Does This Patient with a Pericardial Effusion Have Cardiac Tamponade?

Home , Cardiac tamponade, Jugular venous pressure, Korotkoff sounds, Pericardial effusion, Pulsus paradoxus

THE RATIONAL CLINICIAN’S CORNER CLINICAL EXAMINATION

Does This Patient With a Pericardial Effusion Have Cardiac Tamponade?

Christopher L. Roy, MD Context Cardiac tamponade is a state of hemodynamic compromise resulting from Melissa A. Minor, MD cardiac compression by fluid trapped in the pericardial space. The clinical examination M. Alan Brookhart, PhD may assist in the decision to perform pericardiocentesis in patients with cardiac tamponade diagnosed by echocardiography. Niteesh K. Choudhry, MD, PhD Objective To systematically review the accuracy of the history, physical examination, and basic diagnostic tests for the diagnosis of cardiac tamponade. CLINICAL SCENARIOS Data Sources MEDLINE search of English-language articles published between 1966 Case 1 and 2006, reference lists of these articles, and reference lists of relevant textbooks. A 55-year-old woman with chronic ob- Study Selection We included articles that compared aspects of the clinical exami- structive pulmonary disease presents to nation to a reference standard for the diagnosis of cardiac tamponade. We excluded your office with 2 weeks of progressive studies with fewer than 15 patients. Of 787 studies identified by our search strategy, dyspnea. She had breast cancer diagnosed 8 were included in our final analysis. and treated more than 10 years ago and Data Extraction Two authors independently reviewed articles for study results and has had no evidence of recurrence. She quality. A third reviewer resolved disagreements. was seen in the emergency department 1 Data Synthesis All studies evaluated patients with known tamponade or those re- week ago, where her evaluation included ferred for pericardiocentesis with known effusion. Five features occur in the majority an echocardiogram that demonstrated a of patients with tamponade: dyspnea (sensitivity range, 87%-89%), tachycardia (pooled moderate-sized (15 mm in maximum sensitivity, 77%; 95% confidence interval [CI], 69%-85%), pulsus paradoxus (pooled width), circumferential pericardial effu- sensitivity, 82%; 95% CI, 72%-92%), elevated jugular venous pressure (pooled sen- sion without echocardiographic evidence sitivity, 76%; 95% CI, 62%-90%), and cardiomegaly on chest radiograph (pooled ofcardiactamponade.Duringyourexami- sensitivity, 89%; 95% CI, 73%-100%). Based on 1 study, the presence of pulsus para- nation, you note tachycardia and faint doxus greater than 10 mm Hg in a patient with a pericardial effusion increases the heart sounds. A pulsus paradoxus is 6 likelihood of tamponade (likelihood ratio, 3.3; 95% CI, 1.8-6.3), while a pulsus para- mm Hg with a blood pressure of 100/60 doxus of 10 mm Hg or less greatly lowers the likelihood (likelihood ratio, 0.03; 95% CI, 0.01-0.24). mmHg,thejugularvenouspressureisnot elevated, and auscultation of her lungs re- Conclusions Among patients with cardiac tamponade, a minority will not have dys- veals diminished breath sounds. You ob- pnea, tachycardia, elevated jugular venous pressure, or cardiomegaly on chest radiograph. A pulsus paradoxus greater than 10 mm Hg among patients with a pericardial tain a chest radiograph that is unchanged effusion helps distinguish those with cardiac tamponade from those without. Diag- from previous films and that shows hy- nostic certainty of the presence of tamponade requires additional testing. perinflated, clear lung fields and no en- JAMA. 2007;297:1810-1818 www.jama.com largement of the cardiac silhouette. You suspect that the patient’s symptoms are attributabletoherlungdisease,butshould He is found to have a pulsus para- Author Affiliations: Divisions of General Medicine and Primary Care (Dr Roy) and Pharmacoepidemiology and you evaluate further for tamponade? doxus of 18 mm Hg with a blood pres- Pharmacoeconomics (Drs Brookhart and Choudhry), sure of 130/70 mm Hg and a heart rate Brigham and Women’s Hospital, Harvard Medical Case 2 School, Boston, Mass; Hospitalist Service, Brigham and of 90/min. The remainder of the physi- Women’s Hospital, Boston (Drs Roy and Choudhry); A previously healthy 34-year-old man cal examination is unremarkable. The and Harvard Vanguard Medical Associates, Boston (Dr chest radiograph shows an enlarged car- Minor). presents to the emergency depart- Corresponding Author: Christopher L. Roy, MD, Di- ment with dyspnea for the past 3 days. diac silhouette. You order an echocar- vision of General Medicine and Primary Care, Hospi- diogram that shows a moderate-sized, talist Service, Brigham and Women’s Hospital, 75 Fran- cis St, Boston, MA 02115 ([email protected]). See also Patient Page. circumferential pericardial effusion with The Rational Clinical Examination Section Editors: a left ventricular ejection fraction of David L. Simel, MD, MHS, Durham Veterans Affairs CME available online at 65%. The study is limited by body habi- Medical Center and Duke University Medical Center, www.jama.com Durham, NC; Drummond Rennie, MD, Deputy Edi- tus and rapid respirations, and the right tor, JAMA.

Downloaded from www.jama.com on April 24, 2007 DOES THIS PATIENT WITH A PERICARDIAL EFFUSION HAVE CARDIAC TAMPONADE? atrium and right ventricle are poorly vi- too sensitive and overdiagnose cardiac The Pathophysiology sualized. Do the physical examination tamponade in patients with only subtle of Cardiac Tamponade findings alter your assessment of this evidence of hemodynamic compro- The pericardial sac consists of 2 layers: patient’s pericardial effusion? mise.4,6,7 Nonetheless, because echocar- an outermost parietal pericardium and diography is a noninvasive modality that the inner visceral pericardium (epicar- WHY IS THE CLINICAL is highly accurate, it is an essential com- dium) that reflects directly over the sur- EXAMINATION FOR CARDIAC ponent of the evaluation.8-11 face of the heart. Normally, no more than TAMPONADE IMPORTANT? There are several scenarios in which 15 to 30 mL of fluid exists between the Cardiac tamponade occurs when fluid the clinical examination for cardiac tam- 2 layers, held at pressures approximat- trapped in the pericardial space com- ponade could be useful. If the clinical ex- ing the pleural pressure or approxi- presses the heart and compromises car- amination could rule out tamponade mately 5 mm Hg lower than central ve- diac output. The consequences of car- with a high degree of certainty, it could nous pressure.13,14 diac tamponade range from barely preclude the need for echocardiogra- When pericardial fluid accumulates detectable effects to overt hemody- phy in some patients and allow the cli- slowly, parietal pericardium compli- namic collapse.1-3 When a pericardial ef- nician to pursue other potential diag- ance increases. Increasing amounts of fusion becomes large enough or accu- noses. On the other hand, the clinician fluid causes intrapericardial pressure to mulates rapidly enough to cause is sometimes faced with echocardio- increase, while the central venous pres- hemodynamic consequences, readily ob- graphic findings suggestive of cardiac sure responds by increasing to main- served symptoms and signs herald im- tamponade, and clinical correlation is tain a gradient that allows cardiac fill- pending circulatory collapse that neces- necessary to guide the decision for ur- ing. When pericardial compliance can sitates urgent therapeutic intervention. gent pericardiocentesis.6,12 increase no more, the intrapericardial The invasive reference standard for the diagnosis of cardiac tamponade re- Figure 1. Relationship Between Intracardiac Filling Pressures and Intrapericardial Pressure and quires simultaneous measurement of in- Cardiac Output in Cardiac Tamponade trapericardial and intracardiac pressures. The intrapericardial, right atrial, Normal Cardiac Tamponade pulmonary artery diastolic, and pulmonary capillary wedge pressures are all elevated and equalized in cardiac tamponade (FIGURE 1). In the absence of other pathology (eg, effusive- constrictive disease), pericardiocentesis will normalize pressures and improve the cardiac output, confirm- Pressure ing the diagnosis.4 Echocardiographyisthediagnostictest Left Heart Filling Pressure used most frequently to evaluate cardiac tamponade when it is clinically suspected Intrapericardial Pressure (FIGURE 2). While echocardiography is the noninvasive reference standard for Increasing Pericardial Effusion Volume the diagnosis of pericardial effusion,5 the mere presence of an effusion does not define cardiac tamponade. Rather, sev- Cardiac Output eral echocardiographic findings, includ- Normal ingrightatrialsystoliccollapse,rightven- tricular diastolic collapse, inferior vena caval plethora, and exaggeration of respirophasic changes in flow velocities across the tricuspid and mitral valves, are indicative of tamponade physiology and make the diagnosis with near cer- Increasing Pericardial Effusion Volume tainty when the pretest probability is As pericardial fluid volume increases to the limit of pericardial compliance, intrapericardial pressure rises. Car- high. Because cardiac tamponade occurs diac output begins to fall as intrapericardial pressure equalizes with central venous, right atrial, and right ven- on a continuum of hemodynamic effects, tricular end-diastolic pressures (right heart filling pressures), then rapidly falls as it equalizes with left atrial and left ventricular end-diastolic pressures (left heart filling pressures). the findings on echocardiogram may be

Downloaded from www.jama.com on April 24, 2007 DOES THIS PATIENT WITH A PERICARDIAL EFFUSION HAVE CARDIAC TAMPONADE? pressure first equalizes with the right ven- infrequently that the diagnosis should When and How to Perform tricular diastolic pressure and then with not be considered unless the setting is ap- the Clinical Examination the left (Figure 1). At this threshold, the propriate. As a result of different pa- for Cardiac Tamponade cardiac output drops and circulation is tient populations, practice settings, and In the nonurgent setting, cardiac tam- maintained only through an increase in the intensity of evaluation undertaken, ponade is not typically a consider- heart rate, contractility, and peripheral the reported incidences of various eti- ation unless the patient has an under- arteriolar vasoconstriction.14-16 Because ologies of pericardial effusions vary.17-19 lying cause with relevant symptoms or of pericardial compliance, large amounts In the largest reported series, Sagristà- suggestive signs. Thus, the astute clini- of fluid may accumulate before overt he- Sauleda et al18 reviewed the medical rec- cal epidemiologist will infer correctly modynamic effects occur. Rapid (ie, over ords of 322 Spanish patients with mod- that cardiac tamponade has been stud- minutes to hours) fluid accumulation erate or large effusions. The most ied mostly in selected patients at risk may exceed the ability of the pericar- common diagnoses were acute idio- of having the condition on the basis of dium to stretch; if so, intrapericardial pathic pericarditis (20%), iatrogenic ef- symptoms and signs. In fact, the re- pressure increases rapidly and cardiac fill- fusion (16%), malignancy (13%), chronic cent literature has focused almost en- ing is impaired, with an often dramatic idiopathic effusion (9%), acute myocar- tirely on patients with known pericar- drop in cardiac output. dial infarction (8%), end-stage renal dis- dial effusion on echocardiography. ease (6%), congestive heart failure (5%), These studies will have a tendency to Causes of Cardiac Tamponade collagen vascular disease (5%), and tu- overestimate the sensitivity of symp- The etiology of cardiac tamponade in pa- berculosis or bacterial infection (4%). toms and signs in unselected patients. tients admitted to medical services mir- Pericardial effusions often result from pri- rors the various conditions that cause mary cardiac disease, occurring in 14% Symptoms pericardial effusions. The absence of of patients with congestive heart fail- The important symptoms include dys- population-based incidence estimates of ure, 21% of patients with valvular heart pnea, chest pain, or fullness.15,23 Pa- tamponade and the paucity of data about disease, and 15% of patients with myo- tients may report nausea or abdomi- how often pericardial effusion leads to cardial infarction.20 Effusions are ex- nal pain from hepatic and visceral tamponade prevents establishment of tremely common after cardiac sur- congestion or dysphagia from esopha- pretest probability estimates. We do gery,21 although large effusions causing geal compression.15,23 Nonspecific know that cardiac tamponade occurs so tamponade are rare in this setting.22 symptoms such as lethargy, fever,

Figure 2. Echocardiogram of a 62-Year-Old Woman With Advanced Lung Cancer and Malignant Pericardial Effusion Causing Cardiac Tamponade

Transducer Transducer

Collapse Right Ventricle Right Atrium APEX

Effusion

Left Atrium Left Ventricle

Right Atrium Effusion

Right

PLANEF Ventricle O VIEW

APEX Transducer

In this subcostal view captured in early ventricular diastole, a large, circumferential pericardial effusion measuring 3.3 cm in maximal diameter is compressing the heart, and the right ventricle is completely collapsed (see video at http://jama.com/cgi/content/full/297/16/1810/DC1).

Downloaded from www.jama.com on April 24, 2007 DOES THIS PATIENT WITH A PERICARDIAL EFFUSION HAVE CARDIAC TAMPONADE? cough, weakness, fatigue, anorexia, and lar interdependence, is a mildly de- and right ventricular hypertrophy.23 palpitations also occur.23 Because pa- creased stroke volume and blood pres- Likewise, other conditions can create tients with acute cardiac tamponade are sure during inspiration. Opposite a pulsus paradoxus, including severe critically ill and often in shock, the clini- changes occur during expiration. In car- obstructive pulmonary disease, con- cal history may not be obtainable. diac tamponade, ventricular interdepen- gestive heart failure, mitral stenosis, dence is exaggerated because high peri- massive pulmonary embolism, severe Signs cardial pressure compresses the entire hypovolemic shock, obesity, and tense The physical examination begins with heart; during inspiration, left heart fill- ascites.13,23 an assessment of vital signs and, de- ing is more dramatically restricted by pending on the degree of hemody- right heart filling.29 Other Signs namic compromise, may reveal tachy- Pulsus paradoxus can be detected by Patients with cardiac tamponade of- cardia, hypotension, and tachypnea. The palpating the radial pulse and noting an ten have an elevated jugular venous classic findings of tamponade were de- inspiratory diminution of the pulse dur- pressure at bedside examination, but the scribed in 1935 by thoracic surgeon ing normal respirations. It is most com- sensitivity of this finding may be re- Claude Schaeffer Beck.24 The Beck triad monly measured using a manual sphyg- duced by the patient’s body habitus and, is characterized by decreasing arterial momanometer (FIGURE 3). During theoretically, in the setting of hypovo- blood pressure, increasing jugular ve- normal respirations, the examiner lemia. The Kussmaul sign, an inspira- nous pressure, and a small, quiet heart. slowly deflates the sphygmomanom- tory elevation of the jugular venous This triad was seen in surgical patients eter cuff while listening for the first pressure, should not be seen unless un- with acute tamponade from intraperi- Korotkoff sounds. These sounds are ini- derlying constrictive disease exists.13,15 cardial hemorrhage due to trauma or to tially intermittent and respirophasic, be- On cardiac auscultation, the examiner myocardial or aortic rupture. This “sur- coming audible with expiration and in- may note diminished heart sounds, al- gical tamponade” is distinct from that audible with inspiration. With further though this finding may also be observed seen in medical patients, who gener- cuff deflation, the Korotkoff sounds be- inpatientswithlargechestsorchronicob- ally develop pericardial effusions slowly, come audible throughout the respira- structive pulmonary disease. Although and in whom the findings of the Beck tory cycle. The clinician records the sys- pericardial friction rubs are often believed triad may not be present at all.25 Some tolic pressure at which Korotkoff to occur only in patients with pericardi- medical patients, especially those with sounds are first audible and the sys- tis and relatively small effusions, some pa- preexisting hypertension, may actually tolic pressure at which they are au- tients with tamponade have a pericardial be hypertensive with tamponade physi- dible constantly through a respiratory rub.8,25,35 One study demonstrated that a ology.26 A fever associated with a peri- cycle. The presence of a pulsus para- rub may be present with very large effu- cardial effusion suggests an infectious doxus may also be detected by observ- sions, so the presence of a rub is not a use- etiology but also could be associated with ing the inspiratory diminution of the pe- ful clue to the size of the effusion.36 immune-mediated connective tissue dis- ripheral pulse on an arterial catheter ease such as systemic lupus erythema- tracing or pulse oximeter (Figure 3).30,31 Electrocardiogram tosus. Most textbooks define a greater than and Chest Radiograph 10-mm Hg difference between the ini- The electrocardiogram and chest radio- Pulsus Paradoxus tial detection of sounds on expiration and graph support the diagnosis of a large The examiner should evaluate the pa- the constant presence of sounds with pericardial effusion causing tampon- tient for the presence of pulsus para- each heartbeat through the respiratory ade. Findings on an electrocardiogram doxus, a phenomenon originally de- cycle as a “pulsus paradoxus.”32 Some ex- indicating a large pericardial effusion in- scribed by Adolf Kussmaul in 1873 as a perts suggest that the absolute value of clude low QRS voltage, electrical alter- palpable diminution of the radial pulse pulsus paradoxus should be inter- nans (changes in the amplitude or mor- on inspiration in patients with cardiac preted as a percentage of the pulse pres- phology of the P, QRS, and ST-T waves tamponade.27 Pulsus paradoxus is not a sure or as a percentage of the expira- from one beat to the next, resulting from paradox at all, but an exaggeration of the tory systolic pressure.15,33,34 “Total cardiac oscillation within the pericar- normal inspiratory decrease in blood paradox,” also known as “pulse oblit- dial fluid37), atrial arrhythmias, and, if pressure.28 This decrease occurs be- eration,”is defined as inspiratory disap- pericardial inflammation is present, ST- cause negative intrapleural pressures pearance of the brachial and radial pulses, segment elevation and PR-segment de- during inspiration cause increased ve- with total disappearance of the Korot- pression. Depending on the size of the nous return and filling of the right heart, koff sounds.25 effusion, the chest radiograph may be which result in bowing of the septum to Many conditions may mask the pres- normal or may demonstrate an en- the left, decreasing filling of the left heart. ence of pulsus paradoxus, including hy- larged, globular cardiac silhouette and/or In the normal heart, the consequence of potension, pericardial adhesions, aor- an epicardial fat stripe or “double lu- this phenomenon, known as ventricu- tic regurgitation, atrial septal defects, cency” sign on lateral views.38

Downloaded from www.jama.com on April 24, 2007 DOES THIS PATIENT WITH A PERICARDIAL EFFUSION HAVE CARDIAC TAMPONADE?

METHODS Bayes theorem, cardiac auscultation, pro- additional articles, we reviewed the ref- Literature Search fessional competence, sensitivity and speci- erence lists of articles that were retrieved and Data Collection ficity, reproducibility of results, observer and references cited in cardiology and We performed a structured MEDLINE variation, diagnostic tests, routine, electro- physical diagnosis textbooks.32,39-41 search (January 1, 1966, to December 31, cardiography, pericardium, pericardial dis- We included articles reporting the 2006) to identify English-language ease, pericarditis, pericardial effusion, peri- results of original studies that evaluated articles pertinent to the clinical exami- cardial tamponade, tamponade, and cardiac the history, physical examination, and nation of patients with cardiac tampon- tamponade. Two authors (C.L.R., routine diagnostic tests such as electro- ade. Keywords used included physical M.A.M.) independently reviewed the cardiography or plain chest radiogra- examination, medical history taking, pre- abstracts of the search and retrieved phy compared with a reference stan- dictive value, decision support techniques, potentially relevant articles. To identify dard for the diagnosis of cardiac

Figure 3. Measurement and Mechanism of Pulsus Paradoxus

A Measuring Pulsus Paradoxus Korotkoff Sounds Sphygmomanometer Cuff Deflated First IntermittentContinuous None 120

110 Systolic Blood Pressure 100 Pulsus Paradoxus Sphygmomanometer 90 Pressure, mm Hg 80

70 Diastolic Blood Pressure 60

Presence of Korotkoff Sounds

Arterial Blood Pressure Pulsus Paradoxus Tracing

Expiration Intrathoracic Pressure Inspiration

B Ventricular CARDIAC TAMPONADE NORMAL Interdependence Expiration Expiration

Inspiration Inspiration Inspiration Inspiration

Effusion Right Ventricle Left Ventricle Right Ventricle Left Ventricle

A, The examiner inflates the sphygmomanometer cuff fully, listens for Korotkoff sounds as the cuff is slowly deflated, and then notes the pressure at which Korotkoff sounds are initially audible only during expiration. As the cuff is further deflated, the examiner notes the pressure at which Korotkoff sounds become audible during expiration and inspiration. The difference between these 2 pressures is the pulsus paradoxus. In cardiac tamponade, the pulsus paradoxus measures greater than 10 mm Hg. Inspiratory dimunition in the pulse wave amplitude seen on this arterial tracing demonstrates pulsus paradoxus. A similar phenomenon may be observed on a pulse oximeter waveform. B, During inspiration in the normal heart, negative intrapleural pressures increase venous return to the right ventricle and decrease pulmonary venous return to the left ventricle by increasing pulmonary reservoir for blood. As a result of increased right ventricular distention, the interventricular septum bows slightly to the left, and the distensibility, filling, and stroke volume of the left ventricle are mildly reduced. In expiration, these changes are reciprocal, resulting in the septum bowing to the right and a mild reduction in right ventricular filling. In the presence of cardiac tamponade, the reciprocal changes seen in the normal heart are exaggerated when the pericardial sac is filled with fluid, thus limiting distensibility of the entire heart. This results in a more dramatic reduction in filling of the left ventricle during inspiration, exacerbating the normal inspiratory decrease in stroke volume and blood pressure.

Downloaded from www.jama.com on April 24, 2007 DOES THIS PATIENT WITH A PERICARDIAL EFFUSION HAVE CARDIAC TAMPONADE? tamponade. Acceptable reference stan- the reference standard for the diagnosis However,2studiesexamining80patients dards included pericardiocentesis with of cardiac tamponade. Guberman et al25 report the sensitivities of the clinical his- right heart catheterization and echocar- included patients with clinical findings tory among patients with pericardial ef- diography. We excluded studies deal- consistent with tamponade (elevated fusion referred for pericardiocentesis ing with pericardial diseases other than jugular venous pressure and pulsus para- (TABLE 2).6,47 Most patients with cardiac cardiac tamponade, as well as studies of doxus) and the documentation of peri- tamponade have dyspnea (sensitivity, patients with cardiac tamponade after car- cardial effusion by echocardiography or 87%-88%).Chestpain,cough,fever,leth- diac surgery. We also excluded those with other means. Gibbs et al47 included pa- argy,andpalpitationsoccurin25%orless fewer than 15 patients; similar exclu- tients diagnosed with cardiac tampon- of patients. sions of small studies have been used in ade using echocardiographic criteria. One prior reviews in The Rational Clinical study included control patients with- Accuracy of the Examination series.42,43 out pericardial effusion to evaluate the Physical Examination Articles were graded for methodologi- sensitivity and specificity of pulsus para- for Cardiac Tamponade cal quality.44 Level 1 studies indepen- doxus in diagnosing tamponade.9,33 One study adequately evaluated pul- dently compare a symptom or sign with sus paradoxus among patients with and Accuracy of Symptoms an acceptable reference standard of di- without cardiac tamponade (TABLE 3), agnosis among a large number of con- for Cardiac Tamponade allowing calculation of LRs. Curtiss et secutive patients. Level 2 studies meet Nostudiesprovidedataforcalculatingthe al33 studied 65 patients with known peri- the criteria for level 1 studies but have likelihood ratios (LRs) of symptoms that cardial effusion referred for pericardio- fewer patients. Level 3 studies examine suggest cardiac tamponade. centesis for suspected tamponade and nonconsecutive patients suspected of having the target disorder, but, as in level Table 1. Studies Assessing the Accuracy of the Clinical Examination for Cardiac Tamponade* 1 and 2 studies, the patients are identi- Study Design/ fied independently of the reference stan- Source No. of Patients Patient Population Reference Standard dard and the results of the reference stan- Reddy et al,34 1978 19 Retrospective chart Hemodynamic findings review of patients before and after dard are interpreted blinded to the signs with pericardial pericardiocentesis and symptoms. Level 4 studies are non- effusion referred for independent comparisons of signs and pericardiocentesis symptoms vs a gold standard among pa- Guberman et al,25 1981 56 Retrospective chart Clinical criteria including review of patients elevated jugular tients with, and perhaps without, the tar- with cardiac venous pressure, get condition. Level 4 studies may there- tamponade pulsus paradoxus, and documentation fore include implicit retrospective of pericardial effusion reviews of small groups and case series Singh et al,8 1984 16 Prospective cohort of Hemodynamic findings literature. patients with before and after pericardial effusion pericardiocentesis referred for Data Analysis pericardiocentesis When 3 or more studies presented data Curtiss et al,33 1988 65 Prospective cohort of Hemodynamic findings patients with before and after on a particular finding, we calculated pericardial effusion pericardiocentesis pooled sensitivities and confidence in- referred for tervals (CIs) using a randomeffects pericardiocentesis Levine et al,6 1991 50 Prospective cohort of Hemodynamic findings model; otherwise, we report the point patients with before and after estimate or range. The data were ana- suspected cardiac pericardiocentesis lyzed using R version 2.2.0.45 tamponade referred for pericardiocentesis RESULTS Brown et al,26 1992 18 Prospective cohort of Hemodynamic findings patients with cardiac before and after Study Characteristics tamponade referred pericardiocentesis Of the 787 studies identified using our for pericardiocentesis Cooper et al,46 1995 30 Retrospective chart Relief of symptoms search strategy, 8 that included a total of review of patients of dyspnea by 300 patients met our inclusion criteria with pericardial pericardiocentesis effusion referred for (TABLE 1). All studies were of level 4 pericardiocentesis quality. No studies assessed the preci- Gibbs et al,47 2000 46 Retrospective chart Echocardiographic sion of the clinical examination in car- review of patients evidence of cardiac with pericardial tamponade diac tamponade. effusion referred for Six of the 8 studies used right heart pericardiocentesis catheterization and pericardiocentesis as *All studies were quality level 4.

Downloaded from www.jama.com on April 24, 2007 DOES THIS PATIENT WITH A PERICARDIAL EFFUSION HAVE CARDIAC TAMPONADE? used a 20% or greater increase in car- simultaneous recordings of brachial or Eight studies of 295 patients re- diac output after pericardiocentesis as femoral artery pressure, electrocardio- ported the sensitivity of various compo- the reference standard for tamponade. gram, and a thermistor output for res- nents of the physical examination in car- Pulsus paradoxus was measured using piratory phase identification. The LR for diac tamponade (TABLE 4). All but 1 of a pulsus paradoxus greater than 12 these studies25 were in patients with peri- mm Hg was 5.9 (95% CI, 2.4-14) and cardial effusion referred for pericardio- Table 2. Sensitivity of the Clinical History in the Diagnosis of Cardiac Tamponade for a cutpoint of greater than 10 mm Hg centesis. The most common findings 33 Levine Cooper was 3.3 (95% CI, 1.8-6.3). When the among patients with cardiac tampon- et al,6 et al,46 pulsus paradoxus was 10 mm Hg or less, ade were tachycardia, elevated jugular ve- Sign/ 1991 1995 cardiac tamponade was unlikely (LR, nous pressure, and pulsus paradoxus, Symptom (N = 50) (N = 30) Dyspnea 88 87 0.03; 95% CI, 0.01-0.24) and was simi- with pooled sensitivities ranging from Fever 25 larly unlikely when a higher pulsus 76% to 82%. Hypotension and dimin- Chest pain 20 paradoxus threshold of 12 mm Hg ished heart sounds were insensitive (26% Cough 10 7 Lethargy 3 or less was used (LR, 0.03; 95% CI, and 28%, respectively). Hypertension can Palpitations 3 0-0.21).33 occur in patients with tamponade but primarily occurs in patients with preexisting hypertension (sensitivity, 33%; Table 3. Sensitivity, Specificity, and Positive and Negative Likelihood Ratios (LRs) of Pulsus 95% CI, 11%-55%).26 Guberman et al25 Paradoxus in the Diagnosis of Cardiac Tamponade* noted that tachypnea was a common Pulsus Paradoxus, mm Hg† finding in their study of 56 patients (sen- Ͼ12 Ͼ10 sitivity, 80%; 95% CI, 70%-90%). Sensitivity, % 98 98 Accuracy of Electrocardiography Specificity, % 83 70 and Chest Radiography LR (95% CI) Positive 5.9 (2.4-14) 3.3 (1.8-6.3) The electrocardiographic findings of Negative 0.03 (0-0.21) 0.03 (0.01-0.24) cardiac tamponade lacked sensitivity Abbreviation: CI, confidence interval. (TABLE 5). The pooled sensitivity of low *All data from Curtiss et al (N = 65).33 †Measured using an intra-arterial transducer. QRS voltage was only 42% (95% CI,

Table 4. Sensitivity of the Physical Examination in the Diagnosis of Cardiac Tamponade %

Reddy Guberman Singh Curtiss Levine Brown Cooper Gibbs et al,34 et al,25 et al,8 et al,33 et al,6 et al,26 et al,46 et al,47 Pooled 1978 1981 1984 1988 1991 1992 1995 2000 Sensitivity Sign (N = 19) (N = 56) (N = 16) (N = 65) (N = 50) (N = 18) (N = 25)* (N = 46) (95% CI) Pulsus paradoxus Ͼ10 mm Hg 71†‡ 77§ 75§ 98‡ 86࿣ 56࿣ 80࿣ 82 (72-92) Tachycardia 77 74 65 87 77 (69-85) Hypotension 35 14 30 24 26 (16-36) Hypertension ¶ 33 Tachypnea 80 Diminished heart sounds 34 24 24 28 (21-35) Elevated JVP 88 74 53 87 76 (62-90) Peripheral edema 21 28 Pericardial rub 29 19 Hepatomegaly 55 28 Kussmaul sign 26 Pulse pressure, mm Hg Ͼ054 Ͼ100 12 Total paradox 23 Abbreviations: CI, confidence interval; JVP, jugular venous pressure. *Not all patients had documentation of clinical findings. †Defined pulsus paradoxus as expiratory systolic pressure-inspiratory systolic pressure/expiratory systolic pressure Ͼ10%. ‡Pulsus paradoxus measured with intra-arterial transducer. §Pulsus paradoxus measured with sphygmomanometer or intra-arterial transducer. ࿣Pulsus paradoxus measured with sphygmomanometer. ¶Systolic blood pressure Ͼ140 mm Hg.

Downloaded from www.jama.com on April 24, 2007 DOES THIS PATIENT WITH A PERICARDIAL EFFUSION HAVE CARDIAC TAMPONADE?

Table 5. Sensitivity of the Electrocardiogram in the Diagnosis of Cardiac Tamponade %

Reddy Guberman Singh Levine Cooper Gibbs Pooled et al,34 1978 et al,25 1981 et al,8 1984 et al,6 1991 et al,46 1995 et al,47 2000 Sensitivity (N = 19) (N = 53)* (N = 16) (N = 50) (N = 23) (N = 46)* (95% CI) Low voltage 40 50 56 22 39 42 (32-53) Atrial arrhythmia 0 9 4 6 (1-11) Electrical alternans 21 16 ST-segment elevation 30 18 PR-segment depression 18 Abbreviation: CI, confidence interval. *Not all patients had documentation of clinical findings.

32%-53%), and 2 studies reported that- electrical alternans had a sensitivity of Table 6. Sensitivity of the Chest Radiograph in the Diagnosis of Cardiac Tamponade only 16% to 21%. Atrial arrhythmias are Source Patients, No. Cardiomegaly, % Guberman et al,25 1981 53 95 infrequent in cardiac tamponade, as are 8 findings associated with acute pericar- Singh et al, 1984 16 94 Levine et al,6 1991 50 68 ditis (PR-segment depression and ST- Gibbs et al,47 2000 46 100 segment elevation). Pooled sensitivity (95% CI) 89 (73-100) Cardiomegaly on chest radiography Abbreviation: CI, confidence interval. is fairly useful in the diagnosis of cardiac tamponade, with a pooled sensitivity of 89% (95% CI, 73%-100%) in CONCLUSIONS cardiac tamponade. As a result, we are the 4 studies of 165 patients that evalu- Based on our review of the literature, unable to comment on whether the ated this finding (TABLE 6). dyspnea, tachycardia, elevated jugular clinical examination is useful in other venous pressure, pulsus paradoxus, or scenarios (eg, to exclude tamponade SCENARIO RESOLUTIONS cardiomegaly on chest radiograph is in patients who have not yet had an Case 1 seen in 70% or more of patients with echocardiogram). In addition, the The patient has several important nega- a known pericardial effusion and car- studies reviewed all took place in aca- tive findings, notably an absence of a diac tamponade. However, 50% or demic medical centers, raising con- pulsus paradoxus, a jugular venous less of patients will have diminished cerns of referral bias and questions pressure that is not elevated, and no en- heart sounds, hypotension, or low about generalizability to the commu- largement of the cardiac silhouette on voltage on electrocardiogram. Based nity setting. The available evidence is chest radiograph. Knowing that these on 1 study, the presence of a pulsus also limited in that no studies have are relatively sensitive findings, you are paradoxus greater than 10 mm Hg analyzed the predictive value of a reassured that despite her known effu- increases the likelihood of cardiac combination of multiple findings. sion, cardiac tamponade is unlikely, and tamponade, while a pulsus paradoxus Finally, cardiac tamponade is a con- you entertain other possibilities to ac- of 10 mm Hg or less decreases the tinuum of hemodynamic effects and count for her symptoms. likelihood. The presence and degree the patients included in published of pulsus paradoxus may be helpful to studies are likely heterogeneous as to Case 2 predict the degree of hemodynamic their position on that continuum, so The patient has a pulsus paradoxus in the compromise in tamponade in patients the sensitivities and specificities of setting of a known pericardial effusion, with an effusion who are suspected of findings on clinical examination will and although the echocardiogram is non- having this condition. In general there vary depending on the severity of diagnostic and the rest of his examina- is a paucity of evidence supporting patients studied. tion is normal, this clinical finding alone the diagnostic accuracy of the clinical Two reviewed studies explicitly used increases the likelihood of tamponade examination for cardiac tamponade, invasive monitoring (ie, intra-arterial physiology. You arrange for urgent peri- and it must be stressed that all studies transducers), rather than a sphygmoma- cardiocentesis, which demonstrates el- enrolled patients with a previously nometer,tomeasurepulsusparadoxus.33,34 evated intrapericardial pressures and established pericardial effusion. We While one would expect that invasive equalization with intracardiac pres- found no studies that evaluated inter- measurement would be more accurate sures. Drainage of the effusion leads to rater reliability of clinical findings or than manual measurement, the literature resolution of the elevated pulsus para- that included patients not already lacks studies directly comparing the ac- doxus and improvement in symptoms. known or highly suspected to have curacy of these methods. Nonetheless, we

Downloaded from www.jama.com on April 24, 2007 DOES THIS PATIENT WITH A PERICARDIAL EFFUSION HAVE CARDIAC TAMPONADE? found the sensitivity of pulsus paradoxus 8. Singh S, Wann LS, Schuchard GH, et al. Right ven- diac tamponade: a pathophysiologic continuum. Clin tricular and right atrial collapse in patients with car- Cardiol. 2003;26:215-217. wassimilarinstudiesusingdifferentmea- diac tamponade—a combined echocardiographic and 29. Fowler NO. Physiology of cardiac tamponade and surement techniques.6,8 hemodynamic study. Circulation. 1984;70:966-971. pulsus paradoxus, I: mechanisms of pulsus para- 9. Singh S, Wann LS, Klopfenstein HS, Hartz A, Brooks doxus in cardiac tamponade. Mod Concepts Cardio- In summary, when faced with a pa- HL. Usefulness of right ventricular diastolic collapse vasc Dis. 1978;47:109-113. tient with a known pericardial effu- in diagnosing cardiac tamponade and comparison to 30. Hartert TV, Wheeler AP, Sheller JR. Use of pulse sion, the clinical examination may help pulsus paradoxus. Am J Cardiol. 1986;57:652-656. oximetry to recognize severity of airflow obstruction 10. Engel PJ, Hon H, Fowler NO, Plummer S. Echo- in obstructive airway disease: correlation with pulsus guide decisions about the appropriate- cardiographic study of right ventricular wall motion paradoxus. Chest. 1999;115:475-481. ness of expectant management or more in cardiac tamponade. Am J Cardiol. 1982;50:1018- 31. Frey B, Butt W. Pulse oximetry for assessment of 1021. pulsus paradoxus: a clinical study in children. Inten- urgent, invasive intervention. 11. Klopfenstein HS, Schuchard GH, Wann LS, et al. sive Care Med. 1998;24:242-246. The relative merits of pulsus paradoxus and right ven- 32. Orient JM, Sapira JD. Sapira’s Art & Science of Author Contributions: Dr Choudry had full access to tricular diastolic collapse in the early detection of car- Bedside Diagnosis. 3rd ed. Philadelphia, Pa: Lippin- all of the data in the study and takes responsibility for diac tamponade: an experimental echocardiographic cott Williams & Wilkins; 2005. the integrity of the data and the accuracy of the data study. Circulation. 1985;71:829-833. 33. Curtiss EI, Reddy PS, Uretsky BF, Cecchetti AA. analysis. 12. Fowler NO. Cardiac tamponade: a clinical or an Pulsus paradoxus: definition and relation to the se- Study concept and design: Roy, Choudhry. echocardiographic diagnosis? Circulation. verity of cardiac tamponade. Am Heart J. 1988;115: Acquisition of data: Roy, Minor, Choudhry. 1993;87:1738-1741. 391-398. Analysis and interpretation of data: Roy, Brookhart, 13. Spodick DH. Pericarditis, pericardial effusion, car- 34. Reddy PS, Curtiss EI, O’Toole JD, Shaver JA. Car- Choudhry. diac tamponade, and constriction. Crit Care Clin. 1989; diac tamponade: hemodynamic observations in man. Drafting of the manuscript: Roy, Choudhry. 5:455-476. Circulation. 1978;58:265-272. Critical revision of the manuscript for important in- 14. Fowler NO. Physiology of cardiac tamponade and 35. Spodick DH. Pericardial rub: prospective, mul- tellectual content: Roy, Minor, Brookhart, Choudhry. pulsus paradoxus, II: physiological, circulatory, and tiple observer investigation of pericardial friction in 100 Statistical analysis: Brookhart, Choudhry. pharmacological responses in cardiac tamponade. Mod patients. Am J Cardiol. 1975;35:357-362. Administrative, technical, or material support: Roy, Concepts Cardiovasc Dis. 1978;47:115-118. 36. Markiewicz W, Brik A, Brook G, Edoute Y, Minor, Choudhry. 15. Hancock EW. Cardiac tamponade. Med Clin North Monakier I, Markiewicz Y. Pericardial rub in pericar- Study supervision: Choudhry. Am. 1979;63:223-237. dial effusion: lack of correlation with amount of fluid. Financial Disclosures: None reported. 16. Reddy PS, Curtiss EI. Cardiac tamponade. Car- Chest. 1980;77:643-646. Acknowledgment: We thank Robert L. Trowbridge, MD diol Clin. 1990;8:627-637. 37. Goldberger AL, Shabetai R, Bhargava V, West (Maine Hospitalist Service, Maine Medical Center, Port- 17. Levy PY, Corey R, Berger P, et al. Etiologic diag- BJ, Mandell AJ. Nonlinear dynamics, electrical alter- land), and David H. Newman, MD (Department of nosis of 204 pericardial effusions. Medicine (Baltimore). nans, and pericardial tamponade. Am Heart J. 1984; Emergency Medicine, St Luke’s/Roosevelt Hospital Cen- 2003;82:385-391. 107:1297-1299. ter, New York, NY), for their helpful comments on an 18. Sagristà-Sauleda J, Merce J, Permanyer-Miralda 38. Heinsimer JA, Collins GJ, Burkman MH, Roberts earlier version of this article. Drs Trowbridge and New- G, Soler-Soler J. Clinical clues to the causes of large L Jr, Chen JT. Supine cross-table lateral chest roent- man received no compensation for their contributions. pericardial effusions. Am J Med. 2000;109:95-101. genogram for the detection of pericardial effusion. 19. Atar S, Chiu J, Forrester JS, Siegel RJ. Bloody peri- JAMA. 1987;257:3266-3268. REFERENCES cardial effusion in patients with cardiac tamponade: 39. Zipes DP, Braunwald E. Braunwald’s Heart Dis- is the cause cancerous, tuberculous, or iatrogenic in ease: A Textbook of Cardiovascular Medicine. 7th ed. 1. Reddy PS, Curtiss EI, Uretsky BF. Spectrum of he- the 1990s? Chest. 1999;116:1564-1569. Philadelphia, Pa: WB Saunders; 2005. modynamic changes in cardiac tamponade. Am J 20. Maisch B. Pericardial diseases, with a focus on eti- 40. Bates B, Bickley LS, Hoekelman RA. A Guide to Cardiol. 1990;66:1487-1491. ology, pathogenesis, pathophysiology, new diagnos- Physical Examination and History Taking. 6th ed. Phila- 2. Appleton CP, Hatle LK, Popp RL. Cardiac tamponade tic imaging methods, and treatment. Curr Opin Cardiol. delphia, Pa: Lippincott; 1995. and pericardial effusion: respiratory variation in transval- 1994;9:379-388. 41. Fuster V, Hurst JW. Hurst’s the Heart. 11th ed. vular flow velocities studied by Doppler echocardiography. 21. Weitzman LB, Tinker WP, Kronzon I, Cohen ML, New York, NY: McGraw-Hill; 2004. J Am Coll Cardiol. 1988;11:1020-1030. Glassman E, Spencer FC. The incidence and natural 42. Choudhry NK, Etchells EE. Does this patient have 3. Fowler NO. The significance of echocardiographic- history of pericardial effusion after cardiac sur- aortic regurgitation? JAMA. 1999;281:2231-2238. Doppler studies in cardiac tamponade. J Am Coll gery—an echocardiographic study. Circulation. 1984; 43. Klompas M. Does this patient have an acute tho- Cardiol. 1988;11:1031-1033. 69:506-511. racic aortic dissection? JAMA. 2002;287:2262-2272. 4. Shaver JA, Reddy PS, Curtiss EI, Ziady GM, Reddy 22. Kuvin JT, Harati NA, Pandian NG, Bojar RM, Khab- 44. Holleman DR Jr, Simel DL. Does the clinical ex- SC. Noninvasive/invasive correlates of exaggerated baz KR. Postoperative cardiac tamponade in the mod- amination predict airflow limitation? [published cor- ventricular interdependence in cardiac tamponade. ern surgical era. Ann Thorac Surg. 2002;74:1148-1153. rection appears in JAMA. 1995;273:1334]. JAMA. J Cardiol. 2001;37(suppl 1):71-76. 23. Spodick DH. Acute cardiac tamponade. N Engl J 1995;273:313-319. 5. Engel PJ. Echocardiography in pericardial disease. Med. 2003;349:684-690. 45. R: A Language and Environment for Statistical Cardiovasc Clin. 1983;13:181-200. 24. Beck C. Two cardiac compression triads. JAmMed Computing [computer program]. Version 2.2.0. Vi- 6. Levine MJ, Lorell BH, Diver DJ, Come PC. Impli- Assoc. 1935;104:714-716. enna, Austria: R Foundation for Statistical Comput- cations of echocardiographically assisted diagnosis of 25. Guberman BA, Fowler NO, Engel PJ, Gueron M, ing; 2003. pericardial tamponade in contemporary medical pa- Allen JM. Cardiac tamponade in medical patients. 46. Cooper JP, Oliver RM, Currie P, Walker JM, Swan- tients: detection before hemodynamic embarrassment. Circulation. 1981;64:633-640. ton RH. How do the clinical findings in patients with J Am Coll Cardiol. 1991;17:59-65. 26. Brown J, MacKinnon D, King A, Vanderbush E. pericardial effusions influence the success of aspiration? 7. Merce´ J, Sagrista-Sauleda J, Permanyer-Miralda Elevated arterial blood pressure in cardiac tamponade. Br Heart J. 1995;73:351-354. G, Evangelista A, Soler-Soler J. Correlation between N Engl J Med. 1992;327:463-466. 47. Gibbs CR, Watson RD, Singh SP, Lip GY. Man- clinical and Doppler echocardiographic findings in pa- 27. Kussmaul A. U¨ ber schwielige Mediastino- agement of pericardial effusion by drainage: a survey tients with moderate and large pericardial effusion: im- Perikarditis und den paradoxen Puls. Berl Klin of 10 years’ experience in a city centre general hospi- plications for the diagnosis of cardiac tamponade. Am Wochenschr. 1873;10:433-435. tal serving a multiracial population. Postgrad Med J. Heart J. 1999;138:759-764. 28. Swami A, Spodick DH. Pulsus paradoxus in car- 2000;76:809-813.

Downloaded from www.jama.com on April 24, 2007