| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
Guest Access | Sign In via User Name/Password
|
|
|
|||||||
Guest Access | Sign In via User Name/Password |
|||||||||
* From the Department of Radiology, The Churchill Hospital, Old Road, Headington, Oxford, UK.
Correspondence to: Fergus Gleeson, MB ChB, Department of Radiology, The Churchill Hospital, Old Road, Headington, Oxford OX3 7LJ, UK; e-mail: Fergus.gleeson{at}radiology.oxford.ac.uk
Abstract
Purpose: To determine the diagnostic accuracy of ultrasound in the diagnosis of pneumothorax in patients with COPD with particular regard to false-positive diagnoses.
Materials and methods: This was a single-center, prospective, blinded study. Nine patients with pneumothorax, 9 patients with cystic fibrosis, 17 patients with COPD, and 6 control subjects were studied. Ultrasound clips were recorded at three positions in both hemithoraces of each patient and then reviewed by two observers blinded to patient status. Each clip was scored for the presence or absence of pneumothorax and the degree of observer confidence.
Results: The sensitivity and specificity for a pneumothorax were 100% and 84% for the experienced observer and 78% and 81% for the inexperienced observer, respectively. In the COPD patient group, specificity was 71% for the experienced observer and 65% for the inexperienced observer. There were no false-positive diagnoses in the cystic fibrosis or the control group.
Conclusion: Patients with COPD commonly show signs on ultrasound mimicking a pneumothorax, but this was not seen in patients with cystic fibrosis. In patients with COPD, ultrasound may be used to exclude the presence of a pneumothorax, but it cannot be used to confidently diagnose pneumothorax without using other imaging modalities.
Key Words: diagnosis • pneumothorax • ultrasound
The ultrasound diagnosis of pneumothorax is well established and has been reported to be of value in the acute assessment of patients when an upright chest radiograph (CXR) is not possible to achieve, most notably trauma patients123 and those in the ICU.456 The "focused assessment with sonography for trauma" ultrasound examination for peritoneal fluid in trauma patients has been extended to include thoracic examination, christened eFAST, and has been reported to be simple and easy to learn.3 Reports on the sensitivity of ultrasound for detection of a pneumothorax range from 73 to 100%, with a specificity of near 100%, with only a few studies reporting false-positive results,2 particularly in ICU patients.5 A large study3 of trauma patients excluded those with preexisting lung disease. A single study1 reported a false-positive diagnosis in a patient with bullous emphysema. We sought to determine the sensitivity and specificity of ultrasound in the detection of pneumothorax in patients with chronic lung disease with specific regard to false-positive diagnoses, and also tested the hypothesis that thoracic ultrasound is easy to learn.
Materials and Methods
This was a single-center, prospective, blinded study. Informed consent was obtained for all patients and volunteers. Institutional review board approval was obtained. Thoracic ultrasound examination was performed on four patient groups: known pneumothorax (n = 9), cystic fibrosis (n = 9), COPD (n = 17), and control (n = 6). The pneumothorax group was recruited from patients attending the radiology department for a diagnostic radiograph that demonstrated this diagnosis. The ultrasound examination was performed within 30 min of CXR. Two of the patients in the pneumothorax group had a previous spontaneous pneumothorax. One of the patients had a diagnosis of COPD. The others had a primary spontaneous pneumothorax. The cystic fibrosis patient group was recruited from patients attending the radiology department for CXR as part of an annual review. These patients had a diagnosis of cystic fibrosis from the chest clinic, and all had typical CXR features. Lung function data were available in eight patients. The ultrasound examination was performed within 30 min of CXR. The COPD group was recruited from patients attending a pulmonary rehabilitation clinic and from patients with known COPD admitted to hospital with an exacerbation of chest symptoms. Their pulmonary function data were recorded from the notes. All had a current inpatient CXR that excluded a concurrent pneumothorax. The control group was made up of volunteers from the radiology department staff who had no current or preceding history or symptoms of lung pathology. Lung function and CXRs were not obtained in the control group.
All CXRs were departmental, upright, posteroanterior chest films obtained as part of the patient’s normal management. Chest ultrasonography was performed (Sequoia; Acuson; Mountain View, CA) with either a 6- or 8-MHz linear probe. Short video clips were obtained at three positions on the chest: the second rib interspace anteriorly, the fourth interspace anteriorly, and the sixth interspace laterally in the mid-axillary line, on each side. Scans were performed in three sites to enable identification of small loculated pneumothoraces if present, and to assess whether there was an identifiable difference in pleural movement craniocaudally in the chest in the different patient groups. In all cases, the patients and volunteers were positioned semisupine and breathing normally. These clips were recorded on the hard drive of the machine and later reviewed by two radiologists independently, blinded as to the patient category. The first was an experienced observer with > 10 years of experience of thoracic ultrasound (F.V.G.). The second observer was a radiology resident with 4 years of experience of general ultrasound examinations but no prior experience of thoracic ultrasound (M.G.). He was shown normal pleural movement on a volunteer and then a single video clip of a patient with a pneumothorax.
Both observers recorded the presence of pleural gliding, comet tail artifacts, and observation of the "lung point" at each of the three sites, on each side, and selected a response from the following choices: definitely no pneumothorax, probably no pneumothorax, uncertain, probably a pneumothorax, and definitely a pneumothorax. Patient demographics are shown in Table 1 . Spirometry was available for eight of the cystic fibrosis patients. The median FEV1 was 1.8 L (range, 0.8 to 2.7 L), and the median FVC was 3.6 L (range, 2.0 to 4.8 L). Spirometry had been recorded on all patients in the COPD group; the median FEV1 was 0.8 L (range, 0.6 to 1.3 L), and the median FVC was 2.0 L (range, 1.3 to 2.7 L). There was no evidence of a pneumothorax on the chest radiographs of the cystic fibrosis or COPD groups. In total, the 41 subjects had both hemithoraces assessed, each in three sites and on both sides, resulting in 246 recordings.
|
Observer responses for each patient group are as follows.
Control Group (Six Patients)
All responses were correctly called definitely no pneumothorax by the experienced observer. One of six responses was called probably no pneumothorax by the inexperienced observer, with all five other responses correctly identified as definitely no pneumothorax. An image of normal pleura from a control subject is shown in Figure 1
.
|
Pneumothorax Patients (Nine Patients)
The experienced observer correctly identified all nine clips with confidence, reporting all cases as definitely a pneumothorax. In all patients, lack of pleural gliding and comet tail artifacts was reported in the most superior part scanned on the side of the pneumothorax (ie, the second intercostal space anteriorly). In one patient, pleural sliding was present in the fourth intercostal space, and in this patient plus four additional patients pleural sliding was seen in the sixth intercostal space. The positional distribution of pleural sliding is shown in Table 2
.
|
|
|
|
|
The only part of normal lung visible on ultrasound is the pleura. The visceral and parietal pleura are seen as a single echogenic line that moves with respiration.123 The added use of power Doppler may help to identify pleural movement in difficult cases.7 Bright lines of reverberation artifact called comet tails are seen originating from within discrete points of normal pleura, and these are also seen to move with respiration.5 The cause of these artifacts is not clear, and they are not always seen. Both of these signs indicate the absence of a pneumothorax, and their absence suggests that a pneumothorax is present, but unfortunately their absence may also occur without a pneumothorax.125 Other signs of pneumothorax on ultrasound include the loss of visualization of lung masses abutting the pleural surface after biopsy, as the mass falls away from the parietal pleura beneath the pneumothorax.8 The curtain sign is seen in the presence of a hydropneumothorax, in which the air-fluid level represents the lower edge of a curtain of air that obscures underlying detail. A similar sign is the "lung point" described by Lichtenstein et al6 as visualization of the edge of the pneumothorax, where lung sliding and absence of lung sliding are seen on the same image. This lung point can be seen to move with respiration. In a nonrandomized, unblinded study, Lichtenstein et al6 reported this in feature in 44 of 66 cases of pneumothorax and in no cases in a control group, and consequently suggested that this is the only specific ultrasound sign of a pneumothorax.
The absence of lung sliding in patients without pneumothorax has been previously reported.125 In an unblinded study of 76 patients, 42 with pneumothoraces on ICU including 4 patients with "chronic lung disease" and 34 critically ill control subjects without a pneumothorax, 6 of the control patients had absent lung sliding at ultrasound (specificity, 91%).5 Rowan et al1 performed a prospective blinded study of 11 patients with a traumatic pneumothorax, all of whom underwent ultrasound, supine CXR, and a chest CT. Ultrasound had a sensitivity of 100%, and supine CXR had a sensitivity of 36%, compared to the reference standard of CT. There was one false-positive finding for ultrasound in a patient with bullous emphysema seen on CT. The presence of pleural movement detected on ultrasound in patients with CT proven pneumothorax, due to partial apposition of the visceral and parietal pleura has been previously reported.9 Goodman et al9 performed a prospective blinded study comparing ultrasound, CT, and radiographic detection of pneumothorax in 29 patients after lung biopsy. The ultrasound examination was confined to the site of biopsy. Of 13 patients with a pneumothorax, seven cases were detected by ultrasound (sensitivity, 54%); however, CT demonstrated that in some patients pleural apposition was present at the site of the ultrasound scan, with a small pneumothorax seen on CT at the least dependent part of the thorax away from the biopsy site. If ultrasound imaging had been performed throughout the whole hemithorax, the sensitivity would probably have been higher. The specificity was 100%. Our study confirms that it is important to examine the lung apices to detect pneumothoraces, as small pneumothoraces may be false-negative at the lung bases.
Chung et al10 also performed a study in patients after lung biopsy and showed that ultrasound was more reliable for detection of a pneumothorax than supine CXR. They reported a specificity of 94% for both ultrasound and radiographs. One patient received a false diagnosis of pneumothorax by all four observers, and the authors attributed this to a pleural adhesion. No other information is given about this patient, and neither is there any information about comorbidity provided for any of the other patients. Since the patients attended for lung biopsy, it can be assumed that the majority are smokers.
The previously reported studies and our results confirm that ultrasound is very sensitive for the detection of pneumothorax, allowing for the caveat that either the complete hemithorax needs to be examined, or at a minimum the least-dependent part must be examined, as a simple, nonloculated pneumothorax will be detected in this region, as seen in our study. Pneumothoraces if present were always detected in the second intercostal space, and also in the fourth and sixth intercostal space if the pneumothorax was large enough. Unfortunately, lack of pleural sliding is also most commonly seen in COPD patients in the second intercostal space. Since there is least pleural movement at the apices normally, reduction of lung movement with hyperexpansion will be most manifest at the apices. We hypothesize that lack of pleural sliding may be due to lung hyperexpansion and consequent reduction in respiratory movement. There was not a correlation between the lung function tests and a false-positive diagnosis (Table 5). In our series, no false-positive findings were recorded in patients with cystic fibrosis, perhaps because the degree of hyperexpansion was not as severe. Lack of pleural movement was also seen as an isolated finding in the sixth intercostal space in one COPD patient, with pleural sliding movement seen more cranially in the second and fourth intercostal spaces. No clinical or radiologic factor could be determined that would account for this discrepancy. Other than postulating that pleural adhesions were responsible, it is difficult to account for this.
Comparison of the responses from the experienced and inexperienced observers shows that in experienced hands, thoracic ultrasound is a very sensitive test for a pneumothorax. However, a practitioner new to this particular technique, even one with previous familiarity of general ultrasound, has a lower accuracy. Clearly, more experience than merely seeing a video clip of a pneumothorax on ultrasound is needed to achieve a high sensitivity, especially so for practitioners unfamiliar with ultrasound techniques.
The study is limited in that the person performing the sonography was not blinded to the diagnosis; in practical terms, this is not possible to do. In the same vein, it does not mimic the real-life experience for the observers to try to diagnose a pneumothorax using video clips only. Also, this is a small number of patients, and so further prospective studies are needed for assessing the diagnostic accuracy of ultrasound in the diagnosis of pneumothorax in patients with COPD.
In summary, this study confirms that ultrasound is a sensitive test for the detection of pneumothorax, especially in experienced hands. Although thoracic ultrasound is easily learnt and practiced by an inexperienced observer, the diagnostic accuracy is lower than in experienced hands. Patients with COPD can mimic the signs of a pneumothorax on ultrasound, making this test unreliable in these patients. Ultrasound is a reliable test to exclude a pneumothorax; but if results are positive, further imaging should be used to confirm the diagnosis before treatment is initiated, particularly if the lung function of the patient is unknown.
Footnotes
Abbreviation: CXR = chest radiograph/radiography
Received for publication March 22, 2005. Accepted for publication September 15, 2005.
References
This article has been cited by other articles:
|
|
 |
|
  Diagnosis of Pneumothorax with Thoracic Ultrasound Journal Watch Emergency Medicine, April 25, 2006; 2006(425): 18 - 18. [Full Text]
|
|
Read all eLetters
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
