Importantly, in this study all the patients’ CXRs and intravascular measurements were taken in the supine or semi-supine position and only films graded as satisfactory for positioning (that is, not overly rotated on visual inspection) MG132 Proteasome were included in the analysis. In addition to patient positioning, some have raised concern that the disease process might affect the assessment of VPW. Indeed, the effects of recent trauma, thoracic surgery, or prior radiation therapy alter components of the mediastinal silhouette and compromise the utility of the VPW [18,19]. On the other hand, respiratory factors have been shown to have relatively little effect on VPW measurements. Milne observed comparable VPW measurements during both inspiration and expiration [5].

Although mechanical ventilation may have profound effects upon other radiographic findings such as the pattern and severity of parenchymal infiltrates [20,21], VPW measurements have been found to be consistent between spontaneous and positive pressure breaths [20]. Our data also found only a trend toward a weak correlation between PEEP and VPW measurements. Despite these potential limitations in measuring the VPW, we confirmed prior findings that VPW correlates with PAOP and we found that the VPW correlated 1.5 times better with PAOP than cumulative net fluid balance and 2.5 times better than PEEP. Thus, for patients without or for clinicians who prefer not to use invasive intravascular pressure measurements, VPW represents a better surrogate of PAOP than net fluid balance.

One limitation of our study is that we compare VPW to two surrogate measures of intravascular volume, CVP and PAOP, and not a direct measure of intravascular volume, such as right (RVEDV) or left ventricular end-diastolic volume (LVEDV). Although echocardiography might estimate RVEDP and LVEDP, too few patients had these available on days with VPW measurements to investigate this correlation directly. CVP and PAOP do correlate well with right (RVEDP) and left ventricular end-diastolic pressure (LVEDP), respectively [22-24]. Although a similar correlation with RVEDV and LVEDV is widely presumed, this is not the case in a number of conditions pertinent to acute lung injury, including sepsis [25-27], trauma [28], and acute respiratory failure requiring mechanical ventilation [29]. Observations by Kumar and colleagues suggest that CVP and PAOP do not correlate Brefeldin_A well with RVEDV or LVEDV even in normal, healthy volunteers [30]. This is likely due to varying compliance of the ventricles from patient to patient and heartbeat to heartbeat within the same patient.