Deworming and adjuvant interventions for improving the developmental health and well-being of children in low-and middle-income countries : a systematic review and network meta-analysis

......................................................................................................................... 13 Plain Language Summary ......................................................................................... 17 Summary of findings tables ...................................................................................... 19 1 Background ............................................................................................................ 24


Disease burden
The burden of disease of soil-transmitted helminths and schistosomes was estimated at almost 3 million disability adjusted life years DALYS globally in 2004 according to the World Health Organization's (WHO) Global Burden of Disease (1).The neglected tropical diseases of soiltransmitted helminthiasis and schistosomiasis affect more than a third of the world's population.These infections rarely cause death, and therefore the burden is predominantly due to morbidity.Infections by worms affect the nutritional status of children through various mechanisms, such as feeding on host tissue and interfering with absorption of nutrients (2).These mechanisms lead to anaemia and related micronutrient deficiencies which may contribute to impaired growth, cause fatigue and further hinder school attendance (3,4).Moreover, reduced school attendance at early ages has major implications throughout the life course and can impede upon labour market outcomes, maintaining cycles of poverty and worsening health equity gaps (5).
The four species of soil-transmitted worms most commonly associated with malnutrition and disease in children are: Ascaris lumbricoides (roundworm), Trichuris trichura (whipworm), Ancylostoma duodenale and Necator americanus (hookworms).These worms cause infection through ingestion of eggs from contaminated soil, food (e.g.vegetables) or water or active penetration of the skin by larvae in soil.All four of these parasites are found in areas where there are poor sanitation practices and are often linked to areas of poverty; factors associated with worm infections are poor hygiene practices, poor sanitation and lack of shoes (6).
Schistosomiasis (also known as Bilharzia) is a disease that affects over 200 million people, with over 95% of infections in Africa (7).Schistosomiasis is caused by trematodes (parasitic worms commonly called blood flukes).These worms cause infection when larvae, released by fresh water snails, penetrate host skin during contact with infested water.Like soil-transmitted helminths, schistosomiasis is found in areas with poor sanitation and poor access to safe drinking water.It is also prevalent in areas where there are bodies of fresh water, which is the habitat of the intermediate host, the snail.There are five species of schistosomiasis which infect humans: Schistosoma mansoni, Schistosoma japonicum, Schistosoma mekongi, Schistosoma intercalatum (which cause intestinal schistosomiasis) and Schistosoma haematobium (which causes urinary schistosomiasis).Schistosomiasis was previously thought to occur mainly in school-age children and above because younger children were less likely to be swimming in water bodies, but recent epidemiological monitoring data show that infants and preschool-aged children are also at risk of the disease (8).Schistosome infection is hypothesized to cause iron deficiency anaemia, growth stunting and cognitive impairment, which lead to reduced school attendance, and this hypothesized effect may be mediated by the burden of infection as well as other factors (9).

Pharmacologic interventions
To combat the burden of helminths, the World Health Organization recommends concurrent deworming for soil-transmitted helminths and schistosomiasis in endemic areas, combined with improved sanitation and health education to sustain the effect of deworming and reduce reinfection rates.Pharmacologic therapy may be applied by: 1) mass drug administration to whole communities, 2) targeted pharmacologic treatment of high risk populations (e.g.schoolchildren), or 3) selective pharmacologic treatment of infected individuals (10).Deworming of children has been described as the most cost-effective strategy for improving educational attendance in low-and middle-income countries (11).In the last 10 years, the effects of deworming have been suggested to be improved by synergistic effects with other interventions such as hygiene promotion, iron or vitamin A supplementation or feeding programmes (12)(13)(14)(15)(16).
The treatments for deworming are inexpensive, and because of the safety of the drugs, no medical staff is required for administration.Generic forms of albendazole and mebendazole which are most commonly used to treat soil-transmitted helminths, and praziquantel which is used to treat schistosomiasis, cost less than $0.02USD per dose and have been donated at large to endemic regions since the 1980's from respective drug companies (17).Either albendazole or mebendazole can be administered together with praziquantel.According to the WHO guidelines on deworming, chemopreventive therapy for soil-transmitted helminths with albendazole or mebendazole should be administered once a year where prevalence rates are ≥20 per cent, and twice a year where prevalence rates are ≥50 per cent (18).Schistosomiasis chemoprevention with praziquantel is dependent on the prevalence of infection in school-aged children, with treatment being yearly in high-risk communities (prevalence ≥ 50 %), once every two years in medium risk communities (prevalence ≥ 10% but < 50%) and twice during primary schooling age in low risk communities (prevalence <10 %) (19).This frequency of treatment is to prevent re-infection and maintain low worm burden in persons treated.
These drugs have been proven to be effective in reducing worm loads in school-aged children.The anthelminthic action of praziquantel is not dependant on the location of the parasites within the body.The drug enhances Ca2+ permeability, which leads to an influx of Ca2+ and spastic paralysis in schistosomes.When damage is sufficient to the syncytial tegument, the resulting influx of Ca2+ will disrupt any processes using this ion, which results in parasite death (20).Praziquantel is not effective against soil-transmitted helminths due to differences in morphology.Albendazole and mebendazole are effective in treating infections caused by soil-transmitted helminths.These drugs interrupt microtubule polymerisation, by binding to parasite's β-tubulin in the mitochondria of the worms, leading to the deaths of adult worms in the host (21).Serious adverse effects such as allergic reactions occur in less than one in 1000 treatments.Most adverse effects are mild and short lived, and occur mainly in those infected, which implies that these symptoms are a result of the worms dying.The effects include abdominal pain, headache and nausea.
The issue of potential drug resistance to antihelminthic treatment has received greater attention in more recent years (22), yet the evidence remains inconclusive, with few studies examining the effects among human populations (23).Drug resistance is defined by WHO (1996) when egg reduction rates are less than 70 per cent for ascaris and less than 50 per cent for trichuris post-albendazole treatment however this is often difficult to measure without consistent reporting times between trials (24).Given the current uncertainty around the longer-term implications of mass preventive chemotherapy, it is necessary to ensure ongoing surveillance of worm prevalence and reinfection rates alongside treatment programs.Other authors suggest focusing on the sustainability of funding for studies to ensure consistency and availability of treatment (25) and on interventions that target behavioural change and infrastructure such as hygiene, education and sanitation (26).

Concurrent interventions
In addition to deworming, the WHO and World Bank as well as several other international organizations, such as the World Food Programme and Deworm the World, propose providing feeding or micronutrients (such as vitamin A, iron or multiple micronutrient supplements) in conjunction with deworming (e.g., http://www.povertyactionlab.org/scale-ups/school-baseddeworming).The World Bank recommends deworming as part of the FRESH strategy (Focusing Resources on Effective School Health) (27).Some international organizations that provide aid to developing countries such as UNICEF are increasingly taking an integrated approach to providing deworming in combination with other preventive strategies such as vitamin A, insecticide treated bed nets and immunization.These integrated approaches may produce economies of scale in providing multiple interventions together and increase attendance to the clinics or health days because parents value the number of interventions.

HOW THE INTERVENTION MIGHT WORK
Because this systematic review involves multi-component interventions, we have developed a logic model to elucidate the causal chain from worm infection to nutritional status and educational effects, how deworming in combination with other strategies intervene in these causal pathways, and which factors are important in moderating these effects (Figure 1) (28).This logic model demonstrates generic relationships.However, different worms have specific effects that will be investigated in the interpretation of results.For example, hookworm (Necator americanus and Ancylostoma duodenale) is the only worm expected to cause iron deficiency anaemia by feeding on the hosts' blood.
Deworming treatment, which involves chemopreventive therapy for soil-transmitted helminths or schistosomiasis, depending on endemicity, is administered directly to the child.With many deworming treatment programs taking place in community-based settings, there is the potential to also see a spill-over effect of decreased worm burden among untreated children and their household members.This occurs because there is a lower risk of untreated children contracting the disease from treated children.
Hygiene education and promotion and sanitation programs are designed to reduce the likelihood of reinfection by reducing worm burden in water and surrounding soil, thus reducing the exposure of children to worms in both water and soil.Thus, hygiene education and sanitation interventions are hypothesized to increase the duration of effect, and may also increase the effect size in the long term.
Micronutrient supplementation, including vitamin A, iron or multiple micronutrients (which may also include folic acid, iodine, vitamin B1, vitamin B2, vitamin B6,vitamin B12, vitamin C, copper, niacin, zinc and selenium) (29) aim to reduce anaemia or improve nutrient absorption and therefore nutritional status.These nutritional co-interventions are hypothesized to have synergistic effects.For example, feeding programs such as school feeding or micronutrient supplementation may correct nutritional deficiencies hypothesized to be caused by worms.
Mediating factors affecting the causal pathway include poverty, under-nutrition, hygiene, sanitation, prevalence and intensity of infection and co-infections.With successful implementation and uptake of interventions and co-interventions, intended effects are expected in improved well-being, growth, cognitive development, and educational performance.Overall, deworming programming may improve health equity by benefiting the poorest individuals who are at greatest risk for exposure to worms (e.g. through poor sanitation) and most vulnerable to infection (e.g.due to poor nutritional status).

Figure 1. Proposed logic model (based on CDC community guide methods)
The Campbell Collaboration | www.campbellcollaboration.org

Rationale and Previous Systematic Reviews
A recent Cochrane Collaboration systematic review on deworming for soil-transmitted helminths (30) concluded that deworming programmes may be effective in relation to weight gain in some circumstances but not others, and found no effects on school performance (31).Previous versions of this Cochrane review were criticized in the literature for four main reasons.Firstly, it did not address the possibility of treatment externalities or spill-over effects for individuals that do not receive treatment (both targeted individuals and household members) may experience a reduction in infection and reinfection rates because of their exposure to treated individuals in the same community or school (32).Secondly, the lack of effect on cognitive outcomes was not interpreted in the context of poverty, health status and the learning environment in these studies.Thirdly, effects on school attendance did not consider the validity of school attendance records (used in the studies included in the Taylor-Robinson review) compared to on-site checks (32).Fourthly, this Cochrane review excluded trials of soil-transmitted helminth treatment combined with other interventions unless these interventions were also given in the comparison group.Thus, it does not assess the effects of concurrent schistosomiasis deworming in endemic areas as recommended in the WHO guidelines, nor does it assess the effects of adjuvant interventions, such as nutritional, sanitation or hygiene interventions (9).The 2012 update of this Cochrane review was revised to consider some of these comments.For example, they obtained data on the children in the study by Miguel and Kremer (33) that received only albendazole.The updated review also describes that cluster trials are most likelt to be able to assess whether treatment externalities exist.However, because of the focus on the single treatment for soil-transmitted helminths, this review legitimately does not include treatment groups where praziquantel for schistosomiasis or any other cointervention was included in only one of the treatment arms.
A second review (2) assessed the effects of soil-transmitted helminth treatment (albendazole, mebendazole, pyrantel or piperazone) alone on weight, height, mid-upper arm circumference, skinfold thickness and haemoglobin.Authors found statistically significant improvements in growth measurements but not in haemoglobin levels, when compared with control groups; no cointerventions were taken into consideration.A third review (34) found that concurrent treatment for both soil-transmitted helminths and schistosomiasis was more effective at reducing anaemia than albendazole alone, however this review did not assess the interventions' impact on growth or educational performance.
Thus, there is no systematic review which assesses the impact of the proposed WHO guideline policy of deworming for both soil-transmitted helminths and schistosomiasis on school attendance, growth, well-being and adverse effects, nor the effects of combining with nutrition, hygiene or sanitation interventions.
This systematic review aims to build upon previous systematic reviews conducted on soiltransmitted helminth interventions and address the four criticisms raised in response to the Taylor-Robinson et al. review (19).We will assess: 1) the effects of deworming according to the WHO guidelines (i.e. for both soil-transmitted helminths and schistosomiasis, according to prevalence) as well as synergistic effects of co-interventions (such as micronutrients, feeding, and hygiene and/or sanitation interventions); 2) treatment externalities (spill-over effects) for untreated children living in endemic areas (see above); 3) the effects on school attendance and participation in light of validity of different measures of attendance, and 4) the effects on cognition, taking into account the learning environment, contextualized in areas of poverty and under-nutrition.Long-term effects on weight, height, education and labour market outcomes will be taken into account by including nonrandomised studies.We will also assess effects on HIV, tuberculosis and malaria as secondary outcomes since the WHO Partners for Parasite Control hypothesize that HIV, malaria and tuberculosis outcomes are influenced by deworming (10).These latter outcomes will only be included when studies reporting effects on nutrition or education report them.
In order to assess the effects of multiple components, complex interventions, we will conduct a mixed treatment comparison systematic review and meta-analysis.Mixed treatment comparisons allow the comparison of treatments that have not been compared directly in studies by using indirect comparisons to a common comparator or to other interventions.The advantage of the multiple treatment comparisons approach is that it allows: 1) the assessment of heterogeneity due to multiple components of a complex intervention (35,36) (i.e.hygiene education, sanitation, micronutrients, feeding programs and type of deworming); 2) identification of areas in the network where evidence is limited (e.g.there are likely to be fewer studies of schistosomiasis control because schistosomiasis is endemic in fewer regions, mainly Africa and South-East Asia) and 3) meta-regression in a mixed treatment comparison systematic review allows more complete consideration of covariates (such as nutritional status and intensity of worm infection).By taking these factors into consideration, we will assess the synergistic effects of deworming and other programmes such as school feeding, micronutrient supplements and hygiene promotion, as well as assessing the effects on other outcomes such as HIV, tuberculosis and malaria burden (37).
We consider the inclusion of schistosomiasis treatment to be a crucial component to this systematic review, since the previous Cochrane review of deworming excluded schistosomiasis treatment unless given to both groups.Schistosomes commonly cause anaemia affecting growth and development (20), treatment outcomes differ from those of soil-transmitted helminths (42), and prevalence rates overlap with soil-transmitted helminth endemic areas (43).Treatment of children infected with schistosomiasis has been shown to improve educational outcomes (44).Thus, our review assesses the effects of the deworming strategy recommended by the WHO, which includes pharmacologic therapy with praziquantel for schistosomiasis where prevalence of schistosomiasis is greater than ten per cent.
Two other unique characteristics of this systematic review are: 1) the inclusion of quasiexperimental studies to capture long-term effects on growth and education as well as providing additional data on treatment externalities if treatment and control arms are conducted in different communities; and 2) the assessment of treatment externalities by comparing effects in cluster allocated studies (since cluster allocated studies should have less spill-over effects if the treated and control children are in different communities).

Objective of the review
This systematic review will evaluate the effects of the WHO policy on deworming for soiltransmitted helminths and schistosomiasis in conjunction with other co-interventions (such as hygiene promotion, school feeding and micronutrients) in children (6 months to 16 years) in lowand middle-income countries on growth, educational status, cognition, well-being and adverse effects.
We will assess the evidence base for deworming as a complex intervention that takes into consideration context, synergistic effects, treatment externalities and the internal validity of educational status evaluations.
Research questions: 1. What is the effect of deworming according to the WHO guidelines compared to placebo (or control) in children (6 months to 16 years) in low-and middle-income countries on growth, educational status, cognition, well-being and adverse effects?
2. What is the effect of deworming for soil-transmitted helminths only or schistosomiasis only compared to the combination approach of deworming for both schistosomiasis and geohelminths, in children (6 months to 16 years) where both are endemic in low-and middle-income countries on growth, educational status, cognition, well-being and adverse effects?
3. What is the effect of adding hygiene education, sanitation, micronutrients or feeding programs compared to deworming alone in children (6 months to 16 years) in low-and middle-income countries on growth, educational status, cognition, well-being and adverse effects?
4. What factors, either confounding or effect modifiers (such as worm endemicity, infection intensity, baseline nutritional status, child age and sex, and spill-over effects) contribute to heterogeneity of effect?Definitive evidence on the effects of deworming in combination with other interventions such as feeding, nutritional supplements and hygiene promotion on educational and health outcomes is critical for those who make decisions about funding these programs.

Participants
 Children from six months to 16 years of age in worm endemic areas in low and middle income countries (LMICs) as defined by the World Bank; at risk of infection from Ascaris lumbricoides (roundworm), Trichuris trichura (whipworm), Ancylostoma duodenale and Necator americanus (hookworms) or schistosomiasis.

Interventions
 Mass drug administration or targeted chemoprevention which follows the WHO guidelines for treatment for soil-transmitted helminths and schistosomiasis according to prevalence (see Table 1).The programs may be administered in any location, such as schools, health facilities, community centres or through community outreach;  Any of the commonly used drugs for soil-transmitted helminths or schistosomiasis: albendazole, mebendazole, ivermectin, pyrantel and levamisole and praziquantel in the appropriate dose; or, for other pharmacologic treatments, we will include any licensed drugs, at the appropriate dose levels;  May also include cointerventions of hygiene promotion and education, sanitation improvements (e.g.water treatment), micronutrients (e.g.vitamin A, iron or multiple micronutrients such as iron, folic acid, iodine, vitamin C, vitamin B1, B2, B6, B12, niacin, copper, zinc and selenium) or feeding programs.
We will include all studies that include an arm of deworming according to WHO guidelines for either geohelminths or schistosomiasis (or both) and meet other inclusion criteria.We will assess endemicity of schistosoma and soil-transmitted helminths as reported by the authors.If endemicity is not reported, we will contact the authors.We will also refer to maps of schistosomiasis and soiltransmitted helminthiasis prevalence (e.g.Global Atlas of Helminth Infections: http://www.thiswormyworld.org/).If sufficient studies are available, we will conduct a sensitivity analysis to assess the effect of geohelminth deworming in regions of low (less than 10%), moderate (10-30%) or high (greater than 30%) schistosomiasis prevalence.Interventions of micronutrient supplementation, feeding, hygiene or sanitation compared to placebo will be excluded unless they are combined with or compared with deworming treatment which matches the WHO guidelines (Figure 2).
The helminth species targeted in this review are Ascaris lumbricoides (roundworm), Trichuris trichura (whipworm), Ancylostoma duodenale and Necator americanus (hookworms) and schistosoma-those most commonly associated with malnutrition and greatest disease burden among children in worm endemic areas (45).We will exclude other soil-transmitted helminths (Strongyloides, lymphatic filariasis and onchocerciasis) because different treatments regimens are required or insufficient empirical evidence exists around prevalence, burden and accepted control strategies.The included worms are consistent with those addressed in the current WHO guidelines on helminth control in school-aged children (10).Other helminth diseases, such as lymphatic filariasis and onchocerciasis will be excluded; these use overlapping drug regimens but are covered in separate guidelines (http://www.who.int/lymphaticfilariasis/resources/en;http://www.who.int/apoc/publications/en/)due to non-school-based control approaches.Strongyloidiasis will be excluded since it is less responsive to albendazole or mebendazole and lacks any formal public health strategy (http://www.who.int/neglected_diseass/diseases/strongyloidiasis/en/.Given the independent research efforts and protocols for the variety of worm types, we recognize the importance of ensuring that co-administration of helminth treatment is taken into consideration for analyzing the epidemiology of soil-transmitted helminthiasis and duplication of administrative efforts is minimized to improve effective resource usage.This systematic review focuses on mass drug administration or targeted deworming since the selective treatment of only infected individuals is not widely used due to the cost of screening tests.We will exclude studies where children are screened for infection, and only infected children are treated.

Comparisons
We will accept studies that use one of the following three types of comparisons: a) comparison that singles out the effect of the intervention of interest (e.g.all co-interventions are provided to both intervention and control group; and a control or placebo group is used as a comparison to the active intervention); b) a placebo or "do nothing" control group; or c) studies with a control group that receives an active intervention which is not given to the intervention group.
Comparison groups may use three points before and after the intervention was introduced, or it may use separate individuals, either using equivalent groups design (RCTs, RDDs) or non-equivalent groups using statistical methods to equalise groups.
Based on preliminary searches, we expect that studies are likely to use either a two-group design or a factorial approach.For example, a study conducted in Vietnam assessed the effects of albendazole and multiple micronutrient biscuits using a 2x2 factorial design which allows for three possible comparisons through the isolation of each intervention with paired placebos (14).For studies with a factorial design, we will use all available data to populate the network meta-analysis.Studies must assess at least one of the primary outcomes to be included for this review.Secondary outcomes will be recorded only if at least one of the primary outcomes is also measured.Outcomes will be recorded based on validated scales and calculated using standardized mean difference.We will exclude studies which measure the outcomes of worm burden or worm prevalence alone since this is not demonstrated to be associated with health or educational outcomes.Minimum reporting time is four months since we feel based on clinical expertise that changes in the primary outcomes of growth, educational status and cognition require at least this amount of time to materialize.All time points will be recorded, however we will use data reported at 12 months as the primary analysis, where possible, because this amount of time is deemed sufficient for growth, education and cognition changes.The time period for which the outcome is measured must be the same in both groups given the effects of climate (e.g.rainy season).

Study Types
We will include randomised controlled trials and controlled clinical trials, which may be randomised at the individual or cluster level.
We will also include quasi-experimental studies such as controlled before and after studies and interrupted time series (with at least three times points before and after the intervention, with or without a control group), cohort, case-control and cross-sectional studies.Comparison groups for these study designs must use statistical methods of analysis to match participants with nonparticipants, or statistical methods to account for confounding and sample selection bias.Methods of analysis to match participants and non-participants include regression discontinuity, propensity score matching (PSM) and covariate matching.Methods of analysis to control for confounding and selection bias include multivariate regression analysis using difference-in-differences (DID) estimation and instrumental variables (IV) estimation based on "natural experiments".Comparative studies with only post measurement will be included, provided they use one or more of these techniques.
We decided to include quasi-experimental studies in addition to RCTs because these study designs are more likely to be able to assess long-term effects such as cognition, labour market outcomes and school attendance.
Studies will not be excluded based on date, language or publication status.We will not exclude studies based on risk of bias assessments.

Search strategy development
We have developed a comprehensive search strategy with support from an information scientist (JM) for electronic databases and grey literature sources such as organizations active in deworming.The draft search strategy underwent review with PRESS and John Eyers, information scientist of the Campbell International Development Group, and appropriate changes were made (Appendices A and B to produce the finalized search strategy (Table 2).
We identified relevant studies to inform the search strategy development: four randomised controlled trials and one cross-sectional study that used propensity score matching (46) that compare deworming combined with other interventions such as hygiene education (47), feeding (46) or micronutrient supplementation (29,48,49) to placebo or active comparison groups (Table 3).

Electronic searches
The search includes the following health and non-health electronic databases: MEDLINE, CINAHL, LILACS, EMBASE, the Cochrane Library, Econlit, Internet Documents in Economics Access Service (IDEAS), Public Affairs Information Service (PAIS), Social Services Abstracts, Global Health CABI and CAB Abstracts.Grey literature databases will also be included (e.g.thesis dissertations, System for Information on Grey Literature in Europe (SIGLE)-ends in 2005).
We will also search websites of relevant organizations (UNICEF, Save the Children, Deworm the World, WHO, the World Bank We will also screen the references of included studies and conduct a SCOPUS search to identify any studies which cite included studies, according to the Peer Review of Electronic Search Strategies' PRESS recommendations on developing search strategies (50).

Other Searches
We will use reference lists from previous systematic reviews, conducted within the last ten years, to identify potentially relevant, individual studies and assess them based on the outlined eligibility criteria.

Selection of studies
Two reviewers will independently screen titles and abstracts based on the following questions: a) does the intervention include pharmacologic deworming treatment which is provided by mass or targeted administration to an identified high-risk group?; b) is at least one of growth, well-being, educational attendance, cognition or adverse events outcomes measured?; c) does the population include children between the ages of six months to sixteen years?; d) is the length of time from intervention to followup four months or longer?; and e) does the study design include an appropriate comparison group?(i.e., uses statistical methods to control for confounding such as propensity-score or covariate matching).We will pre-test the title and abstracts screening questions.If any one of these questions is answered as 'no', then the study will be excluded from further consideration.If all questions are answered as 'yes', then the study will be included for full-text screening.Coding for screening will be entered into systematic review software manager, EppiReviewer 4. After each reviewer has independently screened studies, any discrepancies around decisions for inclusion or exclusion will be discussed and reconciled accordingly.Full text will be retrieved for titles and abstracts accepted for inclusion after discussion by both reviewers.Full text will be screened by two reviewers for inclusion according to the pre-specified eligibility criteria.Any disagreements will be settled by discussion with a third party who will review the full text and decide whether it meets the inclusion criteria.For judgments related to appropriate control for confounders, we will consult with a statistician (GAW).

Data extraction
The two reviewers will conduct independent data extraction and risk of bias assessment of all included studies.The data extraction form will model the EPOC data collection form and be expanded where necessary in order to conduct network meta-analyses and assess the criteria specified by the SIGN-50.We will pre-test the data extraction form.Information to be extracted includes data on study design, statistical analysis, details about the participants (including the number in each group), setting (e.g.endemicity, sanitation), intervention (e.g.type of drugs, dose, frequency and process of implementation), comparison, cost-effectiveness, health and cognition outcomes (including whether outcomes are validated).We will extract process data on the implementation of the intervention such as method of delivering deworming (e.g.provision of deworming integrated with other programs), amount of supervision and monitoring of attendance in school and attendance to the deworming sessions.Where possible, we will extract data about socio-demographic variables associated with disadvantage, across factors described by the acronym PROGRESS (Place of residence, Race/ethnicity, Occupation, Gender/sex, Religion, Education, Socioeconomic status and Social capital) (51).We will extract data on any effect modifier analyses (e.g.subgroup analyses and meta-regression) conducted in the primary studies.We will compare the extraction by both reviewers, and reach consensus by discussion and consultation with a third reviewer, if necessary.

Process of implementation
We will extract the following process elements, based on a process evaluation of school feeding ( 52):  multifaceted approaches ( e.g. were other supports or integrated provisions given, such as hygiene promotion, iron supplementation, bed nets or vitamin A, in addition to providing deworming treatment?);

 time of day interventions are given;
 settings (where was the intervention administered?e.g.school, daycare, primary care clinics, immunization days);  prior needs assessment to inform intervention design and delivery (to identify when, where and how to give interventions to maximize uptake efficacy);  who delivered the intervention (e.g.supervised, and if so, by whom)?;  were interventions provided free of charge or for a reduced price according to income?
 were prompts/reminders provided (e.g. was intake of food or medication monitored)?;  cost and time to run program;  proportion of children in the community enrolled in school;  context in which the program is given (e.g.health systems context, community sanitation, availability of water to wash, infection intensity, poverty, whether the program is delivered as a vertical program);  reach of program ( e.g.what was the proportion of eligible children that are covered by the program?For example school-based programs will not reach children who are not enrolled in school and this could affect the reinfection rate in the community);  duration of the study (effects on cognition and education are expected only over longer time frames);  dose and type of drug given;  endemicity.
For each of these process elements, we will accept the definitions provided by the studies.We will consider the comparability of different methods of measuring these factors in grouping studies.For example, hygiene and sanitation may be defined differently in different settings and studies.

Assessment of risk of bias in included studies
Two members (PT and VW) of the team have expertise in assessing risk of bias of different study designs.The majority of study designs (randomised controlled trials, controlled before and after studies, cohort, and case-control studies) will be assessed by the risk of bias tool, SIGN-50 (53) .For other quasi-experimental studies not included under the SIGN-50 checklists (i.e.cross-sectional studies), we will employ the International Development Coordinating Group's risk of bias tool as it explicitly outlines assessment for study designs using propensity score and covariate matching.For interrupted time series studies, we will use the Cochrane Effective Practice and Organization of Care risk of bias checklist.
The main categories of bias that will be assessed are: selection bias, performance bias, detection bias, attrition bias and outcome reporting bias.Risk of bias will be assessed for each outcome in each study.Since we are particularly interested in spill-over effects, we will assess this risk under performance bias and make notes about the likelihood of spill-over effects.
SIGN-50 is a quality assessment instrument (QAI) developed by the Scottish Intercollegiate Guidelines Network and uses independent checklists according to type of study being assessed.Each checklist seeks to a) measure internal validity, b) provide an overall assessment of the study, and c) describe the context and details of the study (54).We selected SIGN-50 (55) for assessing risk of bias based on an overview of available checklists conducted for CADTH which showed that SIGN-50 performed well for randomised controlled trials and non-randomised studies when compared to other risk of bias checklists.SIGN-50 is summarized based on three-point summary table indicating the degree to which the set criteria have been fulfilled and denoted by either (++), (+), or (-) (56).Three SIGN-50 checklists will be used according to study design.These studies will be classified as high risk of bias if they have the following fatal flaws or "kill" factors, defined for each study design as being rated "poorly addressed" or "not addressed". 1) Randomised controlled trials  1.3 Concealment of allocation  1.4 Blinding  1.8 Loss to follow-up  1.9 Intention to treat analysis 2) Case control studies  1.2 Cases and controls are taken from comparable populations  1.6 Cases are clearly defined and differentiated from controls  1.9 Exposure status is measured in a standard, valid and reliable way  1.10 The main potential confounders are identified and taken into account in the design and analysis 3) Cohort studies  1.2 Selection of subjects: comparable in all respects except factor of investigation  1.8 Assessment of outcome is made blind to exposure status  1.10 Measure of assessment of exposure is reliable  1.13 Main confounders are identified and taken into account Any disagreements about the judgment relating to the risk of bias ratings will be resolved by discussion with a third party.

Measures of treatment effect
Two levels of analysis will be conducted: 1) meta-analyses for each outcome for each comparison of interest and comparisons which will be used to inform the network meta-analysis; 2) network metaanalysis for the main comparisons of interest.The network geometry in Figure 2 shows the main comparisons of interest: a) combined deworming (STH and schistosomiasis) or single target deworming (STH or schistosomiasis) vs. placebo; b) combined deworming vs. single target deworming (STH and schistosomiasis vs. STH or schistosomiasis); c) single target deworming vs. each other (STH vs. schistosomiasis); d) deworming vs. deworming and micronutrient(s); and e) deworming vs. deworming and feeding.The recently updated Cochrane systematic review of deworming for geohelminths included 42 randomized and quasi-randomized controlled trials, suggesting the breadth and size of literature will be sufficient to address these questions First, we will conduct meta-analyses for each comparison of interest, where clinically sensible (e.g. for trials of clinically similar populations, interventions, comparisons and outcomes) using Review Manager 5.1 Software.We will not combine results from different study designs since these are of different risk of biases.For RCTs, we will use unadjusted estimates.For quasi-experimental studies, we will use adjusted estimates and generic variance methods.

Meta-analyses for each comparison
Different study designs will be analyzed separately.Different interventions and comparisons will be analyzed separately.Studies with similar outcomes will be grouped for analysis.

Calculating effect size
The effect size of the continuous outcomes for growth (height and weight-for-age) will be analyzed as weighted mean differences of change scores as these will be measured using the same units (Z-scores) across studies.Continuous outcomes of test scores and cognition tests will be analyzed as change scores using standardized mean differences as the scales used may differ across studies.Standard deviations for each effect estimate will also be calculated.We will consult with a specialist in educational measures to group similar tests together.This may vary by adjusted and unadjusted estimates, and will be considered.Dichotomous outcomes will be analyzed as relative risks, using random effect methods.We will use a random effects model since we expect the underlying treatment effect will vary depending on the context, populations and setting.
We will report analyses for each outcome and follow-up period separately.
We will not conduct meta-analyses for the secondary outcomes.
Where the reported outcome data is not in the required format, (e.g.means, standard deviation and sample sizes), effect sizes will be calculated using the appropriate formulae provided in the Cochrane Handbook (57).If the data are adjusted, we will use the adjustment included in the reported estimate.For regression studies, we will use the IDCG protocol and review guidelines.
Costs and resource use data will be synthesized in tables.We will not conduct a cost-effectiveness analysis.

Unit of analysis issues
Where the unit of allocation is by groups (e.g.schools, communities, village, region), we will use the standard deviation adjusted for clustering, if provided by the study.In the case of randomised studies, if the study has not adjusted for clustering, we will adjust the standard deviations using the variance inflation factor, as described in the Cochrane Handbook (57).The variance inflation factor is calculated using the equation: (1 + (m-1) x ICC) where (m) is the cluster size and ICC is the intra cluster correlation.If cluster size is not reported, the number of participants in each analysis or total number of participants (if former not available) will be divided by the number of clusters to calculate cluster size.If ICC is not reported, we will estimate ICC values for the corresponding outcome measure using published ICCs for similar outcome measures.The effect of ICC values will be assessed using sensitivity analysis.
For quasi-randomised studies, where an adjusted estimate has been determined but clustering is not taken into consideration, we will derive procedures to accommodate for clustering in the modelling process, if possible.

Effects of treatment extrernalities
Treatment externalities may occur if there are spill-over effects to those individuals who live in the treatment locality but don't receive deworming (who in an individually randomised study will be the control group).The effects of treatment externalities will be assessed by two methods: 1) assessing the improvement in control groups for studies conducted with clustering versus those without clustering (hypothesizing that those without cluster assignment may have larger beneficial effect in the control group because geographical distance between control and intervention clusters may reduce the externalities to the control group); and 2) assessing whether individual studies have assessed the extent to which control groups benefit.

Assessment of heterogeneity
Heterogeneity will be assessed by visual inspection of forest plots, chi-squared test and I 2 .I 2 will be used to quantify inconsistency across studies, as it describes the percentage of variability in effect estimates that is due to heterogeneity.We will explore heterogeneity using meta-regression and sensitivity analyses.
Heterogeneity will be explored, if possible, by using meta-regression to assess the role of possible effect modifiers such as sanitation, poverty, under-nutrition, prevalence/intensity of infection, coinfection, concomitant interventions (e.g.micronutrients, hygiene) and risk of bias.We have chosen to assess the role of these factors based on prior theory or evidence, as shown in the logic model (3,32,58).Subgroup analyses reported in the included studies will be extracted and compared to these meta-regression analyses.

Summary of findings table
The magnitude of effect and quality of evidence will be presented clearly and concisely in a summary of findings table, with the seven most patient important outcomes (56).We define the human important outcomes as anthropometry, educational participation and achievement, cognitive status, well-being and adverse effects.Quality will be assessed using the GRADE criteria for this summary of findings table, as recommended by the Cochrane Handbook.

Dealing with missing and incomplete data
We will attempt to collect complete data on items in the data extraction form.Information will be sought from the authors of the study if information reported is insufficient to calculate effect size and standard deviation.In the event these attempts are unsuccessful, the studies and or outcomes will be listed as pending.We will not impute missing values (e.g.missing variance or outcome data).

Sensitivity Analysis
Sensitivity analyses will be conducted to assess the impact of outlier individual studies (e.g.very large studies, very large effects, very precise confidence intervals) on the overall effect size.Meta-analysis will be conducted with and without studies that may not fully fit inclusion criteria.We will also assess effects of risk of bias, missing data, treatment compliance, imputed variance inflation factor and the effect of unpublished studies using sensitivity analysis.
3.3.6.8Qualitative Synthesis: Process evaluation Data for all studies will also be synthesized qualitatively.We will summarize process data by assessing the contribution of each process element to the primary outcomes, where possible.We will use the process data to interpret results based on the assumptions about causal pathways shown in the logic model (Figure 1).

Step 2: Network Meta-analysis
At the next level, network meta-analyses (NMAs) will be conducted.Network meta-analysis is increasingly being used in health research as it allows for simultaneous comparison of all available therapies and uses all available relevant data, and is thus extremely relevant to clinicians, researchers and decision makers.A two-phase approach will be used, with the first phase using data from RCTs alone and the second phase using data from observational studies separately.This method allows for estimation of the impact of the observational research on summary estimates.
Following careful assessment of heterogeneity across trials in terms of patient characteristics, trial methodologies, and treatment protocols, NMAs will be conducted for the pre-specified outcomes.We will compare effect estimate according to outcomes of interest (growth, educational status, cognition, well-being, co-morbidities, costs and adverse effects, where available).There are three types of possible comparison including 1) isolation of deworming intervention compared to placebo or "donothing", 2) deworming and active intervention compared to control group and 3) deworming compared to deworming and other active interventions, and 4) Deworming compared to other active interventions will only be assessed in situations where studies assessing one of the former three comparisons provide information that helps to close loops in the network meta-analyses.We have assumed that placebo and "do nothing" control groups result in similar magnitude of effect in our network meta-analysis diagram.We will test this assumption in the assessment of heterogeneity, if possible.
We will use a Bayesian approach to conduct random-effects models with selection based on the Deviance Information Criterion (DIC) and residual deviance.R (R Foundation for Statistical Computing, Vienna, Austria) and WinBUGS (MRC Biostatistics Unit, Cambridge, UK) will be used for Bayesian network meta-analyses according to the routine that accommodates evidence structures, which may consist of multi-arm trials as developed at the Universities of Bristol and Leicester (www.bris.ac.uk/cobm/research/mpes/).Deworming will be the reference group for all Bayesian NMAs.Posterior densities for unknown parameters will be estimated using Markov Chain Monte Carlo (MCMC) methods.Basic parameters will be assigned non-informative or vague prior distributions; more informative priors may be considered after evaluation of the information base and clinical expert advice.Point estimates and 95 per cent credible intervals will be used to summarize findings.The probability of a comparator being optimal will be estimated for each outcome based on the proportion of MCMC simulations in which its relative measure of effect was best.The mean rank for each comparator will also be calculated.Consistency between direct and indirect evidence will be formally assessed using back-calculation and node splitting techniques (59).Graphical methods and numerical summaries will be developed for presenting results from network meta-analysis (60).Model diagnostics will also include trace plots and the Brooks-Gelman-Rubin statistic (55) to assess and ensure model convergence.Two chains will be fit in WinBUGS for each analysis, each usually employing ≥20,000 iterations, with a burn-in of ≥20,000 iterations.Whether novel agent effects are present will be examined and their magnitude of effect estimated (61).
Provided sufficient data are available to inform the evidence network, we will conduct metaregression and/or sub-group analyses to adjust for sex, age, worm intensity and endemicity of worms to test the robustness of reference case analyses (62).In other sensitivity analyses, we will remove studies from the network that have high risk of bias (as defined above).
(See Figure 2 for proposed evidence network geometry for this meta-analysis).
We will report on the added-value of using NMA in the final report.

Assessment of Publication Bias
Publication bias is a risk for any systematic review.We will attempt to minimize publication bias by conducting a comprehensive search strategy of published and grey literature, using trial registries and contacting experts in the field.We will assess the presence of publication bias using funnel plots and trim and fill if possible, and interpret these cautiously based on number of studies retrieved.

8 Tables Table 1 . Recommended mass and/or targeted control strategies for soil- transmitted helminths and schistosomiasis in school-age children
*Modified from 2002 WHO Guidelines Table2.2(10)

Table 3 . Potentially eligible studies identified Biovin 1993 (49)
school children who tested positive for Ankylostome and or Ascaris infection but negative for malaria parasites.Intervention  Decaris (Levamisole or Tetramisole) two 150 mg tablets four days apart + iron supplement (20ml) administered orally for 30 days  Decaris (Levamisole or Tetramisole) two 150 mg tablets four days apart alone  Iron supplement (20ml) administered orally for 30 days alone  Decaris placebo + no iron Comparison No iron and/or placebo Outcomes Cognitive function was evaluated using the mental processing portion of the Kaufman Assessment Battery for Children (K-ABC) test.Single treatment of albendazole (600mg) for geohelminths + single treatment of praziquantel (40mg/kg) for Schistosoma mansoni + micronutrient placebo  daily multi-micronutrient supplementation (iron, vitamin A, folic acid, iodine, vitamin C, vitamin B1, B2, B6, B12, niacin, copper and selenium) + anthelminthics placebo  Micronutrient placebo + anthelminthics placebo Weight for age, height for age, knee-heel length and serum albumin Raven's Coloured Progressive Matrices (RCPM), Rey Auditory Verbal Learning Test (RAVLT), Symbol Digit Modalities Test (SDMT), Group Mathematics Test (GMT).
*Need to contact authors for further detail.