Diagnostic utility of B-type natriuretic peptide in critically ill patients with pulmonary edema: a prospective...

Distinguishing pulmonary edema due to acute lung injury (ALI) or the acute respiratory distress syndrome (ARDS) from hydrostatic or cardiogenic edema is challenging in critically ill patients. B-type natriuretic peptide (BNP) can effectively identify congestive heart failure in the emergency room setting but, despite increasing use, its diagnostic utility has not been validated in the intensive care unit (ICU).

We performed a prospective, blinded cohort study in the medical and surgical ICUs at the University of Chicago Hospitals. Patients were eligible if they were admitted to the ICU with respiratory distress, bilateral pulmonary edema and a central venous catheter suggesting either high-pressure (cardiogenic) or low-pressure (ALI/ARDS) pulmonary edema. BNP levels were measured within 48 hours of ICU admission and development of pulmonary edema and onward up to three consecutive days. All levels were drawn simultaneously with the measurement of right atrial or pulmonary artery wedge pressure. The etiology of pulmonary edema--cardiogenic or ALI/ARDS--was determined by three intensivists blinded to BNP levels.

We enrolled a total of 54 patients (33 with ALI/ARDS and 21 with cardiogenic edema). BNP levels were lower in patients with ALI/ARDS than in those with cardiogenic edema (496 +/- 439 versus 747 +/- 476 pg/ml, P = 0.05). At an accepted cutoff of 100 pg/ml, specificity for the diagnosis of ALI/ARDS was high (95.2%) but sensitivity was poor (27.3%). Cutoffs at higher BNP levels improved sensitivity at considerable cost to specificity. Invasive measures of filling pressures correlated poorly with initial BNP levels and subsequent day BNP values fluctuated unpredictably and without correlation with hemodynamic changes and net fluid balance.

BNP levels drawn within 48 hours of admission to the ICU do not reliably distinguish ALI/ARDS from cardiogenic edema, do not correlate with invasive hemodynamic measurements, and do not track predictably with changes in volume status on consecutive daily measurements.