End-inspiratory pause prolongation in acute respiratory distress syndrome patients: effects on gas exchange and mechanics

Background End-inspiratory pause (EIP) prolongation decreases dead space-to-tidal volume ratio (Vd/Vt) and PaCO2. We do not know the physiological benefits of this approach to improve respiratory system mechanics in acute respiratory distress syndrome (ARDS) patients when mild hypercapnia is of no concern. Methods The investigation was conducted in an intensive care unit of a university hospital, and 13 ARDS patients were included. The study was designed in three phases. First phase, baseline measurements were taken. Second phase, the EIP was prolonged until one of the following was achieved: (1) EIP of 0.7 s; (2) intrinsic positive end-expiratory pressure ≥1 cmH2O; or (3) inspiratory–expiratory ratio 1:1. Third phase, the Vt was decreased (30 mL every 30 min) until PaCO2 equal to baseline was reached. FiO2, PEEP, airflow and respiratory rate were kept constant. Results EIP was prolonged from 0.12 ± 0.04 to 0.7 s in all patients. This decreased the Vd/Vt and PaCO2 (0.70 ± 0.07 to 0.64 ± 0.08, p < 0.001 and 54 ± 9 to 50 ± 8 mmHg, p = 0.001, respectively). In the third phase, the decrease in Vt (from 6.3 ± 0.8 to 5.6 ± 0.8 mL/Kg PBW, p < 0.001) allowed to decrease plateau pressure and driving pressure (24 ± 3 to 22 ± 3 cmH2O, p < 0.001 and 13.4 ± 3.6 to 10.9 ± 3.1 cmH2O, p < 0.001, respectively) and increased respiratory system compliance from 29 ± 9 to 32 ± 11 mL/cmH2O (p = 0.001). PaO2 did not significantly change. Conclusions Prolonging EIP allowed a significant decrease in Vt without changes in PaCO2 in passively ventilated ARDS patients. This produced a significant decrease in plateau pressure and driving pressure and significantly increased respiratory system compliance, which suggests less overdistension and less dynamic strain.


Background
Mechanical ventilation in patients with acute respiratory distress syndrome (ARDS) must combine both low tidal volumes (Vt) and adequate positive end-expiratory pressure (PEEP) [1,2]. However, in patients with ARDS, respiratory acidosis and high airway plateau pressures (P plat ) may limit management of ventilatory adjustments. In particular, the functional consequences of hypercapnia and respiratory acidosis may differ considerably depending on a patient's condition, and they may involve almost any physiological function [3][4][5][6].
Optimization of mechanical ventilation parameters is associated with a reduction in dead space and is a useful strategy to reduce hypercapnia in ARDS patients [7]. Many other strategies have also been developed to decrease hypercapnia at the bedside, such as increases in respiratory rate [8], use of active humidifiers [9] and the tracheal gas insufflation [10] or aspiration of dead space [11]. At bedside, the dead space could be calculated using the Enghoff modification of the Bohr equation. The use of this equation implies the use of PaCO 2 as surrogate for Open Access *Correspondence: jmancebo@santpau.cat 1 Servei de Medicina Intensiva, Hospital de la Santa Creu i Sant Pau, Universidad Autònoma de Barcelona (UAB), Sant Quintí, 89, 08041 Barcelona, Spain Full list of author information is available at the end of the article alveolar carbon dioxide. Therefore, this equation measures a global index of efficiency of gas exchange because it takes also shunt effect into account [12].
Some authors have also shown that prolonging the end-inspiratory pause (EIP) is a feasible maneuver to achieve similar targets [13,14]. In experimental models [15] and in ARDS patients [14,[16][17][18], EIP prolongation has proven effective at enhancing CO 2 elimination and decreasing partial pressure of carbon dioxide in arterial blood (PaCO 2 ) and also physiological dead space (Vd phys ). Prolonging EIP extends the time available for an enhanced diffusion between inhaled Vt and resident alveolar gas, thus facilitating the transfer of CO 2 from alveoli toward the airways [17,18].
Although several of the physiological studies described above have reported that EIP prolongation improves gas exchange, none have investigated the potential physiological benefits of this approach in terms of Vt reduction or improved respiratory system mechanics when hypercapnia is of no concern. To address this gap, the objective of our study was to ascertain whether EIP prolongation decreases PaCO 2 and whether this effect can be used to decrease Vt while keeping PaCO 2 constant. We hypothesized that this approach may have beneficial effects on respiratory system mechanics in ARDS patients.

Methods
The study was performed in the Intensive Care Unit at Hospital de la Santa Creu i Sant Pau, Barcelona (Spain). The institutional ethics committee approved the study (Reference: 10/089), and the patients' relatives gave signed informed consent.

Patients
Fourteen patients who met the criteria for ARDS [19] were included in the study. Exclusion criteria were: age <18 years, pregnancy, hemodynamic or respiratory instability, and variation of more than 0.5 °C in body temperature in the last 12 h before the study was planned [20]. One patient was excluded during the study period (see Results).
All patients were under sedation and analgesia with intravenous perfusion of midazolam and opiates. Neuromuscular blockade was used in all patients to prevent triggering of the ventilator. Careful endotracheal suctioning was performed before the protocol was started. Heated humidifiers (Fisher & Paykel; MR 290 chamber and MR 850 ALU electric heater; Panmure, New Zealand) were used for airway humidification in all patients. These humidifiers were placed in the inspiratory limb of the circuit in accordance with the manufacturer's recommendations. The respiratory rate, FiO 2 , inspiratory flow (square pattern) and PEEP were kept constant throughout the study.

Protocol
All patients were in steady state in the 60-min preceding data recording, and all of them were in a semirecumbent position. The study was performed in three consecutive 30-min phases. Measurements in the first phase (baseline phase) were taken under the mechanical ventilation parameters set by the patient's attending physician. In the second phase (EIP prolongation phase), the EIP was prolonged until one of the following parameters was reached: (1) EIP of 0.7 s; (2) intrinsic positive end-expiratory pressure (PEEPi) ≥1 cmH 2 O; or (3) inspiratory-expiratory ratio (I/E) of 1:1. We chose the EIP prolongation time (0.7 s) based on findings from a previous study by Devaquet et al. [18] in which a 20 % prolongation of the inspiratory time induced a significant decrease in PaCO 2 and dead space. In the third phase (Vt reduction phase), the Vt was diminished in steps of 30 mL every 30 min until PaCO 2 reached baseline levels.
The following data were collected at inclusion: demographic variables (age, sex, height), simplified acute physiology score II, ARDS etiology and days of mechanical ventilation.
During the last minute of each phase, we collected the following respiratory variables: peak airway pressure, Pplat, mean airway pressure, PEEPi, PEEP, driving airway pressure (∆Paw), Vt, dead space-to-Vt ratio (Vd/Vt), static compliance of the respiratory system (Crs) and airway resistance. At the same time, we recorded the following gas exchange variables: pH, partial pressure of arterial oxygen (PaO 2 ), PaCO 2 and end-tidal carbon dioxide concentration in the mixed expired gas (EtCO 2 ). PEEPi was measured with a prolonged end-expiratory pause of 4 s, performed using the ventilator expiratory hold button. EtCO 2 was measured continuously with a CO 2 mainstream sensor (General Electric Capnostat, Milwaukee, WI, USA). The mean value of the last 10 recorded EtCO 2 values in each phase of the study was used for analysis.
Ventilatory settings and airway pressures were recorded directly from the ventilator monitoring system. Plateau pressure was measured during an endinspiratory pause. Dead space was calculated using the Enghoff modification of the Bohr equation [21]: Vd/ Vt = (PaCO 2 − PeCO 2 )/PaCO 2 , being PeCO 2 the partial pressure of carbon dioxide in mixed expired gas. Expired gas was measured by collecting gas for 3 min with a Douglas bag (P-341-60; Warren E. Collins Inc., Boston, MA, USA) attached directly to the expiratory port of the ventilator. An automated analyzer (ABL 520; Radiometer A/S, Copenhagen, Denmark) was used to measure expired and arterial gases. Dead space data were expressed as physiological dead space (Vd phys in mL), defined as the sum of instrumental, anatomic and alveolar dead space [22]. Driving pressure (cmH 2 O) was calculated as Pplat-PEEP. Crs (mL/cmH 2 O) was calculated as Vt/[Pplat-(PEEP + PEEPi)], and airway resistance (cmH 2 O/L/s) was calculated as (peak airway pressure − plateau pressure)/Flow. Predicted body weight (PBW) was calculated as follows: 50 + 0.91(height in cm-152.4) for men and 45.5 + 0.91(height in cm-152.4) for women [8]. Arterial to end-tidal CO 2 gradient (P(a-et) CO 2 ) was calculated in each study phase. We used Puritan Bennett ™ 840 (Covidien, Galway, Ireland) and Dräger Evita XL (Dräger Medical, Lübeck, Germany) ventilators. All the ventilators used have a compressible volume compensation system.

Statistical analysis
Data are expressed as mean ± standard deviation. The results were analyzed using one-way analysis of variance for repeated measures (ANOVA) with the Greenhouse-Geisser correction. We performed the Kolmogorov-Smirnov test to confirm normal data distribution. Since the distribution of the data was normal, we used the Student's t test and the Pearson linear correlation to compare data and correlations between phases and variables, respectively. A two-tailed p value less than 0.05 was considered statistically significant. The SPSS ® Statistics (version 20.0, Chicago, IL, USA) statistical software was used for statistical analysis.

Results
One of the 14 patients enrolled in the study was excluded from the analysis due to fever, tachypnea and unstable EtCO 2 during the second phase of the study. The study was performed 5 ± 4 days after starting mechanical ventilation. Table 1 shows demographic data at admission, ARDS etiology and baseline characteristics at study day.
Between the first and second phase, significant decreases were observed in both the Vd/Vt ratio (0.70 ± 0.07 to 0.64 ± 0.08; p < 0.001) and EtCO 2 (41 ± 6 to 39 ± 6 mmHg; p = 0.006). Basal Vd phys and P(a-et) CO 2 had a close correlation (r = 0.75; p = 0.003). The change in Vd phys and the change in P(a-et)CO 2 between the first and second phase also showed a close correlation (r = 0.68; p = 0.001).
In the third phase (EIP prolongation and Vt reduction), the Vt was significantly reduced as compared to previous phases (6.3 ± 0.8 to 5.6 ± 0.8 mL/Kg PBW; p < 0.001).
The Vd phys decreased progressively and significantly during all phases of the study (267 ± 71 to 244 ± 65 to 216 ± 58 mL; p < 0.001). The Vd phys and Vt at baseline were strongly correlated (r = 0.946; p < 0.001). Additionally, the Vt reduction was tightly correlated with the decrease in Vd phys (r = 0.894; p < 0.001). Respiratory system mechanics, gas exchange, hemodynamics, and temperature data throughout the study are also given in Table 2.

Discussion
The main finding of our study was that the end-inspiratory pause prolongation allowed to decrease tidal volume while maintaining similar PaCO 2 levels. Indeed, the decrease in tidal volume led to a significant decrease in Pplat and ∆Paw, and it also improved the respiratory system compliance.
Several studies have shown that prolongation of EIP enhances CO 2 elimination and decreases dead space and PaCO 2 levels [14][15][16][17][18]. Diffusion of CO 2 is time dependent, and EIP prolongation increases the time available for alveolar gas exchange [14,23,24]. Devaquet et al. [18] extended inspiratory time from 0.7 ± 0.2 to 1.4 ± 0.3 s by increasing the inspiratory pause time from 0 to 20 % of the total breathing cycle. They observed that this modification significantly decreased both Vd/Vt (around 10 %) and PaCO 2 (around 11 %). Despite these beneficial effects of prolonged EIP and the direct relationship between inspiratory time and enhanced CO 2 elimination [16,18], EIP prolongation may lead to potentially adverse effects such as PEEPi production and inversion of the I/E ratio together with increases in mean airway pressure. These effects might also provoke hyperinflation, thus altering cardiac performance [25,26]. Nevertheless, Devaquet and colleagues [18] showed that EIP could be prolonged without significantly increasing PEEPi (I/E ratio 1:1.5). Not surprisingly, and in spite of a significant increase in EIP, we did not induce any significant increase in PEEPi since the expiratory time was long enough to avoid air trapping at the end of a passive expiration (average expiratory time 1.7 ± 0.3 s). Actually (see Table 2), the average product of three time constants (the time needed to    [14][15][16][17][18]. When comparing phase 1 (baseline) and phase 2 (isolated EIP prolongation), we found that the decrease in the Vd/Vt correlated well with the drop in PaCO 2 (r = 0.810; p < 0.001). These changes observed in our patients may be explained by the increase on the time available for distribution and diffusion of inspired tidal gas within resident alveolar gas during EIP prolongation [14]. Indeed, total PEEP levels, airflow, respiratory rate, tidal volume and respiratory mechanics were totally unchanged in this phase of our study [14,27,28].
Comparing the second (isolated EIP prolongation) and third (EIP prolongation and Vt reduction) phases, our data showed that the Vd/Vt ratio remained unchanged. However, the Vd phys , expressed in mL, decreased significantly between phases 2 and 3. This is explained by the significant reduction in Vt (that also provoked a decrease in Vd phys ) during the third phase as compared to the previous phases, and thus Vd/Vt ratio did not change. The fact that the reduction in Vt in the third phase was accompanied by a significant decrease in Vd phys and ΔPaw (with a significant increase in compliance) suggests that some degree of overdistension might be present at baseline.
As previously described, low tidal volume ventilation in ARDS may induce hypercapnia and, secondarily, induce pulmonary artery hypertension that may impair right ventricular function [29] and eventually cause acute cor pulmonale [30]. To reduce hypercapnia in ARDS ventilated patients, active heated humidifiers are often used. These devices significantly decrease dead space, PaCO 2 and ventilator mechanical load [9] without increasing airflow resistance [31]. Although active humidification is recommended over heat and moisture exchangers in ARDS patients [32], two studies focussing on the effects of EIP prolongation on gas exchange [16,17] did not describe the type of humidification used in their patients. A third study used passive or active humidification (10 and 5 patients, respectively) [18]. However, the effects on PaCO 2 in all these studies [16][17][18] were consistently the same, thus suggesting that humidification type per se does not influence the effects of EIP on PaCO 2 .
Another technique used to decrease hypercapnia is to increase the respiratory rate. However, in ARDS patients, several studies have shown that a high respiratory rate led to gas trapping and induced PEEPi [33,34]. In addition, experimental models suggested that higher respiratory rates may contribute to the development of ventilatorinduced lung injury [35,36]. Vieillard-Baron et al. [25] compared two respiratory rate strategies, 30 versus 15 breaths/min. They found that the high respiratory rate did not reduce PaCO 2 levels but produced dynamic hyperinflation and reduced the cardiac index. In our patients, EIP prolongation was achieved with a relatively Fig. 1 Individual values for PaCO 2 during the study. The asterisk denotes statistically significant differences (p < 0.001) during prolongation of end-inspiratory pause. EIP end-inspiratory pause, PaCO 2 partial pressure of carbon dioxide in arterial blood, Vt tidal volume Fig. 2 Individual values for Vd phys during the study. The asterisks denote a significant, progressive decrease in Vd phys (p < 0.001) during prolongation of end-inspiratory pause (EIP) and during Vt reduction. EIP end-inspiratory pause, Vd phys physiological dead space, Vt tidal volume high inspiratory flow rate (1 L/s), thus avoiding inverse I/E ratio. This was a safe strategy to decrease PaCO 2 levels, while keeping respiratory rate constant (22 breaths/ min) and not generating PEEPi.
In our study, the reduction in Vt to maintain isocapnia was modest. Should major reductions in Vt were required, then the use of invasive extracorporeal carbon dioxide removal devices had to be considered in order to avoid acute hypercapnia [37].
Studies analyzing the EIP prolongation did not describe changes in PaO 2 [14,18], except one study by Mercat et al. [16]. This latter study found a slight, but not statistically significant, increase in PaO 2 levels during EIP prolongation. This finding was not confirmed in our study. We speculate that the length of time that patients are maintained with EIP prolongation and the mean airway pressure achieved during extended EIP may have contributed to this finding. Indeed, in Mercat's study [16], EIP prolongation was continued for 1 h with a mean airway pressure of 21 cmH 2 O and an I/E ratio 1.1. In contrast, in Devaquet's study [18] and in our own study, EIP prolongation was shorter (30 min in both), mean airway pressure was lower (15 and 17 cmH 2 O, respectively), and the I/E ratios achieved were 1:1.5 in Devaquet's study and 1:1.7 in ours.
The main novelty of our study is that prolonging EIP allowed to reduce Vt by 11 % (from 6.3 ± 0.8 to 5.6 ± 0.8 mL/kg of PBW; p < 0.001), maintaining PaCO 2 levels equal to baseline. These sequential ventilatory changes were accompanied by a reduction in Vd phys . Also, when PaCO 2 returned to baseline due to a reduction in Vt, we found a significant decrease in Pplat and an increase in Crs. In addition, these changes in ventilatory mechanics were accompanied by a significant decrease in ∆Paw. All those findings could be explained by a degree of baseline overinflation even though our initial Vt was low [38]. We further support our contention by the tight correlation between Vt and Vd phys at the onset of the study and the tight correlation between the decrease in Vt and Vd phys at the end of the study. Our patients were basally ventilated with parameters similar to those used in previous studies [16][17][18] in terms of Vt and PEEP, and Table 2 Respiratory system mechanics, gas exchange and hemodynamic data during the study Data are presented as number (%) or mean ± SD Intergroup differences (p < 0.05): a, phase 1 versus phase 2; b, phase 1 versus phase 3; c, phase 2 versus phase 3 Crs static compliance of the respiratory system, EIP end-inspiratory pause, EtCO 2 end-tidal carbon dioxide concentration in the expired air, FiO 2 fraction of inspired oxygen, HR heart rate, MAP mean arterial pressure, PaO 2 partial pressure of oxygen in arterial blood, PaCO 2 partial pressure of carbon dioxide in arterial blood, PBW predicted body weight, PEEPi intrinsic positive end-expiratory pressure, Pmean mean airway pressure, Ppeak peak airway pressure, Pplat plateau airway pressure, P(a-et)CO 2 arterial to end-tidal CO 2 gradient, Raw airway resistance, Vd phys physiological dead space, Vd/Vt dead space-to-Vt ratio, Vt tidal volume, ∆Paw driving airway pressure  [39]. In our study, a slight but not statistically significant decrease in mean arterial pressure was observed. Such trend could have been the result of complex interactions of PaCO 2 and mean airway pressure in cardiovascular system.
We think that EIP prolongation is a feasible maneuver to optimize the consequences of mechanical ventilation in ARDS patients. Physicians may consider using an EIP prolongation in the early phase of ARDS when patients often require sedation and neuromuscular blocking agents. In our study, we have effectively implemented this strategy by using active humidification, relatively high inspiratory flow rates and close monitoring of PEEPi. This bundle decreases PaCO 2 , which in turn will allow to further decrease Vt and the consequent lung strain when isocapnic conditions are met.
One of the limitations of our study is the relatively small number of patients, the majority with pneumonia, and the fact that the study is short term. Studies with patients with different ARDS etiologies and larger numbers are warranted to confirm our data. Also, we did not measure other parameters such as inflammatory mediators or lung volumes. The calculation of dead space using the Enghoff modification of Bohr equation in patients with large shunt fractions (>20-30 %) could underestimate dead space fraction [12]. In our study, we did not measure intrapulmonary shunt. However, according to the gas exchange values that we obtained, shunt fractions above 30 % are unlikely. Additionally, the EIP prolongation increases the mechanical inflation time and it could extend into neural expiration. Asynchronies may thus develop and cause an inadequate patient-ventilator interaction when the patients are not paralyzed [39][40][41]. Our results could be dependent on our routine management of mechanical ventilation in ARDS patients, but our findings have been consistent in all patients and we consider they could be extrapolated to other ARDS patients. Finally, the absolute decrease in tidal volume, although statistically significant, is moderate.

Conclusions
In conclusion, our data indicate that EIP prolongation is a simple and feasible strategy to decrease dead space and PaCO 2 levels. In addition, when PaCO 2 levels are of no clinical concern, EIP prolongation allows us to further decrease tidal volume. This, in turn, decreases plateau airway pressure, driving airway pressure and improves respiratory system compliance, suggesting less overdistension and less risk of dynamic strain and lung injury. Therefore, the use of this simple ventilator maneuver during mechanical ventilation in sedated and paralyzed ARDS patients merits consideration.
Abbreviations ARDS: acute respiratory distress syndrome; Crs: static compliance of the respiratory system; EIP: end-inspiratory pause; EtCO 2 : end-tidal carbon dioxide concentration in the mixed expired gas; IE: inspiratory-expiratory ratio; PaCO 2 : partial pressure of carbon dioxide in arterial blood; PaO 2 : partial pressure of arterial oxygen; PBW: predicted body weight; PeCO 2 : partial pressure of carbon dioxide in mixed expired gas; PEEP: positive end-expiratory pressure; PEEPi: intrinsic positive end-expiratory pressure; P plat : plateau airway pressure; P(a-et) CO 2 : arterial to end-tidal CO 2 gradient; Vd/Vt: dead space-to-Vt ratio; Vd phys : physiological dead space; Vt: tidal volume; ∆Paw: driving airway pressure.