Analysis of observer variability in measurement of pulmonary artery occlusion pressures

The goal of this study was to determine inter- and intraobserver variabilit y in measurement of pulmonary artery occlusion pressure (Ppao), comparing v alues recorded by critical care nurses and those measured by physician spec ialists. Critical care nurses (CCNs) obtained contiguous pulmonary artery a nd occlusion pressure paper tracings, up to twice a day, between lune 1997 and March 1998. All tracings were interpreted on two separate occasions, in blinded fashion, by our Chiefs of Critical Care (CCMD) and Cardiology (CAR D). Their values of Ppao were compared with those that had been recorded by CCNs. One hundred and forty-seven measurements of Ppao were performed on 4 0 patients with a mean age of 62.5 +/- 2.2 yr and a mean APACHE II score of 21.5 +/- 0.8. Either or both physician readers found 34 tracings as not sa tisfactory for Ppao interpretation. Intraobserver agreement of Ppao measure ments, determined by correlation coefficients, was 0.91 for the CCMD and 0. 87 for the CARD. Correlation coefficients for interobserver comparisons wer e 0.83 for CCMD-CARD, 0.66 for CARD-CCN, and 0.67 for CCMD-CCN. Clinically significant differences were observed between CCMD-CARD (range of differenc es, -11 to 12 mm Hg), CARD-CCN (-13 to 15 mm Hg), and CCMD-CCN (-11 to 15 m m Hg). When Ppao readings were categorized as low (< 5 mm Hg), normal (5-15 mm Hg), and high (> 15 mm Hg), kappa values were 0.57 for CARD-CCMD, 0.51 for CARD-CCN, and 0.41 for CCMD-CCN comparisons. Interobserver variability was not explained by positive pressure ventilation or by the presence of (> 4 mm Hg) ventricular waves. The absolute values of interobserver differenc es in tracings with respiratory phasic variations (RPV) greater than or equ al to 8 mm Hg were significantly greater than for tracings with variations < 8 mm Hg (p < 0.05, except CCMD-CCN, p = 0.10). Intraobserver differences also tended to be higher for tracings with RPV greater than or equal to 8 m m Hg (p = 0.06 and 0.05). When selected tracings were presented to 23 CCNs and 18 physicians, variability of Ppao interpretation was twice as great fo r tracings with large RPVs as compared with those with minimal RPVs. These data suggest that observer variability of Ppao interpretation is of potenti al clinical importance and that the degree of variability is associated wit h the magnitude of respiratory phasic variation of intrathoracic pressures. Although this could represent a local aberration, this study highlights a factor (respiratory phasic variation of Ppao) responsible for significantly increased intra- and interobserver variabilities.