Sen 2009 (C).

Methods	Cluster‐randomised controlled trial. 4‐arm design with randomisation at school level.
Participants	240 school age females, aged 9‐13 years, attending four schools in Vadodara area of India. Females were excluded from the analysis if menstruation commenced. None of the Females were involved in athletic sports on a regular basis. Baseline prevalence of anaemia: 68.3%. Socioeconomic status not described in detail but participants were described as "underprivileged".
Interventions	Schools were allocated to one of the following groups: Group 1 (n = 65): females received 100 mg elemental iron (as ferrous gluconate) and 500 μg (0.5 mg) folic acid folic acid oral once weekly; Group 2 (n = 89): females received the same supplement twice weekly (200 mg of elemental iron per week); Group 3 (n = 59): females received 100 mg elemental iron (as ferrous gluconate) daily; Group 4 (n = 41): females received no supplement. Length of the intervention: 1 year. Groups 1 and 2 were combined and compared with groups 3 and 4 as appropriate.
Outcomes	Physical work capacity, haemoglobin change and adherence.
Notes	Malaria endemicity not reported. Analyses in this review include the estimated effective sample size only, after adjusting the data to account for the clustering effect.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Of 17 schools meeting the inclusion criteria, 4 schools were selected randomly using a random numbers table. Once the four schools were selected, the chit system (chits representing school 1, 2, 3, 4) was used. The order of placing a school in a category was: the first school that is picked up (from the four) goes to daily; the chit is then put back; the next chit picked up goes to twice weekly; the next to once weekly and the one left over, to control so that all schools have an equal probability of being allocated to any of the four groups (information communicated by the author).
Allocation concealment (selection bias)	Low risk	Not reported. Since allocation was at school level it is unlikely that there is selection bias at individual level.
Blinding (performance bias and detection bias) All outcomes	Unclear risk	Participants, personnel and outcome assessors: Each school received a different intervention, although it is unclear if the intervention was blinded.
Incomplete outcome data (attrition bias) All outcomes	High risk	4 schools. In these schools a random sample of 240 children was selected. 163 had pre and postintervention data for work capacity (68% followed up). Females who started their periods were excluded from the analysis. For cognitive tests results relate to a sub‐sample of 161 females available pre and post‐test.
Selective reporting (reporting bias)	Unclear risk	There is insufficient information to permit judgement.
Other bias	High risk	The design effect was not taken into account in the analysis.