[ Original Research Pulmonary Vascular Disease ]

Lung Ultrasound-Implemented Diagnosis of Acute

Decompensated Heart Failure in the ED
A SIMEU Multicenter Study
Emanuele Pivetta, MD; Alberto Goffi, MD; Enrico Lupia, MD, PhD; Maria Tizzani, MD; Giulio Porrino, MD;
Enrico Ferreri, MD; Giovanni Volpicelli, MD, FCCP; Paolo Balzaretti, MD; Alessandra Banderali, MD;
Antonello Iacobucci, MD; Stefania Locatelli, MD; Giovanna Casoli, MD; Michael B. Stone, MD; Milena M. Maule, PhD;
Ileana Baldi, PhD; Franco Merletti, MD; and Gian Alfonso Cibinel, MD; for the SIMEU Group for Lung Ultrasound in
the Emergency Department in Piedmont

BACKGROUND: Lung ultrasonography (LUS) has emerged as a noninvasive tool for the differ-
ential diagnosis of pulmonary diseases. However, its use for the diagnosis of acute decompen-
sated heart failure (ADHF) still raises some concerns. We tested the hypothesis that an
integrated approach implementing LUS with clinical assessment would have higher diagnostic
accuracy than a standard workup in differentiating ADHF from noncardiogenic dyspnea in
the ED.
METHODS: We conducted a multicenter, prospective cohort study in seven Italian EDs. For
patients presenting with acute dyspnea, the emergency physician was asked to categorize
the diagnosis as ADHF or noncardiogenic dyspnea after (1) the initial clinical assessment and
(2) after performing LUS (“LUS-implemented” diagnosis). All patients also underwent chest
radiography. After discharge, the cause of each patient’s dyspnea was determined by indepen-
dent review of the entire medical record. The diagnostic accuracy of the different approaches
was then compared.
RESULTS: The study enrolled 1,005 patients. The LUS-implemented approach had a signifi-
cantly higher accuracy (sensitivity, 97% [95% CI, 95%-98.3%]; specificity, 97.4% [95% CI,
95.7%-98.6%]) in differentiating ADHF from noncardiac causes of acute dyspnea than the
initial clinical workup (sensitivity, 85.3% [95% CI, 81.8%-88.4%]; specificity, 90% [95% CI,
87.2%-92.4%]), chest radiography alone (sensitivity, 69.5% [95% CI, 65.1%-73.7%]; speci-
ficity, 82.1% [95% CI, 78.6%-85.2%]), and natriuretic peptides (sensitivity, 85% [95% CI,
80.3%-89%]; specificity, 61.7% [95% CI, 54.6%-68.3%]; n 5 486). Net reclassification index
of the LUS-implemented approach compared with standard workup was 19.1%.
CONCLUSIONS: The implementation of LUS with the clinical evaluation may improve accuracy
of ADHF diagnosis in patients presenting to the ED.
TRIAL REGISTRY: Clinicaltrials.gov; No.: NCT01287429; URL: www.clinicaltrials.gov
CHEST 2015; 148(1):202-210

Manuscript received October 21, 2014; revision accepted January 3, AFFILIATIONS: From the Cancer Epidemiology Unit (Drs Pivetta, Maule,
2015; originally published Online First February 5, 2015. Baldi, and Merletti), Department of Medical Sciences, CeRMS and
ABBREVIATIONS: ADHF 5 acute decompensated heart failure; University of Turin, Turin, Italy; Department of Emergency Medicine
AUC 5 area under the curve; BNP 5 brain natriuretic peptide; (Drs Pivetta and Cibinel), “E. Agnelli” General Hospital, Pinerolo, Turin,
CXR 5 chest radiography; IQR 5 interquartile range; IS 5 interstitial Italy; Department of Emergency Medicine (Drs Pivetta and Stone),
syndrome; LUS 5 lung ultrasonography; NRI 5 net reclassification Brigham and Women’s Hospital, Boston, MA; Department of Emergency
improvement; NT-pro-BNP 5 N-terminal pro-brain natriuretic pep- Medicine (Drs Pivetta, Lupia, Tizzani, Porrino, Ferreri, and Locatelli),
tide; ROC 5 receiver operating characteristic A.O.U. Città della Salute e della Scienza di Torino and University of Turin,

202 Original Research [ 1 4 8 # 1 C H E S T J U LY 2 0 1 5 ]

Downloaded From: http://journal.publications.chestnet.org/ by a Bibliotheque De L Universite Laval User on 10/06/2015
 Missed or delayed diagnosis of acute decompensated has been correlated with extravascular lung water
heart failure (ADHF) in the ED is associated with pro- content18-20 and with surrogate markers of pulmonary
longed hospital stay, higher rates of ICU admission, edema,21,22 leading a consensus conference to recom-
higher mortality, and increased costs.1-3 However, mend LUS for the diagnosis of ADHF.15 However,
although the worldwide incidence of heart failure is LUS accuracy may be limited by its low specificity
reaching epidemic proportions,4 the initial diagnostic (as low as 54% in some studies), since B-lines may
workup of ADHF has remained fundamentally be detected in lung diseases with diffuse interstitial
unchanged over decades4-6 and may be heavily under- involvement other than ADHF.10,15,18,21,23 Furthermore,
mined by several factors, such as the poor sensitivity of conflicting results on the correlation of B-lines with
the physical examination, ECG inaccuracy, and unreli- pulmonary artery wedge pressure have been pub-
ability of chest radiography (CXR) findings.1,7,8 lished,19,24,25 which led some authors to raise concerns
on LUS use for ADHF diagnosis.19,23,24 In this study,
Lung ultrasonography (LUS) has emerged as a rapid, we evaluated the performance of a diagnostic approach
noninvasive, bedside tool for the diagnosis of several implementing LUS with the clinical assessment in
pulmonary and pleural diseases.9-17 LUS detection of differentiating ADHF from noncardiac causes of
multiple and diffuse reverberation artifacts (B-lines) acute dyspnea in the ED.

Materials and Methods Participants

Study Design We considered eligible all adult subjects (age . 18 years) present-
ing to the ED with acute dyspnea, defined as either sudden onset of
This was a prospective, observational, multicenter, cohort study con- shortness of breath or increase in the severity of chronic dyspnea in
ducted in seven Italian EDs (two academic, one university-affiliated, the last 48 h. We excluded patients with dyspnea obviously not due to
and four community hospitals). Patients were enrolled from October ADHF (eg, traumatic injury), and those already invasively ventilated at
2010 to September 2012; each center recruited for a 6-month period. the time of evaluation. The presence of an emergency physician with
The protocol was approved by the institutional review board of the expertise in LUS (training performed according to Italian Society of
“Città della Salute e della Scienza di Torino” Hospital (project approval Emergency Medicine [SIMEU] guidelines; . 40 examinations com-
number 2CEI-105/0068893) and by the institutional review boards of pleted) was required for patient enrollment.
the other participant hospitals. All patients gave informed written con-
sent to participate. All patient data, after enrollment, were deidentified. Study Protocol
The study was conducted in accordance with the principles of the Dec- After the initial standard workup, which included past medical history,
laration of Helsinki for clinical research involving human subjects and history of the present illness, physical examination, ECG, and arterial
registered on ClinicalTrials.gov (Identifier: NCT01287429). blood gas analysis,6 the emergency physician responsible for patient
care was asked to categorize the diagnosis as ADHF or noncardiogenic
Turin, Italy; Interdepartmental Division of Critical Care Medicine dyspnea. The diagnosis of ADHF was based on the most updated guide-
(Dr Goffi), University of Toronto, Toronto, ON, Canada; Toronto Western lines available.6,26 In a pragmatic “real world” approach, we decided not
Hospital (Dr Goffi), University Health Network, Toronto, ON, Canada; to define a strict diagnostic protocol nor to reinforce existing guidelines
Department of Medical Sciences (Dr Lupia), University of Turin, Turin, for the diagnosis of ADHF in the recruiting centers. Immediately after
Italy; Department of Emergency Medicine (Dr Volpicelli), “San Luigi clinical workup, the same physician performed LUS, and the new pre-
Gonzaga” University Hospital, Orbassano, Turin, Italy; Department of sumptive etiology, based on the results of both clinical assessment and
Emergency Medicine (Dr Balzaretti), “Ordine Mauriziano” Hospital, LUS findings, was recorded (“LUS-implemented” diagnosis).
Turin, Italy; Department of Emergency Medicine (Dr Banderali), “Cardinal
Massaia” Hospital, Asti, Italy; Department of Emergency Medicine After hospital discharge, an expert emergency physician and a cardi-
(Dr Iacobucci), “Santa Croce e Carle” Hospital, Cuneo, Italy; Department ologist, blinded to LUS results, independently reviewed the entire med-
of Emergency Medicine (Dr Casoli), “Martini” Hospital, Turin, Italy; ical record and indicated the cause of patient’s acute dyspnea (“final
and Department of Cardiac, Thoracic, and Vascular Sciences (Dr Baldi), diagnosis”). In case of disagreement, a third expert emergency physi-
University of Padova, Padova, Italy. cian reviewed the medical records and adjudicated the case.
Partial data from this study were presented at the European Society
of Cardiology 2013 Annual Meeting, August 31-September 4, 2013, The etiology of dyspnea was categorized as a dichotomous variable
Amsterdam, The Netherlands; American College of Emergency (ADHF or noncardiogenic). If both etiologies were concomitantly pre-
Physicians 2012 Scientific Assembly, October 8-11, 2012, Denver, CO; sent, the reviewers were asked to indicate the one they considered more
and the American Institute of Ultrasound in Medicine 2013 Annual relevant in determining patient’s acute dyspnea.
Meeting, April 7-11, 2013, New York, NY.
FUNDING/SUPPORT: The authors have reported to CHEST that no For all patients, a CXR report interpreted by a radiologist was available.
funding was received for this study. To evaluate CXR accuracy, mention of pulmonary venous congestion
CORRESPONDENCE TO: Emanuele Pivetta, MD, Department of Medical and/or of bilateral interstitial and/or alveolar edema was considered
Sciences, Cancer Epidemiology Unit, A.O.U. Città della Salute e della diagnostic for ADHF.
Scienza di Torino, Via Santena 7, 10126, Torino, Italy; e-mail: emanuele.
pivetta@gmail.com In a subgroup of patients, brain natriuretic peptide (BNP) or N-terminal
© 2015 AMERICAN COLLEGE OF CHEST PHYSICIANS. Reproduction of pro-BNP (NT-pro-BNP) levels were measured, although it was not
this article is prohibited without written permission from the American expressly required by the study protocol. The cutoff values for diag-
College of Chest Physicians. See online for more details. nosing ADHF were set at 400 pg/mL for BNP,27 and at 450, 900, and
DOI: 10.1378/chest.14-2608 1,800 pg/mL for patients , 50 years old, between 50 and 75 years old,

journal.publications.chestnet.org 203

Downloaded From: http://journal.publications.chestnet.org/ by a Bibliotheque De L Universite Laval User on 10/06/2015

 and . 75 years of age, respectively, for NT-pro-BNP.28 Finally, two a power of 90% to test an absolute difference of 10% in both sensitivity
emergency physicians with limited LUS training (, 10 examinations and specificity between clinical and LUS-implemented evaluation, with
performed) and a physician expert in LUS blindly reviewed LUS images 5% a error (two-sided test). Descriptive data are presented as mean
from 200 randomly selected patients, and both intra- and interobserver (⫾ SD) or median (with interquartile range [IQR]) for continuous vari-
variability were evaluated. ables, and as numbers and percentages for categorical variables.

Lung Ultrasonography The accuracy of each diagnostic tool is expressed as sensitivity, specific-
Patients were examined using a curvilinear transducer (5-3 MHz), ity, predictive values, and likelihood ratios obtained using 2 3 2 tables.
according to a previously described, six-zone scanning protocol (Fig 1).29 For each test, “positive” and “negative” results were considered the diag-
nosis of ADHF or noncardiac dyspnea, respectively. Receiver operating
Diffuse interstitial syndrome (IS) was defined as the bilateral presence
characteristic (ROC) and area under the curve (AUC) statistics31,32 are
of two or more zones showing the presence of at least three B-lines:
also shown.
vertical, hyperechoic reverberation artifacts extending from the pleu-
ral line to the bottom of the screen.9,15 The accuracy of LUS alone was We used the McNemar test for paired data33 to compare the accuracy
determined by reanalyzing, a posteriori, the sonographic images; the in the detection of ADHF of the different diagnostic tests. Diagnos-
presence of diffuse IS was considered diagnostic for ADHF. tic improvement obtained with LUS implementation to the initial
standard workup was assessed by the net reclassification improvement
Statistical Analysis
(NRI), which estimates the percentage of subjects moving from one
The sample size calculation assumed a prevalence of ADHF ranging clinical diagnosis to the other.34 Intra- and interobserver agreement
from 45% to 55%, with sensitivity and specificity of the clinical assess- was assessed using Cohen k with associated 95% CI. Data were col-
ment alone ranging from 70% to 85% and from 75% to 85%, respec- lected in a Microsoft Excel (Microsoft Inc) spreadsheet, and statisti-
tively.1,3,30 We estimated that a sample size of 915 patients could achieve cal analyses were conducted using Stata 11.0/SE (Stata Corp).

Results agreement between the two expert physicians adjudi-

cating the final diagnosis of ADHF was very high
Patients
(k, 0.93; 95% CI, 0.91–0.95; P , .01). Of the 35 discor-
Of 1,007 patients eligible for this study, two denied dant cases, 16 were categorized as ADHF by the third
consent to enrollment; therefore, 1,005 patients expert.
were included in the analyses. Of these, 544 (54.1%)
were enrolled in community hospitals and 461 Admissions were 821 (81.7%), 176 patients were dis-
(45.9%) in academic or university-affiliated centers. charged home, and eight patients died in the ED.
Sixty-two emergency physicians participated in the Among admitted patients, 602 were admitted to general
study, each enrolling a median number of 44 patients internal medicine/cardiology units, 175 to ICU or high-
(IQR, 22). dependency units, 12 to surgical units, and 32 to short-
stay units (with five patients subsequently admitted to
Final diagnosis was ADHF for 463 patients (46%) internal medicine units, and 27 discharged home).
and noncardiac dyspnea for 542 (54%). Interrater
Table 1 shows patient demographic and clinical data.
Table 2 reports symptoms associated with dyspnea and
findings detected during the initial clinical assessment.
Outcomes
Figures 2 and 3 show accuracy, ROC, and AUC of the
different diagnostic approaches evaluated. Clinical
workup and LUS alone had relatively high accuracy in
the identification of ADHF, whereas CXR showed only
moderate accuracy. Sensitivity and specificity were sig-
nificantly higher for LUS alone compared with CXR
(P , .01). The LUS-implemented protocol had the
highest accuracy, with a significant increase in sensitivity
Figure 1 – Six-zone scanning protocol used for lung ultrasonography and specificity in comparison with clinical workup
evaluation. The anterior chest wall was divided into three zones. Two
zones were located anteriorly, in the second intercostal space on the mid- alone (P , .01) (Fig 2). The NRI of LUS-implemented
clavicular line, and in the fourth intercostal space on the anterior axil- approach compared with standard clinical workup was
lary line, respectively. The third zone was located laterally in the fifth
intercostal space on the mid-axillary line. An additional scan was per- 19.1% (95% CI, 14.6%-23.6%; P , .01) (Table 3). The
formed on the posterior axillary line at the base of the lungs to assess accuracy of BNP/NT-pro-BNP values, available in a
for the presence of pleural effusion. AAL 5 anterior axillary line;
MAL 5 mid-axillary line; MC 5 mid-clavicular line. (The patient pro-
subgroup of 486 patients, was significantly lower than
vided written consent for the use of this photograph.) that of the LUS-implemented approach (P , .01) (Fig 2).

204 Original Research [ 1 4 8 # 1 C H E S T J U LY 2 0 1 5 ]

Downloaded From: http://journal.publications.chestnet.org/ by a Bibliotheque De L Universite Laval User on 10/06/2015
 TABLE 1 ] Demographic Characteristics and Medications of Patients Enrolled
Characteristics ADHF (n 5 463) Noncardiac Dyspnea (n 5 542) All Patients (N 5 1,005) P Valuea
Age, median (IQR), y 78 (11) 76 (13) 77 (13) , .01
Women 248 (53.6) 216 (39.8) 464 (46.2) , .01
Baseline characteristics
Tobacco useb 166 (35.8) 254 (46.9) 420 (41.8) , .01
COPD 125 (27.2) 284 (52.2) 409 (40.7) , .01
Asthma 3 (0.7) 35 (6.5) 38 (3.8) , .01
Interstitial lung disease 19 (4.1) 38 (7) 57 (5.7) , .05
Hypertensionc 335 (72.4) 303 (55.9) 638 (63.5) , .01
Congestive heart failure c 148 (32) 55 (10.2) 203 (20.2) , .01
Ischemic cardiomyopathy/CAD 155 (33.5) 104 (19.2) 259 (25.8) , .01
Other cardiomyopathies 169 (36.5) 71 (13.1) 240 (23.9) , .01
Diabetesc 169 (36.5) 112 (20.7) 281 (27.9) , .01
Arrhythmia d 161 (34.8) 106 (19.6) 267 (26.5) , .01
Dyslipidemiac 100 (21.6) 60 (11.1) 160 (15.9) , .01
Cerebrovascular accident c 37 (8) 42 (7.8) 79 (7.9) . .05
CKD/chronic dialysise 108 (23.3) 61 (11.3) 169 (16.8) , .01
Neoplastic disease c 42 (9.1) 90 (16.6) 132 (13.1) , .01
Thromboembolic disorder 11 (2.4) 27 (5) 38 (3.8) . .05
Medications
Diuretics 314 (67.8) 234 (43.2) 548 (54.5) , .01
b-Blockers 206 (44.5) 91 (16.8) 297 (29.5) , .01
ACE inhibitors 248 (53.6) 184 (33.9) 432 (43) , .01
Antiplatelet agentsf 206 (44.5) 182 (33.6) 388 (38.6) , .01
Anticoagulantsg 106 (22.9) 68 (12.5) 174 (17.3) , .01
Bronchodilators 92 (19.9) 270 (49.8) 362 (36) , .01
Antidiabetic agents/insulin 161 (34.8) 111 (20.5) 272 (27.1) , .01
Steroids 39 (8.4) 98 (18.1) 137 (13.6) , .01
Antiarrhythmic agents 101 (21.8) 66 (12.2) 167 (16.6) , .01
Home oxygen 38 (8.2) 80 (14.8) 118 (11.7) , .01

Data given as No. (%) unless otherwise indicated. ACE 5 angiotensin-converting-enzyme; ADHF 5 acute decompensated heart failure; CAD 5 coronary
artery disease; CKD 5 chronic kidney disease; IQR 5 interquartile range.
ax2 test for categorical variables, or Mann-Whitney U test for continuous variables.

bCurrent or remote use.

cAny type and/or grade of disorder/disease.

dAny type of cardiac rhythm disorders (eg, atrial fibrillation, paroxysmal, persistent or permanent; atrial flutter; sick sinus syndrome; atrioventricular

blocks; supraventricular paroxysmal tachycardia).

eCKD defined as chronic renal failure with creatinine level . 2 mg/dL (177 mmol/L).

fAcetylsalicylic acid, clopidogrel, prasugrel, ticagrelor, ticlopidine, tirofiban, and dypiridamole.

gWarfarin, acenocoumarol, any type of heparin, fondaparinux.

Finally, when two emergency physicians with limited LUS Discussion

training (, 10 examinations performed) and a physician In this study, we found that, in adult subjects with acute
expert in LUS blindly reviewed LUS images from 200 ran- dyspnea in the ED, a LUS-implemented approach had
domly selected patients, the k statistic for agreement for higher diagnostic accuracy than other commonly used
the diagnosis of IS was 0.94 (95% CI, 0.89-0.98; P , .01). diagnostic tools (ie, initial clinical workup, CXR, LUS
Intraobserver agreement was 0.97 for the expert operator alone, and BNP/NT-pro-BNP levels) in differentiating
(95% CI, 0.95-0.99; P , .01) and 0.92 for the physicians ADHF from noncardiac causes of acute dyspnea. To
with limited LUS training (95% CI, 0.88-0.96; P , .01). overcome LUS limitations in ADHF diagnosis,1,6,35,36

journal.publications.chestnet.org 205

Downloaded From: http://journal.publications.chestnet.org/ by a Bibliotheque De L Universite Laval User on 10/06/2015

 TABLE 2 ] Symptoms Associated With Acute Dyspnea and Clinical Findings Detected at the Time of Patient
Presentation in the ED
Noncardiac Dyspnea All Patients
Symptoms/Findings ADHF (n 5 463) (n 5 542) (N 5 1,005) P Valuea
Associated symptoms
Feverb 28 (6.1) 164 (30.3) 192 (19.1) , .01
Dry and/or productive cough 77 (16.6) 278 (51.3) 355 (35.3) , .01
Chest pain 54 (11.7) 49 (9) 103 (10.3) . .05
Palpitation 18 (3.9) 10 (1.9) 28 (2.8) .05
Physical examination findings, mean (SD)
Heart rate, beats/min 95.2 (22.7) 98 (21.4) 96.7 (22) , .05
Systolic BP, mm Hg 147.6 (26.8) 138.6 (24.2) 142.6 (25.8) , .01
Diastolic BP, mm Hg 81.3 (15.4) 77.8 (13.7) 79.4 (14.6) , .01
PaO2/FIO2 ratioc 271.5 (72.8) 283.4 (84.5) 278 (79.6) , .05
Respiratory rate, breaths/min 28.3 (7.9) 28.3 (8.1) 28.3 (7.9) . .05
Temperature, °C 36.3 (0.6) 36.9 (1.4) 36.6 (1.2) , .01
Wheezing 97 (21) 210 (38.8) 307 (30.6) , .01
Rales 375 (81) 286 (52.8) 661 (65.8) , .01
Peripheral edema 253 (54.6) 117 (21.6) 370 (36.8) , .01
Noninvasive mechanical ventilationd in the ED 97 (20.9) 48 (8.9) 145 (14.4) , .01

Data given as No. (%) unless otherwise indicated. See Table 1 legend for expansion of abbreviations.
x test for categorical variables, Mann-Whitney U test, or Student t test for continuous variables.
a 2

bTympanic temperature . 38.3°C.

cThe ratio of PaO and FIO was calculated using PaO (mm Hg) measured at the time of the first arterial blood gas analysis and FIO provided, as
2 2 2 2
reported in the case report form. In 83.1% of cases, the blood gas analysis was performed with patients breathing room air (FIO2, 0.21).
dAny type of noninvasive mechanical ventilation.

we adopted an integrated approach that paired LUS The superiority of LUS in the detection of diffuse IS
with the pretest probability of the clinical assessment. may be at least partially related to the more rapid sono-
Moreover, we strictly defined diffuse IS as the bilateral graphic detection of signs of pulmonary congestion, which
presence of two or more positive zones, to decrease have been shown in experimental studies to appear only
false-positive diagnoses due to the appearance of B-lines 15 min after the induction of pulmonary injury and to
in several primary lung diseases (eg, pneumonia, atelec- even precede changes in Pao2/Fio2 ratio.40
tasis). By using this strategy, we obtained sensitivity,
The LUS-implemented approach resulted in absolute
specificity, and AUC values for the diagnosis of ADHF
increases in sensitivity and specificity of 6.5% and 3.9%,
as high as 97.0%, 97.4%, and 0.972, respectively.
respectively, when compared with LUS-alone, although,
In agreement with previous data,1,3,37,38 the accuracy of in our study, the sensitivity, specificity, and AUC of LUS
the initial clinical workup for ADHF diagnosis was only alone were higher than in previous studies.10,11,41 This
moderate. The LUS-implemented approach increased discrepancy may be related to the difference in clinical
the sensitivity and specificity of clinical assessment by setting21,41 or to the timing of LUS performance.10
11.7% and 7.4%, respectively, with an NRI of 19.1%,
In a subgroup of patients (486 of 1,005), we evaluated
which shows a relevant diagnostic impact of LUS in this
the diagnostic accuracy of serum concentrations of
setting.
BNP/NT-pro-BNP, extensively studied biomarkers of
CXR is still considered a fundamental component of the ADHF.1,6,36 In agreement with previous studies,6,11,28,38
initial assessment of a patient with suspected ADHF.39 natriuretic peptides had higher sensitivity (85.0%)
However, in our study, CXR sensitivity, although higher than specificity (61.7%), and a moderate AUC (0.733).
than that reported in other studies,1,7 was as low as It is known that this relatively low specificity is due
69.5%, confirming that the absence of pulmonary to the increase of BNP/NT-pro-BNP levels in several
venous congestion/interstitial edema/alveolar edema other diseases, such as pulmonary embolism, left ven-
is not sufficient to exclude the diagnosis of ADHF.1 tricle dysfunction without ADHF, atrial fibrillation,

206 Original Research [ 1 4 8 # 1 C H E S T J U LY 2 0 1 5 ]

Downloaded From: http://journal.publications.chestnet.org/ by a Bibliotheque De L Universite Laval User on 10/06/2015
 Figure 2 – Diagnostic accuracy of the different approaches to acute dyspnea in the ED. BNP 5 brain natriuretic peptide; LR2 5 negative likelihood ratio;
LR1 5 positive likelihood ratio; LUS 5 lung ultrasonography; NPV 5 negative predictive value; NT-pro-BNP 5 N-terminal pro-brain natriuretic peptide;
PPV 5 positive predictive value.

and cor pulmonale.42-44 In our study, both LUS- However, the differences in the clinical setting, LUS
implemented approach and LUS alone outperformed timing, and/or BNP/NT-pro-BNP sampling may at least
BNP/NT-pro-BNP values in differentiating ADHF from partially account for this discrepancy. Therefore, it
noncardiac causes of dyspnea. Previously published studies can be hypothesized that detection of sonographic IS
reported results apparently not consistent with ours.11,35,41 may be useful in recognizing those conditions where

Figure 3 – A, Receiver operating characteristic (ROC) curve comparing accuracy of clinical workup, LUS-implemented diagnosis, LUS alone, and chest
radiography. B, ROC curve comparing accuracy of clinical workup, LUS-implemented diagnosis, LUS alone, and BNP/NT-pro-BNP levels. See Figure 2
legend for expansion of other abbreviations.

journal.publications.chestnet.org 207

Downloaded From: http://journal.publications.chestnet.org/ by a Bibliotheque De L Universite Laval User on 10/06/2015

 TABLE 3 ] Net Reclassification Improvementa
LUS-Implemented LUS-Implemented
(Post-LUS) (Post-LUS)
Noncardiac Patients With Noncardiac
Patient With ADHF Dyspnea ADHF Total Noncardiac Dyspnea Dyspnea ADHF Total
Clinical workup Noncardiac 5 63 68 Clinical workup Noncardiac 481 7 488
(pre-LUS) dyspnea (pre-LUS) dyspnea
ADHF 9 386 395 ADHF 47 7 54
Total 14 449 463 Total 528 14 542

LUS 5 lung ultrasonography. See Table 1 legend for expansion of other abbreviation.
aNet reclassification improvement was calculated as ([63 / 463] 2 [9 / 463]) 1 ([47 / 542] 2 [7 / 542]) 5 19.1% (95% CI, 14.6%-23.6%; P , .01).

myocardial wall stress is increased but extravascular the ED in the vast majority of patients enrolled in our
lung water content remains unchanged (eg, pulmonary study. LUS was delayed in 16 patients, due to lack of
embolism, atrial fibrillation, and cor pulmonale). In availability of the ultrasound system, but the time
addition, LUS may be helpful in patients with only between ED admission and LUS performance was
slightly elevated BNP/NT-pro-BNP levels (“grey zone”), still , 1 h.
or in situations where this analysis is not available in a
Several limits have to be taken into account in inter-
timely manner (eg, remote areas, peripheral EDs, low-
preting our results. First, we did not enroll consecu-
income countries).
tively all patients presenting to the ED with acute
A strength of our study may be considered the high dyspnea. Moreover, patient enrollment required the
degree of generalizability. We enrolled 1,005 patients presence of an emergency physician with expertise in
with acute dyspnea in seven EDs, from both academic LUS. Therefore, we cannot exclude a selection bias
and community hospitals. As in previous similar leading to an overestimation of LUS accuracy. How-
studies,7,11,45 patients were usually elderly, with several ever, patients’ characteristics, consistent with previous
comorbidities, and taking multiple medications chroni- studies, the number of centers and of physicians par-
cally (Table 1). Pragmatically, the physicians were not ticipating in the study, and the physicians’ heteroge-
limited by a strict diagnostic protocol, but were free neous expertise in LUS make this possibility unlikely.
to evaluate patients according to real-world practices. Second, a common problem in studies on patients with
Moreover, they had heterogeneous clinical and LUS ADHF is the lack of a standard criterion to determine
expertise, ranging from residents to experienced attending the final diagnosis. As others have done,11,38 we used
physicians. the independent review of the medical records by two
expert physicians, with a third physician reviewing
Finally, we found a very good intra- and interobserver discordant cases. Despite the clear limitation of such
agreement between inexperienced operators and a phy- analysis, we observed a high interrater agreement. An
sician expert in LUS when interpreting LUS images. additional limitation exists, since the same emergency
These findings confirm previous results,10,11,29 indicating physician performed both the initial clinical workup
that LUS can be easily learned and its interpretation is and LUS, and so was not blinded to the results of clin-
highly reproducible. ical workup. Therefore, LUS findings may have not
These results are of particular interest when considering been completely independent from the clinical diagno-
sis, and the diagnostic accuracy of the LUS-implemented
that echocardiography, the recommended test for the
approach could be subject to bias. Even the final diag-
assessment of patients presenting with suspected ADHF
nosis, although the reviewers were blinded to LUS
and acute onset of symptoms, requires extensive training,
results, may have been indirectly affected by this bias,
especially to assess diastolic dysfunction. It also is often
since the therapeutic management of the patients may
not readily available in emergency setting.6
have been influenced by the LUS results. Additionally,
Although we did not specifically collect data on the in patients with multiple concomitant causes of dyspnea
time elapsed between onset of symptoms and ED referral (eg, pneumonia and ADHF), we asked both investigators
or between ED admission and LUS performance, LUS and reviewers to indicate the one they considered more
was performed within 40 min from presentation to relevant in determining the patient’s acute dyspnea.

208 Original Research [ 1 4 8 # 1 C H E S T J U LY 2 0 1 5 ]

Downloaded From: http://journal.publications.chestnet.org/ by a Bibliotheque De L Universite Laval User on 10/06/2015
 Therefore, we are not able to assess the diagnostic accu- therefore, it is highly unlikely that a less accurate test
racy of the LUS-implemented approach in those would have outperformed a more accurate test when
patients with acute dyspnea with mixed etiology. Finally, integrated with the clinical assessment. In conclusion,
the study protocol did not allow us to compare the our study demonstrates that the implementation of
accuracy of the LUS-implemented approach with a LUS with the clinical assessment improves the diag-
CXR-implemented approach, which is the standard nostic accuracy for ADHF and has a significant diag-
diagnostic approach currently used. However, LUS nostic impact in adult patients presenting to the ED
alone was significantly more accurate than CXR; with acute dyspnea.

Acknowledgments Livia Ausiello, MD; Letizia Barutta, MD; pensated heart failure. Ann Emerg Med.
Emanuele Bernardi, MD; Alessia Bono, MD; 2006;47(1):13-18.
Author contributions: E. P., A. G., E. L., Daniela Forno, MD; Alessandro Lamorte, MD; 8. Gillespie ND, McNeill G, Pringle T,
F. M., and G. A. C. had full access to all of Davide Lison, MD; Bartolomeo Lorenzati, MD; Ogston S, Struthers AD, Pringle SD.
the data in the study and take responsibility Elena Maggio, MD; Ilaria Masi, MD; Matteo Cross sectional study of contribution of
for the integrity of the data and the accuracy Maggiorotto, MD; Giulia Novelli, MD; clinical assessment and simple cardiac
of the data analysis. E. P. served as principal Francesco Panero, MD; Massimo Perotto, MD; investigations to diagnosis of left ven-
author. E. P., A. G., E. L., M. M. M., I. B., Marco Ravazzoli, MD; Elisa Saglio, MD; tricular systolic dysfunction in patients
F. M., and G. A. C. contributed to the study Flavia Soardo, MD; Alessandra Tizzani, MD; admitted with acute dyspnoea. BMJ.
concept; E. P., A. G., E. L., M. T., G. P., E. F., Pietro Tizzani, MD; Mattia Tullio, MD; Marco 1997;314(7085):936-940.
G. V., P. B., A. B., A. I., S. L., G. C., M. M. M., Ulla, MD; and Elisa Romagnoli, MD. 9. Lichtenstein DA, Mezière GA. Relevance
F. M., and G. A. C. contributed to the study of lung ultrasound in the diagnosis of
design and data acquisition; E. P., A. G., E. L., Other contributions: We would like to thank acute respiratory failure: the BLUE pro-
M. M. M., I. B., and F. M. contributed to all the colleagues of participating emer- tocol. Chest. 2008;134(1):117-125.
statistical analyses; E. P., A. G., E. L., M. M. M., gency medicine units for their work and
collaboration. 10. Volpicelli G, Mussa A, Garofalo G, et al.
and I. B. contributed to preparation of Bedside lung ultrasound in the assess-
figures and tables; E. P., A. G., E. L., M. T., ment of alveolar-interstitial syndrome.
G. P., E. F., G. V., P. B., A. B., A. I., S. L., G. C., References Am J Emerg Med. 2006;24(6):689-696.
M. B. S., M. M. M., I. B., F. M., and G. A. C. 1. Wang CS, FitzGerald JM, Schulzer M,
contributed to interpretation of the results 11. Liteplo AS, Marill KA, Villen T, et al.
Mak E, Ayas NT. Does this dyspneic Emergency thoracic ultrasound in the
and to revising the manuscript; and E. P., A. G., patient in the emergency department differentiation of the etiology of short-
M. B. S., M. M. M., I. B., F. M., and G. A. C. have congestive heart failure? JAMA. ness of breath (ETUDES): sonographic
contributed to writing the manuscript. 2005;294(15):1944-1956. B-lines and N-terminal pro-brain-type
Financial/nonfinancial disclosures: The 2. Peacock WF, Braunwald E, Abraham W, natriuretic peptide in diagnosing con-
authors have reported to CHEST that no et al. National Heart, Lung, and Blood gestive heart failure. Acad Emerg Med.
potential conflicts of interest exist with any Institute working group on emer- 2009;16(3):201-210.
companies/organizations whose products or gency department management of 12. Grimberg A, Shigueoka DC, Atallah
services may be discussed in this article. acute heart failure: research challenges AN, Ajzen S, Iared W. Diagnostic accu-
and opportunities. J Am Coll Cardiol. racy of sonography for pleural effusion:
Collaborators: Members of the Group for 2010;56(5):343-351. systematic review. Sao Paulo Med J.
Lung Ultrasound in the Emergency Depart- 3. Ray P, Birolleau S, Lefort Y, et al. Acute 2010;128(2):90-95.
ment in Piedmont, Italy, are the following: respiratory failure in the elderly: etiology, 13. Reissig A, Copetti R, Mathis G, et al.
Paolo Baron, MD; Stefania Battista, MD; emergency diagnosis and prognosis. Lung ultrasound in the diagnosis and
Giuseppina Buonafede, MD; Valeria Busso, MD; Crit Care. 2006;10(3):R82. follow-up of community-acquired
Andrea Conterno, MD; Paola Del Rizzo, MD; 4. Redfield MM. Heart failure—an epidemic pneumonia: a prospective, multicenter,
Patrizia Ferrera, MD; Paolo Fascio Pecetto, MD; of uncertain proportions. N Engl J Med. diagnostic accuracy study. Chest.
Corrado Moiraghi, MD; Fulvio Morello, MD; 2002;347(18):1442-1444. 2012;142(4):965-972.
Fabio Steri, MD; Giovannino Ciccone, MD; 14. Picano E, Frassi F, Agricola E, Gligorova S,
5. McCullough PA, Nowak RM, McCord J,
Cosimo Calasso, MD; Mimma A Caserta, MD; Gargani L, Mottola G. Ultrasound
et al. B-type natriuretic peptide and
Marina Civita, MD; Carmen Condo’, MD; lung comets: a clinically useful sign
clinical judgment in emergency diagno-
Vittorio D’Alessandro, MD; Sara Del Colle, MD; sis of heart failure: analysis from Breath- of extravascular lung water. J Am Soc
Stefania Ferrero, MD; Giulietta Griot, MD; ing Not Properly (BNP) Multinational Echocardiogr. 2006;19(3):356-363.
Emanuela Laurita, MD, PhD; Alberto Study. Circulation. 2002;106(4):416-422. 15. Volpicelli G, Elbarbary M, Blaivas M,
Lazzero, MD; Francesca Lo Curto, MD; et al; International Liaison Committee
6. McMurray JJ, Adamopoulos S, Anker
Marianna Michelazzo, MD; Vincenza on Lung Ultrasound (ILC-LUS) for
SD, et al; ESC Committee for Practice
Nicosia, MD; Nicola Palmari, MD; Alberto International Consensus Conference
Guidelines. ESC Guidelines for the
Ricchiardi, MD; Andrea Rolfo, MD; Roberto diagnosis and treatment of acute and on Lung Ultrasound (ICC-LUS). Inter-
Rostagno, MD; Fabrizio Bar, MD; Enrico chronic heart failure 2012: the Task national evidence-based recommenda-
Boero, MD; Mauro Frascisco, MD; Ilaria Force for the Diagnosis and Treatment tions for point-of-care lung ultrasound.
Micossi, MD; Alessandro Mussa, MD; of Acute and Chronic Heart Failure 2012 Intensive Care Med. 2012;38(4):577-591.
Valerio Stefanone, MD; Renzo Agricola, MD; of the European Society of Cardiology. 16. Alrajab S, Youssef AM, Akkus NI,
Gabriele Cordero, MD; Federica Corradi, MD; Developed in collaboration with the Caldito G. Pleural ultrasonography
Cristina Runzo, MD; Aldo Soragna, MD; Heart Failure Association (HFA) of the versus chest radiography for the diag-
Daniela Sciullo, MD; Domenico Vercillo, MD; ESC. Eur Heart J. 2012;33(14):1787-1847. nosis of pneumothorax: review of the
Attilio Allione, MD; Nicoletta Artana, MD; 7. Collins SP, Lindsell CJ, Storrow AB, literature and meta-analysis. Crit Care.
Fabrizio Corsini, MD; Luca Dutto, MD; Abraham WT; ADHERE Scientific 2013;17(5):R208.
Giuseppe Lauria, MD; Teresa Morgillo, MD; Advisory Committee, Investigators and 17. Noble VE, Murray AF, Capp R, Sylvia-
Bruno Tartaglino, MD; Daniela Bergandi, MD; Study Group. Prevalence of negative Reardon MH, Steele DJ, Liteplo A.
Ilaria Cassetta, MD; Clotilde Masera, MD; chest radiography results in the emer- Ultrasound assessment for extravascular
Mario Garrone, MD; Gianluca Ghiselli, MD; gency department patient with decom- lung water in patients undergoing

journal.publications.chestnet.org 209

Downloaded From: http://journal.publications.chestnet.org/ by a Bibliotheque De L Universite Laval User on 10/06/2015

 hemodialysis. Time course for resolution. 27. Maisel A. B-type natriuretic peptide 38. Maisel AS, Krishnaswamy P, Nowak
Chest. 2009;135(6):1433-1439. levels: diagnostic and prognostic in RM, et al; Breathing Not Properly
18. Jambrik Z, Monti S, Coppola V, et al. congestive heart failure: what’s next? Multinational Study Investigators.
Usefulness of ultrasound lung comets Circulation. 2002;105(20):2328-2331. Rapid measurement of B-type natri-
as a nonradiologic sign of extra- 28. Januzzi JL, van Kimmenade R, uretic peptide in the emergency diag-
vascular lung water. Am J Cardiol. Lainchbury J, et al. NT-proBNP testing nosis of heart failure. N Engl J Med.
2004;93(10):1265-1270. for diagnosis and short-term prognosis 2002;347(3):161-167.
19. Agricola E, Bove T, Oppizzi M, et al. in acute destabilized heart failure: an 39. Peacock WF, Fonarow GC, Ander DS,
“Ultrasound comet-tail images”: a international pooled analysis of 1256 et al. Society of Chest Pain Centers rec-
marker of pulmonary edema: a com- patients: the International Collaborative ommendations for the evaluation and
parative study with wedge pressure of NT-proBNP Study. Eur Heart J. 2006; management of the observation stay
and extravascular lung water. Chest. 27(3):330-337. acute heart failure patient-parts 1-6.
2005;127(5):1690-1695. 29. Cibinel GA, Casoli G, Elia F, et al. Acute Card Care. 2009;11(1):3-42.
20. Baldi G, Gargani L, Abramo A, et al. Diagnostic accuracy and reproducibility 40. Gargani L, Lionetti V, Di Cristofano C,
Lung water assessment by lung ultraso- of pleural and lung ultrasound in dis- Bevilacqua G, Recchia FA, Picano E.
nography in intensive care: a pilot study. criminating cardiogenic causes of acute Early detection of acute lung injury
Intensive Care Med. 2013;39(1):74-84. dyspnea in the emergency department. uncoupled to hypoxemia in pigs using
Intern Emerg Med. 2012;7(1):65-70. ultrasound lung comets. Crit Care Med.
21. Lichtenstein D, Mézière G, Biderman P,
30. Delerme S, Ray P. Acute respiratory 2007;35(12):2769-2774.
Gepner A, Barré O. The comet-tail
artifact. An ultrasound sign of alveolar- failure in the elderly: diagnosis and prog- 41. Gargani L, Frassi F, Soldati G, Tesorio P,
interstitial syndrome. Am J Respir Crit nosis. Age Ageing. 2008;37(3):251-257. Gheorghiade M, Picano E. Ultrasound
Care Med. 1997;156(5):1640-1646. 31. Lusted LB. Signal detectability and lung comets for the differential diagnosis
22. Xirouchaki N, Magkanas E, Vaporidi K, medical decision-making. Science. of acute cardiogenic dyspnoea: a com-
et al. Lung ultrasound in critically ill 1971;171(3977):1217-1219. parison with natriuretic peptides. Eur J
patients: comparison with bedside Heart Fail. 2008;10(1):70-77.
32. DeLong ER, DeLong DM, Clarke-
chest radiography. Intensive Care Med. Pearson DL. Comparing the areas under 42. Januzzi JL Jr, Chen-Tournoux AA,
2011;37(9):1488-1493. two or more correlated receiver operating Moe G. Amino-terminal pro-B-type
23. Sperandeo M, Varriale A, Sperandeo G, characteristic curves: a nonparametric natriuretic peptide testing for the diag-
et al. Assessment of ultrasound acoustic approach. Biometrics. 1988;44(3):837-845. nosis or exclusion of heart failure in
artifacts in patients with acute dyspnea: patients with acute symptoms. Am J
33. McNemar Q. Note on the sampling error Cardiol. 2008;101(3A):29-38.
a multicenter study. Acta Radiol. 2012; of the difference between correlated pro-
53(8):885-892. portions or percentages. Psychometrika. 43. Maisel AS, Clopton P, Krishnaswamy P,
24. Lichtenstein DA, Mezière GA, Lagoueyte 1947;12(2):153-157. et al; BNP Multinational Study Inves-
JF, Biderman P, Goldstein I, Gepner A. tigators. Impact of age, race, and sex on
34. Pencina MJ, D’Agostino RB Sr, the ability of B-type natriuretic peptide to
A-lines and B-lines: lung ultrasound as D’Agostino RB Jr, et al. Evaluating the
a bedside tool for predicting pulmonary aid in the emergency diagnosis of heart
added predictive ability of a new marker: failure: results from the Breathing Not
artery occlusion pressure in the critically from area under the ROC curve to
ill. Chest. 2009;136(4):1014-1020. Properly (BNP) multinational study.
reclassification and beyond. Stat Med. Am Heart J. 2004;147(6):1078-1084.
25. Volpicelli G, Skurzak S, Boero E, et al. 2008;27(2):157-172.
Lung ultrasound predicts well extravas- 44. Morrison LK, Harrison A, Krishnaswamy
35. Anderson KL, Jenq KY, Fields JM, P, Kazanegra R, Clopton P, Maisel A.
cular lung water but is of limited useful-
Panebianco NL, Dean AJ. Diagnosing Utility of a rapid B-natriuretic peptide
ness in the prediction of wedge pressure.
heart failure among acutely dyspneic assay in differentiating congestive heart
Anesthesiology. 2014;121(2):320-327.
patients with cardiac, inferior vena cava, failure from lung disease in patients pre-
26. Jessup M, Abraham WT, Casey DE, and lung ultrasonography. Am J Emerg senting with dyspnea. J Am Coll Cardiol.
et al. 2009 focused update: ACCF/ Med. 2013;31(8):1208-1214. 2002;39(2):202-209.
AHA Guidelines for the Diagnosis and
Management of Heart Failure in Adults: 36. Knudsen CW, Omland T, Clopton P, et al. 45. Eurlings LW, Sanders-van Wijk S,
a report of the American College of Diagnostic value of B-Type natriuretic van Kimmenade R, et al. Multimarker
Cardiology Foundation/American Heart peptide and chest radiographic findings strategy for short-term risk assessment
Association Task Force on Practice in patients with acute dyspnea. Am J in patients with dyspnea in the emer-
Guidelines: developed in collaboration Med. 2004;116(6):363-368. gency department: the MARKED (Multi
with the International Society for Heart 37. García X, Simon P, Guyette FX, et al. mARKer Emergency Dyspnea)-risk
and Lung Transplantation. Circulation. Noninvasive assessment of acute dyspnea score. J Am Coll Cardiol. 2012;60(17):
2009;119(14):1977-2016. in the ED. Chest. 2013;144(2):610-615. 1668-1677.

210 Original Research [ 1 4 8 # 1 C H E S T J U LY 2 0 1 5 ]

Downloaded From: http://journal.publications.chestnet.org/ by a Bibliotheque De L Universite Laval User on 10/06/2015