Stoltzfus 1997 (Cluster)

Methods	Cluster‐RCT Method to adjust for clustering: generalised estimating equations Cluster unit: school Average cluster size: 255 ICCs: not reported Length of follow‐up: 12 months
Participants	All children living in endemic area Number analysed for primary outcome: 12 schools randomized containing 3063 children Mean age: 10.5 years Inclusion criteria: children in grades 1 to 5 from 12 randomly selected schools on Pemba island; only grades 1 to 4 included in evaluation of nutritional effect Exclusion criteria: none stated
Interventions	Multiple doses vs placebo Mebendazole: 500 mg twice yearly; Mebendazole: 500 mg 3 times a year; Placebo.
Outcomes	Weight gain; Height gain; Change in haemoglobin at 12 months. Estimates are provided from multiple regression models taking into account various baseline differences for 2 subgroups above and below 10 years old. Unadjusted outcomes not presented. (These 2 groups were combined in the Dickson 2000a Cochrane Review.) Other outcomes measured but not reported: micronutrient status (blood) for protoporphyrin and serum ferritin; stool egg count (Kato‐Katz); z‐scores for height‐for‐age and weight‐for‐height; body mass index.
Notes	Location: Zanzibar, Tanzania Community category: 1 Appropriate adjustment made for cluster randomization using general estimating equation Source of funding: funded through cooperative agreement DAN‐5116‐1‐00‐8051‐00 between The Johns Hopkins University and the Office of Health and Nutrition, United States Agency for International Development.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	3 schools randomly selected from each of the 4 districts, and then allocated.
Allocation concealment (selection bias)	Unclear risk	No details reported.
Blinding (performance bias and detection bias) All outcomes	Unclear risk	No details reported.
Incomplete outcome data (attrition bias) All outcomes	Low risk	84% (3063/3605) of randomized participants were evaluated, reasons for losses to follow‐up not reported. Inclusion of all randomized participants (number evaluable/number randomized): 84% (3063/3605).
Selective reporting (reporting bias)	High risk	Not all pre‐specified outcomes reported adequately.
Other bias	Low risk	Recruitment bias: low (Unlikely to change schools) Baseline imbalance: low (no differences apparent) Loss of clusters: low (none reported) Incorrect analysis: cluster adjusted (low risk) Comparability with RCTs randomizing individuals: unclear