In acute lung injury (ALI) and acute respiratory distress syndrome (ARDS), functional residual capacity (FRC) is markedly decreased as a result of numerous factors, including alveolar collapse, pulmonary edema with alveolar flooding, supine position, sedation-induced diaphragm inactivity, and cardiac enlargement [15]. Measuring FRC (or end-expiratory lung volume [EELV] when PEEP is applied) might help to measure the aerated lung available for ventilation and to better monitor the effects of ventilation strategies. Reproducible measurement techniques that can be used at the bedside are needed to minimize overdistention and to determine which patients may benefit from recruitment strategies. Repeated CT scans and gas-dilution techniques are two validated methods of lung-volume measurement but are so complex that their use has been confined to research settings. Recently, washout/washin techniques using oxygen [6, 7] or nitrogen [8, 9] have been made available in ICU ventilators, allowing bedside EELV measurement. A comparison of the nitrogen washout/washin EELV measurement with helium dilution or CT scan had shown good correlations in stable patients ventilated with low-PEEP levels [8]. The limitations of the nitrogen washout/washin technique for EELV measurement under other conditions, such as high FiO2 or high PEEP, have not been fully investigated [10].

PEEP-induced changes in lung volume (referred to as PEEP-volume) can also be assessed simply at the bedside by using passive spirometry. This accurate method requires a long expiration to zero end-expiratory pressure (ZEEP), where FRC is assumed to be reached. When considering the changes induced by two different levels of PEEP in a given patient, the difference in EELV (that is, ΔEELV = EELVhigh PEEP - EELVlow PEEP) should theoretically be similar to the difference in ΔPEEP-volume (PEEP-volumehigh PEEP - PEEP-volumelow PEEP), assuming that the FRC has not been modified by the PEEP changes (see Figure 1).

Figure 1
figure 1

Schematic representation of the pressure-time diagram at the end of each epoch in a single patient. Data at both PEEP levels studied are presented. Patm, atmospheric pressure; EELV, end-expiratory lung volume measured by using the nitrogen technique; PEEP-volume, volume trapped by PEEP above the functional residual capacity, measured by using a long exhalation to atmospheric pressure (zero end-expiratory pressure); Vt, tidal volume; ΔEELV, EELVhigh PEEP - EELVlow PEEP ; ΔPEEP-volume, PEEP-volumehigh PEEP - PEEP-volumelow PEEP.

We designed a multicenter study with the primary objective of investigating the precision (reproducibility) of the nitrogen washout/washin technique for EELV measurement in patients with ALI/ARDS at two PEEP levels, including a high level, with a small variation in oxygen concentration (10%). Our secondary objective was to evaluate the accuracy of the technique by comparing PEEP-induced changes (Δ) in lung volume with the nitrogen technique or the PEEP-volume above the FRC measured with passive spirometry. As PEEP-volume is relatively easy to measure accurately with a calibrated pneumotachograph, it may therefore be considered a "gold standard." Because we expected possible discrepancies between the two techniques, we also compared the measured changes in lung volume (ΔEELV and ΔPEEP-volume) with the minimal predicted increase in lung volume, computed from static compliance (Cstat) at low PEEP and ΔPEEP. The minimal predicted increase in lung volume was considered the smallest-volume increase that can occur. We have also used this method to evaluate alveolar recruitment, as described elsewhere [11].

Material and methods

This was a multicenter study performed in five French medical intensive care units at the Henri Mondor University Hospital in Créteil, European Georges Pompidou University Hospital in Paris, Angers University Hospital in Angers, l'Archet 1 University Hospital in Nice, and Charles Nicolle University Hospital in Rouen. In compliance with French legislation, the institutional review board of the Henri Mondor University Hospital approved the protocol for all centers and waived the need for informed consent, as PEEP optimization was considered part of standard care. The patients or next of kin received information about the study.


Patients were enrolled if they met the standard criteria for acute lung injury (ALI) [12]: partial pressure of arterial oxygen over fraction of inspired oxygen (PaO2/FiO2) less than 300 mm Hg, bilateral pulmonary infiltrates on the chest radiograph, and no clinical evidence of left atrial hypertension. Most patients had ARDS, defined as PaO2/FiO2 less than 200 mm Hg. Exclusion criteria were age younger than 18 years, pregnancy, history of chronic obstructive pulmonary disease and/or lung surgery, and hemodynamic instability, defined as an increase in vasoactive drug (epinephrine, norepinephrine) dosages in the last 6 hours. All bedside anterior-posterior chest radiographs were reviewed by two independent observers (JJR and QL) according to CT Scan ARDS Study Group criteria to determine the pattern of aeration loss: lobar radiologic hyperattenuation predominating in the lower lobes (focal disease), diffuse radiologic hyperattenuation evenly distributed throughout the upper and lower lobes (white lungs), or patchy radiologic hyperattenuation involving the upper and lower lobes with persistent aeration of part of the upper lobes [13]. Patients with diffuse or patchy aeration loss were classified as having nonfocal disease [14].

Ventilation strategies

All patients received volume-assist control ventilation by using an Engström ICU ventilator (Version V4 and V5) with a CVOX module sensor (V4.5) General Electric, Madison (WI). This ventilator provides bedside EELV measurements by using the multibreath nitrogen-washout technique (MBNW) [8, 1518]. The oxygenation goal was achieved by adjusting FiO2, which was maintained constant during the study. Tidal volume was set at 6 ml/kg of predicted body weight. All patients received two PEEP levels, each for 45 minutes, in random order. PEEP levels were set as in the EXPRESS study [19]. In the minimal-distention strategy, PEEP and inspiratory Pplat were kept as low as possible while keeping arterial oxygen saturation at 88% to 92% or more. External PEEP was set to maintain total PEEP (the sum of external and intrinsic PEEP) between 5 and 9 cm H2O. In the optimized recruitment strategy, PEEP was adjusted based on Pplat and was kept as high as possible without increasing the inspiratory Pplat above 28 to 30 cm H2O. All patients were sedated. Neuromuscular blocking agents were administered only if deemed necessary by the clinician in charge.


Lung volume and precision of measurements

At the end of each 45-minute period, blood was drawn for arterial blood gas measurement, and EELV was measured 3 times by using the MBNW technique to assess precision. This technique has been described elsewhere [9, 16]. In brief, continuous measurement of end-tidal O2 and CO2 during a change in FiO2 (here, 10%) allows the calculation of nitrogen washout and washin and subsequently of the aerated lung volume. Two assumptions are made: heterogeneity in alveolar gas distribution is considered constant during the measurement procedure, and cellular metabolism and gas exchange between lung capillaries and alveoli are considered stable during the MBNW procedure. The mean of the washout and washin data is computed automatically if the difference between the two is less than 20% (cut-off determined by the manufacturer). Because FRC is a volume measured without PEEP (that is, at atmospheric pressure), we used the term end-expiratory lung volume (EELV) for the volume measured in our study. Three EELV measurements were performed at each PEEP level.

PEEP-volume (above FRC) by using passive spirometry

Prolonged exhalation (15 seconds) to the elastic equilibrium volume at ZEEP was performed, at the end of a 45-minute period, to standardize lung-volume history. Pressure and flow were recorded by using a dedicated computer linked to the ventilator (sample every 0.04 seconds), pressure, and flow curves were drawn off-line by using the software (Acknowledge 3.7.3) Goleta Ca. Volumes were measured by flow integration. PEEP-volume above FRC was obtained by subtracting the insufflated tidal volume from the flow-signal integration of this long exhalation. PEEP-volume was measured at the end of each of the two PEEP periods.

Measurement of compliance

Cstat of the respiratory system was computed by dividing tidal volume by Pplat (measured during an end-inspiratory pause (1 second)) minus total PEEP. Total PEEP was measured by using an expiratory pause (1 second).

A pressure-volume curve was obtained during low-flow inflation from the low PEEP level to 30 cm H2O to check that compliance (Clin) was linear or not decreasing within this range.

Minimal predicted increase in lung volume

The minimal predicted increase in lung volume [20] is the smallest possible lung-volume increase that can be induced by PEEP. It was computed from Cstat at low PEEP, as follows:

Minimal predicted increase in lung volume (milliliters) = CstatlowPEEP· ΔPEEP

where ΔPEEP is the difference between high and low PEEP.

This minimal increase should be equal to (if no recruitment occurs) or smaller than (if alveolar recruitment occurs) ΔEELV and ΔPEEP-volume. We evaluated the slope of the pressure-volume curve during tidal inflation to check that compliance did not decrease over tidal inflation and, therefore, that the computed minimal increase was indeed the lowest possible increase that could occur.

Statistical analysis

All variables are described as median (interquartile range). Precision of the nitrogen technique results was assessed by calculating the coefficient of variation for the three pairs of washout/washin measurements. The coefficient of variation was calculated as the SD of the differences divided by the mean of all measurements. The Bland and Altman method [21] was used to evaluate reproducibility of the nitrogen technique and to evaluate agreement between ΔEELV and ΔPEEP-volume. The largest difference between the three EELV measurements at each PEEP level was plotted against the mean. Accuracy of the technique was assessed by comparing the changes in lung volume induced by the PEEP increase. ΔEELV was plotted against ΔPEEP-volume. Correlations were evaluated by using linear regression (r2 ). Paired values were compared by using the Wilcoxon test. The Fisher t test and Mann-Whitney U test were used when appropriate. Values of p smaller than 0.05 were considered significant.


We studied 37 patients, of whom three were excluded from the analysis because of poor signal quality (two patients had unstable signals during PEEP-volume recording (spontaneous breathing), and one had greater than 20% differences between washout and washin values). Table 1 reports the main characteristics of the 34 patients included in the analysis, 32 with ARDS and two with ALI. Table 2 reports data on ventilation mechanics, ventilator settings, measured volumes, and calculated volumes at each PEEP level. Both PEEP strategies were well tolerated by all patients. No patients experienced any significant desaturation during the study measurements (EELV or PEEP-volume).

Table 1 Patient characteristics
Table 2 Arterial blood gas values and ventilation during the minimal-distention (low PEEP) and high-recruitment (high PEEP) periods

Precision of the nitrogen technique

The 34 patients had three pairs of EELV measurements at each PEEP level (that is, 204) (Figure 2). Of these measurements, six (2.9%), in six different patients (two at low PEEP and four at high PEEP) showed greater than 20% differences between washout and washin values and were therefore excluded. The coefficient of variability for the remaining measurements was 3.0% at low PEEP and 3.9% at high PEEP (p < 0.0001).

Figure 2
figure 2

Largest difference (absolute values) between the three EELV values obtained in each patient, as a function of EELV. The difference between the measured values was larger at high PEEP (p = 0.004). When expressed as a percentage of EELV, no difference was observed according to PEEP level. Gray diamonds, low PEEP; solid circles, high PEEP; red diamond, mean value at low PEEP; blue circle, mean value at high PEEP; vertical and horizontal bars, the standard deviation.

The largest mean difference between the three pairs of EELV measurements was 81 ± 64 ml. The difference was larger at higher PEEP levels (53 ± 43 ml versus 108 ± 69 ml; p = 0.004) but was similar when expressed as a percentage of EELV (Figure 2). Mean FiO2 was 67 ± 17%; the highest FiO2 levels were not associated with greater EELV variability.

Comparison with PEEP-induced changes in lung volume and accuracy of the method

Minimal predicted increase in lung volume, ΔEELV, and ΔPEEP-volume are shown in Figure 3.

Figure 3
figure 3

Median and interquartile range of minimal predicted increase in lung volume, ΔEELV, and ΔPEEP-volume. Minimal predicted increase in lung volume, 330 (190 to 421) ml. ΔEELV, 402 (263 to 654) ml. ΔPEEP-volume, 585 (325 to 822) ml.

ΔEELV and ΔPEEP-volume were only modestly correlated with each other (Figure 4a) (ΔEELV = 62.4 + [0.73 ΔPEEP-volume]; r2 = 0.47). Bias between these two measuring methods was 97 ± 255 ml, with a 95% confidence interval for limits of agreement of -414 to 608 ml (Figure 5).

Figure 4
figure 4

ΔEELV and ΔPEEP-volume correlation. (a) Correlation between ΔEELV and ΔPEEP-volume in all patients (r2 = 0.47). Straight line is correlation: ΔEELV = 62.4 + 0.7 ΔPEEP-volume. (b) Relation between the minimal predicted increase in lung volume and ΔEELV. Red dots: patients in whom measurement errors were detected; dashed line, identity. (c) Correlation between ΔEELV and ΔPEEP-volume after exclusion of the four patients with obvious ΔEELV measurement errors (r2 = 0.80). Straight line is correlation: ΔEELV = -42.1 + 1.0 ΔPEEP-volume.

Figure 5
figure 5

Comparison according to Bland and Altman [21]of measurements of ΔPEEP-volume and ΔEELV. Bias between the two methods was 97 ± 255 ml with a 95% confidence interval for the limits of agreement (dashed lines) of -414 to 608 ml.

The relation between the minimal predicted increase in lung volume and ΔEELV was dispersed (Figure 4b). In particular, four patients had ΔEELV values that were substantially lower than the minimal predicted increase in lung volume (red dots; Figure 4b), suggesting underestimation of the volume change by EELV measurement. All four patients received PEEP levels ≥ 16 cm H2O, compared with only five of the 30 remaining patients (p = 0.003), and three had focal aeration loss compared with only three (10%) of the 30 remaining patients (p = 0.01). FiO2 was high (80% ± 16%) in these patients but was not significantly higher than that in the other patients (p = 0.1). The cause of ARDS (pulmonary or extrapulmonary) was not associated with measurement discrepancies. The high PEEP values suggested possible occurrence of leaks that could invalidate the measurements. When we excluded these four patients whose ΔEELV values were lower than the predicted minimal increase in lung volume, the correlation between ΔEELV and ΔPEEP-volume became substantially stronger (r2 = 0.80; Figure 4c).


The main results of this physiological study can be summarized as follows: (a) the MBNW technique at two PEEP levels provided reproducible EELV measurements with acceptable precision; and (b) compared with ΔPEEP-volume and the minimal predicted increase in lung volume due to PEEP, ΔEELV measured by using the nitrogen technique seemed accurate for measuring lung-volume variations induced by PEEP. In a few patients, however, the method could give erroneous results, especially in case of high pressures. Comparing with the minimal predicted increase in lung volume may help to detect these errors.

Nitrogen technique variability

The MBNW technique described by Olegard et al. [9] allows bedside EELV measurement by using small and safe FiO2 increases and decreases (± 10%). Precision was greater with larger FiO2 changes [8, 16], because nitrogen changes were greater. The small (10%) FiO2 change used in our study may have contributed to the test-retest variability but was deemed safer for our hypoxemic patients. All measurements were performed at the steady state 45 minutes after a change in PEEP, and no other interventions likely to affect cardiac output were performed, the patients being considered stable. Fewer than 3% of the EELV measurements failed (greater than 20% difference between washout and washin). Because the technique used to measure EELV involves computing the mean of washin and washout values [9], we assessed test-retest variability without comparing washout with washin. The variability we found in patients with ALI or ARDS at each PEEP level was comparable to that reported by Olegard et al. [9], who studied chiefly postoperative patients. As with the helium-dilution technique, absolute variability of the nitrogen technique in our study increased with higher PEEP and higher EELV. However, variability relative to absolute lung volume did not differ for higher EELV values (Figure 2). The lower precision reported by the manufacturer for FiO2 > 70% was not replicated here, but the flawed measurements seemed to occur at higher FiO2 values.

PEEP-induced changes in lung volume

EELV values at low PEEP in our study were very low (less than 1,000 ml at low PEEP) and similar to values obtained previously by using CT scan [2, 22] or helium dilution [23] in ARDS patients. PEEP-volume and EELV represent different volumes obtained with two totally independent methods. We thus compared lung-volume changes induced by PEEP. ΔEELV and ΔPEEP-volume; both evaluated the PEEP-induced lung volume increase. The correlation was good in some patients but poor in others (Figure 5). The variability of EELV values may have contributed to a poor correlation. We sought to detect obviously flawed data by using a third method. Katz et al. [20] demonstrated that the lung-volume increase induced by PEEP changes was larger than expected from the airway-pressure change and compliance at low PEEP, indicating progressive lung recruitment [11]. We therefore calculated the minimal predicted increase in lung volume induced by PEEP, which is easily derived from Cstat at low PEEP [20]. In addition, by tracing a pressure-volume curve over the tidal-volume range at low PEEP, we checked that compliance did not decrease significantly within this volume range, to ensure that no volume increase smaller than the calculated minimal increase could occur. This method might prove useful at the bedside to assess the lower ΔEELV limit. Any difference between ΔEELV and this minimal predicted increase in lung volume may be considered an estimate of alveolar recruitment [11]. ΔPEEP-volume may slightly underestimate the lung-volume change, because of the assumption that FRC is unchanged after exhalation from high or low PEEP (Figure 3). Yet recent data [24] suggest that FRC may increase after high PEEP compared with low-PEEP ventilation. We used a 15-second expiration to ZEEP to minimize this problem. Our analysis, made at two PEEP levels, shown elsewhere, suggested that FRC was stable for our measurements [11].

Obvious discrepancies occurred in four patients. All four patients had the highest set PEEP levels (> 16 cm H2O). Although not proven, it is very possible that microleaks due to high set PEEP may explain discrepancies by decreasing the EELV high PEEP measurement and therefore ΔEELV. The higher set FiO2 values in these four patients may have adversely affected measurement precision, although further studies are needed to evaluate this possibility. Patients with focal aeration loss are at higher risk of hyperinflation versus recruitment [25], and the lung-volume distribution due to PEEP depends closely on disparities in regional lung compliance [26]. Another hypothesis could be that EELV discrepancies in patients with higher PEEP and focal aeration loss may be related to differences in regional gas distribution. MBNW equilibration may be impaired by regional time-constant inequalities [27], and a higher dead space due to higher PEEP [28] and hyperinflation [2931]. In clinical practice, we suggest comparing the increase in EELV with PEEP to the minimal predicted increase in lung volume to detect erroneous measurements.


The MBNW technique exhibits acceptable accuracy and precision for lung-volume measurement at different PEEP levels in patients with ARDS. Substantial underestimation of lung-volume changes may occur, at least in some patients, presumably in case of leaks due to high pressures, and additional measurements may be required to check this accuracy.

Key messages

  • Nitrogen washin/washout technique exhibits acceptable accuracy and precision for lung-volume measurement at different PEEP levels and high FiO2 in patients with ARDS.

  • Underestimation of lung-volume changes may occur in some patients presumably in case of leaks due to high pressures.