Can pulmonary artery pressure be predicted without right heart catheterization in chronic obstructive lung disease?

We have investigated the dependance of observed pulmonary artery mean pressure (PAP) on body surface, age, PaO2, pH, PaCO2, hematocrit, and spirometric data (VC, FEV1/VC, RV/TLC) in 70 patients with chronic obstructive lung disease (COLD). After elimination of all variables that failed to correlate with PAP, multifactorial analysis showed that only two of nine independent variables, namely PaO2 and body surface, correlated significantly with PAP. According to our calculations, 28.9% of total PAP is predictable by PaO2, 1.5% by H+ concentration, 2.8% by RV/TLC, and 2.5% by body surface. Fully 64.2% of total variability was not accounted for by our regression analysis; thus the error of predicted PAP was so great (+/- 17 mm Hg for P = 0.05) as to invalidate the method. We also recalculated our subjects' PAP values by applying Enson's and Grassi's equations to our own lung function and biochemical data, and compared the predicted PAP values thus obtained with those measured directly in our subjects. Both equations proved imprecise and/or inaccurate in the individual case. From this we conclude that whereas available equations may be suitable for predicting the mean PAP value of a large population sample, the same equations cannot give a reliable prediction in individual cases.