Introduction

Mechanically ventilated patients often make strong inspiratory efforts with large falls in pleural pressure (Ppl) that are not evident in the airway pressure. These can be due to anxiety and inadequate sedation, increased respiratory drive from the underlying disease process [1–3], and increased airway resistance caused by excess secretions or too small an endotracheal tube. These efforts can require a considerable work of breathing and produce hemodynamic instability. They also are strong predictors of weaning failure during a spontaneous breathing trial [4]. When recognized, the clinician can try to reduce these efforts by increasing sedation, clearing secretions, or changing the endotracheal tube. It thus would be helpful to have some quantitative assessment of changes in Ppl in mechanically ventilated patient, but unfortunately measurement of changes in Ppl in critically ill patients is not simple. The use of pulmonary artery (PA) catheters has significantly decreased, but they are still frequently used. We reasoned that when present, variations in PA occlusion pressure (Ppao) during spontaneous inspiratory efforts could be used to approximate the magnitude of the fall in Ppl.

The rational is as follows. As long as the PA catheter is in zone III of the pulmonary vasculature, where alveolar pressure is less than pulmonary venous pressure, Ppao reflects left atrial pressure. In turn, changes in Ppl are normally reflected by changes in pressure in the compliant left atrium. However, two factors confound the relationship of changes in left atrial pressure to changes in Ppl by increasing left atrial volume during inspiration. Lung inflation squeezes blood out of zone  III vessels [5–7] and increases left atrial filling and pressure. Second, the fall in Ppl lowers the environment of the heart relative to the aorta which increases left ventricular afterload [8, 9]. Thus we predicted that the inspiratory fall in left atrial pressure would underestimate the fall in Ppl but would still be useful clinically because it would give a lower estimate of the magnitude of Ppl changes and allow assessment of trends. We also predicted that changes in CVP should not as reliably indicate changes in Ppl because of the differences in the sources of volume that fill the right and left hearts.

The reservoir for the left heart is in the pulmonary veins, which are also in the chest and thus their surrounding pressure moves in parallel with that of the left atrium. In contrast, the venous reservoir for the right heart is outside the chest so that the fall in Ppl increases the pressure gradient from the systemic venous reservoir to the right atrium. We examined these predictions in a consecutive series of patients who had esophageal balloons surrounding their nasogastric tubes and PA catheters as part of their routine care. This work was previously presented at the annual meeting of the American Thoracic Society [10].

Methods

We screened 41 consecutive patients who were intubated and had a PA catheter and nasogastric tube in place as part of their routine management and thus did not require separate consent according to our institutional procedures. Exclusion criteria included gastrointestinal bleeding, esophageal anatomical abnormalities including tumors, strictures, or esophageal varices, and hemodynamic or respiratory instability. The nasogastric tubes had a 10 cm balloon attached circumferentially (Mallinckrodt #16) that could be used to measure esophageal pressure (Peso) as an indicator of Ppl variations. Of the initial 41 patients only 24 patients had spontaneous ventilatory efforts and adequate Ppao. This group included 17 men and 7 women with a mean age of 56.9 ± 24.2 years, and all had undergone cardiac surgery, except for one case of acute respiratory distress syndrome postvascular surgery. Descriptive data on the 24 patients evaluated are listed in Table 1. All patients were intubated and ventilated except for one who was on a T-piece, and this patient had a PEEP of 0 cmH2O. All other patients had a PEEP of 5 except one in whom it was 8 cmH2O.

Table 1 Descriptive data of patients enrolled in the study based on baseline status (CVP, central venous pressure; Ppao, pulmonary artery occlusion pressure; PAP, pulmonary artery pressure; BPm, mean arterial pressure; PEEP, positive end-expiratory pressure)
Full size table

Proper positioning of the esophageal balloon was assessed by simultaneously measuring Peso and tracheal pressure (Ptr) during Mueller maneuvers [11]. In this procedure changes in Peso (dPeso) should equal changes in Ptr (dPtr) and dPtr/dPeso should be 0.85 or higher. Only 14 patients met this requirement. Vascular pressures were measured with disposable transducers (Transpac, Abbott Critical Care, N Chicago Ill., USA) which were leveled at 5 cm below the sternal angle. Ppao was deemed adequate and in zone III when a and v waves were evident, and there was a clear change from the pulmonary artery diastolic pressure [12]. Pressures were measured in the supine position and the transducer. Vascular and esophageal pressure signals were acquired with a data acquisition system (direct/pulmonitor) and maximum displacements from the end-expiratory value were assessed in simultaneously recorded signals (see Electronic Supplementary Material). Measurements were excluded when there was evidence of forced-expirations that raise expiratory pressures. We averaged between three and five breaths per condition. Average data are presented as mean ± SD.

Results

The average inspiratory decrease in Ppao (dPpao) was 6.3 ± 4.4 cmH2O, and average dPeso was 8.4 ± 6.5 cm H2O. The Bland–Altman plot of the difference between dPeso and dPpao vs. the average values of dPeso and dPpao (Fig. 1) shows the identity plot for dPpao vs. dPeso. The bias, 1.95 SD, and regression equations for this and all other conditions are shown in Table 2. There was a tendency for the differences to be greater with larger swings in pleural pressure. The bias was significantly different from zero when compared by t test; this was true at p ≤ 0.001 for the 14 patients with dPeso/dPtr of 0.85 or higher as well as for the whole set of 24 subjects. The average inspiratory decrease in CVP (dCVP) was 4.1 ± 2.6 cmH2O for an average change in dPeso of 6.9 ± 5.0 cmH2O. The bias and identity plots are shown in Fig. 2. There was more variation than seen for dPpao, and the R 2 was only 0.43. The bias was not significantly different from zero in the patients with dPeso/dPtr of 0.85 or higher but was significant (p ≤ 0.003 for the whole group). Figure 3a shows the Bland–Altman plot of difference of dPeso and dPpao for positive pressure breaths and Fig. 3b the identity plot. The bias between the measurements was small.

Fig. 1
figure 1

a Bland–Altman plot of change in Peso (dPeso) and change in Ppao (dPpao) during spontaneous inspiratory effort. The average of dPeso and Ppao is on the x-axis and the difference between these two measures in cmH2O on the y-axis. Broken line, bias (2.2 cmH2O); dotted lines, ± 1.95 SD of the bias (8.2, –3.9 cmH2O); black circles, data points in which dPeso/dPtr during a Mueller maneuver was 0.85 or greater; lighter circles, those in which dPeso/dPtr was less than 0.85. b Identity plot of dPpao vs. dPeso. Dotted line, identity; solid line, regression line. Symbols are as in a. Overall R2 was 0.84 whereas the R2 for those with dPeso/dPtr of 0.85 or greater was 0.91

Full size image
Table 2 Bias and standard deviation (± SD), regression equations and correlation coefficients (R 2) for each condition (dPpao, change in pulmonary artery occlusion pressure; dPpao(+), change in pulmonary artery occlusion pressure with positive pressure; dPeso, change in esophageal pressure; dCVP, change in central venous pressure)
Full size table
Fig. 2
figure 2

a Bland–Altman plot of change in Peso (dPeso) and change in CVP (dCVP) during spontaneous inspiratory effort. The average of dPeso and CVP is on the x-axis and difference between these two measures in cmH2O on the y-axis. Broken line, bias (2.9 cmH2O); dotted lines, ± 1.95 SD of the bias (10.3, –4.6 cmH2O); black circles, data points where dPeso/dPtr during a Mueller maneuver was 0.85 or higher; lighter circles, those in which dPeso/dPtr was less than 0.85. b Identity plot of dPpao vs. dCVP. Dotted line, identity; solid line, regression line. Symbols are as in a. Overall R2 was 0.43

Full size image
Fig. 3
figure 3

a Bland–Altman plot of change in Peso (dPeso) and change in Ppao (dPpao) during positive pressure inspiration. The average of dPeso and Ppao is on the x-axis and the difference between these two measures in cmH2O is on the y-axis. Broken line, bias (0.5 cmH2O); dotted lines, ± 1.95 SD of the bias (2.8,–3.8 cmH2O). b Identity plot of dPpao vs. dPeso for positive pressure breaths. Dotted line, identity; solid line, regression line. Overall R2 was 0.73

Full size image

Discussion

Consistent with our prediction, inspiratory changes in Ppao gave good estimates of the change in pleural pressure as assessed by dPeso, and in most instances dPpao underestimated dPeso. We used dPeso as the “gold-standard” to estimate dPpl, but Peso is less than an ideal gold-standard in supine patients [13, 14]. Potential problems include compression of the esophagus by the heart, esophageal contractions, and inhomogeneities in Ppl [13]. Accuracy is improved by positioning the balloon with an “occlusion test” in which the subject inspires against an obstructed airway and the ratio of dPeso to dPtr is measured [11, 13, 15, 16]. A ratio of 0.85 or higher indicates that Peso is a reasonable estimate of Ppl, but this occurred in only 14 subjects. The R 2 for dPeso vs. dPpao was slightly better in these subjects than those with dPeso/dPtr of 0.85 or higher (Table 2) and dPpao was greater than dPeso in only two of these cases and in both the difference was 2 cmH2O or less. Thus dPpao is a reasonable estimate of dPeso, but on average underestimates the actual change.

There was a tendency for the difference between dPeso and dPpao to be greater at higher values of Ppao. A number of factors might explain this. The atrial pressure–volume relationship is curvilinear so that at large left atrial volumes the increased filling of the left heart that occurs with lung inflation [5, 7] results in a larger pressure increase than occurs at lower left atrial volumes. Secondly, pulmonary venous volume is likely larger at higher left atrial pressures so that more volume can be squeezed out of the lungs during inflation. Finally, the compliance of the left atrium is lower at higher volumes, which decreases the transmission of changes in Ppl to the left atrium.

We chose to examine dPpao rather than changes in pulmonary artery pressure (PAP) because the respiratory changes in PAP make it harder to quantify the respiratory variations. This partly due to the technical factor that excursion of PAP during the cardiac cycle is obviously much large than Ppao, and therefore it is harder precisely to quantify the bottom of the respiratory swing. However, respiratory changes in pulmonary artery diastolic pressure still should give estimates of Ppl changes that are not that dissimilar to those of Ppao and have the advantage of being a “continuous” measurement.

There was greater variation in the prediction of dPeso by dCVP than dPpao. This is consistent with our prediction that dCVP does not always reflect changes in Ppl. One group found that dCVP is a good predictor of dPeo in normal subjects in the sitting position, but not always in the supine position [17], and another argued that dCVP even might be more accurate that dPeso [18]. These contradictory findings may be due to differences in volume status. When right heart volume is close to maximum, the heart is not compliant and small changes in volume produce large changes in pressure. The noncompliant heart also blunts transmission of changes in Ppl to cardiac cavities. Furthermore, when CVP is high, abdominal volume reserves are larger and can be recruited by decent of the diaphragm during inspiration. In support of this, the relationship of dCVP to changes in Ppl was decreased at higher values of CVP in animals [19]. Thus dCVP should not be expected to reflect of dPeso at higher values of CVP and frequently does not fall with a fall in Ppl [20].

In contrast to the positive bias for spontaneous inspiratory efforts, the bias of the relationship of dPeso vs. dPpao during positive pressure ventilation was slightly negative and dPpao tracked dPeso closely. The negative bias (i.e., greater dPpao than dPeso) could have been due to improved emptying of the left ventricle during positive pressure breathing [21, 22] which countered increased inspiratory filling. It is also possible that the inflated lungs distorted the esophagus and dPpao was actually the true indicator of the change in Ppl.

Clinical use

The assessment of inspiratory swings in Ppl by dPpao has a number of potential clinical uses. We previously showed that patients who have a fall in CVP of 1 mmHg or more with an inspiratory effort generally respond to a fluid challenge with an increase in cardiac output, whereas cardiac output does not rise in patients in whom the fall in CVP is less than 1 mmHg. We used a dPpao of 2 mmHg or greater to indicate that the fall in Ppl is adequate to test the right heart response. This study supports the use of that measurement in that dPpao likely underestimates the fall in Ppl [20].

The inspiratory fall in Ppao also gives an indication of the effort required to trigger the ventilator. A large inspiratory fall in Ppl should alert the physician to a number of possibilities. These include increased airway resistance because of secretions or bronchospasm or, alternatively, a strong respiratory drive and a need for greater sedation and perhaps an increase in the inspiratory flow rate of the ventilator. In addition to the comfort issue, large inspiratory swings in Ppl increase the afterload on the left ventricle [8, 9]. When poor ventricular function is decreased, the inspiratory rise in afterload can result in pulmonary edema [23–25] and respiratory failure postextubation [26]. One of our patients had a resting Ppao of 10 mmHg and a dPpao of 18 cmH2O in association with a dPeso of 29 cmH2O. This implies that the transmural pressure increased by 8 mmHg [(29 – 18 cmH2O)/1.36] and transmural left atrial pressure was actually 18 mmHg during inspiration.

In conclusion, in critically ill patients who have a pulmonary artery catheter in place as part of their overall management, dPpao gives an estimate of changes in Ppl during spontaneous inspiratory efforts, but generally underestimates the actual fall in Ppl and thus gives a “minimal” estimate of the magnitude of the inspiratory fall in Ppl.