Physician accuracy for diagnosing heart failure in unselected patients hospitalised with dyspnoea

ABSTRACT Aim: To examine variables associated with heart failure (HF) and determine accuracy of Emergency Department (ED) physicians to diagnose HF in a Norwegian teaching hospital without point-of-care natriuretic peptide testing. Methods and results: We included 150 consecutive patients hospitalised for dyspnoea and collected the results of clinical examination and the probability of HF (0-100%) from the ED physicians. HF was adjudicated according to guidelines by two independent senior physicians. Of 150 patients, 68 patients (45%) were diagnosed with HF as the primary cause of the hospitalisation. HF patients were older (75.1 vs. 68.0 y, p<0.001) and more likely to be male (57% vs. 37%, p=0.01). There was no difference in New York Heart Association functional class or the duration of symptoms prior to hospitalization between HF patients and patients with non-HF dyspnoea. Several clinical variables previously reported to be predictive of HF were associated with HF by crude analysis in our patients, but only history of HF (HR 11.14 [OR 2.73-45.47], p<0.001), history of hypertension (HR 3.22 [OR 1.12-9.22], p=0.03), and atrial fibrillation (HR 3.22 [OR 1.21-10.58], p=0.02) were independently associated with HF in multivariate analysis. The area under the curve for ED physician diagnosis of HF was 0.85 (95% CI 0.79-0.91, p