Abstract
BACKGROUND: The performance of biomarkers for heart failure (HF) in older residents in long-term care is poorly understood and has not differentiated between left ventricular systolic dysfunction (LVSD) and HF with preserved ejection fraction (HFpEF). METHODS: This is the first diagnostic accuracy study in this population to assess the differential diagnostic performance and acceptability of a range of biomarkers against a clinical diagnosis using portable echocardiography. A total of 405 residents, aged 65-100 years (mean 84.2), in 33 UK long-term care facilities were enrolled between April 2009 and June 2010. RESULTS: For undifferentiated HF, BNP or NT-proBNP were adequate rule-out tests but would miss one in three cases (BNP: sensitivity 67%, NPV 86%, cut-off 115 pg/ml; NT-proBNP: sensitivity 62%, NPV 87%, cut-off 760 pg/ml). Using higher test cut-offs, both biomarkers were more adequate tests of LVSD, but would still miss one in four cases (BNP: sensitivity 76%, NPV 97%, cut-off 145 pg/ml; NT-proBNP: sensitivity 73%, NPV 97%, cut-off 1000 pg/ml). At these thresholds one third of subjects would test positive and require an echocardiogram. Applying a stricter 'rule out' threshold (sensitivity 90%), only one in 10 cases would be missed, but two thirds of subjects would require further investigation. Biomarkers were less useful for HFpEF (BNP: sensitivity 63%, specificity 61%, cut-off 110 pg/ml; NT-proBNP: sensitivity 68%, specificity 56%, cut-off 477 pg/ml). Novel biomarkers (Copeptin, MR-proADM, and MR-proANP) and common signs and symptoms had little diagnostic utility. CONCLUSIONS: No test, individually or in combination, adequately balanced case finding and rule-out for heart failure in this population; currently, in-situ echocardiography provides the only adequate diagnostic assessment. TRIAL REGISTRATION: Controlled-Trials.com ISRCTN19781227.