Statistical analysis
Patient characteristics were described using means (standard deviation), medians (interquartile range), or counts and frequencies, as appropriate. Primary and secondary endpoints were analysed using regression models with treatment group and haemoglobin at baseline (Hb<105 or Hb≥105) as covariates. Endpoints measured at multiple timepoints were analysed using repeated measures models with a timepoint by treatment group interaction. All analyses were performed in SAS v 9.4 (SAS Institute, Cary NC USA) or R v4.2.0 (R Foundation for Statistical Computing, Vienna, Austria).
Primary endpoint: The difference in proportions of participants who required a repeat iron infusion at each timepoint was analysed using a repeated measures model with generalized estimating equations (GEEs) and an independence working correlation.
Two one-sided tests at the 0.05 level were used to assess equivalence at each time point. Since the equivalence margin was 5%, 90% confidence intervals for absolute differences in proportions between groups at each time point were examined as to whether they lay within the interval (-0.05, 0.05). To establish equivalence, the proportion of participants in both groups must lie completely within this interval.
Secondary endpoints: Repeat infusion rates were analysed using Poisson regression models. Serum iron, transferrin, and transferrin saturation were analysed using mixed effects linear regression models adjusted for baseline. The measurement of ferritin at baseline was subject to threshold effects from limits to detection (<4 µg/l) and was singly imputed half the limit of detection (2 µg /l).
Maternal pregnancy and birth complications, mode of delivery and neonatal outcomes were analysed using linear or logistic regression models, as appropriate. GEEs with an independence working correlation structure were used, where relevant, to account for multiple births.
For secondary clinical outcomes, we report mean differences (MDs) in the change from baseline between treatment groups, rate ratios (RRs) or odds ratios (ORs), as appropriate, with 95% confidence intervals (CI).