Usefulness of Ultrasonography in
Predicting Pleural Effusions > 500 mL in
Patients Receiving Mechanical
Ventilation*
Antoine Roch, MD PHD; Mirela Bojan, MD; Pierre Michelet, MD;
Fanny Romain, MD; Fabienne Bregeon, MD; Laurent Papazian, MD PHD; and
Jean-Pierre Auffray, MD
Study objective: To assess the accuracy of chest ultrasonography in predicting pleural effusions
> 500 mL in patients receiving mechanical ventilation.
Design: Prospective study.
Setting: Surgical and medical ICU in a teaching hospital.
Patients: Forty-four patients receiving mechanical ventilation with indications of chest drainage
of a nonloculated pleural effusion.
Interventions: Diagnosis of pleural effusion was based on clinical examination and chest
radiography. Chest drainage was indicated when considered as potentially useful for the patient
(hypoxemia and/or weaning failure). Sonograms were performed before drainage at the bedside,
in the supine position, and measurements were performed at the end of expiration. Effusions
were classified as > 500 mL or < 500 mL according to the drained volume.
Measurements and results: The drained volume ranged from 100 to 1,800 mL (mean, 730 ⴞ 440
mL [ⴞ SD]). The distance between the lung and posterior chest wall at the lung base (PLDbase)
and the distance between the lung and posterior chest wall at the fifth intercostal space (PLD5)
were significantly correlated with the drained volume (PLDbase, r ⴝ 0.68, p < 0.001; PLD5,
r ⴝ 0.56, p < 0.001). A PLDbase > 5 cm predicted a drained volume > 500 mL with a sensitivity
of 83%, specificity of 90%, positive predictive value of 91%, and negative predictive value of 82%.
Interobserver and intraobserver percentages of error were, respectively, 7 ⴞ 6% and 9 ⴞ 6% for
PLDbase, and 6 ⴞ 5% and 8 ⴞ 5% for PLD5. The PaO2/fraction of inspired oxygen ratio
significantly increased after chest drainage in patients with collected volumes > 500 mL
(p < 0.01).
Conclusions: Bedside pleural ultrasonography accurately predicted a nonloculated pleural
effusion > 500 mL in patients receiving mechanical ventilation using simple and reproducible
measurements.
(CHEST 2005; 127:224 –232)
Key words: chest drainage; ICU; mechanical ventilation; pleural effusion; ultrasonography
Abbreviations: Fio2 ⫽ fraction of inspired oxygen; LD ⫽ lung-diaphragm distance; PLDbase ⫽ distance between the
lung and posterior chest wall at the lung base; PLD5 ⫽ distance between lung and posterior chest wall at the fifth
intercostal space; ROC ⫽ receiver operator characteristic
prevalence of pleural effusions is high in ICU
T hepatients,
reaching 62% in medical patients.
1,2
Apart from traumatic effusions and empyema, most
pleural effusions are free effusions and are associated
with atelectasis, heart failure, fluid overload, hypoalbuminemia, or abdominal diseases.2 Most effusions disappear with the treatment of their cause or
with diuretics. However, in patients receiving mechanical ventilation, drainage of large effusions may
be useful to improve oxygenation3 or respiratory
system compliance. Because thoracentesis and tube
drainage are associated with complications,1,4 – 6
quantifying effusion volume may be of benefit by
influencing decision of whether drainage is required
and by regularly reevaluating the effusion amount.
CT scanning is the “gold standard” for the quantification of pleural effusions,7,8 but is difficult to
perform in patients receiving mechanical ventilation
in the ICU. However, standard chest radiography
was shown to be inaccurate in detecting8 and quantifying effusions9 because of underestimation or
overestimation of the effusion volume. Ultrasonography is a simple noninvasive bedside procedure that
rapidly detects most of liquid effusions.8 –11 It can be
accurately performed after a short training period,
and can be securely repeated over the time. Sonog-
224
Downloaded From: http://publications.chestnet.org/pdfaccess.ashx?url=/data/journals/chest/22020/ on 06/17/2017
Clinical Investigations in Critical Care
raphy was shown to have better sensitivity and
specificity than chest radiography for the positive
diagnosis of pleural effusion, notably in patients
receiving mechanical ventilation.8,9,12–13 Only a few
studies9,14,15 have evaluated sonography for quantification, but none have evaluated this diagnostic tool
in the precise field of patients receiving mechanical
ventilation. In patients receiving mechanical ventilation who are under anesthesia, there are well-known
microatelectases that develop mainly in dependent
areas,16,17 a fall in the functional residual capacity,18,19 and a cephalad displacement of the diaphragm,19 –21 which also has reduced motion. By
altering the chest and lung sizes, these phenomena
could contribute to a modification in the location and
measurements of pleural effusions and, consequently, to their sonographic quantification. The
purpose of this prospective study was to assess the
accuracy of chest ultrasonography in predicting pleural effusions ⬎ 500 mL using simple and reproducible measurements.
Materials and Methods
During a 6-month period, 655 patients were admitted to our
medical and surgical ICU. Among them, 410 patients received
mechanical ventilation for ⬎ 24 h. We included all patients in
whom drainage of a pleural effusion of a nontraumatic origin and
without sign of loculation was considered potentially therapeutic
and who did not meet exclusion criteria. Study protocol was
approved by the local ethics committee. Written consent was
obtained from the next of kin. Patients were sedated under
controlled-mode ventilation or were not sedated under pressuresupport ventilation. No patients received muscle relaxants continuously. Patients who were not sedated at the moment of
inclusion had previously received sedation for 5 ⫾ 6 days (⫾ SD).
The diagnosis of pleural liquid effusion was based on clinical
and radiologic diagnosis. Clinical diagnosis was based on the
association of absence of breath sounds to auscultation, flatness to
percussion, and decrease in thoracic inspiratory movements.
Anteroposterior chest radiographs were obtained in the semirecumbent position, and were reviewed independently by the two
co-attending clinicians in charge of the patient. Effusion was
*From the Service de Réanimation Polyvalente (Drs. Roch,
Bojan, Michelet, and Auffray), Service d’ Information Médicale
(Dr. Romain), and Service de Réanimation Médicale (Dr. Papazian), Hôpitaux Sud; and Laboratoire de Physiologie Respiratoire
(Dr. Bregeon), UPRES EA 2201, Université de la Méditerranée,
Marseille, France.
This work was done in the Service de Réanimation Polyvalente
and the Service de Réanimation Médicale, Hôpitaux Sud, Marseille, France.
Institutional support was provided by the Assistance Publique,
Hôpitaux de Marseille, France.
Manuscript received January 13, 2004; revision accepted July 14,
2004.
Reproduction of this article is prohibited without written permission from the American College of Chest Physicians (e-mail:
[email protected]).
Correspondence to: Antoine Roch, MD, PhD, Département
d’Anesthésie Réanimation, Hôpital Sainte-Marguerite, 13274
Marseille Cedex 9, France; e-mail: [email protected]
www.chestjournal.org
defined as a blunting of the lateral costophrenic angle associated
with an opacification that covered at least the lower lobe without
obscuration of vascular markings. The patient was included if
receiving mechanical ventilation, and if the indication of drainage
was confirmed by the two clinicians, ie, considered as potentially
useful for the patient (hypoxemia and/or weaning failure). If the
pleural effusion was bilateral, the side where the effusion was
considered larger on the chest radiograph was chosen for drainage. Exclusion criteria were as follows: (1) emergency necessity of
a tube thoracostomy, (2) severe hemostasis alterations (platelets
⬍ 50 g/L, fibrinogen ⬍ 2 g/L, prothrombin ⬍ 50% of control, or
cephalin-activated time more than twice the control), and (3) risk
factors for a loculated effusion: any previous history of chest
drainage, thoracic surgery, or chest trauma.
Ultrasonography
Sonography was performed with an ultrasound scanner (Sonos
3000 Hewlett-Packard Ultrasound Scanner; Hewlett-Packard;
Andover, MA) using a 3-MHz transducer and bi-dimensional
gray-scale mode. All sonograms were performed in patients
receiving positive-pressure mechanical ventilation at the bedside
in the supine position with arm abducted. Positive end-expiratory
pressure was not modified for the examination. Sonography was
always performed by the same clinician, an intensivist with
experience of 2 years in sonography (25 h) and 10 years in ICU.
This clinician was only informed of the suspicion of a pleural
effusion and the side it was on, but could not see the chest
radiograph. He did not communicate the results of sonography to
the clinician in charge before chest drainage, except when
detecting no pleural effusion. In this case, a thoracic CT scan was
obtained to confirm the absence of effusion (effusion thickness
⬍ 0.5 cm), and the physician postponed the decision of chest
drainage.
Detection of Pleural Effusion
The transducer was positioned on the posterior axillary line
between the ninth and eleventh ribs to identify the liver on the
right side, the spleen on the left side, and the diaphragm. To
visualize the effusion, the transducer was then advanced cephalad
and a longitudinal view was chosen. The positive diagnosis of
pleural effusion was based on the association of the following: (1)
the presence of an anechoic image above the diaphragm, (2) the
image was bordered by the parietal layer in surface and by the
visceral layer in depth, (3) the identification of the lung behind
the effusion, and (4) the inspiratory decrease in the interpleural
distance. If the effusion looked partitioned and/or suspended, it
was considered as loculated and the patient was excluded.
Measurements of Pleural Effusion
Measurements of pleural effusion were performed at end of
expiration. Recorded distances corresponded to the mean of
three measurements (Fig 1).
Lung-Diaphragm Distance: In a longitudinal view, the diaphragm
and lung were identified by setting the transducer on the posterior
axillary line at a right angle to the chest wall (Figs 1, top, 2). Then,
the widest thickness of the anechoic area between the lung and
diaphragm was recorded as lung-diaphragm distance (LD).
Distance Between the Lung and Posterior Chest Wall at the
Lung Base: When the inferior pole of the lung was identified, a
transducer was placed 30 mm above it and a transversal view was
obtained by rotating the transducer by 90° [Figs 1, bottom, 2].
The effusion appeared as a lenticular shape between the chest
wall and the lung. The anteroposterior thickness of the effusion at
the most dependant place was recorded as the distance between
CHEST / 127 / 1 / JANUARY, 2005
Downloaded From: http://publications.chestnet.org/pdfaccess.ashx?url=/data/journals/chest/22020/ on 06/17/2017
225
Figure 1. Top: Longitudinal ultrasonographic view showing a
moderate effusion (E) delimited by diaphragm (D), parietal
pleura (PP), and visceral pleura (VP) surrounding consolidated
lung (L). In this patient, LD (F) is ⬍ 0.5 cm. Bottom: Transversal
ultrasonographic view at lung base showing a moderate effusion
(E) surrounding consolidated lung (L). In this patient, PLDbase
distance (7), representing distance between lung and posterior
chest wall (PCW), is ⬍ 2.5 cm.
the lung and posterior chest wall at the lung base (PLDbase).
When patients had large amount of pleural effusion, the lower
extremity of the lung was reduced to a floating thin line. In this
case, a transversal view was obtained after a cephalad displacement of the transducer (Fig 2, bottom).
Distance Between the Lung and Posterior Chest Wall at the
Fifth Intercostal Space: A transducer was advanced cephalad
until the fifth intercostal space, and a new transversal view was
obtained. The anteroposterior thickness of the lenticular shape at
the most dependent place was recorded as the distance between
the lung and posterior chest wall at the fifth intercostal space
(PLD5).
Chest Drainage
Chest drainage was performed in the hour following sonography, in the supine position, and after surgical disinfection. In
Figure 2. Top: Longitudinal view of the chest showing the place
of ultrasonographic measurements. PLDbase and PLD5 measurements are performed using a transversal view by rotating a
transducer by 90°. PLDbase is measured 30 mm above the
inferior pole of the lung. Bottom: If the lung base is reduced to
a floating line, PLDbase is measured 30 mm above the superior
pole of this collapsed zone after a cephalad displacement of the
transducer. The dashed arrow would represent a false measurement of PLDbase.
patients not sedated (n ⫽ 25), drainage was preceded by local
anesthesia and sedation. A chest tube (24F) was inserted in the
fifth intercostal space on the midaxillary line after disconnection
from the ventilator when possible and advanced toward the
posterior costophrenic sinus.22 Attention was paid to avoid the
loss of liquid before connection of the aspiration system. The
chest tube was connected to a closed system of drainage, and
suction was settled to 50 cm H2O. The patient was replaced in a
semirecumbent position. Chest radiography was performed to
confirm the position of the tube. A sonographic examination was
performed 3 h after drainage. If no residual effusion was detected
(no anechoic area between the lung, chest, and diaphragm), the
volume of effluent was noted and suction was stopped. If a
residual effusion was present, the chest tube was moved and
226
Downloaded From: http://publications.chestnet.org/pdfaccess.ashx?url=/data/journals/chest/22020/ on 06/17/2017
Clinical Investigations in Critical Care
sonographic control was performed. Then, the volume of effluent
was noted 1 h after. If a residual effusion persisted, the patient
was excluded. Ventilator settings could be modified after drainage without any protocol. Occurrence of thoracic bleeding or
pneumothorax was recorded during the duration of the drainage.
Samples of pleural effusion were obtained at the moment of the
drainage and before tube removal for microbiologic analysis. The
chest tube was removed when the effluent was ⬍ 100 mL/d.
Recorded Data
Demographic data are exposed in Table 1. The cardiothoracic
ratio was measured on the inclusion chest radiography. The body
surface area was calculated using a standard formula. Concerning
pleural effusion, recorded data were the most probable cause, the
side, and the collected volume. Effusions were classified as parapneumonic and/or associated with atelectasis (without empyema),
heart failure, hypoalbuminemia, or intraabdominal diseases.
Assessment of Reproducibility of Sonographic Measurements
Intraobserver and interobserver reproducibilities were assessed in the 44 patients. Intraobserver reproducibility was
assessed using video-recorded data by repeating the measurements on two occasions (15 to 30 days after initial examination).
To assess the interobserver reproducibility, the measurements
were performed from video recordings by one other observer on
three occasions. This observer was an intensivist with an experience of 1 year in pleural sonography (12 h), and who was unaware
of the results of the first observer. Reproducibility, expressed as
percentage of error, was calculated as the maximal difference
between the three sets of measurements divided by the mean
value of the measurement.
Statistical Analysis
Statistical calculation was performed using a statistical software
package (SPSS 11.0 package; SPSS; Chicago, IL). Distribution
was checked. Data are expressed as mean ⫾ SD. Respiratory and
hemodynamic parameters before and after chest drainage were
compared using paired Student t test. Correlation between
sonographic measurements and drained volume was performed
by mean of the Spearman correlation test. Effusions were
classified a priori in two groups according to the drained volume:
⬎ 500 mL or ⱕ 500 mL. The ␹2 test was used to analyze
differences in proportions of sex and side of the effusion in the
Table 1—Patient Characteristics According to the Amount of Drained Liquid*
Characteristics
Drained volume, mL
LD, cm
PLDbase, cm
PLD5, cm
Side of effusion
Right
Left
Age, yr
ICU length of stay, d
Total duration of ventilation, d
Time between admission and inclusion, d
Duration of sedation before inclusion, d
Sepsis before inclusion
SAPS II score on admission
Body size, cm
Body weight, kg
Body area, m2
Male/female gender
Cardiothoracic ratio
Ventilatory mode on inclusion
Controlled mode
PSV
Pao2/Fio2, mm Hg
Before drainage
After drainage
PEEP, cm H2O
Before drainage
After drainage
Heart rate, beats/min
Before drainage
After drainage
MAP, mm Hg
Before drainage
After drainage
Effusion ⱕ 500 mL
(n ⫽ 20)
Effusion ⬎ 500 mL
(n ⫽ 24)
p Value†
346 ⫾ 107
1.3 ⫾ 1.7
3.3 ⫾ 1.8
3.6 ⫾ 1.7
1,050 ⫾ 344
3 ⫾ 3.2
6.9 ⫾ 2.5
5.9 ⫾ 1.9
0.4
⬍ 0.001
⬍ 0.001
10
9
59 ⫾ 9
38 ⫾ 21
27 ⫾ 15
14 ⫾ 12
4⫾5
7
45 ⫾ 11
169 ⫾ 7
71 ⫾ 12
1.81 ⫾ 0.1
12/8
0.56 ⫾ 0.06
11
13
61 ⫾ 12
36 ⫾ 17
29 ⫾ 14
13 ⫾ 12
7⫾5
10
40 ⫾ 15
171 ⫾ 8
74 ⫾ 13
1.85 ⫾ 0.1
16/8
0.55 ⫾ 0.07
9
11
10
14
0.8
214 ⫾ 83
232 ⫾ 110
206 ⫾ 62
251 ⫾ 91‡
0.8
5⫾2
5⫾2
6⫾2
6⫾2
0.8
93 ⫾ 22
89 ⫾ 21
75 ⫾ 16
77 ⫾ 17
0.3
88 ⫾ 15
84 ⫾ 13
80 ⫾ 14
78 ⫾ 12
0.5
0.8
0.9
0.7
0.7
0.8
0.5
0.5
0.3
0.8
0.7
0.7
0.5
0.9
*Data are presented as mean ⫾ SD or No. SAPS ⫽ simplified acute physiology score; PEEP ⫽ positive end-expiratory pressure; MAP ⫽ mean
arterial pressure.
†Univariate analysis.
‡p ⬍ 0.01 vs before drainage by paired Student t test.
www.chestjournal.org
CHEST / 127 / 1 / JANUARY, 2005
Downloaded From: http://publications.chestnet.org/pdfaccess.ashx?url=/data/journals/chest/22020/ on 06/17/2017
227
two groups. Student t test was used to compare the means of the
following continuous variables between the two groups: the three
sonographic measurements (LD, PLDbase, and PLD5), age,
body size, weight and surface area, cardiothoracic ratio, simplified acute physiology score II on admission, time between
admission and inclusion, length of ICU stay, duration of ventilation, Pao2/fraction of inspired oxygen (Fio2) ratio, heart rate, and
the mean arterial pressure before drainage. Factors with p ⬍ 0.20
by univariate analysis and additional variables that had potential
clinical importance were introduced in a stepwise logistic regression analysis. For the factors with a p ⬍ 0.05 by the logistic
regression analysis, a receiver operator characteristic (ROC)
curve was constructed to determine the optimal cut-off point to
predict a volume ⬎ 500 mL or ⱕ 500 mL. Sensitivity, specificity,
and predictive values were calculated.
Results
Among the 52 patients with indication of chest
drainage, sonography could not detect pleural effusion in 5 patients and CT scan confirmed the
absence of effusion. In three other patients, effusion
looked partitioned on sonography. Consequently,
chest drainage was not performed and those patients
were excluded. The 44 included patients were admitted for coma (n ⫽ 13), trauma without chest
injury (n ⫽ 12), acute heart failure (n ⫽ 4), pneumonia (n ⫽ 6), peritonitis (n ⫽ 5), or acute pancreatitis
(n ⫽ 4). Patient characteristics are listed in Table 1.
Characteristics of Pleural Effusions
Pleural effusion was predominantly parapneumonic and/or consecutive to atelectasis in 18 patients
(41%), consecutive to heart failure in 9 patients
(20%), hypoalbuminemia in 8 patients (18%), peritonitis in 5 patients (11%), and acute pancreatitis in
4 patients (10%). Pleural effusion was right sided in
22 patients and left sided in 22 patients. Drained
volume ranged from 100 to 1,800 mL (mean,
730 ⫾ 440 mL; Fig 3), whatever the side location.
No cases of infected or bloody effusion were observed at the moment of tube insertion or during the
period of drainage. The chest tube was removed
3 ⫾ 1 days after drainage. Moderate thoracic bleeding (⬍ 100 mL) occurred in two patients at the
moment of the insertion.
The Pao2/Fio2 ratio significantly increased 12 h
after chest drainage in patients with drained effusions ⬎ 500 mL (p ⬍ 0.01; Table 1), whereas it did
not increase in patients with drained effusion ⬍ 500
mL. Considering all patients, there was no correlation between drained volume and improvement in
oxygenation. However, there was a positive correlation between drained volume and improvement in
oxygenation in patients with drained effusions ⬎ 500
mL (r ⫽ 0.50, p ⫽ 0.01; Fig 4).
Correlation Between Drained Volume and
Ultrasonographic Measurements
PLDbase and PLD5 were significantly correlated
with the collected volume ⬎ 3 h after drainage
(PLDbase, r ⫽ 0.68, p ⬍ 0.001; PLD5, r ⫽ 0.56,
p ⬍ 0.001; Fig 5), whereas LD was not (r ⫽ 0.24,
p ⫽ 0.13). LD was nil in 16 patients, even with
significant volume effusions (up to 1,200 mL). The
interobserver and intraobserver reproducibilities of
sonographic measurements are exposed in Table 2.
Prediction of a Drained Volume ⬎ 500 mL by
Ultrasonography
The variables introduced in the logistic regression
analysis were PLDbase, PLD5, and side of the
effusion. PLDbase was predictive of a drained vol-
Figure 3. Distribution of patients according to the drained volume over 3 h.
228
Downloaded From: http://publications.chestnet.org/pdfaccess.ashx?url=/data/journals/chest/22020/ on 06/17/2017
Clinical Investigations in Critical Care
Figure 4. Correlation between drained volume and evolution in Pao2/Fio2 ratio before and 12 h after
drainage in effusions ⬎ 500 mL.
ume ⬎ 500 mL (p ⬍ 0.001; odds ratio, 2.66; 95%
confidence interval, 1.5 to 4.7). PLD5 measurement
and effusion side did not provide any additional
information. The ROC curve analysis was performed
using cut-off points of 0.5 cm of PLDbase between 0
cm and 15 cm (Fig 6). A value of 5 cm was chosen as
the threshold for PLDbase (Table 3). A PLDbase
⬎ 5 cm predicted a drained volume ⬎ 500 mL, with
a sensitivity of 83%, specificity of 90%, positive
predictive value of 91%, and negative predictive
value of 82%.
Among the 44 patients included, pleural effusion
was ⬍ 500 mL in 19 patients (43%), and was ⬍ 300
mL in 5 patients (11%). As inclusion was notably
based on radiographic signs, this result emphasizes
the lack of specificity of chest radiography for diagnosing significant effusions in ICU patients receiving
mechanical ventilation.
Discussion
The present study shows the following in patients
receiving mechanical ventilation: (1) anteroposterior
diameter of a nonloculated pleural effusion when
Table 2—Reproducibility of Sonographic
Measurements*
Figure 5. Correlation between PLDbase and drained volume.
www.chestjournal.org
Variables
Interobserver
Error, %
Intraobserver
Error, %
PLDbase, cm
PLD5, cm
LD, cm
7⫾6
6⫾5
22 ⫾ 14
9⫾6
8⫾5
10 ⫾ 8
*Data are presented as mean ⫾ SD.
CHEST / 127 / 1 / JANUARY, 2005
Downloaded From: http://publications.chestnet.org/pdfaccess.ashx?url=/data/journals/chest/22020/ on 06/17/2017
229
Figure 6. ROC curve for PLDbase, showing sensitivity and
1-specificity of different PLDbase thresholds for diagnosis of an
effusion ⬎ 500 mL.
measured by sonography at the lung base or at the
fifth intercostal space (PLDbase or PLD5) correlates
with its volume, (2) PLDbase ⬎ 5 cm predicts an
effusion volume ⬎ 500 mL with very good operating
characteristics, (3) PLDbase and PLD5 can be measured with good interobserver and intraobserver
reproducibilities by a physician with a short training
period.
Chest thoracentesis has proved to contribute to
etiologic diagnosis and treatment of pleural effusions.1 Moreover, patients with pleural effusion
could have longer ICU stays and longer duration of
mechanical ventilation.2 Therefore, chest tube drainage was shown to be effective in voluminous effusions by improving oxygenation and thoracic compliance. Talmor et al3 studied the effects of tube
drainage in 19 patients receiving mechanical ventilation with hypoxemia despite high positive endexpiratory pressure levels. A mean liquid removal of
863 ⫾ 124 mL (⫾ SD) produced an increase in Pao2
by 60% at 24 h after drainage. In patients not
receiving ventilation with an acute exacerbation of
congestive heart failure, Miyamoto et al15 showed
that drainage of pleural effusions ⬎ 500 mL led to a
shorter term of oxygen supply and to a lower furosemide consummation. However, as thoracentesis
and tube drainage are associated with unpredictable
complications such as a pneumothorax,1,4 lung perforation,6 or pleural infection,5 they should be considered only in patients likely to benefit from chest
tube drainage.
Sonography has many advantages in evaluating
pleural effusions in ICU patients. First, it is portable,
and consequently superior to CT scan, the “gold
standard” method,7 in patients with reduced mobility. Second, it was shown to be superior to chest
radiography in detecting minimal pleural effusions.8,9 –13,15 In ICU patients, signs of pleural effusion are regularly overshadowed by parenchymal
lung disorders, and the supine chest radiograph was
found to have a sensitivity of only 39% and an
accuracy of 47% in detecting pleural effusion in
patients receiving mechanical ventilation.8 If thoracentesis is indicated for etiologic diagnosis, sonography makes it safer and more successful, notably by
guiding puncture23 and determining the nature of
pleural effusion.24,25 Finally, previous studies9,14,15
showed that sonography was more accurate than
radiography in quantifying pleural effusions. In 33
patients with an acute exacerbation of congestive
heart failure, Miyamoto et al15 measured the angle
between diaphragm, lung (right side), or pericardium (left side) in an interscapular view and supine
position. They found a significant correlation
(r ⫽ 0,77) with the aspirated effluent. Eibenberger
et al9 measured 51 nonloculated pleural effusions.
Sonographic measurements correlated better with
effusion volume (r ⫽ 0,80) than the radiographic
results. Using a standardized formula, an effusion
width of 20 mm had a mean volume of 380 ⫾ 130
mL, while an effusion of 40 mm had a mean volume
of 1,000 ⫾ 330 mL. The average prediction error
was 224 mL, while it was 465 mL with radiography.
The present study was conducted in patients with
mechanical ventilation. Mechanical ventilation induces alterations in the functional residual capacity,18,19 and a cephalad displacement of the diaphragm.19 –21 These phenomena could contribute to
Table 3—Operating Characteristics of PLDbase to Predict a Drained Volume > 500 mL According to Different
Thresholds
Threshold for
PLDbase, cm
4
4.5
5
5.5
6
Sensitivity
Specificity
Positive Predictive
Value
Negative Predictive
Value
Accuracy
0.88
0.88
0.83
0.75
0.58
0.65
0.7
0.9
0.95
0.95
0.75
0.78
0.91
0.95
0.94
0.81
0.82
0.82
0.76
0.65
0.77
0.79
0.86
0.84
0.75
230
Downloaded From: http://publications.chestnet.org/pdfaccess.ashx?url=/data/journals/chest/22020/ on 06/17/2017
Clinical Investigations in Critical Care
reduction in chest volume. The consecutive variation
in the location and dimensions of pleural effusions
could alter their sonographic quantification. We
found that an effusion width ⬎ 5 cm predicted an
effusion ⬎ 500 mL with a positive predictive value of
91%. This result is quite different from those by
Eibenberger et al.9 Using their results, we would
have largely overestimated the amount of liquid in
our patients. The differences in chest morphology
observed in patients receiving mechanical ventilation
could explain this discrepancy.
Our results confirm that sonography performed by
a clinician with relatively short experience in sonography can predict an effusion ⬎ 500 mL with good
operating characteristics. This result is important,
since this examination could be realized in the
absence of an available radiologist.
The correlation between PLDbase measurement
and effusion volume looked weaker when the effusion was ⬎ 1,000 mL. PLDbase could indeed evaluate only partially massive effusions. When the
effusion is very large, it surrounds the lung and
PLDbase does not evaluate the part that is localized
between lung and lateral or anterior chest wall,
leading to an underestimation of effusion amount.
Nevertheless, a very precise quantification of pleural
effusion ⬎ 1,000 mL may probably be not very
useful in patients receiving mechanical ventilation
since drainage will be systematically performed.
We did not demonstrate any relation between
drained volume and LD. Moreover, LD was nil in 16
patients with effusions up to 1,200 mL. The lack of
reliability of LD could be explained, first by the
supine position that directed effusion between lung
and posterior chest wall, and second by the variable
size and position of the triangular ligament between
lung and diaphragm.
We supposed that the left situation of the heart
would have influenced sonographic measurements
of the left-sided effusions. We did not find any
difference in the ability of PLDbase to predict a
significant effusion regarding its side or the cardiothoracic ratio. Consequently, PLDbase could be
equally used for left and right effusions using the
same thresholds.
This study has some limitations. First, inclusion
was mainly based on the presence of radiographic
signs of pleural effusion. This could have led to a
selection bias, since we did not include patients with
pleural effusion without a radiographic sign. In these
patients, liquid could have a different distribution.
Second, we only searched to predict a quantitatively
important effusion, but did not study the impact of
effusion on lung expansion. The extent of lung
atelectasis could perhaps help in indicating chest
drainage. Third, we a priori chose 500 mL as a
www.chestjournal.org
significant effusion volume. This was based on the
data from previous studies3,15 and from our clinical
practice. However, the benefit of chest drainage in
pleural effusion ⬎ 500 mL in the ICU patient
remains to be demonstrated. Indeed, even if we
observed a significant improvement in oxygenation
after drainage of effusions ⬎ 500 mL, this study was
not designed to evaluate this parameter. Finally, we
could not definitely eliminate the presence of a
loculated effusion in some patients and consequently
a residual effusion after drainage. However, we took
several precautions to prevent it. First, we excluded
a priori the patients considered at risk for loculated
effusion (history of drainage, trauma, or thoracic
surgery). Second, inclusion criteria based on the
chest radiograph corresponded to a free effusion.
Third, the physician who performed sonography was
particularly attentive in detecting any partitioning or
suspension of the effusion, what permitted us to
exclude three patients. Finally, sonographic control
was performed after drainage and confirmed the lack
of residual effusion in all patients.
In summary, our study showed the ability of
bedside pleural sonography to accurately predict an
effusion ⬎ 500 mL using simple and reproducible
measurements. Therefore, it could be useful in
indicating pleural drainage and in follow-up. Concerning effusions ⬍ 500 mL, the accuracy of sonography in indicating drainage should be evaluated,
notably by analyzing the extent of lung collapse.
References
1 Fartoukh M, Azoulay E, Galliot R, et al. Clinically documented pleural effusions in medical ICU patients: how useful
is routine thoracocentesis? Chest 2002; 121:178 –184
2 Mattison LE, Coppage L, Alderman DF, et al. Pleural
effusions in the medical ICU, prevalence, causes and clinical
implications. Chest 1997; 111:1018 –1023
3 Talmor M, Hydo L, Gershenwald JG, et al. Beneficial effects
of chest tube drainage of pleural effusion in acute respiratory
failure refractory to positive end-expiratory pressure ventilation. Surgery 1998; 123:137–143
4 Colt HG, Brewer N, Barbur E. Evaluation of patient-related
and procedure-related factors contributing to pneumothorax
following thoracocentesis. Chest 1999; 116:134 –138
5 Collop NA, Kim S, Sahn SA. Analysis of tube thoracostomy
performed by pulmonologists at a teaching hospital. Chest
1997; 112:709 –713
6 Resnick DK. Delayed pulmonary perforation: a rare complication of tube thoracostomy. Chest 1993; 103:311–313
7 Mergo PJ, Helmberger T, Didovic J, et al. New formula for
quantification of pleural effusions from computed tomography. J Thorac Imaging 1999; 14:122–125
8 Lichtenstein D, Goldstein I, Mourgeon E, et al. Comparative
diagnostic performances of auscultation, chest radiography,
and lung ultrasonography in acute respiratory distress syndrome. Anesthesiology 2004; 100:9 –15
9 Eibenberger KL, Dock WI, Ammann ME, et al. Quantification of pleural effusions: sonography versus radiography.
Radiology 1994; 191:681– 684
CHEST / 127 / 1 / JANUARY, 2005
Downloaded From: http://publications.chestnet.org/pdfaccess.ashx?url=/data/journals/chest/22020/ on 06/17/2017
231
10 Rozycki GS, Pennington SD, Feliciano DV. Surgeon-performed ultrasound in the critical care setting: its use as an
extension of the physical examination to detect pleural effusion. J Trauma 2001; 50:636 – 642
11 Sisley AC, Roycki GS, Ballard RB, et al. Rapid detection of
traumatic effusion using surgeon-performed ultrasonography.
J Trauma 1998; 44:291–296
12 Wu RG, Yuan A, Liaw YS, et al. Image comparison of
real-time gray-scale ultrasound and color Doppler ultrasound
for use in diagnosis of minimal pleural effusion. Am J Respir
Crit Care Med 1994; 150:510 –514
13 Kataoka H, Takada S. The role of thoracic ultrasonography
for evaluation of patients with decompensated chronic heart
failure. J Am Coll Cardiol 2000; 35:1638 –1646
14 Schmidt O, Simon S, Schmitt R, et al. Volumetry of pleural
effusion in multi-morbidity, postoperative patients of a surgical intensive care unit: comparison of ultrasound diagnosis
and thoracic bedside image. Zentralbl Chir 2000; 125:375–
379
15 Miyamoto T, Sasaki T, Kubo T, et al. Non-invasive determination of the quantity of pleural effusion and evaluation of the
beneficial effect of pleuracentesis in patients with acute
exacerbation of chronic congestive heart failure. J Cardiol
1997; 30:205–209
16 Brismar B, Hedenstierna G, Lundquist H, et al. Pulmonary
densities during anesthesia with muscular relaxation, a proposal of atelectasis. Anesthesiology 1985; 62:422– 428
17 Lundquist H, Hedenstierna G, Strandberg A, et al. CT-
18
19
20
21
22
23
24
25
assessment of dependent lung densities in man during general anaesthesia. Acta Radiol 1995; 36:626 – 632
Hedenstierna G, Jarnberg PO, Gottlieb I. Thoracic gas
volume measured by body plethysmography during anesthesia and muscle paralysis: description and validation of a
method. Anesthesiology 1981; 55:439 – 443
Hedenstierna G, Strandberg A, Brismar B, et al. Functional
residual capacity, thoracoabdominal dimensions, and central
blood volume during general anesthesia with muscle paralysis
and mechanical ventilation. Anesthesiology 1985; 62:247–254
Froese AB, Bryan AC. Effects of anesthesia and paralysis on
diaphragmatic mechanics in man. Anesthesiology 1974; 41:
242–255
Drummond GB, Allan PL, Logan MR. Changes in diaphragmatic position in association with the induction of anaesthesia. Br J Anaesth 1986; 58:1246 –1251
Laws D, Neville E, Duffy J. BTS guidelines for the insertion
of a chest drain. Thorax 2003; 58(suppl II):53–59
Lichtenstein D, Hulot JS, Rabiller A, et al. Feasibility and
safety of ultrasound-aided thoracentesis in mechanically ventilated patients. Intensive Care Med 1999; 25:955–958
Yu CJ, Yang PC, Chang DB. Diagnostic and therapeutic use
of chest sonography: value in critically ill patients. AJR Am J
Roentgenol 1992; 159:695–701
Yang P, Luh K, Chang D. Value of sonography in determining
the nature of pleural effusion: analysis of 320 cases. AJR Am J
Roentgenol 1992; 159:29 –33
232
Downloaded From: http://publications.chestnet.org/pdfaccess.ashx?url=/data/journals/chest/22020/ on 06/17/2017
Clinical Investigations in Critical Care