Discharge anaemia after cardiac surgery: timing, surgical risk and clinical actionability

I read with interest the multicentre study by Boyko et al. examining the association between discharge anaemia and long-term survival after cardiac surgery [1]. The authors distinguish the strong prognostic signal of pre-operative haemoglobin from the weaker signal measured after surgery. Two analytical clarifications could strengthen the clinical interpretation of the findings. Discharge haemoglobin was the exposure, but survival follow-up began on the date of surgery. A patient must reach the point at which discharge haemoglobin can be measured before being assigned to an anaemia category. It would therefore be helpful to report the number and handling of deaths during the index admission; the availability of haemoglobin measurement in such patients; and whether time before exposure ascertainment entered the risk set. A prespecified analysis restricted to patients alive at discharge, with discharge as time zero and in-hospital mortality reported separately, would estimate the association with post-discharge survival. This would answer a conditional prognostic question and complement the analysis starting at surgery. The primary model appropriately avoided adjustment for transfusion and cardiopulmonary bypass duration when estimating an association that may include peri-operative pathways. Surgical urgency and procedure category, however, are known before discharge haemoglobin is measured. Moderate-to-severe anaemia occurred more often after urgent or emergency surgery, and after valve or combined procedures. These features may influence both discharge haemoglobin and later survival, without necessarily being consequences of discharge haemoglobin itself. Sequential models that add urgency, procedure category, centre and surgical era before intra-operative variables could show how much prognostic information discharge haemoglobin retains after pre-exposure operative risk is characterised. This would also clarify the attenuation observed after adjustment for EuroSCORE. These issues matter for patient blood management. Anaemia was present in 93.1% of patients at discharge, whereas the adjusted spline remained broadly flat across most haemoglobin values. A common recovery state should not be treated automatically as a uniform therapeutic phenotype. The current clinical uncertainty is illustrated by a randomised trial in which early intravenous iron increased haemoglobin after cardiac surgery but did not significantly improve its primary composite outcome or functional measures at 4 weeks [2]. Conversely, anaemia persisting into recovery has been associated with mortality after valvular surgery [3]. Prognostic thresholds, treatment selection thresholds and restrictive transfusion thresholds therefore address different decisions. The AABB guideline supports a restrictive transfusion threshold of 7.5 g.dl-1 in patients who are haemodynamically stable after cardiac surgical, not a target for routine correction of discharge haemoglobin [4]. Future trials could use serial trajectories, iron phenotype and patient-important outcomes to identify patients most likely to benefit.