When appropriately managed, modern intravenous iron therapy is safe 6 : among the 1746 patients included in Aksan's study, there were no severe hypersensitivity reactions. Aksan's analysis perhaps overemphasises differences in safety between the intravenous products: the reported adverse events were almost all minor, with a rate of 12%-17%.So how should IDA in patients with IBD be managed? ECCO recommends up to 100 mg/day elemental Fe as oral iron for patients with mild anaemia and inactive IBD.1 Patients with active disease, evidenced by markedly raised CRP levels, tend to respond poorly to oral iron, 7,8 and like those intolerant of oral iron or having Hb <100 g/L, should usually proceed directly to intravenous iron.Whether ferric maltol, a new, more expensive and better tolerated oral iron 9 will be an option in this situation is under further evaluation (ClinTrials.gov NCT02680756).For intravenous treatment, we do not feel that the current analysis establishes beyond doubt the superior efficacy of FCM. We agree withAksan that we should be cautious about applying rank probabilities to important decision-making, 10 and that further head-to-head efficacy and safety trials are needed. Meanwhile, the intravenous preparation selected in most centres will be guided mainly by its local cost, convenience of administration and individual patients' tolerance. Kennedy and colleagues propose an alternative analysis using the random effects model. However, they have moved the goalposts, altering the efficacy endpoint to haemoglobin increase ≥2 g/dL only.Our efficacy endpoint, haemoglobin normalisation or increase ≥2 g/dL was carefully chosen to match the goal of iron replacement therapy specified in the current ECCO anaemia guidelines. 6 We recommend using this endpoint as a measure of treatment success in future trials of intravenous iron in IBD.Kennedy and colleagues suggest that removing certain trials from the analysis would change its results. Since only five studies fulfilled the inclusion criteria, it is unsurprising that removing any one study could affect the result-especially the large Evstatiev trial, 7 exclusion of which would reduce the NMA population by two-fifths (483/1143) and remove the majority of patients treated with IS(239/306) and FCM(244/380). This is the only study included which directly compares two IV products, and we believe its inclusion is entirely justified, despite the simplified dosing schedule, since this is a reflection of differing current dosage recommendations. Removal of the trial