Journal Club: Iron Before Cardiac Surgery (ITACS)

Something different this episode. No 3am emergency — instead, a pre-assessment clinic on a Tuesday afternoon, a woman booked for an aortic valve replacement in six weeks, and a haemoglobin of 118. Do you give her intravenous iron? ITACS, published in The BMJ in August 2026, is the best answer we have ever had to that question. It is also a trial with a great deal to teach about how to read a paper properly — so this is a journal club, and we go through the methods slowly, because that is where the meaning lives. The question first. A third of patients coming for cardiac surgery are anaemic, twenty to fifty per cent are transfused, and both are powerfully associated with complications, longer stays and death — in a specialty that consumes around ten per cent of the entire NHS blood supply. Correcting the anaemia in clinic attacks both problems at once. But anaemia may be a marker as well as a mechanism: if patients do badly because of the kidney disease or inflammation that made them anaemic, then fixing the number fixes the screen and changes nothing about the patient. A strong prognostic marker is not automatically a treatment target — and that idea runs through the whole episode. Then the methods. Thirty-three hospitals across ten countries, 955 anaemic adults for elective cardiac surgery, a single 1,000 mg dose of intravenous iron or placebo one to twenty-six weeks beforehand — and, deliberately, no requirement to prove iron deficiency. We look at how you blind a brown drug (a black syringe and an opaque line), and at the detail that separates a good trial from a trial that merely says "double blind" in its abstract: they checked whether the masking had worked by asking patients to guess their allocation. The primary outcome is days alive and at home at 90 days — what it captures, and the two things it hides. And then the finding that isn't in the abstract at all: the primary outcome was originally days at home at thirty days, and was amended to ninety in December 2020. We give the defence and the concern, and then we look at what the original outcome showed. It was null. Had the investigators kept it, this would be a negative trial, and that belongs in any honest summary of the paper. The results deserve care. The iron worked biochemically — ferritin rose from around 110 to over 500 — but haemoglobin rose by under 4 g/L, and three quarters of treated patients were still anaemic on the day of surgery, which explains the size of everything that follows. The primary result is one day, with a confidence interval touching zero, in a trial powered for a day and a half. Transfusion is the solid finding: 68% down to 61%, about fifteen patients treated to prevent one transfusion, and roughly 44 units of blood saved per hundred patients. But length of stay was identical, complications were identical — so where did the extra day come from? The paper answers it, and the answer is a smaller claim than the headline sounds. The best thing in the trial isn't in the abstract either. Rather than only reporting the median, the investigators reported the treatment effect across the whole distribution — and it turns out that patients who recovered well gained a fraction of a day, while at the 25th centile the difference was 6.1 days. All the benefit sits with the patients who did badly. It is post hoc, it was requested at peer review, and one of its confidence intervals is enormous — but it is biologically coherent, and it may be the most important idea to come out of this trial. We also correct an argument we would have made before reading the full paper. The obvious criticism — that enrolling patients without proven iron deficiency dilutes the effect — was pre-specified, tested, and not supported. Calum then supplies the more sophisticated objection, which is that an interaction test in a trial this size cannot exclude a subgroup difference. Plus the safety signal that matters to us specifically: a sixteen-fold increase in hypophosphataemia. We finish with the four-point critique, the strengths the trial genuinely deserves, and what to do in clinic on Monday. Chapters (00:00) Cold open — a haemoglobin of 118 and six weeks to go (01:00) Why the question matters, and the ten per cent of the blood supply (02:10) Marker or mechanism? The idea that runs through everything (03:00) What we knew before ITACS (03:40) The methods: 33 hospitals, 955 patients, no iron deficiency required (04:40) How do you blind a brown drug — and how do you prove it worked? (05:40) Days alive and at home: what it captures and what it hides (07:10) The primary outcome was changed mid-trial (08:30) What the original outcome showed (09:20) Sample size, two interim analyses, and the 95.4% interval (10:30) Did the iron actually do anything? Under 4 g/L (11:40) The primary result — one day, and an interval touching zero (12:50) Transfusion: the solid finding (13:50) Same length of stay, same complications — so where did the day come from? (15:00) The result the headline misses: who actually benefited (16:30) Iron deficiency — an argument corrected, and a better objection (18:00) Safety, and the phosphate (18:40) The critique in four parts (19:30) The strengths, and Monday morning (20:30) Wrap-up Key takeaways Anaemia is a marker as well as a mechanism — a strong prognostic marker is not automatically a treatment target ITACS: 33 hospitals, 10 countries, 955 anaemic adults, a single 1,000 mg dose of IV iron or placebo, 1–26 weeks before elective cardiac surgery The primary outcome was changed from days at home at 30 days to 90 days in December 2020 — and the original outcome was null Always ask what a trial's original primary outcome was, and what it showed The iron filled the stores but raised haemoglobin by under 4 g/L, and three quarters of treated patients were still anaemic on the day of surgery Primary result: one extra day at home in 90, with a confidence interval touching zero, in a trial powered for 1.5 days Transfusion is the robust finding: 68% to 61%, NNT about 15, and roughly 44 units of blood saved per 100 patients treated Length of stay and complications were identical — the extra day accrued after discharge as shorter readmissions, not fewer of them A post hoc quantile analysis suggests all the benefit sits with the poor recoverers: 6.1 days at the 25th centile against a fraction of a day at the 75th Benefit was the same with or without demonstrable iron deficiency — but that interaction test is underpowered, so read it as no reason to restrict treatment rather than proof that deficiency is irrelevant Intravenous iron caused a sixteen-fold increase in hypophosphataemia, which is not a footnote in a patient about to be weaned The 95.4% confidence interval is the fingerprint of pre-planned interim analyses and alpha spending Find the anaemia early: the biggest barrier in this trial was time, not biology Read the methods before the abstract, and convert every relative number into an absolute one References / further reading Myles PS, Klein AA, Smith JA, et al. Intravenous iron to treat anaemia before cardiac surgery (ITACS): international, double blind, placebo controlled randomised trial. BMJ 2026; 394: e100407 Myles PS, Richards T, Klein A, et al. Rationale and design of the intravenous iron for treatment of anemia before cardiac surgery trial. Am Heart J 2021; 239: 64–72 Boer C et al. 2017 EACTS/EACTA Guidelines on patient blood management for adult cardiac surgery. J Cardiothorac Vasc Anesth 2018 Agarwal S, Choi SW, Fletcher SN, Klein AA, Gill R. The incidence and effect of resternotomy following cardiac surgery on morbidity and mortality: a 1-year national audit on behalf of the Association of Cardiothoracic Anaesthesia and Critical Care. Anaesthesia 2021; 76: 19–26 Myles PS et al. Validation of days at home as an outcome measure after surgery. BMJ Open 2017 Follow the podcast Bluesky: @cardiacoutput.bsky.social X: @CardiacOutputMC If you've found this useful, a follow or a share genuinely helps other trainees find it — and do get in touch if there's a paper you'd like taken apart. This podcast is for medical education for healthcare professionals. It is not clinical advice. Discussion of published research reflects our own reading and interpretation — always read the primary source and follow your own centre's guidelines.