Ethics, consent and permission
The study was approved by the Ethics Committee of the University Medical Center of Freiburg (vote # 268/15) on June 29, 2015 and registered at the German Register for Clinical Trials (DRKS00008924). This study adhered to the CONSORT guidelines.
Study design and patient population
After obtaining written informed consent from the participants, we studied respiratory mechanics, hemodynamic variables, and regional ventilation in 60 consecutive patients with American Society of Anesthesiologists (ASA) physical status I-III, who underwent otorhinolaryngeal surgery at the Medical Center of the University of Freiburg, Germany. The study was performed as a prospective parallel-arm, randomized controlled trial with an allocation ratio of 1:1. Randomization was carried out in blocks of 30 by a computer-generated allocation sequence. Participants were enrolled and assigned to the interventions by a study-related anesthetist. The exclusion criteria were ASA physical status > III, age < 18 years, pregnancy, emergency procedure, cardiac pacemaker and other active implants, obesity (body mass index ≥ 30 kg·m-2), a history of pulmonary disease, laparoscopic surgery, or refusal to participate.
After primary recruitment and preoperative evaluation, the patients received routine monitoring (electrocardiography, SpO2 and noninvasive blood pressure measurement; Infinity Delta XL, Dräger Medical, Lübeck, Germany). After preoxygenation to an expiratory fraction of oxygen of 0.8, anesthesia was induced and maintained as total intravenous anesthesia with a continuous infusion of propofol (Propofol 1%; Fresenius Kabi, Bad Homburg, Germany; target-controlled infusion, effect site target concentration for induction: 6–8 µg·mL-1; effect site target concentration for maintenance: 3–5 µg·mL-1, Agilia, Schnider Model; Fresenius Kabi) and remifentanil (TEVA GmbH, Ulm, Germany; induction: 1–2 µg·kg-1, maintenance: 0.15–0.3 µg·kg-1·min-1). During the study protocol, a Bispectral Index™ (BIS™) monitoring (Medtronic, Minneapolis, USA) was used as an additional monitor of anesthesia depth (BIS value target 40–60). Tracheal intubation was facilitated with 0.15 mg·kg-1 predicted body weight  iv cisatracurium (Fresenius Kabi). Potential hypotension, defined as mean arterial pressure < 65 mmHg, was treated with a continuous norepinephrine infusion (0.03–0.2 µg·kg-1·min-1). Volume requirements were addressed individually, according to clinical judgement, with a crystalloid solution (Jonosteril; Fresenius Kabi). For tracheal intubation, we used tracheal tubes with low pressure cuffs, with an internal diameter of 7.0–7.5 mm for women and 8.0 mm for men (Mallinckrodt Hallo-Contour; Covidien, Neustadt an der Donau, Germany). All patients were ventilated in the volume-controlled mode with a tidal volume (VT) of 7 mL·kg-1 predicted body weight. Ventilation frequency was set to maintain an end-tidal carbon dioxide partial pressure between 35 and 40 mmHg. In all patients, the initial PEEP was set to 5 cmH2O. Following these baseline measurements, the randomization was disclosed. In the control group, the PEEP was maintained for the whole procedure. In the intervention group, the PEEP was adjusted dynamically according to the recommendations resulting from the intratidal compliance profile analysis (see below).
To calculate nonlinear intratidal CRS profiles via the gliding-SLICE method, we chose 21 equidistant slices as a tradeoff between calculation effort and reasonable resolution. The resulting intratidal compliance curves were classified into six different compliance profiles, as described earlier [19, 22, 23]. In brief, a second-order polynomial was fitted into the compliance-volume curve, and the resulting segment of a parabola was assumed to represent the compliance-volume curve in a filtered form. If the segment showed an increase of more than 20% of the compliance maximum, the profile was classified as containing an increasing part. A segment decreasing by more than 20% of the compliance maximum was classified as containing a decreasing part. A segment containing the angular point of the parabola was classified as containing the horizontal part. A compliance profile with less than 20% change was classified as horizontal (Fig. 1) . The decision support system suggested a PEEP increase of 2 cmH2O in the case of a merely increasing compliance profile, 1 cmH2O in the case of an increasing compliance profile with a horizontal component, a PEEP decrease of 2 cmH2O for a merely decreasing compliance profile, and 1 cmH2O in the case of a decreasing compliance profile with horizontal component. A merely horizontal compliance profile resulted in the suggestion to maintain PEEP as it was.
Electrical impedance tomography
Regional ventilation was measured via electrical impedance tomography (EIT) (PulmoVista 500, Dräger Medical) every 10 minutes for a duration of 2 minutes. EIT recordings were evaluated offline using software developed in Matlab (MATLAB R2014a, The Mathworks Inc., Natick, MA, USA). As a first step, the relevant lung areas were determined for each patient by applying the lung area estimation method [24, 25] to the raw EIT data. The functional region of interest was selected by deleting all pixels with an impedance change smaller than 20% of the maximum tidal impedance change. The remaining pixels were mirrored to compensate for potential atelectatic areas. The obtained lung area was then applied to all the recorded raw EIT images. After this preprocessing, the functional impedance images were generated by subtracting the frames corresponding to the start of inspiration from the frames corresponding to the end of inspiration. These functional images (f-EIT) thus represented the distribution of the tidal volume for each breath. To assess potential changes in regional ventilation, tidal variation as well as a gain and loss calculations were performed and compared between the two groups. The gain and loss calculations were based on subtracting the functional impedance images of different time points to directly compare differences in ventilation between them. In this study, the averaged f-EIT images of the first EIT recording (baseline measurement, prior to the surgical procedure) and the averaged f-EIT images of the last EIT recording (after the surgical procedure was finished) were subtracted for each patient. The resulting differential images were split into ventral and dorsal parts and the number of positive (‘gain’) and negative (‘loss’) pixels were calculated for each such region. A gain was represented by the number of pixels that exhibited an increase in aeration in the last measured EIT sequence compared to the first (baseline) measured EIT sequence and loss was shown by a decrease in aeration. The results were compared between the two different groups. The change in tidal volume (ΔVT) was calculated as the difference between gain (TVG) and loss (TVL) (ΔVT = TVG - TVL) for the previously defined ventral (ΔVT,v) and dorsal (ΔVT,d) lung areas. This provided a measure for changes in regional ventilation. If this difference was positive, we assumed an increase in regional ventilation in the respective lung area, whereas a negative difference indicated a decrease in regional ventilation .
Tidal variation (impedance distribution) is the percentage of tidal volume going to the ventral (TVv) and the dorsal areas (TVd). This was calculated for all the functional impedance images using Equation 1, (see Equation 1 in the Supplementary Files)
where xi,v are the impedance values in the ventral region, xi,d the impedance values in the dorsal region and xi the sum of all impedance values of the f-EIT under consideration. Tidal variation was calculated for each averaged f-EIT image of each 2-minute EIT recording.
End points and data collection
The frequency of each type of nonlinear intratidal CRS profile (measured using the gliding-SLICE method) was the primary endpoint of this study. The secondary endpoints were regional ventilation (ventral and dorsal ventilation distribution, ventral and dorsal gain and loss and tidal variation), the respiratory system variables (peak inspiratory pressure [PIP], PPlat, mean tracheal pressure [Pmean], PEEP) and hemodynamic variables (systolic blood pressure [BPsys], diastolic blood pressure [BPdias], heart rate and mean arterial pressure [MAP]). The intratidal compliance profiles, respiratory, and hemodynamic variables were recorded continuously during the study protocol. EIT measurements were performed every 10 minutes for a duration of 2 minutes.
Sample size calculation and statistical evaluation
No data are available concerning the variance of frequencies of compliance profiles. Therefore, we based our sample size calculation on estimation of a general standardized effect size e, being the quotient of differences in means and standard deviation. With regard to our approach, which adapted PEEP according to the measured compliance profile, we assumed a large effect size and therefore chose e = 0.8 . In regard to the trial design (unpaired test conditions) and an assumed e of 0.8, 50 patients were required to reach a test power of 0.8 with a desired level of significance of 0.05.
To compensate for potentially incomplete data sets, 60 patients were recruited. Data are presented as means (standard deviation). Differences between the two groups were assessed with the unpaired Students t-test. Statistical significance was considered for p < 0.05. Shapiro–Wilk tests were used to confirm that the assumed normal distribution could not be rejected. For data not normally distributed, differences between the two groups were assessed with Mann–Whitney U tests.