Primary outcomes

1. Stunting (moderate: height‐for‐age (HAZ) below ‐2 standard deviations (SD); severe: HAZ below ‐3 SD)

2. Wasting (moderate: weight‐for‐height (WFH) below ‐2 SD; severe: WFH below ‐3 SD)

3. Underweight (moderate: weight‐for‐age (WFA) below ‐2 SD; severe: WFA below ‐3 SD)

4. Anemia (as defined by trialists)

5. Psychomotor development outcomes (as defined by trialists)

6. Neuro‐developmental outcomes (as defined by trialists)

7. Any adverse effects, including allergic reactions, as diagnosed by clinical assessment (atopic dermatitis, urticaria, oedema (oral), ophthalmic pruritus, allergic rhinitis, asthma, anaphylaxis)

Secondary outcomes

1. Mid upper arm circumference (MUAC; the circumference of the left upper arm, measured at the mid‐point between the tip of the shoulder and the tip of the elbow)

2. Haemoglobin (g/L)

3. Morbidity (incidence of diarrhoea, acute respiratory illness (ARI) and fever as defined by trialists)

4. Mortality

International databases

1. Cochrane Central Register of Controlled Trials (CENTRAL; current issue) in the Cochrane Library, and which includes the Cochrane Developmental, Psychosocial and Learning Problems Specialised Register.

2. MEDLINE Ovid (1946 onwards).

3. MEDLINE In‐Process and Other Non‐Indexed Citations Ovid (current issue).

4. MEDLINE E‐pub ahead of print Ovid (current issue).

5. Embase Ovid (1980 onwards).

6. CINAHL EBSCOhost (Cumulative Index to Nursing and Allied Health Literature; 1937 onwards).

7. Social Sciences Citation Index Web of Science (SSCI; 1970 onwards).

8. Science Citation Index Web of Science (SCI; 1970 onwards).

9. Conference Proceedings Citation Index ‐ Science Web of Science (CPCI‐S; 1990 onwards).

10. Conference Proceedings Citation Index ‐ Social Science & Humanities Web of Science (CPCI‐SS&H; 1990 onwards).

11. Cochrane Database of Systematic Reviews (CDSR; current issue), part of the Cochrane Library.

12. Epistemonikos (epistemonikos.org; current issue).

13. POPLINE (www.popline.org, current issue).

14. ClinicalTrials.gov (clinicaltrials.gov).

15. World Health Organization International Clinical Trials Registry Platform (WHO ICTRP; who.int/trialsearch).

Regional databases

1. IBECS (Índice Bibliográfico Español en Ciencias de la Salud; ibecs.isciii.es; current issue).

2. Scieclo (Scientific Electronic Library Online; www.scielo.br; current issue).

3. AIM Africa Global Index Medicus (Africa Index Medicus; search.bvsalud.org/ghl/?lang=en&submit=Search&where=REGIONAL; current issue).

4. IMEMR Global Index Medicus (Index Medicus for the Eastern Mediterranean Region; search.bvsalud.org/ghl/?lang=en&submit=Search&where=REGIONAL; current issue).

5. LILACS (Latin American and Caribbean Health Sciences Literature; lilacs.bvsalud.org/en; current issue).

6. PAHO/WHO Institutional Repository for Information Sharing (iris.paho.org/xmlui; current issue).

7. WHOLIS Global Index Medicus ( WHO Library Database; search.bvsalud.org/ghl/?lang=en&submit=Search&where=REGIONAL; current issue).

8. WPRIM Global Index Medicus (Western Pacific Index Medicus; search.bvsalud.org/ghl/?lang=en&submit=Search&where=REGIONAL; current issue).

9. IMSEAR Global Index Medicus (Index Medicus for the South‐East Asian Region; search.bvsalud.org/ghl/?lang=en&submit=Search&where=REGIONAL; current issue).

10. IndMED (indmed.nic.in/indmed.html; current issue).

11. Native Health Research Database (hscssl.unm.edu/nhd; current issue).

We will search using both keyword and controlled vocabulary (when available), using the search terms set out in Appendix 1. We will not apply language or date restrictions for any database. If we identify studies written in a language other than English, we will commission their translation into English. We will record any such studies as 'Studies awaiting classification' until a translation becomes available.

(1) Random sequence generation (checking for selection bias)

We will assess whether the method used to generate the allocation sequence is described in sufficient detail to allow an assessment of whether it produces comparable groups.

• Low risk of bias (any truly random process, for example, random number table; computer random number generator).

• High risk of bias (any non‐random process, for example, odd or even date of birth; hospital or clinic record number).

• Unclear risk of bias (where there is insufficient information provided to permit judgement of high or low risk of bias).

(2) Allocation concealment (checking for possible selection bias)

We will assess whether the method used to conceal the allocation sequence is described in sufficient detail to determine whether intervention allocations could have been foreseen in advance of, or during, enrolment.

• Low risk of bias (for example, telephone or central randomisation; consecutively numbered, sealed, opaque envelopes).

• High risk of bias (open random allocation; unsealed or non‐opaque envelopes).

• Unclear risk of bias (where there is insufficient information provided to permit judgement of high or low risk of bias).

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

We will describe all measures used, if any, to blind study participants and personnel from knowledge of which intervention a participant received.

We will assess the risk of performance bias associated with blinding of participants and personnel as follows.

• Low risk of bias (blinding of participants and personnel and unlikely that the blinding could have been broken, or no blinding or incomplete blinding but outcome unlikely to be influenced).

• High risk of bias (participants and personnel not blinded, incomplete or broken blinding, and outcome likely to be influenced).

• Unclear risk of bias for participants and personnel (where there is insufficient information provided to permit judgement of high or low risk of bias).

Whilst assessed separately, we will combine the results into a single evaluation of risk of bias associated with blinding (Higgins 2011a).

(4) Blinding of outcome assessment (checking for possible detection bias)

We will describe all measures used, if any, to blind outcome assessors from knowledge as to which intervention a participant received.

• Low risk of bias (blinding of outcome assessment and unlikely that the blinding could have been broken, or no blinding but measurement unlikely to have been influenced).

• High risk of bias (for example, no blinding of outcome assessment, where measurement is likely to be influenced by lack of blinding, or where blinding could have been broken).

• Unclear risk of bias (where there is insufficient information provided to permit judgement of high or low risk of bias).

(5) Incomplete outcome data (checking for possible attrition bias through withdrawals, dropouts, protocol deviations)

We will assess the outcomes in each included study as follows.

• Low risk of bias (either there are no missing outcome data or the missing outcome data are unlikely to bias the results based on the following considerations: study authors provide transparent documentation of participant flow throughout the study, the proportion of missing data is similar in the intervention and control groups, the reasons for missing data are provided and balanced across intervention and control groups, or the reasons for missing data are not likely to bias the results (for example, moving house)).

• High risk of bias (missing outcome data are likely to bias the results. Trials will also receive this rating if an 'as‐treated (per protocol)' analysis is performed with substantial differences between the intervention received and that assigned at randomisation, or if potentially inappropriate methods for imputation have been used).

• Unclear risk of bias (where there is insufficient information provided to permit judgement of high or low risk of bias).

(6) Selective outcome reporting (checking for possible reporting bias)

Selective reporting can lead to reporting bias. We will compare methods to results and look for outcomes measured (or likely to have been measured) but not reported.

• 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, are reported).

• High risk of bias (where not all the study’s prespecified outcomes have been reported; one or more reported primary outcomes are not prespecified; outcomes of interest are reported incompletely and so cannot be used; the study fails to include the results of a key outcome that would have been expected to have been reported).

• Unclear risk of bias (where there is insufficient information provided to permit judgement of high or low risk of bias).

(7) Other sources of bias (checking for other possible sources of bias not covered by the domains above)

We will assess if the study was free of other potential bias, including industry funding bias or conflict of interest in primary studies, as follows.

• Low risk of bias (where there is similarity between outcome measures at baseline, similarity between potential confounding variables at baseline, or adequate protection of study arms against contamination).

• High risk of bias (where there is no similarity between outcome measures at baseline, similarity between potential confounding variables at baseline, or adequate protection of study arms against contamination).

• Unclear risk of bias (where there is insufficient information provided to permit judgement of high or low risk of bias).

(8) Overall risk of bias

We will summarise the risk of bias at two levels: within trials (across domains), and across trials using GRADE (GRADEpro 2014).

We will assess the likely magnitude and direction of bias in each of the above‐mentioned domains and whether we consider they are likely to impact on the findings. We will consider trials at high risk of bias if all the individual domains for 'Risk of bias' assessment are assessed as high risk, and at overall low risk of bias if all the domains are assessed as low risk of bias. We will consider trials at unclear risk of bias if the information in majority of the domains is unclear. We will explore the impact of the level of bias through a Sensitivity analysis.

For non‐randomised studies, we will use the Risk of Bias in Non‐randomized Studies – of Interventions (ROBINS‐I) assessment tool to assess: bias due to confounding, bias in selection of participants into the study, bias in classification of interventions, bias due to deviations from intended interventions, bias due to missing data, bias in measurement of outcomes, bias in selection of the reported result, and overall bias (see sites.google.com/site/riskofbiastool/welcome/home).

Dichotomous data

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

Continuous data

For continuous data, we will use the mean difference (MD) with 95% CI, if outcomes are measured in the same way between trials. We will use the standardised mean difference (SMD) with 95% CI to combine trials that measure the same outcome but use different measurement methods.

When some trials report endpoint data and other reports change from baseline data (with errors), we will combine these in the meta‐analysis if the outcomes are reported using the same scale.

Rates

For rates, if they represent events that could have occurred more than once per participant, we will report the rate difference using the methodologies described in Deeks 2011.

Cluster‐randomised trials

We will adjust their sample sizes or standard errors using the methods described in theCochrane Handbook using an estimate of the intracluster correlation co‐efficient (ICC) derived from the trial (if possible), from a similar trial or from a study of a similar population (Higgins 2011b). If we use ICCs from other sources, we will report this and conduct sensitivity analyses to investigate the effect of variation in the ICC (see Sensitivity analysis). If we identify both cluster‐randomised trials and individually‐randomised trials, we plan to synthesise the relevant information. We will consider it reasonable to combine the results from both if there is little heterogeneity between the study designs, and the interaction between the effect of intervention and the choice of randomisation unit is considered to be unlikely. We will also acknowledge heterogeneity in the randomisation unit and perform a sensitivity analysis to investigate the effects of the randomisation unit (Sensitivity analysis).

Trials with more than two treatment groups

For trials with more than two intervention groups (multi‐arm trials), we will include the directly relevant arms only. If we identify trials with various relevant arms, we will combine the groups into a single pair‐wise comparison (Higgins 2011b), and include the disaggregated data in the corresponding subgroup category. If the control group is shared by two or more study arms, we will divide the control group (events and total population) over the number of relevant subgroup categories to avoid double counting the participants. We will note the details in the 'Characteristics of included studies' tables.

Summary of findings

For the assessment across trials, we will set out the main findings of the review in 'Summary of findings' table(s), prepared using GRADE software (GRADEpro 2014). We will list the primary outcomes for each comparison with estimates of relative effects along with the number of participants and trials contributing data for those outcomes; these outcomes would be stunting, wasting, underweight, anaemia, psychomotor development and neuro‐developmental outcomes. For each individual outcome, we will assess the quality of the evidence using the GRADE approach (Balshem 2010), which involves consideration of within‐study risk of bias (methodological quality), directness of evidence, heterogeneity, precision of effect estimates and risk of publication bias, and results in one out of four levels of quality (high, moderate, low or very low). This assessment will be limited only to the trials included in the review.