Chi et al.

Annals of Intensive Care (2022) 12:22

https://doi.org/10.1186/s13613-022-00995-w

RESEARCH Open Access

Prevalence and prognosis of respiratory

pendelluft phenomenon in mechanically
ventilated ICU patients with acute respiratory
failure: a retrospective cohort study
Yi Chi1, Zhanqi Zhao2,3, Inéz Frerichs4, Yun Long1* and Huaiwu He1*

Abstract
Background: Respiratory pendelluft phenomenon, defined as intrapulmonary gas redistribution caused by asyn-
chronous alveolar ventilation, could be potentially harmful by inducing lung injury. The aim of the present study was
to investigate its prevalence and prognosis in intensive care unit (ICU) patients with acute respiratory failure (ARF).
Methods: This was a retrospective observational study on 200 mechanically ventilated ARF patients treated in a ter-
tiary ICU. The presence of pendelluft was determined using electrical impedance tomography (EIT) within 48 h after
admission. Its amplitude was defined as the impedance difference between the sum of all regional tidal impedance
variation and the global tidal impedance variation. A value above 2.5% (the 95th percentile from 30 healthy volun-
teers) was considered confirmative for its occurrence.
Results: Pendelluft was found in 61 patients (39 in 94 patients with spontaneous breathing, 22 in 106 receiving
controlled ventilation), with an overall prevalence of 31%. Existence of spontaneous breathing and higher global
inhomogeneity index were associated with pendelluft. Patients with pendelluft had a longer ICU length of stay [10 (6,
14) vs. 7 (4, 11) days; median (lower, upper quartile); p = 0.022] and shorter 14-day ventilator-free days [8 (1, 10) vs. 10
(6, 12) days; p = 0.015]. Subgroup survival analysis suggested the association between pendelluft and longer ventila-
tion duration, which was significant only in patients with ­PaO2/FiO2 ratio below 200 mmHg (log-rank p = 0.042). ICU
mortality did not differ between the patients with and without pendelluft.
Conclusions: Respiratory pendelluft occurred often in our study group and it was associated with longer ventilation
duration. Early recognition of this phenomenon should trigger interventions aimed at alleviating pendelluft.
Keywords: Pendelluft, Mechanical ventilation, Acute respiratory failure, Intensive care unit

Introduction chest [1], obstructive lung disease [2], and acute respira-
Respiratory pendelluft is the phenomenon of intrapulmo- tory distress syndrome (ARDS) [3]. Evidence from ani-
nary gas redistribution caused by asynchronous alveo- mal experiments has suggested that pendelluft could be
lar ventilation. It has been spotted in patients with flail potentially harmful by inducing local overdistension and
tidal recruitment [4–6]. Hence, the early recognition of
pendelluft is warranted for timely adjustment of treat-
*Correspondence: iculong_yun@163.com; hehuaiwu@pumch.cn
1
State Key Laboratory of Complex Severe and Rare Disease, Department
ment and ventilation strategy, especially in critically ill
of Critical Care Medicine, Peking Union Medical College Hospital, patients. Earlier techniques (positron imaging, multi-
Peking Union Medical College, Chinese Academy of Medical Sciences, 1 channel lung sound analysis, darkfield microscopy, etc.)
shuaifuyuan, Dongcheng District, Beijing, China
Full list of author information is available at the end of the article

© The Author(s) 2022. Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which
permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the
original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or
other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line
to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory
regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this
licence, visit http://creativecommons.org/licenses/by/4.0/.
Chi et al. Annals of Intensive Care (2022) 12:22 Page 2 of 9

used to monitor pendelluft are not suitable for critically within the measurement plane and then calculates the
ill patients. regional ventilation map by subtracting the inspiration-
The chest electrical impedance tomography (EIT) is a begin from the expiration-begin image as well as the
non-invasive monitoring technique that can obtain real- global tidal signal variation.
time images of regional lung ventilation at the bedside by To establish a reference value, EIT measurements were
detecting the bio-impedance changes during consecutive performed in 30 healthy volunteers without any under-
respiratory cycles [7]. The application of EIT has enabled lying lung disease (demographics are summarized in
bedside detection of respiratory pendelluft in patients in Additional file 1: Table S1). The following EIT-based
the intensive care unit (ICU). It allows not only qualita- parameters were calculated in both patients and healthy
tive, but also quantitative analysis of pendelluft [8, 9]. subjects:
Despite the progress, little is known about the epide-
miology of pendelluft and its association with clinical Pendelluft
outcome. A recent published EIT-based algorithm for pendelluft
The primary objective of this retrospective study was detection [9] was adopted by our study. According to the
to explore the prevalence and prognosis of pendelluft in theory of pendelluft proposed by Otis et al. [11], when
mechanically ventilated ICU patients with acute respira- pendelluft occurs, the sum of the regional tidal volumes is
tory failure (ARF). greater than the overall tidal volume, their difference rep-
resenting the pendelluft volume. Similarly, the EIT-based
Methods pendelluft amplitude was defined as the impedance dif-
From January 2020 to November 2021, ICU patients ference between the sum of all regional tidal impedance
with ARF, defined by the ratio of partial pressure of arte- variation (TIV) and the global TIV (Additional file 1: Fig.
rial oxygen to fraction of inspired oxygen ­(PaO2/FiO2) S1). Since this pixel-based algorithm is so sensitive, the
below 300 mmHg within 48 h of admission, and receiving occurrence of pendelluft was considered only when its
mechanical ventilation were eligible for study inclusion. amplitude exceeded 2.5% of global TIV (which was the
The exclusion criteria were: age < 18 years, pregnancy, 95th percentile from 30 healthy volunteers) (Additional
body mass index over 50 kg/m2, ribcage malformation, file 1: Table S1).
and any contraindication to the use of EIT (automatic
implantable cardioverter defibrillator, chest wounds lim- Ventilation defect score
iting electrode belt placement, implantable pumps, etc.). Its calculation was based on a semi-quantitative method
This retrospective study was approved by the Institu- for analyzing heterogeneity of ventilation distribution
tional Research and Ethics Committee of Peking Union validated by a previous study [12]. The global ventila-
Medical College Hospital (S-K1859). tion map was separated into four non-overlapping quad-
The following parameters were documented: age, sex, rants of equal size to trace gas distribution into different
predicted body weight (PBW), Acute Physiology and regions of interest (ROIs): lower left (LL), lower right
Chronic Health Evaluation II (APACHE II) score, arte- (LR), upper left (UL) and upper right (UR). Distribution
rial blood gas and respiratory parameters at the time defects in each quadrant were scored as follows: 0 (quad-
point of EIT recording [tidal volume, respiratory rate, rant distribution% ≥ 15%), 1 (15% > quadrant distribu-
positive end-expiratory pressure (PEEP), existence of tion% ≥ 10%) and 2 (quadrant distribution% < 10%). The
spontaneous breathing (defined as EIT-based respira- total ventilation defect score was the sum of the scores of
tory rate higher than set respiratory rate under con- each quadrant (Additional file 1: Fig. S2).
trolled ventilation)], ventilatory ratio (minute ventilation
[mL/min] × ­PaCO2 [mmHg])/(PBW [kg] × 100 [mL/ Global inhomogeneity (GI) index
min] × 37.5 [mmHg]) [10], outcome measures such as The GI index was designed to describe the overall degree
ICU length of stay, 14-day ventilator-free days (VFD) and of spatial heterogeneity of ventilation [13]. A smaller
ICU mortality. global inhomogeneity index represents a more homoge-
neous distribution.
EIT‑based measurements
EIT measurements were performed within 24 h of proven Statistical analysis
ARF diagnosis with PulmoVista 500 (Dräger Medical, Descriptive data are expressed as numbers and per-
Lübeck, Germany). An EIT belt with 16 electrodes was centages for categorical variables and median (lower,
placed around the patient’s thorax at the 4–5th intercos- upper quartile) for continuous variables. Categori-
tal space level. EIT measures changes in voltages between cal variables were compared using the Pearson Chi-
electrode pairs, reconstructs the impedance changes square test, whereas continuous variables distributed
Chi et al. Annals of Intensive Care (2022) 12:22 Page 3 of 9

nonparametrically between groups were compared using Higher proportion of higher ventilation defect score
the Mann–Whitney U test. Pendelluft amplitudes among was seen in patients with pendelluft (p = 0.042). Patients
groups of different defect scores were compared using with higher defect score also had larger pendelluft ampli-
the Kruskal–Wallis H test. tude [1.2 (0.5, 2.5) versus 1.5 (0.7, 3.8) versus 2.1 (0.9,
To evaluate independent factors associated with pen- 6.8) in % of global TIV for defect score 0, 1, ≥ 2, respec-
delluft, significant univariate risk factors or variables con- tively; Kruskal–Wallis p = 0.011] (Additional file 1: Fig.
sidered clinically relevant to pendelluft were examined S3). Subsequent multiple comparisons showed the differ-
using backward stepwise multivariable logistic regres- ence was significant only between defect score 0 and ≥ 2
sion analysis. To avoid overfitting, a maximal number of (p = 0.008).
six variables in the pendelluft model was entered in view
of the 61 events observed (APACHE II score, P ­ aO2/FiO2
ratio, PEEP, spontaneous breathing, respiratory rate and Outcome
GI index). The linear relationship of PEEP and pendelluft The outcome of patients with or without pendelluft is
amplitude was explored with Spearman correlation coef- displayed in Tables 1 and 3 and Figs. 2 and 3. ICU mortal-
ficient. To analyze the relationship between pendelluft ity was 12% for the whole study population and did not
and ventilation duration, cause-specific Cox proportional differ between the groups (p = 0.151). ICU length of stay
hazard models were implemented to predict the relative was longer [10 (6, 14) versus 7 (4, 11) days; p = 0.022] and
hazard of successful discontinuation from ventilator with 14-day ventilator-free days was shorter [8 (1, 10) versus
95% confidence intervals; the model was adjusted for 10 (6, 12) days; p = 0.015] in patients with pendelluft.
APACHE II score and ­PaO2/FiO2 ratio. Restricted cubic Survival analysis revealed marginally non-significant
spline was used to explore the possible nonlinear rela- effect of pendelluft on discontinuation from ventilation
tionship between pendelluft amplitude and the relative within 14 days in the overall study population (log-rank
hazard of discontinuation from ventilator. P values of less p = 0.066). When the study population was divided into
than 0.05 were considered statistically significant. Sta- two subgroups according to ­ PaO2/FiO2 ratio, pendel-
tistical analyses were performed using SPSS 25.0 (SPSS, luft was associated with significantly longer 14-day ven-
Chicago, IL) and R version 4.0.3. tilation duration in patients with P ­ aO2/FiO2 ratio below
200 mmHg (log-rank p = 0.042) while it had no effect
on 14-day ventilation duration in patients with ­PaO2/
Results FiO2 ratio above 200 mmHg (log-rank p = 0.930). Cox
Risk factors for pendelluft regression also identified pendelluft as an independent
A total of 200 patients (135 men and 65 women) were risk factor for longer 14-day ventilation duration, after
included, with a median age of 62 years and a median being adjusted by APACHE II score and P ­ aO2/FiO2 ratio
APACHE II score of 16 at 24 h of ICU admission. (Table 3). Higher pendelluft amplitude was associated
Patients received mechanical ventilation with a median with lower likelihood of discontinuing from mechanical
­PaO2/FiO2 ratio of 200 (164, 246) mmHg, tidal volume of ventilation within 14 days, taken pendelluft amplitude
428 (396, 501) mL, PEEP of 7 (5, 8) ­cmH2O, and respira- 2.5% as a reference (Fig. 3).
tory rate of 17 (15, 19) cycles/minute at the time point Outcome analyses were also performed in patients with
of EIT recording (Table 1). 94 (47%) of them had spon- and without spontaneous breathing separately (Addi-
taneous breathing. Pendelluft was detected in 61 of the tional file 1: Fig. S4). In either group, pendelluft was not
200 ARF patients (prevalence of 31%), 39 of 94 sponta- associated with longer 14-day ventilation duration, but
neously breathing patients (41%) and 22 of 106 patients further investigation into patients with spontaneous
receiving fully controlled ventilation (21%). Patients breathing and simultaneously P/F ratio below 200 mmHg
with pendelluft had a higher proportion of spontaneous showed a marginal effect of pendelluft with longer 14-day
breathing, respiratory rate and higher GI index (Table 1). ventilation duration (log rank p = 0.081).
Tidal volume, PEEP, P ­ aCO2 and ventilatory ratio did not
differ between the groups. Multivariable logistic regres-
sion analysis identified that the existence of spontane- Discussion
ous breathing and higher GI index were associated with The main findings of our study were that: (1) pendelluft
pendelluft (Table 2). The relationship between PEEP and was detected in 31% of 200 ARF patients ventilated in the
pendelluft amplitude was explored in one subgroup of ICU; (2) higher GI index and the existence of spontane-
patients with P/F ratio below 150 mmHg (39 patients; ous breathing were the independent factors associated
Spearman r = − 0.37, p = 0.02) and the other subgroup of with pendelluft; (3) pendelluft was associated with longer
spontaneous breathing patients (94 patients; Spearman 14-day ventilation duration among patients with ­PaO2/
r = − 0.22, p = 0.03) (Fig. 1). FiO2 ratio below 200 mmHg.
Chi et al. Annals of Intensive Care (2022) 12:22 Page 4 of 9

Table 1 Clinical characteristics and outcomes of patients having acute respiratory failure with or without pendelluft
All patients No pendelluft Pendelluft p value
(n = 200) (n = 139) (n = 61)

Age, years 62 (51, 69) 62 (51, 68) 62 (50, 73) 0.549

Female gender, n (%) 65 (32%) 44 (32%) 21 (34%) 0.825
Predicted body weight, kg 64.0 (56.2, 68.5) 64.0 (54.1, 68.5) 64.9 (56.8, 69.4) 0.290
APACHE II score 16 (13, 20) 16 (13, 20) 17 (14, 23) 0.153
Respiratory parameters
Spontaneous breathing, n (%) 94 (47%) 55 (40%) 39 (64%) 0.002
Tidal volume, mL 428 (396, 501) 430 (398, 500) 421 (390, 510) 0.810
Respiratory rate, ­minute−1 17 (15, 19) 16 (15, 18) 18 (16, 21) 0.002
PEEP, ­cmH2O 7 (5, 8) 7 (5, 8) 6 (5, 8) 0.152
Arterial blood gas
 ­PaO2/FiO2 ratio, mmHg 200 (164, 246) 204 (167, 249) 194 (151, 239) 0.225
pH 7.4 (7.4, 7.5) 7.4 (7.4, 7.5) 7.4 (7.4, 7.5) 0.763
 ­PaCO2, mmHg 40.8 (37.0, 43.9) 41.0 (37.6, 44.0) 39.4 (36.0, 42.4) 0.189
Ventilatory ratio 1.3 (1.2, 1.5) 1.3 (1.2, 1.5) 1.4 (1.2, 1.6) 0.170
EIT parameters
Dorsal ventilation, % 41 (32, 47) 41 (32, 48) 41 (32, 47) 0.984
Defect Score, n (%) 0.042
  0 88 (44%) 68 (49%) 20 (33%)
  1 61 (30%) 42 (30%) 19 (31%)
  ≥ 2 51 (26%) 29 (21%) 22 (36%)
  GI index 0.36 (0.34, 0.38) 0.36 (0.34, 0.37) 0.37 (0.34, 0.39) 0.023
  Pendelluft amplitude, % 1.4 (0.7, 3.5) 0.9 (0.5, 1.6) 6.4 (3.9, 13.3) < 0.001
Outcome
MV duration, days 5 (3, 8) 4 (2, 7) 6 (3, 8) 0.052
ICU length of stay, days 7 (5, 13) 7 (4, 11) 10 (6, 14) 0.022
14-day ventilator-free days 9 (5, 11) 10 (6, 12) 8 (1, 10) 0.015
ICU mortality, n (%) 23 (12%) 13 (9%) 10 (16%) 0.151
Data are presented as median (lower, upper quartile) unless otherwise specified
APACHE Acute Physiology and Chronic Health Evaluation, PEEP positive end-expiratory pressure, EIT electrical impedance tomography, GI global inhomogeneity, MV
mechanical ventilation, ICU intensive care unit

Table 2 Univariate and multivariate logistic regression analysis for pendelluft

Variables Univariate logistic regression Multivariate logistic regression
Odds ratio (95% CI) p Odds ratio (95% CI) p

APACHE II score 1.043 (0.994–1.096) 0.086 1.053 (0.999–1.111) 0.052

PaO2/FiO2 ratio 0.997 (0.991–1.002) 0.211 0.995 (0.989–1.002) 0.164
PEEP 0.930 (0.792–1.082) 0.357 0.944 (0.789–1.123) 0.520
Spontaneous breathing 2.707 (1.463–5.112) 0.002 2.375 (1.175–4.883) 0.017
Respiratory rate 1.133 (1.044–1.235) 0.003 1.070 (0.975–1.178) 0.157
GI index (per 0.01 increase) 1.153 (1.049–1.275) 0.004 1.141 (1.030–1.272) 0.014
APACHE Acute Physiology and Chronic Health Evaluation, PEEP positive end-expiratory pressure; GI global inhomogeneity

Definition and prevalence of pendelluft [3], EIT has been increasingly used to detect the pendel-
Since the first report of “occult pendelluft” phenomenon luft in critically ill patients [14–17]. At least two studies
in anesthetized pigs with acute lung injury and a patient tried to quantitatively assess the gas volume subjected
with acute respiratory distress syndrome (ARDS) in 2013 to pendelluft and moving within the lungs through
Chi et al. Annals of Intensive Care (2022) 12:22 Page 5 of 9

Fig. 1 Exploration of correlation between PEEP and pendelluft amplitude in A P/F ratio below 150 mmHg and B presence of spontaneous
breathing, respectively

Table 3 Univariate and multivariate Cox regression analysis for discontinuation from ventilation at Day 14 among patients with ­PaO2/
FiO2 ratio < 200 mmHg
Variables Univariate Cox regression Multivariate Cox regression
Hazard ratio (95% CI) p Hazard ratio (95% CI) p

Pendelluft 0.592 (0.354–0.991) 0.046 0.562 (0.334–0.946) 0.030

APACHE II score 0.930 (0.891–0.970) 0.001 0.922 (0.882–0.964) < 0.001
PaO2/FiO2 ratio 1.008 (1.002–1.015) 0.016 1.009 (1.002–1.016) 0.015
PEEP 1.020 (0.980–1.122) 0.684 Not included
APACHE Acute Physiology and Chronic Health Evaluation, PEEP positive end-expiratory pressure, PaO2 arterial partial pressure of oxygen, FiO2 fraction of inspired
oxygen

Fig. 2 Kaplan–Meier 14-day probability of discontinuation from ventilation curve for patients with (blue) or without pendelluft (red) in A the overall
study population, B in patients with P
­ aO2/FiO2 ratio below 200 mmHg and C between 200 and 300 mmHg

EIT-based algorithms. Coppadoro et al. [8] defined the spontaneous breathing test. Sang et al. [9] introduced a
increased regional impedance from four quadrant ROIs method to detect the amplitude of pendelluft by com-
during the global expiratory phase and vice versa as paring the sum of all pixel TIV with the global TIV, and
pendelluft. They reported a median pendelluft volume expressed it as percent, where 1% of pendelluft amplitude
of 3.3 (2.1, 8.8) mL in 20 patients who had just failed a was equal to 1 mL pendelluft volume per 100 mL tidal
Chi et al. Annals of Intensive Care (2022) 12:22 Page 6 of 9

the ventilation defect score [12], a semi-quantitative

parameter to describe the severity of uneven gas distribu-
tion to four quadrant ROIs, ranging from 0 to 6. Higher
scores, reflecting higher heterogeneity in gas distribution,
were associated with larger amplitudes of pendelluft. The
other was the GI index [13], widely used in the evaluation
of lung recruitment, PEEP titration and weaning process.
The calculation of the GI index was based on the devia-
tion of each pixel tidal impedance variation. Higher GI
values denoted higher degree of lung heterogeneity. Our
study recognized GI index as an independent factor asso-
ciated with pendelluft.
Spontaneous breathing effort during mechanical ven-
tilation might improve gas exchange and lung aeration,
but excessive effort could also cause uneven distribu-
tion of intrathoracic pressure in already injured lung,
which was proposed as another mechanism eliciting
Fig. 3 Relationship between pendelluft amplitude and risk ratio for
ventilation discontinuation in the study population. 2.5% amplitude
respiratory pendelluft [3]. Previous studies noticed the
of pendelluft was set as the reference disappearance of pendelluft in ARDS patients when neu-
romuscular blockers were applied [3, 18], supporting the
association between spontaneous effort and pendelluft.
In our study population, pendelluft was more likely to
volume. The latter algorithm was closer to the original occur in patients with spontaneous breathing, but there
theory of pendelluft proposed by Otis et al. [11] and was was no direct evaluation of breathing effort. Respiratory
adopted in our study. According to Otis et al., pendelluft rate was an indirect indicator reflecting the breathing
could occur where heterogeneity of respiratory time con- effort [19, 20]. It was higher in patients with pendelluft
stants (compliance * resistance) existed between adjacent but without statistical significance after multivariable
alveoli. The time shift due to heterogeneity of time con- regression. Relatively low respiratory frequency (below
stants within the lungs could be assessed by the present 21 cycles/minute in 75% of spontaneously breathing
pendelluft evaluation. On the other hand, heterogeneous patients) might obscure its effect. Careful monitoring of
time constants may also result in a so-called “regional spontaneous breathing effort (e.g., P0.1, pressure muscle
ventilation delay”: heterogeneous regional inflation as index, negative deflection of pressure during end-expir-
compared to the global due to collapse of alveoli and/or atory occlusion, etc.) and its association with pendelluft
airways without time shift at the end of inspiration. Such amplitude is needed. It should be also noted that zero
regional ventilation delay was not captured by our cal- spontaneous effort did not exclude the possibility of pen-
culation. Although the definition of pendelluft was rela- delluft, as was seen in around 21% of ARF patients with-
tively clear, the problem was, the EIT-based algorithm out presence of spontaneous breathing and was proven
was so sensitive to detect a small amount of pendelluft, by dynamic computed tomography in an experimental
possibly without pathological significance, in healthy study conducted on a swine model of mild acute respira-
adults without underlying lung diseases. Therefore, we tory distress syndrome [18].
investigated the pendelluft amplitude in 30 healthy vol- Some studies found that applying higher PEEP
unteers and then set the 95th percentile, i.e., 2.5% as a could alleviate pendelluft in ARDS [4, 5]. As pendel-
threshold, only above which the occurrence of clinically luft was mainly associated with lung heterogeneity
significant pendelluft was considered. Based on that defi- and dynamic pleural pressure variations, higher PEEP
nition, we were able to report the incidence of pendel- may reduce the magnitude of pendelluft by lowering
luft for the first time among ventilated ICU patients with the level of spontaneous effort via neuromechanical
ARF. uncoupling and by reducing atelectasis. Both mecha-
nisms promote a more homogeneous lung expansion
Risk factors for pendelluft
[21]. Opposite evidence also existed showing PEEP
Respiratory time constants inequality, also interpreted as had no effect on pendelluft [22], but the ventilation
alveolar heterogeneity, is the basis of respiratory pendel- mode, baseline P/F ratio and calculation of pendel-
luft. In the present study, we used two easily accessible luft in the study were all different from previous ones.
parameters to describe the lung heterogeneity. One was Hence, we made subgroup analysis and found a week
Chi et al. Annals of Intensive Care (2022) 12:22 Page 7 of 9

but significant negative correlation between PEEP and Study limitations

pendelluft amplitude in the more hypoxemic popula- Our study has some limitations. First, the overall study
tion and patients with spontaneous breathing. The dif- population was heterogeneous, with both assisted and
ferent results among studies could be partly explained controlled ventilation. The subjects had only mild-to-
by the disease severity and whether spontaneous moderate impaired oxygenation on average, which might
breathing was present. weaken the effect of pendelluft on clinical outcome. That
was also the reason why we chose 14-day rather than
28-day ventilator-free day as the primary mechanical
Clinical implication and prognosis of pendelluft ventilation-related outcome. Second, the out-of-phase
Pendelluft has the potential to cause lung injury as impedance change generated by diaphragm movement
it could increase local lung stress and cause regional [23] or pleural effusion [24] could be mistaken as pendel-
overdistension even under protective ventilator set- luft. Since an upward shift of diaphragm was common in
tings. Previous animal experiments suggested that patients with obesity or increased intra-abdominal pres-
pendelluft was associated with tidal recruitment, and sure, electrode belts were placed at a higher position in
that effort-dependent lung injury occurred in the same these patients (3–4th intercostal space) according to their
region where pendelluft appeared [4–6]. Our study chest X-ray or lung ultrasound findings to avoid the sig-
revealed for the first time that pendelluft was asso- nal interference from the moving diaphragm. Future EIT
ciated with longer duration of mechanical ventila- algorithm needs be updated to distinguish lung regions
tion among ICU patients with P ­ aO2/FiO2 ratio below with present diaphragm movement or pleural effu-
200 mmHg, after APACHE II score and P ­ aO2/FiO2 sion. Third, the respiratory management was probably
ratio was adjusted. The effect of pendelluft on venti- influenced by EIT results in some cases, introducing a
lation duration was dependent on the severity of ARF potential bias when assessing the impact of pendelluft on
as similar effect was not seen in mild impaired oxy- patient outcome.
genation. Results from subgroup analysis according
to whether spontaneous breathing was present might Conclusions
suggest different clinical impact of pendelluft in active In conclusion, pendelluft was identified in 31% of a sin-
or passive condition (Additional file 1: Fig. S4). We gle-center ARF patients ventilated in the ICU. Pendelluft
supposed that pendelluft associated with spontaneous occurred more often in cases with spontaneous breath-
effort in patients with moderate-to-severe impaired ing and higher lung heterogeneity. It was associated with
oxygenation could be an injurious ventilation pat- longer ventilation duration in patients with P­ aO2/FiO2
tern that possibly lengthen ventilation duration, while ratio below 200 mmHg. Careful monitoring and thera-
pendelluft in passive condition might only imply lung pies aimed at alleviating pendelluft should be tested in
inhomogeneity but no direct evidence to lung injury. patients with severe hypoxia in the future.
Results from analysis of restricted cubic spline sug-
gested that pendelluft amplitude below the reference
Abbreviations
of 2.5% had an unclear influence on the probability of APACHE: Acute Physiology and Chronic Health Evaluation; ARDS: Acute
successful discontinuation from mechanical ventila- respiratory distress syndrome; ARF: Acute respiratory failure; CI: Confidence
tion, while higher amplitude of pendelluft above the interval; EIT: Electrical impedance tomography; FiO2: Fraction of inspiratory
oxygen; GI: Global inhomogeneity; ICU: Intensive care unit; LL: Lower left; LR:
threshold was associated with prolonged ventilation Lower right; MV: Mechanical ventilation; PBW: Predicted body weight; PaO2:
duration. The optimal threshold of pendelluft ampli- Partial pressure of arterial oxygen; PEEP: Positive end-expiratory pressure; ROI:
tude or volume for predicting clinical outcome war- Region of interest; TIV: Tidal impedance variation; UL: Upper left; UR: Upper
right; VFD: Ventilator-free day.
rants further investigation.
We also explored the relationship between pendel-
Supplementary Information
luft amplitude and ventilatory ratio, a variable reflect-
The online version contains supplementary material available at https://​doi.​
ing ventilation efficiency of the lung [10]. A positive org/​10.​1186/​s13613-​022-​00995-w.
correlation was hypothesized because the pendelluft
gas moving within the lung was not expected to con- Additional file 1: Figure S1. Schematic diagram of EIT-measured
tribute to gas exchange, possibly resulting in reduc- pendelluft amplitude. Pixel 1 and 2 are impedance-time curves from two
tion of ventilation efficiency. However, our results did representative pixels with large ventilation shift. The EIT-based pendelluft
amplitude is calculated as the impedance difference between the sum of
not support the hypothesis. The effect of pendelluft all pixel TIV and the global TIV. TIV, tidal impedance variation. A.U., arbitrary
on ventilation efficiency might have been too weak or unit. Figure S2. Schematic diagram of ventilation defect score. Figure S3.
masked by confounders. Relationship between ventilation defect score and pendelluft amplitude.
Figure S4. Survival analyses performed in patients with and without
Chi et al. Annals of Intensive Care (2022) 12:22 Page 8 of 9

during mechanical ventilation: maximal injury with less positive end-

spontaneous breathing separately. Kaplan–Meier 14-day probability of expiratory pressure. Crit Care Med. 2016;44(8):e678-688.
discontinuation from ventilation curve for patients with (blue) or without 5. Morais CCA, Koyama Y, Yoshida T, Plens GM, Gomes S, Lima CAS, Ramos
pendelluft (red) in the patients A with spontaneous breathing, B absence OPS, Pereira SM, Kawaguchi N, Yamamoto H, et al. High positive end-
of spontaneous breathing, C spontaneous breathing and P/F ratio below expiratory pressure renders spontaneous effort noninjurious. Am J
200 mmHg and D absence of spontaneous breathing and P/F ratio below Respir Crit Care Med. 2018;197(10):1285–96.
200 mmHg 6. Yoshida T, Nakahashi S, Nakamura MAM, Koyama Y, Roldan R, Torsani
V, De Santis RR, Gomes S, Uchiyama A, Amato MBP, et al. Volume-
controlled ventilation does not prevent injurious inflation during
Acknowledgements
spontaneous effort. Am J Respir Crit Care Med. 2017;196(5):590–601.
Not applicable.
7. Frerichs I, Amato MB, van Kaam AH, Tingay DG, Zhao Z, Grychtol B,
Bodenstein M, Gagnon H, Böhm SH, Teschner E, et al. Chest electrical
Authors’ contributions
impedance tomography examination, data analysis, terminology, clini-
YC, HH and YL designed and planned the study. YC and HH were responsible
cal use and recommendations: consensus statement of the TRansla-
for collection and assembly of data. YC, ZZ and HH was responsible for data
tional EIT developmeNt stuDy group. Thorax. 2017;72(1):83–93.
analysis and interpretation. ZZ developed the software for the GI index and
8. Coppadoro A, Grassi A, Giovannoni C, Rabboni F, Eronia N, Bronco A,
pendelluft analysis. YC, HH, ZZ, IF wrote the initial manuscript draft, and all
Foti G, Fumagalli R, Bellani G. Occurrence of pendelluft under pressure
authors were involved in critical revision of the final manuscript. All authors
support ventilation in patients who failed a spontaneous breathing
read and approved the final manuscript.
trial: an observational study. Ann Intensive Care. 2020;10(1):39.
9. Sang L, Zhao Z, Yun PJ, Frerichs I, Möller K, Fu F, Liu X, Zhong N, Li
Funding
Y. Qualitative and quantitative assessment of pendelluft: a simple
This study was supported by Capital’s Funds for Health Improvement and
method based on electrical impedance tomography. Ann Transl Med.
Research (No. 2020-2-40111) and Excellence Program of Key Clinical Specialty
2020;8(19):1216.
of Beijing in 2020 for Critical Care Medicine, Beijing Municipal Science and
10. Sinha P, Calfee CS, Beitler JR, Soni N, Ho K, Matthay MA, Kallet RH.
Technology Commission (No. Z201100005520051).
Physiologic analysis and clinical performance of the ventilatory ratio
in acute respiratory distress syndrome. Am J Respir Crit Care Med.
Availability of data and materials
2019;199(3):333–41.
The datasets used or analyzed in the study are available from the correspond-
11. Otis AB, McKerrow CB, Bartlett RA, Mead J, McIlroy MB, Selver-Stone NJ,
ing author on reasonable request.
Radford EP Jr. Mechanical factors in distribution of pulmonary ventila-
tion. J Appl Physiol. 1956;8(4):427–43.
Declarations 12. He H, Chi Y, Long Y, Yuan S, Zhang R, Yang Y, Frerichs I, Möller K, Fu F,
Zhao Z. Three broad classifications of acute respiratory failure etiolo-
Ethics approval and consent to participate gies based on regional ventilation and perfusion by electrical imped-
The ethics review board of Peking Union Medical Collage Hospital approved ance tomography: a hypothesis-generating study. Ann Intensive Care.
the study protocol. Informed consent was not applicable for the retrospective 2021;11(1):134.
study. 13. Zhao Z, Möller K, Steinmann D, Frerichs I, Guttmann J. Evaluation of
an electrical impedance tomography-based Global Inhomogene-
Consent for publication ity Index for pulmonary ventilation distribution. Intensive Care Med.
Not applicable. 2009;35(11):1900–6.
14. Lopes FA, Souza LAM, Bernardi JTN, Rocha CE, Figueiredo LC, Agostini
Competing interests A, Dragosavac D, Faez D. Pendelluft diagnosed from ventilator weaning
The authors declare that they have no competing interests. indexes obtained through bioelectrical impedance tomography: a
case report. Sao Paulo Med J. 2017;135(3):302–8.
Author details 15. Rossi FS, Costa ELV, Iope DDM, Pacce PH, Cestaro C, Braz LZ, Bousso A,
1
State Key Laboratory of Complex Severe and Rare Disease, Department Amato MB. Pendelluft detection using electrical impedance tomog-
of Critical Care Medicine, Peking Union Medical College Hospital, Peking raphy in an infant. Keep those images in mind. Am J Respir Crit Care
Union Medical College, Chinese Academy of Medical Sciences, 1 shuaifuyuan, Med. 2019;200(11):1427–9.
Dongcheng District, Beijing, China. 2 Department of Biomedical Engineer- 16. Gonçalves-Ferri WA, Rossi FS, Costa ELV, Correa L, Iope D, Pacce PD,
ing, Fourth Military Medical University, Xi’an, China. 3 Institute of Technical Martins-Celini F, Bernardes A, Ribeiro M, Amato MBP. Lung recruitment
Medicine, Furtwangen University, VS‑Schwenningen, Germany. 4 Department and pendelluft resolution after less invasive surfactant administration
of Anesthesiology and Intensive Care Medicine, University Medical Center in a preterm infant. Am J Respir Crit Care Med. 2020;202(5):766–9.
of Schleswig-Holstein Campus Kiel, Kiel, Germany. 17. Enokidani Y, Uchiyama A, Yoshida T, Abe R, Yamashita T, Koyama Y,
Fujino Y. Effects of ventilatory settings on pendelluft phenomenon
Received: 10 December 2021 Accepted: 11 February 2022 during mechanical ventilation. Respir Care. 2021;66(1):1–10.
18. Pellegrini M, Hedenstierna G, Larsson AS, Perchiazzi G. Inspiratory
efforts, positive end-expiratory pressure, and external resistances influ-
ence intraparenchymal gas redistribution in mechanically ventilated
injured lungs. Front Physiol. 2020;11: 618640.
19. Tulaimat A, Trick WE. DiapHRaGM: a mnemonic to describe the work
References of breathing in patients with respiratory failure. PLoS ONE. 2017;12(7):
1. Harada K, Saoyama N, Izumi K, Hamaguchi N, Sasaki M, Inoue K. Experi- e0179641.
mental pendulum air in the flail chest. Jpn J Surg. 1983;13(3):219–26. 20. Apigo M, Schechtman J, Dhliwayo N, Al Tameemi M, Gazmuri RJ.
2. Vyshedskiy A, Murphy R. Pendelluft in chronic obstructive lung disease Development of a work of breathing scale and monitoring need of
measured with lung sounds. Pulm Med. 2012;2012: 139395. intubation in COVID-19 pneumonia. Crit Care. 2020;24(1):477.
3. Yoshida T, Torsani V, Gomes S, De Santis RR, Beraldo MA, Costa EL, Tucci 21. Borges JB, Morais CCA, Costa ELV. High PEEP may have reduced
MR, Zin WA, Kavanagh BP, Amato MB. Spontaneous effort causes occult injurious transpulmonary pressure swings in the ROSE trial. Crit Care.
pendelluft during mechanical ventilation. Am J Respir Crit Care Med. 2019;23(1):404.
2013;188(12):1420–7. 22. Santini A, Mauri T, Dalla Corte F, Spinelli E, Pesenti A. Effects of inspira-
4. Yoshida T, Roldan R, Beraldo MA, Torsani V, Gomes S, De Santis RR, tory flow on lung stress, pendelluft, and ventilation heterogeneity in
Costa EL, Tucci MR, Lima RG, Kavanagh BP, et al. Spontaneous effort ARDS: a physiological study. Crit Care. 2019;23(1):369.
Chi et al. Annals of Intensive Care (2022) 12:22 Page 9 of 9

23. Karsten J, Stueber T, Voigt N, Teschner E, Heinze H. Influence of dif-

ferent electrode belt positions on electrical impedance tomography
imaging of regional ventilation: a prospective observational study. Crit
Care. 2016;20:3.
24. Becher T, Bußmeyer M, Lautenschläger I, Schädler D, Weiler N, Frerichs I.
Characteristic pattern of pleural effusion in electrical impedance tomog-
raphy images of critically ill patients. Br J Anaesth. 2018;120(6):1219–28.

Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in pub-
lished maps and institutional affiliations.