GENERAL THORACIC
Comparative Study of Subxiphoid Versus
Video-Thoracoscopic Pericardial Window
Patrick K. H. OBrien, MD, John C. Kucharczuk, MD, M. Blair Marshall, MD,
Joseph S. Friedberg, MD, Zhen Chen, PhD, Larry R. Kaiser, MD, and
Joseph B. Shrager, MD
Section of General Thoracic Surgery and Department of Biostatistics and Epidemiology, University of Pennsylvania School of
Medicine, and Department of Surgery, Philadelphia Veterans Affairs Medical Center, Philadelphia, Pennsylvania
Background. It remains undefined whether surgical
subxiphoid drainage or thoracoscopic pericardial win-
dow is the optimal operative approach to pericardial
effusion. We hypothesized that the true window into the
pleural space created by the latter might improve the
duration of freedom from recurrent effusion.
Methods. We conducted a retrospective chart review of
indications, preoperative and intraoperative variables,
morbidity, recurrence, and survival.
Results. Fifty-six patients underwent the subxiphoid
procedure and 15 underwent the thoracoscopic proce-
dure. Echocardiographic evidence of tamponade was
present before 8 of 10 thoracoscopic procedures (80%)
and 43 of 56 subxiphoid procedures (81%) for which
descriptions of hemodynamics were available. In addi-
tion, non-pericardial procedures were performed in 10
(67%) and 18 (32%) patients, respectively (p 0.020).
Anesthesia time was longer at thoracoscopy (117.1 32.4
vs 81.1 25.5 minutes; p < 0.001). Procedural morbidity
P ericardial effusion complicates many disease pro-
cesses, both benign and malignant. A variety of
treatments are available, ranging from observation, to
anti-inflammatory or anti-neoplastic chemotherapy, to
pericardiocentesis with or without percutaneous catheter
drainage, and finally to surgical procedures. We believe
that there are two surgical options most commonly used
today: (1) subxiphoid drainage (SXD) and (2) video-
thoracoscopic pericardial window (TPW). We are un-
aware of any available data to guide the choice between
these approaches.
The most common operation is likely SXD, which as
generally performed amounts to placing a large-bore
tube from an upper abdominal incision through the site
of a pericardial biopsy [1]. Symphysis between the epi-
cardium and pericardium develops after several days of
drainage [2]. Because it is done in the supine position,
access is readily available to perform pericardiocentesis if
instability occurs after induction.
Accepted for publication May 18, 2005.
Address correspondence to Dr Shrager, Hospital of the University of
Pennsylvania, Silverstein 6, 3400 Spruce St, Philadelphia, PA 19104;
e-mail: joseph.shrager@uphs.upenn.edu.
2005 by The Society of Thoracic Surgeons
Published by Elsevier Inc
was higher after thoracoscopy (4 [27%] vs 1 [2%]; p
0.006), but was generally minor. Hospital mortality
tended to be higher after the subxiphoid procedure (7
[13%] vs 0 [0%]; p 0.332), but none of the deaths was
procedure-related. Follow-up was complete for 65 pa-
tients (92%). Recurrence occurred in 1 thoracoscopy pa-
tient (8%) and 5 subxiphoid patients (10%) (p 1.000).
Mean time to recurrence by Kaplan-Meier analysis
trends were longer after thoracoscopy (36.1 vs 11.4
months; p 0.16), and multivariate analysis identified
the thoracoscopic approach as an independent predictor
of freedom from recurrence (relative risk, 0.41; p 0.014).
Conclusions. Operative time and minor procedural
morbidity are higher with thoracoscopic pericardial win-
dow, but long-term control of effusion seemed to be
better than after subxiphoid surgical drainage.
(Ann Thorac Surg 2005;80:20139)
2005 by The Society of Thoracic Surgeons
A pericardial window is a larger biopsy or partial
pericardiectomy that creates a passage presumably al-
lowing longer-term drainage into an adjacent space,
usually the pleural space. Pericardial resection creating
such a window can be performed through a limited
anterior thoracotomy, formal thoracotomy [3], or since
the 1990s, thoracoscopy [4 10]. The subxiphoid drainage
procedure is often erroneously referred to as a win-
dow, when in fact no such connection to an adjoining
space is made during standard SXD.
Thoracoscopic pericardial window does create a true
window, but it requires one-lung ventilation and two or
three intercostal incisions. The thoracoscopic access
allows concurrent performance of additional proce-
dures, such as biopsy of the lung, pleural, or medias-
tinal masses, or management of a concomitant pleural
effusion [4 10]. However, need for single-lung ventila-
tion and the preference by some surgeons for lateral
positioning, which impedes easy access for an emergent
drainage, may limit the role of TPW in patients with
hemodynamically significant effusions [5, 6, 10]. Although
TPW is minimally invasive when compared with thoracot-
omy, it may in fact be more invasive than SXD.
The choice among the surgical approaches to the
0003-4975/05/$30.00
doi:10.1016/j.atthoracsur.2005.05.059
 2014 OBRIEN ET AL
GENERAL THORACIC
PERICARDIAL WINDOW COMPARISON
management of pericardial effusion must be based
mainly on (1) the effectiveness of the procedure in pre-
venting recurrent effusion, and (2) the morbidity and
mortality of the procedure. Other considerations include
the relative simplicity of the technique and cost. Pub-
lished recurrence rates have ranged from 2.5% to 16%
after SXD [1, 2, 1121] and 0% to 8% after TPW [4 6, 8 10,
22, 23]. However, duration and intensity of follow-up, as
well as definitions of recurrence have been variable. The
only study directly comparing a subxiphoid with a tho-
racoscopic approach [24] does not clearly describe
whether a true window was created by the former pro-
cedure, and does not focus on surgical morbidity or other
issues of surgical importance.
We reviewed our experience with SXD and TPW at a
single institution, comparing preoperative and intraop-
erative variables, morbidity, and recurrence. Our goal
was to provide data that might guide the choice between
the two approaches. A central hypothesis was that the
creation of a true window by TPW would confer an
improvement in freedom from recurrent pericardial
effusion.
Patients and Methods
Data was collected from the inpatient and outpatient
records of patients who underwent SXD or TPW proce-
dures between 1992 and 2002. These procedures were
performed by 5 surgeons at three different hospitals
within the University of Pennsylvania Health System.
The study was granted an exemption from the require-
ment for informed consent from the Institutional Review
Board of the University of Pennsylvania.
A total of 76 patients underwent SXD or TPW proce-
dures. Five patients were excluded because the medical
record of the index admission was not available. The
remaining 71 patients are the subjects of this review. A
thoracoscopic pericardial window procedure was per-
formed in 15 patients (21%), and the SXD procedure was
performed in 56 patients (79%).
Echocardiographic data prior to surgical drainage were
reviewed for effusion size and the presence of tamponade
physiology as indicated by right atrial compression or
right ventricular diastolic collapse, or both [25, 26]. In
cases in which echocardiography was unavailable, size
was graded according to computed tomography. In cases
in which percutaneous drainage was performed prior to
surgical drainage, echocardiographic data after the per-
cutaneous procedure was used if it existed; otherwise, the
data was omitted.
We recorded any adverse event after the procedure
that occurred during the same hospitalization, but was
not preoperatively present as morbidity. We recorded
any adverse event that clearly would not have taken
place had the patient not undergone the procedure as a
procedural morbidity; in this category we searched the
records specifically for wound infections or wound heal-
ing problems, intraoperative technical complications, in-
traoperative or postoperative bleeding, and drainage
tube-related complications (eg, pneumothorax).
Ann Thorac Surg
2005;80:20139
Follow-Up
Length of stay was calculated from the day of the surgical
procedure until discharge or in-hospital death. Recur-
rence was defined as an effusion on postoperative echo-
cardiography that was either moderate or greater, or
hemodynamically significant, regardless of the need for
further therapy. In the case of patients who were not
followed-up at the University of Pennsylvania Health
System, the treating physician was contacted by tele-
phone. The follow-up period was calculated until recur-
rence, death, or last clinical contact. Additional survival
data was obtained from the Social Security Death Index.
Operative Techniques
The choice of operative procedure was made by the
surgeon. All patients were hemodynamically stable when
brought to the operating room, although many had
tachycardia or mild pulsus paradoxus, or both. In several
instances in both groups, preoperative percutaneous
drainage by the cardiology service had been performed
for a hemodynamically significant effusion to urgently
stabilize a patient. One patient had hemodynamic insta-
bility develop with the induction of general anesthesia.
This patient improved after urgent pericardiocentesis
and an SXD procedure was immediately performed.
General anesthesia was used in all cases. Anesthesia time
was calculated as time from induction until extubation, or
until skin closure if the patient was not extubated or if
additional procedures were to be performed at a separate
site.
Our technique of SXD is made through a 5-cm upper
midline incision. Blunt extraperitoneal dissection is car-
ried out (usually with resection of the xiphoid) to expose
the pericardium. After excision of an approximately 3
3 cm piece of pericardium, and after drainage of the
effusion, a single thoracostomy tube (20 to 32 FF) is placed
and left to drain for at least 96 hours. A closed-system
silastic drain was placed in 3 patients. During SXD, no
attempt was made to create a true window to either the
pleural or peritoneal space.
For TPW, single-lung ventilation using a double-lumen
endotracheal tube was used in all of our cases. Thoracos-
copy was performed through a 10-mm camera port,
typically in the seventh intercostal space in the mid-
axillary line, and the procedure was completed through
one or two working incisions. Nine of the TPW proce-
dures were done from the right side and six from the left.
A piece of pericardium approximately 4 cm in diameter
was resected from anterior to the phrenic nerve, creating
a window into the pleural space. A single chest tube (20
to 32 FF) was placed into the operative pleural space in all
instances; a second chest tube was placed directly into
the pericardium in 2 patients. A thoracoscopic window
had been planned for 2 patients who ultimately under-
went subxiphoid drainage. This conversion was due to
preoperative deterioration in 1 patient and inability to
place a double-lumen endotracheal tube in the other
patient. All other planned TPWs were successfully ac-
complished by that technique.
Ann Thorac Surg OBRIEN ET AL 2015

GENERAL THORACIC
2005;80:20139 PERICARDIAL WINDOW COMPARISON

Table 1. Demographic Characteristics of Patients Table 3. Comparison of Perioperative and Postoperative

Undergoing Surgical Pericardial Drainagea Resultsa
All Patients Thoracoscopic
Characteristics (n 71) Pericardial Subxiphoid
Window Drainage
Age (y) 56.3 15.4 (n 15) (n 56) p Value
Female 33 (46)
Anesthesia time (min) 117.1 32.4 81.1 25.5 0.001b
Cause of effusion
Extubated at the 15 (100%) 49 (91%)c 1.000
History of malignancy 50 (70) conclusion of the
Lung 23 (32) operation
Breast 9 (13) Additional procedure at 10 (67%) 18 (32%) 0.020b
Lymphoma/Lymphoproliferative 8 (11) pericardial drainage
Other 10 (14) Duration of chest 3.3 1.4 4.0 1.6 0.117
Documented malignant effusion 16 (22) tube(s)
Renal failure 4 (6) Pericardium sclerosed 0 (0) 5 (9) 0.577
Collagen-vascular disease 3 (4) Total amount drained 735 742 433 417 0.026b
(mL)
Post cardiotomy 2 (3)
Length of stay 12.4 22.8 10.4 12.2 0.640
Idiopathic, other 12 (17)
Morbidity 7 (47) 29 (52) 0.777
a
Data are presented as mean standard deviation or number (percent). Procedural morbidity 4 (27) 1 (2) 0.006b
Mortality 0 (0) 8 (14) 0.189

Pericardial and pleural tube management, including a

Data are presented as mean standard deviation or number (per-
the use of sclerosants, was at the discretion of the cent). b
p value 0.05. c
Excludes patients with ongoing respira-
tory failure.
surgeon. The duration and amount of postoperative
drainage were calculated for the pericardial drain in
cases of SXD and for all drains (pleural and pericardial)
placed on the operative side in cases of TPW.
(70%). The most common primary malignancies were
Statistical Methods lung (23), breast (9), and lymphoma or lymphoprolifera-
Two-sample t test and two-tailed Fishers exact test were tive disorders (8). Prior pericardial drainage had been
used to compare continuous and dichotomous variables performed in 19 patients (27%), either by pericardiocen-
between groups, respectively. Differences were consid- tesis (17) or by pericardiocentesis with subsequent in-
ered statistically significant at p 0.05. Time to recur- dwelling catheter (2). Two patients had undergone more
rence and time to death were modeled using the Kaplan- than one percutaneous procedure. The effusion was
Meier method. Risk of recurrence and risk of death were proven malignant in 16 patients (22%) (ie, by preopera-
estimated using Cox proportional hazards models. tive cytology in 3 patients [4.2%], operative cytology in 9
[13%], and pericardial biopsy in 10 [14%]).
Results
Demographics Preoperative Echocardiography
Table 1 lists the demographic characteristics of all pa- Echocardiographic data were available for 66 patients, in-
tients. A history of malignancy was present in 50 patients cluding 11 TPW patients (73%) and all SXD patients. Three
of the 4 TPW patients who did not have an echocardiogram
Table 2. Comparison of Demographic and had a preoperative computed tomographic scan. Overall,
Echocardiographic Characteristicsa the effusions were recognized preoperatively as moderate
or greater in 9 TPW patients (77%) and 53 SXD patients
Thoracoscopic
Pericardial Subxiphoid (95%). Echocardiographic descriptions of hemodynam-
Window Drainage p ics were available in 62 patients, including 10 TPW
(n 15) (n 56) Value patients (91%) and 53 SXD patients (95%). Tamponade
Age (y) 54.5 18.4 57.2 14.6 0.366 physiology was present in 8 TPW patients (80%) and 43
Female 5 (33%) 28 (50%) 0.383 SXD patients (81%).
History of malignancy 9 (60%) 41 (73%) 0.351 Data shown in Table 2 demonstrates that there was no
Documented malignant 4 (26%) 12 (21%) 0.731 significant difference between the TPW and SXD groups in
effusion any of the demographic factors or echocardiographic
Prior pericardiocentesis 6 (40%) 13 (23%) 0.206 characteristics.
Effusion moderate or 9 (77) 53 (95) 0.063
greater Perioperative and Postoperative Results
Any tamponade 8 (80) 43 (81) 1.000
Perioperative and postoperative results are summarized
a
Data are presented as mean standard deviation or number (percent). in Table 3.
 2016 OBRIEN ET AL
GENERAL THORACIC
PERICARDIAL WINDOW COMPARISON
CONCOMITANT PROCEDURES. Twenty-eight patients (39%)
underwent additional procedures at the time of pericar-
dial drainage, excluding flexible bronchoscopy or instil-
lation of sclerosants. Among the patients undergoing
TPW, 10 (67%) had additional procedures, with 8 (53%)
having thoracoscopic procedures, including drainage of
an ipsilateral pleural effusion and pleurodesis (4), lung
biopsy (3), pleural biopsy (3), and mediastinal biopsy (1).
Other concomitant procedures in the TPW patients in-
cluded insertion of a chest tube for a contralateral effu-
sion (2) and transesophageal echocardiography (TEE) (1).
Among patients undergoing SXD, a significantly lower
fraction of patients (n 18; 32%) had additional proce-
dures. Additional procedures in the SXD patients in-
cluded insertion of a chest tube for a pleural effusion (8),
TEE (3), mediastinoscopy (2), laser and stenting of the
airway (2), and right middle lobectomy (1), exploratory
laparotomy (1), mediastinal lymph node biopsy (1),
esophagoscopy (1), and tracheostomy (1).
Instillation of a sclerosing agent (talc or doxycycline)
into the pericardial space was performed in 5 patients
(7%), all in the SXD group (intraoperatively in 4 and
postoperatively in 1). Instillation of sclerosants into the
pleural space was done in 8 patients (10%), including 6
TPW patients (intraoperatively in 4 [including 2 con-
tralateral to the TPW through a chest tube] and postop-
eratively in 2), and 2 SXD patients (both postoperatively
through an additional chest tube placed at the time of
surgery).
OPERATIVE TIME. Anesthesia time could be calculated in 67
patients (94%). Not unexpectedly, there was a statistically
significantly longer anesthesia time for TPW (117.1 32.4
minutes) versus SXD (81.1 25.5 minutes) (p 0.001).
SENSITIVITY OF PERICARIDAL BIOPSY. Pericardial biopsies were
done in all patients in the TPW group and in 45 in the
SXD group, including 9 patients with a history of malig-
nancy among the former and 34 such patients among the
latter. Biopsy was positive for malignancy in 4 of 9
patients (44%) and 6 of 34 (18%), respectively, a difference
that did not achieve statistical significance (p 0.177). No
patients without a history of malignancy had positive
pericardial cytology or biopsy.
POSTOPERATIVE DATA AND MORBIDITY. All of the TPW patients
were extubated in the operating room, excluding 2 pa-
tients who were intubated preoperatively for ongoing
respiratory failure; 49 of 54 SXD patients (91%) were
extubated, a difference that did not achieve statistical
significance.
Duration of tube drainage ranged from 1 to 10 days.
The duration was similar in the groups (p 0.117). The
total amount drained was significantly higher in the TPW
group (735 742 mL vs 433 417 mL), but recall that this
drainage included pleural tubes in the former. Length of
stay was similar in the two groups, 12.4 22.8 days (range,
2 to 94; median, 7 days) in the TPW group and 10.4 12.2
days (range, 1 to 76; median, 7 days) in the SXD group.
Morbidity occurred in 36 patients (ie, 7 TPW patients
Ann Thorac Surg
2005;80:20139
[47%] and 29 SXD patients [52%]), and it was primarily
related to coexisting medical problems in these complex
patients that were unrelated to the procedure. Specific
procedural morbidity was greater after TPW procedures
(p 0.006), although the morbidities were considered
minor. Procedural morbidity occurred in 4 patients (27%)
in the TPW group and in 1 in the SXD group (2%). Of the
4 TPW patients, 2 required additional chest tubes for
pneumothorax after chest tube removal, 1 was dis-
charged home with a Heimlich valve for ongoing air leak
from injury to a trapped lung, and 1 was readmitted for
drainage from a chest tube site that was self-limited. In
the SXD patient with procedural morbidity, a pleural
chest tube was placed intraoperatively for a clinical
tension pneumothorax after pericardial drainage. There
were no cases of postoperative pneumothorax or wound-
related problems in the SXD group.
There were no direct procedure-related mortalities
(deaths due to procedural morbidity or early failure to
control effusion). Seven patients died within 30 days or
prior to discharge, all in the SXD group (13%), a differ-
ence that did not achieve statistical significance. All 7
deaths were due to either advancing malignancy or
worsening of an underlying medical illness in the ab-
sence of recurrent effusion.
Recurrence of Effusion and Survival
Crude follow-up, effusion recurrence, and survival data
is shown in Table 4.
Complete follow-up data was available for 65 patients
(92%). Endpoints reached were 6 patients with recur-
rence of effusion, 18 alive without evidence of recurrence,
41 dead without evidence of recurrence. Mean duration
of follow-up was 22.1 31.1 months.
There was one recurrence (8%) in the TPW group and
five (10%) in the SXD group (p 1.000). None of the
recurrences required invasive therapy. The sole recur-
rence after TPW occurred in a 72-year-old man with a
recurrent idiopathic effusion that recurred again 18
months after TPW; he expired 1 month later of congestive
heart failure. Two of the recurrences in the SXD group
occurred early; each patient was found to have a moder-
ate residual effusion at the time of discharge that was not
treated further. The three late SXD recurrences were
noted at a mean of 3.2 months.
From the Social Security Death Index data, we find that
a total of 52 patients had died (ie, 9 in the TPW group
[60%] and 43 in the SXD group [77%]) at a mean of 28.1
35.9 months. Median overall survival was 70.5 months.
Crude, median survival was not statistically significantly
different between the groups (ie, 59.2 months after SXD
and 79.7 months after TPW p 0.68). In patients with a
history of malignancy and documented malignant effu-
sions, median survival was 79.7 months and 23.6 months,
respectively.
Kaplan-Meier Analysis
The Kaplan-Meier curves of freedom from recurrent
effusion after SXD and TPW are shown in Figure 1.
Median time to recurrence was 8.0 months after SXD and
Ann Thorac Surg OBRIEN ET AL 2017

GENERAL THORACIC
2005;80:20139 PERICARDIAL WINDOW COMPARISON

Table 4. Crude Follow-Up, Survival, and Recurrence Dataa Table 5. Results of Cox Proportional Hazards Model of the
Risk of Recurrent Effusion After Surgical Drainage
Thoracoscopic
Pericardial Subxiphoid Relative 95% Confidence
Window Drainage p Variable Risk Interval p Value
(n 15) (n 56) Value
Thoracoscopic pericardial 0.41 0.200.83 0.014a
Follow-up available 13 (87) 54 (96) 0.194 window
Mean follow-up (mo) 32.0 41.5 19.3 27.5 0.186 Prior drainage 1.36 0.692.70 0.380
Recurrence of effusion 1 (8) 5 (9) 0.673 Malignant effusion 2.23 1.064.72 0.035a
Median time to recurrence 21.7 8.0 0.160 History of lung cancer 3.59 1.637.89 0.001a
(mo) History of other cancer 3.79 1.738.32 0.001a
Died 9 (60) 43 (77) 0.206
a
p value 0.05.
a
Data are presented as mean standard deviation or number (percent).

degree of statistical significance (p 0.014; 95% confi-

21.7 months after TPW, and mean time to recurrence was
11.4 and 36.1 months, respectively, but the difference did
not achieve statistical significance (p 0.16).
Cox Models
Cox proportional hazards models were used to examine
relative risk of recurrence of effusion and death after each
procedure (Tables 5, 6). Covariates chosen for the models
included treatment group (SXD vs TPW), history of lung
cancer, history of other malignancy, malignant effusion,
and prior drainage.
The relative risk of recurrence after TPW was 0.41
compared with SXD, and this result achieved a high
dence interval, 0.20 to 0.83). Other independently signif-
icant variables in the recurrence model included history
of lung cancer, history of other malignancy, and malig-
nant effusion (Table 5).
None of the variables achieved significance in the
survival model (Table 6). Documented malignant effusion
nearly achieved significance, with a relative risk of 7.7
(p 0.051; 95% confidence interval, 0.99 to 60.77).
Comment
When an operation is required for the management of
pericardial effusion, there are two main options in recent
years that have been considered reasonable: (1) surgical
SXD and (2) video TPW. If it could be shown that there
was a clear difference in morbidity, mortality, diagnostic
accuracy, recurrence of effusion, or cost between these
approaches, then this information would be useful to
surgeons who are trying to decide between the two
options. Therefore we reviewed our experience with the
two procedures to compare them with respect to these
outcomes. The importance of these issues is highlighted
by the fact that there is an ongoing discussion at a
national level to organize a multicenter, prospective,
randomized study to address this question.
In this series, patients who underwent SXD and those
who underwent TPW had effusions that were similar in
terms of the percentage that were (1) moderate or greater
in size and (2) associated with echocardiographic abnor-
malities suggesting tamponade physiology. General an-
esthesia was well tolerated in both groups, with only 1
patient who had urgent needle decompression per-

Table 6. Results of Cox Proportional Hazards Model of Risk

of Death After Surgical Drainage of Pericardial Effusion
Relative 95% Confidence
Variable Risk Interval p Value

Prior drainage 0.29 0.061.43 0.128

Thoracoscopic pericardial 0.61 0.182.08 0.433
window
History of lung cancer 0.70 0.143.55 0.660
Fig 1. Kaplan-Meier curves of freedom from recurrent effusion. History of other cancer 1.35 0.355.23 0.660
(SXD subxiphoid drainage; VATS thoracoscopic pericardial Malignant effusion 7.75 0.9960.7 0.051
window.)
 2018 OBRIEN ET AL
GENERAL THORACIC
PERICARDIAL WINDOW COMPARISON
formed in the operating room for instability after induc-
tion. Also, single-lung ventilation was well tolerated in
the TPW group. In our practice and in many series [35],
patients who are clinically unstable from tamponade are
often temporized with percutaneous, echocardiographic-
guided drainage prior to any surgical procedure to avoid
instability associated with the induction of general anes-
thesia. In clinically stable patients, our data suggest that
effusion size and presence of mild tamponade on echo-
cardiography do not prohibit TPW.
One potential benefit of TPW in comparison with SXD
is that it allows certain intrathoracic procedures to be
performed simultaneously with the pericardial proce-
dure. In this review, the percentage of patients having
additional procedures was twice as high as in the TPW
group. Most of these were intrapleural procedures ac-
complished thoracoscopically. However, whether there is
any benefit to the patient having a thoracoscopic pleu-
rodesis for pleural effusion combined with a TPW rather
than a tube thoracostomy under the same anesthetic as
SXD, and subsequent pleurodesis, is questionable.
In this series, follow-up was available in 92%, and
recurrence was confirmed by echocardiography. Recur-
rence was uncommon, occurring in 8% and 10% of TPW
and SXD patients, respectively, comparable with previ-
ous series [124].
We had hypothesized that the creation of a true peri-
cardial window into the pleural space might produce a
lower recurrence rate in the TPW group. Although the
crude recurrence rate and time to recurrence were not
significantly different between the two groups, Cox pro-
portional hazards models determined that TPW con-
ferred a statistically significant reduction (relative risk,
0.41) in the risk of recurrent effusion. In addition, this
model confirmed the previous findings of others that
patients with a history of malignancy and especially
those with documented malignant effusions have a
greater risk of recurrence.
Procedural morbidity was low overall, but it was sig-
nificantly higher in the TPW group and related mainly to
complications associated with accessing the pleural space
(particularly pneuomothoraces). In that sense, the higher
overall morbidity seems to reflect the greater complexity
of the procedure and the fact that the surgeon is entering
an additional cavity. Several series of SXD report com-
plication rates of less than 4% [15, 18, 20], but higher rates
have been reported [16, 17]. Most series of TPW report no
complications [4 7, 10]; however Geissbuhler and col-
leagues [8] reported a 12% incidence. In our study,
hospital mortality was 13% among patients having SXD,
which is similar to rates reported in other series [1, 2,
1120], and zero among TPW patients. However, because
none of the mortality was specific to the procedure, we
interpret this as reflecting that patients with greater
comorbidities were selected for the SXD procedure,
which was perceived by the surgeons to be less invasive.
Patients with concomitant critical illness or very ad-
vanced malignancy, and those who otherwise seemed
not to be clinically suited to the somewhat more complex
Ann Thorac Surg
2005;80:20139
TPW procedure were likely offered SXD as the simplest
option.
Anesthesia time was dramatically longer in the TPW
group, a finding that we had anticipated given the added
time necessary for (1) placement of a double-lumen
endotracheal tube, (2) lateral decubitus positioning, and
(3) concomitant thoracoscopic procedures. Mean opera-
tive times have ranged from 27 to 57 minutes in previous
series [8, 9, 22] that presumably measured skin-to-skin
time. We elected to measure anesthesia time to reflect the
total time invested by the care team and as the best
reflection of operating room costs.
The only other study we are aware of that reported a
concurrent series of subxiphoid and thoracoscopic surgi-
cal procedures for drainage of pericardial effusion fo-
cused not on the type of procedure but on other prog-
nostic factors [24]. In that publication, the authors found
that morbidity was not dependent on which procedure
was performed. A careful reading of their publication
allows one to glean that crude recurrence rates were
similar according to procedure (ie, 2 of 14 after the
subxiphoid procedure and 1 of 30 after the thoracoscopic
procedure), but this study did not report recurrence data
according to procedure by multivariate analysis as in our
study.
We did not measure postoperative pain in this study.
However, we believe it is likely that the small, upper
midline incision for SXD is less painful than the two to
three intercostal incisions required for TPW. Certainly
the abdominal incision confers no risk of the prolonged
intercostal neuralgia that can occur after a thoracoscopic
operation [27]. We also did not measure costs, but one
would expect that the documented longer anesthesia
times and the use of thoracoscopic disposables would
favor SXD in this regard, particularly with the finding
that the length of stay was similar between the groups.
Regarding diagnostic accuracy, it might be expected
that the opportunity to provide a larger biopsy of peri-
cardium at TPW would afford improved diagnostic capa-
bility. However, our data do not clearly establish this, as
there was no significant difference in the rate of biopsies
positive for malignancy obtained by the two methods.
Limitations of this series include its retrospective de-
sign, the small sample size in the thoracoscopic group,
and the lack of routine surveillance echocardiograms.
Although a follow-up echocardiogram was available in
56%, routine surveillance might have picked up more
recurrent effusions. However, if such undetected recur-
rences were asymptomatic or occurred in an end-stage
patient, their clinical significance is likely to be small. We
cannot exclude the possibility though that some patients
died of recurrent tamponade that went undiagnosed. In
general, our conclusions must be tempered by all of the
biases inherent in a retrospective study.
To summarize this retrospective series, TPW and SXD
were both reasonably effective in controlling pericardial
effusions. However, when adjusted for confounding vari-
ables, TPW conferred a significantly lower risk of recur-
rent effusion. The thoracoscopic procedure was associ-
ated with a longer operative time and higher rate of
Ann Thorac Surg
2005;80:20139
minor, procedure-related morbidity. Echocardiographic
evidence of mild tamponade did not render the thoraco-
scopic approach unsafe, and TPW was more likely to be
used when concomitant intrapleural procedures were
required.
We conclude from these results that because it is
simpler, faster, and slightly less morbid, SXD should be
the preferred approach in the surgical management of
pericardial effusion if a patients life expectancy is likely
to be extremely limited due to major comorbidities or
extensive metastatic disease. Patients with benign dis-
ease, those with malignancy that has not metastasized
extensively or is exquisitely responsive to chemotherapy
and thus may enjoy prolonged survival, and those re-
quiring concomitant intrapleural procedures should be
considered for TPW. We look forward to a prospective,
randomized trial that may shed further light on this
question.
The authors thank Julien Bonifacio for database work that made
this publication possible.
References
1. Moores DWO, Allen KB, Faber LP, et al. Subxiphoid drain-
age for pericardial tamponade. J Thorac Cardiovasc Surg
1995;109:546 52.
2. Sugimoto JT, Little AG, Ferguson MK, et al. Pericardial
window: mechanisms of efficacy. Ann Thorac Surg 1990;50:
4425.
3. Piehler JM, Pluth JR, Schaff HV, et al. Surgical management
of effusive pericardial disease. J Thorac Cardiovasc Surg
1985;90:506 16.
4. Mack MJ, Landreneau RJ, Hazelrigg SR, et al. Video thora-
coscopic management of benign and malignant pericardial
effusions. Chest 1993;103:390S3S.
5. Hazelrigg SR, Mack MJ, Landreneau RJ, et al. Thoracoscopic
pericardiectomy for effusive pericardial disease. Ann Thorac
Surg 1993;56:7925.
6. Shapira OM, Aldea GS, Fonger JD, Shemin RJ. Video-
assisted thoracic surgical techniques in the diagnosis and
management of pericardial effusion in patients with ad-
vanced lung cancer. Chest 1993;104:12623.
7. Liu HP, Chang CH, Lin PJ, et al. Thoracoscopic management
of effusive pericardial disease: indications and technique.
Ann Thorac Surg 1994;58:16957.
8. Geissbuhler K, Leiser A, Fuhrer J, Ris H-B. Video-assisted
thoracoscopic pericardial fenestration for loculated or recur-
rent effusions. Eur J Cardio-Thorac Surg 1998;14:403 8.
9. Robles R, Pinero A, Lujan JA, et al. Thoracoscopic partial
pericardiectomy in the diagnosis and management of peri-
cardial effusion. Surg Endosc 1997;11:253 6.
OBRIEN ET AL 2019
GENERAL THORACIC
PERICARDIAL WINDOW COMPARISON
10. Nataf P, Cacoub P, Regan M, et al. Video-thoracoscopic
pericardial window in the diagnosis and treatment of peri-
cardial effusions. Am J Cardiol 1998;82:124 6.
11. Santos GH, Frater RWM. The subxiphoid approach in the
treatment of pericardial effusion. Ann Thorac Surg 1977;23:
46770.
12. Prager RL, Wilson CH, Bender HW. The subxiphoid ap-
proach to pericardial disease. Ann Thorac Surg 1982;34:6 9.
13. Hankins JR. Satterfield JR, Aisner J, Wiernik PH, McLaughlin
JS. Pericardial window for malignant pericardial effusion Ann
Thorac Surg 1980;30:46571.
14. Campbell PT, Van Tright P, Wall TC, et al. Subxiphoid
pericardiotomy in the diagnosis and management of large
pericardial effusions associated with malignancy. Chest
1992;101:938 43.
15. Van Trigt P, Douglas J, Smith PK, et al. A prospective trial of
subxiphoid pericardiotomy in the diagnosis and treatment of
large pericardial effusion. Ann Surg 1993;218:777 82.
16. Levin BH, Aaron BL. The subxiphoid pericardial window.
Surg Gynecol Obstet 1982;155:804 6.
17. Alcan KE, Zabetakis PM, Marino ND, et al. Management of
acute cardiac tamponade by subxiphoid pericardiotomy.
JAMA 1982:247:1143 8.
18. McDonald JM, Meyers BF, Guthrie TJ, et al. Comparison of
open subxiphoid pericardial drainage with percutaneous
catheter drainage for symptomatic pericardial effusion. Ann
Thorac Surg 2003;76:811 6.
19. Palatianos GM, Thurer RJ, Pompeo MQ, Kaiser GA. Clinical
experience with subxiphoid drainage of pericardial effu-
sions. Ann Thorac Surg 1989;48:3815.
20. Allen KB, Faber LP, Warren WH, Shaar CJ. Pericardial
effusion: subxiphoid pericardiotomy versus percutaneous
drainage. Ann Thorac Surg 1999;67:437 40.
21. Dosios TD, Theakos N, Angouras D, Asimacopoulos P. Risk
factors affecting the survival of patients with pericardial
effusion submitted to subxiphoid pericardiostomy. Chest
2003;124:242 6.
22. Ohtsuka T, Wolf RK, Wurnig P, Park SE. Thoracoscopic
limited pericardial resection with an ultrasonic scalpel. Ann
Thorac Surg 1998;65:855 6.
23. Ohtsuka T, Takamoto S, Nakajima J, Miyairi T, Kotsuka Y.
Minimally invasive limited pericardiectomy: the hybrid ap-
proach. Ann Thorac Surg 2000:70:1429 30.
24. Cullinane CA, Paz IB, Smith D, Carter N, Grannis FW Jr.
Prognostic factors in the surgical management of pericardial
effusion in the patient with concurrent malignancy. Chest
2004;125:1328 34.
25. Fowler NO. Cardiac tamponade: a clinical or an echocardio-
graphic diagnosis. Circulation 1993;87:1738 40.
26. Shabetai R. Pericardial effusion; haemodynamic spectrum.
Heart 2004;90:255 6.
27. Landreneau RJ, Mack MJ, Hazelrigg SR, et al. Prevalence of
chronic pain after pulmonary resection by thoracostomy or
video-assisted thoracic surgery. J Thorac Cardiovasc Surg
1994;107:1079 86.