Diagnostic uncertainties in older inpatients with suspected heart failure with preserved ejection fraction: a pilot cohort study

Heart failure (HF) with preserved ejection fraction (HFpEF) is increasingly suspected in older adults. However, diagnosis remains challenging due to nonspecific symptoms, age-related structural cardiac changes, and confounding comorbidities. Reliability of algorithms developed in younger populations may be limited in this population. This prospective study enrolled inpatients aged > 75 years with at least one ESC-listed typical sign/symptom suggestive of HF and no prior HF diagnosis were evaluated for HFpEF using ESC 2021 criteria. Structural/functional transthoracic echocardiographic (TTE) abnormalities and age-adjusted NT-proBNP thresholds were recorded. Diagnostic accuracy of the H2FPEF and HFA-PEFF scores was evaluated. We also tested a modified HFA-PEFF score incorporating age-adjusted thresholds for NT-proBNP and for septal/lateral e' velocity. Among 200 inpatients (median age 86.6 years, 58.0% women), 96.0% fulfilled ESC criteria, with TTE abnormalities almost universal (95.0%), mainly left atrial enlargement (70.0%) and increased wall thickness (66.5%). However, only 52.0% had elevated age-adjusted NT-proBNP. Notably, confounding comorbidities were very frequent (98.0%), particularly chronic kidney disease (60.0%) and infections (40.0%). Using rule-in cut-offs, H2FPEF score identified few high-probability cases (7.0%) with very low sensitivity (0.07) and perfect specificity (1.00; AUROC 0.676). HFA-PEFF classified 67.5% of patients as high probability, with better accuracy (sensitivity 0.69, specificity 0.75; AUROC 0.856). The modified HFA-PEFF increased specificity (1.00) but reduced sensitivity (0.36), improving AUROC to 0.908 and lowering false positives. HFpEF diagnosis in older inpatients is complicated by overlapping comorbidities and age-related cardiac changes. Standard algorithms may misclassify patients and do not reflect true disease burden. Incorporating age-adjusted parameters in scores may improve specificity and support better clinical decision-making, but validation in larger studies is needed.