PT045 Epidemiological profile of patients with suspected Heart Failure, subjected to BNP test, assisted in the Emergency Room of a Tertiary Hospital in south Brazil

Abstract

Introduction: Objectives: Validate and compare alternative cut-offs for BNP test in subgroups with altered response with suspected Heart Failure. Methods: This study will be developed using data collected from patients with suspected Heart Failure assisted in the Emergency Room of Mae de Deus Hospital from March 2003 to March 2013. The study consisted of two samples. The first sample, named Test Sample, included 318 patients from March, 2003 to November, 2008; and the second, Validation Sample, included 391 patients from December, 2008 to March, 2013. BNP dosage was performed using the POCT Biosite method, and the Framingham score was used as gold standard. For all patients, one and two-dimensional Doppler echocardiography and chest X-ray were performed. Clinical data on the patients were obtained from the records. In order to adjust accuracy in subgroups with specific clinical characteristics, a multivariate analysis was done with logistic regression. The patients with gold standard Heart Failure(HF) or Not Heart Failure (NHF) in the Test Sample will be analyzed relating the BNP cut-off value to the age, gender, BMI and ECC (renal function was assessed via the Cockroft-Gault formula) and comparing to the cut-off values in the literature. BNP correction formula will be calculated by multiple logistic regression and the probability of having Heart Failure or not will be calculated as well. The calculated formula will be applied to the Validation Sample and the probability of having Heart Failure or not will be estimated. Moreover, AUC, BNP cut-offs will be compared to sensibility values, specificity, positive predictive value, negative predictive value, RV+, RV-, and BNP cut-offs. Results: Below, the results of the stratified analysis of BNP and AUC medians in subgroups BMI>30kg/m2, ECC 80 and gender of the Test Sample. The results showed significant difference in BNP andAUCmedians in the studied subgroups, compared to the cutoffs of theGeneral Sample. Theobserveddifferences suggest the calculationbymultiple logistic regression of a correction factor (Beta) to calculate a newBNP cut-off in subgroupswith altered response to improve the biomarker accuracy for diagnosis and prognosis of Heart failure.