Individualized versus Fixed Positive End-expiratory Pressure for Intraoperative Mechanical Ventilation in Obese Patients: A Secondary Analysis

This analysis included all obese patients of a previous single-center study³  (German Clinical Trials Register: No. DRKS00004199, www.who.int/ictrp/network/drks2/en/) and all patients included in the PROBESE multicenter trial (ClinicalTrials.gov No. NCT02148692, https://clinicaltrials.gov/ct2/show/NCT02148692) from the University of Leipzig Medical Center (Germany). Approvals for both trials were granted by the Leipzig University Ethics Committee (No. 334/16-lk; No. 196-11-ff-8042011). Prospective electrical impedance tomography measurements were performed on the PROBESE patients to facilitate the comparison presented here, for which we obtained an addendum from the local ethics committee before the start of patient recruitment of the PROBESE study at our center. All patients gave written informed consent before inclusion. Detailed descriptions of the methods and the study protocols of both studies were recently published elsewhere,^3,20,21  and the original protocols are available in Supplemental Digital Content 1 (https://links.lww.com/ALN/C578) and Supplemental Digital Content 2 (https://links.lww.com/ALN/C579).

Patients were recruited between November 2012 and July 2013 for our single-center trial (n = 54) and between October 2016 and January 2018 for the PROBESE trial (whole study, n = 2,013; at the Leipzig site, n = 42; fig. 1). Both trials included obese patients with a body mass index greater than or equal to 35 kg/m², age greater than or equal to 18 yr, and a medium or high risk of postoperative pulmonary complications (Assess Respiratory Risk in Surgical Patients in Catalonia [ARISCAT] score greater than or equal to 26)²²  scheduled for elective abdominal surgery.

Both in the single-center and PROBESE trials, patients were randomized to an intervention and a control group. The intervention group received a recruitment maneuver followed by either individualized PEEP (single-center trial) or a fixed PEEP of 12 cm H₂O (PROBESE trial). Patients of the control group received PEEP of 5 cm H₂O (single-center trial) or 4 cm H₂O (PROBESE trial), and in the current analysis, they are combined in one “low PEEP” group (fixed PEEP of 4 to 5 cm H₂O).

All patients received a standardized infusion of crystalloid fluid and total intravenous anesthesia with propofol and remifentanil (all patients in the single-center trial; two patients in the PROBESE trial) or balanced anesthesia with desflurane/sevoflurane and sufentanil/remifentanil (38 patients in the PROBESE trial).^3,21  Routine perioperative monitoring included measurement of invasive arterial blood pressure.

Constant-flow, volume-controlled mechanical ventilation was provided by an intensive care ventilator (EVITA-XL; Dräger Medical AG, Germany) with a tidal volume (V_T) of 7 ml/kg predicted body weight in the PROBESE trial and 8 ml/kg predicted body weight in the single-center trial. Fio₂ was set to 0.4 or higher if necessary to achieve a target oxygen saturation measured by pulse oximetry greater than or equal to 92% in both trials. The initial respiratory rate was set to 12 breaths per minute. The inspiratory-to-expiratory ratio and inspiratory time were set to achieve end-inspiratory and end-expiratory zero flow with inspiratory pause. During anesthesia, respiratory rate, inspiratory flow, and inspiratory-to-expiratory ratio were adjusted as required to maintain normocapnia (Paco₂ greater than or equal to 35 and less than or equal to 45 mmHg).

Timing of measurements, ventilation and PEEP setting, and the positioning of all patients are illustrated in figure 2. All patients in the PROBESE trial at the University of Leipzig Medical Center prospectively received additional electrical impedance tomography measurements at time points to coincide with the single-center trial. Electrical impedance tomography recordings were performed before induction of anesthesia and 2 h after extubation; arterial blood gas analysis and electrical impedance tomography data were measured in the spontaneously breathing patient at ambient air pressure. After induction of anesthesia, baseline measurements were performed in all patients 10 min after the start of mechanical ventilation at a PEEP of 4 or 5 cm H₂O. After the last measurement point, Fio₂ was set to 1.0 before extubation according to current clinical standards.

The patients randomized to control groups remained at fixed PEEP of 4 to 5 cm H₂O without receiving recruitment maneuvers at any time.^3,21 

According to the PROBESE protocol, patients randomized to the fixed PEEP of 12 cm H₂O group received recruitment maneuvers (inspiration-to-expiration ratio, 1:1; PEEP, 12 cm H₂O; respiratory rate, 6 breaths per minute; increasing V_T in steps of 4 ml/kg of predicted body weight until plateau pressure reaches 40 cm H₂O followed by three breaths while maintaining plateau pressure of 40 to 50 cm H₂O) after tracheal intubation (and after ventilator disconnection) and hourly throughout mechanical ventilation.³ 

The individualized PEEP group received a recruitment maneuver (inspiration-to-expiration ratio, 1:1; peak pressure, 50 cm H₂O; PEEP, 30 cm H₂O; respiratory rate, 6 breaths per minute) for 10 breaths followed by a unique decremental PEEP titration before the start of surgery and before insufflation of pneumoperitoneum. Starting with a PEEP of 26 cm H₂O and descending in steps of 2 cm H₂O to a PEEP of 4 cm H₂O, the regional ventilation delay index was calculated at every PEEP step, and the PEEP value corresponding to the lowest value of regional ventilation delay index was identified as individual best PEEP,^3,19,23  which was maintained throughout mechanical ventilation. An additional recruitment maneuver was performed immediately before extubation at the end of the surgical procedure.³ 

Hypotension was defined as mean arterial pressure less than 60 mmHg. Because we wanted to exclude possible confounding effects on pulmonary circulation, norepinephrine was the only vasopressor applied for treating hypotension. Bradycardic episodes during the recruitment maneuver were exclusively treated with atropine.

Electrical impedance tomography monitoring consists of an electrode belt containing 16 or 32 electrodes that is placed around the chest cranial to the diaphragm at the height of the third to fourth intercostal space. An alternating current (typically 5 mA at a frequency of 50 kHz) is applied to a first pair of electrodes, and the resulting surface potentials are registered in the remaining electrode pairs. Subsequently, the location of the current injection and voltage measurements is rotated continuously around the chest. Thus, a complete rotation of the injecting electrode pair results in a total of 208 voltage measurements when using a 16-electrode belt. Impedance of the tissue passed by the current is calculated using the Ohm law. After algorithm-based reconstruction, a two-dimensional, real-time, cross-sectional image of pulmonary ventilation is displayed on a monitor. Changes in relative impedance during a respiratory cycle are displayed using a white to dark blue color scheme (with darker color tones of blue indicating a decrease in impedance, which is to be equated with a decrease in regional ventilation).

For electrical impedance tomography–based individualization of PEEP in the single-center study, the regional ventilation delay for each pixel in the electrical impedance tomography image was determined during a standardized low-flow maneuver of 12 ml/kg predicted body weight, and the regional ventilatory delay index was then defined as the SD for all pixels’ regional ventilatory delay. The lower the regional ventilation delay index, the more homogeneous the regional ventilation of the lung with reduction of overdistention as well as atelectasis formation. The PEEP corresponding to the lowest regional ventilatory delay index was set as individualized PEEP (method described earlier in Nestler et al. including its supplement³ ; see also Muders et al.¹⁹ ).

Ventilation parameters were recorded by the display of the mechanical ventilator. Blood gases, lactate, and pH were determined immediately after sampling with a standard blood gas analyzer (ABL 800; Radiometer, Denmark). Ventilation distribution images were obtained with a commercially available electrical impedance tomography system (PulmoVista 500; Dräger Medical AG). Tidal distributions of ventilation to dependent and nondependent lung zones²⁴  were quantified offline using the commercially available Dräger electrical impedance tomography analysis tool (version 6.1; Dräger Medical AG) and customized institutional software. The ratio of dependent to nondependent ventilation is provided.

The analysis and statistical plan were conceived after the data were accessed. However, the plan to carry out this secondary analysis was documented before inclusion of local patients in the PROBESE trial after discussing it with the steering committee and arranging for the electrical impedance tomography measurements. Visual inspection of the primary variables was performed to detect potential differences between the low PEEP group from the PROBESE and the single-center trial. Given this inspection, the similarities in trial design and inclusion criteria, the small time difference between patient recruitment, and the study team being the same, the assumption was made that the local site data from the trials can be pooled. The primary endpoint Pao₂/Fio₂ immediately before extubation was adopted from the single-center study (registered in 2012) and chosen for comparability between the primary analysis and the current subanalysis. Further endpoints are intraoperative percentage of V_T to nondependent/dependent lung, which is the percentage of the tidal ventilation of the dependent lung in relation to total ventilated lung area as defined in Mauri et al.,²⁴  and the ventilatory ratio, which divides the measured by the expected product of minute ventilation and Paco₂ as defined in Sinha et al.²⁵ 

With the observed SDs for the individualized PEEP group of the single-center study being 10.2 percentage points for ventilation distribution and 85.5 mmHg for Pao₂/Fio₂, the current analysis would have 80% power in detecting a difference of 10.0 percentage points and 11.2 kPa, respectively, compared to the fixed PEEP of the 12 cm H₂O group. These estimates use a significance level of 2.5% for Bonferroni correction.

Differences between groups at a prespecified point in time were compared with a linear model with the baseline value as a covariate and a Bonferroni correction for multiple groups. For these models, a Wald CI was used, and the P value corresponded to it. In data with heavily skewed distributions, group comparisons were made using a Kruskal–Wallis test. For repeated measures analyses, a mixed effects model was used, optimizing the restricted maximum likelihood (using the lme4 package²⁶ ), which will provide unbiased estimates despite missing data, assuming that they are “missing at random.” Patient was taken as the random term with group and time including their interaction as fixed effects. A type two ANOVA was performed with a Wald chi-square test using the default settings from the “Anova” command for the “mer” class from the “car” package.²⁷  Differences in categorical variables were assessed by the chi-square test. For analysis of ventilation distribution at different PEEP levels during the titration process, a paired t test with Bonferroni correction was used. The association between continuous variables is assessed with linear correlation, where the CI is based on Fisher z transformation. Comparison of two dependent correlations was performed with the Zou method from the “cocor” package.^28,29  All statistical analyses were performed using R Version 3.6.1 (RStudio, Version 1.2.1335, http://www.R-project.org/; R Foundation for Statistical Computing, Austria). Summary statistics are presented as mean ± SD, all tests were two-tailed, and P < 0.05 was considered to be statistically significant.

A total of 90 patients were included in this analysis: 50 patients from the single-center trial³  and 40 patients from the PROBESE trial.²⁰  The patients were categorized into three different groups subject to their random allocation in the initial clinical trials: fixed PEEP of 4 to 5 cm H₂O (44 patients), individualized PEEP (25 patients), and fixed PEEP of 12 cm H₂O (21 patients). Visual inspection did not indicate any relevant differences in important variables between the low PEEP groups from the PROBESE and single-center trials. The patient flow (fig. 1) demonstrates that all patients who underwent surgery were included in the analysis of the primary outcome and that the intended intervention was provided to each of these patients. Their baseline characteristics are listed in table 1.

Seventy-three patients underwent a gastric bypass surgery, 16 patients had a sleeve gastrectomy (nine from the fixed PEEP of 4 to 5 cm H₂O, two from the individualized PEEP, and five from the fixed PEEP of 12 cm H₂O group), and one patient from the fixed PEEP of 4 to 5 cm H₂O group had rectal cancer surgery. All surgeries were performed laparoscopically and lasted for a mean ± SD of 173 ± 36, 128 ± 42, and 143 ± 47 min in the individualized PEEP, fixed PEEP of 12 cm H₂O, and fixed PEEP of 4 to 5 cm H₂O groups, respectively. Surgery was performed in the reverse Trendelenburg position, with exception of the patient who received rectal cancer surgery in the Trendelenburg position. The intraoperative characteristics provided in table 2 show that titration of individualized PEEP led to an average PEEP of 18.5 cm H₂O (median, 18 [interquartile range, 16 to 22; range, 10 to 26 cm H₂O]).³  The body mass index in the individualized PEEP ranged from 38.1 to 65.1 kg/m², and the supplement in the single-center trial³  shows that the correlation between optimal PEEP and body mass index is too weak to use body mass index for choosing a PEEP value.

In the primary variables, a small number of data are missing for technical reasons (fig. 3). The Pao₂/Fio₂ ratio before tracheal extubation was 515.3 mmHg in the individualized PEEP group, 369.8 mmHg in the fixed PEEP of 12 cm H₂O group, and 304.5 mmHg in the fixed PEEP of 4 to 5 cm H₂O group. The difference to the individualized PEEP group was 145.5 mmHg ([95% CI, 90.9 to 200.4 mmHg; P < 0.001] fixed PEEP of 12 cm H₂O) and 210.8 mmHg ([95% CI, 164.5 to 257.6 mmHg; P < 0.001] fixed PEEP of 4 to 5 cm H₂O]). Figure 3A depicts the course of Pao₂/Fio₂ through time. After extubation, the differences between the groups vanished.

The individualized PEEP group also showed better respiratory system mechanics as indicated by lower driving pressure and, thus, higher dynamic compliance (table 2; fig. 3B). Notably, Paco₂ and end-tidal partial pressure of carbon dioxide do not differ significantly between groups. The ventilatory ratio increased during the course of the operation, but did not differ between the groups.

Induction of anesthesia and mechanical ventilation shifted regional distribution of V_T to the nondependent lung, which could be reversed in the individualized PEEP group, but remained almost unchanged in both fixed PEEP groups until patients returned to spontaneous breathing (fig. 3C). Intraoperatively, the percentage V_T in the dependent lung was 43.9% in the individualized PEEP group. It was 25.9% in the fixed PEEP of 12 cm H₂O group (difference to individualized PEEP, 18% [95% CI, 8.0 to 20.7]; P < 0.001) and 26.8% in the fixed PEEP of 4 to 5 cm H₂O group (difference to individualized PEEP, 17.1% [95% CI, 10.0 to 20.6]; P < 0.001).

At each PEEP titration stage in the individualized PEEP group, electrical impedance tomography measurements were performed. Thus, it is possible to provide values at each PEEP level. Figure 4 shows that the mean V_T percentage in the dependent lung was 9.7 percentage points (95% CI, 6.9 to 12.7; P < 0.001) lower at a PEEP of 4 cm H₂O compared to 12 cm H₂O, which was, in turn, 11.3 points (95% CI, 7.9 to 14.8; P < 0.001) lower compared with the individualized PEEP setting.

The linear correlation coefficient between regional distribution of V_T and the ventilatory ratio was −0.08 (95% CI, −0.31 to 0.16; P = 0.496) and was −0.51 (95% CI, −0.67 to −0.31; P < 0.001) between regional distribution of V_T and Pao₂/Fio₂. The difference between the two correlation coefficients is 0.43 (95% CI, 0.10 to 0.73; P = 0.010). In a linear regression model, Pao₂/Fio₂ decreased by 51.8 mmHg for every 10 percentage points change in regional distribution of V_T. Furthermore, there was a strong inverse correlation between driving pressure and Pao₂/Fio₂, with a linear correlation coefficient of −0.68 (95% CI, −0.79 to −0.53; P < 0.001).

Basic hemodynamic data and a list of the complications by treatment arm are provided in table 3. During anesthesia and surgery, a considerable number of patients in all three arms had arterial hypotension and required vasopressors (in all cases, norepinephrine only) despite crystalloid fluid infusion of about 4 ml · kg^-1 · h^-1, with a slightly—but significantly—larger infusion rate in the individualized PEEP arm (table 3). The number of patients requiring continuous norepinephrine infusion did not significantly differ between groups (P = 0.764), and overall rates were low. Moreover, 80% of patients in the individualized PEEP, 86% in the fixed PEEP of 12 cm H₂O, and 89% in the fixed PEEP of 4 to 5 cm H₂O groups also received boluses of vasoactive injections during anesthesia outside the recruitment maneuvers (P = 0.585). The bolus equivalents differed significantly between the groups (P = 0.017), but were low overall. The course of lactate was significantly lower in the individualized PEEP group (P = 0.007), but we did not observe differences in the course of base excess, nor did we observe any signs of cardiovascular insufficiency or shock. During recruitment maneuvers, almost all of the individualized PEEP patients (92%) but only half of the fixed PEEP of 12 cm H₂O patients (48%) required vasoactive medication.

We showed that for obese patients undergoing laparoscopic surgery, recruitment maneuver followed by electrical impedance tomography–guided individualized PEEP resulted in (1) higher PEEP, (2) better arterial oxygenation, (3) lower driving pressures, and (4) redistribution of ventilation to dependent lung areas, when compared with recruitment maneuver followed by a fixed PEEP of 12 cm H₂O or low PEEP of 4 or 5 cm H₂O without recruitment maneuver. Taken together and considering indicators of dead space and overdistention, these findings strongly suggest that individually titrated PEEP values further reduce atelectasis formation in dependent lung areas as compared with both groups of the PROBESE study, which used a pragmatic approach comparing lower and higher fixed PEEPs not necessarily aiming for full lung recruitment.²⁰  Changes in end-expiratory lung volume previously reported in the patients from the single-center trial provide evidence that atelectasis occurs at lower PEEP.³  Lung recruitment potential above a PEEP of 12 cm H₂O is confirmed by intraindividual comparisons of regional ventilation distribution in the individualized PEEP subgroup (fig. 4), showing significantly higher ventilation of the dependent lung with individualized PEEP. The difference between the fixed PEEP of 4 cm H₂O and the fixed PEEP of 12 cm H₂O arms in driving pressure before extubation in our local subgroup was comparable to that of the whole PROBESE trial (−6.6 vs. −5.6 cm H₂O, respectively), suggesting that the effects we observed locally might be representative of the whole study population. This implies that a recruitment maneuver followed by a high PEEP of 12 cm H₂O is not sufficient to recruit the lung completely, at least for some of the patients.

Measurements of regional ventilation by electrical impedance tomography during a low-flow breath show delayed filling of (mainly dependent) lung areas if PEEP is set below the closing pressure of those lung units.^20,23  If airway pressure exceeds the opening pressure of the collapsed lung unit during the low-flow breath, a delayed change in local impedance (ventilation) can be detected by electrical impedance tomography. Minimizing this temporal inhomogeneity of regional ventilation by setting PEEP according to the regional ventilation delay index approach^3,20,23  should thus result in the lowest PEEP that minimizes tidal alveolar collapse. Recent experimental evidence demonstrates that the regional ventilation delay index approach improves the ventilation/perfusion ratio with a lower PEEP than PEEP optimization using an open-lung approach based on highest oxygenation.³⁰  In the single-center study,³  individualized PEEP determined by this regional ventilation delay index approach resulted in PEEP values of 10 to 26 (median, 18) cm H₂O, which were significantly higher than the 12 cm H₂O used in the PROBESE study.²⁰  Other studies in bariatric surgery using other optimization parameters regularly found PEEP levels greater than 15 cm H₂O.^12,31  In laparoscopic nonobese patients, mean PEEP was about 15 cm H₂O.³²  Our findings are in line with a recently published study by Tharp et al.⁴  that showed a correlation between optimal PEEP and patient body mass index, but also found that body mass index could not be used to prescribe PEEP, further highlighting the need for individualized ventilation in this specific patient group. Practical issues that may hinder the application of an electrical impedance tomography−based PEEP strategy in the operating room include the limited availability of electrical impedance tomography and the fact that some anesthesia machines may not allow PEEP values greater than 20 cm H₂O.

Recruitment maneuvers are frequently used to apply high pressures to the lung tissue and may themselves contribute to ventilator-associated lung injury.³³  However, the intraoperative improvements in the individualized PEEP group were achieved without repetitive intraoperative recruitment maneuvers³  as performed in the PROBESE trial.²⁰  An intraoperative repetition of recruitment maneuvers does not improve gas exchange and lung function³⁴ ; for patients with mainly primary pulmonary acute respiratory distress syndrome (ARDS), it is critical³⁵  and may additionally lead to severe hemodynamic compromise.

Individualized PEEP was associated with better respiratory system mechanics compared with a fixed PEEP, regardless of whether it was 4, 5, or 12 cm H₂O. The parameters indicating this were better distribution of gas to dependent lung areas, lower driving pressures, and higher oxygenation. The ventilatory ratio suggests that there was a change in neither overdistention nor dead space ventilation. Moreover, there was no correlation between distribution of ventilation and ventilation ratio, but there was a correlation between distribution of ventilation and oxygenation. This supports the notion of higher ventilation without an increase in dead space. The significantly higher driving pressure of both the fixed PEEP of 4 to 5 cm H₂O and fixed PEEP of 12 cm H₂O groups may contribute to increased postoperative pulmonary complications.³⁶  Note that there was a small difference in intraoperative tidal volumes between groups, which is due to the fact that different, but fixed, tidal volumes were prescribed in the two studies (7 ml/kg predicted body weight in the PROBESE study vs. 8 ml/kg predicted body weight in the single-center study). However, we do not expect this to be clinically relevant with regard to postoperative pulmonary complications, as these tidal volumes agree with the concept of modern lung protective ventilation strategies,³⁷  and even higher tidal volumes up to 10 ml/kg do not lead to more frequent occurrence of postoperative pulmonary complications.³⁸ 

The large PROBESE trial could not confirm the hypothesis that a ventilation approach with a higher PEEP of 12 cm H₂O and recruitment maneuvers reduced postoperative pulmonary complications, despite reduction in driving pressures.²⁰  This seems to be in contrast to patients with ARDS, where lower driving pressures proved to be associated with improved outcome.^39,40  However, the relative risks in the ARDS studies would have equated to a risk difference of fewer than 3 percentage points given the observed driving pressures in PROBESE, and the study was not powered to detect such small differences. It is quite conceivable that individualized PEEP, which leads to lower driving pressures, might have greater impact on clinical outcome.

The lack of difference in postoperative pulmonary complications between high and low PEEP groups in the PROBESE trial may be partially explained by our consistent observations that lung recruitment associated with higher PEEP in obese patients was not preserved after tracheal extubation (fig. 3, A and C).³  Measures to prevent lung collapse such as avoiding ventilation with 100% oxygen before tracheal extubation or the use of continuous positive airway pressure in the early postoperative phase are therefore needed and are currently under investigation (ADIPO-CPAP, German Clinical Trials Register No. DRKS00013984). If these measures help prevent postoperative lung collapse and preserve the effects of an individualized PEEP strategy compared to a standard PEEP, it is quite plausible that postoperative pulmonary complications can indeed be reduced by mechanical ventilation strategies in obese patients. However, we cannot completely rule out the possibility that those postoperative lung stabilization strategies would be beneficial also in patients who were ventilated intraoperatively with low PEEP, rendering again the use of individualized intraoperative PEEP not superior to low PEEP. Furthermore, the potential gains of individualized PEEP have to be weighed against hemodynamic depression and increased use of vasoactive medication.

Analyses combining data from different studies spanning several years may contain systematic bias that is difficult to detect. In our case, there were slight differences in the prescribed tidal volumes and the recruitment maneuvers. However, the patients included in this analysis were all from a single center, had the same inclusion criteria, and were treated by the same team of physicians within 6 yr.

The choice of oxygenation as primary endpoint may not be convincing to all readers. It is important to keep in mind that here we are in the context of predefined low V_Ts and that the analysis of compliance, driving pressure, and ventilation index provide important information on distention in nondependent lung areas.

For practical reasons, PEEP titration was performed in a 30-degree reverse Trendelenburg position, but before initiation of capnoperitoneum. Therefore, individualized PEEP did not take the effects of capnoperitoneum on respiratory system mechanics into account. Furthermore, the capnoperitoneum measurement was done without readjustment of individualized PEEP. However, surgery was performed in the reverse Trendelenburg position in our study, which is supposed to lower pleural pressure counterbalancing the intraabdominal pressure applied during capnoperitoneum. Even if the effect of capnoperitoneum on pulmonary mechanics seems to be limited in obese patients,^41,42  we cannot exclude the possibility that different PEEP values would have been obtained if PEEP titration had been performed after insufflation of capnoperitoneum. Furthermore, our results cannot be fully transferred to other patient populations or to laparoscopic procedures with different patient positioning, even if a recent publication found that comparable individualized PEEP values are necessary in normal-weight patients undergoing laparoscopic surgery in extreme Trendelenburg positioning.³² 

Our study was neither designed nor powered to investigate the impact of an individualized PEEP strategy on the incidence of postoperative pulmonary complications. Considering the slightly higher rate of hemodynamic complications and the time needed for the PEEP titration, we strongly encourage (1) testing whether an individualized PEEP strategy also translates into a lower rate of postoperative pulmonary complications, and (2) searching for alternative methods for individualization of mechanical ventilation that would be more applicable in clinical routine.

In obese patients undergoing laparoscopic surgery, an electrical impedance tomography−based individualized ventilation strategy resulted in variable PEEP that, on average, exceeded the fixed PEEP of 12 cm H₂O used in the PROBESE study. Compared with the strategy that uses PEEP of 12 cm H₂O, individualized PEEP was associated with higher intraoperative respiratory system compliance and better oxygenation and distribution of ventilation to dependent lung areas—all consistent with higher alveolar recruitment. These benefits vanished after extubation. If alveolar recruitment with individualized PEEP can be preserved by applying adequate measures after tracheal extubation, it will be important to determine if this translates to a lower rate of postoperative pulmonary complications.

The authors thank the Steering Committee for the Protective Intraoperative Ventilation with Higher versus Lower Levels of Positive End-Expiratory Pressure in Obese Patients (PROBESE) Collaborative Group (listed in the appendix) for critical revision of important intellectual content. The authors would like to thank Arne Dietrich, M.D., Ph.D. (Department of Surgery, University of Leipzig Medical Center, Leipzig, Germany) for his great support in recruiting the bariatric patients as well as planning and designing the single-center study.

Part of this trial was funded by the Federal Ministry of Education and Research, Germany (Integrated Research and Treatment Center Adiposity Diseases; FKZ: 01E01001), and by departmental funding.

Dr. Wrigge received research funding, lecture fees, and technical support from Dräger Medical (Lübeck, Germany); funding from Pfizer (Investigator Initiated Trial Program; Berlin, Germany); funding and lecture fees from InfectoPharm (Heppenheim, Germany); lecture fees from GE Healthcare (Freiburg, Germany), Maquet (Rastatt, Germany), and MSD (Konstanz, Germany); advisory board honoraria from Liberate Medical (Crestwood, Kentucky); and technical support from Swisstom Corp. (Landquart, Switzerland). Dr. Simon received funding and lecture fees from InfectoPharm (Heppenheim, Germany). Dr. Gama de Abreu received fees for consulting activities or contract research from Ambu (Bad Nauheim, Germany), ZOLL Medical Deutschland GmbH (Cologne, Germany), GE Healthcare (Freiburg, Germany), and Dräger Medical (Lübeck, Germany). Dr. Bluth received a Research Grant European Society of Anesthesiology (now European Society of Anaesthesiology and Intensive Care) for administrative issues regarding the PROBESE clinical randomized controlled trial (for which study patients provided part of the source data for the current analysis). Dr. Beda is a full-time employee of Hamilton AG (Bonaduz, Switzerland), but this relationship started after his major contribution to the study. The other authors declare no competing interests.

Thomas Bluth, M.D. (Department of Anesthesiology and Critical Care Medicine, Pulmonary Engineering Group, University Hospital Carl Gustav Carus, Dresden Institute of Technology [Technische Universität Dresden], Dresden, Germany); Ilona Bobek, M.D., Ph.D. (South Pest Central Hospital−National Institute of Hematology and Infectious Diseases, Budapest, Hungary); Jaume C. Canet, M.D., Ph.D. (Department of Anesthesiology and Postoperative Care, Germans Trias i Pujol University Hospital, Badalona, Barcelona, Spain); Luc De Baerdemaeker, M.D., Ph.D. (Department of Anesthesiology and Perioperative Medicine, Ghent University Hospital, Ghent, Belgium); Cesare Gregoretti, M.D., Ph.D. (Department of Surgical, Oncological and Oral Science, University of Palermo, Palermo, Italy; G. Giglio Institute of Cefalù Foundation, Palermo, Italy); Göran Hedenstierna, M.D., Ph.D. (Department of Medical Sciences Clinical Physiology, Academic Hospital−Uppsala, Uppsala, Sweden); Sabrine N. T. Hemmes, M.D., Ph.D. (Department of Intensive Care, Amsterdam University Medical Centers, Location Academic Medical Center, Amsterdam, The Netherlands); Michael Hiesmayr, M.D., Ph.D. (Division Cardiac, Thoracic, Vascular Anesthesia and Intensive Care, Medical University Vienna, Vienna, Austria); Markus Hollmann, M.D., Ph.D. (Department of Anesthesiology, Amsterdam University Medical Centers, Location Academic Medical Center); Samir Jaber, M.D., Ph.D. (Department of Anesthesia and Intensive Care, Saint Eloi Montpellier University Hospital, Montpellier, France; Physiology and Experimental Medicine, University of Montpellier, Montpellier, France); John Laffey, M.D., Ph.D. (Department of Anesthesia and Intensive Care Medicine, Galway University Hospitals, Galway, Ireland; School of Medicine, Regenerative Medicine Institute at the CÚRAM Center for Medical Devices, National University of Ireland, Galway, Ireland); Marc J. Licker, M.D., Ph.D. (Department of Anesthesiology, Pharmacology and Intensive Care, Faculty of Medicine, University Hospital of Geneva, Geneva, Switzerland); Klaus Markstaller, M.D., Ph.D. (Department of Anesthesia, Intensive Care Medicine and Pain Medicine, Medical University of Vienna, Vienna, Austria); Idit Matot, M.D., Ph.D. (Division of Anesthesia, Intensive Care and Pain Medicine, Tel Aviv Medical Center, Sackler Faculty of Medicine, Tel Aviv University, Tel Aviv, Israel); Gary Mills, M.D., Ph.D. (Operating Services, Critical Care and Anesthesia, Sheffield Teaching Hospitals and University of Sheffield, Sheffield, United Kingdom); Jan Paul Mulier, M.D., Ph.D. (Department of Anesthesia, General Hospital Sint-Jan Brugge-Oostende AV, Campus Sint-Jan, Brugge, Belgium); Christian Putensen, M.D., Ph.D. (Department of Anesthesiology and Intensive Care Medicine, University Hospital Bonn, Bonn, Germany); Rolf Rossaint, M.D., Ph.D. (Department of Anesthesiology, Faculty of Medicine, Rheinisch-Westfälische Technische Hochschule Aachen University, Aachen, Germany); Jochen Schmitt, M.D., Ph.D. (Center for Evidence-Based Healthcare, University Hospital Dresden, Dresden Institute of Technology [Technische Universität Dresden]); Mert Senturk, M.D., Ph.D. (Department of Anesthesiology and Intensive Care of Istanbul Medical Faculty, Istanbul University, Istanbul, Turkey); Paolo Severgnini, M.D., Ph.D. (Department of Biotechnology and Sciences of Life, ASST Sette Laghi Circolo Hospital and Macchi Foundation, University of Insubria, Varese, Italy); Juraj Sprung, M.D., Ph.D. (Department of Anesthesiology and Perioperative Medicine, Mayo Clinic, Rochester, Minnesota); Marcos Francisco Vidal Melo, M.D., Ph.D. (Department of Anesthesia, Critical Care and Pain Medicine, Massachusetts General Hospital, Boston, Massachusetts); Hermann Wrigge, M.D., Ph.D. (Integrated Research and Treatment Center Adiposity Diseases, University of Leipzig, Leipzig, Germany; Department of Anesthesiology, Intensive Care and Emergency Medicine, Pain Therapy, Bergmannstrost Hospital Halle, Halle, Germany); Ary Serpa Neto, M.D., M.Sc, Ph.D. (Department of Intensive Care, Amsterdam University Medical Centers, Location Academic Medical Center; Department of Critical Care Medicine, Hospital Israelita Albert Einstein, Sao Paulo, Brazil); Marcus J. Schultz, M.D., Ph.D. (Department of Intensive Care, Academic Medical Center, University of Amsterdam, Amsterdam, The Netherlands); Paolo Pelosi, M.D., Ph.D. (Department of Surgical Sciences and Integrated Diagnostics, University of Genoa, Genoa, Italy; Anesthesia and Intensive Care, San Martino Policlinico Hospital, Scientific Institute for Research, Hospitalization and Healthcare for Oncology and Neurosciences, Genoa, Italy); and Marcelo Gama de Abreu, M.D., Ph.D. (Department of Anesthesiology and Critical Care Medicine, Pulmonary Engineering Group, University Hospital Carl Gustav Carus, Dresden Institute of Technology [Technische Universität Dresden]).