HIF-PHIs Not Linked to Higher Heart Failure Hospitalization
The primary finding supports a narrow comparative association, not a causal safety conclusion: hospitalization estimates were compatible with no difference, and unmeasured differences between treatment groups remain a possible explanation.
A nationwide Japanese observational study found no evidence of higher heart failure hospitalization with hypoxia-inducible factor prolyl hydroxylase inhibitors (HIF-PHIs) than with erythropoiesis-stimulating agents (ESAs). The adults studied had heart failure and chronic kidney disease and were not receiving dialysis. The result does not prove that the two drug classes have equal effects.
Evidence
Heart failure outcomes with HIF-PHIs is a single-study explainer covering research published September 1, 2026. Researchers looked back through a nationwide Japanese claims database, comparing 14,995 new HIF-PHI users with 22,889 new ESA users. Everyone was already receiving diuretics—medicines that increase urine output. The study used an active comparator, meaning another treatment rather than no treatment.
The primary outcome was heart failure hospitalization. Its hazard ratio was 0.93, with a 95% confidence interval of 0.87–1.01. A hazard ratio compares the rate at which an event occurs during follow-up; 1 indicates no difference. The confidence interval describes uncertainty around the estimate and includes 1, so the result is compatible with no difference.
Heart failure hospitalization incidence was 157.8 per 1,000 person-years among HIF-PHI users and 177.0 per 1,000 person-years among ESA users. Person-years combine participants’ time under observation; these rates are not percentages of patients hospitalized.
For the secondary outcome, hospitalization requiring intravenous diuretics—given into a vein—the hazard ratio was 0.87, with a 95% confidence interval of 0.81–0.94. This corresponds to an association with a 13% lower hazard in the HIF-PHI group.
Analysis
Interpretation: The primary result supports the limited conclusion that this analysis did not detect increased heart failure hospitalization with HIF-PHIs relative to ESAs. It does not establish equivalence or a protective effect.
The lower secondary estimate is relevant but answers a different outcome question. It cannot turn the uncertain primary comparison into proof that HIF-PHIs prevent heart failure admissions. Both findings came from the same cohort, not independent studies.
Limitations
Treatment was not randomly assigned. Statistical weighting can address measured differences, but residual confounding—differences between treatment groups that remain unaccounted for—cannot be excluded. A claims-based comparison therefore cannot establish that treatment choice caused the observed outcomes.
The population also matters: these findings do not establish effects in people receiving dialysis or those without heart failure. The primary confidence interval includes both a lower hazard and a small increase, rather than demonstrating identical hospitalization risk.
What to watch
A useful next test would be a randomized comparison in the same patient population, addressing treatment-selection differences more directly. A heart failure hospitalization hazard ratio with a confidence interval entirely above 1 in that comparison would materially increase concern about higher hospitalization hazard and challenge the present absence of an increased-hospitalization signal.