Real-world identification of patients with HFpEF in a tertiary hospital in the Netherlands: the IDENTIFY-HFpEF study

Background

Heart failure with preserved ejection fraction (HFpEF) accounts for a substantial proportion of heart failure (HF) cases but often remains under-recognised due to diagnostic complexity and unstandardised electronic health records (EHR). As a result, many patients remain undiagnosed and undertreated, even in hospital settings where they have frequent contact with healthcare providers.

Purpose

To evaluate and compare existing HFpEF identification strategies in a tertiary hospital population.

Methods

The study population included all adult patients with an echocardiogram with at least 2 echo parameters recorded in a tertiary hospital in the Netherlands between 1/1/2018 - 13/11/2025. We excluded all patients with a verified reduced ejection fraction (EF) or mildly reduced EF diagnosis recorded for the Dutch HF registry, an echocardiogram with EF measured <50% or ARNI (Angiotensin Receptor Neprilysin Inhibitor) users. We labelled all adult patients with a clinician confirmed HFpEF diagnosis recorded for the Dutch HF registry with gold labels. Any patients with a HF diagnosis, without a prior EF <50%, were labelled with data-approximated silver labels for HFpEF. We compared the prediction of HFpEF according to the European Society of Cardiology (ESC) 2021 guidelines (based on NT-proBNP and objective evidence of abnormalities consistent with the presence of left ventricular diastolic dysfunction), the HFA-PEFF algorithm and the H2FPEF score.

Results

The total study population was 16.091 patients with an echocardiogram. Among these were 1.346 silver labelled HFpEF patients and 52 gold labelled HFpEF patients. Patients with gold label HFpEF had a median age of 73 years [IQR 59-78] and 70% were women. Similarly, silver label HF patients had a median age of 70 years [IQR 59-78], 52% were women. Patients without HF were on average younger, with a median age of 52 years [IQR 33-66] and less women (46%).

Predicting HFpEF in the gold label set showed a sensitivity of the ESC, HFA-PEFF and H2FPEF models of 0.88 [95% CI 0.65-0.99], 0.67 [95% CI 0.30-0.93] and 0.40 [95%CI 0.05-0.85] respectively (Figure 1). For predicting HFpEF in the silver labelled set the sensitivity dropped to 0.67 [95% CI 0.64-0.70], 0.52 [95% CI 0.47-0.57] and 0.58 [95%CI 0.53-0.64] respectively (Figure 2).

Conclusions

In this study, ESC-based criteria demonstrated the most robust and consistent performance for identifying HFpEF across both gold- and silver-labelled populations. This supports the pragmatic use in real-world clinical settings where echocardiographic data are often incomplete. In contrast, the H2FPEF and HFA-PEFF scores were limited by substantial missing echocardiographic data, limiting the number of patients the scores could be applied to, reducing their applicability in routine care. These findings highlight the need for HFpEF identification strategies that are scalable and remain informative in heterogeneous hospital populations.Gold label predictions of HFpEFFor image description, please refer to the figure legend and surrounding text.Silver label predictions of HFpEFFor image description, please refer to the figure legend and surrounding text.