Key result
Adding NT-proBNP and ECG-detected AF improves detection of undiagnosed heart failure to an AUC of ~0.85.
Why the study?
Heart failure is a leading cause of hospitalization and death worldwide, and variables helpful for early HF diagnosis in primary care needed identification.
Does a clinical predictive model incorporating history, symptoms, ECG, and NT-proBNP accurately detect undiagnosed heart failure in adults aged ≥ 40 years in the general population?
Cross-Sectional (n=1,863)
No
Does a clinical predictive model incorporating history, symptoms, ECG, and NT-proBNP accurately detect undiagnosed heart failure in adults aged ≥ 40 years in the general population?
Easily available clinical information, ECG, and NT-proBNP can accurately identify undiagnosed heart failure in the general population and guide referrals for echocardiography in primary care.
Model shows promise for unrecognized HF detection in primary care; hypothesis-generating and requires prospective validation before clinical use.
Background: Heart failure (HF) is one of the most common causes of hospitalization and death worldwide. We aimed at identifying variables that can be helpful for early diagnosis of HF in primary care. Methods: In 2015–16, we included 2007 participants aged ≥ 40 years in the 7th Tromsø study. They underwent echocardiography and other examinations needed for establishing a diagnosis of HF according to European guidelines from 2016. We analyzed associations with HF among the 1863 participants who reported not to have HF. Variables significantly associated with HF were included in multivariable logistic regression analyses. Diagnostic models were further analyzed by ROC curves and areas under curve (AUC) were calculated, optimism corrected by bootstrapping. Results: The HF prevalence in the age groups 40–64 years, 65–74 years, and ≥ 75 years were 3.5%, 11.7% and 29.4%, respectively. A predictive model based on self-reported hypertension, myocardial infarction, atrial fibrillation, body-mass index and moderate to severe dyspnea had an AUC of 0.813 (95% CI 0.785–0.843). In a significantly stronger model, in which NT-proBNP was included and self-reported atrial fibrillation was replaced by atrial fibrillation on ECG, an AUC of 0.849 (95% CI 0.821–0.880) was reached. Conclusion: Easily available clinical information may be used both to rule out HF an to identify patients needing further examinations. A direct referral to echocardiography should be considered for elderly patients in primary care with a known cardiovascular disease and severe shortness of breath. With less abundant history and symptoms, ECG and NT-proBNP can guide further investigations.
No takes yet. Share an insight, caveat, or question.
Schirmer et al. (2025) conducted a cross-sectional in Heart failure (n=1,863). Diagnostic model including clinical variables, ECG atrial fibrillation, and NT-proBNP vs. Predictive model based on clinical variables and self-reported history alone was evaluated on Diagnosis of heart failure according to 2016 European guidelines (AUC 0.849, 95% CI 0.821-0.880). Adding NT-proBNP and ECG atrial fibrillation to a clinical diagnostic model improved detection of undiagnosed heart failure (AUC 0.849, 95% CI 0.821–0.880 vs 0.813).
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: