Background: This study was designed to assess the use of clinical criteria
and biochemical testing to detect systolic dysfunction. Our goal is to deve
lop strategies that may enhance the detection and treatment of patients wit
h early ventricular dysfunction while reducing the use of echocardiogrphy.
Methods and Results: We compared the predictive characteristics of the plas
ma brain natriuretic peptide (BNP) concentration with that of a 5-point cli
nical score derived from elements of the history, electrocardiogram. and ch
est radiograph in outpatients (n = 466) referred for echocardiography becau
se of symptoms of heart failure or risk factors for systolic dysfunction. S
ystolic dysfunction was defined as an ejection fraction (EF) less than 45%
and was present in 10.9% of patients. By receiver operating characteristic
analysis, BNP was sensitive and specific for the detection of systolic dysf
unction, with an area under the receiver operating characteristic curve for
the detection of EF less than 45% of 0.79. The BNP assay was abnormal in 4
1% of patients and identified a group with a high prevalence of systolic dy
sfunction (21% with an EF less than 45%), whereas a normal BNP value identi
fied a group with a low prevalence of systolic dysfunction (24% with an EF
less than 45%). The clinical score was positive in 43% of the population an
d identified a group with a high prevalence of systolic dysfunction (24% wi
th an EF less than 45%). A normal score identified a group with a low preva
lence of systolic dysfunction (1% with an EF less than 45%).
Conclusion: This study supports previous studies, which showed that BNP ass
ay predicts systolic dysfunction with acceptable sensitivity and specificit
y, and it underscores the effectiveness of additional readily available cli
nical criteria. Both of these strategies should he considered in screening
for left ventricular dysfunction in populations at risk while limiting expe
nsive cardiac imaging modalities.