Introduction

Respiratory mechanics is a key element in clinical practice to monitor mechanically ventilated patients and guide ventilator settings [1]. Respiratory system compliance (CRS) has been shown to correlate with the amount of aerated lung [2]. In addition, an increased respiratory system driving pressure (DPRS) has been shown to be associated with an increased risk of mortality in patients with ARDS [3, 4]. Beside this “basic” respiratory mechanics assessment, some more complex explorations, including in particular the evaluation of chest wall mechanics and the detection of complete airway closure, may allow to better characterize respiratory mechanics and personalize ventilator settings [5,6,7]. Instead of considering the respiratory system as a whole, partitioning it into the lung and the chest wall using esophageal pressure measurements has thus been proposed to estimate transpulmonary pressures and determine the amount of applied airway pressure, which is spent to inflate the lung and the one spent to displace the chest wall [8,9,10]. The airway closure, a phenomenon recently highlighted in ARDS patients, may also impact respiratory mechanics assessment and ventilatory management [5]. Complete airway closure leads to an absence of communication between proximal airways and alveoli when airway pressure is below the level of the so-called airway opening pressure (AOP). Practically, in case of complete airway closure, no gas enters the lung during insufflation until AOP has been overcome. Complete airway closure may thus lead to driving pressure overestimations and compliance underestimations when AOP is not considered in their calculations. In addition, a PEEP setting below the AOP level may promote inflammation due to cyclic airway closure and favor atelectasis [11]. Its detection at the bedside requires performing a low flow pressure–volume or pressure–time curve [5]. Some data suggest that this “advanced” respiratory mechanics assessment may be particularly relevant in obese patients [12,13,14]. Obesity is a major public health issue with a prevalence around 35% in adults in the United States of America and 13% worldwide and impacts critically ill patients’ management [15,16,17]. Most of the literature describing respiratory mechanics in obese patients comes however from the postsurgical setting and data concerning respiratory mechanics and in particular chest wall mechanics of critically ill obese patients remain scarce. Furthermore, the impact of obesity on respiratory mechanics has not been specifically assessed in patients fulfilling or not ARDS criteria. The main aim of this study was to comprehensively assess respiratory mechanics in obese and non-obese patients with or without ARDS and to study the additional value of an advanced respiratory mechanics evaluation (including esophageal pressure measurements and complete airway closure detection) in these patients compared to a basic evaluation (based on airway pressure monitoring). For this purpose, we prospectively measured gas exchange and respiratory mechanics in standardized conditions with the same positive end-expiratory pressure (PEEP) level and the same normalized tidal volume (Vt) in all intubated patients in two ICUs.

Patients and methods

Patients’ selection

All patients admitted from March 2018 to January 2020 in two academic hospital ICUs (Angers and Strasbourg, France) intubated and mechanically ventilated for any cause were prospectively included in the study within 24 h after intubation. Exclusion criteria were age < 18 years, pneumothorax, contraindication to esophageal pressure measurement, and use of extracorporeal membrane oxygenation (ECMO) at the time of inclusion. Patients admitted after a cardiac arrest were excluded from the analysis and reported in another publication [18]. Airway pressure and flow recordings of these patients were used to describe a novel approach to assess AOP [19].

Study protocol

Settings

All patients received deep sedation and neuromuscular blockers at the time of measurements and were ventilated using an Engström® or R860® ventilator (GE Healthcare, Madison, WI, USA) in supine semi-recumbent position (head of the bed elevated at 30°). Respiratory mechanics and gas exchange were assessed under standardized conditions (volume-controlled ventilation with a Vt of 6 mL kg−1 of predicted body weight (PBW) and constant inspiratory flow of 60 L min−1 and PEEP of 5 cm H2O). Respiratory rate was adjusted by the attending physician (up to 35/min), and the fraction of inspired oxygen (FiO2) was set for pulsed oxygen saturation between 92 and 98%.

Esophageal pressure was measured with a specific nasogastric feeding tube equipped with an esophageal balloon (Nutrivent® catheter, Sidam, San Giacomo Roncole, Italy) and connected to the auxiliary pressure transducer of the ventilator [9, 20]. The balloon was consecutively inflated to target a filling volume of 2, 3, and 4 mL. For the measurements, the filling volume was set as the lowest volume between 2 and 4 mL associated with the largest tidal swing of esophageal pressure during the insufflation of the Vt [21]. In addition, to avoid balloon overfilling, the balloon filling was stopped if a sudden and significant increase of the baseline esophageal pressure was observed. The correct position of the esophageal balloon was then checked by chest X-rays and an occlusion test [9, 20]. The ratio of esophageal pressure swing over airway pressure swing (ΔPes/ΔPaw) during the occlusion test was considered as acceptable if it was between 0.8 and 1.2.

To normalize volume history, a recruitment maneuver was performed in volume-controlled ventilation in absence of hemodynamic instability by increasing PEEP level up to 20 cmH2O for 1 min (maximum plateau pressure (PPlat) of 40 cmH2O). PEEP level was then switched back to 5 cmH2O.

Flow, airway pressure, and esophageal pressure–time curves were recorded using a dedicated computer connected to the ventilator with a 40 ms sampling time for offline analysis.

Measurements

All esophageal pressure signal recordings were independently inspected by two investigators blind to the other clinical data, and those considered non-valid were excluded from the analyses including esophageal pressure data.

Inspiratory and expiratory occlusion maneuvers were performed to measure PPlat, inspiratory esophageal pressure (Peso inspi), total PEEP, and expiratory esophageal pressure (Peso expi).

Abdominal pressure (Pabdo) was measured using an intravesical catheter.

An arterial blood gas was performed after 15 min free of any occlusion maneuver, and EELV was measured at PEEP 5 cmH2O using the nitrogen washout-washin technique (E-COVX module sensor®, GE Healthcare) [22].

A low-flow inflation (5 L min−1, Vt = 8 mL kg−1 PBW) was then performed after a prolonged exhalation to PEEP 5 cmH2O.

Complete airway closure and corresponding AOP were identified by the inspection of the pressure–volume curves as previously described [5, 23].

Calculated variables

DPRS was computed as the difference between PPlat and total PEEP. CRS was computed as the expired tidal volume (Vte) divided by DPRS. The respiratory system elastance (ERS) was equal to 1/CRS. The respiratory system resistance was computed as the difference between peak airway pressure and PPlat divided by the inspiratory flow.

The difference between P1 and PPlat (ΔP1–PPlat) with P1 defined as airway pressure at first zero flow was computed to assess viscoelastic properties of the lung and chest wall tissues and pendelluft phenomenon [24].

Inspiratory (PLinspi) and expiratory transpulmonary pressures (PLexpi) were computed as the difference between PPlat and Peso inspi, and between total PEEP and Peso expi, respectively.

The lung driving pressure (DPL) was computed as the difference between PLinspi and PLexpi [25]. The lung compliance (CL) was computed as Vte divided by DPL. The lung elastance (EL) was equal to 1/CL. The chest wall compliance (CCW) was computed as Vte divided by the difference between Peso inspi and Peso expi.

The elastance ratio (EL/ERS) was calculated to assess the respective effects of the airway pressure on the lung and the chest wall [8, 10]. Plateau pressure of the lung (PPlat Lung) was computed as PPlat multiplied by EL/ERS [26].

As PPlat Lung and PLexpi can be considered respectively as good surrogates of the inspiratory transpulmonary pressure in the nondependent lung and the expiratory transpulmonary pressure in the dependent lung [27], we computed the lung stress as the difference between PPlat lung and PLexpi, to assess the real stress applied to the lung across the ventral to dorsal axis.

DPRS-AOP and CRS-AOP were computed as DPRS and CRS using AOP instead of total PEEP in the calculations in patients with an AOP higher than total PEEP.

The ratios CRS/PBW, CL/PBW, and EELV/PBW were calculated to normalize CRS, CL, and EELV to the gender and the height of the patient. PBW was calculated using the previously published formula [28].

Dead space was assessed using ventilatory ratio, which was computed as minute ventilation (mL/min) × PaCO2 (mmHg)] / (PBW (kg) × 100 × 37.5) [29, 30].

Other collected data

Age, height, weight, past medical history of chronic respiratory disease, immunodepression, Sequential Organ Failure Assessment (SOFA) score [31], and Simplified Acute Physiologic Score II (SAPS II) [32] were collected on the day of admission.

The lung opacities were independently assessed on chest X-rays by two experienced investigators blind to clinical data.

The diagnosis of ARDS was performed using the criteria of the Berlin definition by an adjudication committee blind to respiratory mechanics data [33].

Survival was assessed at day 60 after inclusion.

The number of ventilator-free days at day 28 was defined as the number of days between day 1 and day 28 on which patients breathed without assistance. A value of 0 ventilator-free day was assigned for patients who died before day 28.

Statistical analysis

Results are presented as median [interquartile range] and number (percentage). Normality of the variables was assessed using the D’Agostino & Pearson test. The study population was divided into four groups according to the presence of obesity or not (BMI < or ≥ 30 kg m−2) and the presence of ARDS or not at the time of respiratory mechanics assessment [33]. All the patients with ARDS were also compared to those without ARDS, and all obese patients were compared to non-obese patients. In addition, the population was divided into three groups according to whether EL/ERS ratio was lower than the first quartile of EL/ERS of the study population (Low EL/ERS), higher than the third quartile (high EL/ERS), or in-between (medium EL/ERS). We planned to enroll more than 140 patients to be able to detect 10% absolute changes between the defined patients’ groups in the main physiological variables assessed in the study (with 80% power at a two-sided type I error of 0.05). The groups were compared using Kruskal–Wallis test, Mann–Whitney U-test, t-test, Fischer or Chi2 test, as appropriate according to data distribution. Bonferroni correction was applied for multiple pairwise comparisons. Correlations were analyzed using Spearman test. All tests were performed with a type I error set at 0.05. The statistical analysis was performed using R version 3.6.2 (R Core Team 2019, Vienna, Austria, https://www.R-project.org/).

Results

Main patients’ characteristics

One hundred and sixty-four patients were included in the study. One hundred and forty-nine of them were included in the analysis (15 patients were excluded because of lack of data due to technical issues in recordings or major deviations in study protocol). Valid esophageal pressure measurements were analyzed in 124 patients. Respiratory mechanics and gas exchange were assessed 10 [3.5–22] hours after intubation.

Fifty-two patients (34.9%) were obese. Ninety (60.4%) patients fulfilled ARDS criteria (65.4% of obese and 57.8% of non-obese patients, p = 0.385). The main patient’s characteristics categorized according to the presence or not of obesity and/or ARDS are presented in Table 1 and Additional file 1: Table S1.

Table 1 Main characteristics of the patients
Full size table

Gas exchange

Gas exchange in the patients categorized according to the presence or not of obesity and/or ARDS is presented in Table 2 and Additional file 1: Table S2. PaO2/FiO2 ratio was not different between obese non-ARDS, obese ARDS, and non-obese ARDS patients but was lower in non-obese non-ARDS patients than in the other groups of patients. It tended to be lower in obese than in non-obese patients and was lower in ARDS than in non-ARDS patients. Ventilatory ratio was not different between obese non-ARDS, obese ARDS, and non-obese ARDS patients but was lower in non-obese non-ARDS patients than in non-obese ARDS and obese ARDS patients. It was higher in obese than in non-obese patients and in ARDS than in non-ARDS patients. PaO2/FiO2 ratio and ventilatory ratio correlated with BMI in non-ARDS patients but not in ARDS patients (Additional file 1: Fig. S1A and B).

Table 2 Gas exchange and respiratory mechanics
Full size table

Airway closure and driving pressure

A complete airway closure assessed with a PEEP of 5 cmH2O was found in 23.5% of the patients. It was found in some patients of the four groups but was more frequent in obese ARDS patients than in non-obese non-ARDS and non-obese ARDS patients (Table 2). It was more frequent in obese than in non-obese patients (40.4% vs. 14.4%, p < 0.001) and in ARDS than in non-ARDS patients (30% vs. 13.6%, p = 0.029) (Additional file 1: Table S2).

DPRS and DPRS-AOP in the patients categorized according to the presence or not of obesity and/or ARDS are presented in Table 2 and Additional file 1: Table S2. Considering the whole population, DPRS-AOP was different from DPRS in 15 (10.1%) patients. In these 15 patients, the difference between DPRS and DPRS-AOP was 1.5 [1–3] cmH2O.

ΔP1-PPlat was not different between obese and non-obese patients but was higher in patients with ARDS than in those who did not meet ARDS criteria (Table 2 and Additional file 1: Table S2).

Lung volumes and compliances

CRS/PBW, CL/PBW, and EELV/PBW were not different between obese non-ARDS patients and obese or non-obese ARDS patients but were higher in non-obese non-ARDS patients than in the other groups of patients (Fig. 1). Those parameters were lower in obese than in non-obese patients and in ARDS than in non-ARDS patients (Additional file 1: Fig. S2). Similar results were found when considering the AOP in the calculation of respiratory system compliance (CRS-AOP/PBW, Additional file 1: Fig. S3). The correlations between CRS/PBW and BMI and between EELV/PBW and BMI in ARDS and non-ARDS patients are presented in additional data (Additional file 1: Fig. S4A and B).

Fig. 1
figure 1

Distribution of respiratory system compliance (CRS/PBW, A), lung compliance (CL/PBW, B), and end-expiratory lung volume (EELV/PBW, C) normalized to predicted body weight in patients categorized according to the presence or not of obesity and acute respiratory distress syndrome (ARDS). Boxplots display medians, 10th, 25th, 75th, and 90th percentiles. p-values represent the overall comparisons between the four groups of patients. *, p < 0.05; **, p < 0.01; ***, p < 0.001 (pairwise comparisons with Bonferroni correction)

Full size image

CRS correlated well with EELV in obese and non-obese patients with or without ARDS (Additional file 1: Fig. S5).

Chest wall mechanics

CCW was not different between the four groups of patients and between obese and non-obese patients or between ARDS and non-ARDS patients but Peso expi was higher in obese patients with or without ARDS than in non-obese non-ARDS patients (Fig. 2 and Additional file 1: Fig. S6). Peso expi was higher in obese patients than in non-obese patients but was not significantly different between ARDS and non-ARDS patients. BMI was not correlated with CCW but was correlated with Peso expi in patients with or without ARDS (Additional file 1: Fig. S7A and B). Esophageal pressure–volume and esophageal pressure–time curves during low-flow insufflation of obese and non-obese patients are presented in Fig. 3.

Fig. 2
figure 2

Distribution of chest wall compliance (CCW, A) and end-expiratory esophageal pressure (Peso expi, B) in patients categorized according to the presence or not of obesity and acute respiratory distress syndrome (ARDS). Boxplots display medians, 10th, 25th, 75th, and 90th percentiles. p-values represent the overall comparisons between the four groups of patients. *, p < 0.05 (pairwise comparisons with Bonferroni correction)

Full size image
Fig. 3
figure 3

Esophageal pressure–volume (A) and esophageal pressure–time (B) curves during low-flow insufflation (5L min−1) in obese and non-obese patients. A Black and gray lines represent median values of all obese and non-obese patients included in the study, respectively. B Black and gray lines represent median values and interquartile range of all obese and non-obese patients included in the study, respectively

Full size image

Peso expi and CCW were not correlated with Pabdo (p = 0.331, rho = 0.181 and p = 0.183, rho = − 0.242, respectively; n = 32). Peso expi was correlated with EELV/PBW (p < 0.001, rho = − 0.365) and CRS/PBW (p = 0.010, rho = − 0.231) but not with AOP (p = 0.306, rho = 0.193).

The EL/ERS ratio in the whole population was 0.64 [0.57–0.72]. It was not different between the four groups of patients and between those with and without obesity or between those with and without ARDS (Additional file 1: Fig. S8). The main characteristics of the patients with low EL/ERS (i.e., in whom chest wall compliance markedly impacts respiratory system compliance) were not different from those with medium or high EL/ERS (Additional file 1: Table S3). But patients with low EL/ERS had higher CRS, CL, and EELV/PBW and were less hypoxemic than patients with higher EL/ERS (Additional file 1: Table S4).

Lung driving pressure, plateau pressure of the lung and lung stress

DPL, PPlat lung, and lung stress were not different between obese non-ARDS patients and obese or non-obese ARDS patients but were lower in non-obese non-ARDS patients than in the other groups of patients (Table 2 and Additional file 1: Table S2).

Discussion

The main results of the study can be summarized as follows:

  1. (1)

    Oxygenation, CRS, CL, and EELV were similarly altered in obese patients without ARDS and patients with ARDS (either obese or non-obese).

  2. (2)

    Peso expi was higher in obese patients than in non-obese patients but CCW did not differ between these groups of patients. Chest wall contribution to CRS expressed by the EL/ERS ratio was widely distributed and was not predictable by patient’s general characteristics.

  3. (3)

    Complete airway closure was observed in all groups of patients but was more frequently found in obese than in non-obese patients and in ARDS than in non-ARDS patients. Ignoring airway closure led to an overestimation of DPRS in almost 17% of obese patients.

In the present series, gas exchange, CRS, CL, and EELV were similarly affected by obesity and ARDS in comparison with non-obese non-ARDS patients. These original findings can be explained by the reduction in lung volumes reported in both obese and ARDS patients. Chest wall mechanics differ however between obese and ARDS patients with higher Peso expi in obese patients despite similar CCW. This increased Peso expi may be related to the decreased EELV and the increased frequency of complete airway closure in these patients.

The impairments in lung volumes and chest wall mechanics that we measured in critically ill obese ARDS and non-ARDS patients are consistent with the observations previously reported by Coudroy et al. in a post hoc pooled analysis of two small cohorts of patients with ARDS [13]. In this work, Peso expi, but not CCW, was shown to correlate with BMI. Airway closure was also found to be more frequently observed in patients with higher BMI [13]. Based on CT scan analyses, Chiumello et al. reported lower lung gas volume and higher total superimposed pressure in obese ARDS compared to non-obese ARDS patients [34]. Noticeably, in this series, CCW was similar in obese and non-obese ARDS patients, which is consistent with our observations but PL expi did not differ.

In addition, our observations in critically ill patients are consistent with those reported in obese surgical patients [35,36,37]. Pelosi et al. found however a lower CCW in morbidly obese patients [36]. This discrepancy with our results may be related to the higher BMI, and the strict supine position in which measurements were performed in Pelosi et al. study [36].

Our study is the first to systematically assess, soon after intubation, and according to a well-standardized protocol, the complete respiratory mechanics in a large series of non-selected patients including ARDS and non-ARDS patients. This methodological strength is of particular relevance to appreciate properly the roles played by obesity and ARDS since respiratory mechanics have been shown to significantly change over time under mechanical ventilation due to several confounding factors as the progressive increase in lung weight.

Our findings have important clinical implications especially since obesity is frequent in ICU patients [15]. The respiratory mechanics features observed in both obese and ARDS patients suggest that lung protective ventilation strategy could overall be similarly managed in these patients but no interventional study has so far specifically evaluated the potential benefit of such a strategy in obese non-ARDS patients [14]. In addition, our data suggest that advanced explorations may be of particular value to better assess respiratory mechanics and individualize ventilator settings. Interestingly, whereas basic respiratory mechanics assessments showed similar alterations in obese non-ARDS and non-obese ARDS patients, advanced explorations revealed that mechanisms involved were different in these two groups of patients. Differences in Peso expi may thus lead to different PEEP settings when a positive PL expi is considered as a goal to optimize ventilation [38, 39]. Moreover, our study shows that the EL/ERS ratio may significantly differ between patients and cannot be easily predicted by the main patients’ characteristics. Esophageal pressure monitoring is thus needed to assess the contribution of CCW to CRS. Furthermore, an assessment of airway closure may be systematically considered as this phenomenon impacts driving pressure measurements in around 10% of the patients (and even 17% of obese patients). Such alterations in obese patients respiratory mechanics may contribute to explain the absence of association between DPRS and mortality observed in obese ARDS patients contrary to what was observed in non-obese ARDS patients [40]. In addition, a PEEP level set below the AOP may be associated with a higher risk of ventilator induced lung injury because of the heterogeneity of tidal ventilation distribution and atelectrauma [11].

Our study has several limitations. First, gas exchange and respiratory mechanics were assessed at only one PEEP level and lung recruitability was not directly evaluated. Higher PEEP could have been associated with different observations, but our study design allowed to assess all the patients in similar standardized and safe conditions. Second, we did not deduct the estimated pressure generated by the esophagus wall from the directly measured esophageal pressure [21]. However, our calibration procedure allowed to adjust the balloon filling volume to limit the risk of balloon overstretching, and the amplitude of the difference between the directly measured non-corrected esophageal pressure and the corrected value using the strategy proposed by Mojoli et al. is likely to be very small in this setting. Third, obesity may appear as a heterogeneous disease and some features may be observed only in morbidly obese (BMI > 40 kg m−2) or may vary according to the distribution of fat tissue. ARDS may also be considered as a heterogenous syndrome, and we did not distinguish ARDS caused by pulmonary and non-pulmonary disease. Last, ARDS Berlin definition may be difficult to apply in obese patients who are often hypoxemic and for whom condensations may be difficult to assess on chest X-rays. This difficult classification may contribute to explain why some authors found that obesity was associated with a higher risk of ARDS development [41]. Interestingly, in our study in which chest X-rays were independently assessed by two experienced investigators, and ARDS diagnosis was defined by an adjudication committee, ARDS was not found to be more frequent in obese than in non-obese patients.

Conclusion

Basic respiratory mechanics and gas exchange features of obese patients are similar to those observed in non-obese ARDS patients. But an advanced assessment of respiratory mechanics allows to show that end-expiratory esophageal pressure, although largely distributed, is higher in obese patients. Chest wall compliance is not altered in obese or ARDS patients and is not easily predictable by patients’ general characteristics. A complete airway closure can be found in around 25% of critically ill patients ventilated with a PEEP of 5 cmH2O. Although it is more frequent in obese or ARDS patients, it can be observed in around 10% of non-obese non-ARDS patients. Advanced explorations including esophageal pressure and airway closure assessment can allow to better characterize individual respiratory mechanics and adjust ventilation strategies in some patients.