Arcanjo 2011 (C).

Methods	Cluster‐randomised, placebo‐controlled double‐blind trial. 2 arm design with randomisation at classroom level.
Participants	106 preschool children, both sexes (56 females (52.8%), aged 5 years. The study was conducted in a public school located in the City of Sobral, in the northeast of Brazil between September and December 2009. Exclusion criteria: current supplement intake. Baseline prevalence of anaemia: 58.5%. Forty per cent of the families had an income <300 USD.
Interventions	Classrooms were allocated to one of the following groups: Group 1 (3 classrooms, 52 children): children received once a week 50 mg of elemental iron (as ferrous sulphate heptahydrate) once a week; Group 2 (3 classrooms, 54 children): children received once a week a placebo (on Wednesdays). The placebo contained 2 ml of natural colour additive, annatto, which is odourless and tasteless, providing a yellow–orange colour similar to that of the elemental iron used in the study. Length of the intervention: 14 weeks.
Outcomes	Haemoglobin, hematocrit and anaemia (Hb less than 115 g/L)
Notes	The supplements were administered on Wednesdays. The supplement was administered by a teacher using a plastic medical syringe with scale to squirt the composition into the child’s mouth. The syringes were prepared on an individual basis by medical staff. We adjusted the results of this study to account for the effect of clustering in data; the estimated effective sample size was used in the analyses.  Malaria endemicity not reported.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	An allocation code was generated with a table of random numbers for randomizations of schools and classes.
Allocation concealment (selection bias)	Low risk	The study used a placebo. Since randomisation occurred at classroom level, it is unlikely a selection bias at individual level.
Blinding (performance bias and detection bias) All outcomes	Low risk	Participants: were not aware of different interventions. Personnel: the teacher was not aware of the treatment nor involved in data collection. Ouctome assessors: the staff involved in data collection was blinded with regard to the intervention and placebo groups.
Incomplete outcome data (attrition bias) All outcomes	Low risk	During the study, there were 2 (3.8%) dropouts in group 1, and 5 (9.2%) dropouts in group 2. Intention to treat analysis.
Selective reporting (reporting bias)	Unclear risk	There is insufficient information to permit judgement.
Other bias	Unclear risk	The data was not adjusted by the effect of clustering. Anaemia prevalence at baseline was not balanced between groups: 48% in group 1 and 69% in group 2 (but similar concentrations of haemoglobin).