Diagnostic evaluation and recognition of HFpEF in routine clinical practice: a multicentre retrospective study
K Buitenhuis, S Sanders-Van Wijk, S Koudstaal, A Schut, N Ourahou, T Van Den Berg, M Varwijk, J J Brugts, J SchaapAbstract
Aim
Heart failure (HF) with preserved ejection fraction (HFpEF) accounts for approximately half of all heart failure cases worldwide, with a prevalence that continues to increase. (1, 2) However, HFpEF often remains unrecognised in clinical practice, reflecting its heterogeneous presentation and the challenges of establishing a definitive diagnosis. To support clinicians in estimating the likelihood of HFpEF, several diagnostic likelihood scores, such as H2FpEF, have been proposed. (3) However, little is known about the availability of diagnostic parameters and the use of these tools in routine clinical practice. This retrospective study aimed to evaluate the prevalence of a probable HFpEF diagnosis according to the H2FpEF-score in patients with a primary referral to the outpatient cardiology clinic for HF-like symptoms.
Methods and results
We conducted a retrospective cohort study, in 306 patients with HF-like symptoms (e.g. dyspnoea, oedema, and/or fatigue) without a prior history of heart failure, referred for a first evaluation to the outpatient clinic from three Dutch general hospitals during February 2024 until August 2025. In none of these hospitals a HFpEF likelihood score is part of routine clinical care. We recorded the cardiologists (differential) diagnosis at 2-6 months after referral and used a H2FpEF-score ≥6 as the reference standard for HFpEF. For the analysis, only patients with an available echocardiogram were included (n=280, 81.5%). Among patients with preserved left ventricle ejection fraction (≥50%) and no severe valvular disease (n=179, 58.5%), the cohort showed a high prevalence of HFpEF risk factors: age >60 years (n=163, 91.1%) female sex (n=104, 58.1%) and BMI >30 kg/m2 (n=57, 37.7%). NT-proBNP was elevated (>125 pg/mL) in 69.3% (n=125). In 78.2% (n=140) of cases, an H2FpEF score could be calculated, of whom 51 (36.4%) had a score ≥5 and 28 (20%) had a score ≥6, representing a moderate (≥84%) and high (≥91%) probability of HFpEF, respectively. (3) HFpEF was included in the differential diagnosis in 24.6%. Sensitivity and specificity of the physician’s clinical (differential) diagnosis of HFpEF were 53.6% and 87.5% respectively.
Conclusions
These real-world findings demonstrate that HFpEF is under-recognised in symptomatic patients despite a high likelihood of the condition according to the H2FpEF-score. The observed gap between likelihood-based assessment and clinical diagnosis support the rationale for the implementation of H2FpEF-likelihood scores in routine clinical practice and the prospective evaluation of the impact thereof in the identification of patients with HFpEF.