Zinc supplementation for preventing mortality and morbidity, and promoting growth, in children aged 6 months to 12 years of age

  • Protocol
  • Intervention

Authors


Abstract

This is the protocol for a review and there is no abstract. The objectives are as follows:

To assess the effects of zinc supplementation for preventing mortality and morbidity, and promoting growth, in children aged six months to 12 years of age.

Background

Description of the condition

Zinc is an essential micronutrient. Consistent dietary intake is necessary because the human body cannot produce zinc and does not have an adequate mechanism for storing or releasing zinc (Hotz 2004; Maggini 2010). Severe zinc deficiency affects numerous organ systems, including the immune, gastrointestinal, skeletal, reproductive, and central nervous systems (Tuerk 2009). Even marginal deficiency can lead to immune system dysfunction and restricted physical development (Prasad 1963; Shankar 1998). Children are especially vulnerable to deficiency because their periods of rapid growth create increased zinc needs that may remain unmet (Gibson 2006).

Each year, zinc deficiency accounts for approximately 453,207 deaths (4.4% of all mortalities) among children between six months and five years of age (Fischer Walker 2009). Intervention studies suggest that deficiency leads to deaths due to diarrhoea (247,068), pneumonia (117,997), and malaria (88,142), which are the leading causes of mortality in this age group (Bryce 2005; Fischer Walker 2009; WHO 2009; Black 2010). Zinc deficiency also impairs growth and contributes to childhood stunting (Williams 1970; Hess 2009b; Prasad 2009). High stunting prevalence suggests population-level zinc deficiency (Engle-Stone 2007; Hess 2009b).

The global prevalence of zinc deficiency is approximately 31%, and rates of deficiency approach 73% in some regions (Caulfield 2004; Black 2008). Countries in most of South and Southeast Asia, sub-Saharan Africa, and parts of Latin America have relatively high rates of deficiency (Caulfield 2004; Hotz 2004; Black 2008). At the national level, low- and middle-income countries often struggle with food insecurity and poor infrastructure (Smith 2000; Straub 2008). Poor water and sanitation systems lead to frequent exposure to gastrointestinal pathogens and high rates of infectious disease and diarrhoea (Hotz 2004). Morbidity and mortality contribute to reduced economic productivity (Behrman 2004). Thus, the consequences of national-level zinc deficiency can reduce the availability of resources to expand access to zinc-rich foods.

In both low- and high-income countries, zinc deficiency is often related to individual-level poverty (Hotz 2004). Foods from animal sources, which are rich in zinc, are often expensive. Particularly in low-income countries, poor individuals may primarily eat foods such as cereals, grains, and legumes (Hotz 2004). These foods have relatively low concentrations of zinc; they also have relatively high concentrations of fibre and phytate molecules, which reduce zinc absorption by the intestine (Sandstead 1991; Hotz 2004).

In the long term, eliminating deficiency at the national and individual levels requires universal and reliable access to zinc-rich foods (Brown 2009a). Fortification may also be beneficial in areas where food production and processing systems could enrich foods with zinc (Hotz 2004; Hess 2009a). In the short term, supplementation is inexpensive and can be easier to implement than national-level dietary change or fortification (Shrimpton 2005).

Description of the intervention

Zinc supplements come in various physical forms, including liquid solutions, syrups, pills, tablets, capsules, powders, and pastes (Hotz 2004). Supplements also come in various chemical forms, such as zinc sulfate and zinc acetate, with water soluble compounds preferred because they are more efficiently absorbed (Hotz 2004; Brown 2009b). In addition, zinc is sometimes administered with other micronutrients, such as vitamin A or iron (Brown 2009b). Zinc supplements have been provided at various doses, daily and weekly, for a few weeks to over a year (Brown 2009b).

Recommendations for normal zinc consumption among children range between two and nine mg per day, depending on age and diet (Institute of Medicine 2001; Hotz 2004). The World Health Organization (WHO) recommends a supplemental dose of 20 mg per day for 10 to 14 days to treat diarrhoea in children six to 59 months of age (WHO/UNICEF 2004). A dose of 10 mg per day for six months may significantly reduce stunting (Imdad 2011), and five or 10 mg per day may be appropriate for preventive supplementation among children under age 14 (Hotz 2004). However, there are no standard recommendations for dose, frequency, and duration of preventive zinc supplementation.

How the intervention might work

Zinc is in every cell of the human body, and is required for normal functioning. It plays critical catalytic, structural, and regulatory roles (Cousins 1996; Fischer Walker 2004; Tuerk 2009). Zinc enables hundreds of enzymes to function, facilitates protein synthesis and folding, and regulates processes such as gene expression and apoptosis (MacDonald 2000; Hotz 2004; Stefanidou 2006; Aggarwal 2007; Hambidge 2007; Tuerk 2009). Zinc is also important for DNA and RNA metabolism, as well as cellular replication, differentiation, and growth (MacDonald 2000; Stefanidou 2006). Zinc is involved in both non-specific and specific immune system processes, including phagocytosis, maintenance of gastrointestinal and respiratory tract linings, and development and function of T- and B-cells (Shankar 1998). In addition, zinc deficiency is associated with impaired growth hormone function, reduced production of insulin-like growth factor I, and poor appetite (Ploysangam 1997; Cole 2008; Prasad 2009). By increasing the availability of zinc for these biological processes, supplementation may improve health outcomes.

Most importantly, zinc supplementation may reduce all-cause mortality among children by reducing mortality due to diarrhoea, lower respiratory tract infection (LRTI), and malaria. Trials show that preventive supplementation may reduce the incidence of these three morbidities (Bhutta 1999; Brown 2009b). Trials also show that therapeutic supplementation reduces the duration of acute and persistent diarrhoea (Lazzerini 2008). In addition, some trials indicate that zinc supplementation promotes linear growth and weight gain (Brown 2009b).

However, not all trials have found zinc supplementation to be effective (Brown 2009b; Ramakrishnan 2009). In addition, the effects of zinc may be influenced and complicated by several factors. For example, children with more severe deficiency, such as those who are stunted, may benefit more from supplementation than children with less severe deficiency (Umeta 2000). Children with certain chronic diseases and severe protein-energy malnutrition may have different zinc requirements and growth trajectories than children without these co-morbidities (Brown 2002). Supplementation might also affect children with non-chronic illnesses differently than healthy children. For instance, presence of infection generally causes zinc to be sequestered in the liver, and conditions that affect intestinal function and integrity can influence zinc homeostasis (Hotz 2004). Despite such complications, it has been proposed that short-term therapeutic zinc supplementation, such as the kind recommended by the WHO for diarrhoea, might also result in some long-term preventive effects (Haider 2009).

Another set of factors influencing the effects of zinc supplementation are interactions between zinc, iron, and copper. Iron supplementation may interfere with the absorption of zinc and zinc may interfere with iron and copper absorption (Brown 2009b). The evidence is mixed as to whether supplemental zinc contributes to anaemia, iron deficiency, and/or copper deficiency (Fosmire 1990; Brown 2009b). Other potential adverse effects of zinc occur primarily when zinc is given in very high doses (such as 225 to 450 mg) (Fosmire 1990). These adverse effects include abdominal pain, nausea, vomiting, and diarrhoea (Fosmire 1990; Larson 2008).

Why it is important to do this review

Zinc supplementation in children has been investigated in several non-Cochrane Reviews, some of which have had conflicting results (Bhutta 1999; Brown 2002; Aggarwal 2007; Brown 2009b; Ramakrishnan 2009; Roth 2010; Imdad 2011; Yakoob 2011). For example, two such reviews found that zinc supplementation has a significant effect on height and weight gain (Brown 2002; Brown 2009b). However, another recent review found that zinc did not significantly affect these outcomes (Ramakrishnan 2009). This review seeks to resolve such discrepancies.

Previous Cochrane Reviews of the effects of zinc supplementation in children and mothers have focused on outcomes such as otitis media, pneumonia, and the common cold (Mahomed 2007; Ojukwu 2009; Abba 2010; Humphreys 2010; Irlam 2010; Lassi 2010; Lazzerini 2008; Singh 2011). A protocol has also been published for a review examining whether zinc supplementation improves malaria outcomes in areas where the disease is endemic (Okoye 2008). However, zinc supplementation may have multiple and complex effects, and no Cochrane Review has investigated its impact on all-cause mortality as well as the illnesses responsible for a plurality of child deaths worldwide.

Objectives

To assess the effects of zinc supplementation for preventing mortality and morbidity, and promoting growth, in children aged six months to 12 years of age.

Methods

Criteria for considering studies for this review

Types of studies

Randomised controlled trials (RCTs) and cluster RCTs with a parallel group design, in which intervention and control groups were enrolled concurrently. We will exclude quasi-RCTs, such as trials in which allocation is determined by alternation or date of birth.

Types of participants

Children six months to 12 years of age, inclusive.

We will exclude the following:

  • children less than six months of age (the WHO recommends exclusive breastfeeding for children less than six months of age, and trials assessing zinc for lactating mothers will be excluded);

  • hospitalised children;

  • children with any of the following: severe protein-energy malnutrition; HIV; chronic diseases, such as cystic fibrosis and sickle cell disease, or conditions such as Down syndrome that could affect growth.

If some, but not all, of a study’s participants are eligible for our review, then we will ask the study authors for disaggregated data. If we are unable to obtain the appropriate disaggregated data, then we will include a study if the majority (at least 51%) of its participants are eligible for our review. If we are unable to determine the exact percent of a study’s participants who are eligible, then we will include the study if its participants are eligible on average (for example, the mean participant age is less than 13 years).

Types of interventions

Intervention

Orally administered zinc given as a supplement, regardless of compound, formulation, dose, duration, or frequency.

We will exclude the following:

  • food fortification or intake;

  • sprinkles (a review is already being undertaken on this micronutrient delivery method (Vist 2011));

  • trials evaluating the therapeutic effects of zinc (i.e. trials in which children receive zinc while they are ill with diarrhoea, LRTI, or malaria, but stop receiving zinc after recovering from illness).

Comparisons

Placebo, no intervention, and waiting list controls. A control comparison group can have been administered a non-zinc co-intervention (such as a vitamin A supplement), as long as the intervention group to which it is being compared was administered the same co-intervention. Comparisons between two different dosages of zinc (i.e. a high dose and a low dose) will not be eligible; nor will comparisons between different zinc compounds, durations of supplementation, or frequencies at which doses are given.

Types of outcome measures

We will assess the preventive effects of zinc supplementation by extracting data for the following outcomes. In studies reporting more than one measure of an outcome, we will extract measures for meta-analysis using methods described below (see Measures of treatment effect).

Primary outcomes

1. All-cause mortality

2. Cause-specific mortality

            2.1 Due to all-cause diarrhoea

            2.2 Due to lower respiratory tract infection (LRTI, including pneumonia)

            2.3 Due to malaria.

Secondary outcomes

1. Diarrhoea

            1.1 Incidence of all-cause diarrhoea

            1.2 Prevalence of all-cause diarrhoea

            1.3 Hospitalisation due to all-cause diarrhoea

            1.4 Incidence of severe diarrhoea

            1.5 Prevalence of severe diarrhoea

            1.6 Hospitalisation due to severe diarrhoea

            1.7 Incidence of persistent diarrhoea

            1.8 Prevalence of persistent diarrhoea

            1.9 Hospitalisation due to persistent diarrhoea

2. Lower Respiratory Tract Infection

            2.1 Incidence of LRTI (including pneumonia)

            2.2 Prevalence of LRTI

            2.3 Hospitalisation due to LRTI

3. Malaria

            3.1 Incidence of malaria

            3.2 Prevalence of malaria

            3.3 Hospitalisation due to malaria

4. Growth

            4.1 Height

            4.2 Weight

            4.3 Weight-to-height ratio

            4.4 Prevalence of stunting

5. Zinc status

            5.1 Serum or plasma zinc concentration

            5.2 Prevalence of zinc deficiency

6. All-cause hospitalisation

Adverse Events

7. Side effects (for example, abdominal pain, nausea, vomiting, diarrhoea)

8. Haemoglobin status

            8.1 Blood haemoglobin concentration

            8.2 Prevalence of anaemia

9. Iron status

            9.1 Serum or plasma ferritin concentration

            9.2 Prevalence of iron deficiency

10. Copper status

            10.1 Serum or plasma copper concentration

            10.2 Prevalence of copper deficiency

We will include the following outcomes in the 'Summary of findings' table: all-cause mortality, mortality due to all-cause diarrhoea, mortality due to LRTI, mortality due to malaria, incidence of all-cause diarrhoea, incidence of severe diarrhoea, incidence of persistent diarrhoea, incidence of LRTI, incidence of malaria, height, and side effects. Measures of treatment effect describes the measures we will report in the 'Summary of findings' table.

Search methods for identification of studies

Electronic searches

We will search the following databases without date or language restrictions.

  • Cochrane Central Register of Controlled Trials (CENTRAL), part of the Cochrane Library

  • MEDLINE

  • MEDLINE In-Process & Other Non-Indexed Citations

  • EMBASE

  • African Index Medicus

  • Global Health

  • Latin American Caribbean Health Sciences Literature (LILACS)

  • metaRegister of Controlled Trials

  • WHO Library & Information Networks for Knowledge Database (WHOLIS)

  • IndMED

We will use the following search strategy in MEDLINE, and will modify this strategy as needed to search other databases.

1       zinc/ or zinc compounds/ or zinc oxide/ or zinc sulfate/ or zinc acetate/

2       (zinc or Zn).tw.

3       1 or 2

4       exp infant/ or exp child/ or adolescent/

5       (newborn$ or neonat$ or neo-nat$ or infan$ or baby or babies or toddler$ or preschool$ or pre-school$ or pediatric$ or paediatric$ or child$ or girl$ or boy$ or preteen$ or pre-teen$ or teen$ or preadolescen$ or pre-adolescen$ or adolescen$ or prepubert$ or pre-pubert$ or pubert$).tw.

6       4 or 5

7       randomized controlled trial.pt.

8       controlled clinical trial.pt.

9       randomized.ab.

10     placebo.ab.

11     drug therapy.fs.

12     randomly.ab.

13     trial.ab.

14     groups.ab.

15     7 or 8 or 9 or 10 or 11 or 12 or 13 or 14

16     exp animals/ not humans.sh.

17     15 not 16

18     3 and 6 and 17

Searching other resources

Grey literature

We will search the WHO International Clinical Trials Registry Platform (ICTRP) to identify unpublished and ongoing trials. We will search Conference Proceedings Citation Index (formerly known as ISI Proceedings) to identify conference proceedings. We will search the ProQuest Dissertations & Theses Database to identify dissertations and theses.

Reference lists

We will search the reference lists of relevant review articles and included and excluded studies to identify additional studies in the published or unpublished literature.

Correspondence

We will contact the authors of included studies to identify additional studies that are ongoing or unpublished.

Data collection and analysis

Selection of studies

Two authors (JJ and SD) will independently screen the titles and abstracts of all reports yielded by the search to determine which are eligible for inclusion in the review. JJ and SD will obtain and independently screen the full text of all potentially relevant studies to determine whether or not they meet the inclusion criteria. If JJ and SD disagree about the eligibility of a study, then they will discuss the disagreement between themselves and with a third author (EMW) in order to reach a consensus about the study’s eligibility. We will seek additional information from study authors to help clarify any uncertainties about eligibility. During the study selection process, we will not be blinded to study authors, institutions, journal of publication, or results.

Data extraction and management

We will draft a data extraction form to capture the following characteristics of each study:

General

  • Year of study

  • Country

  • Setting (i.e. urban or rural, specific region or city if provided)

  • Unit of analysis (for example, individual or cluster randomisation)

Participants

  • Total number of study participants and clusters

  • Number of study participants and clusters randomised to each included group

  • Age

  • Gender

  • Inclusion and exclusion criteria

  • Comorbidities

For each intervention or comparison group of interest

  • Dose of zinc supplement

  • Duration of zinc supplementation

  • Frequency of zinc supplementation

  • Co-interventions (if any)

For each outcome of interest

  • Time points (i) collected and (ii) reported

  • Missing data (exclusion of participants, attrition)

For each study, we will also extract data on risk of bias (see Assessment of risk of bias in included studies).

Two authors (JJ and SD) will independently pilot the data extraction form on several studies included in the review. After this pilot, we will revise the form. JJ and SD will use the revised form to independently extract data from the rest of the studies. If a disagreement arises about the data extracted, then JJ and SD will discuss the disagreement between themselves and with EMW in order to reach a consensus. We will provide details of the data extracted for each study in a 'Characteristics of included studies' table.

Assessment of risk of bias in included studies

Two authors (JJ and SD) will code each included study using the Cochrane Collaboration’s tool for assessing risk of bias (Higgins 2011). We will use this tool to judge whether each study is at low, high, or unclear risk of bias relating to sequence generation; allocation concealment; blinding of study participants; blinding of personnel; blinding of outcome assessors; incomplete outcome data; selective outcome reporting; and other sources of bias. If a disagreement arises concerning a risk of bias assessment, then JJ and SD will discuss the disagreement between themselves and with EMW in order to reach a consensus. During the risk of bias assessment process, we will not be blinded to study authors, institutions, journal of publication, or results. We will provide details on each study’s risk of bias in a 'Risk of bias' table.

Measures of treatment effect

Studies often report outcomes using multiple definitions and outcome measures. When studies report outcomes in this way, we will do the following.

Multiple outcomes
Diarrhoea

If a trial presents data on diarrhoea overall, undifferentiated by severity level, then we will include this data in meta-analyses for all-cause diarrhoea outcomes. We will do the same for outcomes that trial authors define as acute diarrhoea. If a trial reports severe diarrhoea outcomes, persistent diarrhoea outcomes, and acute or all-cause diarrhoea outcomes, then we will include the average of these outcomes in all-cause diarrhoea meta-analyses. We will define diarrhoea as severe if it is defined this way by trial authors. We will define persistent diarrhoea as lasting 14 or more days.

LRTI

Several systematic reviews have found that LRTI is often defined inconsistently across studies (Bhutta 1999; Imdad 2010; Roth 2010). To deal with this inconsistency, we will only include LRTI data from a study if the study’s LRTI definition matches any of the following definitions. The first two of these definitions can be diagnosed by someone who is not a medical professional (for example, a field worker). The third definition must be diagnosed by a medical professional (e.g. a physician).

  • Difficulty breathing and/or rapid breathing.

  • Difficulty breathing or cough, along with one or more of the following: age-specific rapid breathing rates, lower chest wall indrawing, chest auscultation signs of pneumonia (decreased breath sounds, bronchial breath sounds, crackles, abnormal voice resonance, pleural rub), nasal flaring, grunting, fever, central cyanosis, inability to breastfeed or drink, vomiting everything, convulsions, lethargy, unconsciousness, or severe respiratory distress (for example, head nodding) (WHO 2000; WHO/UNICEF 2005).

  • Clinical evidence of LRTI based on chest auscultation (decreased breath sounds, bronchial breath sounds, crackles, abnormal voice resonance, pleural rub) or chest radiograph.

We will give preference to more severe or rigorously diagnosed LRTI outcome data. For example, if a study provides LRTI data based on rapid breathing and LRTI data based on both rapid breathing and symptoms such as chest indrawing, then we will extract the latter. If a study provides data from caregivers of children and from study field workers, then we will extract the latter. If a study provides fieldworker-reported and physician-reported data, we will extract the latter. If a trial reports LRTI outcomes and pneumonia outcomes separately, then we will use the LRTI outcomes. We will not include upper respiratory tract infection data.

Malaria

As with LRTI, we will give preference to more virulent forms of malaria.

Growth

If a study reports height outcomes as both raw lengths and height-for-age z-scores, then we will preferentially extract the height-for-age z-scores. However, if a study only reports raw lengths, then we will extract these. We will do the same for raw weights and weight-for-age z-scores.

Zinc Status and Adverse Events

We will include measures of zinc deficiency, anaemia, iron deficiency, or copper deficiency, as defined by trial authors. If authors report more than one measure for a particular outcome, then we will give preference to the one defined as most severe.

Multiple outcome measures

To avoid review author bias, we have predetermined the order of preference for extracting outcomes when data are available in several formats.

For studies that randomised individuals, we will give preference to data that requires the least manipulation by authors or inference by review authors. We will extract raw values (for example, means and standard deviations) rather than calculated effect sizes (for example, Cohen’s d). If outcomes are reported as both final values and changes from baseline, then we will preferentially extract the final values. In the case of cluster RCTs, we will (i) use adjusted estimates reported by the authors, or (ii) use raw data and inflate the standard error (SE) using procedures described below.

For mortality data, we will give preference to denominators in the following order: number with definite outcome known (or imputed, as described in Dealing with missing data), number randomised, and child-years. For other dichotomous outcomes to which both survivors and non-survivors may contribute data, we will give preference to child-years, number with definite outcome known, and number randomised.

Summary Measures

Whenever possible, we will use a risk ratio (RR) as the effect measure for each outcome for which there is dichotomous data. For incidence data, we will combine risk ratios (events per child) and rate ratios (events per child year), because these ratios use the same scale and can be interpreted in the same way for these studies. Since we expect the duration of studies to be short, we do not anticipate interaction between the intervention and time at risk. We will estimate time at risk if appropriate, as when authors report incidence rate, study duration, and number of children in a group. We will use Hedges’ (adjusted) g (a standardized mean difference) for each outcome for which there is continuous data. We will use a hazard ratio as the effect measure for each outcome for which there is time-to-event (survival) data. We will report all outcomes with a 95% confidence interval.

Unit of analysis issues

Cluster-randomised trials

Cluster-randomised trials randomise groups of people rather than individuals. For each cluster-randomised trial, we will first determine whether or not its data incorporate sufficient controls for clustering (such as robust standard errors or hierarchical linear models). If the data do not have proper controls, then we will attempt to obtain an appropriate estimate of the data’s intra-cluster correlation coefficient (ICC). If we cannot find an estimate in the report of the trial, then we will request an estimate from the trial report authors. If the authors do not provide an estimate, then we will obtain one from a similar study and conduct a sensitivity analysis to determine if the results are robust when different values are imputed. We will use the ICC estimate to control the trial’s data for clustering, according to procedures described in Higgins 2011. This process will prevent meta-analyses from being based on clustered data that have not been properly controlled.

Cross-over trials

For cross-over trials, we will extract and analyse data from the first period only.

Studies with multiple treatment groups

For factorial studies, we will include all comparisons that differ only in the presence or absence of zinc. For example, in a 2x2 factorial study of zinc and vitamin A supplementation, we will include two comparisons: (1) zinc versus placebo, and (2) zinc and vitamin A versus vitamin A alone. For other studies, multiple eligible intervention groups will be combined.

Outcomes measured at multiple time points

For outcomes measured at multiple time points, we will only include the time point that occurred the most days after randomisation in our meta-analyses.

Dealing with missing data

Missing data, and methods for imputing such data, may affect the magnitude and direction of a point estimate and its standard error. For all analyses, we will attempt to include all randomised study participants. When analyses are reported for completers as well as controlling for dropout (for example, imputed using regression methods), we will extract the latter. If data are missing for some cases, or if reasons for dropout are not reported, then we will contact study authors to request missing data and further information on dropouts.

For the primary outcome, data are likely to be missing at random. Secondary outcome data may be missing for reasons related to group assignment (for example, early mortality in the comparison group). We will report reasons for missing data, including reasons for dropout, and number of dropouts. We will also conduct a sensitivity analysis to evaluate the potential impact of incomplete outcome data on the findings of this review (see Sensitivity analysis). Finally, we will discuss the potential impact of missing data on review findings.

Assessment of heterogeneity

We will discuss the similarities and differences between included studies in terms of their participants, interventions, outcomes, and methods. For each meta-analysis, we will use three methods to identify statistical heterogeneity: visually inspecting forest plots to see if the confidence intervals of individual studies have poor overlap – a rough indication of statistical heterogeneity; conducting a Chi2 test, and calculating an I2 statistic. We will deem a meta-analysis to have substantial heterogeneity if its Chi2 P value is less than 0.10 and its I2 statistic is greater than 50%.

Assessment of reporting biases

We will create a funnel plot for each meta-analysis that includes 10 or more studies, and look to see if any funnel plot appears asymmetrical. We will judge a meta-analysis with an asymmetrical funnel plot to be potentially biased by small-study effects or reporting bias. For all analyses, we will compare the combined effects calculated using a fixed-effect model to the combined effects calculated using a random-effects model (see Sensitivity analysis).

Data synthesis

We will use Review Manager (RevMan) Version 5.1 software (Review Manager 2011) to conduct all meta-analyses. We will use Mantel-Haenszel methods to meta-analyse dichotomous data that can be combined directly in RevMan. If studies report dichotomous data in multiple formats that cannot be combined in RevMan, we will use Comprehensive Meta-Analysis Version 2 software (Borenstein 2005) to calculate log risk ratios and standard errors for the data, enter these log risk ratios and standard errors into RevMan, and then meta-analyse these using inverse-variance methods. We will also use the inverse-variance method to meta-analyse continuous and time-to-event data. We will use fixed-effect methods for all meta-analyses. Although there may be some differences across trials (for example, dose and population), the biological mechanism should be similar across trials. However, we will conduct a sensitivity analysis in which random-effects methods are used (see Sensitivity analysis).

Subgroup analysis and investigation of heterogeneity

We plan to conduct the following subgroup analyses for outcomes with at least 10 studies measuring the relevant characteristic. We will report a Chi2 test for each analysis to determine whether or not the effects of zinc are statistically significantly different for different subgroups.

  1. Country Income Level: low- and middle-income countries versus high-income countries, as defined by the World Bank’s country income classification system (World Bank 2011);

  2. Age: children six months to under five years versus five years to 13 years;

  3. Stunting: children with a height-for-age z-score of < -2 versus children with a height-for-age z-score ≥ -2;

  4. Dose: daily dose equivalent less than five mg per day, versus five mg to under 10 mg, versus 10 mg to under 15 mg, versus 15 mg to under 20 mg, versus 20 mg or more per day;

  5. Duration: supplementation lasting zero to five months, versus six to 11 months, versus 12 months or more;

  6. Iron co-Interventions: iron+Zinc versus iron alone, zinc versus no-supplementation;

  7. Formulation: solution versus pill versus capsule.

Sensitivity analysis

We will conduct the following sensitivity analyses to examine whether or not our findings are robust to certain decisions we made while conducting the review.

  1. We will repeat the analyses using random-effects methods.

  2. We will repeat the primary meta-analysis excluding studies at high risk of bias due to incomplete outcome data.

  3. We will repeat the primary meta-analysis if it includes one or more cluster-randomised trials for which we had to impute the ICC. We will repeat the meta-analysis using an ICC value at least as large as the largest observed ICC.

Acknowledgements

We thank the Cochrane Developmental Psychosocial and Learning Problems Group for help with the preparation of this protocol, the Cochrane Methods Group for statistical advice and assistance, and anonymous peer reviewers for their feedback.

History

Protocol first published: Issue 10, 2011

Contributions of authors

JJ, SD, AI, EMW contributed to the background. EMW and JJ were primarily responsible for the methods. JJ, EMW, and AI developed the search strategy.

Declarations of interest

Jean A Junior - none known.
Sohni Dean - none known.
Evan Mayo-Wilson - none known.
Aamer Imdad - none known.
Zulfiqar A Bhutta - none known.

Sources of support

Internal sources

  • Aga Khan University, Pakistan.

    Aamer Imdad, Sohni Dean, and Zulfiqar A Bhutta are supported by Aga Khan University, Karachi, Pakistan.

  • The Centre for Evidence-Based Intervention, UK.

    Jean Junior and Evan Mayo-Wilson are supported by the Centre for Evidence-Based Intervention, University of Oxford, UK.

External sources

  • No sources of support supplied

Ancillary