Year : 2011  |  Volume : 14  |  Issue : 3  |  Page : 183--187

Positive end expiratory pressure during one-lung ventilation: Selecting ideal patients and ventilator settings with the aim of improving arterial oxygenation

Nir Hoftman, Cecilia Canales, Matthew Leduc, Aman Mahajan 
 Department of Anesthesiology, David Geffen School of Medicine, University of California, Los Angeles, CA, USA

Correspondence Address:
Aman Mahajan
Ronald Reagan UCLA Medical Center, 757 Westwood Plaza, Suite 2331G, Los Angeles, CA


The efficacy of positive end-expiratory pressure (PEEP) in treating intraoperative hypoxemia during one-lung ventilation (OLV) remains in question given conflicting results of prior studies. This study aims to (1) evaluate the efficacy of PEEP during OLV, (2) assess the utility of preoperative predictors of response to PEEP, and (3) explore optimal intraoperative settings that would maximize the effects of PEEP on oxygenation. Forty-one thoracic surgery patients from a single tertiary care university center were prospectively enrolled in this observational study. After induction of general anesthesia, a double-lumen endotracheal tube was fiberoptically positioned and OLV initiated. Intraoperatively, PEEP = 5 and 10 cmH 2 O were sequentially applied to the ventilated lung during OLV. Arterial oxygenation, cardiovascular performance parameters, and proposed perioperative variables that could predict or enhance response to PEEP were analysed. T-test and c2 tests were utilized for continuous and categorical variables, respectively. Multivariate analyses were carried out using a classification tree model of binary recursive partitioning. PEEP improved arterial oxygenation by ≥20% in 29% of patients (n = 12) and failed to do so in 71% (n = 29); however, no cardiovascular impact was noted. Among the proposed clinical predictors, only intraoperative tidal volume per kilogram differed significantly between responders to PEEP and non-responders (mean 6.6 vs. 5.7 ml/kg, P = 0.013); no preoperative variable predicted response to PEEP. A multivariate analysis did not yield a clinically significant model for predicting PEEP responsiveness. PEEP improved oxygenation in a subset of patients; larger, although still protective tidal volumes favored a positive response to PEEP. No preoperative variables, however, could be identified as reliable predictors for PEEP responders.

How to cite this article:
Hoftman N, Canales C, Leduc M, Mahajan A. Positive end expiratory pressure during one-lung ventilation: Selecting ideal patients and ventilator settings with the aim of improving arterial oxygenation.Ann Card Anaesth 2011;14:183-187

How to cite this URL:
Hoftman N, Canales C, Leduc M, Mahajan A. Positive end expiratory pressure during one-lung ventilation: Selecting ideal patients and ventilator settings with the aim of improving arterial oxygenation. Ann Card Anaesth [serial online] 2011 [cited 2021 Dec 5 ];14:183-187
Available from:

Full Text


Lung isolation and one-lung ventilation (OLV) are commonly employed clinically during thoracic surgery. Improved lung isolation devices coupled with fiberoptic bronchoscopy have reduced the incidence of significant hypoxemia during OLV, yet when it does occur treatment options are often suboptimal or ineffective.

Continuous positive airway pressure (CPAP) delivered to the nonventilated lung can be effective in improving oxygenation, yet CPAP may cause lung inflation that can obscure the surgical field, especially during minimally invasive video-assisted thoracoscopic surgery (VATS). [1],[2],[3] Intermittent two-lung ventilation effectively treats hypoxemia, but disrupts the surgical flow, making VATS difficult to perform. Given these limitations, positive end-expiratory pressure (PEEP) applied to the ventilated lung has been gaining popularity. By improving lung mechanics, balancing functional residual capacity, and aiding alveolar recruitment, PEEP is purported to reduce atelectasis in the ventilated lung and thus improve oxygenation. However, increasing airway pressures in the ventilated lung can theoretically increase pulmonary resistance and thus divert blood from the ventilated to the nonventilated lung, thus increasing shunt fraction and worsening patient oxygenation.

Several small clinical studies looking at the effects of PEEP on oxygenation during OLV have yielded mixed results; some demonstrated improvement while others showed a lack thereof. Some trials employed a small number of patients undergoing up to six different ventilation strategies during a short period of time, complicating analysis of their results. Further clouding interpretation, tidal volumes employed during OLV were generally larger than currently recommended, and the inspired oxygen fraction varied among the studies. [4],[5],[6],[7] We, therefore, chose to design a study that would employ a simple PEEP protocol during a standardized general anesthetic for thoracic surgery. Furthermore, cardiovascular parameters would be measured simultaneously to help elucidate the mechanism of PEEP's actions given the interdependence of the cardiovascular and respiratory organs. The goals of this study were: (1) to evaluate the efficacy of PEEP during OLV, (2) to assess the utility of preoperative predictors of response to PEEP, and (3) to explore optimal intraoperative settings that would maximize the effects of PEEP on oxygenation.

 Materials and Methods

After obtaining UCLA's institutional review board approval, 41 ASA status I-III thoracic surgery patients requiring OLV were prospectively enrolled in this observational study. Patients unable to tolerate OLV or undergoing short procedures that could not allow for completion of the protocol were excluded from the study. Experts in thoracic anesthesia provided intraoperative care for all patients utilizing a standardized clinical protocol described below.

Clinical protocol

After radial artery cannulation and induction of general anesthesia, a double-lumen tracheal tube was fiberoptically positioned and OLV initiated. The patient was placed in the lateral decubitus position, anesthesia was maintained with 1 minimum alveolar concentration of desflurane in oxygen (FiO 2 = 1.0), and neuromuscular blocking drugs administered. Mechanical volume control ventilation during OLV employed smaller tidal volumes (mean ± SD ml/kg = 6.0 ± 1.0) and lower plateau pressures to protect the lung from ventilator-induced injury; the respiratory rate was adjusted as needed to maintain arterial carbon dioxide partial pressure in the desired range (30-40 mmHg). [8]

Intraoperatively following a minimum of 30 min of OLV, PEEP = 5 cm H 2 O was applied to the nonoperative lung for 15 min and then increased to 10 cmH 2 O for a further 15 min. A lung recruitment maneuver (valsalva maneuver held for 5 s at a pressure of 25 cm H 2 O) was first applied to the nonoperative lung, followed immediately by the dialing in of PEEP on the anesthesia machine ventilator (Datex/Ohmeda Aestiva 5). PEEP values were measured using in-line spirometry (Datex/Ohmeda). Data from arterial blood gases were recorded at: pre-induction (baseline), during OLV before PEEP, after PEEP 5 cm H 2 O, and after PEEP 10 cm H 2 O. Blood pressure and heart rate were also continuously measured. In a subset of 20 patients in whom a FloTrac/Vigileo™ monitor (Edwards LifeSciences LLP, Irvine CA; Software version 1.3) was used for clinical care, PEEP's effect on the cardiovascular system was studied by continuously measuring cardiac index (CI) and stroke volume (SV). This device uses the arterial pressure waveform coupled with proprietary algorithms to determine CI and has been validated in previous studies. [9] Data collection was confined to the time period before surgical clamping, stapling, or other form of manipulation of lung tissue.

An increase in arterial oxygen partial pressure (PaO 2 ) ≥ 20% during either PEEP5 or PEEP10 was designated as clinically significant; patients with such findings were defined as "PEEP responders". Seven preoperative clinical variables [age, BMI, forced expiratory volume in 1 s (FEV1), forced expiratory volume in 1 sec divided by forced vital capacity (FEV1/FVC), diffusion lung capacity for carbon monoxide (DLCO), baseline PaO 2 , surgical side] and six intraoperative clinical variables [baseline CI, CI on OLV before PEEP, PaO 2 on OLV before PEEP, pulmonary compliance, plateau pressure, tidal volume per kilogram (Vt/kg)] that could possibly differentiate positive responders to PEEP from non-responders were analysed.

Statistical analysis

The mean values of each continuous variable in responders vs. non-responders were calculated and compared parametrically using the t-test. Surgical side, a categorical variable, was compared in responders vs. Non-responders using the c2 -test. We assessed the percent change in PaO 2 from PEEP 0 to PEEP 5 and from PEEP 0 to PEEP 10 using one-sample t-tests. Cardiovascular parameters were compared across time points using repeated measures ANOVA. Multivariate analyses were carried out using the classification tree model of binary recursive partitioning that allowed for the modeling of nonadditive (synergistic) and nonlinear effects (SPSS Answer Tree version 3.0 Copyright© SPSS Inc. 1997-2001). The sensitivity, specificity, and accuracy under this model were computed. A P value < 0.05 was considered statistically significant.


All 41 patients enrolled in the study completed the protocol. Baseline patient demographics and medical/surgical characteristics are presented in [Table 1]; the 13 clinical variables analysed (preoperative and intraoperative) are presented in [Table 2]. Application of PEEP did not increase arterial PaO 2 (defined as an increase ≥20%) from its pre-PEEP baseline when analysing the group of patients as a whole. The mean percent changes in PaO 2 from PEEP 0 to PEEP 5 and from PEEP 0 to PEEP 10 were 3.3% and 2.0%, respectively; neither was statistically significant (P = 0.50 and 0.72, respectively). Twelve patients (29%) demonstrated increase in PaO 2 during PEEP administration [Figure 1]a, and 29 (71%) did not [Figure 1]b. Of the 13 clinical variables analysed, only intraoperative Vt/kg differed significantly between PEEP responders and non-responders (mean 6.6 vs. 5.7 ml/kg, P=0.01) [Table 2]. None of the preoperative variables proved useful in predicting PEEP responsiveness. The classification tree model (multivariate analysis) derived several combinations of Vt/kg, plateau pressure, and FEV1 (% predicted) that were statistically significant predictors of PEEP responsiveness, although the clinical utility of these formulas remains unclear.{Figure 1}{Table 1}{Table 2}

The application of PEEP to the ventilated lung was well tolerated and did not impact heart rate or blood pressure in any of the 41 patients [a in [Table 3]]. In the subset of 20 patients in whom the FloTrac™/Vigileo™ monitor was utilized, no statistical difference in SV or CO was observed during PEEP administration [b in [Table 3]].{Table 3}


The application of intraoperative PEEP to the ventilated lung during OLV failed to improve arterial oxygenation in the majority (71%) of thoracic surgery patients. However, a significant minority of patients (29%) did improve with PEEP. Our study protocol utilized a combination of low tidal volume and high fraction of inspired oxygen during OLV, which should have created a fertile ground for PEEP to demonstrate improvement given the atelectasis expected to form under such conditions. The fact that such improvement was the exception and not the rule places a damper on the routine use of PEEP during OLV in clinical practice. However, our study was not powered to determine the effect of PEEP on patients with a PaO 2 below 60 mmHg, a group that may stand to benefit most from recruitment of atelectasis. The very low response rate to PEEP of 13% seen in the subset of patients with PaO 2 ≤ 80 mmHg (n = 8), who stood to benefit most from PEEP, was especially disappointing. This outcome seems to contradict the published results of Cohen and Eisenkraft who reported a 91% improvement rate in a small (n = 10) patient population whose PaO 2 prior to application of PEEP = 10 was ≤ 80 mmHg. [4] However, closer scrutiny of the data demonstrates that the two studies had very similar findings, differing only in their definition of "responder to PEEP". Cohen et al. defined "responder" as any patient whose PaO 2 improved with PEEP by any amount, whereas we defined response as "an increase in PaO 2 ≥ 20%". Interestingly, applying our definition of response to Cohen et al.'s data set reduces their response rate from 91% to 45%, highlighting the importance of methodology in determining scientific outcomes.

We chose to define a clinical improvement as an increase in PaO 2 ≥ 20% because we felt it both (1) achieved clinical significance and (2) was unlikely to be a random fluctuation or "background noise". We believe most practitioners would feel obligated to treat an intraoperative PaO 2 = 50 mmHg, and an intervention that would raise the PaO 2 to 60 mmHg (20% improvement and an oxygen saturation ≈90%) would be deemed successful. We also calculated the results using instead an increase in PaO 2 ≥ 15%, and found no difference in outcomes. Our protocol used sequential incremental titration of PEEP because we felt it most closely resembled actual clinical care. Since the PEEP level treatment order was not randomized confounding variables could potentially have skewed the results. By allowing at least 30 min of OLV prior to data collection and avoiding lung manipulation during the experiment, we hoped to reach a steady state in gas exchange, thus reducing clinical variability during the two treatment phases.

Among the several intraoperative clinical variables examined, only Vt/kg demonstrated a statistically significant difference between PEEP responders and non-responders (6.6 vs. 5.7 ml/kg). Perhaps utilizing slightly larger, although still protective, tidal volumes improves the efficacy of PEEP. The classification tree model (multivariate analysis) was able to predict responders and non-responders with an overall accuracy of 87%. However, mathematical branch-points in the model may not necessarily represent clinically important values, and statistically significant differences among subgroups do not necessarily imply clinical significance. Future validation studies on a larger patient population could help determine this model's utility under varying clinical conditions. None of the seven preoperative clinical variables demonstrated utility at predicting response to PEEP, thus placing a damper on the process of identifying useful patient selection criteria. One may therefore conclude that although we currently lack the tools to preselect ideal PEEP candidates, we can nevertheless optimize intraoperative settings to maximize the chances of success of this intervention.

This study does however corroborate previous studies' findings that PEEP applied during OLV is very well tolerated; no deterioration in heart rate, blood pressure, CI, or SV was noted in any of the patients. [4],[10] Perhaps applying PEEP to only half the thorax preserves venous return better than when it is applied to the entire thorax, thus maintaining preload, SV, and CI during surgery.

Although this study was not designed to determine the mechanism that differentiates PEEP responders from non-responders, several hypotheses may explain this difference. First, it is probable that each patient requires a different, "customised" level of PEEP for maximal clinical benefits. Slinger et al. demonstrated this concept by correlating PEEP response to a position on the pulmonary compliance curve. [11] By utilizing smaller protective tidal volumes and incremental PEEP dosing, we assumed most patients would converge on the compliance inflection point during one of the PEEP settings. However, since we did not titrate PEEP levels or plot the pulmonary compliance curve in each individual, it is possible that ideal PEEP was not achieved in all subjects. Second, it is possible that patients in the study had different levels of intrinsic PEEP during OLV. Inomata et al. demonstrated in a small population of thoracic surgery patients that intrinsic PEEP may significantly contribute to the total PEEP, and that intrinsic PEEP varies from patient to patient. [12] Baseline intrinsic PEEP could have led to excessive total PEEP in some patients, thus contributing to the response failure. Since standard anesthesia machines cannot measure intrinsic PEEP we chose not to include this variable in the protocol. However, no patient in the study demonstrated significant air trapping on intraoperative spirometry suggestive of excessive intrinsic PEEP. Furthermore, FEV1/FVC, a measure of airway obstruction that is often associated with intrinsic PEEP, showed no correlation with response to PEEP administration, further suggesting that intrinsic PEEP did not play a major role.

In conclusion, this study demonstrates that PEEP = 5 or 10 cmH 2 O applied to the ventilated lung during thoracic surgery, although hemodynamically well tolerated, does not improve PaO 2 in a majority of patients. However, a subset of patients did respond positively to intraoperative PEEP. No preoperative variables successfully predicted positive response to PEEP. Of the intraoperative variables studies, Vt/kg differed statistically between PEEP responders and non-responders, with slightly larger tidal volumes favoring response.


The authors would like to thank Daniella Markovic for her expert statistical analysis.


1Capan LM, Turndorf H, Patel C, Ramanathan S, Acinapura A, Chalon J. Optimization of arterial oxygenation during one-lung anesthesia. Anesth Analg 1980;59:847-51.
2Fischer GW, Cohen E. An update on anesthesia for thoracoscopic surgery. Curr Opin Anaesthesiol 2010;23:7-11.
3Fujiwara M, Abe K, Mashimo T. The effect of positive end-expiratory pressure and continuous positive airway pressure on the oxygenation and shunt fraction during one-lung ventilation with propofol anesthesia. J Clin Anesth 2001;13:473-7.
4Cohen E, Eisenkraft JB. Positive end-expiratory pressure during one-lung ventilation improves oxygenation in patients with low arterial oxygen tensions. J Cardiothorac Vasc Anesth 1996;10:578-82.
5Katz JA, Laverne RG, Fairley HB, Thomas AN. Pulmonary oxygen exchange during endobronchial anesthesia: Effect of tidal volume and PEEP. Anesthesiology 1982;56:164-71.
6Leong LM, Chatterjee S, Gao F. The effect of positive end expiratory pressure on the respiratory profile during one-lung ventilation for thoracotomy. Anaesthesia 2007;62:23-6.
7Senturk NM, Dilek A, Camci E, Sentürk E, Orhan M, Tuðrul M, et al . Effects of positive end-expiratory pressure on ventilatory and oxygenation parameters during pressure-controlled one-lung ventilation. J Cardiothorac Vasc Anesth 2005;19:71-5.
8Schultz MJ, Haitsma JJ, Slutsky AS, Gajic O. What tidal volumes should be used in patients without acute lung injury? Anesthesiology 2007;106:1226-31.
9Mayer J, Boldt J, Wolf MW, Lang J, Suttner S. Cardiac output derived from arterial pressure waveform analysis in patients undergoing cardiac surgery: Validity of a second generation device. Anesth Analg 2008;106:867-72, table of contents.
10Aalto-Setala M, Heinonen J, Salorinne Y. Cardiorespiratory function during thoracic anaesthesia: A comparison of two-lung ventilation and one-lung ventilation with and without PEEP5. Acta Anaesthesiol Scand 1975;19:287-95.
11Slinger PD, Kruger M, McRae K, Winton T. Relation of the static compliance curve and positive end-expiratory pressure to oxygenation during one-lung ventilation. Anesthesiology 2001;95:1096-102.
12Inomata S, Nishikawa T, Saito S, Kihara S. "Best" PEEP during one-lung ventilation. Br J Anaesth 1997;78:754-6.