Utility of a bedside acoustic cardiographic model to predict elevated left ventricular filling pressure.

BACKGROUND: The authors previously described an acoustic cardiographic model that predicted echocardiographic correlates of elevated left ventricular (LV) filling pressure. This study evaluated this bedside acoustic cardiographic model against invasive measurements of LV filling pressure. METHODS AND RESULTS: Data were prospectively obtained from 68 adults referred for right heart catheterisation. Acoustic cardiographic measurements were obtained during right heart catheterisation. Elevated LV filling pressure was defined as a pulmonary capillary wedge pressure (PCWP) > or =15 mm Hg. Parameters generated from a previous dataset used for the current analysis were measures of LV systolic time, maximum negative area of the P wave, QTc interval and third heart sound (S3) score. Logistic regression was used to calculate area under the curve (AUC). Of the 66 patients included, 39 had elevated PCWP. Estimating the probability of an elevated PCWP from the derived model resulted in an AUC of 0.72 (95% CI 0.60 to 0.85). When the regression model's parameters were held constant but the parameter estimates were allowed to vary, the AUC in the validated model was 0.76 (95% CI 0.64 to 0.88). At a specificity of 90% the positive likelihood ratio (LR+) was 5.0 (1.7 to 15.3) and the negative likelihood ratio was 0.49 (0.34 to 0.71). CONCLUSION: These data demonstrate that the four-variable model predicts elevated filling pressure at the bedside with high specificity and an intermediate LR+. With improvements in sensitivity and further prospective validation of this model in a cohort of emergency department patients with undifferentiated dyspnoea this may be a useful bedside diagnostic modality.

Duke Scholars

Author Christopher Lindsell Biostatistics & Bioinformatics, Division of Biostatistics