Effects of the Prone Position on Respiratory Mechanics

and Gas Exchange during Acute Lung Injury

PAOLO PELOSI, DANIELA TUBIOLO, DANIELE MASCHERONI, PIERLUIGI VICARDI,
STEFANIA CROTTI, FRANCO VALENZA, and LUCIANO GATTINONI
Istituto di Anestesia e Rianimazione, Università degli Studi de Milano and Servizio di Anestesia e Rianimazione, Ospedale Maggiore,
Istituto di Ricovero e Cura a Carattere Scientifico (IRCCS), Milan, Italy

We studied 16 patients with acute lung injury receiving volume-controlled ventilation to assess the
relationships between gas exchange and respiratory mechanics before, during, and after 2 h in the
prone position. We measured the end-expiratory lung volume (EELV, helium dilution), the total respi-
ratory system (Cst,rs), the lung (Cst,L) and the thoracoabdominal cage (Cst,w) compliances (end-
inspiratory occlusion technique and esophageal balloon), the hemodynamics, and gas exchange. In
the prone position, PaO2 increased from 103.2 6 23.8 to 129.3 6 32.9 mm Hg (p , 0.05) without sig-
nificant changes of Cst,rs and EELV. However, Cst,w decreased from 204.8 6 97.4 to 135.9 6 52.5 ml/
cm H2O (p , 0.01) and the decrease was correlated with the oxygenation increase (r 5 0.62, p ,
0.05). Furthermore, the greater the baseline supine Cst,w, the greater its decrease in the prone posi-
tion (r 5 0.82, p , 0.01). Consequently, the oxygenation changes in the prone position were predict-
able from baseline supine Cst,w (r 5 0.80, p , 0.01). Returning to the supine position, Cst,rs
increased compared with baseline (42.3 6 14.4 versus 38.4 6 13.7 ml/cm H2O; p , 0.01), mainly
because of the lung component (57.5 6 25.1 versus 52.4 6 23.3 ml/cm H2O; p , 0.01). Thus, (1)
baseline Cst,w and its changes may play a role in determining the oxygenation response in the prone
position; (2) the prone position improves Cst,rs and Cst,L when the supine position is resumed.
Pelosi P, Tubiolo D, Mascheroni D, Vicardi P, Crotti S, Valenza F, Gattinoni L. Effects of the
prone position on respiratory mechanics and gas exchange during acute lung injury.
AM J RESPIR CRIT CARE MED 1998;157:387–393.

Patients with acute lung injury (ALI) may be characterized by unchanged or deteriorate. In a small sample of patients with
various degrees of hypoxemia, alterations in respiratory me- ALI studied with computed tomography, we previously ob-
chanics, and reduction in end-expiratory lung volume (EELV). served that changes in oxygenation were somewhat related to
The prone position has been proposed as a relatively simple the lung and thoracic cage shape (9).
maneuver to improve oxygenation (1), and an increased In this study we investigated the effects of the prone posi-
EELV was originally suggested as the main mechanism for tion on EELV, lung and chest wall mechanics, and the rela-
oxygenation improvement (2). More recently, experimental tionships between the changes in oxygenation and in respira-
studies in animal model of ALI found that the prone position tory mechanics.
causes a venous admixture reduction and a more even ventila-
tion distribution, without affecting EELV (3, 4). METHODS
In the adult respiratory distress syndrome (ARDS), the Study Population
most severe form of ALI, similar improvement in oxygen-
ation, decrease in venous admixture (5), and more homoge- We studied 16 consecutive patients with ALI. Eleven of the 16 met
the ARDS criteria of the American European Consensus Conference
neous regional lung inflation (6) have been reported. How-
on ARDS (10). None of them had asthma or chronic lung diseases or
ever, no data are available on the EELV change, which is still cardiogenic pulmonary edema. The main clinical characteristics, the
considered a possible mechanism to explain the oxygenation time from onset, the ventilatory settings, and the gas exchange at the
improvement in the prone position (7, 8). Furthermore, oxy- moment of the study, as well as the outcome, are summarized in Table
genation improvement is not a constant finding during prone 1. Thirteen patients were nasotracheally intubated with a cuffed en-
positioning, and no evidence is available so far to explain why dotracheal tube (7.5 to 8 mm inner diameter) and three had tracheos-
the majority of the patients improve, whereas others remain tomies (8 to 10 mm inner diameter). In five of the 16 patients, prone
positioning had been already used before performing the study,
whereas the remaining 11 patients were studied at the first time of
prone positioning.
(Received in original form April 3, 1997 and in revised form July 16, 1997)
Correspondence and requests for reprints should be addressed to Dr. Luciano Ventilatory Setting
Gattinoni, Istituto de Anestesia e Rianimazione, Ospedale Maggiore, IRCCS, via F. All patients were studied while sedated with fentanyl (1.5 to 5.5 mg/kg/h)
Sforza 35, 20122, Milan, Italy. and diazepam (0.03 to 0.18 mg/kg/h), paralyzed with pancuronium
Am J Respir Crit Care Med Vol 157. pp 387–393, 1998 bromide (0.05 to 0.1 mg/kg/h), and ventilated in the volume control
388 AMERICAN JOURNAL OF RESPIRATORY AND CRITICAL CARE MEDICINE VOL 157 1998

TABLE 1
PATIENTS’ CHARACTERISTICS, VENTILATORY SETTING, GAS EXCHANGE,
AND RESPIRATORY MECHANICS AT THE TIME OF THE STUDY

· Outcome of Respiratory Disease

Sex Age Weight Onset VE PEEP PaO2 PaCO2 EELV Cst,rs Cst,L Cst,w
Patient No. (M/F ) (yr) (kg) (days) (L/min) (cm H2O) FIO2 (mm Hg) (mm Hg) (L) (ml/cm H2O) (ml/cm H2O) (ml/cm H2O) (S/D)

1 F 56 60 8.0 7.0 12 0.90 95 73 0.98 46.4 58.2 228.5 S Legionella pneumonia

2* M 67 55 4.0 11.5 5 0.50 109 48 1.36 43.0 71.5 107.8 S Mycoplasma pneumoniae
3* F 79 50 3.0 7.5 12 0.60 166 40 1.10 23.3 27.6 149.7 S Caustic agent inhalation
4 M 23 60 4.0 7.8 18 0.55 70 57 1.28 33.7 45.4 131.5 S Hemorragic alveolitis
5 M 65 85 1.0 13.7 18 0.90 96 46 1.21 42.1 93.1 76.8 S Sepsis
6 M 30 55 8.0 10.5 10 0.55 104 47 0.61 22.4 25.7 178.8 D CMV pneumonia in immunosuppression
7 F 25 65 1.0 7.2 10 0.55 78 45 0.59 21.1 23.6 196.0 S Pneumonia of unknown ethiology
8* M 19 90 18.0 9.9 14 0.50 117 40 1.74 43.2 63.0 137.6 S Multiple trauma, lung contusion
9* F 25 50 3.0 7.0 14 0.40 120 44 1.09 28.1 31.7 249.6 D Legionella pneumonia in immunosuppression
10 M 51 80 1.0 10.8 12 0.90 99 52 1.36 39.4 43.2 443.1 D Pneumonia of unknown ethiology
11* M 53 75 1.0 8.6 10 0.50 125 32 2.17 76.7 99.0 341.4 S Pneumonia of unknown ethiology
12 M 44 67 4.0 10.9 8 1.00 111 66 0.62 27.2 31.5 198.5 D Viral pneumonia in immunosuppression
13 F 62 69 1.0 9.3 15 1.00 64 38 0.87 36.7 45.1 197.2 D Aspergillus pneumonia in immunosuppression
14 M 29 70 2.0 10.6 12 0.60 95 47 1.13 41.8 68.4 107.6 D Viral pneumonia in immunosuppression
15 M 34 70 1.0 11.0 15 0.55 104 35 1.39 51.4 68.6 204.1 S Pneumocystis pneumonia
16 M 62 90 8.0 9.0 12 0.80 98 59 1.26 37.9 42.9 328.0 D Pneumonia of unknown ethiology
Mean 11 M/5 F 45 68 4.3 9.5 12.3 0.68 103 48 1.17 38.4 52.4 204.8 9 S/7 D
SD 19 13 4.5 1.9 3.5 0.20 24 11 0.41 13.7 23.3 97.4
·
Definition of abbreviations: PEEP 5 positive end-expiratory pressure; FIO2 5 fraction of inspired oxygen; V E 5 minute ventilation; EELV 5 end-expiratory lung volume; Cst,rs 5 total
respiratory system static compliance; Cst,L 5 lung static compliance; Cst,w 5 thoracoabdominal cage static compliance.
* Patients with acute lung injury.

mode with constant inspiratory flow. Mechanical ventilation was pro- gas mixture (13% helium in oxygen) was connected to the airway
vided with a Servo Ventilator 900 C (Siemens-Elema, Solna, Sweden). opening previously clamped at end-expiration to maintain the PEEP
The ventilatory setting, positive end-expiratory pressure (PEEP), and level, and 10 deep manual breaths were performed. The helium con-
inspired oxygen fraction (FIO2) were set according to the clinical needs centration in the anesthesia bag was then measured with a helium an-
and kept constant throughout the study. Tidal volume was 0.682 6 alyzer (PK Morgan Ltd, Chatham, Kent, UK), and EELV was com-
0.144 L, respiratory rate was 13.5 6 2.5 breaths/min, TI/Ttot was puted according to the following formula:
0.39 6 0.08, inspiratory flow was 0.386 6 0.077 L/s, PEEP was 12.3 6
3.5 cm H2O, FIO2 was 0.68 6 0.20, and peak inspiratory pressure was EELV = Vi [ He ] i ⁄ [ He ] fin – Vi
37.8 6 6.3 cm H2O. where Vi is the initial gas volume in the anesthesia bag and [He]i and
[He]fin are the initial and final helium concentrations in the bag. The
limitations of this method have been fully discussed elsewhere (12).
Protocol and Measurements Intra-abdominal pressure. Average intra-abdominal pressure was
Change of position was manually performed by four or five attendants measured through a transurethral bladder catheter (13). Using a ster-
(usually three nurses and two doctors). In the prone position the head ile technique, an average of 100 ml of normal saline were infused
was turned laterally and the arms were parallel to the body. A roll un- through the urinary catheter to fill the catheter tubing. The catheter
der the upper part of the chest wall and a pillow under the pelvis were was then clamped distally to the sampling port. A 20-gauge needle
positioned in an effort to minimize restriction of abdominal move- was inserted through the catheter sampling membrane, and the blad-
ments. der catheter pressures were measured using a water manometer ze-
Measurements were obtained in the supine position (baseline) af- roed at the pubis level.
ter 30 and 120 min in the prone position and 30 min after returning to Intra-abdominal pressure was measured only at the baseline, at the
the supine position. In 13 patients additional measurements were ob- end of prone position period, and 120 min after returning to the su-
tained at 120 min after returning to the supine position. pine position. The intra-abdominal pressure data were not collected in
Gas exchange and hemodynamics. All of the patients already had the first four patients.
an arterial and pulmonary artery thermodilution catheter inserted at Respiratory mechanics. Airway pressure (Paw) and gas flow were
the time of the study. Arterial and mixed venous samples were ana- measured by a self-calibrating flow transducer (Varflex; Bicore Moni-
lyzed for gas tensions and pH immediately after sampling (IL BGM, toring System, Irvine, CA) connected at the endotracheal or tracheos-
1312; Instrumentation
. Laboratory,
. Spa, Italy). The right to left intra- tomy tube opening. Esophageal pressure (Pes) was measured by an
pulmonary shunt ( QS/ QT) was computed by the shunt equation, as- esophageal balloon (Bicore Monitoring System) automatically in-
suming a respiratory quotient equal to 1. Mean arterial, pulmonary ar- flated with 0.5 to 1 ml of air and positioned at the lower third of the
tery, pulmonary wedge, and central venous pressures were measured esophagus as shown by chest roentgenograph. All the signals were ac-
with pressure transducers (Transpac IV L974; Abbott Ireland, Sligo, quired on a CP100 Pulmonary Monitor (Bicore Monitoring System),
Republic of Ireland). Zero level was set at the midaxillary line in both converted to digital form and processed separately for subsequent
positions. Cardiac output (CO) was measured in triplicate by the ther- analysis with a dedicated computerized program.
modilution method, and the cardiac index (CI) was computed normal- To measure the compliance of the respiratory system and to parti-
izing the CO for the body surface area. tion it into its pulmonary and chest wall components the occlusion
Physiologic dead space. Expired gases were collected over a 2-min method was used. The end-inspiratory hold button of the Servo 900 C
period in a Douglas bag at the same time as blood gas samples were was pressed for brief (7 to 8 s) airway occlusion (14). Occlusion was
collected. Expired gas was analyzed for mixed expired CO2 concentra- maintained until both Paw and Pes decreased from the maximal value
tion (Model 760 Capnograph; Teledyne Electronic Devices, Bellma- (Pawmax and Pesmax, respectively) to an apparent plateau (Paw2 and
fiok, Mirandola, Italy). The physiologic dead space fraction (VD/VT) Pes2, respectively). Similarly the end-expiratory airway pressure (PEaw)
was computed according to the standard formulas. and the end-expiratory esophageal pressure (PEes) were recorded
End-expiratory lung volume. EELV was measured using a simpli- after a brief end-expiratory hold maneuver. The static respiratory sys-
fied closed-circuit helium dilution method (11) at PEEP during an tem (Cst,rs) and thoracoabdominal (Cst,w) compliances were ob-
end-expiratory pause. An anesthesia bag filled with 1.5 L of a known tained dividing the tidal volume, respectively, by the difference be-
Pelosi, Tubiolo, Mascheroni, et al.: Respiratory Mechanics and Oxygenation in the Prone Position 389

TABLE 2
ARTERIAL BLOOD GAS VALUES AND PHYSIOLOGIC
DEAD SPACE DURING THE STUDY*

Baseline Prone Prone Supine Supine

0 min 30 min 120 min 30 min 120 min

PaO2, mm Hg 103.19 6 23.79 119.06 6 31.81 129.25 6 32.86† 123.13 6 50.02 117.92 6 54.99
PaCO2, mm Hg 48.04 6 11.13 50.23 6 11.76 49.61 6 11.74 49.96 6 12.57 48.21 6 12.64
pHa 7.346 6 0.086 7.337 6 0.067 7.334 6 0.068 7.336 6 0.07 7.348 6 0.065
· ·
QS/QT 0.28 6 0.12 0.27 6 0.11 0.25 6 0.11 0.27 6 0.10 0.28 6 0.11
VD/VT 0.55 6 0.08 0.58 6 0.07 0.55 6 0.08 0.55 6 0.09 0.55 6 0.09
· ·
Definition of abbreviations: QS/ QT 5 right to left intrapulmonary shunt; VD/VT 5 physiologic dead space.
* Values are expressed as mean 6 SD. Data at 120 min after repositioning supine refer to 13 patients.
†
p , 0.05 compared with baseline.

tween Paw2 2 PEaw and Pes2 2 PEes. The static lung compliance RESULTS
(Cst,L) was obtained with the following formula:
Gas Exchange and Hemodynamics
Cst,L = V T ⁄ [ ( Paw 2 – Pes 2 ) – ( P E aw – P E es ) ] .
In this series of patients the prone position resulted in a signif-
One could question the comparison of esophageal balloon mea- icant increase in oxygenation within 120 min (Table 2), even if
surements in the supine and the prone positions. In fact, it is possible there was considerable individual variation in oxygenation re-
that the transmission of the alveolar pressure to the pleural space is sponse in the prone position. Indeed in 12 patients, PaO2 in-
different in the two conditions because of the different relative posi- creased (range, 9 to 73 mm Hg), whereas in four patients it de-
tion of the heart, which moves ventrally on turning prone. However, creased (range, 27 to 216 mm Hg) compared with baseline.
no alternative methods are available in humans, and this method was
When the patients were repositioned supine the oxygenation
considered adequate also in the prone position, and it was adopted to
perform lung-volume curves in both awake (15) and paralyzed sub- was not significantly different from that in the prone position
jects (11, 16). Moreover, if the change of position would result in arte- or that at baseline. Shunt fraction, PaCO2, and dead space did
facts when measuring the esophageal pressure, this would lead to sys- not significantly change when the patients were turned prone
tematic changes in the same direction (false increase or false decrease), and when they were repositioned supine. These responses
but this was not the case in our study (see RESULTS). were similar both in the patients with ALI and in those with
Total resistance of the respiratory system (Rtot,rs)· and of the chest ARDS. We could not find any relationship between the oxy-
· (Pawmax 2 Paw2)/ i and (Pesmax 2
wall (Rtot,w) were computed as V genation response and the time elapsed from disease onset.
·
Pes2)/V i, respectively, where V i is the flow immediately before the As shown in Table 3, mean pulmonary artery pressure,
occlusion. Total lung resistance (Rtot,L) was obtained as the differ-
wedge pressure, and central venous pressure slightly, but sig-
ence between Rtot,rs and Rtot,w. Rtot,rs and Rtot,L include the en-
dotracheal or tracheostomy tube resistance.
nificantly, increased during the prone position and returned to
All respiratory mechanics data (except the EELV) were obtained baseline values after returning to the supine position. How-
as an average of three measurements. ever, since we kept the midaxillary line as a zero reference
level, these minimal changes could be due to a shift of the real
Statistics zero pressure level because of ventral movement of the heart.
All data are expressed as mean 6 1 standard deviation. Comparisons Other hemodynamic variables remained unchanged during
between the different periods were performed using Friedman’s non- the prone position.
parametric analysis of variance. Individual comparison with baseline
were performed with a nonparametric paired t test; Bonferroni’s cor-
Respiratory Mechanics
rection was applied for multiple comparisons. The least-squares method
was used to perform linear regression analysis; p < 0.05 was considered Overall results are shown in Table 4. No significant EELV dif-
statistically significant. ferences were observed between supine and prone positions

TABLE 3
HEMODYNAMIC VALUES DURING THE STUDY*

Baseline Prone Prone Supine Supine

0 min 30 min 120 min 30 min 120 min

HR, beats/min 111 6 24 113 6 21 115 6 23 113 6 19 116 6 14

Pa, mm Hg 79 6 15 82 6 15 84 6 18 82 6 15 83 6 16
Ppa, mm Hg 28 6 7 32 6 7† 32 6 6† 28 6 5 29 6 5
Pw, mm Hg 14 6 4 16 6 4† 17 6 4† 14 6 3 13 6 4
Pcv, mm Hg 11 6 5 12 6 5† 13 6 4† 11 6 4 11 6 4
CI, L/min/m2 4.4 6 1.6 4.4 6 1.4 4.5 6 1.6 4.7 6 1.4 4.7 6 1.2
·
V O2, ml/min/m2 126.9 6 35.4 139.6 6 46.2 139.9 6 48.4 148.6 6 50.4 150.7 6 49.3
TSVR, dyne*s/cm5 746 6 258 759 6 239 763 6 273 723 6 236 710 6 206
PVR, dyne*s/cm5 151 6 50 163 6 41 160 6 50 133 6 42 157 6 45

Definition of abbreviations: HR 5 heart rate; Pa 5 mean arterial pressure; Ppa 5 pulmonary artery pressure; Pw 5 wedge pressure; Pcv 5
·
central venous pressure; CI 5 cardiac index; V O2 5 oxygen consumption; TSVR 5 total systemic vascular resistance; PVR 5 pulmonary vas-
cular resistance.
* Values are expressed as mean 6 SD. Data at 120 min after repositioning supine refer to 13 patients.
†
p , 0.05 compared with baseline.
390 AMERICAN JOURNAL OF RESPIRATORY AND CRITICAL CARE MEDICINE VOL 157 1998

TABLE 4
RESPIRATORY MECHANICS VALUES DURING THE STUDY*

Baseline Prone Prone Supine Supine

0 min 30 min 120 min 30 min 120 min

EELV, L 1.17 6 0.41 1.25 6 0.49 1.29 6 0.57 1.20 6 0.58 1.29 6 0.66
IAP, cm H2O 11.4 6 7.2 — 14.8 6 6.6 — 10.4 6 7.2
Cst,rs, ml/cm H2O 38.4 6 13.7 36.8 6 11.8 35.9 6 10.7 42.3 6 14.4‡§ 43.0 6 15.2†
Cst,L, ml/cm H2O 52.4 6 23.3 55.3 6 26.2 53.9 6 23.6 57.5 6 25.1‡§ 58.5 6 27.3†
Cst,w, ml/cm H2O 204.8 6 97.4 146.8 6 55.5 135.9 6 52.5‡ 219.1 6 100.9 232.0 6 84.0
Rtot,rs, cm H2O/L/s 17.5 6 6.4 17.6 6 4.8 17.6 6 5.7 17.9 6 5.2 16.6 6 4.5
Rtot,L, cm H2O/L/s 15.2 6 6.5 15.1 6 5.3 15.0 6 5.5 15.8 6 5.7 14.5 6 4.5
Rtot,w, cm H2O/L/s 2.3 6 1.8 2.5 6 1.6 2.6 6 1.3 2.1 6 1.4 2.1 6 1.2

Definition of abbreviations: EELV 5 end-expiratory lung volume; IAP 5 intra-abdominal pressure; Cst,rs 5 total respiratory system static
compliance; Cst,L 5 lung static compliance; Cst,w 5 thoracoabdominal cage static compliance; Rtot,rs 5 total resistance of the respiratory
system (including the endotracheal tube); Rtot,L 5 total lung resistance (including the endotracheal tube); Rtot,w 5 total thoracic cage re-
sistance.
* Values are expressed as mean 6 SD. Data at 120 min after repositioning supine refer to 13 patients. Data regarding IAP refer to 12 pa-
tients.
†
p , 0.05 compared with baseline.
‡
p , 0.01 compared with baseline.
§
p , 0.01 compared with prone position.

despite a trend toward increases while prone. Similarly, the in-

tra-abdominal pressure was not significantly affected by the
positional changes. The prone position did not significantly af-
fect Cst,rs and Cst,L (Figure 1), but it resulted in a significant
reduction in Cst,w. Thirty minutes after returning to the su-
pine position, Cst,rs increased compared with either baseline
or prone position values (p , 0.01). This improvement was
mainly due to the lung component (p , 0.01), but the Cst,w
returned to baseline levels. The improvement in total respira-
tory system and lung compliance was consistent in the major-
ity of our patient population. Indeed, in 14 of the 16 patients,
plateau pressure was reduced after returning supine compared
with that at baseline, with a range of 1.3 to 4.2 cm H2O (aver-
age, 2.32 6 0.92 cm H2O) (p , 0.01). This effect also persisted
120 min after returning supine. Airway resistances were not
significantly affected by postural changes.
Oxygenation-Respiratory Mechanics Relationships
The changes in oxygenation induced by the prone position
were not related to changes in EELV or total respiratory sys-
tem or lung compliance. However, the oxygenation changes in
the prone position were predictable from the baseline supine
Cst,w. As shown in Figure 2, higher Cst,w were associated
with greater PaO2 improvements in the prone position (DPaO2 5
232.4 1 0.24 * Cst,r; r 5 0.80, p , 0.01). Moreover, Cst,w
changes in the prone position were correlated with the base-
line Cst,w values (DCst,w 5 88.5 2 0.72 * Cst,w; r 5 0.82, p ,
0.01) (Figure 3).
Furthermore, we found a fair correlation between the early
changes of oxygenation (30 min after prone positioning) and
the changes in the compliance of the thoracoabdominal cage.
(DPaO2 5 3.9 2 0.2 * DCst,w; r 5 0.62, p , 0.05). The greater
the decrease of thoracoabdominal compliance, the greater the
improvement in oxygenation (Figure 4). At 120 min the PaO2
further increased, and the Cst,w further decreased. However,

Figure 1. Total respiratory system, lung and thoracoabdominal

compliance during the study. Data are expressed as mean 6 SD.
Data at 120 min after repositioning supine refer to 13 patients.
*p , 0.05, compared with baseline; **p , 0.01 compared with
baseline; §p , 0.01 compared with prone position.
Pelosi, Tubiolo, Mascheroni, et al.: Respiratory Mechanics and Oxygenation in the Prone Position 391

Figure 2. Relationship between baseline thoracoabdominal cage

compliance (Cst,w) and oxygenation changes (DPaO2) (prone 30 Figure 4. Relationship between changes (prone 30 min minus base-
min minus baseline). DPaO2 5 232.4 1 0.24 * Cst,w; r 5 0.80, line) in the thoracoabdominal cage compliance (DCst,w) and oxy-
p , 0.01. genation (DPaO2). DPaO2 5 3.9 2 0.2 * DCst,w; r 5 0.62, p , 0.05.

the correlation between the thoracoabdominal compliance Supine to Prone Shifting

changes and oxygenation changes was only close to statistical Oxygenation. An increase in EELV was originally proposed
significance (r 5 0.45; p 5 0.07). by Douglas and colleagues (2) to explain the improvement in
oxygenation observed in their pioneering study on the prone
DISCUSSION position in ALI, and it has since been considered a possible
As previously shown by other studies (5, 17, 18), in this series explanation for the improvement in gas exchange associated
of patients we observed a great variability of individual re- with the prone position (7, 8, 17). This hypothesis seems un-
sponses in oxygenation. On average, however, the PaO2 after likely in adult patients with ALI since no significantly EELV
120 min in the prone position was significantly higher than changes were observed during prone positioning, and the
that at baseline. This improvement was not due to EELV EELV variations were not related to the changes in oxygen-
changes. At the same time, the compliance of the thoracoab- ation. The lack of correlation between improvement in oxy-
dominal cage was significantly lower than that at baseline. genation and changes in EELV is in keeping with previous
When returning to the supine position, the total respiratory findings in oleic-acid-injured dogs (3), in normal (16) and in
system and lung compliance values were significantly higher obese (11) anesthetized subjects, and in pediatric patients with
than baseline values. ALI (19). Moreover, in a previous study using computed to-
mography, we did not find any difference in the average slice
densities between the prone and the supine positions, suggest-
ing an unmodified total gas volume (6). However, because of
the relatively small patient population, we cannot safely ex-
clude a possible marginal increase in EELV during prone po-
sitioning; in fact, a trend towards increased EELV was present
in this study and in a previous one on animals (3). Nonethe-
less, this study clearly shows that an increase in global EELV
is not necessary to obtain improved gas exchange in the prone
position.
Therefore, the rise in oxygenation during prone positioning
may only be explained by regional changes of inflation/venti-
lation or perfusion with an unmodified global EELV. Indeed,
data obtained in experimental animal settings attributed the
improvement in oxygenation primarily to a regional variation
of ventilation, which was found more evenly distributed be-
tween the dorsal and the ventral lung regions in the prone po-
sition (20). Moreover, in patients with ALI we found, with the
CT scan, a more homogeneous regional inflation in the prone
position than in the supine position (6), which was actually as-
Figure 3. Relationship between baseline thoracoabdominal cage sociated with better oxygenation (9). In summary, the avail-
compliance (Cst,w) and its modification in the prone position able human and experimental data suggest that the main de-
(prone 30 min minus baseline) (DCst,w). DCst,w 5 88.5 20.72 * terminant of the improvement in oxygenation in the prone
Cst,w; r 5 0.82, p , 0.01. position is the redistribution of the inflation gradient from
392 AMERICAN JOURNAL OF RESPIRATORY AND CRITICAL CARE MEDICINE VOL 157 1998

dorsal to ventral regions (6), resulting in more even regional ated with an unmodified lung volume. An alternative explana-
inflation/ventilation along the vertical axis, in the absence of a tion could be a change of specific compliance.
gravitary perfusion distribution (4). However, the regional as-
sessment of perfusion in patients is technically difficult, and Relationships between Changes in Oxygenation
this hypothesis remains to be proved. and Changes in the Mechanical Properties
Respiratory mechanics. Shifting from the supine to the of the Respiratory System
prone position resulted in a significant reduction in the com- We submit an integrated interpretation of the findings ob-
pliance of the thoracoabdominal cage. This decrease could be tained in this and in other studies of ALI.
explained by a decrease of the compliance of either the rib Paralyzed patients with ALI in the supine position at end-
cage or diaphragmatic component of the chest wall or both. expiration exhibit an inflation gradient (i.e., relative differ-
We can reasonably assume that the diaphragmatic wall com- ence of regional inflation) much greater than that in normal
pliance as a whole (notwithstanding differences in regional subjects; this is due to the increase in lung weight. The pro-
distribution) remained substantially unchanged in the prone gressive decrease of regional inflation along the vertical axis
position because the intra-abdominal pressure was not signifi- results in regional collapse of the most dorsal regions (6, 23).
cantly different in the prone and supine positions. The de- At PEEP levels usually applied when these lungs are insuf-
crease of thoraco-abdominal compliance should therefore be flated, the tidal volume preferentially distributes to the ventral
explained by a greater rigidity of the rib cage component of lung regions, causing higher regional Cst,rs than in the dorsal
the chest wall in the prone position when compared with the regions (22). When shifting from the supine to the prone posi-
supine position. tion, the pattern is almost completely reverted. At end-expira-
Indeed, the compliance of the rib cage component of the tion, the regional inflation decreases from dorsal to ventral,
chest wall is known to be nonhomogeneous, the ventral part and the most dependent ventral regions collapse.
(sternal) presenting a “larger freedom to move,” than the dor- It must be noted, however, that the inflation gradient is de-
sal part (vertebral) (21). The dishomogeneity is further en- creased compared with that in the supine position, i.e., the dif-
hanced in the supine position where movement remains unim- ferences of regional inflation are less pronounced, resulting in
peded in only the ventral part (the dorsal part of the chest lies an overall more homogenous regional inflation. The decrease
on the bed). This is one of the possible mechanisms causing of the inflation gradient in the prone position could be due to
the preferential distribution of ventilation to ventral lung ar- the changed position of the heart, which no longer compresses
eas in supine, paralyzed subjects with ALI (22). In the prone the dorsal lung regions (24, 25). The more even regional infla-
position, the stiffer component of the rib cage (dorsal part) is tion per se could result in better oxygenation. In fact, we pre-
free to move, whereas movement of the more compliant (ven- viously found by CT scan at end-expiration a correlation be-
tral part) becomes impeded by lying on the bed. This results in tween the increased homogeneity of lung inflation induced by
a decrease in net rib cage compliance. However, because of the prone position and the increase of oxygenation (9).
this rearrangement, at least during volume-controlled ventila- In the present study we found, in addition, that the de-
tion, the prone position should partly redistribute the tidal crease of the thoracoabdominal compliance in the prone posi-
ventilation towards the dependent ventral regions, resulting in tion when compared with the supine position was associated
an overall more even regional distribution than that in the su- with improvement of oxygenation. We hypothesize a different
pine position. distribution of tidal volume compared with that in the supine
position as a possible mechanism. Indeed, in the prone posi-
Returning Supine after the Prone Position tion, because of the greater stiffness of the dorsal part of the
Oxygenation. When returning to the supine position oxygen- thoracic cage, the insufflated gases should distribute more to-
ation slightly declined, but no significant differences were ob- wards the ventral and diaphragmatic regions (now dependent)
served when compared with baseline or the prone position. where minimal inflation or collapse are present at end-expira-
However, because of the small population, we cannot exclude tion. The greater the stiffness of the dorsal part of the thoracic
a type II error. Once again, these average values were derived cage, the greater should be insufflation of the less aerated ven-
from variable individual responses. Different oxygenation re- tral regions, and the greater the improvement in oxygenation.
sponses when returning to the supine position have been pre- Although a thorough understanding of the mechanisms un-
viously described, including both a reversal of oxygenation im- derlying the changes in oxygenation also requires knowledge
provement (5) and a maintenance of improved oxygenation of regional perfusion, this study showed that the mechanical
(17, 18). We were unable to find any relationship between res- properties of the thoracoabdominal cage and its variations
piratory mechanics and oxygenation when the supine position play an important role in dictating the oxygenation response
was resumed. in the prone position.
Respiratory mechanics. A main finding of this study was the
significant increase of the compliance of the total respiratory Long-term Effects of Prone Position
system above baseline values when returning to the supine po- In recent years a very large number of studies have reported
sition. This improvement was mainly due to an increase of the an improvement in oxygenation in patients with ALI when
lung compliance. It is somewhat surprising to find that the placed in the prone position (7–9). Indeed, this manuever is
changes in lung compliance were not associated with changes now widely accepted as a useful tool to improve oxygenation
in EELV, which did not significantly change. It is worth con- in patients with severe ALI.
sidering, however, that the way we measured the compliance However, besides the improvement in oxygenation associ-
included a possible intratidal recruitment (i.e., recruitment ated with prone positioning, some of these studies reported
during the inspiratory phase of tidal ventilation) (22). In fact other interesting observations such as the persistent improve-
the substantial redistribution of regional lung collapse ob- ment above baseline values when returning to the supine posi-
served during the prone position (6) could well have resulted tion (17, 18), a dramatic redistribution of CT-lung densities in
in an increased portion of the lung becoming recruitable dur- the prone position (6), and a decrease of densities when re-
ing tidal insufflation when returning to the supine position turning supine after 4 h in the prone position, compared with
(22), thus explaining the increase in lung compliance associ- baseline (26). More recently, Broccard and colleagues (27)
Pelosi, Tubiolo, Mascheroni, et al.: Respiratory Mechanics and Oxygenation in the Prone Position 393

showed that the prone position may decrease the damaging ef- coordination. Am. J. Respir. Crit. Care Med. 149:818–824.
fect of large tidal volume ventilation in dogs with preinjured 11. Pelosi, P., M. Croci, E. Calappi, D. Mulazzi, M. Cerisara, P. Vercesi, P.
Vicardi, and L. Gattinoni. 1996. Prone positioning improves pulmo-
lungs. In the present study, we have shown that the prone po-
nary function in obese patients during general anesthesia. Anesth.
sition leads, independent of gas exchange, to an improvement Analg. 83:578–583.
in the mechanical properties of the respiratory system when 12. Pelosi, P., M. Croci, I. Ravagnan, M. Cerisara, P. Vicardi, A. Lissoni, and
returning to the supine position, thus adding another piece of L. Gattinoni. 1997. Respiratory system mechanics in sedated, para-
evidence to the hypothesis that the prone position, besides lyzed, morbidly obese patients. J. Appl. Physiol. 83:811–818.
improving gas exchange, induces modifications in underlying 13. Iberti, T. J., C. E. Lieber, and E. Benjamin. 1989. Determination of in-
tra-abdominal pressure using a transurethral bladder catheter: clinical
lung conditions that persist after returning to the supine posi-
validation of the technique. Anesthesiology 70:47–50.
tion. 14. Pelosi, P., M. Cereda, G. Foti, M. Giacomini, and A. Pesenti. 1995. Al-
We believe that these findings taken together, represent a terations of lung and chest wall mechanics in patients with acute lung
solid base to justify a prospective trial evaluating whether the injury: effects of positive end-expiratory pressure. Am. J. Respir. Crit.
prone position may affect the clinical course of acute lung in- Care Med. 152:531–537.
jury. 15. Milic-Emili, J., J. Mead, and J. M. Turner. 1964. Topography of esoph-
ageal pressure as a function of posture in man. J. Appl. Physiol. 19:212–
Acknowledgment : The writers wish to thank the physicians and the nursing 216.
staff of the Intensive Care Unit of the Ospedale Maggiore for their coopera- 16. Pelosi, P., M. Croci, E. Calappi, M. Cerisara, D. Mulazzi, P. Vicardi, and
tion in the management of patients during the study, and Prof. Antonio Pe- L. Gattinoni. 1995. The prone positioning during general anesthesia
senti and Prof. Lawrence R. Goodman for their helpful suggestions in pre- minimally affects respiratory mechanics while improving functional
paring the manuscript. residual capacity and increasing oxygen tension. Anesth. Analg. 80:955–
960.
17. Langer, M., D. Mascheroni, R. Marcolin, and L. Gattinoni. 1988. The
References prone position in ARDS patients: a clinical study. Chest 94:103–107.
1. Piehl, M. A., and R. S. Brown. 1976. Use of extreme position changes in 18. Chatte, G., J. M. Sab, J. M. Dubois, M. Sirodot, P. Gaussorgues, and D.
acute respiratory failure. Crit. Care Med. 4:13–14. Robert. 1997. Prone position in mechanically ventilated patients with
2. Douglas, W. W., K. Rheder, F. M. Beynen, A. D. Sessler, and H. M. severe acute respiratory failure. Am. J. Respir. Crit. Care Med. 155:473–
Marsh. 1977. Improved oxygenation in patients with acute respiratory 478.
failure: the prone position. Am. Rev. Respir. Dis. 115:559–566. 19. Wagaman, M. J., J. G. Shutack, A. S. MoomJian, J. G. Schwartz, T. H.
3. Albert, R. K., D. Leasa, M. Sanderson, H. T. Robertson, and M. P. Shaffer, and W. W. Fox. 1979. Improved oxygenation and lung com-
Hlastala. 1987. The prone position improves arterial oxygenation and pliance with prone positioning of neonates. J. Pediatr. 94:787–791.
reduces shunt in oleic-acid-induced acute lung injury. Am. Rev. Respir. 20. Mutoh, T., R. J. Guest, W. J. E. Lamm, and R. K. Albert. 1992. Prone
Dis. 135:628–633. position alters the effect of volume overload on regional pleural pres-
4. Lamm, W. J. E., M. M. Graham, and R. K. Albert. 1994. Mechanism by sures and improves hypoxemia in pigs in vivo. Am. Rev. Respir. Dis.
which the prone position improves oxygenation in acute lung injury. 146:300–306.
Am. J. Respir. Crit. Care Med. 150:184–193. 21. Osmon, G. O. 1995. Functional anatomay of the chest wall. In Ch. Rous-
5. Pappert, D., R. Rossaint, K. Slama, T. Gruning, and K. J. Falke. 1994. sos, editor. The Thorax, 2nd ed. Marcel Dekker, New York. 413–443.
Influence of positioning on ventilation-perfusion relationships in se- 22. Gattinoni, L., P. Pelosi, S. Crotti, and F. Valenza. 1995. Effects of posi-
vere adult respiratory distress syndrome. Chest 106:1511–1516. tive end-expiratory pressure on regional distribution of tidal volume
6. Gattinoni, L., P. Pelosi, G. Vitale, A. Pesenti, L. D’Andrea, and D. and recruitment in adult respiratory distress syndrome. Am. J. Respir.
Mascheroni. 1991. Body position changes redistribute lung computed- Crit. Care Med. 151:1807–1814.
tomographic density in patients with acute respiratory failure. Anes- 23. Pelosi, P., L. D’Andrea, G. Vitale, A. Pesenti, and L. Gattinoni. 1994.
thesiology 74:15–23. Vertical gradient of regional lung inflation in adult respiratory distress
7. Broccard, A., and J. J. Marini. 1995. Effect of position and posture on syndrome. Am. J. Respir. Crit. Care Med. 149:8–13.
the respiratory system. In J. L. Vincent, editor. Yearbook of Intensive 24. Liu, S., S. S. Margulies, and T. A. Wilson. 1990. Deformation of the dog
Care and Emergency Medicine, 1st ed. Springer-Verlag, Berlin. 165– lung in the chest wall. J. Appl. Physiol. 68:1979–1987.
184. 25. Hyatt, R. E., E. Bar-Yishay, and M. D. Abel. 1985. Influence of the heart
8. Joliet, P., P. Bulpa, and J. C. Chevrolet. 1996. Ventilation en decubitus on the vertical gradient of transpulmonary pressure in dogs. J. Appl.
ventral lors du syndrome de detresse respiratorie aigue (SDRA). Physiol. 58:52–57.
Schweiz. Med. Wochenschr. 126:879–882. 26. Priolet, B., G. Tempelhoff, J. Miller, A. Cannamela, and M. J. Carton.
9. Gattinoni, L., P. Pelosi, F. Valenza, and D. Mascheroni. 1994. Patient 1993. Ventilation assiste’ en decubitus ventral: evaluation tomodensit-
positioning in acute respiratory failure. In M. J. Tobin, editor. Princi- ometrique de son efficacite’ dans le traitment des condensations pul-
ples and Practice of Mechanical Ventilation, 1st ed. McGraw-Hill, monaires. Rean. Urg. 2:81–85.
New York. 1067–1076. 27. Broccard, A. F., R. S. Shapiro, L. L. Schmitz, S. A. Ravenscraft, and J. J.
10. Bernard, G. R., A. Artigas, K. L. Brigham, J. Carlet, K. Falke, L. Hud- Marini. 1997. Influence of prone position on the extent and distribu-
son, M. Lamy, J. R. Legall, A. Morris, R. Spragg, and the Consensus tion of lung injury in a high tidal volume oleic acid model of acute res-
Committee. 1994. The American-European consensus conference on piratory distress syndrome. Crit. Care Med. 25:16–27.
ARDS: definitions, mechanisms, relevant outcomes, and clinical trial