Mathematical Model Diagnoses Acute Heart Failure

Researchers at St. Michael's Hospital (Toronto, ON, Canada) sought to derive and validate a prediction model by using the N-terminal pro-B-type natriuretic peptide (NT-proBNP) and additional clinical variables to improve the diagnosis of acute heart failure (AHF). To do so, physician estimates of the probability of AHF in 500 patients treated in the ER--which had participated in a previous congestive heart failure (CHF) trial between December 2004 and December 2005--who were classified into low (0% - 20%), intermediate (21% - 79%), or high (80% - 100%) probability groups for AHF. These probabilities were then compared with the actual determined AHF diagnosis. Likelihood ratios were calculated into an AHF prediction model that was validated both internally (by bootstrapping) and externally by applying the model to another 573 patients from a different study.

The researchers found that the likelihood ratios for AHF with NT-proBNP were 0.11 for cut-point values below 300 pg/mL; increasing to 3.43 for values ranging between 2,700-8,099 pg/mL; and 12.80 for values equal or greater than 8,100 pg/mL. The variables used to predict AHF included age, pre-test probability, and log NT-proBNP. When applied to the external data using the determined final diagnosis as the gold standard, the model appropriately reclassified 44% of patients to either low or high probability of AHF, with a negligible classification of inappropriate redirection. The study was published in the October 13, 2009, issue of the Journal of the American College of Cardiology.

"In many cases, when a patient arrives in an emergency department complaining of shortness of breath, physicians are challenged to correctly diagnose patients,” said lead author Brian Steinhart, M.D. "Our model does not require extensive clinical information, which makes it relatively simple-to-use. When the result is greater than 80% probability for heart failure, it suggests that the physician should treat for AHF and when it is less than 20%, the physician should be looking elsewhere for diagnosis.”

NT-proBNP is a biologically inactive 76 amino acid N-terminal fragment of brain natriuretic peptide (BNP); BNP itself is a 32 amino acid polypeptide secreted by the ventricles of the heart in response to excessive stretching of heart muscle cells (cardiomyocytes). Both BNP and NT-proBNP levels in the blood are used for screening and diagnosis of acute CHF and may be useful to establish prognosis in heart failure, as both markers are typically higher in patients with worse outcome.

Related Links:
St. Michael's Hospital