Prognostic Value of Echocardiography in Hospitalized Patients with Pericardial Effusion Mark J. Eisenberg, MD, MPH, Keith Oken, MD, Salvador Guerrero, MD, Mohammad Ali Saniei, MD, and Nelson B. Schiller, MD ostpericardial effusionsobservedin hospitalized patients resolvespontaneouslyand causelittle if any clinical dilemma. Others, however,may lead to hemodynamic impairment and eventually require pericardiocentesisor surgical drainage. It is often difficult to predict which effusions will progress and which will resolve spontaneously. If clinical outcome could be predicted from information available at the time of diagnosis, patients at high risk for progression could be identified and treated aggressively. At the same time, patients at low risk for progression could also be identified, and conservativemanagementcould be used with greater assurance.Previous studies identified effusion etiology, rate of fluid accumulation and hemodynamic status as clinical factors that may be help ful in assessingprognosis.1-5Echocardiographic signs seenat the time of diagnosismay also provide prognostic information.6 Bight ventricular collapse,right atria1 collapse, and inferior vena cava (IVC) plethora with blunted responseto respiration are sensitiveand specific signs for the diagnosis of cardiac tamponade.‘-l6 Becausethesesigns are reflective of hemodynamic impairment, their presencein patients who are not in overt tamponademay help identify those at high risk for pre gression.To addressthis issue, we examined the echocardiogramsand clinical coursesof 187 hospitalized patients with pericardial effusion. Our purpose was to de termine the prognostic value of echocardiography in hospitalized patients with pericardial effusion.
M
It is often dHfkuR to predkt outcome in hospitalized patknts with perkardial effuskn. To address
this issue, the prognostic value of echocardiography was studied in 187 hospitalized patients diagnosed with perkardial effusions over a l-year period. The index echecardiogram showed that 11 &dons were large (6%), 39 were moderate (21944, and 137 were small (73%). Wght ventrkular cotbpse was present in 7% of cases (13 of 178), right atrtal collapse in 12% (21 of MS), and hderior vena cava (WC) pkthora with bhmted re= sponse to respiration in 35% (46 of 132). During the course of hospitalkatton, 9 patknts (5%) had cardiac tampenade and 16 (9%) had cardiac tamponade, perkardtocentests and/or surgkal drainage (combhred end point). By untvariate analysts, each echocardkgraphk sign was assocfated with both cardiac tamponads and the combtned end point (p 10.01 for comparkons with dze and right-shied chamber collapse; p SO.07 for comparisons with IVC plethora). When the data were analyzed wtth kgktk regresston modeling, effusien size was ths most pewerful predktor of eutcome (card&c tampenade: odds ratio Sl,S!5% conftdence interval 3.!5-729, p = 0.004; combhred end pohrt: odds ratto 78,S6% confidence interval 14421, p = O.OOOl), and neither rtght-&ted chamber coltapse nor IVC plethora wfth blunted mspense to resptratton retained stgntfkant assoctatiens. It is cenctuded that echocardiographkally determhred effuskn ske is a powerful predictor of outcome in hespttalized patknts wtth perkardtal effusion, and that right-stded chamber collapse and IVC plethora with blunted response to respfration add ltttle H any addtttonal prognostk information. (Am J Cardid lSS2;70:934-939
From the CardiovascularResearchInstitute, the Cardiology Division of the Denartment of Medicine. and the John Henrv Mills l?chocardiop;raphy L&oratory, University.of California, !SanWFrancisco, Califoha. Dr. F&.&erg was supported by Institutional National ResearchService Award HL 07192from the Training Program in Heart and Vascular Diseases,National Heart, Lung, and Blood Institute, Bethesda, Maryland. Manuscript received March 16, 1992; revised manuscript receivedand acceptedJune 11.1992. Addressfor reprints: Mark J. Eisenberg,MD, MPH, Moflitt-Long Hospital, University of California, 505PamassusAvenue, San Francisco, California 94143-0214. 934
THE AMERICAN JOURNAL OF CARDIOLOGY VOLUME 70
METHODS Patknt populrlion: The Moffitt-Long Hospital is a tertiary care medical center that serves as the main teaching hospital for the University of California at San Francisco (UCSF). We assembleda retrospective cohort of 187 consecutivepatients from the UCSF Echocardiography Data Base.This cohort comprised all patients who were diagnosedwith pericardial effusion by 2-dimensional echocardiography over a l-year period. Patients came from all medical and surgical services, including the Emergency Department. To be included, patients had to be either hospitalized at the time of the index echocardiogram or admitted to the hospital immediately after the study. If multiple echocardiograms were obtained during an admission, the first to document an effusion was used as the index study. Two hundred forty-two patients were initially identified as having pericardial effusions. Fifty-live patients were subsequentlyexcluded from the analysis (39 be cause they were outpatients, 13 becausetheir echocardiograms were technically inadequate or showed questionable evidence of effusion, 2 becausetheir echocar-
OCTOBER 1. 1992
TABLE I Admission Diagnoses with Pericardial Effusion
of 187 Hospitalized
Patients
TABLE
II
Etiology
Effusion
in 187 Hospitalized
30 26 16 14 11 11 10 8 7 6 5 5 4 3 3 2
(16) (14) (9) (7) (6) (6) (5) (4) (4) (3) (3) (3) (2) (2) (2) (1) 2 (1) 2 (1) 1 (0.5) 1 10.5) 1 (0.5) 1 (0.5) 1 (0.5) 1 (0.5) 1 (0.5) 1 (0.5) 1 (0.5) 1 (0.5) 1 (0.5) 1 (0.5) 1 (0.5) 1 (0.5) 1 (0.5) 1 (0.5) 1 (0.5) 1 (0.5) 1 (0.5) 1 (0.5) 1 (0.5) 1 (0.5) 187 (100)
No. (%)
Etiology
No. (%I Chest pain/unstable angina/possible myocardial infarction Cancer complication Congestive heart failure End-stage liver disease Valvular heart disease End-stage renal failure Organ transplant complication Arrhythmia Complication of acquired immunodeficiency syndrome Elective bypass surgery Accessory pathway ablation procedure Possible sepsis Heart transplant Endocarditis Pericarditislpericardial effusion/cardiac tamponade Abdominal pain Peripheral edema Gastrointestinal bleeding Amyloidosis Skin infection Pelvic pain Elective repair of congenital heart condition Nausea/vomiting Stroke Pneumonia Prosthetic valve dehiscence Secondary hyperparathyroidism Exacerbation of chronic obstructive pulmonary disease Aortic dissection Pulmonary atresia Abdominal aortic aneurysm Premature rupture of membranes Jaundice Syncope Hypertension Sudden death Pleural effusion Dysphagia Pathologic fracture Fever of unknown origin Total
of Pericardial
Patients
Idiopathic Cardiothoracic surgery Neoplasm End-stage renal disease Congestive heart failure Myocardial infarctron Infectron Accessory pathway ablation procedure Total
83 21 21 20 18 10 8 6 187
(45) (11) (11) (11) (10) (5) (4) (3) (100)
Effusion etiologm were based on clinical and historical data. Vety few pericard~al effusions were definitively diagnosed by pericardiocentesis CNexamination of pericardial tissue.
were defined as a collapse of >‘A of chamber area in any view. IVC plethora with blunted responseto respiration was defined as a decreaseof 100 beats/mm and systolic blood pressurel the effusions at the time of diagnosis (Table II). Index cm in width at any point. (Small effusionswere estimat- echocardiogramsshowed that 11 effusions were large ed to be < 100 ml in volume, moderate 100 to 500 and (6%), 39 were moderate (21%), and 137 were small large >500.) Bight ventricular and right atria1 collapse (73%). Bight ventricular collapse was present in 7% of PERICARDIAL
EFFUSION
IN HOSPITALIZED
PATIENTS
935
rA6l.E III Clinical and Echocardiographic Characteristics of the 16 Patients with Susion-Related Outcomes Index Echocardiogram
Effusion Etiology
RV Collapse
RA Collapse
IVC Plethora
Largetn = 11) Neoplasm End-stage renal disease Idiopathic Neoplasm Neoplasm Cardiothoracic surgery Idiopathic Idiopathic
+ + + + + +
+ t + t + + + t
t t + + + + -I-
-
t -
+
-I-
VO.
1 2 3 4 5 6 7 8
1 2 3 4
1 2 3 4
Outcomes*
Moderate (n = 39) Ablation procedure Ablation procedure Infection Infection Small (n = 137) Infection Myocardial infarction Congestive heart failure Congestive heart failure
t -
t -
Pericardiocentesis
Surgical Drainage
Cardiac Tamponade
1 0 0 0
-
-
1
-
1
-
7 0 12
-
0 O
t t +
0 0 0
-
-
t + t
0 -
-
0 0 0
0 0
3 -
0 0 -
2
21 -
-
-
7 9
9
*Numbers refer to the number of days between the index echocardiogram and the effusion-relatedoutcome. Outcome on ay 0 occurred on same day as index echocardiogram. tlmage not adequatefor assessment. IVC = inferiorvena cava; RA = right atrial; RV = right ventricular; + = present; - = absent,
caseswhen the right ventricle was adequatelyvisualized (13 of 178), right atria1 collapsein 12%(21 of 168), and IVC plethora with blunted responseto respiration in 35% (46 of 132). During the course of hospitalization, 9 patients (5%) had cardiac tamponade, 11 (6%) had pericardiocentesis, 8 (4%) had surgical drainage, and 16 (9%) had >,1 of
iiE 2 T1 0
Ai
3
INDEX
5
7
9
11
13
15
17
19
21
DAYS
ECHOCARDIOGRAM
939
THE AMERICANJOURNALOF CARDIOLOGY VOLUME70
these outcomes(Table III). Overall mortality was 11% (20 of 187). Almost all outcomes occurred within the first few days after the index echocardiogram (Figure 1). Of 9 patients who had tamponade,7 were diagnosed on the same day as the index echocardiogram, 1 was diagnoseda day later, and 1 was diagnosed9 days later. This last patient had a small effusion on the index echocardiogram; the patient subsequentlyhad coronary artery bypasssurgery, after which a retrocardiac hematoma (adjacent to the right atrium) developedthat was responsiblefor cardiac tamponade. Of 7 patients who had drainage procedures but never developedcardiac tamponade,most had incidental drainage of their effusions during cardiac surgery that was performed for other reasons. Rognorticv~of~Amongpatients with large effusions, 73% (8 of 11) had tamponade and/or drainage procedures while in the hospital (Table III) comparedwith 44% (4 of 39) among those with moderateeffusionsand 3% (4 of 137) among those with small effusions.Among patients with right ventricular collapse, 54% (7 of 13) had tamponade and/or drainage procedures compared with 29% (6 of 21) among those with right atrial collapse and 15% (7 of 46) among those with IVC plethora with blunted re sponseto respiration. By univariate analysis, each echocardiographic sign was associatedwith both cardiac tamponade and the combinedend point (Table IV). Theseassociationswere present when all 187 patients were examined and when only the 50 with large and moderate effusions were examined. When the echocardiographicsigns were examined in a multivariate logistic regressionanalysis after effusion size was entered, neither right-sided chamber
OCTOBER1, 1992
TABLE IV Univariate and Multivariate Analyses of Echocardiographic Outcome
Cardiac Tamponade
Multivariate models Size* Right ventricular collapse Right atrial collapse IVCplethora Large & moderate effusions (n = 50) Univariate models Size Right ventricular collapse Right atrial collapse IVC plethora Multivariate models Size Right ventricular collapse Right atrial collapse IVC plethora
Cardiac Tamponade, Pericardiocentesis and/or Surgical Drainage
p Value
Odds Ratio (95% Cl)
p Value
46 (10-224) 12 (2-77) 8 (l-75)
0.0001 0.0001 0.009 0.07
78 (14-421) 23 (6-84) 11 (3-42) 5 (l-20)
0.0001 0.0001 0.0003 0.03
51 (4-729) 8 (2-55) -
0.004 0.04 -
78 (14-421) -
0.0001 -
22 (4-142) 22 (3-138) 8 (l-82) 10 (l-100)
0.001 0.0012 0.09 0.05
23 (4-126) 12 (3-54) 10 (260) 9 (2-55)
0.0002 0.002 0.01 0.02
9 (l-72)
0.04t
23 (4-126)
9 (l-72) -
0.04t -
0.0002 -
Odds Ratio (95% Cl) All effusions (n = 187) Univariate models Size* Right ventricular collapse Right atrial collapse IVC plethora
Signs and In-Hospital
215
(20-2,343)
*Odds ratios refer to comparw~ns between patients with large effus~ons and those with small + moderate effusions. tThere was no significant difference between the multivariate model that included size alone and the model that included both size and right 65%, and negative predictive values were L95%, but positive predictive values were
M
It is often dHfkuR to predkt outcome in hospitalized patknts with perkardial effuskn. To address
this issue, the prognostic value of echocardiography was studied in 187 hospitalized patients diagnosed with perkardial effusions over a l-year period. The index echecardiogram showed that 11 &dons were large (6%), 39 were moderate (21944, and 137 were small (73%). Wght ventrkular cotbpse was present in 7% of cases (13 of 178), right atrtal collapse in 12% (21 of MS), and hderior vena cava (WC) pkthora with bhmted re= sponse to respiration in 35% (46 of 132). During the course of hospitalkatton, 9 patknts (5%) had cardiac tampenade and 16 (9%) had cardiac tamponade, perkardtocentests and/or surgkal drainage (combhred end point). By untvariate analysts, each echocardkgraphk sign was assocfated with both cardiac tamponads and the combtned end point (p 10.01 for comparkons with dze and right-shied chamber collapse; p SO.07 for comparisons with IVC plethora). When the data were analyzed wtth kgktk regresston modeling, effusien size was ths most pewerful predktor of eutcome (card&c tampenade: odds ratio Sl,S!5% conftdence interval 3.!5-729, p = 0.004; combhred end pohrt: odds ratto 78,S6% confidence interval 14421, p = O.OOOl), and neither rtght-&ted chamber coltapse nor IVC plethora wfth blunted mspense to resptratton retained stgntfkant assoctatiens. It is cenctuded that echocardiographkally determhred effuskn ske is a powerful predictor of outcome in hespttalized patknts wtth perkardtal effusion, and that right-stded chamber collapse and IVC plethora with blunted response to respfration add ltttle H any addtttonal prognostk information. (Am J Cardid lSS2;70:934-939
From the CardiovascularResearchInstitute, the Cardiology Division of the Denartment of Medicine. and the John Henrv Mills l?chocardiop;raphy L&oratory, University.of California, !SanWFrancisco, Califoha. Dr. F&.&erg was supported by Institutional National ResearchService Award HL 07192from the Training Program in Heart and Vascular Diseases,National Heart, Lung, and Blood Institute, Bethesda, Maryland. Manuscript received March 16, 1992; revised manuscript receivedand acceptedJune 11.1992. Addressfor reprints: Mark J. Eisenberg,MD, MPH, Moflitt-Long Hospital, University of California, 505PamassusAvenue, San Francisco, California 94143-0214. 934
THE AMERICAN JOURNAL OF CARDIOLOGY VOLUME 70
METHODS Patknt populrlion: The Moffitt-Long Hospital is a tertiary care medical center that serves as the main teaching hospital for the University of California at San Francisco (UCSF). We assembleda retrospective cohort of 187 consecutivepatients from the UCSF Echocardiography Data Base.This cohort comprised all patients who were diagnosedwith pericardial effusion by 2-dimensional echocardiography over a l-year period. Patients came from all medical and surgical services, including the Emergency Department. To be included, patients had to be either hospitalized at the time of the index echocardiogram or admitted to the hospital immediately after the study. If multiple echocardiograms were obtained during an admission, the first to document an effusion was used as the index study. Two hundred forty-two patients were initially identified as having pericardial effusions. Fifty-live patients were subsequentlyexcluded from the analysis (39 be cause they were outpatients, 13 becausetheir echocardiograms were technically inadequate or showed questionable evidence of effusion, 2 becausetheir echocar-
OCTOBER 1. 1992
TABLE I Admission Diagnoses with Pericardial Effusion
of 187 Hospitalized
Patients
TABLE
II
Etiology
Effusion
in 187 Hospitalized
30 26 16 14 11 11 10 8 7 6 5 5 4 3 3 2
(16) (14) (9) (7) (6) (6) (5) (4) (4) (3) (3) (3) (2) (2) (2) (1) 2 (1) 2 (1) 1 (0.5) 1 10.5) 1 (0.5) 1 (0.5) 1 (0.5) 1 (0.5) 1 (0.5) 1 (0.5) 1 (0.5) 1 (0.5) 1 (0.5) 1 (0.5) 1 (0.5) 1 (0.5) 1 (0.5) 1 (0.5) 1 (0.5) 1 (0.5) 1 (0.5) 1 (0.5) 1 (0.5) 1 (0.5) 187 (100)
No. (%)
Etiology
No. (%I Chest pain/unstable angina/possible myocardial infarction Cancer complication Congestive heart failure End-stage liver disease Valvular heart disease End-stage renal failure Organ transplant complication Arrhythmia Complication of acquired immunodeficiency syndrome Elective bypass surgery Accessory pathway ablation procedure Possible sepsis Heart transplant Endocarditis Pericarditislpericardial effusion/cardiac tamponade Abdominal pain Peripheral edema Gastrointestinal bleeding Amyloidosis Skin infection Pelvic pain Elective repair of congenital heart condition Nausea/vomiting Stroke Pneumonia Prosthetic valve dehiscence Secondary hyperparathyroidism Exacerbation of chronic obstructive pulmonary disease Aortic dissection Pulmonary atresia Abdominal aortic aneurysm Premature rupture of membranes Jaundice Syncope Hypertension Sudden death Pleural effusion Dysphagia Pathologic fracture Fever of unknown origin Total
of Pericardial
Patients
Idiopathic Cardiothoracic surgery Neoplasm End-stage renal disease Congestive heart failure Myocardial infarctron Infectron Accessory pathway ablation procedure Total
83 21 21 20 18 10 8 6 187
(45) (11) (11) (11) (10) (5) (4) (3) (100)
Effusion etiologm were based on clinical and historical data. Vety few pericard~al effusions were definitively diagnosed by pericardiocentesis CNexamination of pericardial tissue.
were defined as a collapse of >‘A of chamber area in any view. IVC plethora with blunted responseto respiration was defined as a decreaseof 100 beats/mm and systolic blood pressurel the effusions at the time of diagnosis (Table II). Index cm in width at any point. (Small effusionswere estimat- echocardiogramsshowed that 11 effusions were large ed to be < 100 ml in volume, moderate 100 to 500 and (6%), 39 were moderate (21%), and 137 were small large >500.) Bight ventricular and right atria1 collapse (73%). Bight ventricular collapse was present in 7% of PERICARDIAL
EFFUSION
IN HOSPITALIZED
PATIENTS
935
rA6l.E III Clinical and Echocardiographic Characteristics of the 16 Patients with Susion-Related Outcomes Index Echocardiogram
Effusion Etiology
RV Collapse
RA Collapse
IVC Plethora
Largetn = 11) Neoplasm End-stage renal disease Idiopathic Neoplasm Neoplasm Cardiothoracic surgery Idiopathic Idiopathic
+ + + + + +
+ t + t + + + t
t t + + + + -I-
-
t -
+
-I-
VO.
1 2 3 4 5 6 7 8
1 2 3 4
1 2 3 4
Outcomes*
Moderate (n = 39) Ablation procedure Ablation procedure Infection Infection Small (n = 137) Infection Myocardial infarction Congestive heart failure Congestive heart failure
t -
t -
Pericardiocentesis
Surgical Drainage
Cardiac Tamponade
1 0 0 0
-
-
1
-
1
-
7 0 12
-
0 O
t t +
0 0 0
-
-
t + t
0 -
-
0 0 0
0 0
3 -
0 0 -
2
21 -
-
-
7 9
9
*Numbers refer to the number of days between the index echocardiogram and the effusion-relatedoutcome. Outcome on ay 0 occurred on same day as index echocardiogram. tlmage not adequatefor assessment. IVC = inferiorvena cava; RA = right atrial; RV = right ventricular; + = present; - = absent,
caseswhen the right ventricle was adequatelyvisualized (13 of 178), right atria1 collapsein 12%(21 of 168), and IVC plethora with blunted responseto respiration in 35% (46 of 132). During the course of hospitalization, 9 patients (5%) had cardiac tamponade, 11 (6%) had pericardiocentesis, 8 (4%) had surgical drainage, and 16 (9%) had >,1 of
iiE 2 T1 0
Ai
3
INDEX
5
7
9
11
13
15
17
19
21
DAYS
ECHOCARDIOGRAM
939
THE AMERICANJOURNALOF CARDIOLOGY VOLUME70
these outcomes(Table III). Overall mortality was 11% (20 of 187). Almost all outcomes occurred within the first few days after the index echocardiogram (Figure 1). Of 9 patients who had tamponade,7 were diagnosed on the same day as the index echocardiogram, 1 was diagnoseda day later, and 1 was diagnosed9 days later. This last patient had a small effusion on the index echocardiogram; the patient subsequentlyhad coronary artery bypasssurgery, after which a retrocardiac hematoma (adjacent to the right atrium) developedthat was responsiblefor cardiac tamponade. Of 7 patients who had drainage procedures but never developedcardiac tamponade,most had incidental drainage of their effusions during cardiac surgery that was performed for other reasons. Rognorticv~of~Amongpatients with large effusions, 73% (8 of 11) had tamponade and/or drainage procedures while in the hospital (Table III) comparedwith 44% (4 of 39) among those with moderateeffusionsand 3% (4 of 137) among those with small effusions.Among patients with right ventricular collapse, 54% (7 of 13) had tamponade and/or drainage procedures compared with 29% (6 of 21) among those with right atrial collapse and 15% (7 of 46) among those with IVC plethora with blunted re sponseto respiration. By univariate analysis, each echocardiographic sign was associatedwith both cardiac tamponade and the combinedend point (Table IV). Theseassociationswere present when all 187 patients were examined and when only the 50 with large and moderate effusions were examined. When the echocardiographicsigns were examined in a multivariate logistic regressionanalysis after effusion size was entered, neither right-sided chamber
OCTOBER1, 1992
TABLE IV Univariate and Multivariate Analyses of Echocardiographic Outcome
Cardiac Tamponade
Multivariate models Size* Right ventricular collapse Right atrial collapse IVCplethora Large & moderate effusions (n = 50) Univariate models Size Right ventricular collapse Right atrial collapse IVC plethora Multivariate models Size Right ventricular collapse Right atrial collapse IVC plethora
Cardiac Tamponade, Pericardiocentesis and/or Surgical Drainage
p Value
Odds Ratio (95% Cl)
p Value
46 (10-224) 12 (2-77) 8 (l-75)
0.0001 0.0001 0.009 0.07
78 (14-421) 23 (6-84) 11 (3-42) 5 (l-20)
0.0001 0.0001 0.0003 0.03
51 (4-729) 8 (2-55) -
0.004 0.04 -
78 (14-421) -
0.0001 -
22 (4-142) 22 (3-138) 8 (l-82) 10 (l-100)
0.001 0.0012 0.09 0.05
23 (4-126) 12 (3-54) 10 (260) 9 (2-55)
0.0002 0.002 0.01 0.02
9 (l-72)
0.04t
23 (4-126)
9 (l-72) -
0.04t -
0.0002 -
Odds Ratio (95% Cl) All effusions (n = 187) Univariate models Size* Right ventricular collapse Right atrial collapse IVC plethora
Signs and In-Hospital
215
(20-2,343)
*Odds ratios refer to comparw~ns between patients with large effus~ons and those with small + moderate effusions. tThere was no significant difference between the multivariate model that included size alone and the model that included both size and right 65%, and negative predictive values were L95%, but positive predictive values were
