Iron deficiency in HFpEF: a true clinical divergence or a competing risk artefact?

Dear Editor, Alnuwaysir and colleagues recently concluded that conventional definitions of iron deficiency (ID) lack prognostic relevance for heart failure (HF) hospitalizations and cardiovascular (CV) mortality in patients with preserved ejection fraction (HFpEF).1 While the cohort is well-characterized, a glaring methodological flaw undermines their core conclusion: the complete omission of competing risk analyses. In the highly comorbid HFpEF population, non-CV mortality is a formidable competing risk.2 Standard Cox proportional hazards regression—the sole survival model utilized by the authors—treats non-CV death as an independent, non-informative right-censoring event. This mathematical approach assumes that a patient who dies from a non-CV cause, such as sepsis or malignancy, remains ‘at risk’ and could hypothetically experience a future HF hospitalization.3 This impossible assumption routinely overestimates the cumulative incidence of the primary endpoint and severely distorts hazard ratios. This statistical distortion becomes fatal when evaluating a systemic condition like ID. The authors’ own multivariable models demonstrate that ID is strongly driven by advanced age, renal impairment, and severe hypoalbuminemia—all of which are potent drivers of non-CV death. Consequently, patients with ID experience non-CV mortality at a disproportionately rapid rate. By treating these non-CV deaths as standard censoring events, the Cox model actively deletes the sickest ID patients from the risk set early in the follow-up period. This severe informative censoring inevitably forces the hazard ratio for CV events toward the null.