Introduction

During acute respiratory distress syndrome (ARDS), functional residual capacity (FRC) is dramatically decreased [1, 2]. Improving oxygenation and limiting maximal alveolar pressure are current goals of mechanical ventilation (MV), which include positive end expiratory pressure (PEEP) titration [3], prone positioning [4] or, less frequently, recumbent positioning [5]. Computed tomographic lung studies have observed the presence of sterno-vertebral and cephalocaudal gradients in the distribution of densities [6–8], which provide a rationale for the positioning of ARDS patients. In contrast with prone position, very few adverse events have been described with semirecumbent position [9, 10], a position frequently recommended in the ICU [11]. Preliminary encouraging results of verticalization have been shown on oxygenation [5], but very few data are available. In this study [5], patients responding to vertical positioning (defined as a 40 % PaO2/FiO2 increase) depicted an increase in lung volume above FRC which well correlated with a better oxygenation. However, PaO2/FiO2 is known as a poor indicator of recruitment during ARDS [12]. Optimizing lung volume may help to decrease the strain induced by tidal inflation and participate to protective ventilation [13]. Thanks to dedicated intensive care beds, recumbent or semi-seated position can be easily obtained for ventilated patients. Recently, the multibreath washout/washin of nitrogen [14–16] has been implemented in ICU ventilators, allowing bedside end expiratory lung volume (EELV) measurements with acceptable accuracy and reproducibility [15, 17]. We then hypothesize that lung volume monitoring at the bedside could help to define which patient may benefit of verticalization.

The aims of our study were, firstly, to evaluate the concomitant effects of verticalization on lung volume and oxygenation following a change from supine to semirecumbent, seated, and back to supine position. Secondly, to evaluate the contribution of lung volume monitoring to spot patients who may decrease their strain while verticalized.

Patients and methods

This was a multicenter study performed in five French medical intensive care units: Henri Mondor University Hospital in Créteil, Hôpital Européen Georges Pompidou in Paris, Angers University Hospital, l’Archet 1 University Hospital in Nice and Charles Nicolle University Hospital in Rouen. According to the French legislation, the institutional review board of Henri Mondor hospital approved the protocol for all centers and waived the need for informed consent considering the optimization of oxygenation using body positioning as a standard of care; information was given to the subjects or their next of kin.

Patients

Patients were enrolled if they met the standard criteria for ARDS or previously acute lung injury [18, 19]: ratio of the partial pressure of arterial oxygen to the fraction of inspired oxygen (PaO2/FiO2) of less than 300 mmHg, presence of bilateral pulmonary infiltrates on the chest radiograph, and no clinical evidence of left atrial hypertension. The exclusion criteria were an age under 18, pregnancy, chronic obstructive pulmonary disease according to the patient’s medical history, history of pulmonary surgery, hemodynamic instability defined as increase of vasopressor (epinephrine, norepinephrine) in the last 6 h.

Patients with PaO2/FiO2 <100 with PEEP ≤5 cmH2O at inclusion (supine 1) were considered as having severe ARDS [18].

Ventilatory strategies

All patients received volume assist control ventilation using an ICU ventilator (Engström, General Electrics, Madison, Michigan, USA). This ventilator provides bedside EELV measurements using the nitrogen washout/washin technique. Oxygenation goal (SaO2 ≥90 %) was obtained by adjusting FiO2 and PEEP which were maintained during the study, and tidal volume was set at 6 mL/kg of predicted body weight (PBW). Patients were sedated; neuromuscular blocking agents were occasionally administered only if judged necessary by the clinician in charge.

Positioning

All patients leaned on the “TotalCare” beds [Hill-Rom (Batesville, IN, USA)]. During the first epoch, patients were supine positioned at 15°–20° for routine care, this was referred as “supine 1″. For the second epoch, trunk was elevated at 45°, and this was referred to as “semirecumbent position”. For the third epoch, trunk was elevated at 60° with legs down at 45°; this was referred to as “seated position”. Lastly, the patient was put back in the initial supine position referred to as “supine 2”.

Each position was maintained during 45–60 min; head of beds were equipped with an angle indicator to confirm the correct positioning.

Measurements

Lung volume: At the end of each 45-min period arterial blood gases were sampled, then three consecutive measures of EELV were performed using the nitrogen washout/washin technique. (See additional method ESM).

Strain: Lung strain was computed as the ratio between tidal volume and EELV [20].

Compliance: Static compliance (C stat) of the respiratory system was computed as tidal volume divided by the inspiratory plateau pressure, measured during an end-inspiratory pause, minus total PEEP. Total PEEP (PEEPtot) was measured using a 5-s expiratory pause.

Responders: PaO2/FiO2 increase >20 % between supine and seated position defined responders.

Statistical analysis

All values are given as median and (25th–75th interquartile range). Nonparametric tests were used because of the small size of the population. Friedman’s nonparametric test was used to compare the values of EELV/PBW, PaO2/FiO2 ratio and compliance in the different epochs. Differences at the level of p < 0.05 were considered as significant and a Wilcoxon matched pairs signed-rank test was performed to compare the different epochs to supine 1. To correct for multiple comparisons (Bonferroni correction), we considered a p value equal to or smaller than 0.01 to indicate significance. The Mann–Witney U-test was used to compare responders and non-responders, as well as mild ARDS versus moderate and severe [18]. A p value <0.05 was considered as significant.

Results

Forty patients were studied, including five with mild ARDS, 30 moderate and five severe [18]. All postures were well tolerated, allowing all patients to complete the study, but four patients not receiving vasopressor infusion received saline infusion when seated to maintain blood pressure. FiO2 was maintained stable along the study for all patients. Table 1 reports the main characteristics of the patients. Respiratory mechanics, lung volumes and blood gases during the four epochs are shown in Table E1.

Table 1 Main demographic and physiological characteristics of the patients studied
Full size table

Oxygenation

PaO2/FiO2 was significantly higher in seated than in supine position (p < 0.003), especially for moderate and severe ARDS (Fig. E1). PaO2/FiO2 increased by more than 20 % in 13 responders [39 % (30–63)]. In responders, PaO2 increased from 83 mmHg (77; 90) to 108 (105; 127) (p = 0.001) while it did not change in non-responders. Figure 1 represents the individual changes of PaO2/FiO2 between supine and seated positions against the lung volume changes. No correlation was found.

Fig. 1
figure 1

Representation of PaO2/FiO2 increase between supine and seated position against lung volume increase. Dotted line represents ΔPaO2/FiO2 = 20 %, the cutoff of for responders. Straight line represents regression. r 2 = 0.07, p = ns

Full size image

Lung volume and compliance

Values of EELV/PBW measured in the four successive positions for the whole population and for responders are presented in Fig. 2. When patients were verticalized, EELV expressed in mL/kg of PBW increased from 16 (13–22) to 19 mL/kg (15–25) (p < 0.05). Only responders had a significant EELV/PBW increase compared to baseline. Responders had significantly lower EELV/PBW values than non-responders during all epochs except seated.

Fig. 2
figure 2

EELV/PBW measured using the nitrogen washin/washout technique at the four epochs studied. Panel A represents the whole population. Panel B: blue boxes represents non-responders and red boxes responders. $ p < 0.05 responders versus non-responders. Friedman’s non-parametric test showed p < 0.05 for all patients and responders allowing Wilcoxon paired test. *p < 0.005 versus supine 1 (Bonferroni correction)

Full size image

Strain was significantly higher supine 0.38 (0.27; 0.45) versus 0.31 [0.24; 0.40] (p = 0.0006) and decreased significantly only for responders 0.42 (0.40; 0.47) versus 0.38 (0.30; 0.43) (p = 0.007).

Compliance decreased with verticalization in nonresponders [42 mL/cmH2O (30–46) vs. 32 mL/cmH2O (24–41) supine and seated, respectively (p < 0.002)], while it was lower at baseline but remained stable in responders: 30 mL/cmH2O (22–38) versus 29 mL/cmH2O (22–32) (Fig. 3).

Fig. 3
figure 3

Static compliance calculated at the four epochs studied. Panel A represents the whole population. Panel B: blue boxes represents non-responders and red boxes responders. $ p < 0.05 responders vs non-responders. Friedman’s nonparametric test showed p < 0.05 for all patients and non-responders allowing Wilcoxon paired test: * p < 0.001 versus supine 1 (Bonferroni correction). * p < 0.002 vs supine 1 (Bonferroni correction)

Full size image

Discussion

In this series of ARDS patients ventilated supine, verticalization (seated position) resulted in a rapid and significant concomitant increase in lung volume and oxygenation. Interestingly, patients responding to verticalization exhibited lower EELV/PBW at baseline and higher increase in EELV/PBW compare to non-responders.

Seated position

We used the seated position provided by dedicated intensive care beds. One previous study used a comparable seated position using conventional beds [5]. Here, the position was simply achieved by repositioning the bed electronically and was then controlled with the angle measurer included at the bed’s head [21–23]. A Position was thus easily obtained and maintained. Benedik et al. [24] described in spontaneously breathing obese patients that the change in EELV was only observed in a seated position but not in semirecumbent. In the present study, both semirecumbent and seated position resulted in higher EELV. This discrepancy might be due to the Fowler position (45° vs. 30°) they studied [24] which can generate an extra abdominal pressure potentially participating to respiratory system compliance decrease. The semirecumbent position (45°) can generate an extra abdominal pressure close to 9 mmHg, which may contribute to a decrease in respiratory system compliance [25]. The seated position with legs down we used has not yet been precisely tested in terms of abdominal pressure.

Oxygenation and lung volume

Effect of verticalization on PaO2/FiO2 ratio for responders was quick and started from the semi recumbent position epoch. Completion of this study on short periods, allowed to point out only fast responders. Among the mechanisms that can explain oxygenation increase, recruitment of previously nonaerated alveoli is supported by the stability of static compliance while the EELV/PBW increased. The compliance decrease of nonresponders has already been described in surgical population [26]. Richard et al. showed in a medical ARDS population of 16 patients comparable to ours, a bimodal distribution of oxygenation and EELV increase. In their study, responders had a PaO2/FiO2 increase above 40 % associated with a PEEP-volume (measured with prolonged exhalation to elastic equilibrium) increase associated with chord-compliance decrease. Past studies have yielded controversial results on the effect of position on oxygenation after surgical procedures [26]. Recent meta-analysis on prone position [27] suggested a mortality reduction mostly for the most hypoxemic patients (PaO2/FiO2 below 140 mmHg) at the cost of increased complications (pressure ulcer, endotracheal tube dislodgment or obstruction). Already recommended in the bundles for prevention of ventilator-associated pneumonia [11], the semirecumbent position can be routinely achieved during ventilation and could be proposed as an alternative to prone position. Gas exchange are not sufficiently precise to assess recruitment [12] or lung volume increase due to positioning (Fig. 1). Lung volume is a component of ventilatory-induced strain (Vt/EELV) [20, 28, 29], if its implication in ventilatory-induced lung injury is relevant in line with recent biological [20], human [28] and animal findings [29], we believe its bedside measurements might be helpful especially to discriminate which patient may benefit of positioning. The strain decrease found (specially for responders) could be an interesting tool, sharper than oxygenation, to indicate positioning but further studies are needed to test this hypothesis.

This study has limitations. Firstly, the sequence of positions has not been randomized. This prevented us to distinguish precisely the specific effect of the seated position (with legs down and higher trunk verticalization angle) from a time dependent effect of the verticalization [5]. EELV/PBW of responders, however, already increased in the semirecumbent epoch suggesting an immediate effect while it remained stable for non-responders. Although, a delayed effect on oxygenation and lung volume can not be eliminated [30]. Secondly, some variability of the nitrogen technique may exist with PEEP, although it has been shown to be acceptable [15]. Last, we conventionally chose PaO2/FiO2 increase to define responders albeit oxygenation variations are multifactorial especially during ARDS and mechanical ventilation [31]. Nonetheless FiO2 was stable all along the study. Physiological data, including abdominal pressure, esophageal pressure, cardiac output for shunt estimation in future studies would certainly improve our understanding of PaO2/FiO2 improvement during verticalization.

Patient positioning, especially the seated position, is a relatively simple way to improve oxygenation, notably for moderate and severe ARDS. The tools used in our study (specific beds and nitrogen washin/washout) are now routinely available outside of the research field, offering the prospect of a simple and rational use in ARDS.