Original Research

Cardiology 2015;130:37–45 Received: July 14, 2014

Accepted after revision: September 30, 2014
DOI: 10.1159/000368796
Published online: December 10, 2014

Pericardiocentesis Adverse Event Risk

Factors: A Nationwide Population-Based
Cohort Study
Ming-Yun Ho a Jian-Liang Wang b Yu-Sheng Lin c Chun-Tai Mao d
Ming-Lung Tsai a Ming-Shien Wen a Chun-Chieh Wang a I-Chang Hsieh a
Kuo-Chun Hung a Chao-Yung Wang a Huang-Pin Wu e Tien-Hsing Chen a, f, g
a
Division of Cardiology, Chang-Gung Memorial Hospital, Linkou, b Division of Cardiology, Landseed Hospital,
Pingzhen City, c Division of Cardiology, Chang-Gung Memorial Hospital, Chaiyi, Divisions of d Cardiology and
e
Pulmonary, Critical Care and Sleep Medicine, Chang-Gung Memorial Hospital, Keelung, and f Chang Gung University
College of Medicine, Taoyuan, Taiwan; g Department of Cardiology, Chang-Gung Memorial Hospital, Xiamen, China

For editorial comment see p. 34

Key Words p = 0.003). In-hospital death numbers and complication risks

Pericardial effusion · Pericardiocentesis · Recurrence rate · (OR 2.38, p < 0.001; OR 1.27, p = 0.01) were greater in the
Complication · Mortality catheter-related cardiac procedure group than in the other
groups. Conclusions: Malignant neoplasms and catheter-
based cardiac procedures have become major risk factors for
Abstract adverse events in patients receiving pericardiocentesis in
Objectives: Echocardiography-guided pericardiocentesis Taiwan. Malignancy leads to an increase in recurrence and
has been the leading procedure for diagnosis and therapy of in-hospital mortality but is associated with a lower rate of
pericardial effusion. We aimed to identify risk factors for re- acute complications. Cardiac catheterization procedures
currence, complications, and mortality in pericardial effu- and surgery increase both complications and in-hospital
sion patients treated with pericardiocentesis. Methods: We mortality. © 2014 S. Karger AG, Basel
identified and collected data from 8,101 patients receiving
pericardiocentesis between 1997 and 2010 from the Taiwan
National Health Insurance Research Database. A multivariate
regression model was used to investigate risk factors for re- Introduction
currence, complications, and death. Results: There were
8,565 admissions among 8,101 patients. The most common Over the past few decades, percutaneous pericardio-
underlying condition was malignancy (41%), especially lung centesis has been the primary diagnostic tool for massive
cancer (23%), tuberculosis (9.0%), and acute pericarditis pericardial effusion and the primary treatment for this
(8.2%). Surgical drainage was required in 12.7% of cases. Re-
currence was more likely in patients with malignancy (HR
Chang Gung Memorial Hospital - LinKou Branch

2.20, p < 0.001), but complications were less likely (OR 0.52, Ming-Yun Ho and Jian-Liang Wang contributed equally to this work.
60.248.19.7 - 12/10/2014 10:16:56 AM

© 2014 S. Karger AG, Basel Dr. Tien-Hsing Chen

0008–6312/14/1301–0037$39.50/0 Division of Cardiology, Department of Internal Medicine
Chang Gung Memorial Hospital, No. 5, Fu-Shin Street
Downloaded by:

E-Mail karger@karger.com
Kweishan, Taoyuan 333 (Taiwan)
www.karger.com/crd
E-Mail skyheart0826 @ gmail.com
condition in addition to surgical treatment [1, 2]. This cation of Diseases, 9th revision (ICD-9) codes 512 and 512.2] and
technique is not only an effective approach for the imme- any cardiac perforation repaired via surgery while the pericardio-
centesis was being conducted. In-hospital deaths were identified
diate relief of symptomatic pericardial effusion or acute using the hospitalization dataset of the NHIRD. Those marked as
cardiac tamponade [3–5] but also a potential treatment dead in the ICD-9 or who withdrew from the NHI program were
for the recurrent pericardial effusion caused by the termi- all considered dead.
nal stage of malignancy. However, the most effective
strategy for managing pericardial effusions in cases of im- Covariate Assessment
To investigate the effects of covariates on outcome, we includ-
minent or recurrent tamponade has not been determined ed the following covariates: underlying condition, coexisting car-
yet [6, 7]. diac surgery, catheter-based cardiac procedure, and surgical drain-
Although it is a rapid and effective treatment for these age requirement. All underlying conditions were recorded based
conditions, percutaneous pericardiocentesis can cause on the ICD-9 among hospitals affiliated with the NHI program,
complications such as heart chamber perforation and and they were categorized into: (1) malignancy (140–208), (2) car-
diovascular disease (acute myocardial infarction, 410; aortic dis-
pneumothorax [8, 9]. Previous large-scale studies have section, 441; heart failure, 448, and ischemic heart disease, 410,
reported a greater number of malignant neoplasm-relat- 411, 412, and 413), (3) pericardial disease (acute pericarditis, 420;
ed effusions, a greater number of catheter-based cardiac rheumatic pericarditis, 391.0 and 393, and adhesive and constric-
procedures used for treatment, and an increasing inci- tive pericardium, 423.0 and 423.2), (4) systemic disease (autoim-
dence of complications and recurrences in recent times mune disease, 136.1, 443.1, 446, 555, 556, 694.4, 710, and 714; hy-
pothyroidism, 243 and 244, and amyloidosis, 277.3), (5) chronic
compared to earlier times [9–11]. However, there have kidney disease with or without dialysis (403, 404, and 585), and (6)
been few investigations on risk factors associated with re- infectious disease (tuberculosis, 011–018, and HIV, 042, and
currence, complications, and survival. 079.53). Coexisting cardiac surgery included valve repair, coro-
To identify such risk factors, we conducted a compre- nary artery bypass surgery, and aortic or aneurysm repair. Cathe-
hensive study on a registry-based population of individu- ter-related cardiac procedures – decoded from the reimbursed in-
patient dataset of the NHIRD – were temporary or permanent
als who underwent pericardiocentesis between 1997 and pacemaker insertion, radiofrequency ablation, atrial transseptum
2010 in Taiwan. We examined risk factors for recurrence, puncture, and percutaneous transluminal coronary angioplasty.
complications, and survival, as well as the patterns of un-
derlying conditions, coexisting catheter-based cardiac Statistical Analysis
procedures, and cardiothoracic surgery over time. Continuous variables are presented as means ± SD, and cate-
gorical variables are presented as numbers (%). Differences were
compared using analysis of variance (ANOVA) for continuous
variables and a χ2 test or Fisher’s exact test with Yate’s correction
Methods if a cell number was <5 or close to zero for categorical variables.
Multiple linear regression was used to analyze associations be-
Study Source and Population tween pulmonary function and independent variables. Logistic re-
This population-based cohort study was approved by the Ethics gression models were used to examine associated risk factors for
Institutional Review Board of Chang Gung Memorial Hospital. complications and in-hospital death. Moreover, Cox proportional
We retrieved data from the National Health Insurance Research hazards models were used to scrutinize the associated risk factors
Database (NHIRD), a nationwide registry with health care infor- for recurrence and mortality. The results of logistic regression
mation on more than 25 million people covering up to 99.91% of models and Cox regression models were summarized using odds
the population of Taiwan that has been used for a number of stud- ratios (OR) and hazard ratios (HR), respectively, both with 95%
ies on risk factors [12–15]. The accuracy of this database is main- confidence intervals (CI). All statistical assessments were two-sid-
tained through regular reviews by the National Health Institute ed, and p < 0.05 was considered statistically significant. Statistical
(NHI) Bureau. analyses were performed using SPSS 18.0 statistical software (SPSS
Patients in this database receiving pericardiocentesis between Inc., Chicago, Ill., USA).
1997 and 2010 were enrolled into this study. Their initial pericar-
diocentesis date was considered to be the index hospitalization date,
and their inpatient records were reviewed and followed-up until
either the date of death or December 31, 2010, whichever occurred Results
earlier. All subjects were analyzed by study period (1997–2000,
2001–2005, and 2006–2010) based on the following outcome status: A total of 8,101 patients admitted 8,565 times for treat-
recurrence, complications, in-hospital death, and all-cause death. ment with percutaneous pericardiocentesis were identi-
fied. The average age was 61.3 ± 18.5 years; the median
Outcomes of Interest
Recurrence requiring repeat pericardiocentesis was defined as follow-up duration was 0.65 years (IQR 0.12–3.00).
any pericardiocentesis performed after the index hospitalization Table 1 summarizes the annual patient number, ad-
Chang Gung Memorial Hospital - LinKou Branch

date; complications referred to pneumothorax [Internal Classifi- missions, recurrences, complications, in-hospital deaths,
60.248.19.7 - 12/10/2014 10:16:56 AM

38 Cardiology 2015;130:37–45 Ho/Wang/Lin/Mao/Tsai/Wen/Wang/

DOI: 10.1159/000368796 Hsieh/Hung/Wang/Wu/Chen
Downloaded by:
 Table 1. Annual number of procedures, recurrences, complications, and deaths in pericardiocentesis patients in
Taiwan during 1997–2010

Year Patients, n Admissions, n Recurrences, n Complications, n In-hospital Total

deaths, n deaths, n

1997 382 401 27 5 58 282

1998 394 433 42 7 48 283
1999 419 444 26 7 58 300
2000 468 500 30 8 58 356
2001 493 525 39 6 45 354
2002 548 590 43 9 57 380
2003 563 602 40 5 67 410
2004 661 703 46 10 80 441
2005 636 673 36 11 81 432
2006 666 709 55 8 88 432
2007 660 727 52 13 144 426
2008 738 780 46 11 149 410
2009 750 803 51 19 176 407
2010 723 766 22 13 157 276
Total 8,101 8,656 555 132 1,266 5,189

and total deaths by year during the study period. The dial disease, chronic kidney disease, postpericardiotomy
number of patients, admissions, and in-hospital deaths syndrome or catheter-related cardiac procedures, and a
gradually increased over this time period from 382, 401, surgical drainage requirement had significantly decreased
and 58, respectively, in 1997 to 723, 766, and 157 in 2010. risks of recurrence (HR 0.17–0.79; table 3). Catheter-re-
The total number of deaths decreased to 293 by 2010 after lated cardiac procedures and a requirement for surgical
increasing to a summit of 441 in 2004, and the number of drainage were significantly associated with an increased
recurrences and complications remained relatively stable risk of complications (HR 2.38 and 1.85, respectively).
during the study period. Malignancy was associated with a lower risk of complica-
Table 2 shows the patient characteristics by study pe- tions (HR 0.52).
riod (1997–2000, 2001–2005, and 2006–2010). The aver- Conditions or events significantly associated with an
age patient age was significantly lower in the first study increased risk of in-hospital death were: aortic dissection
period than in the 2 later periods. The proportion of pa- (HR 2.29), malignancy (HR 1.58), pericardiocentesis per-
tients with any malignancy, or with gastrointestinal tract formed during the 2006–2010 time period (HR 1.57), co-
cancer, liver and biliary tract cancer, or lung and pleural existing cardiac surgery (HR 1.46), pericardiocentesis
cancer increased over the 3 time periods, as did the pro- performed in a local community hospital (HR 1.42), and
portion of patients with ischemic heart disease, chronic male gender (HR 1.2). Pericardial disease and infectious
kidney disease, and HIV carrier status. The proportion of disease were associated with a lower risk of in-hospital
patients with postpericardiotomy syndrome and receiv- death (HR 0.65 and 0.59, respectively). Conditions or
ing the catheter-related cardiac procedures of radiofre- events associated with an increased risk of all-cause death
quency ablation of arrhythmias and percutaneous trans- were malignancy (HR 2.88), aortic dissection (1.49),
luminal coronary angioplasty also increased over the chronic kidney disease (HR 1.27), age per 10 years (HR
3 time periods. As for deaths, in-hospital deaths increased 1.14), and male gender (HR 1.14). Catheter-related car-
and overall deaths decreased over the 3 periods (table 2). diac procedures (HR 0.89), infection (HR 0.72), pericar-
Table 3 shows the HR for clinical characteristics asso- dial disease (HR 0.68), and systemic disease (0.69) were
ciated with recurrence, complications, in-hospital death, associated with a significantly decreased risk of all-cause
and overall death. Malignancy and pericardiocentesis death.
performed in a local community hospital were signifi- Table 4 shows a subgroup analysis of recurrence ac-
cantly associated with an increased risk of recurrence cording to malignancy type. Lung and pleural cancer ac-
Chang Gung Memorial Hospital - LinKou Branch

(HR 2.07 and 2.14, respectively). Male gender, pericar- counted for 9.9% of recurrences (HR 3.01, p < 0.001),
60.248.19.7 - 12/10/2014 10:16:56 AM

Risk Factors in Pericardiocentesis Cardiology 2015;130:37–45 39

DOI: 10.1159/000368796
Downloaded by:
 Table 2. Characteristics of 8,101 patients with pericardiocentesis by time period
40
Variable Overall Period p value
1 2 3
1997–2000 2001–2005 2006–2010

Admissions, n (% of total) 8,565 1,778 (20.5) 3,093 (35.7) 3,785 (43.7)

Demographics
DOI: 10.1159/000368796
Cardiology 2015;130:37–45

Patients, n (% of total) 8,101 1,663 (20.5) 2,901 (35.8) 3,537 (43.7)

Mean age, years 61.3±18.5 58.6±20.0 61.5±18.1 62.5±18.0 <0.001 ,f

Age group (years) <0.001

<10 176 (2.2) 72 (4.3) 52 (1.8) 52 (1.5)
10–19 124 (1.5) 31 (1.9) 48 (1.7) 45 (1.3)
20–29 221 (2.7) 45 (2.7) 81 (2.8) 95 (2.7)
30–39 414 (5.1) 113 (6.8) 146 (5.0) 155 (4.4)
40–49 976 (12.0) 198 (11.9) 358 (12.3) 420 (11.9)
50–59 1,357 (16.8) 235 (14.1) 442 (15.2) 680 (19.2)
60–69 1,735 (21.4) 420 (25.3) 657 (22.6) 658 (18.6)
70–79 2,054 (25.4) 403 (24.2) 764 (26.3) 887 (25.1)
80–89 963 (11.9) 138 (8.3) 323 (11.1) 502 (14.2)
≥90 81 (1.0) 8 (0.5) 30 (1.0) 43 (1.2)
Male gender 4,718 (58.2) 932 (56.0) 1,729 (59.6) 2,057 (58.2) 0.063
Underlying conditions
Malignancya
All 3,317 (40.9) 626 (37.6) 1,154 (39.8) 1,537 (43.5) <0.001
Head and neck 82 (1.0) 15 (0.9) 35 (1.2) 32 (0.9) 0.427
GI tract 284 (3.5) 43 (2.6) 99 (3.4) 142 (4.0) 0.031
Liver and biliary tract 84 (1.0) 13 (0.8) 18 (0.6) 53 (1.5) 0.001
Lung and pleural 1,874 (23.1) 344 (20.7) 653 (22.5) 877 (24.8) 0.003
Heart and mediastinum 28 (0.3) 3 (0.2) 15 (0.5) 10 (0.3) 0.122
Hematologic 211 (2.6) 53 (3.2) 64 (2.2) 94 (2.7) 0.130
Hsieh/Hung/Wang/Wu/Chen
Ho/Wang/Lin/Mao/Tsai/Wen/Wang/

Breast 261 (3.2) 54 (3.2) 88 (3.0) 119 (3.4) 0.754

Other 493 (6.1) 101 (6.1) 182 (6.3) 210 (5.9) 0.854
Cardiovascular diseasea
Acute myocardial infarction 568 (7.0) 118 (7.1) 191 (6.6) 259 (7.3) 0.508
Aortic dissection 171 (2.1) 29 (1.7) 63 (2.2) 80 (2.3) 0.385
Heart failure 919 (11.3) 173 (10.4) 343 (11.8) 403 (11.4) 0.344
Ischemic heart disease 1,050 (13.0) 173 (10.4) 371 (12.8) 506 (14.3) <0.001
Pericardial diseasea
Acute pericarditis 666 (8.2) 146 (8.8) 264 (9.1) 256 (7.2) 0.017
Rheumatic pericarditis 5 (0.1) 1 (0.1) 4 (0.1) 0 (0.0) 0.086
Adhesive and constrictive pericardium 90 (1.1) 24 (1.4) 34 (1.2) 33 (0.9) 0.326

Downloaded by:
Chang Gung Memorial Hospital - LinKou Branch
60.248.19.7 - 12/10/2014 10:16:56 AM
 Table 2. (continued)
Risk Factors in Pericardiocentesis
Variable Overall Period p value
1 2 3
1997–2000 2001–2005 2006–2010

Systemic diseasea
Autoimmune disease 194 (2.4) 41 (2.5) 72 (2.5) 81 (2.3) 0.863
Hypothyroidism 124 (1.5) 23 (1.4) 52 (1.8) 49 (1.4) 0.358
Amyloidosis 5 (0.1) 0 (0.0) 2 (0.1) 3 (0.1) 0.507
Chronic kidney diseasea 614 (7.6) 80 (4.8) 207 (7.1) 327 (9.2) <0.001
Infectious diseasea
Tuberculosis 731 (9.0) 150 (9.0) 302 (10.4) 279 (7.9) 0.002
HIV carrier 10 (0.1) 0 (0.0) 1 (0.03) 9 (0.3) 0.012
Coexisting cardiac surgeryb 555 (6.9) 143 (8.6) 183 (6.3) 229 (6.5) 0.006
Coexisting catheter procedure
TPM 323 (4.0) 108 (6.5) 89 (3.1) 126 (3.6) <0.001
PPM 90 (1.1) 18 (1.1) 32 (1.1) 40 (1.1) 0.987
RFA for arrhythmias 51 (0.6) 0 (0.0) 21 (0.7) 30 (0.8) 0.001
Atrial transseptal puncture 8 (0.1) 2 (0.1) 1 (0.03) 5 (0.1) 0.378
PTCA 698 (8.6) 99 (6.0) 224 (7.7) 375 (10.6) <0.001
Complications
Pneumothoraxc 70 (0.9) 15 (0.9) 19 (0.7) 36 (1.0) 0.289
Cardiac perforation with surgical repaird 58 (0.7) 12 (0.7) 21 (0.7) 25 (0.7) 0.996
DOI: 10.1159/000368796
Cardiology 2015;130:37–45

Surgical drainage requirede 1,030 (12.7) 209 (12.6) 359 (12.4) 462 (13.1) 0.698
Recurrence 475 (5.9) 105 (6.3) 174 (6.0) 196 (5.5) 0.504
In-hospital mortality 1,266 (15.6) 222 (13.3) 330 (11.4) 714 (20.2) <0.001
Overall mortality 5,189 (64.1) 1,221 (73.4) 2,017 (69.5) 1,951 (55.2) <0.001
b Inclusive of valve repair, coronary artery bypass surgery, and aortic or aneurysm
Values are presented as numbers (%) unless otherwise stated. p < 0.05. PTCA =
Percutaneous transluminal coronary angioplasty; PPM = permanent pacemaker; repair.
c ICD-9, pneumothorax (512 and 512.2).
RFA = radiofrequency ablation; TPM = temporary pacemaker; GI = gastrointestinal.
a d Any cardiac perforation repaired via surgery, if needed, while conducting the
ICD-9, malignancy (140–208); cardiovascular disease (acute myocardial infarc-
tion, 410; aortic dissection, 441; heart failure, 448, and ischemic heart disease, 410, pericardiocentesis.
e Including pericardiostomy, pericardiectomy, and throcoscopic pericardial win-
411, 412, and 413); pericardial disease (acute pericarditis, 420; rheumatic pericarditis,
391.0 and 393, and adhesive and constrictive pericardium, 423.0 and 423.2); system- dow.
f The mean age in period 1 was significantly lower than those in periods 2 and 3
ic disease (autoimmune disease, 136.1, 443.1, 446, 555, 556, 694.4, 710, and 714; hy-
pothyroidism, 243 and 244, and amyloidosis, 277.3); chronic kidney disease with or using the Games-Howell post hoc test.
without dialysis (403, 404, and 585); infectious disease (tuberculosis, 011–018); HIV
(042 and 079.53).
41

Downloaded by:
Chang Gung Memorial Hospital - LinKou Branch
60.248.19.7 - 12/10/2014 10:16:56 AM
 Table 3. Associated risk factors for recurrence of effusion, complications, in-hospital death, and overall mortality (n = 8,656)
42
Variable Cases, n Recurrence Complications In-hospital death Overall death
event, % HR (95% CI) p value event, % OR (95% CI) p value event, % OR (95% CI) p value event, % HR (95% CI) p value

Gender
Male 4,984 (57.6) 5.3 0.79 (0.66–0.93) 0.005 1.8 1.38 (0.95–1.99) 0.090 16.1 1.20 (1.06–1.35) 0.004 65.2 1.13 (1.07–1.19) <0.001
Female 3,672 (42.4) 7.9 1 1.2 1 14.4 1 62.7 1
Age (per 10 years) NA NA 1.00 (0.95–1.05) 0.961 NA 1.03 (0.93–1.14) 0.521 NA 1.01 (0.97–1.04) 0.637 NA 1.14 (1.12–1.16) <0.001
DOI: 10.1159/000368796
Cardiology 2015;130:37–45

Year of procedure
1997–2000 1,778 (20.5) 7.0 1 1.5 1 13.2 1 73.5 1
2001–2005 3,093 (35.7) 6.6 1.06 (0.85–1.32) 0.622 1.3 0.88 (0.54–1.45) 0.619 11.3 0.84 (0.70–1.01) 0.059 69.5 1.04 (0.97–1.12) 0.281
2006–2010 3,785 (43.7) 6.0 0.96 (0.77–1.20) 0.723 1.7 1.12 (0.71–1.78) 0.631 19.8 1.57 (1.33–1.85) <0.001 55.3 0.95 (0.88–1.02) 0.144
Malignancy
No 4,987 (57.6) 4.5 1 1.9 1 13.2 1 51.9 1
Yes 3,669 (42.4) 9.0 2.07 (1.71–2.51) <0.001 1.0 0.52 (0.33–0.81) 0.003 18.4 1.58 (1.37–1.82) <0.001 80.7 2.88 (2.70–3.08) <0.001
Aortic dissection
No 8,482 (98.0) 6.5 1 1.5 1 15.1 1 64.1 1
Yes 174 (2.0) 0.6 0.21 (0.03–1.50) 0.120 0.6 0.38 (0.05–2.81) 0.344 29.3 2.29 (1.61–3.26) <0.001 66.7 1.49 (1.23–1.81) <0.001
Pericardial disease
No 7,873 (91.0) 6.7 1 1.6 1 16.3 1 66.1 1
Yes 783 (9.0) 3.3 0.52 (0.33–0.82) 0.005 1.0 0.54 (0.23–1.28) 0.161 6.5 0.65 (0.47–0.90) 0.009 44.3 0.68 (0.60–0.77) <0.001
Systemic disease
No 8,301 (95.9) 6.4 1 1.6 1 15.6 1 65.1 1
Yes 355 (4.1) 7.3 0.91 (0.62–1.37) 0.638 0.6 0.38 (0.09–1.58) 0.183 9.6 0.74 (0.51–1.07) 0.112 42.3 0.69 (0.59–0.82) <0.001
Chronic kidney disease
No 8,020 (92.7) 6.7 1 1.6 1 15.5 1 64.3 1
Yes 636 (7.3) 3.1 0.54 (0.34–0.85) 0.007 1.1 0.60 (0.28–1.31) 0.202 14.3 1.03 (0.81–1.31) 0.810 62.3 1.27 (1.14–1.41) <0.001
Infectious disease
No 7,875 (91.0) 6.6 1 1.5 1 16.3 1 65.7 1
Yes 781 (9.0) 4.6 0.79 (0.53–1.17) 0.235 1.3 0.97 (0.44–2.13) 0.943 5.9 0.50 (0.35–0.70) <0.001 48.1 0.72 (0.63–0.81) <0.001
Coexisting surgery
No 8,095 (93.5) 6.8 1 1.5 1 15.0 1 64.6 1
Yes 561 (6.5) 0.7 0.17 (0.06–0.45) <0.001 1.2 0.45 (0.20–1.01) 0.051 20.9 1.46 (1.15–1.86) 0.002 56.9 1.05 (0.93–1.19) 0.442
Coexisting catheter procedure
Hsieh/Hung/Wang/Wu/Chen
Ho/Wang/Lin/Mao/Tsai/Wen/Wang/

No 7,669 (88.6) 7.1 1 1.2 1 15.0 1 65.7 1

Yes 987 (11.4) 0.9 0.17 (0.08–0.32) <0.001 3.7 2.38 (1.52–3.72) <0.001 18.2 1.27 (1.05–1.55) 0.016 51.7 0.84 (0.76–0.92) <0.001
Surgical drainage requirement
No 7,519 (86.9) 6.8 1 1.4 1 15.9 1 63.2 1
Yes 1,137 (13.1) 3.8 0.42 (0.31–0.58) <0.001 2.1 1.85 (1.16–2.94) 0.009 12.4 0.72 (0.59–0.87) 0.001 70.0 0.95 (0.88–1.03) 0.202
Center volume
Low (≤12) 2,987 (34.5) 6.1 1 1.3 1 14.5 1 65.2 1
Medium (13–31) 2,709 (31.3) 6.3 1.23 (0.95–1.59) 0.112 1.7 1.07 (0.62–1.84) 0.807 13.9 0.95 (0.79–1.14) 0.569 63.7 0.95 (0.87–1.03) 0.174
High (≥32) 2,960 (34.2) 6.9 1.35 (1.00–1.82) 0.051 1.6 0.93 (0.51–1.71) 0.809 17.7 1.29 (1.04–1.60) 0.019 63.4 0.88 (0.80–0.97) 0.007
Hospital level
Medical center 5,367 (62.0) 6.3 1 1.7 1 15.4 1 63.9 1
Metropolitan 2,771 (32.0) 6.1 1.29 (0.98–1.69) 0.067 1.3 0.68 (0.39–1.19) 0.174 14.8 1.14 (0.94–1.38) 0.186 64.2 0.98 (0.90–1.06) 0.570
Local community 518 (6.0) 9.3 2.14 (1.50–3.06) <0.001 0.8 0.43 (0.15–1.26) 0.125 18.1 1.42 (1.08–1.86) 0.013 65.8 1.21 (1.06–1.37) 0.004

NA = Not available. p < 0.05 indicates a significant difference.

Downloaded by:
Chang Gung Memorial Hospital - LinKou Branch
60.248.19.7 - 12/10/2014 10:16:56 AM
 Table 4. Subgroup analysis for recurrence of effusion by malignancy neoplasm (n = 8,101)

Malignancy type Cancers, n Recurrences, n Recurrences, % HR (95% CI) p value

All 3,317 280 8.4 3.06 (2.53–3.69) <0.001

Head and neck 82 7 8.5 2.79 (1.32–5.90) 0.007
Gastrointestinal tract 284 17 6.0 1.37 (0.84–2.22) 0.208
Liver and biliary tract 84 0 0.0 NA NS
Lung and pleural 1,874 186 9.9 3.01 (2.49–3.63) <0.001
Heart and mediastinum 28 1 3.6 0.52 (0.07–3.70) 0.514
Hematologic 211 8 3.8 0.70 (0.35–1.41) 0.317
Breast cancer 261 25 9.6 1.71 (1.14–2.55) 0.009
Other cancer 493 36 7.3 1.53 (1.09–2.15) 0.014

p < 0.05 indicates a significant difference. NA = Not available; NS = not significant.

breast cancer accounted for 9.6% of recurrences (HR 1.71, iopathic pericardial effusion [19], 21–62% for malignan-
p = 0.009), and head and neck cancer accounted for 8.5% cy-related pericardial effusion [20], and 4% for cardiac
of recurrences (HR 2.79, p = 0.007). surgery-related pericardial effusions [21, 22] in previous
reports. This difference compared to other studies is due
to differences in the definition of recurrence: our study
Discussion defined recurrence as another pericardial effusion requir-
ing pericardiocentesis that occurred after the index hos-
Unlike previous studies [16], the current study includ- pitalization and within the follow-up period. The propor-
ed the entire population of Taiwan (the NHI coverage is tion of metastases to the heart and adjacent cardiac struc-
up to 99.91% of the population of Taiwan, a country tures in this study was consistent with that reported in
where not-for-profit health insurance is implemented). previous studies (1–10%) [16, 19].
This is the first and largest pericardiocentesis study ever A subgroup analysis stratified by cancer type showed
conducted in this country. higher recurrence rates in head and neck, breast, and lung
In our study, the proportion of those aged greater than and pleural malignancy. In liver and biliary tract cancer,
80 increased from 8.4% in 1997 to 15% in 2010, although no recurrences were recorded. The difference in recur-
the overall population growth in Taiwan during this pe- rence rates between cancers can be attributed to differ-
riod was only 11%. A substantial increase in the number ences in their characteristics. Lung and breast cancer are
of patients, admissions, and pericardiocentesis proce- primary proliferations of tumor cells in the thoracic area,
dures occurred during the study period (table 1) and the and head and neck cancer easily metastasizes to the pul-
proportion of pericardiocenteses performed on patients monary and esophageal area and can result in pericardial
who had catheter-based cardiac procedures, especially invasion. The ability of some cancer types to invade the
percutaneous transluminal coronary angioplasty, also in- chest cavity also explains why malignancy as a disease cat-
creased rapidly – from 6% in the first period of the study egory is associated with an increased risk for recurrence
to 10% in the third period. (table 1) [23, 24]. Surgical drainage seemed to lower the
As reported in previous studies [17, 18], malignancy risk of recurrence regardless of the malignancy history
was the most common underlying condition, and lung (table 4). A previous study reported that [5, 20] recur-
and pleural cancer was the most common type of malig- rence decreased from 6.8% in a simple pericardiocentesis
nancy. Among the underlying conditions, malignancy group to 3.8% in a surgical drainage group. Even if a pa-
has also previously been reported to be the most likely risk tient had a history of malignancy, as long as his or her
factor for recurrence in Taiwan [16], and it was associ- physical condition permitted surgical drainage, this pro-
ated with a doubling of the risk in our study. In the cur- cedure could provide relief and a lower risk of recurrence
rent study, recurrence rates were 9% for malignant peri- [25].
cardial effusion and 4.5% for nonmalignant pericardial In contrast to malignancy, underlying conditions that
Chang Gung Memorial Hospital - LinKou Branch

effusion compared to a recurrence rate of 6–58% for id- can be cured or effectively treated were less likely to be
60.248.19.7 - 12/10/2014 10:16:56 AM

Risk Factors in Pericardiocentesis Cardiology 2015;130:37–45 43

DOI: 10.1159/000368796
Downloaded by:
seen in patients having a recurrence. Chronic kidney dis- aortic dissection patients survived pericardiocentesis and
ease (including renal replacement), pericarditis, constric- the subsequent hospital experience. Patients with curable
tive pericardial disease and cardiac surgery, and catheter- diseases – pericardial disease and infectious disease – had
based cardiac procedures did not require repeated peri- better survival. Though catheter-based cardiac proce-
cardiocentesis. dures were associated with a greater number of in-hospi-
The overall complication rate in our study (1.42%) tal deaths and a poorer overall survival, if patients could
was similar to that in previous studies (1.2%) [8, 20] and survive the hospital admission their survival time would
remained quite stable, i.e. from 1.17% in 1997 to 1.56% be prolonged.
in 2010. Catheter-based cardiac surgery and malignan- There are several known limitations when using the
cy had opposite effects on the complication rate (ta- NHIRD. Firstly, it is hard to determine a causal relation-
ble 3), which was increased in catheter-based cardiac ship with certainty due to the lack of personal chart infor-
surgery and decreased in malignancy. The difference mation such as basic clinical features (body weight and
could be explained by differences between the two un- height) or details of pericardial effusion (cytology or cul-
derlying conditions in terms of the volume of pericar- ture reports) [15]. To avoid this limitation, we removed
dial effusion and whether or not the tamponade was any uncertain confounding factors. For example, trau-
acute. The larger the volume of the pericardial effusion, matic patients with pericardiocentesis showed an ex-
the larger the pericardial space available for needle in- tremely high HR for complications like pneumothorax
sertion is and the less likely it is that insertion of the and cardiac rupture – which needed surgical repair in this
pericardiocentesis needle will be inaccurate and cause study. As we could not distinguish the complications
damage to the surrounding tissue. Accumulation of ma- from the underlying disease, we excluded these patients
lignancy-related pericardial effusion is a chronic pro- from the analysis. Secondly, the NHIRD adopted the
cess that produces a massive pericardial effusion amount ICD-9 for disease coding so that no further details of the
regardless of whether or not tamponade is present at disease severity or the stage of malignancy among pa-
diagnosis. Pericardial effusion induced by cardiac cath- tients were collected (the inevitable limitations of the ex-
eter-based procedures is usually acute, and a small isting ICD). Nevertheless, the information on insurance-
amount effusion fluid produces a tamponade that re- paid procedures (pericardiocentesis, cardiac catheter-
quires intervention. This physiologic difference be- based procedures, or cardiac surgery) was accurate and
tween malignancy-based and cardiac catheter-based complete because it was strictly stipulated by the NHI.
pericardial effusions had opposite effects on complica- Moreover, in the clinical setting, pericardiocentesis is
tions of pericardiocentesis and explains why the overall considered to be a minor complication of cardiac cathe-
complication rate remained stable during the study pe- ter-based procedures (the reimbursement value was rela-
riod. As the number of malignancy and cardiac cathe- tive low compared to other medical treatments, such as a
ter-based procedures will continue to increase, we as- coronary stent or pacemaker) and physicians may not re-
sume that in the future the complication number will quest reimbursement for pericardiocentesis even though
keep climbing but the rate will remain stable. it would be covered by the NHI. Thus, the collected peri-
The survival rate of patients receiving pericardiocen- cardiocentesis number may be underestimated.
tesis was poor. The mean follow-up period and overall In conclusion, malignant neoplasm and catheter-re-
mortality was 2.08 years and 68.5%, respectively. The me- lated cardiac procedures were the most commonly seen
dian survival time and overall mortality of patients with underlying conditions in patients receiving pericardio-
malignant disease was 0.73 years and 80.7%, respectively. centesis. Malignancy was the major cause of recurrence
Aortic dissection and malignancy were the main risk fac- and in-hospital mortality, but it was associated with a
tors for in-hospital death. Aortic dissection is considered lower rate of acute complications after pericardiocentesis.
to be a major contraindication for pericardiocentesis [4] On the other hand, complications and in-hospital mor-
because rapid withdrawal of blood from the pericardium tality were higher among patients with cardiac catheter
might reduce the pressure on the dissection from the procedures and surgery.
leaked blood and increase the blood flow from the dissec-
tion into the pericardium [26]. However, sometimes it is
an urgent matter to reduce the cardiac tamponade and Conflict of Interest
stabilize the patient enough so that the desired surgical
Chang Gung Memorial Hospital - LinKou Branch

treatment can be performed. Seventy-one percent of our The authors have no conflicts of interests.
60.248.19.7 - 12/10/2014 10:16:56 AM

44 Cardiology 2015;130:37–45 Ho/Wang/Lin/Mao/Tsai/Wen/Wang/

DOI: 10.1159/000368796 Hsieh/Hung/Wang/Wu/Chen
Downloaded by:
 References
1 Callahan JA, Seward JB, Tajik AJ: Cardiac 10 Tsang TS, Barnes ME, Hayes SN, Freeman 17 Lam KY, Dickens P, Chan AC: Tumors of the
tamponade: pericardiocentesis directed by WK, Dearani JA, Butler SL, Seward JB: Clini- heart: a 20-year experience with a review of
two-dimensional echocardiography. Mayo cal and echocardiographic characteristics of 12,485 consecutive autopsies. Arch Pathol
Clin Proc 1985;60:344–347. significant pericardial effusions following Lab Med 1993;117:1027–1031.
2 Tsang TS, Freeman WK, Sinak LJ, Seward JB: cardiothoracic surgery and outcomes of echo- 18 Klatt EC, Heitz DR: Cardiac metastases. Can-
Echocardiographically guided pericardiocen- guided pericardiocentesis for management: cer 1990;65:1456–1459.
tesis: evolution and state-of-the-art tech- Mayo Clinic experience, 1979–1998. Chest 19 Sagrista-Sauleda J, Angel J, Permanyer-Miral-
nique. Mayo Clin Proc 1998;73:647–652. 1999;116:322–331. da G, Soler-Soler J: Long-term follow-up of
3 Spodick DH: Acute cardiac tamponade. N 11 Tsang TS, Barnes ME, Gersh BJ, Bailey KR, idiopathic chronic pericardial effusion. N
Engl J Med 2003;349:684–690. Seward JB: Outcomes of clinically significant Engl J Med 1999;341:2054–2059.
4 Maisch B, Seferović PM, Ristić AD, Erbel R, idiopathic pericardial effusion requiring in- 20 Tsang TS, Seward JB, Barnes ME, Bailey KR,
Rienmüller R, Adler Y, Tomkowski WZ, tervention. Am J Cardiol 2003;91:704–707. Sinak LJ, Urban LH, Hayes SN: Outcomes of
Thiene G, Yacoub MH; Task Force on the Di- 12 Yang YW, Chen YH, Xirasagar S, Lin HC: In- primary and secondary treatment of pericar-
agnosis and Management of Pericardial Dis- creased risk of stroke in patients with bullous dial effusion in patients with malignancy.
eases of the European Society of Cardiology: pemphigoid: a population-based follow-up Mayo Clin Proc 2000;75:248–253.
Guidelines on the diagnosis and management study. Stroke 2011;42:319–323. 21 Ashikhmina EA, Schaff HV, Sinak LJ, Li Z,
of pericardial diseases executive summary – 13 Wu CY, Wu MS, Kuo KN, Wang CB, Chen Dearani JA, Suri RM, Park SJ, Orszulak TA,
the Task Force on the Diagnosis and Manage- YJ, Lin JT: Effective reduction of gastric can- Sundt TM 3rd: Pericardial effusion after car-
ment of Pericardial Diseases of the European cer risk with regular use of nonsteroidal anti- diac surgery: risk factors, patient profiles, and
Society of Cardiology. Eur Heart J 2004; 25: inflammatory drugs in Helicobacter pylori-in- contemporary management. Ann Thorac
587–610. fected patients. J Clin Oncol 2010; 28: 2951– Surg 2010;89:112–118.
5 Eichler K, Zangos S, Thalhammer A, Jacobi V, 2957. 22 Ben-Horin S, Bank I, Guetta V, Livneh A:
Walcher F, Marzi I, Moritz A, Vogl TJ, Mack 14 Cheng CL, Kao YH, Lin SJ, Lee CH, Lai ML: Large symptomatic pericardial effusion as the
MG: CT-guided pericardiocenteses: clinical Validation of the National Health Insurance presentation of unrecognized cancer: a study
profile, practice patterns and clinical out- Research Database with ischemic stroke cases in 173 consecutive patients undergoing peri-
come. Eur J Radiol 2010;75:28–31. in Taiwan. Pharmacoepidemiol Drug Saf cardiocentesis. Medicine (Baltimore) 2006;
6 Burazor I, Imazio M, Markel G, Adler Y: Ma- 2011;20:236–242. 85:49–53.
lignant pericardial effusion. Cardiology 2013; 15 Wu CY, Chen YJ, Ho HJ, Hsu YC, Kuo KN, 23 Day GL, Blot WJ: Second primary tumors in
124:224–232. Wu MS, Lin JT: Association between nucleo- patients with oral cancer. Cancer 1992;70:14–
7 Lazaros G, Stefanadis C: Malignant pericar- side analogues and risk of hepatitis B virus- 19.
dial effusion: still a long road to Ithaca. Cardi- related hepatocellular carcinoma recurrence 24 Licciardello JT, Spitz MR, Hong WK: Multi-
ology 2013;125:15–17. following liver resection. JAMA 2012; 308: ple primary cancer in patients with cancer of
8 Tsang TS, Freeman WK, Barnes MF, Reeder 1906–1914. the head and neck: second cancer of the head
GS, Packer DL, Seward JB: Rescue echocar- 16 Cho BC, Kang SM, Kim DH, Ko YG, Choi D, and neck, esophagus, and lung. Int J Radiat
diographically guided pericardiocentesis for Ha JW, Rim SJ, Jang Y, Chung N, Shim WH, Oncol Biol Phys 1989;17:467–476.
cardiac perforation complicating catheter- Cho SY, Kim SS: Clinical and echocardio- 25 Laham RJ, Cohen DJ, Kuntz RE, Baim DS, Lo-
based procedures: the Mayo Clinic experi- graphic characteristics of pericardial effusion rell BH, Simons M: Pericardial effusion in pa-
ence. J Am Coll Cardiol 1988;32:1345–1350. in patients who underwent echocardiograph- tients with cancer: outcome with contempo-
9 Tsang TS, Enriquez-Sarono M, Freeman WK, ically guided pericardiocentesis: Yonsei Car- rary management strategies. Heart 1996; 75:
Barnes ME, Sinak LJ, Gersh BJ, Bailey KR, diovascular Center experience, 1993–2003. 67–71.
Seward JB: Consecutive 1,127 therapeutic Yonsei Med J 2004;45:462–468. 26 Isselbacher EM, Cigarroa JE, Eagle KA: Car-
echocardiographically guided pericardiocen- diac tamponade complicating proximal aortic
teses: clinical profile, practice patterns, and dissection: is pericardiocentesis harmful? Cir-
outcomes spanning 21 years. Mayo Clin Proc culation 1994;90:2375–2378.
2002;77:429–436.

Chang Gung Memorial Hospital - LinKou Branch

60.248.19.7 - 12/10/2014 10:16:56 AM

Risk Factors in Pericardiocentesis Cardiology 2015;130:37–45 45

DOI: 10.1159/000368796
Downloaded by: