Aims: Aims were to evaluate (1) reclassification of patients from heart failure with mildly reduced (HFmrEF) to reduced (HFrEF) ejection fraction when an EF = 40% was considered as HFrEF, (2) role of EF digit bias, ie, EF reporting favouring 5% increments; (3) outcomes in relation to missing and biased EF reports, in a large multinational HF registry.
Methods and results: Of 25,154 patients in the European Society of Cardiology (ESC) HF Long-Term registry, 17% had missing EF and of those with available EF, 24% had HFpEF (EF≥50%), 21% HFmrEF (40%-49%) and 55% HFrEF (<40%) according to the 2016 ESC guidelines´ classification. EF was "exactly" 40% in 7%, leading to reclassifying 34% of the HFmrEF population defined as EF = 40% to 49% to HFrEF when applying the 2021 ESC Guidelines classification (14% had HFmrEF as EF = 41% to 49% and 62% had HFrEF as EF≤40%). EF was reported as a value ending with 0 or 5 in ∼37% of the population. Such potential digit bias was associated with more missing values for other characteristics and higher risk of all-cause death and HF hospitalization. Patients with missing EF had higher risk of all-cause and CV mortality, and HF hospitalization compared to those with recorded EF.
Conclusions: Many patients had reported EF = 40%. This led to substantial reclassification of EF from old HFmrEF (40%-49%) to new HFrEF (≤40%). There was considerable digit bias in EF reporting and missing EF reporting, which appeared to occur not at random and may reflect less rigorous overall care and worse outcomes.
Copyright © 2023 The Author(s). Published by Elsevier Inc. All rights reserved.