1. Objectives

To evaluate the effects of deworming drugs for the treatment of soil-transmitted helminths in children on weight, haemoglobin concentrations, cognition, physical wellbeing, school attendance, school performance, and mortality

2. How studies were identified

The following databases were searched in April 2014:

• Cochrane Infectious Diseases Group Specialized Register
• CENTRAL (The Cochrane Library 2015, Issue 4)
• MEDLINE
• EMBASE
• LILACS
• CINAHL
• metaRegister of Controlled Trials (mRCT)

Reference lists were also searched

3. Criteria for including studies in the review

3.1 Study type

Randomized controlled trials, quasi-randomized controlled trials, and cluster-randomized controlled trials with at least two clusters per arm

3.2 Study participants

Children, under 16 years of age, living in areas endemic for soil-transmitted helminths or identified as being infected by screening in community trials

(Studies that exclusively enrolled sick children or children being treated for malnutrition were excluded)

3.3 Interventions

Deworming drugs for soil-transmitted helminths, administered either as a single dose or in multiple doses, compared to placebo or no treatment

(Included deworming drugs were those listed in the WHO Model List of Essential Medicines 2006: albendazole, levamisole, mebendazole, pyrantel, and ivermectin)

(Studies providing co-interventions, such as micronutrient supplementation or growth monitoring, were included if both treatment and control arms received the co-intervention)

3.4 Primary outcomes

• Weight
• Haemoglobin
• Formal psychometric tests of cognition

Secondary outcomes included other nutritional indicators: height, mid-upper arm circumference, skinfold thickness (including triceps and subscapular skinfolds), body mass index (BMI); measures of physical wellbeing (e.g., Harvard Step Test); school attendance: days present at school, number of children dropping out; school performance (as measured by examination results); and death. Adverse events included serious adverse events (death, life-threatening events, or events leading to hospitalization), and other adverse events

4. Main results

4.1 Included studies

Forty-five randomized controlled trials, enrolling more than one million children, were included in this review:

• Forty-four studies included a total of 67,672 children, and one study reporting on the outcome of child mortality involved more than one million children
• Thirty-seven studies employed mass treatment of an unscreened population; eight trials recruited children on the basis of having high worm loads
• Follow-up ranged from one to 21 months for trials administering a single dose, and from post-intervention to five years in trials using multiple doses
• Twenty-eight trials assessed albendazole, some of these included co-interventions such as praziquantel, ivermectin, diethylcarbamazine, secnidazole and health promotion in both treatment and control arms
• Seven trials assessed mebendazole, two trials used mebendazole plus pyrantel; other agents included pyrantel pamoate, piperazine, piperazine citrate, tetrachloroethylene, and levamisole
• Nine trials included health promotion activities, and four trials had a factorial design with arms including vitamin A, micronutrient-fortified soup, and iron-fortified noodles

4.2 Study settings

• Bangladesh (4 trials), Benin, Botswana, Cameroon, China (2 trials), Democratic Republic of the Congo, Ethiopia (2 trials), Guatemala, Haiti (2 trials), India (5 trials), Indonesia (4 trials), Jamaica (2 trials), Kenya (5 trials), Malaysia (2 trials), Nigeria, the Philippines (2 trials), Sierra Leone, South Africa (2 trials), Uganda (2 trials), the United Republic of Tanzania (3 trials), Viet Nam (3 trials), , and multi-centre (2 trials: China, the Philippines and Kenya; China and Myanmar)
• Children were recruited from school populations (26 trials), communities (12 trials), and in health facilities or by health workers (7 trials); in one trial children were recruited on discharge from hospital
• Nine studies were cluster-randomized; units of randomization included urban slums, rural administrative blocks, schools, households, parishes and villages
• Fifteen trials were conducted in populations with a high prevalence or intensity of helminth infection, 12 were in populations with moderate prevalence and low intensity, and ten were in populations with low prevalence and low intensity

4.3 Study settings

How the data were analysed
Four comparisons against placebo or no treatment were made: i) infected children, single/first dose; ii) infected children, multiple doses; iii) all children living in an endemic area, single/first dose; and iv) all children living in an endemic area, multiple doses. Endemic areas were further categorized for analysis by prevalence and intensity (high/moderate/low; WHO 2002 technical guidelines classification). Continuous data were summarized using mean differences (MD) with corresponding 95% confidence intervals (CI), and dichotomous data were planned to be summarized using risk ratios. For multiple-dose trials, data from the longest follow-up time point was used in analyses. Data from cluster-randomized trials was adjusted for clustering. Fixed-effects meta-analysis was used except where significant heterogeneity was detected, in which case random-effects models were employed. The following subgroup analyses were planned for the investigation of heterogeneity:

• by age group: < five years versus ≥ five years
• by manufacturer
• by treatment setting: community, school, health post, hospital

To investigate whether the effects of the intervention were modified by length of follow-up, a random-effects meta-regression for the outcome weight in children in an endemic area after multiple doses was conducting using length of follow-up as a covariate. A forest plot sorted by date was also visually inspected. Sensitivity analyses included trials with a low risk of bias for allocation concealment.

Results
Comparison 1. Children with infection: single dose of deworming drugs versus no intervention
Weight
In five trials involving children known to have helminth infection, those children given a single dose of deworming drugs gained three-quarters of a kilogram more weight than those who did not receive deworming drugs (MD 0.75 kg, 95% CI [0.24 to 1.26], p=0.0038; 627 children).

Haemoglobin
No statistically significant difference was observed between treatment and control groups in haemoglobin (MD 0.10 g/dL, 95% CI [-0.65 to 0.86], p=0.79; 2 studies/247 children).

Psychometric tests of cognition
One study reported improvements with deworming drugs in three of the ten psychometric tests conducted (fluency, digit span forwards, digit span backwards).

Additional outcomes
Gains were also observed after deworming treatment in pooled analyses of height (MD 0.25 cm, 95% CI [0.01 to 0.49], p=0.041; 5 trials/647 participants), mid-upper arm circumference (MD 0.49 cm, 95% CI [0.39 to 0.58], p<0.00001; 4 trials/396 children), and skinfold thickness (triceps: MD 1.34 mm, 95% CI [0.72 to 1.97], p<0.0001; 3 trials/352 children; subscapular: MD 1.29 mm, 95% CI [1.13 to 1.44], p<0.00001; 2 trials/339 children). BMI did not differ between groups in one trial of 407 children, and other trials reported nutritional status data in formats that could not be included in pooled analysis. Physical wellbeing was reportedly greater with deworming in a non-randomized analysis from one trial, while no treatment effect was found in another trial. None of the identified trials provided data for the outcomes school attendance, school performance, mortality, or adverse events.

Comparison 2. Children with infection: multiple doses of deworming drugs versus no intervention
Weight
In one study enrolling 284 children, weight was almost 1 kg greater (MD 0.90 kg, 95% CI [0.54 to 1.26]) in the deworming group after two doses in comparison to the controls.

Psychometric tests of cognition
Deworming was reported to have no effect on intellectual development scores in one trial enrolling 392 children.

Additional outcomes
Mid-upper arm circumference (MD 0.40 cm, 95% CI [0.27 to 0.53], triceps skinfold thickness (MD 1.80 mm, 95%CI [1.52 to 2.08]), and subscapular skinfold thickness (MD 1.50 mm, 95%CI [1.23 to 1.77]) gains were greater in those receiving deworming treatment in one study enrolling 284 children, but change in height did not differ between groups. No differences were observed in BMI, height-for-age Z-scores or school attendance between treatment and control groups in one trial involving 407 participants.

Comparison 3. All children living in an endemic area: single dose of deworming drugs
Weight
In seven trials conducted in areas of high, moderate and low helminth endemicity, change in weight did not differ between those treated with deworming agents and those who were not overall (MD -0.04 kg, 95% CI [-0.11 to 0.04], p=0.33; 2719 participants), or when subgrouped by level of endemicity.

Haemoglobin
Haemoglobin concentrations did not differ between groups in pooled analysis overall (MD 0.06 g/dL, 95% CI [-0.05 to 0.17], p=0.30; 3 studies/1005 children), or when subgrouped by level of endemicity.

Psychometric tests of cognition
No treatment effect or a negative treatment effect was found in two studies that reported on cognitive testing, although data were not presented.

Additional outcomes
In pooled analysis, height and mid-upper arm circumference did not differ between those treated with deworming drugs and those who were not. Other studies presented data in formats unsuitable for meta-analysis; one of these trials reported significant increases in weight-for-height, weight-for-age, and height-for-age Z-scores. No other studies reported significant differences between treatment and control groups overall for nutritional outcomes or haemoglobin concentrations. In one trial of 46 patients given albendazole, no adverse events were noted, however, in another trial using tetrachloroethylene–a drug no longer used–a number of adverse events were reported. In sensitivity analyses including only those trials judged to be at low risk for bias due to inadequate allocation concealment, no differences between treatment and control groups in weight, height, mid-upper arm circumference, or haemoglobin were detected.

Comparison 4. All children living in an endemic area: multiple doses of deworming drugs, longest follow-up
Weight
Overall, no evidence for a difference in weight gain between deworming and control groups was observed in pooled analysis (MD 0.08 kg, 95% CI [-0.11 to 0.27], p=0.43; 10 trials/2656 children). Heterogeneity was high (I²=83%), and was partially explained by prevalence of infection in the area, although no significant effect was found in any subgroup. When restricted to trials at low risk of bias due to allocation concealment, the effect remained statistically non-significant. The MD in change in weight between deworming and control groups did not differ by length of follow-up or by publication year.

Haemoglobin
In meta-analysis of seven trials, no treatment effect was observed for the outcome haemoglobin overall (MD -0.02 g/dL, 95% CI [-0.08 to 0.04], p=0.54; 3595 participants), when subgrouped by endemicity, or when restricted to trials at low risk of selection bias due to inadequate allocation concealment.

Psychometric tests of cognition
No treatment effect was noted for any cognitive test outcomes in five trials involving 32486 children.

Additional outcomes
In pooled analysis, children treated with deworming drugs did not differ from controls for the outcomes change in height, mid-upper arm circumference or triceps skinfold thickness. In one trial of 359 children, mebendazole significantly reduced the prevalence of mild wasting malnutrition among children less than 30 months of age. In another trial, treated children under ten years of age had significantly greater weight gain in comparison to controls (2.38 kg, standard error (SE) [0.08] versus 2.11 kg, SE [0.08], p<0.05; 3064 children). In meta-analysis of the outcome school attendance, the number of days children were present at school did differ between treatment and control groups (MD 2% higher, 95% CI [-4 to 8], p=0.59; 2 trials/293 children). No other studies reported significant differences between treatment and control groups for nutritional outcomes or absenteeism. School examination performance was measured in two trials; no significant difference was found between treatment groups. In the trial enrolling over one million children, death did not differ between deworming and control groups (MD in deaths per child-care centre 0.16, SE [0.11], mortality ratio 0.95, 95% CI [0.89 to 1.02]). In two other trials, the risk of death did not appear to be affected by deworming treatment.

5. Additional author observations*

The overall quality of evidence was graded as low or moderate in most of the outcomes for community deworming in endemic areas, meaning that further research could alter the effects described. Nevertheless, the review included a reasonable amount of evidence from trials across a range of settings, including areas of varying endemicity, which enabled stratification by intensity of infection. Ten trials assessed the effects of multiple doses of deworming in endemic areas, including four cluster-randomized controlled trials, which are important as they can detect community-level treatment effects. The review included results from a replication reanalysis study (Aiken et al., 2014 International Initiative for Impact Evaluation, 3ie) of data from a cluster-randomized trial in which externalities (benefits seen in untreated individuals living close to treatment areas) were originally reported. The pure replication study, using the same statistical methods as the original study authors, did not find externalities to be present after correcting for coding errors.

Deworming drugs may increase average weight gain in infected children, but single or multiple doses of deworming drugs have little or no effect on weight or other nutritional outcomes when all children living in endemic areas are treated. There is no clear evidence for a treatment effect on haemoglobin, cognition, physical wellbeing, school exam performance, school attendance or death. In sum, current evidence does not support large public health programmes of deworming in developing countries, although children known to be infected with helminths should receive deworming treatment. The review did not discuss the cost-benefits of screening for infection prior to treatment versus mass deworming.

Further research may help to clarify the effects of community deworming in endemic areas. However, if considered, additional research should be balanced against the decline in helminth infection worldwide and the current absence of any clear evidence for an effect.