of micronutrients by study group was as follows: CaD (n = 61) received 700 mg calcium and 200 IU vitamin D; CaDEA (n = 51) received 700 mg calcium, 200 IU vitamin D, 12 IU vitamin E, and 2000 IU vitamin A as beta carotene; Calcium‐only group (Ca) (n = 64) received 700 mg calcium a day. Orange juice preparations were isocaloric and provided 110 kcal/240 mL for a total contribution of 220 kcal/day. Orange juice was home‐delivered every 2 weeks, and log sheets of deliveries were maintained. Study participants were instructed to drink 2 x 240‐mL glasses of orange juice a day using a re‐useable cup holding 8 oz (240 mL) to measure juice Duration: 12 weeks The interventions arms eligible for the review were CaDEA (2) and Ca alone (3).
Outcomes	Calcium, phosphorous, albumin, alkaline phosphatase, 25‐hydroxyvitamin D, parathyroid hormone, retinol, a‐tocopherol
Notes	Sponsored by The Beverage Institute for Health & Wellness, The Coca Cola Company, Atlanta, GA Study duration: January through June during 2005 and 2006.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Quote: "Participants were randomised by a computer‐generated code into one of the three beverage intervention groups." Comment: Adequately done
Allocation concealment (selection bias)	Unclear risk	Insufficient information to permit judgement
Blinding of participants and personnel (performance bias) All outcomes	Low risk	Quote: "Beverages were blind‐packaged in colour and number coded containers by the manufacturer (Minute Maid)." Comment: Adequately done
Blinding of outcome assessment (detection bias) All outcomes	Low risk	Quote: "Beverages were blind‐packaged in colour and number coded containers by the manufacturer (Minute Maid)." Comment: Adequately done
Incomplete outcome data (attrition bias) All outcomes	High risk	CaD 8/61; CaDEA 11/51; Calcium‐only group 16/64 Total loss to follow‐up 25% Comment: High attrition rate
Selective reporting (reporting bias)	Low risk	Comment: Trial registration or published protocol not identified, but the outcomes specified in the methodology section have been reported in the Results
Other bias	Low risk	No additional bias identified

Faber 2005.

Methods	Randomised controlled trial
Participants	The study area was in The Valley of a Thousand Hills in KwaZulu‐Natal province, South Africa. All eligible infants (n = 361) who were aged 6 – 12 months at baseline were asked to participate
Interventions	Intervention (n = 180):The finely milled maize meal was fortified to supply 3 mg carotene, 11 mg iron (ferrous fumarate), and 3 mg zinc (zinc sulfate) per 40 g dry product, Ascorbic acid (sodium ascorbate) was added (56 mg/40 g dry product) to enhance iron absorption. The maize meal was further fortified with certain nutrients that are limited in the diet of South African children, so that it supplied 110 g copper, 10 g selenium, 0.4 mg riboflavin, 0.15 mg vitamin B6, 0.25 g vitamin B12, and 2.5 mg vitamin E per 40 g dry product Control (n = 181): Same porridge with no added nutrients Dose: The mothers helped to identify a suitable portion size, which was set at 20 g dry product, mixed with 125 mL milk or water. The dry product was packed in individual 25 g colour‐coded sachets; the additional 5 g/sachet allowed for spillage and the mother’s tasting. An intake of 2 sachets a day was recommended, consumed as either 1 or 2 meals Food vehicle: Porridge Duration: 6 months
Outcomes	Motor development, anthropometrics (weight, length, LAZ, WAZ, WLZ), serum haemoglobin, ferritin, retinol and zinc, CRP
Notes	Supported by the Thrasher Research Fund and the Community‐based Health Programme of The Valley Trust. Tiger Food Brands Limited donated the fortified‐porridge product Study duration: First phase: February to August 2002 Second phase: September 2002 to March 2003.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Quote: "The allocation schedule was generated in blocks of 8 by the drawing of a sticker from a container that contained 4 yellow and 4 green stickers. Infants were randomly assigned in the order that they completed the baseline survey." Comment: Adequately done
Allocation concealment (selection bias)	Unclear risk	"Color‐coding was used to distinguish between the 2 treatment groups. The project leader was aware of which porridge each of the groups was receiving, because the fortified porridge had a slight yellow colour due to the carotene used as fortificant." Comment: Cannot ascertain if this affected outcome
Blinding of participants and personnel (performance bias) All outcomes	Unclear risk	Quote: "The mothers and community health workers were not aware of which porridge was fortified. All baseline and postintervention measurements were done in a blinded manner." "The project leader was aware of which porridge each of the groups was receiving, because the fortified porridge had a slight yellow colour due to the ‐carotene used as fortificant." Comment: Participant blinding conducted adequately, but cannot ascertain if personnel blinding was adequate
Blinding of outcome assessment (detection bias) All outcomes	Unclear risk	Quote: "All baseline and postintervention measurements were done in a blinded manner." "The project leader was aware of which porridge each of the groups was receiving, because the fortified porridge had a slight yellow colour due to the carotene used as fortificant." Comment: Cannot ascertain if personnel blinding was adequate
Incomplete outcome data (attrition bias) All outcomes	Low risk	Intervention group: 36/180 Control group: 36/181 Overall 19.12% loss to follow‐up Comment: Low attrition rate
Selective reporting (reporting bias)	Low risk	Comment: Trial registration or published protocol not identified, but the outcomes specified in the methodology section have been reported in the Results
Other bias	Low risk	No additional bias identified

Gibson 2011.

Methods	Randomised controlled trial
Participants	This 12‐month RCT was conducted on 6‐month‐old Zambian infants (n = 743) living in Chilenje, a middle‐income area of Lusaka, Zambia. The study was conducted from October 2005 to July 2009. Eligible infants aged 6 months ± 2 weeks whose parents or guardians gave written informed consent were randomly assigned to receive either a richly micronutrient‐fortified porridge or a basal porridge
Interventions	Intervention (n = 373): Caregivers received individual instructions by the project nutritionist about how to prepare and cook the porridge according to the package directions and how to feed the Chilenje porridge in place of other porridges. They were each supplied with a spoon (5 mL) to measure the amount of porridge flour (i.e. 7 level spoonfuls of dry flour, ˜ 30 g dry flour), and a plastic feeding cup graduated in millilitres to measure the volume of water required (i.e. 250 mL). The slurry was cooked (5 – 10 minutes) and transferred into the graduated plastic feeding cup, which was then used to feed the child. The volume of porridge fed to the child was noted by the caregiver at the end of each feeding (in millilitres). Porridges prepared according to the package directions had an energy density of 0.76 kcal/g (1 kcal = 4.18 kJ), 16% energy from protein, and an analysed phytate content of 5.8 g/kg (dry weight) Control (n = 370): Placebo Food vehicle: Porridge Dose: Vitamin A, retinol equivalents 6.5 ug/kg, vitamin C, 2 g/kg, cholecalciferol, 0.1 mg/kg, thiamine (mononitrate), 9 mg/kg, riboflavin, 11.2 mg/kg, niacin (niacinamide), 140 mg/kg, pyridoxine (HCl), 8.6 mg/kg, folate, 2.21 mg/kg, vitamin B12, 9.75 mg/kg, pantothenic acid, 40.3 mg/kg, iron (ferrous fumarate), 250 mg/kg, zinc (oxide), 200 mg/kg, copper (gluconate), 3.2 mg/kg, manganese (sulphate monohydrate), 12 mg/kg, selenium (sodium selenite) 0.2 mg/kg, Calcium 6.8 g/kg, Phosphorous 5.3 g/kg, Magnesium (oxide) 943 mg/kg. Duration: 1 year Additional interventions: All infants in the RCT were supplemented with vitamin A capsules at their 6‐, 12‐, and 18‐month clinic visits, according to the standard protocol of care, through the government national vitamin A supplementation programme
Outcomes	Anaemia, iron deficiency, iron deficiency anaemia, zinc deficiency
Notes	Supported by the Bill and Melinda Gates Foundation. Micronutrients were provided by DSM Nutritional Products, Isando, South Africa Study duration: October 2005 to July 2009
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Quote: " Eligible infants aged 6 months ± 2 wk whose parents or guardians gave written informed consent were randomly assigned to receive either a richly micronutrient‐fortified porridge or a basal porridge using a block randomisation scheme, with a block length of 20." Comment: Adequately done
Allocation concealment (selection bias)	Unclear risk	Insufficient information to permit judgement
Blinding of participants and personnel (performance bias) All outcomes	Low risk	Quote: "The investigators, clinic staff, outcomes assessors, and participants were unaware of the intervention assignment and knowledge of treatment groups became known only after the database was finalized. An exit questionnaire indicated that the blinding was effective." Comment: Adequately done
Blinding of outcome assessment (detection bias) All outcomes	Low risk	Quote: "The investigators, clinic staff, outcomes assessors, and participants were unaware of the intervention assignment and knowledge of treatment groups became known only after the database was finalized. An exit questionnaire indicated that the blinding was effective." Comment: Adequately done
Incomplete outcome data (attrition bias) All outcomes	High risk	Intervention: 83/373 Control: 84/370 Loss to follow‐up of 23.3% and 23.7% in control and intervention groups respectively Comment: High attrition rate
Selective reporting (reporting bias)	Low risk	This trial was registered at the UK National Institute for Health Research, Current Controlled Trials, ISRCTN Register (www.controlled‐trials.com/mrct/trial/835053/ISRCTN37460449) as ISRCTN37460449
Other bias	Low risk	No additional bias identified.

Hieu 2012.

Methods	Randomised controlled trial
Participants	School children (n = 403) aged 6 – 9 years in grade 1 ‐ 3 of 5 primary schools were recruited. The schools were located in 3 communes of 2 districts (Bac Tra My and Tien Phuoc) of Quang Nam province, 900 km south of Hanoi, Vietnam, where micronutrient deficiencies were known to exist. The schools were selected based on their proximity to the general hospital of Tam Ki so that blood samples could be processed within 4 hours of collection
Interventions	Intervention: The treatment groups were as follows: Daily fortified biscuit group and placebo tablet once a week (FB); Daily non‐fortified biscuits and placebo tablet once a week (control group, C); Fe tablet once a week and daily non‐fortified biscuits (weekly Fe pharmaceutical supplementation group, SUP). Biscuits were distributed for 6 months during the break time (09.00 – 09.30 hours), 5 days a week excluding school holidays, weekends and public holidays Control: Placebo and Iron supplement Food vehicle: Biscuit Dose: A daily ration of 5 biscuits (approximately 30 g) covered 50% of the RNI of a 9‐year‐old child for vitamin A (all‐transretinol), Fe (iron fumarate), Zn (zinc sulphate) and iodine, 40 % of the requirements of Cu, vitamin C, thiamin, riboflavin, vitamins B6, B12, E and niacin, 35% of the requirements of Mg, 20% of the requirements of Ca, vitamin D and folate and 7% of the requirements of Mn, Se, K, chloride, Na, fluoride, pantothenic acid, vitamin K and biotin Duration: 6 months Additional interventions: All the children were de‐wormed by the health services of the Quang Nam province with mebendazole (500 mg) a few days after the start of the study The intervention arms eligible for this review were FB and C
Outcomes	Serum haemoglobin, transferrin receptor, ferritin, retinol, zinc, body iron
Notes	Supported by Decentralized French co‐operation, Sight and Life and IRD Study duration: November 2005 to May 2006
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Quote: "From an alphabetical name list of the children attending the five selected schools, children were randomly assigned into three treatment groups using a computer‐generated random list and all children from all schools were allocated to the three groups." Comment: Adequately done
Allocation concealment (selection bias)	Low risk	Quote: "All field staff and researchers as well as teachers were blinded for the group allocation that was kept in a sealed envelope at the NIN until the end of data analysis." Comment: Adequately done
Blinding of participants and personnel (performance bias) All outcomes	Low risk	Quote: "All field staff and researchers as well as teachers were blinded for the group allocation that was kept in a sealed envelope at the NIN until the end of data analysis." Comment: Adequately done
Blinding of outcome assessment (detection bias) All outcomes	Low risk	Quote: "All field staff and researchers as well as teachers were blinded for the group allocation that was kept in a sealed envelope at the NIN until the end of data analysis." Comment: Adequately done
Incomplete outcome data (attrition bias) All outcomes	High risk	Overall loss to follow‐up: 106/403 Total loss to follow‐up 26.3%. Comment: High attrition rate
Selective reporting (reporting bias)	Low risk	Comment: Trial registration or published protocol not identified, but the outcomes specified in the methodology section have been reported in the Results
Other bias	Low risk	No additional bias identified

Hyder 2007.

Methods	Randomised controlled trial
Participants	Conducted in 54 non‐formal primary education (NFPE) schools operated by the Bangladesh Rural Advancement Committee (BRAC, one of the largest national non‐governmental organisations in the world) in Sherpur district, 300 km northeast of Dhaka city. Participants were adolescent girls (n = 1125). Adolescent boys attending the schools (around 30% of the students) were included in the randomisation process and were provided the same beverages to avoid sharing. However, they were not included in any aspect of the data collection or analysis
Interventions	Intervention (n = 559): Fortified orange‐flavoured powdered beverage. The contents of 2 sachets, which contained 90 g powder, were dissolved in 1000 mL of tube‐well water. Each student received 200 mL of the reconstituted fortified or non‐fortified beverage daily Control (n = 566): equal quantity of a non‐fortified orange‐flavoured powdered beverage (identical to the fortified beverage in terms of weight, colour, flavour, and appearance) as a control Dose: micronutrient‐fortified powder in 1 serving (200 mL): Iron, mg 7.0, vitamin A, IU (RE) 1296 (389), iodine, mg 75, zinc, mg 7.5, vitamin C, mg 120, riboflavin, mg 0.91, folic acid, mg 120, vitamin B12, mg 1.0, vitamin B6, mg 1.0, vitamin E, mg 10, niacin, mg 5.0 Food Vehicle: Powdered orange‐flavoured beverage Duration: 6 days a week for 12 months
Outcomes	Haemoglobin and serum levels of ferritin, retinol, zinc, and CRP. Anthropometric measurements including height, weight, and MUAC
Notes	Supported by the Micronutrient Initiative, Ottawa, Canada. Supplement was provided by Procter & Gamble Study duration: Not specified.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Quote: "Randomization was done by listing all selected children, assigning them with random numbers, and dividing the odd numbers from the even numbers to form the 2 groups." Comment: Insufficient information to permit judgement
Allocation concealment (selection bias)	Unclear risk	Comment: Insufficient information to permit judgement
Blinding of participants and personnel (performance bias) All outcomes	Low risk	Quote:"The sachets containing the fortified and non‐fortified beverage differed only in the sachet’s colour (blue or yellow). Researchers, school teachers, shastho shebikas (BRAC community health workers), and students did not know whether the blue or yellow Coloured sachets contained the fortified beverage." "The decoding was done only by the manufacturer after the study was completed and the data analysed." Comment: Adequately done
Blinding of outcome assessment (detection bias) All outcomes	Low risk	Quote:"The sachets containing the fortified and non‐fortified beverage differed only in the sachet’s colour (blue or yellow). Researchers, schoolteachers, shastho shebikas (BRAC community health workers), and students did not know whether the blue or yellow Coloured sachets contained the fortified beverage." "The decoding was done only by the manufacturer after the study was completed and the data analysed." Comment: Adequately done
Incomplete outcome data (attrition bias) All outcomes	Low risk	Intervention group: 77/559 Control group: 59/566 12.1% total loss to follow‐up Comment: Low attrition rate unlikely to affect results
Selective reporting (reporting bias)	Low risk	Comment: Trial registration or published protocol not identified, but the outcomes specified in the methodology section have been reported in the Results
Other bias	Low risk	No additional bias identified

Jinabhai 2001.

Methods	A community‐based double‐blind, randomised, placebo‐controlled trial in 11 rural South African primary schools in 1995
Participants	The study randomly allocated 579 children aged between 8 and 10 years into 6 study groups
Interventions	The intervention groups were managed as follows: Group 1: de‐worming, biscuits fortified with a combination of micronutrients (vitamin A and iron) and other nutrients (defined below); Group 2: de‐worming, biscuits fortified with vitamin A; Group 3: de‐worming, non‐fortified biscuits (no micronutrients); Group 4: not de‐wormed, biscuits fortified with vitamin A, iron and other nutrients; Group 5: not de‐wormed, biscuits fortified with vitamin A; Group 6: not de‐wormed, non‐fortified biscuits. The 2 biscuits given daily to groups 1 and 4 contained vitamin B (25% RDA, 0.25 mg), vitamin A (50% RD, 350 J.tg), iron in the form of FeEDTA (50% RDA, 5 mg), calcium (25% RDA, 200 mg) and zinc (25% RDA, 2.5 mg) The 2 biscuits given daily to groups 2 and 5 supplied 100% (700 J.tg) RDA vitamin A Duration: 4 months For this review, we have included data from Group 4 and Group 6 only
Outcomes	Micronutrient status (serum retinol, haemoglobin, hematocrit, serum ferritin, serum iron and percentage transferrin saturation); helminthic infections (prevalence and intensity); nutritional status (weight, height and knee‐heel length were measured, and serum albumin levels were assessed) and scholastic and cognitive tests
Notes	This study was supported by a research grant from the Human Sciences Research Council, the British Council supported the UK academic Iink, the fortified biscuits were supplied by SASKO, the albendazole tablets by Smith Kline Beecham and the praziquantel tablets by Bayer Study duration: 4 months in 1995
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Quote: "The subjects were randomly allocated into six study groups." Comment: Insufficient information
Allocation concealment (selection bias)	Unclear risk	Comment: Insufficient information
Blinding of participants and personnel (performance bias) All outcomes	Unclear risk	Quote: "A community‐based, double‐blind, randomised, placebo‐controlled trial" Comment: Insufficient information
Blinding of outcome assessment (detection bias) All outcomes	Unclear risk	Quote: "A community‐based, double‐blind, randomised, placebo‐controlled trial" Comment: Insufficient information
Incomplete outcome data (attrition bias) All outcomes	Unclear risk	Comment: The CONSORT flow diagram was not provided
Selective reporting (reporting bias)	Low risk	No information on trial registration provided. Outcomes specified in the methods section have been reported in the Results
Other bias	Low risk	Comment: No additional bias identified

Järvenpaa 2007.

Methods	Randomised controlled trial
Participants	Conducted in Finland. 72 pregnant women from the city of Oulu were recruited for the study
Interventions	Intervention (n = 40): During the 8‐week intervention period, the women followed their habitual diet, except that 1000 mL of their daily liquids were replaced by fortified or normal mineral water Control (n = 32): Placebo Food vehicle: Fortified mineral water Dose: Potassium (mg) 141, magnesium (mg) 53, calcium (mg) 800, sodium (mg) 6, vitamin B6 (mg) 1.5, vitamin B12 (mg) 2.1, folic acid (mg) 470, vitamin D (mg) 5.0 Duration: 8 weeks
Outcomes	Serum folate, vitamin B12, erythrocyte folate, plasma homocysteine
Notes	Supported by Olvi PLC, Iisalmi, Finland Study duration: Not specified.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Insufficient information to permit judgement
Allocation concealment (selection bias)	Unclear risk	Insufficient information to permit judgement
Blinding of participants and personnel (performance bias) All outcomes	Unclear risk	Quote: "A randomised, placebo‐controlled, double‐blind parallel study design was used." Comment: Insufficient information on how blinding was performed
Blinding of outcome assessment (detection bias) All outcomes	Unclear risk	Quote: "A randomised, placebo‐controlled, double‐blind parallel study design was used." Comment: Insufficient information on how blinding was performed
Incomplete outcome data (attrition bias) All outcomes	Low risk	Overall loss to follow‐up: 8/74 8.4% loss to follow‐up Comment: Low attrition rate
Selective reporting (reporting bias)	Low risk	Trial registration or published protocol not identified, but the outcomes specified in the methodology section have been reported in the Results
Other bias	Low risk	No additional bias identified

Liu 1993.

Methods	Cluster‐randomised controlled trial
Participants	Conducted in China from February 1990 to June 1990 on healthy full‐term infants (n = 226) born without complication and with birth weights > 2.5 kg aged 6 ‐ 13 months at the outset
Interventions	Intervention (n = 77): Fortified rusks were either eaten dry or were taken in liquid form. after dispersion in water Control (n = 87): Unfortified rusk Dose: wheat flour, sugar and vegetable oil. MMN per rusk (17g): calcium 300 mg, iron 5 mg, zinc 3 mg, vitamin A 224 ug, vitamin D 4 ug, thiamine 0.15 ug, riboflavin 0.2 mg, niacin 2.5 mg, cyanocobalamin 0.3 ug, folic acid 25 ug Food vehicle: Rusk Duration: 3 months
Outcomes	Anthropometric measurements (body weight, length), a clinical examination, blood samples (free erythrocyte porphyrin, plasma ferritin, erythrocyte glutathione reductase activation coeff, vitamin E and retinol), and diet histories (24‐hour recall)
Notes	Supported in part by a grant from the United Kingdom Department of Trade and Industry Study duration: February 1990 to June 1990
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Insufficient information provided to permit judgement
Allocation concealment (selection bias)	Unclear risk	Insufficient information provided to permit judgement
Blinding of participants and personnel (performance bias) All outcomes	Unclear risk	Insufficient information provided to permit judgement
Blinding of outcome assessment (detection bias) All outcomes	Unclear risk	Insufficient information provided to permit judgement
Incomplete outcome data (attrition bias) All outcomes	High risk	Total loss to follow‐up was 24.9% Comment: High attrition rate may affect outcomes
Selective reporting (reporting bias)	Low risk	No information on trial registration provided. Outcomes specified in the methods section have been reported in the Results
Other bias	Low risk	No additional bias identified (i) recruitment bias: Low risk (ii) baseline imbalance: Low risk (iii) loss of clusters: Low risk (iv) incorrect analysis: Low risk (v) comparability with individually randomised trials: Low risk

Lopriore 2004.

Methods	Randomised controlled trial
Participants	Conducted between May 1998 and January 1999 at Saharawi refugee camps near the town of Tindouf in southwest Algeria, on children (n = 374) aged 3 – 6 years with height‐for‐age z scores (HAZ) ≤ −2.0 with use of World Health Organization and National Center for Health Statistics (WHO/NCHS) reference median were eligible
Interventions	Children were assigned to 1 of 5 groups: Fortified spread (FS) (n = 75) Fortified spread plus metronidazole (FSM) (n = 75) Unfortified spread (US) (n = 75) Unfortified spread plus metronidazole (USM) (n = 75) Control (n = 74) Dose: per 100 g; Calcium (mg) 1000, potassium (mg) 1134, phosphorus (mg) 635, magnesium (mg) 156, iron (mg) 42, zinc (mg) 41, copper (mg) 2, vitamin A (ug) 2000, vitamin D (ug) 50, vitamin E (mg) 20, vitamin C (mg) 125, vitamin B1 (mg) 4, vitamin B2 (mg) 4, vitamin B6 (mg) 4, vitamin B12 (ug) 4, folate (ug) 500, pantothenic acid (mg) 25, niacin (mg) 50 Food vehicle: Spread Duration: 6 months Additional: metronidazole or mebendazole treatment For this review, we have merged data from FS and FSM groups as the intervention group, and have merged data from US and USM groups as the control group
Outcomes	Growth: knee‐heel length, weight, height, WAZ, WHZ, HAZ, stunting, underweight and wasting, haemoglobin levels, anaemia, morbidity, faecal macroscopy and egg counts
Notes	Supported by the Italian nongovernmental organization Comitato Internazionale per lo Sviluppo dei Popoli (CISP) as part of a grant from the European Commission Humanitarian Office (ECHO) Study duration: May 1998 and January 1999
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Quote: "With use of a simple computer‐generated randomisation method." Comment: Adequately done
Allocation concealment (selection bias)	Unclear risk	Insufficient information provided to permit judgement
Blinding of participants and personnel (performance bias) All outcomes	Low risk	Quote: "Neither the field assistants nor the investigator was aware of group assignment. The codes were revealed only after all subjects had completed the trial." "The supplements were colour coded at production, and the key revealing the code was kept by the manufacturer in France." Comment: Adequately done
Blinding of outcome assessment (detection bias) All outcomes	Low risk	Quote: "Neither the field assistants nor the investigator was aware of group assignment. The codes were revealed only after all subjects had completed the trial." Comment: Adequately done
Incomplete outcome data (attrition bias) All outcomes	High risk	Fortified spread (FS) (24/75) Fortified spread plus metronidazole (FSM) (23/75) Unfortified spread (US) (23/75) Unfortified spread plus metronidazole (USM) (21/75) Control (29/74) Total loss to follow‐up 32.1% Comment: High attrition rate may affect outcome
Selective reporting (reporting bias)	Low risk	Comment: No information on trial registration provided. Outcomes specified in the methods section have been reported in the Results
Other bias	Low risk	No additional bias identified

Mardones 2007.

Methods	A non‐blinded, controlled before‐after study conducted among pregnant women at 19 antenatal clinics and delivered at 2 maternity hospitals in Santiago, Chile between May 2002 and February 2003
Participants	970 pregnant women
Interventions	Pregnant women were assigned to receive regular powdered milk (n = 477) or a milk product fortified with multiple micronutrients and omega‐3 fatty acids (n = 493). Women in the experimental group received 2 kg per month of powdered milk (Mamans or product M, produced by Parmalat SpA, Parma, Italy), fortified with multiple micronutrients: Energy (kcal): 521.0 Protein (g): 25.0 Fats (g): 21.0 Milk fat: 10.5 Vegetable fat: 10.5 Polyunsaturated fatty acids: 5.3 Omega‐3 fatty acids: 0.9 Omega‐6 fatty acids" 4.4 Carbohydrates (g): 58 Lactose: 3 Vitamins A (mg): 1200, thiamine (B1) (mg): 1.0, riboflavin (B2) (mg): 1.0, pyridoxine (B6) (mg): 2.0, B12 (mg): 1.5, C (mg): 110, D3 (mg): 15, E (mg): 45, niacin (PP) (mg): 10, biotin (mg): 45, folic acid (mg): 600, Ca (mg): 960, P (mg): 720, Mg (mg): 90, Zn (mg): 12, Fe (mg): 27, Bioavailable Fe (mg): 4.5, Se (mg): 15
Outcomes	Maternal anthropometry, birthweight, duration of gestation, infant length, infant head circumference, preterm birth, low birthweight
Notes	Supported by Parmalat SpA, Italy, the company that provided the Maman product. Study duration: May 2002 and February 2003
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	High risk	Quote: "A total of 1173 women were considered eligible, and they were recruited and randomised." Comment: No details of randomisation provided.
Allocation concealment (selection bias)	High risk	Quote: "A total of 1173 women were considered eligible, and they were recruited and randomised." Comment: No details of allocation provided.
Blinding of participants and personnel (performance bias) All outcomes	High risk	Quote: "Non‐blinded, randomised controlled study" Comment: Not done
Blinding of outcome assessment (detection bias) All outcomes	High risk	Quote: "Non‐blinded, randomised controlled study" Comment: Not done
Incomplete outcome data (attrition bias) All outcomes	High risk	Comment: Fortified milk group: 224/589 Regular powdered milk group: 219/552 Overall 41.4% attrition rate
Selective reporting (reporting bias)	Low risk	Comment: No information on trial registration provided. Outcomes specified in the methods section have been reported in the results
Other bias	Unclear risk	Baseline outcome measurements: Quote: "Selected baseline biological and social variables were similar between the control and the experimental groups with the exception of gestational age at recruitment, which was slightly higher in group M (intervention group)." Comment: Low risk Baseline characteristics: Quote: "Selected baseline biological and social variables were similar between the control and the experimental groups with the exception of gestational age at recruitment, which was slightly higher in group M (intervention group)." Comment: Low risk Protection against contamination: Low risk

Nesamvuni 2005.

Methods	Randomised controlled trial
Participants	Using the creche and clinic as entry points into the community in Oukasie, Brits, in the North West Province of South Africa, all 1 – 3‐year‐old children (n = 60) at the creches and the well‐baby clinic were screened and the first 60 undernourished children who had weight‐for‐age or height‐for‐age below the 5th percentile of the National Center for Health Statistics (NCHS) reference identified
Interventions	Intervention: undernourished 1 – 3‐year‐old children and their households were randomly allocated to either an experimental (n = 30): or control group (n = 30). The households (families) in the experimental group received a vitamin‐fortified maize meal and those in the control group unfortified maize meal. Between 25 and 50 kg (depending on usual monthly consumption) of maize meal flour was provided to the families per month to replace all maize meal consumed by these households Control: Unfortified maize meal Dose: 1700 IU vitamin A, 0.61 mg thiamine, 0.62 mg riboflavin and 0.56mg pyridoxine Food vehicle: Maize meal porridge Duration: 12 months
Outcomes	Weight, height, haemoglobin, hematocrit, serum retinol, serum RBP
Notes	The study was funded by grants from the National Research Foundation, Potchefstroom University for Christian Higher Education, Hoffman La Roche (Switzerland), Roche Vitamin and Fine Chemicals and a gift of maize from Maizecor. Study duration: Not specified.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Insufficient information to permit judgement
Allocation concealment (selection bias)	Unclear risk	Insufficient information to permit judgement
Blinding of participants and personnel (performance bias) All outcomes	Unclear risk	Quote: "The study design was a randomised, parallel, single‐blind intervention (families were blinded)" Comment: No blinding of personnel
Blinding of outcome assessment (detection bias) All outcomes	High risk	Quote: "The study design was a randomised, parallel, single‐blind intervention (families were blinded)." Comment: Not done
Incomplete outcome data (attrition bias) All outcomes	High risk	Intervention group: 9/30 Control group: 7/30 Comment: High attrition rate
Selective reporting (reporting bias)	Low risk	Comment: Trial registration not specified. Outcomes specified in the methods section have been reported in the Results
Other bias	Low risk	No additional bias identified

Nga 2009.

Methods	Randomised controlled trial
Participants	Conducted in Vietnam. Pupils were recruited from 2 schools that had been selected on the basis of a high prevalence of anaemia and parasite infestations among school children during an earlier survey. In total, 642 children aged 6 – 8 years in 20 classes were available, of which 510 children were randomly selected
Interventions	Intervention: The 4 intervention groups were: Non‐fortified biscuit plus placebo de‐worming treatment (placebo) (n = 128); Multi‐micronutrient‐fortified biscuit plus placebo de‐worming treatment (MMF) (n = 128); Non‐fortified biscuit plus de‐worming treatment with albendazole (Alb) (n = 127); Multi‐micronutrient–fortified biscuits plus de‐worming treatment with Alb (MMF+Alb) (n = 127). Food vehicle: Fortified biscuit Dose: Iron (ferrous fumarate), 6 mg, zinc (zinc sulfate), 5.6 mg, iodine (potassium iodide), 35 ug, vitamin A (retinyl acetate), 300 ug RE, thiamine (thiamine mononitrate), 1 mg, riboflavin, 0.9 mg, vitamin B6, 1.1 mg, niacin (niacinamid), 10.5 mg NE, vitamin B12, 1.5 ug, folic acid, 120 ug, vitamin C, 28.4 mg, calcium (CaHPO4), 150 mg, cholecalciferol, 74 ug, magnesium, 40 mg, selenium (sodium salt), 6.8 ug, potassium (citrate), 378 mg, phosphorus, 70 mg, pantothenic acid, 3 mg, vitamin E, 2.8 ug, vitamin K, 10 ug, biotin (D‐biotin), 18 ug Duration: 6 months Additional interventions: A single dose of intestinal anthelminthic treatment as orange‐flavoured chewable tablets containing 400 mg albendazole (Vidoca) or identical placebo tablet was given. De‐worming with albendazole was given to all children at the end of the study after the final stool sample collection. The intervention arms eligible for this review were MMF and the placebo groups
Outcomes	Haemoglobin, serum ferritin, transferrin receptor, retinol. zinc, body iron
Notes	Supported by the Neys‐van Hoogstraten Foundation, The Netherlands, and Ellison Medical Foundation Study duration: January to June 2007
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Quote: "510 pupils were allocated to 1 of the 4 intervention groups based on a computer‐generated list, matched on age (12‐mo age groups) and sex, and using a block size of 8 by one of the researchers not involved in the field work." Comment: Adequately done
Allocation concealment (selection bias)	Unclear risk	Insufficient information to permit judgement
Blinding of participants and personnel (performance bias) All outcomes	Low risk	Quote: "The codes of fortified and non‐fortified biscuits, Alb, and placebo were kept by the manufacturers and by a member of the institute staff not directly involved in the study until the data analysis was finished." Comment: Adequately done
Blinding of outcome assessment (detection bias) All outcomes	Low risk	Quote: "The code of fortified and non‐fortified biscuits, Alb, and placebo were kept by the manufacturers and by a member of the institute staff not directly involved in the study until the data analysis was finished." Comment: Adequately done
Incomplete outcome data (attrition bias) All outcomes	Low risk	Non‐fortified biscuit plus placebo de‐worming treatment (placebo) (10/128); Multi‐micronutrient‐fortified biscuit plus placebo de‐worming treatment (MMF) (14/128); Non‐fortified biscuit plus de‐worming treatment with albendazole (Alb) (10/127); or Multi‐micronutrient–fortified biscuits plus de‐worming treatment with Alb (MMF+Alb) (10/127). 8.6% total loss to follow‐up Comment: Low attrition rate
Selective reporting (reporting bias)	Low risk	Trial registration not specified. Outcomes specified in the methods section have been reported in the Results
Other bias	Low risk	No additional bias identified

Oelofse 2003.

Methods	A community‐based randomised trial conducted among 6‐ to 12‐month‐old infants from a black urban disadvantaged community in the Western Cape, South Africa between March 1999 and June 2000
Participants	60 children aged approximately 6 months were randomly selected from all mothers visiting the local clinic with their infants
Interventions	The experimental group (n = 30) received a micronutrient‐fortified complementary food throughout the 6‐month period, while the control group (n = 30) did not receive any complementary food, but continued their normal diet Energy (kJ): 1304 Protein (g): 12 Fat (g): 6 Carbohydrate (g): 54.8 Vitamin A (iu): 1200, vitamin C (mg): 40, vitamin B1 (mg): 0.64, vitamin B2 (mg): 0.24, niacin (mg): 3.2, calcium (mg): 368, iron (mg): 8, vitamin D (iu): 160, vitamin E (iu): 4, biotin (mg): 20, folic acid (mg): 17.6, pantothenic acid (mg): 0.6, vitamin B12 (mg): 0.6, vitamin B6 (mg): 0.24, phosphorous (mg): 232, iodine (mg): 26, zinc (mg): 5.6, potassium (mg): 632, sodium (mg): 272, chloride (mg): 440
Outcomes	Serum retinol, iron, haemoglobin, zinc, weight, length, weight for age Z‐score, height for age Z‐score, weight for height Z‐score
Notes	No sample size calculations done a priori. Funding was not specified Study duration: March 1999 to June 2000
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Quote: "Each infant was randomly allocated to either an experimental or a control group." Comment: Insufficient information
Allocation concealment (selection bias)	Unclear risk	Quote: "Each infant was randomly allocated to either an experimental or a control group." Comment: Insufficient information
Blinding of participants and personnel (performance bias) All outcomes	High risk	Probably not done
Blinding of outcome assessment (detection bias) All outcomes	High risk	Probably not done
Incomplete outcome data (attrition bias) All outcomes	High risk	Intervention group: 14/30 Control group: 16/30 High attrition rates
Selective reporting (reporting bias)	Low risk	Comment: Trial registration not specified. Outcomes specified in the methods section have been reported in the Results
Other bias	Low risk	Comment: No additional bias identified

Osendarp 2007.

Methods	Randomised controlled trial
Participants	Conducted in Australia and Indonesia. The trials were conducted from August 2003 to April 2005 in children (n = 396) aged 6 – 10 years from South Australian government metropolitan schools of higher socio‐economic status in Adelaide and from schools in the central district of Jakarta of middle to low socio‐economic status (n = 384)
Interventions	Intervention: The studies in Australia and Indonesia both used a 2‐x‐2 factorial design in which the children were individually randomly allocated to 1 of 4 intervention groups: A mix of micronutrients (n = 106 + 94) N3 fatty acids (n = 96 + 97) Both (n = 92 + 98) Placebo (n = 102 + 95) A fruit‐flavoured drink (soy 0.6%) was used as the vehicle for all treatments, which were added as powders Food vehicle: Fruit‐flavoured drink Dose: Iron as NaFeEDTA 10 mg, zinc as zinc sulfate 5 mg, vitamin A as retinol acetate 400 ug, folate 150 ug, vitamin B6 1 mg, vitamin B12 1.5 ug, vitamin C 45 mg Duration: 1 year The intervention arms eligible for this review were Arms 1 and 4
Outcomes	Haemoglobin, serum ferritin, vitamin B12, zinc, transferrin receptor, body iron, general intelligence, verbal learning and memory, visual attention
Notes	Supported by Unilever Netherlands BV Study duration: August 2003 to April 2005
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Quote: "In Australia, the children were randomly assigned to intervention groups on entry in the study. In Indonesia, the children were stratified by school before being randomly assigned. Random assignment was done by means of a computer‐generated list in both countries." Comment: Adequately done
Allocation concealment (selection bias)	Unclear risk	Insufficient information to permit judgement
Blinding of participants and personnel (performance bias) All outcomes	Low risk	Quote: "The supplement powders were indistinguishable in colour and taste and were color‐coded. The codes remained unknown to both investigators and participants until the study was completed, all data had been entered, and initial analyses had been performed." Comment: Adequately done
Blinding of outcome assessment (detection bias) All outcomes	Low risk	Quote: "The supplement powders were indistinguishable in colour and taste and were color‐coded. The codes remained unknown to both investigators and participants until the study was completed, all data had been entered, and initial analyses had been performed." Comment: Adequately done
Incomplete outcome data (attrition bias) All outcomes	High risk	Australia 37%, Indonesia 7.1% loss to follow‐up Comment: High attrition rate for the study in Australia may have affected outcomes
Selective reporting (reporting bias)	Low risk	The trial is registered with the Netherlands Trials Registry as Trial N324 (NTR362). No evidence of selective reporting
Other bias	Low risk	No additional bias identified

Perignon 2016.

Methods	Cluster‐Randomised controlled trial
Participants	The study was conducted between November 2012 and July 2013 in 20 primary schools (n = 2440 children) from 5 districts of Kampong Speu province in Cambodia. Children attending the selected schools were eligible to be part of the study if they were 6 – 16 years of age, written informed consent was obtained from parent/caregiver, and the child did not have a mental or severe physical handicap. Children with severe anaemia (defined as haemoglobin concentration < 70 g/L) were excluded
Interventions	Intervention: The four intervention groups were: Fortified cold‐extruded rice UltraRice original formulation (URO) (n = 476); Fortified hot‐extruded rice UltraRice new formulation (URN) (n = 496); Fortified hot‐extruded rice Nutririce (n = 499); Non‐fortified rice (placebo) (n = 479); Control: (n = 490) Breakfast was distributed 6 days a week for 6 months Food vehicle: Fortified rice Dose: URO, URN, NutriRice Iron (mg) 10.67, 7.55, 7.46, zinc (mg) 3.04, 2.02, 3.68, vitamin B1 (mg) 1.06, 1.43, 0.69, folic acid (mg) 0.17, 0.28, 0.14, vitamin A (IU) 0, 2140, 960, vitamin B3 (mg) 0, 12.57, 7.98, vitamin B12 (µg) 0, 3.8, 1.26, vitamin B6 (mg) 0, 0, 0.92 Duration: 6 months Additional interventions: Children were de‐wormed using mebendazole just after the baseline and endline, according to the standard procedures of the Ministry of Health, Cambodia The intervention arms eligible for this review were Arms 3 and 4.
Outcomes	Haemoglobin, serum ferritin, transferrin receptor, body iron
Notes	Supported by USDA/FAS, WFP‐DSM consortium, and IRD Study duration: November 2012 to July 2013
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Quote: "The 16 selected schools were randomly allocated to one of the four intervention groups using a computer generated list with predefined criteria of group size." Comment: Adequately done
Allocation concealment (selection bias)	Low risk	Quote: "Rice was packaged in bags containing a letter (A‐H) according to allocation.." Comment: Adequately done
Blinding of participants and personnel (performance bias) All outcomes	Low risk	Quote: "Randomization was done by one of the researchers (M.A.D.) not involved in the field work and the codes were not known by any researchers or field staff during implementation, thus assuring the study was double‐blind." Comment: Adequately done
Blinding of outcome assessment (detection bias) All outcomes	Low risk	Quote: "Randomization was done by one of the researchers (M.A.D.) not involved in the field work and the codes were not known by any researchers or field staff during implementation, thus assuring the study was double‐blind." Comment: Adequately done
Incomplete outcome data (attrition bias) All outcomes	Low risk	Fortified cold‐extruded rice UltraRice original formulation (URO) 30/476; Fortified hot‐extruded rice UltraRice new formulation (URN) 32/496; Fortified hot‐extruded rice Nutririce 39/499; Non‐fortified rice (placebo) 3/479) Control: 39/490 7.9% total loss to follow‐up. Comment: Low attrition rate
Selective reporting (reporting bias)	Low risk	Comment: The trial was registered at ClinicalTrials.gov (Identifier: NCT01706419). No evidence of selective reporting
Other bias	Low risk	No additional bias identified. (i) recruitment bias: Low risk (ii) baseline imbalance: Low risk (iii) loss of clusters: Low risk (iv) incorrect analysis: Low risk (v) comparability with individually randomised trials: Low risk

Petrova 2019.

Methods	A randomised, controlled, double‐blind trial conducted among children at 3 schools in Granada, Spain
Participants	119 children aged 8 – 14 years
Interventions	Children were randomly allocated to a fortified milk group or a regular full‐milk control group for a duration of 5 months. Children in the Fortified group (n = 60) consumed 0.6 L/day of a fortified milk beverage containing vitamins (A, B complex, C, D and E), minerals (calcium, phosphorus, zinc), fish oils (with high levels of DHA and EPA), oleic acid, and carbohydrates (sugar and honey) (Puleva Ma®) Energy (Kcal/kJ): 69/288 Proteins (g): 3.0 Carbohydrates (g): 7.4 Total fat (g): 3.0 Saturated fatty acids (g): 1.2 Monounsaturated fatty acids (g); 1.5 Polyunsaturated fatty acids (g): 0.3 Omega‐3 (mg): 35, docosahexaenoic acid (DHA) (mg): 20, eicosapentaenoic acid (EPA) (mg): 10, vitamin A (retinol) (mg): 120, vitamin B1 (mg): 0.21, vitamin B2 (mg): 0.24, vitamin B3 (mg): 2.7, pantothenic acid (mg): 0.9, vitamin B6 (mg): 0.3, biotin (mg): 22.5, folic acid (mg): 30.0, vitamin B12 (mg): 0.15, vitamin C (mg): 9.0, vitamin D (mg): 0.75, vitamin E (mg): 1.5, calcium (mg): 140, zinc (mg): 2.25 Children in the Control group (n = 59) consumed 0.6 L/day of regular full milk
Outcomes	Biochemical indicators (HDL, LDL, TG, DHA, ferritin, iron, calcium, vitamin D, vitamin E); Anthropometric measures (BMI, waist circumference); Cognitive tests (digital span, letter number sequencing, coding, symbol/animal search)
Notes	The study was funded by Lactalis Puleva SL. One of the authors is currently employed and one of the authors was employed by Biosearch Life, which is part of Lactalis Study duration: January to June (year not specified).
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Quote: "Randomization was done with the program SIGESMU." Comment: Adequately done
Allocation concealment (selection bias)	Low risk	Quote: "The two beverages were labelled Product A and Product B" Comment: Adequately done
Blinding of participants and personnel (performance bias) All outcomes	Low risk	Quote: "all persons involved in the execution of the study were blind to their true content. The beverages were supplied in vacuum‐sealed tetrabrik containers with blank surfaces, without any trademarks or identification." Comment: Adequately done
Blinding of outcome assessment (detection bias) All outcomes	Low risk	Quote: "all persons involved in the execution of the study were blind to their true content. The beverages were supplied in vacuum‐sealed tetrabrik containers with blank surfaces, without any trademarks or identification." Comment: Adequately done
Incomplete outcome data (attrition bias) All outcomes	Low risk	Fortified milk group: 52/60 Regular milk group: 51/59
Selective reporting (reporting bias)	Low risk	Comment: Trial registration not specified. Outcomes specified in the methods section have been reported in the Results
Other bias	Low risk	No additional bias identified

Pinkaew 2013.

Methods	Randomised controlled trial
Participants	The study was conducted in Satun province, on the west coast of southern Thailand, where most of the population is Muslim. The study was performed in 8 primary schools (n = 203) in the Muang district, which included mainly children from low‐income families. The schools had 4‐ to 12‐year‐old children (kindergarten to grade 6) who were provided with a school lunch programme (5 days a week), which was partly subsidised by the government
Interventions	Intervention (n = 101): The fortified rice was mixed with the natural rice and cooked by local cooks at a central kitchen in Satun town, which had been specifically set up for the study. The cooked rice was weighed into individual portions of 140 g into a color‐coded container that was labelled with the child's name. The weight was regularly controlled by research assistants. The rice was transported to the 8 schools by the research assistants and the 140 g of cooked rice (triple‐fortified rice or unfortified rice) was given to each child. The rice was consumed with foods such as soup or curry, which was provided by the school lunch programme. The rice meal was fed 5 days a week Control: Placebo (n = 102) Food vehicle: Fortified rice Dose: 10 mg iron, 9 mg zinc, and 1050 mg vitamin A/g extruded rice Duration: 6 months Additional interventions: After completion of the study, all children who remained deficient in any of the micronutrients in either group received supervised treatment of the respective micronutrient(s) according to local policies
Outcomes	Serum zinc, retinol, haemoglobin, ferritin
Notes	The study was supported by Medicor Foundation (Triesen, Liechtenstein) and The Royal Thai Government Scholarship. Dr. Paul Lohmann GmbH (Emmerthal, Germany) provided iron and zinc compounds and DSM Nutritional Products Ltd. (Basel, Switzerland) provided the vitamin A compound Study duration: July 2009 to March 2010
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Comment: Insufficient information to permit judgement
Allocation concealment (selection bias)	Unclear risk	Comment: Insufficient information to permit judgement
Blinding of participants and personnel (performance bias) All outcomes	Unclear risk	Comment: Insufficient information to permit judgement
Blinding of outcome assessment (detection bias) All outcomes	Unclear risk	Comment: Insufficient information to permit judgement
Incomplete outcome data (attrition bias) All outcomes	Low risk	Intervention group: 7/101 Control group: 14/102 In total, 21 children (˜ 10%) were lost for final analysis and 182 completed the study according to the protocol Comment: Low attrition rate
Selective reporting (reporting bias)	Low risk	Comments: This trial was registered at www.clinicaltrials.gov as NCT01061307. No evidence of selective reporting
Other bias	Low risk	Comment: No additional bias identified

Pinkaew 2014.

Methods	Randomised controlled trial
Participants	The study was performed in a peri‐urban area of the Muang district, Satun province, on the west coast of southern Thailand. Most of the population was Muslim and the participants were primarily from low‐income families. One primary school in the Muang district, with children aged 4 – 12 years, was selected for the study (n = 50). The school provided a free lunch meal (5 days a week) that was partly subsidised by the government
Interventions	Intervention(n = 25): 1 group was given the triple‐fortified rice containing Fe, Zn, and vitamin A (fortified group) Control: Unfortified rice (n = 25) Food vehicle: Triple‐fortified rice Dose: 10 mg Fe, 9 mg Zn, and 1.05 mg vitamin A/g extruded rice Duration: 2 months
Outcomes	Serum retinol, vitamin A deficiency
Notes	Supported by Medicor Foundation (Triesen, Liechtenstein), the International Atomic Energy Agency (Vienna, Austria), and the Royal Thai Government Scholarship Study duration: August to November 2010
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Insufficient information to permit judgement
Allocation concealment (selection bias)	Unclear risk	Insufficient information to permit judgement
Blinding of participants and personnel (performance bias) All outcomes	Unclear risk	Quote: "The VA efficacy study was a double‐blind, randomised, controlled trial." Comment: Insufficient information to permit judgement
Blinding of outcome assessment (detection bias) All outcomes	Unclear risk	Insufficient information to permit judgement
Incomplete outcome data (attrition bias) All outcomes	Low risk	Quote: "Of the 50 children who started the intervention, 45 children completed it." Intervention group: 2/25 Control group: 3/25 Comment: Low attrition rate
Selective reporting (reporting bias)	Low risk	This trial was registered at clinicaltrials.gov as NCT01199445. No evidence of selective reporting
Other bias	Low risk	No additional bias identified

Powers 2016.

Methods	A randomised, double‐blind, placebo‐controlled intervention trial was conducted in girls recruited at ages 16 – 19 years, from schools and colleges in Sheffield, UK between July 2012 and May 2013
Participants	71 adolescent girls aged 16 – 19 years were selected from schools, colleges and Universities within the Sheffield area
Interventions	Intervention (n = 34): Girls were randomised to receive 50 g fortified with 150 ml semi‐skimmed milk daily for 12 weeks, as a breakfast or as a supper Energy (kcal): 257 Fat (g): 3.1 Carbohydrate (g): 47.8 Sugars (g): 15.9 Vitamin D (μg): 4.15, vitamin C (mg): 51.5, vitamin B1 (mg): 1.21, vitamin B2 (mg): 1.71, niacin (mg): 17.8, vitamin B6 (mg): 1.74, folic acid (μg): 176, vitamin B12 (μg): 1.45, iron (mg): 6.5, calcium (mg): 215 Control (n = 37): Unfortified cereal
Outcomes	Plasma ferritin, haemoglobin, micronutrient intake
Notes	The Kelloggs Company of Great Britain provided the cereal for the study and financial support for the research. The salary of one of the authors was provided by Kelloggs Company of Great Britain Study duration: July 2012 and May 2013
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Quote: "Volunteers were randomised in blocks of twelve to receive a daily intake of either fortified or unfortified cereal." Comment: Insufficient information
Allocation concealment (selection bias)	Unclear risk	Not specified.
Blinding of participants and personnel (performance bias) All outcomes	Low risk	Quote: "Unfortified and fortified cereal was provided and the identity of each was blinded to the researchers and participants." Comment: Adequately done
Blinding of outcome assessment (detection bias) All outcomes	Low risk	Quote: "Unfortified and fortified cereal was provided and the identity of each was blinded to the researchers and participants." Comment: Adequately done.
Incomplete outcome data (attrition bias) All outcomes	Low risk	Retained in study: 73/78 CONSORT flow diagram does not specify the loss to follow‐up according to the study group assignment
Selective reporting (reporting bias)	Low risk	Registered with Current Controlled Trials (Registration: ISRCTN55141306) and prespecified outcomes reported
Other bias	Low risk	The study enrolled voluntary participants and an Amazon voucher for GBP 30 was offered on completion of the study, although it was not considered a source of bias

Rahman 2015.

Methods	Cluster‐randomised controlled trial
Participants	The study sites included 7 out of the total 16 unions (approximately 65 villages) (n = 352 children) of Mirsarai sub‐district in the south‐eastern part of Bangladesh. Assuming that 7 – 9 eligible children (aged 6 – 15 years) would be available from each bari and using a statistics book generated random‐number table, a total of 44 baris were randomly selected from the total listed baris for distribution of the flour. Among the 44 selected baris, 22 baris were randomly assigned to the intervention group and 22 baris to the control group (control)
Interventions	Intervention (n = 203): Throughout the trial period, the project staff distributed the flour once every week. In order to prevent participants sharing of chapattis with other members of a bari, the same amount of flour was also allocated to other members of that bari during this period. Children received chapattis made from 100 g of fortified or unfortified wheat flour daily for 6 months Control (n = 149): Unfortified flour Food vehicle: Chappattis made from fortified wheat flour Dose: Vitamin A 212 ug, iron 6.6 mg, thiamine 0.64 mg, riboflavin 0.40 mg, folic acid 0.15 mg, zinc oxide 3.3 mg, niacin as niacinamide 5.3 mg Duration: 6 months
Outcomes	Serum retinol, ferritin, transferrin receptor, haemoglobin
Notes	Funded by a grant from the MOST project (Contract No. HRN‐AA‐00–98‐00047‐00) and by support to the Mirsarai field area by USAID Cooperation Agreement number 388‐A‐00–97‐00032‐00 Study duration: Not specified.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Quote: "Using a statistics book generated random number table, a total of 44 baris were randomly selected from the total listed baris for distribution of the flour. Among the 44 selected baris, 22 baris were randomly assigned to the intervention group and 22 baris to the control group (control)." Comment: Adequately done
Allocation concealment (selection bias)	Low risk	Quote: "A person not involved with the study assigned the baris to six different codes of flour (A, B, C, D, E and F) for distribution of the flour bags." Comment Adequately done
Blinding of participants and personnel (performance bias) All outcomes	Low risk	Quote: "It was only after completion of the analysis, the groups were unblinded." Comment: Adequately done
Blinding of outcome assessment (detection bias) All outcomes	Low risk	Quote: "It was only after completion of the analysis, the groups were unblinded." Comment: Adequately done
Incomplete outcome data (attrition bias) All outcomes	Low risk	Intervention group: 12/203 Control group: 6/149 5% total loss to follow‐up Comment: Low attrition rate
Selective reporting (reporting bias)	Low risk	Comment: Trial registration not specified. Outcomes specified in the methods section have been reported in the Results
Other bias	Low risk	No additional bias identified (i) recruitment bias: Low risk (ii) baseline imbalance: Low risk (iii) loss of clusters: Low risk (iv) incorrect analysis: Low risk (v) comparability with individually randomised trials: Low risk

Sazawal 2007.

Methods	Randomised controlled trial
Participants	The trial was carried out from April 2002 to April 2004, in Sangam Vihar, a peri‐urban community located on the outskirts of New Delhi, India. All permanent resident families in the area with children aged 1 – 3 years (n = 633) were invited to participate in the study, and their consent sought. Children who were exclusively or predominantly breast‐fed or allergic to milk were excluded. Children with severe malnutrition needing rehabilitation or chronic/severe illness requiring hospitalisation or special treatment were to be excluded
Interventions	Intervention (n = 316): Fonterra Brands (Singapore) Pte. Ltd. provided 32 g single‐serve sachets of fortified milk powder and control for the study. At enrolment, the procedure for preparing milk was clearly explained and demonstrated to mothers. Each week, the milk assistants delivered 21 sachets at home and advised the mother to feed the child 3 sachets a day Control (n = 317): Unfortified milk Food vehicle: Fortified powdered milk Dose: Fortified milk (3 servings a day) was designed to deliver additional amounts of zinc (7.8 mg), iron (9.6 mg), selenium (4.2 ug), copper (0.27 mg), vitamin A (156 ug), vitamin C (40.2 mg), vitamin E (7.5 mg) Duration: 1 year Additional interventions: At enrolment, all children who had severe anaemia (haemoglobin < 70 g/L) were given therapeutic doses of iron for 3 months in addition to their assigned intervention
Outcomes	Haemoglobin, hematocrit, protoporphyrin, ferritin, transferrin receptor, zinc, weight velocity, height velocity, WHZ, WAZ, HAZ
Notes	Supported from the grants of Fonterra Brands (Singapore) Pte. Ltd Study duration: April 2002 to April 2004
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Quote: "Letter codes A through D were used to identify four groups (across two separate trials). In‐house computer software generated a random sequence of group codes with permuted block length of 16." Comment: Adequately done
Allocation concealment (selection bias)	Low risk	Quote: "Group codes from 1 to 6 were used to identify the fortified and non‐fortified yoghurt" Comment: Adequately done
Blinding of participants and personnel (performance bias) All outcomes	Low risk	Quote: "The supplementation sachets were identical in colour, size (weight 32 g), taste and packaging and were labelled with a letter code. The investigators and the study team were blinded to the identity of the letter codes. Fonterra Brands Pte. Ltd. provided code identification to investigators after finishing of the trial, at the time of analysis." Comment: Adequately done
Blinding of outcome assessment (detection bias) All outcomes	Low risk	Quote: "The supplementation sachets were identical in colour, size (weight 32 g), taste and packaging and were labelled with a letter code. The investigators and the study team were blinded to the identity of the letter codes. Fonterra Brands Pte. Ltd. provided code identification to investigators after finishing of the trial, at the time of analysis." Comment: Adequately done
Incomplete outcome data (attrition bias) All outcomes	Low risk	Intervention group: 27/316 Control group: 36/317 Comment: Low attrition rate
Selective reporting (reporting bias)	Low risk	This trial was registered at ClinicalTrials.gov: NCT00980733 and there was no evidence of selective reporting
Other bias	Low risk	Comments: No additional bias identified

Sazawal 2013.

Methods	Randomised controlled trial
Participants	The study was conducted between June 2008 and March 2010 in primary schools of Gabtali town of Bogra district in the Rajshahi Division (n = 1010), Northern Bangladesh. The selected schools were in close proximity to a yoghurt factory. The inclusion criteria for enrolment into the study were children aged 6 to 9 years attending selected schools, who were likely to remain in the same school, and parents providing consent. Children with severe malnutrition needing nutritional rehabilitation or chronic/severe illness requiring hospitalisation or special treatment were excluded and referred for treatment
Interventions	Intervention (n = 501): Children allocated to the yoghurt groups received 1 cup of the yoghurt (60 g) daily during the lunch break of the school for 1 year. The feeding session was strictly monitored and supervised by the field workers and class teachers and the compliance to the intervention was recorded in compliance record forms. A separate list was prepared for children who were absent from school and their respective yoghurt cups were delivered at home in the afternoon Control (n = 509): Unfortified yogurt Food vehicle: Fortified yogurt Dose: Calcium 85 (mg), phosphorus 67 (mg), iron 3.3 (mg), zinc 3.0 (mg), iodine 40 (μg), vitamin A 140 (μg) Duration: 12 months
Outcomes	Haemoglobin, serum ferritin, transferrin receptor, zinc, iodine, RBP, body iron stores, weight velocity, height velocity, WAZ, HAZ, BMIz
Notes	Global Alliance for Improved Nutrition funded the study Study duration: June 2008 and March 2010
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Quote: "Using in‐house computer software, a random sequence of group codes with a permuted block length of 6 was generated to randomly allocate the individual child to one of the two yoghurt groups." Comment: Adequately done
Allocation concealment (selection bias)	Low risk	Quote: "Group codes from 1 to 6 were used to identify the fortified and non‐fortified yoghurt." Comment: Adequately done
Blinding of participants and personnel (performance bias) All outcomes	Low risk	Quote: "The codes of the groups were not known to the investigators, field team, teachers, children or anyone involved in the study during the field implementation. Cups were prepared and labelled with group codes a day in advance at a factory in Bogra. The yoghurt for the two intervention groups was identical in packaging, appearance, taste and smell." Comment: Adequately done
Blinding of outcome assessment (detection bias) All outcomes	Low risk	Quote: "The codes of the groups were not known to the investigators, field team, teachers, children or anyone involved in the study during the field implementation. Cups were prepared and labelled with group codes a day in advance at a factory in Bogra. The yoghurt for the two intervention groups was identical in packaging, appearance, taste and smell." Comment: Adequately done
Incomplete outcome data (attrition bias) All outcomes	High risk	Intervention group: 227/501 Control group: 216/509 Comment: High attrition rate
Selective reporting (reporting bias)	Low risk	Comment: This trial was registered at ClinicalTrial.gov: NCT00980733 and all prespecified outcomes have been reported in the Results
Other bias	Low risk	Comment: No additional bias identified

Solon 2003.

Methods	An individually‐randomised, double‐blind, placebo‐controlled field efficacy trial among schoolchildren in the municipality of Balete, located in the province of Batangas in the Philippines
Participants	831 children in grades 1 – 6 were enrolled from 4 elementary schools
Interventions	Participants were randomised into 1 of the 4 following groups and received beverage for 16 weeks: Group 1 received fortified beverage with anthelmintic therapy (n = 203); Group 2 received fortified beverage with placebo anthelmintic therapy (n = 209); Group 3 received non‐fortified beverage with anthelmintic therapy (n = 213); Group 4 received non‐fortified beverage with placebo anthelmintic therapy (n = 206). The fortified beverage contained a single serving (25 g sachets) with iron (4.8 mg), vitamin A (700 IU), iodine (48 μg), zinc (3.75 mg), vitamin C (75 mg), riboflavin (0.46 mg), folic acid (0.06 mg), vitamin B12 (0.5 μg), vitamin B6 (0.5 mg), vitamin E (2.5 mg), and niacin (2.5 mg) For this review, we have only included data from Group 2 and Group 4
Outcomes	Weight, height, weight for age Z‐score, height for age Z‐score, weight for height Z‐score, haemoglobin, anaemia, urinary iodine, physical fitness and cognitive performance
Notes	The trial was funded by the Nutrition Center of the Philippines and Procter & Gamble Co Study duration: October 1998 to March 1999
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Quote: "Study participants were assigned, through randomizations at the individual level, to one of four different treatment groups." Comment: Insufficient information
Allocation concealment (selection bias)	Unclear risk	Not specified.
Blinding of participants and personnel (performance bias) All outcomes	Low risk	Quote: "Placebo beverage and placebo anthelmintic pills were indistinguishable from their counterparts in appearance, smell, and taste." Comment: Adequatelyt done
Blinding of outcome assessment (detection bias) All outcomes	Low risk	Quote: "Placebo beverage and placebo anthelmintic pills were indistinguishable from their counterparts in appearance, smell, and taste." Comment: Adequately done
Incomplete outcome data (attrition bias) All outcomes	Low risk	Overall loss to follow‐up 43/851
Selective reporting (reporting bias)	Low risk	Trial registration not specified. Outcomes specified in the methods section have been reported in the Results
Other bias	Low risk	No additional bias identified.

Taljaard 2013.

Methods	Randomised controlled trial
Participants	Conducted in primary school children (n = 414) between the ages of 6 and 11 years in a peri‐urban settlement in the North West province in South Africa. The study was conducted in 3 preselected primary schools chosen by the Department of Education. Learners at all 3 schools were provided a single daily meal, sponsored by the National School Nutrition Programme. The inclusion criteria were as follows: (1) no health condition that would make cognitive testing impractical (e.g. dyslexia and hearing difficulties); (2) 6 – 10 years old by January 2010; (3) no use of medication or supplements that could affect nutritional status
Interventions	Intervention: The four different formulations of the beverages were as follows: Micronutrients with sugar (MNS) (n = 104); No micronutrients (control beverage) with sugar (CS) (n = 104); Micronutrients with a non‐nutritive sweetener (MNNS) (n = 103); No micronutrients (control beverage) with a non‐nutritive sweetener (CNS) (n = 103). Food vehicle: Fortified beverage Dose: Vitamin A (ug RE) 400, vitamin E (mg) 7·5, vitamin C (mg) 60, vitamin B2 (mg) 0·4, nicotinamide (mg) 2·7, vitamin B6 (mg) 0·5, folic acid (ug) 140, vitamin B12 (ug) 1·0, calcium (mg) 120, iron (mg) 7·0, zinc (mg) 3·75, iodine (ug) 60 Duration: 8 months Additional interventions: Children were de‐wormed at the baseline with 200 mg (100 mg twice daily) of mebendazole for 3 consecutive days For this review, we have only included data from MNNS and CNS groups.
Outcomes	Haemoglobin, serum ferritin, protoporphyrin, transferrin receptor, zinc, retinol
Notes	Supported by a research grant from Coca Cola South Africa (Pty) Study duration: March 2010 to November 2010
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Insufficient information to permit judgement
Allocation concealment (selection bias)	Unclear risk	Insufficient information to permit judgement
Blinding of participants and personnel (performance bias) All outcomes	Low risk	Quote: "The colourant Yellow Sunset E110 was used to give a similar colour to the micronutrient‐containing beverages that had b‐carotene. All the beverage formulations were, therefore, identical in colour and taste. The participants, investigators and school assistants were blinded to treatment assignments." Comment: Adequately done
Blinding of outcome assessment (detection bias) All outcomes	Unclear risk	Comment: Insufficient information to permit judgement
Incomplete outcome data (attrition bias) All outcomes	Low risk	Micronutrients with sugar (MNS) (2/104); No micronutrients (control beverage) with sugar (CS) (3/104); Micronutrients with a non‐nutritive sweetener (MNNS) (8/103); No micronutrients (control beverage) with a non‐nutritive sweetener (CNS) (3/103). 3.9% total loss to follow‐up Comment: Low attrition rate
Selective reporting (reporting bias)	Low risk	Comment: Trial registration not specified. Outcomes specified in the methods section have been reported in the Results
Other bias	Low risk	Comment: No additional bias identified

Tapola 2004.

Methods	Randomised controlled trial
Participants	66 participants were recruited to take part in the study from eastern Finland. They fulfilled the following inclusion criteria: age 26 – 65 years, normal liver, kidney and thyroid function, and no history of unstable coronary artery disease (i.e. myocardial infarction, coronary artery bypass graft (CABG), or percutaneous transluminal coronary angioplasty (PTCA) within the previous 6 months), transient Ischaemic attack, kidney stones or malignant diseases. People with alcohol abuse (above 45 g of ethanol a day) or those that had used vitamin supplements (B or D vitamins) within 2 months prior to the study were excluded
Interventions	Intervention (n = 31): During the intervention period, the participants followed their habitual diet except that 750 ml of liquids were replaced with the mineral water. They were not allowed to use calcium supplements and foods fortified with calcium and/or any B group vitamin during the study Control (n = 29): Placebo Food vehicle: Fortified mineral water Dose: Potassium (mg) 141, magnesium (mg) 53, calcium (mg) 563, sodium (mg) 6, vitamin B6 (mg) 1, vitamin B12 (ug) 7.5, folic acid (ug) 563, vitamin D (ug) 0.6 Duration: 8 weeks
Outcomes	Serum folate, erythrocyte folate, serum vitamin B12, calcium, alkaline phosphatase, plasma homocysteine
Notes	Supported by Olvi PLC, Iisalmi, Finland Study duration: Not specified.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Insufficient information to permit judgement
Allocation concealment (selection bias)	Unclear risk	Insufficient information to permit judgement
Blinding of participants and personnel (performance bias) All outcomes	Unclear risk	Insufficient information to permit judgement
Blinding of outcome assessment (detection bias) All outcomes	Unclear risk	Insufficient information to permit judgement
Incomplete outcome data (attrition bias) All outcomes	Low risk	Quote: "Of the 62 subjects who were randomised, 60 (39 men and 21 women) completed the study. Two subjects discontinued the study: one dropped out due to adverse effects (abdominal discomfort, diarrhoea, nausea and vomiting) of the test mineral water and one subject moved away." Comment: Low attrition rate
Selective reporting (reporting bias)	Low risk	Trial registration not specified. Outcomes specified in the methods section have been reported in the Results
Other bias	Low risk	Quote: "Men and women were randomised separately. In addition, the randomisation was carried out separately for couples and single persons, in order to ensure the same mineral water for both spouses." Comment: Low risk of bias as randomisation was performed adequately on all groups

Tatala 2002.

Methods	Randomised controlled trial
Participants	Conducted in pregnant women 12 and 34 weeks pregnant in Tanzania from August 1999 to October 1999 (n = 439).
Interventions	Intervention (n = 129): The fortified beverage mix was packaged in 25 g packets. Each woman was asked to consume the beverage thus produced twice a day with meals Control (n = 130): non‐fortified beverage mix of identical appearance, colour and taste. Was packaged in similar, but different coloured 25 g packets and served as the placebo Dose: Iron 10.8 mg, vitamin A 1050 RE, iodine 90 ug, zinc 10.5 mg, vitamin C 144 mg, riboflavin 1.2 mg, folic acid 280 ug, vit B12 6 ug, B6 1.4 mg, niacin 10 mg, vit E 21 mg. Food vehicle: orange‐flavoured micronutrient‐fortified powdered beverage mix containing 11 micronutrients Duration: 8 weeks Additional Interventions: Just before the women left the antenatal clinic, staff provided the mothers with a 2‐week supply of an iron/folic acid supplement that contained 60 mg of elemental iron and 500 g of folic acid to be taken on a daily basis. Women who were found to have parasitic infections were treated with a single dose of albendazole (400 mg)
Outcomes	Height, weight, mid‐upper arm circumference (MUAC) and skinfold thickness, Hemoglobin, TSH, retinol, CRP, ferritin
Notes	The micronutrient supplement was developed and produced by Procter & Gamble Company. Supported by a Micronutrient Initiative grant, Procter & Gamble Company, UNICEF Tanzania, Tanzania Food and Nutrition Centre and Cornell University. Study date: August 1999 to October 1999
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Quote: "At each of the six study centres, a block randomisation (10 subjects in each block) was used to assign women into either the micronutrient‐fortified (experimental) group or the non‐fortified (control) group." Comment: Adequately done
Allocation concealment (selection bias)	Unclear risk	Insufficient information to permit judgement
Blinding of participants and personnel (performance bias) All outcomes	Unclear risk	Quote: "This study was a randomised, placebo‐controlled double‐blind effectiveness trial." Comment: Insufficient information to permit judgement
Blinding of outcome assessment (detection bias) All outcomes	Unclear risk	Quote: "This study was a randomised, placebo‐controlled double‐blind effectiveness trial." Comment: Insufficient information to permit judgement
Incomplete outcome data (attrition bias) All outcomes	High risk	Quote: "Of those enrolled (439), 121 women were lost to follow‐up and 59 mothers delivered before their 8‐wk postintervention visit (Fig. 1). Because delivery affects haemoglobin and other variables that were measured, the main statistical analyses were restricted to the 259 (59% of enrolled women; 127 experimental and 132 placebo) women who were still pregnant and were still in the study after 8 wk of supplementation, regardless of their gestational age at entry into the study." Comment: Attrition rate may have affected outcomes
Selective reporting (reporting bias)	Low risk	Trial registration not specified. Outcomes specified in the methods section have been reported in the Results
Other bias	Low risk	No additional bias identified

Thankachan 2012.

Methods	Randomised controlled trial
Participants	The study was carried out in a convenience sample of children attending 4 primary schools in the Bangalore Urban District of Karnataka State in South India. These schools were selected because they cater to the educational needs of children from low socio‐economic neighbourhoods and are situated within the Bangalore urban region. 258 consenting children aged 6 – 12 years were included. Weight and height were measured and venous blood was collected to identify children with haemoglobin (Hb) concentrations < 115 g/L for 6 – 11 years and < 120 g/L for 12‐year‐olds. Those who did not intend to use micronutrient supplements other than those administered at school during the study, or who did not intend to migrate or withdraw from school during the study period were eligible for inclusion in the study. Participants with Hb < 90 g/L or who had any chronic illness requiring long‐term use of medication, physical handicaps, or severe malnutrition (weight‐for‐age Z‐score or height‐for‐age Z‐score < −3) were excluded
Interventions	Intervention: The 3 types of rice: 1. High iron: 12.5 mg Fe/100 g (n = 86) 2. Low iron: 6.5 mg Fe/ 100 g (n = 86) 3. Control, ˜ 100 g raw rice/meal (n = 86) Food Vehicle: Two types of fortified rice; high and low iron Dose: All mg/100g, vitamin A 0.5, thiamine 0.38, niacin 5, vitamin B6 0.38, vitamin B12 0.00075, folate 0.075, iron 12.5 high iron group; 6.25 low iron group, zinc 3 Duration: 6 months Additional Interventions: All study children were de‐wormed under the supervision of the research staff with 400 mg albendazole (Low‐Cost Pharmaceuticals) before the study and near the study midpoint
Outcomes	Haemoglobin, serum ferritin, transferrin receptor, protoporphyrin, retinol, zinc, thiamine, vitamin B12, plasma homocysteine
Notes	Supported by DSM, Mumbai, India Study duration: July 2009 to March 2010
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Quote: "A block randomisation with a computer‐generated list in blocks of 20 was used to assign children to one of the 3 intervention groups." Comment: Adequately done
Allocation concealment (selection bias)	Unclear risk	Quote: "Each group was randomly assigned a distinct colour code, which remained unknown to both the study staff and the children until the completion of the study." Comment: Insufficient information to permit judgement
Blinding of participants and personnel (performance bias) All outcomes	Low risk	Quote: "Each group was randomly assigned a distinct colour code, which remained unknown to both the study staff and the children until the completion of the study." Comment: Adequately done
Blinding of outcome assessment (detection bias) All outcomes	Low risk	Quote: "All data were entered and initial analyses were performed prior to unmasking of the study." Comment: Adequately done
Incomplete outcome data (attrition bias) All outcomes	Low risk	10% total loss to follow‐up Comment: Low attrition rate
Selective reporting (reporting bias)	Low risk	Comment: This trial was registered at the Clinical Trials Registry of India as CTRI/20/09/091/000941 and there was no evidence of selective reporting
Other bias	Low risk	Comment: No additional bias identified

Thankachan 2013.

Methods	Randomised controlled trial
Participants	This study was carried out in children attending the St Joseph primary school in Kolar and Franciscan School, Bangalore, South India. 246 children aged 6 – 12 years were included. Weight and height were measured and 8 ml of venous blood was collected for determination of Hb and serum ferritin (SF). Children with SF levels < 20 ug/l and who were not intending to use micronutrient supplements during the study, to migrate or withdraw from school during the study period, were eligible for inclusion in the study. Children with anaemia (Hb < 8 g/dl), chronic illness, physical handicaps or severe malnutrition (weight‐for‐age (WAZ) or height‐for‐age (HAZ) Z‐score <‐3) were excluded
Interventions	Intervention (n = 122): The MMN‐fortified beverage provided between 20% and 70% of the recommended daily allowance of micronutrients (vitamin A, B2, B12, C, folic acid, iron and zinc) for children between 6 and 12 years. The beverage contained 6 mg iron/serving as ferrous gluconate along with 27 mg of vitamin C in an orange‐flavoured base. The drinks were provided 6 days/week for a period of 8 weeks Control (n = 124): Unfortified beverage Food vehicle: Fortified beverage Dose: Vitamin A 243 (ug), vitamin C 27 (mg), vitamin B2 0.63 (mg), vitamin B12 1.27 (ug), folic acid 35 (ug), iron 5.9 (mg), zinc 1.2 (mg) Duration: 8 weeks
Outcomes	Haemoglobin, serum ferritin, transferrin receptor, protoporphyrin, vitamin A, B12, C, zinc, folate, body iron stores
Notes	Supported by Coca‐Cola, India Study duration: January 2010 to March 2010
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Quote: "A block randomisation with a computer‐generated list in blocks of 20 each was used to assign children to one of the two intervention groups." Comment: Adequately done
Allocation concealment (selection bias)	Unclear risk	Quote: "Each group was randomly assigned a distinct colour code, which remained blinded to both the study staff and children until the completion of the study." Comment: Insufficient information to permit judgement
Blinding of participants and personnel (performance bias) All outcomes	Low risk	Quote: "Each group was randomly assigned a distinct colour code, which remained blinded to both the study staff and children until the completion of the study. All data were entered and initial analyses were performed before unmasking the study." Comment: Adequately done
Blinding of outcome assessment (detection bias) All outcomes	Low risk	Quote: "Each group was randomly assigned a distinct colour code, which remained blinded to both the study staff and children until the completion of the study. All data were entered and initial analyses were performed before unmasking the study." Comment: Adequately done
Incomplete outcome data (attrition bias) All outcomes	Low risk	Intervention group: 2/122 Control group: 1/124 0.01% total loss to follow‐up Comment: Low attrition rate
Selective reporting (reporting bias)	Low risk	Trial registration not specified. Outcomes specified in the methods section have been reported in the Results
Other bias	Low risk	No additional bias identified

Tucker 2004.

Methods	Randomised controlled trial
Participants	Conducted in the USA. Volunteers were recruited through advertisements in local newspapers, posters, radio, and mailing lists. 215 adults (93 men and 122 women) 50 to 85 years old, completed the protocol, of whom 196 were non‐Hispanic white, 12 were African‐American, 4 were Asian‐American, and 3 were of another ethnicity
Interventions	Intervention (n = 93): The enrolled participants were randomly assigned to consume breakfast cereal fortified with the RDAs of folic acid, vitamin B‐6, and vitamin B‐12 (400 ug, 2 mg, and 6 ug, respectively) per 1‐cup (0.24 L) serving or an identical cereal without the addition of these vitamins Control (n = 96): Unfortified cereal Food vehicle: Fortified cereal Dose: Analysis of the ready‐to‐eat cereal after fortification showed that the actual content was 440 ug folic acid, 1.8 mg vitamin B‐6, and 4.8 ug vitamin B‐12 per serving. Both cereals contained 100 kcal, 24 g carbohydrate, 1 g dietary fibre, 0.35 mg thiamine, 0.34 mg riboflavin, and 4.0 mg niacin per serving Duration: 12 weeks
Outcomes	Serum folate, vitamin B12, B6, plasma homocysteine
Notes	Supported by a grant from the Kellogg Company and by the US Department of Agriculture Agricultural Research Service (contract 53‐3K06‐01) Study duration: Not specified
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Quote: "The study statistician (GED) randomly assigned the subjects to 1 of the 2 groups." Comment: Insufficient information to permit judgement
Allocation concealment (selection bias)	Unclear risk	Insufficient information to permit judgement
Blinding of participants and personnel (performance bias) All outcomes	Unclear risk	Quote: "All staff members that interacted with the subjects were blind to the group assignments." Comment: Blinding of participants is not mentioned
Blinding of outcome assessment (detection bias) All outcomes	Unclear risk	Insufficient information to permit judgement
Incomplete outcome data (attrition bias) All outcomes	Low risk	10% total loss to follow‐up Comment: Low attrition rate
Selective reporting (reporting bias)	Low risk	Trial registration not specified. Outcomes specified in the methods section have been reported in the Results
Other bias	Low risk	No additional bias identified

Van het Hof 1998.

Methods	Randomised controlled trial
Participants	Conducted in the Netherlands. Non‐smoking adults, healthy as assessed by a medical investigation, between 18 and 65 years old, were eligible for participation in the study. The volunteers (n = 31) did not use vitamin C, E, carotenoid, selenium or zinc supplements. They were not using a medically prescribed diet or slimming regimen and had been weight‐stable for at least 1 month prior to the start of the study. Women were not pregnant or lactating. Volunteers were recruited from employees of Unilever Research Laboratorium and from inhabitants of Vlaardingen and the surrounding district
Interventions	Intervention (n = 15): 15 g/d of an antioxidant fortified margarine Control (n = 16): 15 g/d of an ordinary unfortified margarine Food vehicle: Fortified margarine Dose: vitamin C 121 mg, vitamin E 31 mg, a‐carotene 2.7 mg, b‐carotene 5.3 mg Duration: 9 months
Outcomes	Serum vitamin E, a‐carotene, b‐carotene, vitamin C, albumin, uric acid, total antioxidant activity
Notes	Funding not specified Study duration: Not specified
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Insufficient information to permit judgement
Allocation concealment (selection bias)	Unclear risk	Insufficient information to permit judgement
Blinding of participants and personnel (performance bias) All outcomes	High risk	Quote: "Due to the difference in colour of the margarines, the study was only blind to the analysts analysing the blood samples." Comment: Not done
Blinding of outcome assessment (detection bias) All outcomes	Low risk	Quote: "Due to the difference in colour of the margarines, the study was only blind to the analysts analysing the blood samples." Comment: Adequately done
Incomplete outcome data (attrition bias) All outcomes	Low risk	Quote: "Fifteen men and sixteen women completed the study. The volunteer who withdrew from participation in the study did so for medical reasons not related to the experimental treatment." Comment: Low attrition rate
Selective reporting (reporting bias)	Low risk	Trial registration not specified. Outcomes specified in the methods section have been reported in the Results
Other bias	Low risk	No additional bias identified

Van Stuijvenberg 1999.

Methods	Randomised controlled trial
Participants	The study population consisted of children aged 6 – 11 years in grades 1 – 5 of the Ndunakazi Primary School, a school in a rural mountainous area ˜ 60 km northwest of Durban, KwaZuluNatal, South Africa, and serving a community characterised by low socio‐economic status (n = 228)
Interventions	Intervention (n = 115): The biscuits and cold drinks were distributed daily during the school week during the first 2 hours of the school day. No intervention took place during school holidays, weekends, or public holidays; the supplement was provided for a total of 215 days, or 43 weeks Control (n = 113): Placebo Food vehicle: Fortified biscuit Dose: Iron 5.9 (mg), b‐carotene 2.0 (mg), iodine 95.4 (ug) Duration: 12 months Additional Interventions: To enhance the absorption of iron, a vitamin C–fortified cold drink was given to the intervention group; the control group received an unfortified cold drink (placebo) All the children were dewormed (400 mg albendazole) at 4‐monthly intervals during the 12‐month randomised controlled trial, and on a further three occasions during the subsequent 18‐month follow‐up period.
Outcomes	Serum retinol, ferritin, iron, transferrin saturation, haemoglobin, hematocrit, urinary iodine, white blood cell count
Notes	Supported by a grant from SASKO Pty Ltd, who also donated the fortified products and placebo; the anthelmintic tablets were donated by SmithKline Beecham Pharmaceuticals Pty, Ltd Study duration: May 1995 to June 1996
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Insufficient information to permit judgement
Allocation concealment (selection bias)	Unclear risk	Insufficient information to permit judgement.
Blinding of participants and personnel (performance bias) All outcomes	Low risk	Quote: "The other group received an unfortified biscuit similar to the fortified biscuit in macronutrient composition, taste, and appearance. To avoid the exchange of biscuits and cold drinks between classmates, the intervention and control groups were seated on opposite sides of the classroom." Comment: Adequately done
Blinding of outcome assessment (detection bias) All outcomes	Unclear risk	Insufficient information to permit judgement
Incomplete outcome data (attrition bias) All outcomes	Low risk	5.3% total loss to follow‐up Comment: Low attrition rate
Selective reporting (reporting bias)	Low risk	Trial registration not specified. Outcomes specified in the methods section have been reported in the Results
Other bias	Low risk	No additional bias identified

Vaz 2011.

Methods	A double‐blind, placebo‐controlled, randomised trial in children aged between 7 and 10½ years from 3 schools in Bangalore, India
Participants	300 clinically healthy school‐age children aged 7 to 10½ years old
Interventions	Children were allocated to 1 of 3 study arms: The test group (F) was given a fortified choco‐malt beverage powder (n = 95) One of the control groups (U) was administered an energy‐equivalent unfortified choco‐malt beverage powder (n = 95) Second control group received no intervention (C) (n = 97). Duration: 120 days. For this review, we included data for groups F and U
Outcomes	Micronutrient status included thiamine, riboflavin, folate, niacin, iron, pyridoxal phosphate, and vitamins B12 and C
Notes	This trial was sponsored by GlaxoSmithKline Consumer Healthcare Ltd, Gurgaon, India Study duration:
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Quote: "The block randomisation technique was employed to generate 20 blocks (10 each of girls and boys to ensure equal gender distribution) each of size 15. The participants in each block were individually randomised to 1 of the 3 treatment groups based on a computer‐generated randomisation sequence." Comment: Adequately done
Allocation concealment (selection bias)	Low risk	Quote: "The computer‐generated sequence of randomisation with study arm allocation was restricted to a single person (T.T.)" Comment: Adequately done
Blinding of participants and personnel (performance bias) All outcomes	Low risk	Quote: "The sponsor retained the codes for the product (F and U groups) and a copy was kept with a faculty member not involved with the study at the site in the event of an emergency. These codes were broken once all biochemical assessments (except thiamine and niacin) were completed and after database lock." Comment: Adeqautely done
Blinding of outcome assessment (detection bias) All outcomes	Low risk	Quote: "The sponsor retained the codes for the product (F and U groups) and a copy was kept with a faculty member not involved with the study at the site in the event of an emergency. These codes were broken once all biochemical assessments (except thiamine and niacin) were completed and after database lock." Comment: Adeqautely done
Incomplete outcome data (attrition bias) All outcomes	Low risk	Comment: Overall loss to follow‐up: 13/287 Low risk of attrition bias
Selective reporting (reporting bias)	Low risk	Comment: This trial was registered at clinicaltrials.gov as NCT00876018 and the prespecified outcomes have been reported
Other bias	Low risk	No additional bias identified

Villalpando 2006.

Methods	Randomised controlled trial
Participants	This randomised clinical trial was carried out in a poor peri‐urban community of 5000 inhabitants in the outskirts of Puebla, a city located 120 km east of Mexico City. Healthy children (n = 115), 10 – 30 months of age at the beginning of the study, were selected from a registry of children younger than 5 years of age living in the community
Interventions	Intervention: Children were randomly assigned to drink 400 mL/d (200 mL in the morning, 200 mL in the evening) of cow’s whole milk (distributed as milk powder) either fortified (FM) with 5.28 mg/400 mL of iron as ferrous gluconate and other micronutrients (n = 58) or non‐fortified milk (NFM) (iron concentration: 0.2 mg/400 mL) (n = 57). Food vehicle: Fortified milk Dose: unit/kg dry powder Iron (ferrous gluconate), 109.8 mg, zinc (zinc oxide), 109.8 mg, retinol palmitate, 449 ug, vitamin C (sodium ascorbate), 998 mg, folic acid, 669 ug Duration: 6 months
Outcomes	Haemoglobin, serum ferritin, transferrin receptor, zinc
Notes	Supported in part by The Ministry of Social Development of Mexico and Instituto Nacional de Salud Publica Study duration: Not specified
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Quote: "Children were randomly assigned to drink 400 mL/d (200 mL in the morning, 200 mL in the evening) of cow’s whole milk .. " Comment: Insufficient information to permit judgement
Allocation concealment (selection bias)	Unclear risk	Insufficient information to permit judgement
Blinding of participants and personnel (performance bias) All outcomes	Low risk	Quote: "The packages of FM and NFM were undistinguishable, except for a color‐coded band in the upper corner of the sachet. The colour code was unknown to researchers, field workers, and users and was disclosed after data analysis." Comment: Adequately done
Blinding of outcome assessment (detection bias) All outcomes	Low risk	Quote: "The packages of FM and NFM were undistinguishable, except for a color‐coded band in the upper corner of the sachet. The colour code was unknown to researchers, field workers, and users and was disclosed after data analysis." Comment: Adequately done
Incomplete outcome data (attrition bias) All outcomes	Low risk	Intervention group: 10/68 Control group: 5/62 11.5% total loss to follow‐up Comment: Low attrition rate
Selective reporting (reporting bias)	Low risk	Trial registration details not specified. All the prespecified outcomes in the methods section have been reported in the Results section
Other bias	Low risk	No additional bias identified

Vinodkumar 2009.

Methods	Cluster‐randomised controlled trial
Participants	3 residential schools were randomly selected as the experimental schools and 3 other residential schools as the controls in the city of Chennai, Tamilnadu, South India (n = 402)
Interventions	Intervention (n = 213): Multiple micronutrient‐fortified salt Control (n = 189): Iodised salt Food vehicle: Fortified salt Dose: Vitamin A 3000 IU, vitamin B1 1 mg, vitamin B2 1 mg, vitamin B6 1 mg, niacin 5 mg, iron 1000 ppm, iodine 40 ppm, folic acid 100 mcg, vitamin B12 4 mcg, zinc 10 mg Duration: 9 months Additional interventions: Both the experimental and the control children were given a tablet of albendazole (400 mg) at baseline, at 4 months, and post‐intervention after 9 months
Outcomes	Haemoglobin, serum ferritin, transferrin receptor, CRP, vitamin A, B12, folate, zinc, body iron stores, prevalence of angular stomatitis
Notes	Supported by Task Force Sight and Life Study duration: July 2005 to April 2006
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Quote: "Three residential schools were randomly selected as the experimental schools and three other residential schools as the controls.." Comment: Insufficient information to permit judgement
Allocation concealment (selection bias)	Unclear risk	Insufficient information to permit judgement
Blinding of participants and personnel (performance bias) All outcomes	Unclear risk	Insufficient information to permit judgement
Blinding of outcome assessment (detection bias) All outcomes	Unclear risk	Insufficient information to permit judgement
Incomplete outcome data (attrition bias) All outcomes	Low risk	Comment: 11.7% total loss to follow‐up Low attrition rate
Selective reporting (reporting bias)	Low risk	Comment: Trial registration details not specified. All the prespecified outcomes in the methods section have been reported in the Results section
Other bias	Low risk	No additional bias identified

Wang 2017.

Methods	A cluster‐randomised controlled trial among healthy Chinese middle‐school students, aged 12 to 14 years, between June 2015 and January 2016
Participants	360 students were enrolled from Xi’an Middle School
Interventions	Participating children were allocated to either an intervention group (n = 177) or a control group (n = 183). Intervention group students were given 250 mL micronutrient‐fortified milk (Future Star, Mengniu Dairy Company Limited, Hohhot, China) per day for 6 months; students of the control group were provided with pure milk with approximately the same caloric value of the fortified milk (Milk Deluxe, China Mengniu Dairy Company Limited, Hohhot, China) Energy KJ: 332 Protein g: 3.1 Fat g: 3.6 Carbohydrate g: 8.6 Sodium mg: 58 Vitamin A g RE: 78, vitamin D g: 1.5, vitamin E mg ‐TE: 2.0, vitamin B2 mg: 0.09, pantothenic acid mg: 0.2, phosphorus mg: 70, calcium mg: 100, Zinc mg: 0.34
Outcomes	Micronutrient deficiencies (iron, vitamin D, vitamin B2, vitamin B12, selenium); academic performance; motivation and learning strategy scores
Notes	This work was financially supported by a grant (Grant No. 81101333) from the National Natural Science Foundation of China and a grant (Grant No. 13‐168‐201608) from China Medical Board Study duration: June 2015 and January 2016
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Quote: "Participating children were allocated to either an intervention group (n = 177) or a control group (n = 183) with random number table by the research staff." Comment: Adequately done
Allocation concealment (selection bias)	High risk	Not done
Blinding of participants and personnel (performance bias) All outcomes	High risk	Quote: "Children, study investigators and the data analyst were not blinded to the treatment allocation" Comment: Not done
Blinding of outcome assessment (detection bias) All outcomes	High risk	Quote: "Children, study investigators and the data analyst were not blinded to the treatment allocation" Comment: Not done
Incomplete outcome data (attrition bias) All outcomes	Low risk	Intervention group: 40/177 Control group: 24/183
Selective reporting (reporting bias)	Low risk	Trial registration details not specified. All the pre‐specified outcomes in the methods section have been reported in the Results section
Other bias	Low risk	Comment: No additional biases identified. (i) recruitment bias: Low risk (ii) baseline imbalance: Low risk (iii) loss of clusters: Low risk (iv) incorrect analysis: Low risk (v) comparability with individually randomised trials: Low risk

Zimmerman 2004.

Methods	Randomised controlled trial
Participants	Conducted in Morocco. The participants were children 6 – 14 years old from 2 neighbouring primary schools
Interventions	Intervention: Group 1 (IS group) was given IS, i.e. salt fortified with 25 ug I/g salt (n = 83) Group 2 (TFS group) was given TFS, i.e. salt triple‐fortified with 25 ug iodine, 60 ug vitamin A, and 2 mg iron/g salt (n = 74) Food vehicle: Triple‐fortified salt Dose: 25 ug iodine, 60 ug vitamin A, and 2 mg iron/g salt. Duration: 10 months
Outcomes	Haemoglobin, serum transferrin, ferritin, zinc, protoporphyrin, body iron, retinol, RBP
Notes	Supported by the Thrasher Research Fund (Salt Lake City), the Foundation for Micronutrients in Medicine (Rapperswil, Switzerland), and the Swiss Federal Institute of Technology (Zurich, Switzerland) Study duration: Not specified
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	High risk	Quote: "Because each participating family shared a monthly salt portion, children were randomly divided by household into 2 groups." Comment: Not adequately done
Allocation concealment (selection bias)	Unclear risk	Comment: Insufficient information to permit judgement
Blinding of participants and personnel (performance bias) All outcomes	Low risk	Quote: "Both investigators and households were blind to group assignment" Comment: Adequately done
Blinding of outcome assessment (detection bias) All outcomes	Low risk	Quote: "Both investigators and households were blind to group assignment" Comment: Adequately done
Incomplete outcome data (attrition bias) All outcomes	Low risk	Quote: "Of the 159 children who began the study, 157 completed it; 2 children in the TFS group moved away." Comment: Low attrition rate
Selective reporting (reporting bias)	Low risk	Comment: Trial registration details not specified. All the pre‐specified outcomes in the methods section have been reported in the results section
Other bias	Low risk	No additional bias identified

BMI: body mass index; CRP: C‐reactive protein; DHA: docosahexaenoic acid; HAZ: height‐for‐age Z‐score; HDL: high‐density lipoprotein; LAZ: length‐for‐age Z‐score; LD: low‐density lipoprotein; MUAC: middle upper arm circumference; RBP: retinol binding protein; RDA: recommended daily allowance; RNI: recommended nutritional intake; TG: triglycrides; WAZ: wight‐for‐age Z‐score; WHZ: weight‐for‐height Z‐score; WLZ: weight‐for‐length Z‐score

Characteristics of excluded studies [ordered by study ID]

Study	Reason for exclusion
Aburto 2010	The study does not have an appropriate control group
Agte 2006	This study involves comparison of use of alternative natural foods with standard foods rather than the use of fortified food vehicles
Anand 2007	This study only used 2 micronutrients (Iron and vitamin A)
Angeles Agdeppa 2011	This study included children who were anaemic at baseline
Angeles‐Agdeppa 2017	This study included participants who were anaemic at baseline
Barkley 2015	This study reported multiple cross‐sectional surveys
Baró 2003	This study was a pre‐post study without a control group
Berasategi 2011	This study compares diets instead of singular food vehicles with little reference to the exact included foods in those diets
Bishop 1996	This study did not mention the specific micronutrients
Chen 2011	This study assessed in‐home fortification using micronutrient powders
Cheung 2016	This study evaluates supplementation
Christian 2015	This study provided fortified blended foods
Colker 2002	This study included participants with osteoarthritis
Filteau 2010	This study focused on specific population group (HIV exposed children)
Gathwala 2007	This study assessed in‐home fortification using micronutrient powders
Gershoff 1977	This study was a pre‐post study without a control group
Glosz 2018	This study compares various fortified blended foods
Goyle 2010	This study was a pre‐post study without a control group
Grieger 2009	This study was a pre‐post study without a control group
Hoffman 2007	This study was a pre‐post design without a control group
Hund 2013	This is a cross‐sectional study
Huybregts 2009	This study focused on blended foods
Iannotti 2016	This study assess ready‐to‐use therapeutic food
Jaatinen 2014	This study was a pre‐post study without a control group
Janmohamed 2016	This study assessed fortified blended foods
Kanellakis 2012	This study did not have an appropriate control group
Krebs 2012	This study did not have an appropriate control group
Kruger 2010	This study included non‐isocaloric supplement
Kumar 2007	This study did not have an appropriate control group
Kumar 2008	This study assessed point‐of‐use fortification
Kumar 2014	This study did not have an appropriate control group
Kuriyan 2016	This study assessed point‐of‐use fortification
Kuusipalo 2006	This study focused on ready‐to‐use therapeutic food and spreads
Lartey 1999	This study focused on ready‐to‐use therapeutic food
Layrisse 1996	This study had a pre‐post design without a control group
Loui 2004	This study did not have an appropriate control group
Lucas 1996	This study compares milk MMN fortifier with MMN supplement
Lutter 2007	This study assessed ready‐to‐use therapeutic food
Lönnerdal 1994	This study assessed iron and selenium fortification only
Malpeli 2013	This is a pre‐post study without a control group
Manders 2009	This study supplemented a drink; did not assess fortification
Manno 2011	This study did not have an appropriate control group
Matilsky 2009	This study did not have an appropriate control group
McNulty 1996	This study did not have an appropriate control group
Mendez 2012	This study focused on a specific population group (critically ill people)
Mendoza 2004	This study did not have an appropriate control group
Mishaan 2004	This study was a before‐after study without a control group
Mukhopadhyay 2007	This is study assessed point‐of‐use home fortification
Muthayya 2009	This study did not have an appropriate control group
Osei 2010	This is study assessed point‐of‐use home fortification
Ouédraogo 2010	This study provided multiple micronutrient supplement, not fortification
Parker 2015	This study included participants who were already anaemic
Pettifor 1989	This is study assessed point‐of‐use home fortification
Phu 2010	This study assessed ready‐to‐use therapeutic food
Phuka 2008	This study assessed ready‐to‐use therapeutic food.
Phuka 2009	This study assessed ready‐to‐use therapeutic food
Pullakhandam 2011	This study assessed bioavailability in cell cultures
Ramakrishnan 2004	This study assessed supplementation; not fortification
Ramírez‐Silva 2013	This study reported nutrient intakes only
Rohner 2016	This study is a cross‐sectional survey
Rosado 2010	This study did not have an appropriate control group
Schümann 2009	This study assessed supplementation; not fortification
Seal 2008	This is a pre‐post study without a control group
Semba 2011	This study is a cross‐sectional survey
Shatrugna 2006	This study involves calcium supplementation
Stuetz 2012	This is a pre‐post study without a control group
Sun 2011	This is study assessed point‐of‐use home fortification
Tazhibayev 2008	This is a pre‐post study without a control group
Thomas 2012	This study did not have an appropriate control group
Torrejón 2004	This is an observational study
Troesch 2011	This is study assessed point‐of‐use home fortification
Ueland 2007	The study focused on folic acid fortification alone
Unger 2017	This study assessed lipid‐based nutrient supplement
Van Stuijvenberg 2001	This is a pre‐post study without a control group
Varea 2012	This is a pre‐post study without a control group
Varma 2007	This study assessed only 2 micronutrients for fortification (iron and vitamin A)
Yeh 2013	This study assessed supplementation rather than fortification
Zagré 2007	This study assessed supplementation rather than fortification

Differences between protocol and review

• Background has been updated to reflect current information.

• We have modified the search strategy from the protocol, in consultation with the Information Scientist.

• We have now specified that we excluded studies conducted among special population groups including critically‐ill people, anaemic people or people diagnosed with any specific diseases. We also excluded studies on therapeutic blended food and food supplementation.

• We have removed the text pertaining to contacting the trial authors for more data, since this was not needed and hence not done.

• We have removed subgroup analysis by study design (RCTs/non‐RCTs, CBA/ITS).

• We have specified that we have only conducted subgroup analysis where there were at least three studies in each subgroup.

• We have modified the methods used to assess reporting bias to indicate that we used visual assessment rather than statistical tests, and that we explored any asymmetric funnel plots using sensitivity analysis to compare the fixed‐effect and random‐effects meta‐analyses.

• One of our prespecified primary outcomes was anthropometric measures and we had intended to include stunting, wasting and underweight. However, since none of the included studies reported these outcomes, we could not report them. We reported WAZ, HAZ/LAZ and WHZ as anthropometric outcomes, since they were reported in the included studies.

• We could not conduct all of the planned subgroup analyses under every comparison due to a limited number of included studies reporting on the relevant comparisons.

Contributions of authors

All authors contributed to the development of the review. Rohail Kumar (RK), Anoosh Moin (AM), Kashif Mukhtar (KM) and Salman Bin Mahmood (SBM) developed and ran the search strategy and obtained copies of the studies; Jai K Das (JKD), SBM and Rehana A Salam (RAS) selected which studies to include; RK, AM, KM, SBM and Zohra Lassi (ZL) extracted data from studies and entered data into Review Manager 5; SBM, JKD, ZL and RAS carried out and interpreted the analysis. RAS and JKD evaluated the studies according to the PROGRESS‐PLUS criteria. JKD, RAS, SBM, ZL and Zulfiqar A Bhutta (ZAB) drafted the final review, with input from all the authors.

Sources of support

Internal sources

• Aga Khan University, Pakistan.

External sources

• No sources of support supplied

Declarations of interest

JKD: no competing interests.

RAS: no competing interests.

SBM: no competing interests.

AM: no competing interests.

RK: no competing interests.

KM: no competing interests.

ZL: Participated in a Nestlé Nutrition Institute workshop on Health and Nutrition in Adolescents and Young Women: preparing for the next generation for the related publication Nestlé Nutrition Institute Series Volume 80 (2015)

ZAB: Participated in a Nestlé Nutrition Institute workshop on Health and Nutrition in Adolescents and Young Women: preparing for the next generation and co‐edited the related publication Nestlé Nutrition Institute Series Volume 80 (2015). ZAB declares previous travel support from the Nestlé Nutrition Institute for attendance at a meeting on fortification strategies at the University of Winterthur, Winterthur, Switzerland, in October 2011. ZAB received an institutional grant from GAIN on fortification program evidence review in 2014. The paper is currently in press.

Edited (no change to conclusions)