Corrigan 1936.

Methods	Quasi‐randomised trial with allocation by odd or even numbers. 2‐arm trial.
Participants	200 normal pregnant women attending antenatal care clinic with 3‐7 months of gestational age at Boston City Hospital, Boston, USA.
Interventions	Participants were assigned a number in order. Patients who had been assigned an odd number received 0.2 g of ferrous sulphate (3 tablets daily to be taken after meals ‐ total daily dose 0.6 g); patients with even numbers received placebo that were identical in appearance and size and contained lactose but not ferrous sulphate. Supplements were from recruitment until delivery. Women who took less than 1 of the 2 tablets prescribed daily were excluded. Setting and health worker cadre: the intervention was performed by physicians at the antepartum clinic of Boston City Hospital, Boston, Massachusetts, United States of America.
Outcomes	Number of women with anaemia at 1‐week postpartum. (Figures were also provided for the mean Hb level at 1‐week postpartum but no SD was provided and we were not able to include these data in the analysis.)
Notes	Mean Hb in the intervention group 117 g/L and 112 g/L in the control group. Gestational age at start of supplementation: mixed gestational age. Anaemic status at start of supplementation: not specified. Daily iron dose: higher daily dose (60 mg or more). Iron release formulation: not specified. Iron compound: ferrous sulphate. Malaria setting: non‐malarial setting. As of 2011: Malaria: no risk.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	High risk	Quasi‐randomised trial, odd/even numbers.
Allocation concealment (selection bias)	High risk	Alternate allocation.
Blinding of participants and personnel (performance bias) All outcomes	Unclear risk	This study was placebo‐controlled but the method of randomisation may have meant that staff were aware of treatment allocation. Also women who did not comply were excluded (and it is possible that there were systematic differences between groups in compliance).
Blinding of outcome assessment (detection bias) All outcomes	Unclear risk	Outcomes reported unlikely to have been affected by lack of blinding (although it was not clear how many women were excluded after randomisation).
Incomplete outcome data (attrition bias) All outcomes	Unclear risk	200 women were randomised and it was stated that women that did not comply (that took on average less than 1 of the prescribed tablets daily) or in whom sepsis or haemorrhage developed during pregnancy, birth or the early postnatal period were excluded. It was not clear how many women were excluded for these reasons and it was not clear whether or not there was any ITT analysis.
Selective reporting (reporting bias)	Unclear risk	There is insufficient information to permit judgement.
Other bias	Unclear risk	Most of the results were provided in graphs and were not simple to interpret and we have included all of these results in the analyses. Women were described as similar at baseline. The denominators for results were not clear.