Assessment of dead-space ventilation in patients with acute respiratory distress syndrome: a prospective observational study

Background Physiological dead space (VD/VT) represents the fraction of ventilation not participating in gas exchange. In patients with acute respiratory distress syndrome (ARDS), VD/VT has prognostic value and can be used to guide ventilator settings. However, VD/VT is rarely calculated in clinical practice, because its measurement is perceived as challenging. Recently, a novel technique to calculate partial pressure of carbon dioxide in alveolar air (PACO2) using volumetric capnography (VCap) was validated. The purpose of the present study was to evaluate how VCap and other available techniques to measure PACO2 and partial pressure of carbon dioxide in mixed expired air (PeCO2) affect calculated VD/VT. Methods In a prospective, observational study, 15 post-cardiac surgery patients and 15 patients with ARDS were included. PACO2 was measured using VCap to calculate Bohr dead space or substituted with partial pressure of carbon dioxide in arterial blood (PaCO2) to calculate the Enghoff modification. PeCO2 was measured in expired air using three techniques: Douglas bag (DBag), indirect calorimetry (InCal), and VCap. Subsequently, VD/VT was calculated using four methods: Enghoff-DBag, Enghoff-InCal, Enghoff-VCap, and Bohr-VCap. Results PaCO2 was higher than PACO2, particularly in patients with ARDS (post-cardiac surgery PACO2 = 4.3 ± 0.6 kPa vs. PaCO2 = 5.2 ± 0.5 kPa, P < 0.05; ARDS PACO2 = 3.9 ± 0.8 kPa vs. PaCO2 = 6.9 ± 1.7 kPa, P < 0.05). There was good agreement in PeCO2 calculated with DBag vs. VCap (post-cardiac surgery bias = 0.04 ± 0.19 kPa; ARDS bias = 0.03 ± 0.27 kPa) and relatively low agreement with DBag vs. InCal (post-cardiac surgery bias = −1.17 ± 0.50 kPa; ARDS mean bias = −0.15 ± 0.53 kPa). These differences strongly affected calculated VD/VT. For example, in patients with ARDS, VD/VTcalculated with Enghoff-InCal was much higher than Bohr-VCap (VD/VTEnghoff-InCal = 66 ± 10 % vs. VD/VTBohr-VCap = 45 ± 7 %; P < 0.05). Conclusions Different techniques to measure PACO2 and PeCO2 result in clinically relevant mean and individual differences in calculated VD/VT, particularly in patients with ARDS. Volumetric capnography is a promising technique to calculate true Bohr dead space. Our results demonstrate the challenges clinicians face in interpreting an apparently simple measurement such as VD/VT. Electronic supplementary material The online version of this article (doi:10.1186/s13054-016-1311-8) contains supplementary material, which is available to authorized users.


Background
Physiological dead space (V D,phys ) represents the fraction of ventilation not participating in gas exchange, including the airway (or anatomical) dead space (V D,aw ; i.e., ventilation of the conducting airways) and alveolar dead space (V D,alv ; i.e., ventilation receiving no pulmonary artery perfusion). In patients with acute respiratory distress syndrome (ARDS), dead space has prognostic value [1][2][3][4] and can be used to guide ventilator settings [5][6][7][8]. However, dead space is rarely calculated in clinical practice, because assessment of dead space is perceived as challenging and misunderstanding exists on different methods of calculation.
The first method used to calculate dead-space fraction (V D /V T ) was introduced in 1891 by Christian Bohr [9]: where V D is dead-space volume (i.e., volume not participating in gas exchange), V T is total exhaled volume, PACO 2 is the partial pressure of carbon dioxide in alveolar air, and PeCO 2 is the partial pressure of carbon dioxide in mixed expired air. V D calculated using Bohr's equation accurately measures V D,phys [10]. However, difficulties with measurement of PACO 2 led to rejection of this method. In 1938, Enghoff proposed replacement of PACO 2 by partial pressure of carbon dioxide in arterial blood (PaCO 2 ), also known as the Enghoff modification [11]. This modification is in general use today, but it comes with limitations. By substituting PaCO 2 for PACO 2 , intrapulmonary shunt and diffusion limitations are taken into the equation, resulting in a falsely elevated dead-space fraction [10,12]. Therefore, the Enghoff modification of Bohr's equation is not a measure of dead space as such but a global index of gas exchange impairment. Nevertheless, in clinical practice, the Enghoff modification is often falsely referred to as V D,phys . Another modification of the traditional Bohr formula uses the end-tidal partial pressure of carbon dioxide (PETCO 2 ) instead of PACO 2 [13]. In healthy subjects at rest, PETCO 2 almost equals PaCO 2 (and PACO 2 ), but during heavy exercise PETCO 2 overestimates PaCO 2 and in lung disease PETCO 2 underestimates PaCO 2 [14][15][16]. Recently, a novel technique for determining PACO 2 based on volumetric capnography was developed and validated [17,18]. With this technique, the eliminated concentration of CO 2 is plotted against the expired tidal volume, which allows breath-to-breath calculation of PACO 2 and Bohr dead space. However, in humans, volumetric capnography-based PACO 2 has been applied only to healthy and anesthetized subjects [19].
In addition to the difficulties with measurement of PACO 2 in Bohr's formula, there are different techniques for measuring its second component, PeCO 2 . First, with a Douglas bag, expired air can be collected and analyzed for the fraction of CO 2 . However, this method is laborintensive, and, in mechanically ventilated patients, gas compression and ventilator bias flow dilute expired air and should be corrected for [20]. Second, indirect calorimetry measures CO 2 production :VCO 2 ð Þ, which can be used to calculate PeCO 2 . Third, the most commonly used and easiest method to determine PeCO 2 is volumetric capnography.
The purpose of the present study was to evaluate how different techniques of measuring PACO 2 and PeCO 2 affect calculated dead-space ventilation in mechanically ventilated patients with ARDS and normal lung function. PACO 2 was calculated using volumetric capnography or replaced with PaCO 2 . PeCO 2 was calculated using the Douglas bag, indirect calorimetry, and volumetric capnography.

Study subjects
We conducted a prospective, observational study in the intensive care unit of the Radboud University Medical Center in Nijmegen, The Netherlands. The protocol was approved by the institutional review board (CMO regio Arnhem-Nijmegen) and was in accordance with the ethical standards laid down in the 1964 Declaration of Helsinki and its later amendments. The institutional review board waived the need for informed consent.

Study design
Two patient groups were studied: 15 patients who underwent elective post-cardiac surgery and 15 patients fulfilling the Berlin Definition of ARDS [21]. Exclusion criteria were hemodynamic instability (mean arterial pressure <65 mmHg despite vasopressors) in both groups and past medical history of lung disease in the post-cardiac surgery patients. All patients were ventilated with a SERVO-i ventilator (Maquet Critical Care, Sölna, Sweden) and disposable tubing (patients with ARDS, Evaqua breathing circuit, Fisher & Paykel Healthcare, Auckland, New Zealand; post-cardiac surgery patients, Limb-O breathing circuit, GE Healthcare, Little Chalfont, UK). Mechanical ventilator settings were not adjusted during the study. Fraction of inspired oxygen (FiO 2 ) and positive end-expiratory pressure (PEEP) were set according to the lower PEEP/higher FiO 2 arm of the ARDSNet protocol.

Enghoff-DBag
Dead space with Enghoff-DBag was calculated using the Enghoff modification: PaCO 2 was determined using an arterial blood gas sample derived from an arterial catheter. Expired air was collected during 2 to 3 minutes to obtain a representative sample from the expiratory port of the ventilator in a 25-L Douglas bag. PeCO 2 was determined using a sample taken from the bag with a 50-ml syringe (BD Plastipak; BD, Drogheda, Ireland), which was analyzed using the Siemens Rapidlab 865 (Diamond Diagnostics, Holliston, MA, USA). The coefficient of repeatability of the Rapidlab was 0.03 kPa.
PeCO 2 in the expired air was corrected for dilution due to gas compression in the ventilator circuit [20], as well as for ventilator bias flow (2 L/min): bias flow volume ¼ expiratory time Â bias flow ð4Þ where P peak is inspiratory peak pressure. The compliance of the ventilator circuit was determined during an internal ventilator test in each patient.
Gas sampling was performed via side-stream sampling with a connection piece (dead space 9.5 ml) distal to the Y-piece. PeCO 2 was calculated as follows: where k is the gas constant (0.115 when expressing PeCO 2 in kilopascals), :VCO 2 is CO 2 production (in milliliters per minute standard temperature dry pressure] and :V : is minute ventilation (in liters per minute body temperature standard pressure). :VCO 2 and :V : were stored per minute on the monitor. An average of at least 5 minutes was used for calculations.

Bohr-VCap
For Bohr-VCap, dead space was calculated using the Bohr equation (Eq. 1). Flow and arterial carbon dioxide tension (PCO 2 ) were measured using the NICO capnograph (Philips Respironics, Murrysville, PA, USA). The capnograph consists of a mainstream CO 2 sensor (CAPNOSTAT; Philips Respironics) using infrared absorption technology and a flow sensor connected to the CAPNOSTAT attached distal to the Y-piece (dead space 8.5 ml). Flow and PCO 2 were acquired at a sampling rate of 200 Hz and stored for offline analysis. Offline analysis was performed with an algorithm developed for MATLAB (MathWorks, Natick, MA, USA). The volumetric capnogram was obtained per breath by plotting PCO 2 against expired volume. The volumetric capnogram was averaged over a period of 2 minutes, selected by visual inspection to ensure no artifacts. The latter was necessary to correct for respiratory variability (particularly with pressure support ventilation) and thus obtain a representative breath (Additional file 1: Fig. S1). PACO 2 , PeCO 2 , and V D,aw were determined from the volumetric capnogram using model fitting (Additional file 1: Fig. S2) as described by Tusman and colleagues [17]. Briefly, mean PACO 2 was calculated as the midpoint of phase III in the volumetric capnogram, and PeCO 2 was calculated as the area under the curve of the volumetric capnogram divided by expiratory volume. The position of the airway-alveolar interface (V D,aw ) was calculated as the inflection point of phase II of the volumetric capnogram. Consequently, V D,alv could be calculated as follows:

Enghoff-VCap
For Enghoff-VCap, dead space was calculated using the Enghoff modification (Eq. 2). PeCO 2 was determined from the volumetric capnogram as described in the preceding subsection.

Statistical analysis
Statistical analysis was performed with Prism 5 software (GraphPad Software Inc., San Diego, CA, USA). The normality of the distribution of the data was determined with the D' Agostino-Pearson test. Normally distributed variables were expressed as mean ± standard deviation. Nonparametric data were expressed as median [interquartile range]. Paired t tests and Bland-Altman analysis were used for comparisons. P < 0.05 was considered statistically significant. Table 1 reports patient characteristics and ventilator settings. Figure 1 shows representative examples of the volumetric capnogram of post-cardiac surgery patients and patients with ARDS. Average values of PACO 2 , PaCO 2 , PeCO 2 , and V D /V T for both groups, measured and calculated with the different methods, are given in Table 2.

Results
PACO 2 , PETCO 2 , and PaCO 2 For both patient groups, there was a significant difference between PACO 2 , PETCO 2 , and PaCO 2 , confirming that these parameters are not interchangeable (Fig. 2). As expected, these differences were much more pronounced in patients with ARDS (  Fig. S3). PeCO 2 after correction for bias flow and compressible volume was 0.65 ± 0.11 kPa and 0.39 ± 0.16 kPa higher than when uncorrected for post-cardiac surgery and patients with ARDS, respectively (Additional file 1: Fig. S4).

Dead space
Large differences in calculated dead space were present between the four different methods ( Table 2). Compared with Bohr-VCap, dead space calculated with Enghoff-VCap (PACO 2 replaced with PaCO 2 , but similar PeCO 2 ) increased dead space by 31 ± 18 % and 52 ± 15 % for the post-cardiac surgery patients and patients with ARDS, respectively. Figure 4 shows Bland-Altman plots of dead space obtained with different methods. In post-cardiac surgery patients, the mean bias in V D /V T between Enghoff-DBag vs. Bohr-VCap was 10 ± 6 %, and between Enghoff-DBag vs. Enghoff-InCal it was 22 ± 10 %. In patients with ARDS, the mean bias in V D /V T between Enghoff-DBag vs. Bohr-VCap was 23 ± 7 %, and between Enghoff-DBag vs. Enghoff-InCal it was 2 ± 8 %.
Changes in intrapulmonary shunt and diffusion have a greater effect on Enghoff-VCap than Bohr-VCap. Partial pressure of oxygen in arterial blood (PaO 2 )/FiO 2 ratio (PF ratio) may be used as an indicator of these lung parameters. Figure 5 shows the correlation between dead space (Bohr-VCap and Enghoff-VCap) and PF ratio.
Values of V D,aw and V D,alv calculated from the volumetric capnogram are presented and discussed in Additional file 1: Fig. S5.

Discussion
The present study demonstrates the consequences of applying different techniques for measuring PACO 2 and PeCO 2 to calculate dead space in mechanically ventilated patients with ARDS and normal lung function. To our knowledge, we are the first to evaluate a novel method to calculate PACO 2 using volumetric capnography in patients with ARDS. We show that the differences introduced by replacing PACO 2 with PaCO 2 are more pronounced in patients with ARDS than in mechanically ventilated patients with normal lung function. Furthermore, the different techniques used to measure PeCO 2 introduce potential and clinically relevant sources of error in calculating dead space. These findings have important implications for calculating dead space in daily clinical practice.

Alveolar and arterial PCO 2
PACO 2 is the mean value of CO 2 in the alveolar compartment, which depends on the balance between perfusion and ventilation of the lung units [10]. The replacement of PACO 2 with PaCO 2 in the Bohr formula (Enghoff modification) was proposed to avoid the difficulties of identifying an appropriate PACO 2 . However, in contrast to PACO 2 , PaCO 2 is affected by intrapulmonary shunt and diffusion impairment [22,23]. In a healthy lung, the difference between PACO 2 and PaCO 2 is minimal but will increase for any gas exchange abnormality. Indeed, we found that the gradient between PACO 2 and PaCO 2 is much higher in patients with ARDS than in patients without lung disease (Fig. 2). The former has a strong effect on the calculated dead space in patients with ARDS (52 % increase). Hence, the Enghoff modification of Bohr dead space is not a dead-space measurement as such, but a global index of gas exchange impairment. This is illustrated in Fig. 5, where dead space calculated with Enghoff-VCap shows a strong correlation (r 2 = 0.54) with PF ratio, whereas this correlation is weak (r 2 = 0.12) with Bohr-VCap. In other words, the use of true alveolar PCO 2 makes dead-space calculation less dependent on intrapulmonary shunt and diffusion impairment. Even with Bohr-VCap, we found that dead space in patients with ARDS was higher than in post-cardiac surgery patients. This may be explained by the difference in lung condition but also by the difference in tidal volume between the groups. A lower tidal volume relatively increases dead space.   In the present study, we used three techniques (DBag, VCap, and InCal) to measure PeCO 2 . In the last decade, researchers in several clinical studies compared these techniques as well [24][25][26]. None of the studies included comparisons of all three techniques, but a high agreement in PeCO 2 was found previously between VCap and DBag [25] and between VCap and InCal [24,26]. In accordance with these results, we found a high agreement between DBag and VCap in patients with ARDS and in post-cardiac surgery patients (Fig. 3). However, the accuracy of indirect calorimetry to measure PeCO 2 appeared lower. First, the 95 % limits of agreement were larger with DBag vs. InCal compared with DBag vs. VCap in both patient groups. Second, the mean bias between DBag and InCal showed large offset in postcardiac surgery patients. It is important to note that, with the Douglas bag, expired air is collected at the expiratory limb of the ventilator circuit and is consequently diluted by CO 2 -free air coming from compressed volume and bias flow volume. In the present study, dilution of expired air lowered PeCO 2 . The effect of dilution becomes larger as the ratio between bias flow volume and expired volume increases. This ratio is higher in post-cardiac surgery patients, who have, in general, a relatively long expiration time compared with patients with ARDS. The reliability of using a correction factor to estimate the degree of dilution depends primarily upon the accuracy of the recorded peak pressure and expired tidal volume for compressed volume [24] and the expiratory time for bias flow volume. Volumetric capnography measures expired CO 2 distal to the Y-piece of the ventilator circuit and is unaffected by compression volume and bias flow.

Clinical implications
The techniques used in the present study cause large differences in calculated dead space (Table 2 and Fig. 4) and demonstrate the difficulties encountered in clinical practice. These differences are dependent on the choice of dead-space formula (Bohr or Enghoff modification) and the technique used to measure PeCO 2 , as discussed above, and they have important clinical implications. First, one should never use different techniques to calculate dead space in follow-up of a patient. Second, several studies have demonstrated that elevated dead space in patients with ARDS is associated with an increased risk of mortality [1][2][3][4]. The researchers in these studies calculated the Enghoff modification of Bohr dead space and thus calculated an index of global gas exchange impairment and not true dead space. Therefore, it is unknown whether true Bohr dead space measured with VCap has similar prognostic value. Third, a question remains regarding which method clinicians should use at the Dotted lines represent 95 % limits of agreement, and dashed lines represent mean bias bedside to determine dead space. The answer depends on the clinical problem to be addressed and the techniques available. Nowadays, there are several capnographs available that provide dead-space values at the bedside. These include stand-alone monitors (e.g., NICO capnograph) or modules incorporated into the mechanical ventilator (e.g., Evita Infinity V500, Dräger Medical, Lübeck, Germany; HAMILTON-G5, Hamilton Medical, Bonaduz, Switzerland). However, these capnographs are not able to calculate alveolar PCO 2 (and thus Bohr dead space), as demonstrated in our study, and still require manual entry of PaCO 2 to determine dead space according Enghoff 's modification. If one's goal is to improve or follow up overall gas exchange, it complies is appropriate to take an arterial blood gas samples and use the Enghoff modification. However, if one wants to evaluate the effect of different ventilator settings on alveolar dead space, one must calculate Bohr dead space (i.e., physiological dead space). For example, differences in endexpiratory lung volume and extrinsic PEEP levels greatly affect airway and alveolar dead space [27][28][29]. In case of high PEEP, vessels can be compressed by overdistention of alveoli, which causes alveolar perfusion to decrease and consequently increases alveolar dead space.
However, high PEEP may also overcome atelectasis and thereby increase alveolar recruitment and reduce pulmonary shunting. If dead space is measured using the Enghoff modification, it is not possible to discriminate between the effects of PEEP on pulmonary shunt and alveolar dead space.

Study limitations
The gold standard for calculating Bohr dead space is the mathematical algorithm of the multiple inert gas elimination technique (MIGET), an approach that allows quantification of all the pulmonary and extrapulmonary determinants of arterial oxygenation. Due to the complexity of the MIGET technique, it is never used in clinical practice and rarely in clinical studies. Nevertheless, it is reasonable to assume that Bohr dead space calculated using volumetric capnography in our study provided an accurate estimate. First, the concept of obtaining PACO 2 from the midportion of phase III with volumetric capnography has recently been validated against the MIGET technique in lung-lavaged pigs [18]. Second, our values of dead space were comparable with the only clinical study in patients with ARDS in the current era of low tidal volumes in which researchers calculated dead space using both the MIGET technique and the Enghoff modification [30]. In that study, V D /V T,Bohr was 40 % and V D /V T,Enghoff was 65 %, compared with 45 % and 68 %, respectively, in our present study.
Previously, using the similar volumetric capnography technique as used in the present study, V D /V T,Bohr was found to be 23 % in healthy subjects and 28 % in anesthetized patients undergoing elective, noncomplex, and neither laparoscopic nor thoracic surgeries in supine position [19]. In our post-cardiac surgery patients, V D / V T,Bohr was markedly higher at 38 %. This difference is most likely the result of a longer surgical procedure, open chest surgery, hypovolemia, and higher PEEP in our post-cardiac surgery patients.
With volumetric capnography, the calculation of PACO 2 depends on the determination of the intersections of the tangents of phases II and III (Additional file 1: Fig. S2) [17]. In post-cardiac surgery patients and in most patients with ARDS, this intersection is present. However, in some patients with ARDS, phase III can be very steep due to severe heterogeneity of the lung. Consequently, there is no definite transition from phase II to phase III and hence no intersection of the tangent of phases II and III, which leads to false calculation of PACO 2 . The latter occurred in one of our patients, who was excluded from analysis.

Conclusions
Use of different techniques to measure PACO 2 and PeCO 2 results in clinically relevant mean and individual Correlation between dead space and PaO 2 /FiO 2 ratio. Dead space was calculated using volumetric capnography with PaCO 2 (Enghoff-VCap) and PACO 2 (Bohr-VCap). Dead space calculated with Enghoff-VCap shows a strong correlation (r 2 = 0.54) with PaO 2 /FiO 2 ratio (PF ratio), whereas this correlation is weak (r 2 = 0.12) with Bohr-VCap. Thus, the use of PACO 2 makes dead-space calculation less dependent on intrapulmonary shunts and diffusion impairment. PaO 2 partial pressure of oxygen in arterial blood, PaCO 2 partial pressure of carbon dioxide in arterial blood, PACO 2 partial pressure of carbon dioxide in alveolar air, PF PaO 2 /FiO 2 ratio, V D /V T dead-space fraction