Neth Heart J. 2025 Nov 4. doi: 10.1007/s12471-025-02000-y. Online ahead of print.
ABSTRACT
BACKGROUND: Diagnosing heart failure with preserved ejection fraction (HFpEF) remains challenging. While diagnostic algorithms support clinical evaluation, their performance across sexes is understudied, despite HFpEF being more prevalent in females, which may result in sex-specific underdiagnosis.
PURPOSE: To assess the diagnostic accuracy of three HFpEF algorithms-HFA-PEFF, H2FPEF, and the ESC HF 2016 criteria-and to evaluate sex-related differences in performance.
METHODS: Two prospective cohorts with unexplained dyspnoea were analysed. The Amsterdam cohort (2014-2020; n = 135) had HFpEF confirmed or excluded via (exercise) right heart catheterization (RHC). The Maastricht cohort (2015-2021; n = 659) had HFpEF confirmed or excluded based on expert consensus with utilisation of HFpEF scores, and RHC when needed. Sex-specific diagnostic performance of three HFpEF algorithms was assessed using ROC curves, AUC, and a panel of metrics with cut-offs determined by the rule-in/rule-out strategies.
RESULTS: HFpEF prevalence was high in both cohorts (84.4% and 82.5%), with a female majority (69.6% and 66.5%). Across all algorithms, males consistently showed lower AUC values than females, although differences were not statistically significant. The highest diagnostic performance within the Amsterdam cohort was observed with H2FPEF (AUC 0.86 and 0.82 for females and males), while HFA-PEFF performed best within Maastricht cohort (AUC 0.85 and 0.83, respectively). Performance for ruling-in and ruling-out HFpEF was numerically lower in males than females; Amsterdam cohort HFA-PEFF and ESC 2016 specificity were 83% versus 93% and 50% versus 73%, Maastricht cohort H2FPEF specificity was 81% versus 89%.
CONCLUSIONS: HFpEF diagnostic algorithms may perform better in females than males in referral outpatient settings. Inconsistent performance of diagnostic algorithms between different sexes warrants further optimisation to diagnose HFpEF.
PMID:41186885 | DOI:10.1007/s12471-025-02000-y
