Objective
Pulmonary embolism (PE) is a common diagnosis with significant mortality if not appropriately treated. The use of transthoracic echocardiography in patients with PE is common; however, its diagnostic capabilities in this use are unclear. With the increased use of ultrasonography in medical settings, it is important to understand the strengths and limitations of echocardiography for the diagnosis of PE.
Methods
We conducted a systematic review of PubMed, CINAHL, and EMBASE through 2016 for articles assessing the diagnostic accuracy of transthoracic echocardiography for PE. Two authors independently abstracted relevant data from the studies. We assessed quality using the QUADAS-2 tool for diagnostic studies.
Results
Undefined “right heart strain” was the most common sign used, and it had a sensitivity of 53% (95% CI, 45%–61%) and a specificity of 83% (95% CI, 74%–90%). Eleven other distinct signs were identified: ventricle size ratio, abnormal septal motion, tricuspid regurgitation, 60/60 sign, McConnell’s sign, right heart thrombus, right ventricle hypokinesis, pulmonary hypertension, right ventricular end-diastolic diameter, tricuspid annular plane systolic excursion, and right ventricular systolic pressure.
Conclusions
Studies show a consistently high specificity and low sensitivity for echocardiography in the diagnosis of PE, making it potentially adequate as a rule-in test at the bedside in critical care settings such as the emergency department and intensive care unit for patients with a suspicion of PE, especially those unable to get other confirmatory studies. Future research may continue to clarify the role of bedside echocardiography in conjunction with other tests and imaging in the overall management of PE.
Highlights
- •
The current article represents one of the most thorough meta-analyses of transthoracic echocardiography for pulmonary embolism to date and confirms a moderate sensitivity and high specificity across multiple echocardiographic parameters.
- •
The test characteristics of echocardiography for pulmonary embolism are similar between cardiologists, cardiac sonographers, and physicians trained in point of care ultrasound.
- •
Overall, echocardiography should not be used alone as test to rule out pulmonary embolism.
- •
Findings of right heart strain on transthoracic echocardiography can guide management, especially in scenarios where confirmatory studies are either not available (resource-limited settings) or feasible (critically ill patients).
Venous thromboembolism (VTE), a clinical syndrome involving venous thrombosis followed by distant embolism, is the third most common cause of cardiovascular death and the leading cause of preventable hospital death. Estimates of the incidence of pulmonary embolism (PE) and VTE range from 1.05 to 1.48 per 1,000 patient-years. Of the nearly 300,000 patients who experience VTE each year, 15%–25% present with sudden death or die within 30 days of diagnosis. Approximately half of all PEs are diagnosed in the emergency setting, and prompt diagnosis and treatment have been shown to improve outcomes and survival.
Echocardiographic signs of right heart (RH) strain are often used as prognostic markers in patients with acute PE ; however, the ability of echocardiography to diagnose PE is not well understood. It is suspected to be relatively specific for PE in patients without known chronic cardiorespiratory disease. Recent guidelines by the European Society of Cardiology do not recommend echocardiography as part of the diagnostic workup in non-high-risk patients; however, signs of right ventricular (RV) overload in high-risk patients without another leading diagnosis justify emergent treatment if computed tomography (CT) or other confirmatory test is not immediately available.
Most systematic reviews on the diagnosis of PE have not focused on echocardiography. The last systematic review that reported test characteristics on echocardiography for diagnosing PE was done in 2005 and found only two studies on the topic. It is possible that the test characteristics have improved in light of the increased use of ultrasonography in the emergency department (ED) and the improvements in technology in recent years. Therefore, we sought to do a systematic review and meta-analysis to summarize the test characteristics of echocardiography using gold standards of CT and surgery in the diagnosis of PE in adult patients presenting to any health care provider with signs or symptoms of PE.
Methods
Search Techniques
We conducted a thorough and systematic literature search of English language articles published on transthoracic echocardiography used for the diagnosis of PE from January 1, 1980, to the end of year 2014 (exact search dates in Supplemental Appendix 1 , available at www.onlinejase.com ) using PubMed (National Library of Medicine), CINAHL (EBSCO), and EMBASE (Elsevier). For the searches, we selected medical subject headings (MeSH) and keywords to capture the concepts of ultrasonography, echocardiography, RV, and PE. The search results were combined and exported to EndNote bibliographic management tool, and duplicate results were removed. The search was conducted in compliance with the Preferred Reporting in Systematic Reviews and Meta Analyses (PRISMA) guidelines. We updated the search using the same criteria in November 2016. Complete details of the search strategies including search terms are available ( Supplemental Appendix 1 , available at www.onlinejase.com ).
All titles were independently reviewed for possible inclusion by two trained reviewers (J.D. and L.G.). Prior to beginning the review, both reviewers agreed to err on the side of inclusion. If either reviewer selected a reference, the full text was ordered for further review. Using this strategy, 464 articles were selected for further review. The reference sections of all included articles were checked for additional potentially relevant articles ( Figure 1 ).
Inclusion and Exclusion Criteria
Articles meeting the following criteria were eligible for review: English language, research of any design that reported quantitative data, focus on diagnosis of acute PE using echocardiography, and reporting of data that allowed for calculation of both sensitivity and specificity. Exclusion criteria included articles that were anecdotal or had no data; on >50% pregnant patients; prognostic studies; single case reports or had a sample size less than 10; studies that used healthy controls; or experimental studies using animals, animal tissue, or cadavers.
A priori, we defined the positive reference standard as either CT, pulmonary angiogram, high-probability ventilation/perfusion (V/Q) scan, intermediate-probability V/Q scan with any other confirmatory study (including leg ultrasound for venous thrombosis), surgery, or autopsy. We chose a wide range of gold standard studies to reflect the large amount of diagnostic studies that are used in practice and to ensure we did not bias our sample if different confirmatory tests were used on patients with different pretest probabilities of PE.
Data Collection and Processing
The team developed an abstraction tool designed to confirm final eligibility for full review, assess article characteristics, and extract data relevant to the study question, which was used by two reviewers (J.D. and L.G.) to independently abstract data from the articles. Data included test characteristics, reported results, location of study, gold standard test used, sonographer experience, and other ultrasonography studies done. If data important to the meta-analysis, specifically data related to the sensitivity and specificity, were needed, the corresponding author of the study was asked for the information by a single e-mail. The two abstractors and two senior team members (A.A. and J.M.F.) discussed the two independent abstractions and combined them into a final abstraction. All abstraction disagreements were minor and were quickly resolved via consensus discussion among the four team members.
In order to perform subgroup analyses, articles were categorized into those performed only in the ED, as focused cardiac ultrasound (FOCUS, previously referred to as “point of care”), and performed by a physician (instead of an cardiac sonographer, nurse, or other health care professional). These groups were not mutually exclusive, and studies could be in more than one group. For example, studies performed in ED patients could be comprehensive exams performed by cardiac sonographers or focused exams performed by physicians. Studies that did not report this information specifically were not included in these analyses.
Quality Assessment
Two reviewers (J.D. and L.G.) independently assessed the quality of the articles using the QUADAS-2 tool (University of Bristol, Bristol, UK) modified to conform to this meta-analysis of diagnostic accuracy. The resultant QUADAS-2 tool was used to assess studies for potential risk of bias involving methodology, reporting, and validity. Study heterogeneity was assessed using χ 2 analysis. All disagreements were minor and resolved via consensus of several members of the team (A.A., L.G., J.D., and J.M.F.).
We decided a priori the characteristics of each study that would meet the criteria for low risk of bias in each section of the QUADAS-2 tool. A more detailed description of the QUADAS-2 tool can be found in the Supplemental Appendix 2 (available at www.onlinejase.com ).
Statistical Analysis
All statistical analyses were performed using R version 3.1. For interrater reliability, we determined both percent agreement and Cohen’s kappa. For the diagnostic accuracy meta-analysis, a hierarchical summary receiver operator curve (HSROC) was constructed that allows both fixed and random effects (for threshold and accuracy). The HSROC models were used to obtain sensitivity, specificity, positive and negative likelihood ratios, and diagnostic odds ratios. HSROC was chosen because it has been shown to be superior to other commonly used methods, it takes into account the correlation between sensitivity and 1-specificity and models the observations for each study, and it allows the meta-analyst to investigate heterogeneity between studies taking into account both within- and between-study variability. The use of a ROC to show the relationship between sensitivity and 1-specificity is especially important when different values for cutoffs are available, as in this study.
The distribution of the estimates of the random effects was taken into account, and a 95% confidence region and 95% prediction region around these values were calculated within which the operating characteristics may lie. All pooled meta-analytic statistics are reported with their respective 95% CIs. No attempt was made to assess for publication bias, as our explicit strategy was a controlling method. Subgroup analyses were carried out each of the tests separately and on those studies that reported being conducted in the ED, utilizing point-of-care ultrasonography, and those that were performed by physician.
Results
Our systematic search strategy yielded 5,905 articles without duplicates (7,164 total). Of these, 464 were selected for full text review and 21 were deemed to meet all inclusion criteria and no exclusions for this review ( Figure 1 and Table 1 ). The percent agreement between the two reviewers for article selection was 94.3%, and Cohen’s kappa was κ = 0.908 (95% CI, 0.836–0.980; P < .001). We included one study that reported that one patient used a positron emission tomography as a reference standard, as all of the other patients met our inclusion criteria and it was impossible from the data provided to exclude this one patient. No new articles to include were found in the reference sections of the included articles. We updated our search using the same strategy in November 2016, and we identified three new articles to include out of the 1,079 references ( Figure 1 ). The interrater reliability of this new search was 98.1% and Cohen’s kappa was κ = 0.744.
| First author | Year | Country | Setting | Patient selection | Type of echocardiography | Who performed echocardiography |
|---|---|---|---|---|---|---|
| Bova | 2003 | Italy | ED and inpatient | Consecutive patients referred to or hospitalized with signs or symptoms of PE | Formal | Three “trained cardiologists” |
| Ceccarelli | 2003 | Italy | Inpatient | Patients ages 75–95 admitted to geriatric ward with a suspected PE | Unclear | Unclear |
| Cetiner | 2014 | Turkey | ED | Retrospective selection of patients with suspicion of PE, select patients with PE compared with patients proven to not have a PE | Unclear | Unclear |
| Chen | 2006 | Taiwan | ED to inpatient | Retrospectively identified patients with PE compared with consecutive patients presenting with dyspnea | Formal | Two cardiologists “offline” |
| Daley | 2016 | USA | ED | Convenience sample of patients who were undergoing a CT angiogram for evaluation of a possible PE | FOCUS | Three point-of-care ultrasound fellowship-trained attendings, four emergency medicine ultrasound fellows, one emergency medicine resident, and one medical student with no prior ultrasound experience who received 2 weeks of training |
| Dresden | 2014 | USA | ED | Convenience sample of suspected or confirmed PE | FOCUS | Trained MD/investigator, quality checked by an expert sonographer |
| Firdous | 2013 | India | Inpatient | Retrospective review of hospitalized patients with a Wells score >2 | Unclear/formal | Blinded senior cardiologists |
| Grifoni | 1998 | Italy | ED | Patients with suspected PE | Formal “standard” | Unclear |
| Jackson | 2000 | USA | ED | Convenience sample of patients who were ordered a diagnostic test for suspected PE | FOCUS | Nurse and an cardiac sonographer |
| Kalkan | 2016 | Turkey | ED | Consecutive patients admitted to emergency service for acute chest pain and/or dyspnea and underwent pulmonary CT angiography due to suspicion of PE | Formal | “Two expert cardiologists” |
| Kasper | 1997 | Germany | Unclear | Patients with suspected PE | Unclear (internal medicine/nuclear medicine) | Physician |
| Kjaergaard | 2006 | Denmark | Unclear | Patients referred for V/Q scan | Unclear/formal | Unclear |
| Kurzyna | 2002 | Poland | Unclear (tertiary referral center) | Consecutive patients with suspicion of PE referred to the center | Unclear/formal | Unclear |
| Laursen | 2013 | Denmark | ED | Convenience sample of ED patients with a respiratory signs or symptoms | FOCUS | Physician |
| Lodato | 2008 | USA | Various | Retrospective review of patients who had CT to rule out PE | Unclear | Unclear (reviewed by a blinded reader) |
| Mansencal | 2008 | France | ED | Moderate to high PE suspicion with an elevated D-Dimer | FOCUS | Physician |
| McConnell | 1996 | USA | Unclear (inpatient) | Retrospective identification of all patients with an echo seeking RV function | Unclear/formal | Unclear, “two experienced echocardiographers” |
| Miniati | 2001 | Italy | ED | Clinically suspected PE | Unclear/formal | Physician |
| Nazerian | 2014 | Italy | ED | Clinically suspected PE | FOCUS | One of 13 physicians, including principal investigator, eight emergency medicine physicians, and four trained residents (two internal medicine and two emergency medicine) |
| Nazeyrollas | 1996 | France | Cardiac intensive care unit | Suspected PE or recent dyspnea or chest pain | FOCUS | Physician |
| Rudoni | 1998 | USA | Unclear | Retrospective review of patients with dyspnea and suspected PE who underwent echocardiography | Formal | Cardiac sonographer, read by a cardiologist |
| Sakuma | 2006 | Japan | Unclear | Prospective case registry of suspected PE | Unclear | Unclear |
| Tuzovic | 2016 | USA | ED | Retrospective review of ED patients with suspected PE who had a CT, biomarkers, and an echocardiogram within 2 days | Formal | Unclear |
| Viellard-Baron | 1998 | France | Inpatient | Patients admitted with suspected PE who had an echocardiogram completed | Unclear | “Experienced echocardiographer” |
Data were collected from 24 unique studies on 12 unique signs suggestive of acute PE. The prevalence of disease in our study was 40.8%. Several studies included sensitivity and specificity data for multiple signs. As a result, 75 measures of sensitivity and specificity were collected. We treated these as 75 different studies. Three articles and two signs in two articles did not report the number of true positives, true negatives, false positives, and false negatives. Two of these authors returned raw data after an email request. The other signs without enough data were excluded, and the remaining 71 studies used in our analyses are displayed in Table 2 .
| Sign | Studies | True negative | False negative | True positive | False positive | n | Sensitivity (95% CI) | Specificity (95% CI) | Chi-square test for heterogeneity | |
|---|---|---|---|---|---|---|---|---|---|---|
| Sensitivity | Specificity | |||||||||
| 60/60 | 2 | 49 | 82 | 26 | 10 | 167 | 0.24 (0.16, 0.33) | 0.84 (0.45, 0.97) | 0.864 | 0.031 |
| Hypokinesis | 6 | 348 | 153 | 92 | 34 | 627 | 0.38 (0.31, 0.44) | 0.91 (0.88, 0.94) | <0.001 | <0.001 |
| McConnell | 6 | 372 | 147 | 42 | 10 | 571 | 0.22 (0.16, 0.29) | 0.97 (0.95, 0.99) | <0.001 | 0.027 |
| PAH | 5 | 259 | 134 | 129 | 43 | 565 | 0.44 (0.19, 0.72) | 0.84 (0.70, 0.92) | <0.001 | <0.001 |
| RH strain | 16 | 886 | 456 | 487 | 157 | 1986 | 0.53 (0.45, 0.61) | 0.83 (0.74, 0.90) | <0.001 | <0.001 |
| RHT | 4 | 252 | 157 | 5 | 0 | 414 | 0.05 (0.02, 0.09) | 0.99 (0.96, 1.00) | 0.724 | 0.480 |
| RV:LV | 8 | 501 | 133 | 162 | 83 | 879 | 0.55 (0.49, 0.60) | 0.86 (0.83, 0.89) | <0.001 | <0.001 |
| RVEDD | 4 | 228 | 40 | 156 | 49 | 473 | 0.80 (0.61, 0.92) | 0.80 (0.67, 0.89) | <0.001 | 0.006 |
| Septal | 8 | 530 | 269 | 97 | 29 | 925 | 0.26 (0.22, 0.31) | 0.95 (0.93, 0.97) | 0.862 | 0.003 |
| TR | 7 | 438 | 188 | 125 | 92 | 843 | 0.40 (0.35, 0.46) | 0.83 (0.79, 0.86) | <0.001 | <0.001 |
| TAPSE | 2 | 190 | 35 | 95 | 62 | 382 | 0.64 (0.54, 0.73) | 0.61 (0.56, 0.67) | 0.428 | <0.001 |
| RVSP | 3 | 110 | 28 | 48 | 29 | 215 | 0.47 (0.34, 0.61) | 0.73 (0.65, .80) | 0.022 | <0.001 |
| All signs | 71 | 4,163 | 1,822 | 1,464 | 598 | 8,047 | ||||
The most commonly cited use of echocardiography to detect PE was through a combination of findings suggestive of PE. These findings were termed and defined variably across 15 articles, but we combined them with the term RH strain. Terms for combined measures included RV dysfunction, RV strain, abnormal RH, and acute cor pulmonale. For the studies used in this combined term, these were the terms used in the original studies and no specific sign was mentioned in the original studies. The positive likelihood ratio for this combined term was 3.12, and the negative likelihood ratio was .57.
Other signs reported include the 60/60 sign ([1] tricuspid regurgitation (TR) jet gradient of less than 60 mmHg OR TR velocity less than 3.9 m/sec AND [2] pulmonary flow acceleration time of <60 msec), RV hypokinesis, McConnell’s sign (distinctive akinesia of the mid free wall with normal motion at the apex), pulmonary arterial hypertension (PAH), visualization of an RH thrombus (RHT), RV to left ventricle ratio (RV:LV), RV end-diastolic diameter (RVEDD), abnormal or paradoxical septal motion (septal), TR, tricuspid annular plane systolic excursion (TAPSE), and RV systolic pressure (RVSP). Other signs that only occurred in one study are not reported: inferior vena caval distension, RV isovolumetric acceleration, and LV end-diastolic diameter.
Pooled estimates and 95% CIs for the sensitivity and specificity for each of the 10 signs considered are displayed in Table 2 . Forest plots and summary ROC were created for the RH strain sign ( Figures 2 and 3 ), and ROCs are available for the other three signs with the greatest number of studies on them: RV:LV ratio, abnormal septal motion, and TR ( Figure 3 ). Forest plots for these three signs are available upon requests from the authors. Only four signs showed a low chance of heterogeneity: RHT (sensitivity and specificity), 60/60 sign (sensitivity), TAPSE (sensitivity), and abnormal septal motion (sensitivity).
We also conducted subgroup analyses on the studies that reported that they were performed by a physician, ( Table 3 ), performed as FOCUS ( Table 4 ), and performed only in the ED ( Table 5 ). For physician-performed studies, 60/60 sign and visualization of a RHT had only one study on it. For FOCUS studies, only RH strain, RV:LV ratio, and abnormal/paradoxical septal motion had more than one study on them. For ED studies, only 60/60 sign and McConnell’s sign had 0 or 1 studies on them.
| Sign | Studies | True negative | False negative | True positive | False positive | Subjects | Sensitivity (95% CI) | Specificity (95% CI) | Chi-square test for heterogeneity | |
|---|---|---|---|---|---|---|---|---|---|---|
| Sensitivity | Specificity | |||||||||
| 60/60 | 0 | — | — | — | — | — | — | — | — | — |
| Hypokinesis | 3 | 272 | 67 | 62 | 3 | 404 | 0.47 (0.33, 0.61) | 0.98 (0.95, 0.99) | 0.068 | 0.276 |
| McConnell | 2 | 157 | 62 | 15 | 4 | 238 | 0.19 (0.11, 0.30) | 0.98 (0.94, 0.99) | 1 | <0.123 |
| PAH | 2 | 90 | 39 | 75 | 24 | 228 | 0.65 (0.29, 0.91) | 0.74 (0.50, 0.89) | <0.001 | 0.095 |
| RH strain | 5 | 491 | 221 | 232 | 56 | 1000 | 0.61 (0.32, 0.84) | 0.88 (0.83, 0.92) | <0.001 | 0.002 |
| RHT | 1 | 92 | 54 | 2 | 0 | 148 | 0.04 (0.01, 0.14) | 0.99 (0.92, 1.00) | <0.001 | <0.001 |
| RV:LV | 4 | 351 | 74 | 56 | 32 | 513 | 0.46 (0.32, 0.62) | 0.91 (0.76, 0.97) | 0.013 | <0.001 |
| RVEDD | 3 | 209 | 22 | 133 | 42 | 406 | 0.86 (0.72, 0.93) | 0.83 (0.69, 0.92) | 0.041 | 0.003 |
| Septal | 2 | 204 | 62 | 24 | 4 | 294 | 0.28 (0.20, 0.39) | 0.96 (0.91, 0.98) | 1.000 | 0.104 |
| TR | 3 | 207 | 84 | 70 | 32 | 393 | 0.46 (0.28, 0.65) | 0.87 (0.82, 0.90) | 0.004 | 0.952 |
| All signs | 24 | 2,073 | 685 | 669 | 197 | 3,624 | ||||
| Sign | Studies | True negative | False negative | True positive | False positive | Subjects | Sensitivity (95% CI) | Specificity (95% CI) | Chi-square test for heterogeneity | |
|---|---|---|---|---|---|---|---|---|---|---|
| Sensitivity | Specificity | |||||||||
| 60/60 | 0 | — | — | — | — | — | — | — | — | — |
| Hypokinesis | 3 | 272 | 67 | 62 | 3 | 404 | 0.47 (0.33, 0.61) | 0.98 (0.95, 0.99) | 0.068 | 0.276 |
| McConnell | 1 | 116 | 24 | 6 | 0 | 146 | 0.21 (0.10, 0.39) | 1.00 (0.94, 1.00) | 0.002 | <0.001 |
| PAH | 3 | 217 | 91 | 55 | 26 | 389 | 0.30 (0.10, 0.62) | 0.90 (0.79, 0.95) | 0.001 | 0.023 |
| RH strain | 6 | 554 | 174 | 172 | 82 | 982 | 0.55 (0.32, 0.76) | 0.87 (0.82, 0.90) | <0.001 | 0.031 |
| RHT | 3 | 243 | 142 | 4 | 0 | 389 | 0.04 (0.02, 0.09) | 0.99 (0.97, 1.00) | 0.830 | 0.949 |
| RV:LV | 3 | 269 | 29 | 36 | 27 | 361 | 0.54 (0.41, 0.66) | 0.91 (0.64, 0.98) | 0.248 | <0.001 |
| RVEDD | 3 | 209 | 22 | 133 | 42 | 406 | 0.86 (0.72, 0.93) | 0.83 (0.69, 0.92) | 0.041 | 0.003 |
| Septal | 4 | 351 | 132 | 44 | 8 | 535 | 0.26 (0.20, 0.33) | 0.96 (0.93, 0.98) | 0.807 | 0.208 |
| TR | 3 | 227 | 78 | 48 | 29 | 382 | 0.36 (0.17, 0.61) | 0.88 (0.83, 0.92) | <0.001 | 0.266 |
| All signs | 29 | 2,458 | 759 | 560 | 217 | 3,994 | ||||
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
