The prevalence of pleural effusions is high in ICU

Transcription

1 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, ml [ 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 (PLD 5 ) were significantly correlated with the drained volume (PLDbase, r 0.68, p < 0.001; PLD 5, r 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 6% and 9 6% for PLDbase, and 6 5% and 8 5% for PLD 5. The PaO 2 /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: ) Key words: chest drainage; ICU; mechanical ventilation; pleural effusion; ultrasonography Abbreviations: Fio 2 fraction of inspired oxygen; LD lung-diaphragm distance; PLDbase distance between the lung and posterior chest wall at the lung base; PLD 5 distance between lung and posterior chest wall at the fifth intercostal space; ROC receiver operator characteristic The prevalence of pleural effusions is high in ICU patients, 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 oxygenation 3 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 detecting 8 and quantifying effusions 9 because of underestimation or overestimation of the effusion volume. Ultrasonography is a simple noninvasive bedside procedure that rapidly detects most of liquid effusions It can be accurately performed after a short training period, and can be securely repeated over the time. Sonog- 224 Clinical Investigations in Critical Care

2 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 studies 9,14,15 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, 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, Reproduction of this article is prohibited without written permission from the American College of Chest Physicians ( permissions@chestnet.org). Correspondence to: Antoine Roch, MD, PhD, Département d Anesthésie Réanimation, Hôpital Sainte-Marguerite, Marseille Cedex 9, France; Antoine.Roch@mail.ap-hm.fr 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,

3 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 (PLD 5 ). 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 PLD 5 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 H 2 O. 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 Clinical Investigations in Critical Care

4 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 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 Effusion 500 ml (n 20) Effusion 500 ml (n 24) p Value Drained volume, ml , LD, cm PLDbase, cm PLD 5,cm Side of effusion Right Left 9 13 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, m Male/female gender 12/8 16/8 0.5 Cardiothoracic ratio Ventilatory mode on inclusion Controlled mode PSV Pao 2 /Fio 2,mmHg Before drainage After drainage PEEP, cm H 2 O Before drainage After drainage Heart rate, beats/min Before drainage After drainage MAP, mm Hg Before drainage After drainage *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. CHEST / 127 / 1/ JANUARY,

5 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 PLD 5 ), 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, Pao 2 /fraction of inspired oxygen (Fio 2 ) 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 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, 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 Pao 2 /Fio 2 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 PLD 5 were significantly correlated with the collected volume 3 h after drainage (PLDbase, r 0.68, p 0.001; PLD 5, 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, PLD 5, 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 Clinical Investigations in Critical Care

6 Figure 4. Correlation between drained volume and evolution in Pao 2 /Fio 2 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). PLD 5 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* Variables Interobserver Error, % Intraobserver Error, % Figure 5. Correlation between PLDbase and drained volume. PLDbase, cm PLD 5,cm LD, cm *Data are presented as mean SD. CHEST / 127 / 1/ JANUARY,

7 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 PLD 5 ) correlates with its volume, (2) PLDbase 5 cm predicts an effusion volume 500 ml with very good operating characteristics, (3) PLDbase and PLD 5 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 al 3 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 ml ( SD) produced an increase in Pao 2 by 60% at 24 h after drainage. In patients not receiving ventilation with an acute exacerbation of congestive heart failure, Miyamoto et al 15 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 puncture 23 and determining the nature of pleural effusion. 24,25 Finally, previous studies 9,14,15 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 al 15 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 al 9 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 ml, while an effusion of 40 mm had a mean volume of 1, 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 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 Sensitivity Specificity Positive Predictive Value Negative Predictive Value Accuracy Clinical Investigations in Critical Care

8 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 significant effusion volume. This was based on the data from previous studies 3,15 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: Mattison LE, Coppage L, Alderman DF, et al. Pleural effusions in the medical ICU, prevalence, causes and clinical implications. Chest 1997; 111: 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: Colt HG, Brewer N, Barbur E. Evaluation of patient-related and procedure-related factors contributing to pneumothorax following thoracocentesis. Chest 1999; 116: Collop NA, Kim S, Sahn SA. Analysis of tube thoracostomy performed by pulmonologists at a teaching hospital. Chest 1997; 112: Resnick DK. Delayed pulmonary perforation: a rare complication of tube thoracostomy. Chest 1993; 103: Mergo PJ, Helmberger T, Didovic J, et al. New formula for quantification of pleural effusions from computed tomography. J Thorac Imaging 1999; 14: 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: Eibenberger KL, Dock WI, Ammann ME, et al. Quantification of pleural effusions: sonography versus radiography. Radiology 1994; 191: CHEST / 127 / 1/ JANUARY,

9 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: Sisley AC, Roycki GS, Ballard RB, et al. Rapid detection of traumatic effusion using surgeon-performed ultrasonography. J Trauma 1998; 44: 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: Kataoka H, Takada S. The role of thoracic ultrasonography for evaluation of patients with decompensated chronic heart failure. J Am Coll Cardiol 2000; 35: 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: 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: Brismar B, Hedenstierna G, Lundquist H, et al. Pulmonary densities during anesthesia with muscular relaxation, a proposal of atelectasis. Anesthesiology 1985; 62: Lundquist H, Hedenstierna G, Strandberg A, et al. CTassessment of dependent lung densities in man during general anaesthesia. Acta Radiol 1995; 36: 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: 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: Froese AB, Bryan AC. Effects of anesthesia and paralysis on diaphragmatic mechanics in man. Anesthesiology 1974; 41: Drummond GB, Allan PL, Logan MR. Changes in diaphragmatic position in association with the induction of anaesthesia. Br J Anaesth 1986; 58: Laws D, Neville E, Duffy J. BTS guidelines for the insertion of a chest drain. Thorax 2003; 58(suppl II): Lichtenstein D, Hulot JS, Rabiller A, et al. Feasibility and safety of ultrasound-aided thoracentesis in mechanically ventilated patients. Intensive Care Med 1999; 25: Yu CJ, Yang PC, Chang DB. Diagnostic and therapeutic use of chest sonography: value in critically ill patients. AJR Am J Roentgenol 1992; 159: 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: Clinical Investigations in Critical Care