ORIGINAL ARTICLE: Clinical Endoscopy

Jackson Pratt drain fluid-to-serum bilirubin concentration ratio

for the diagnosis of bile leaks
Peter Darwin, MD, Eric Goldberg, MD, Lance Uradomo, MD, MPH
Baltimore, Maryland, Washington, DC, USA

Background: Jackson Pratt (JP) drain fluid bilirubin levels may be assayed in the evaluation of possible bile leaks.
Although fluid color and bilirubin level may prompt additional evaluation, there are no reference data available.
Objective: To assess the JP drain fluid-to-serum bilirubin ratio in patients with documented bile leaks.
Design: Prospective case series.
Setting: Tertiary referral center.
Methods: Patients referred for ERCP for the management of documented bile leaks with a JP drain in place were
included. Demographic data, bile leak etiology, and serum bilirubin levels were recorded. JP drain fluid was sent
for color evaluation and bilirubin concentration. Control subjects included both patients after nonbiliary surgery
with a JP drain in place and medical patients with ascites undergoing paracentesis.
Results: JP drain fluid-to-serum bilirubin concentration and fluid color evaluation was performed on 23 patients
with documented bile leaks by ERCP and compared with 26 controls (16 surgical and 10 medical). The JP drain
fluid/ascites-to-serum bilirubin ratio was significantly higher in those with bile leaks (mean ratio 45.6) compared
with combined controls (mean ratio 0.9). Use of a cutoff JP drain fluid-to-serum bilirubin ratio of 5 would be
100% sensitive and specific for the prediction of a bile leak in the selected control group. There was overlap
in fluid color evaluation between the groups.
Limitations: Controls did not include those with suspected bile leaks and negative technetium 99m–HIDA scin-
tigraphy or ERCP findings.
Conclusions: JP drain fluid-to-serum bilirubin concentration ratio greater than 5 seems to be highly sensitive
and specific for the detection of a bile leak. Used along with clinical criteria, this ratio could be used to select
patients to proceed directly to ERCP. (Gastrointest Endosc 2010;71:99-104.)

Endoscopic therapy has proven beneficial as a first-line
treatment for the management of most bile leaks.1,2 Tech-
netium 99 m–HIDA scintigraphy (HIDA) has a high accu-
racy for predicting the presence of bile leaks and is often
used as a noninvasive guide to determine those in need
of endoscopic management.3 However, it may be difficult
to mobilize trauma patients for the transport to the nuclear
medicine department, and an occasional unstable patient
Abbreviations: HIDA, technetium 99m–HIDA scintigraphy; JP, Jackson
Pratt.
DISCLOSURE: All authors disclosed no financial relationships relevant
to this publication.
See CME section; p. 139.
Copyright ª 2010 by the American Society for Gastrointestinal Endoscopy
0016-5107/$36.00
doi:10.1016/j.gie.2009.08.015
has difficulty lying flat for the approximate 1 hour required
for the study. There are also potential logistical consider-
ations during off hours, recent issues with HIDA tracer
availability, and the additional time and cost required.4
Patients with a Jackson Pratt (JP) drain in place may have
fluid visually inspected or bilirubin concentration assayed
to determine the presence of a potential bile leak. This is
often done when the output is high, the output increases,
or color appears compatible with bile. However, color eval-
uation may be subjective, and there are few reference data
to use as a guide when evaluating JP drain fluid bilirubin
levels. If reliable for detecting leaks, JP drain fluid evalua-
tion for bilirubin concentration would be less time-con-
suming and costly than an HIDA scan before ERCP.
We performed this study to evaluate JP drain fluid-to-
serum bilirubin level ratio in patients with bile leaks docu-
mented by ERCP, as well as in both surgical and medical

www.giejournal.org Volume 71, No. 1 : 2010 GASTROINTESTINAL ENDOSCOPY 99

JP bilirubin for bile leak diagnosis
controls. The goal was to determine a ratio that is accurate
for detecting a biliary leak. Clinically, JP drain fluid evalua-
tion could be used to determine those at high risk of
a leak without the need for HIDA scan evaluation.
PATIENTS AND METHODS
The study was approved by the University of Maryland
Institutional Review Board for research. All patients re-
ferred for ERCP for the management of documented bile
leaks from September 2005 to February 2007 were asked
to participate. All patients had high (defined as O200
mL/24 h for at least 2 days) or persistent (O20 mL/24 h
that continued for more than a week) output from the
JP drain or HIDA scan results that were available for re-
view. Patients were either from our own institution or re-
ferred for ERCP. All were older than 18 years of age, and
informed study consent was obtained before the proce-
dure. Information recorded included patients’ age, sex,
bile leak etiology, 24-hour JP drain fluid volume when
available, and serum bilirubin within 7 days of referral.
Before the procedure, JP drain fluid was emptied into
a chemistry tube and sent for analysis. Color determina-
tion was performed by visual inspection by a laboratory
technician blinded to the patient’s history. A color was se-
lected from a computerized list that included the follow-
ing choices: clear, slightly cloudy, cloudy, turbid, bloody,
straw, yellow, orange, amber, green, and brown as either
clear or cloudy. Routine bilirubin analysis was then per-
formed and results recorded in milligrams per deciliter.
ERCP was performed with therapy as indicated by the
findings. Endoscopic procedure data were recorded in
a standard computerized report system. The patients
were observed in recovery and then discharged or re-
turned to their hospital bed or referring institution. If a bil-
iary stent was placed, follow-up was arranged in 3 to 4
weeks for stent removal. JP drains were removed when
the output decreased to less than 20 mL in 24 hours.
Surgical controls included those patients admitted for
postoperative care for nonbiliary tract surgery. All had in-
traperitoneal JP drains in place. Because the aim of the
study was to compare fluid bilirubin levels in patients
who definitively had a leak with those who did not, we ex-
cluded controls with possible small or sealed leaks. Specif-
ically, the surgical controls excluded were those with
biliary anastomosis, liver resection, and hepatic trauma.
The posttrauma patients included were those explored
and found to have no hepatic injury. There were no post-
cholecystectomy control subjects with negative HIDA or
ERCP findings. Medical controls were those undergoing
diagnostic or therapeutic paracentesis. Data similar to
those of the bile leak group were collected, and the fluid
samples were handled in a similar manner.
100 GASTROINTESTINAL ENDOSCOPY Volume 71, No. 1 : 2010
Darwin et al
Capsule Summary
What is already known on this topic
d Patients with a Jackson Pratt (JP) drain in place may have
fluid visually inspected or bilirubin concentration assayed
to determine the presence of a potential bile leak, but
color evaluation may be subjective and few reference
data are available for evaluating JP drain fluid bilirubin
levels.
What this study adds to our knowledge
d In a pilot study comparing 23 patients with documented
bile leaks with 26 medical and surgical controls, JP drain
fluid/ascites–to–serum bilirubin ratio was significantly
higher in those with bile leaks compared with all
controls, but there was overlap in fluid color evaluation
among the groups.
Statistical analysis
Mean (95% CI) bilirubin levels are expressed in milli-
grams per deciliter. Mean bilirubin and the ratio of JP drain
fluid-to-serum bilirubin were compared among groups by
using the t test. Dichotomous variables were analyzed with
the Pearson c2 test and the Fisher exact test. A 2-sided P
value !.05 was considered significant. All analyses were
performed by using statistical software (Intercooled Stata,
version 8.2; StataCorp LP, College Station, Tex).
RESULTS
A total of 49 patients were enrolled; 23 patients were in
the study group of bile leaks. Eighteen (78.2%) of the 23
patients were post-cholecystectomy (13 laparoscopic and
5 open), 4 were post-trauma, and 1 was a living related
liver donor. All bile leak patients had high or persistent
JP drain outputs, positive HIDA findings, or both. Fifteen
(65.2%) of 23 patients had HIDA scans performed, and
all were positive. All study patients had leaks documented
by ERCP (Fig. 1).
The control group consisted of 26 patients with 10
medical controls and 16 surgical controls. The medical
control group was composed of 8 patients with underlying
cirrhosis/portal hypertension and 2 with malignant ascites.
The surgical control group consisted of 4 patients with
small bowel resection or repair, 3 patients with abdominal
trauma (no liver injury by exploration), 2 patients with co-
lonic resection, 2 patients with ventral hernia repair, and 1
each with open cyst duodenostomy, gastric bypass, total
gastrectomy, enterocutaneous fistula repair, and pancre-
atic necrosectomy.
Bilirubin analysis
The mean bilirubin in the JP drain fluid/ascites of
the combined control group was 1.13 mg/dL (95% CI,
www.giejournal.org
Darwin et al
Figure 1. Cholangiogram demonstrating a cystic stump leak with a JP
drain in place.
0.75-1.52) (Table 1). The mean ascites bilirubin of the med-
ical patients was 1.12 mg/dL (95% CI, 0.19-2.05) and that of
the surgical patients was 1.14 mg/dL (95% CI, 0.76–1.51; P Z
.96 for the difference between medical and surgical). The
mean bilirubin in the JP drain fluid of the study patients
was markedly elevated (39.69 mg/dL; 95% CI, 28.78-50.61).
The difference was dramatic between the study group and
the medical (P! .0001), surgical (P! .0001), and combined
control (P! .0001) groups.
The ratio of JP drain fluid to serum bilirubin was calcu-
lated to account for differences in serum bilirubin levels
among the groups (Fig. 2). The combined control group
JP drain fluid–serum ratio was 0.94 (95% CI, 0.62–1.26).
The medical and surgical control patient JP drain fluid–
serum ratios were 0.43 (95% CI, 0.15–0.70) and 1.27
(95% CI, 0.83–1.70), respectively (P Z .006 for the differ-
ence between medical and surgical control patients).
These figures were markedly lower than the JP drain
fluid–serum bilirubin ratio for the patients with bile leaks.
Their mean ratio was 45.67 (95% CI, 34.08–57.25; P !
.0001 for comparison with all control groups). The mean
ratio for those with high-volume versus persistent leaks
was similar at 47.8 versus 41.2, respectively.
Receiver operating characteristic analysis was per-
formed to identify the JP drain fluid bilirubin and the JP
drain fluid–serum ratio that would optimize their sensitiv-
ity and specificity in the identification of bile leaks. For JP
drain fluid bilirubin, use of a cutoff of 4.9 mg/dL or greater
resulted in 100% sensitivity and 100% specificity. For the
drain fluid-to-serum ratio, a cutoff of 5.4 or greater was
100% sensitive and specific. The receiver operating charac-
teristic area for both measures was 1.
There was a trend toward a higher ratio if the bile leak
occurred after cholecystectomy (50.75; 95% CI, 37–64.4)
compared with other causes (37.52; 95% CI, 10.33–44.72;
P Z .087). Both of these subgroups had significantly
higher ratios than the controls (P ! .0001).
www.giejournal.org
JP bilirubin for bile leak diagnosis
Color and clarity
The color and clarity characterizations for each JP drain
or ascites sample are listed in Table 2. Straw color was
more likely to be described in the control group (16.7%)
compared with the study group (0%; P Z .04). Fluid
was described as brown in only study patients (P Z
.003). Smaller numbers of clear, red, and green fluid
were seen in only the control (clear, red) or study (green)
group. There was no difference among groups in the
clarity of the fluid (P Z .38), although bloody fluid was
observed only in surgical controls.
Fluid volume assessment
Nineteen (82.6%) of 23 of the patients with bile leak had
24-hour JP drain fluid measurements available, with a range
of output from 40 to 950 mL daily and a mean output of
281.8 mL daily. There were 12 patients with high-volume
JP drain output (O200 mL/day) and 7 with persistent leaks
(O20 mL/day). The volume in the surgical control group
ranged from 5 to 580 mL daily (available in records re-
viewed from 12 of 16 subjects), with a mean of 137.5 mL/
24 h (P Z .065 compared with the bile leak group). A re-
ceiver operating characteristic analysis was performed to
evaluate the utility of drain volume identifying the
patients with bile leaks. The area under the fitted curve
was 0.78. The characteristics of fluid volume were inferior
to those for bilirubin concentrations. A cutoff of 75 mL/24
h carried a sensitivity of 89% and a specificity of 59%,
whereas at 110 mL/24 h, the sensitivity was 77% and spec-
ificity was 69%.
DISCUSSION
Biliary leakage after laparoscopic cholecystectomy is
the most commonly encountered postoperative complica-
tion.3 Causes may include a cystic stump or duct of Luszka
leak and main duct injury. Bile leaks after open cholecytec-
tomy or biliary anastomosis and posttraumatic injury are
also seen. Liver injury often is encountered after blunt ab-
dominal trauma, with damage to either the intrahepatic or
extrahepatic biliary tree.5 Penetrating abdominal injuries
as well commonly result in liver injury. These patients re-
quire a ‘‘damage-control’’ laparotomy that may proceed in
stages along with other modalities.6,7
HIDA scans have a high accuracy for predicting the
presence of bile leaks after laparoscopic cholecystectomy
and other biliary surgeries.3,8,9 Conversely, elevated liver
enzymes and bilirubin levels after cholecystectomy may
transiently be seen and often do not predict postoperative
complications.10,11 Bile leaks may also be seen without
liver enzyme elevation. Those patients with postoperative
JP drains in place have monitoring of fluid output volume
and color to help determine whether there is a postopera-
tive leak. Patients with difficult cholecystectomy or those
deemed at high risk of a postoperative leak may have
Volume 71, No. 1 : 2010 GASTROINTESTINAL ENDOSCOPY 101
JP bilirubin for bile leak diagnosis Darwin et al

TABLE 1. Mean serum bilirubin, Jackson Pratt drain output bilirubin, and ratio of Jackson Pratt drain output-to-serum bilirubin in
patients with bile leak compared with medical and surgical controls

Mean serum bilirubin, mg/dL Mean JP drain output bilirubin, mg/dL Mean ratio: JP drain output–serum
Controls (95% CI) (95% CI) bilirubin, mg/dL (95% CI)

Combined 2.88 (0.97-4.79) 1.13 (0.75-1.52) 0.94 (0.62–1.26)

Medical 5.6 (0.72-10.48) 1.12 (0.19-2.05) 0.43 (0.15–0.70)
Surgical 1.18 (0.54-1.82)* 1.14 (0.76–1.51) 1.27 (0.83–1.70)*
Bile leak 1.00 (0.77-1.24)y 39.69 (28.78-50.61)z 45.67 (34.08–57.25)z
CI, Confidence interval; JP, Jackson Pratt.
*P ! .05 compared with medical controls.
yP ! .005 compared with Medical controls.
zP ! .0001 compared with all control groups.

TABLE 2. Color and clarity of Jackson Pratt drain fluid

and ascites samples in patients with bile leak
compared with medical and surgical controls

Surgical Medical Study

Color controls controls group Total

Clear 1 1 0 2
Yellow 5 6 8 19

Straw 2 2 0 4
Amber 2 0 5 7
Orange 1 1 1 3
Red 3 0 0 3
Figure 2. Mean JP drain fluid-to-serum bilirubin ratio: patients with bile
leak compared with medical and surgical controls. *P ! .0001 for com- Brown 0 0 7 7
parison with all control groups. Green 0 0 2 2

a JP drain left in place at the time of surgery. In addition,
postoperative pain or collections may prompt laparo-
scopic or percutaneous drain placement. To further assess
patients with a high drain output (O200 mL/day), ERCP or
a HIDA scan may be performed.2 However, JP drain fluid
volume may be confounded by ascites from portal hyper-
tension or other causes. A persistent or abruptly increas-
ing JP output may also indicate an ongoing leak. Our
general policy is to monitor serial 24-hour output and
pull such drains when the output decreases to less than
20 mL/day. Clinically, fluid color that appears bilious or
fluid bilirubin concentration may be used to stratify those
at high risk of leak who require intervention, but there are
few reference data to guide management.
ERCP has proved to be effective for bile leak manage-
ment. Endoscopic obliteration of the transpapillary pres-
sure gradient with short stents allows flow of bile into
the duodenum and is an effective method of pressure
equalization in postsurgical leaks.12 Endoscopic sphincter-
otomy alone or in combination with nasobiliary drainage
has also been used as therapy. Similarly, traumatic bile leaks
may be initially managed with endoscopic therapy, provid-
ing successful outcomes in a majority of cases.13 However,
endoscopic therapy is not without risks and is often
pursued only when leaks are documented by HIDA scan.
Although HIDA scintigraphy is noninvasive, there are
logistical considerations in ill patients as well as added
time and cost. There has also been a recent shortage of
tracer available for clinical use. If reliable in patients with
a drain in place, JP drain fluid-to-serum bilirubin concentra-
tion ratio for the diagnosis of bile leaks could be used to
select those in need of endoscopic therapy and avoid
HIDA scan or diagnostic ERCP. However, HIDA scintigraphy
before ERCP may still play a role in selected cases to help
define the site of the leak before therapy.
Our results demonstrate that subjective color evalua-
tion of JP drain fluid has variable results, and there is
not a classic fluid appearance in proven cases of bile
leak. However, JP drain fluid bilirubin is significantly
higher in those with bile leaks compared with surgical
and medical controls. Using the JP drain output/ascites-
to-serum bilirubin ratio also demonstrates a marked

102 GASTROINTESTINAL ENDOSCOPY Volume 71, No. 1 : 2010 www.giejournal.org

Darwin et al
difference between the groups. A similar ascites-to-serum
bilirubin ratio has been evaluated in patients with cirrhosis
and malignancy. Bilirubin levels in cirrhotic ascites are
know to be lower than serum levels.14 A ratio greater
than 0.6 has been used as a marker for distinguishing
transudates from exudates,15 but the serum-to-ascites
albumin ratio has been most widely used for this purpose.
Runyon16 measured this ratio in 65 patients with various
types of ascites to determine the normal range. The
mean ratio was similar to our medical control group at
0.38  0.44. One patient in this series had a ruptured gall-
bladder with a ratio of 7.1. An ascites–to-serum bilirubin
ratio greater than 1.0 was suggested as a threshold for
the diagnosis of choleperitoneum. However, the mean ra-
tio in our surgical controls without biliary tract manipula-
tion was higher than this threshold.
Posthepatic resection bilirubin levels have been shown
to increase in drain discharge late in the postoperative
course,17 but there is otherwise scant literature on fluid/
serum bilirubin level in postsurgical patients. Truedson18
evaluated drainage fluid–conjugated bilirubin in a series
of patients after open cholecystectomy, but no correlation
with serum values or evaluation of possible leaks was per-
formed. In our study, using a JP drain fluid bilirubin level
greater than 4.9 mg/dL or a JP drain fluid–serum ratio cut-
off greater than 5.4 would include all patients with a docu-
mented leak. The combination of cirrhosis/ascites with
a bile leak could not be evaluated because there were
no such patients in our series.
Although not a primary study aim, JP drain volume alone
was evaluated as a predictor of a bile leak. The 24-hour vol-
ume in both study subjects and surgical controls was
recorded when available. All study subjects had either
high-volume or persistent JP drain output. Although vol-
ume evaluation had a lower sensitivity and specificity
than JP drain output bilirubin, the daily drain output after
cholecystectomy is an important factor in determining
the need for additional evaluation and therapy. For study
purposes, those with outputs that continued to be more
than 200 mL/day were considered high output and those
that remained more than 20 mL/week were persistent. We
continue to follow these parameters in our clinical practice.
Limitations of this study include a relatively small sam-
ple size and no blinded comparison of HIDA scan with the
JP drain fluid–serum bilirubin concentration ratio. In addi-
tion, this was a single-center study with no controls with
presumed bile leaks that had negative HIDA scan or
ERCP findings. The control group selected had no possi-
bility of a bile leak by design. Additional control patients
could have included those after cholecystectomy, biliary
anastomosis, liver resection, and hepatic trauma. Our ini-
tial study design excluded evaluation of such patients be-
cause we could not exclude a small or sealed leak nor
justify the added cost/risk of a HIDA scan or ERCP solely
for study purposes. The control group selected therefore
does not allow us to address those patients with a sus-
www.giejournal.org
JP bilirubin for bile leak diagnosis
pected leak, high JP drain fluid bilirubin, and negative
HIDA/ERCP findings or distinguish between those leaks
that may or may not respond to endoscopic therapy. Fur-
ther study would be required to address these questions
with a HIDA scan and ERCP in all patients. Based on the
above limitations, the JP drain fluid–serum bilirubin con-
centration ratio should not be used as an isolated indica-
tor of those patients who could benefit from endoscopic
therapy. The management of patients with suspected
bile leaks still relies on clinical criteria including serial
drain fluid volume assessment if a JP drain is in place.
However, the control group used in this study does give
baseline reference JP drain fluid bilirubin levels in patients
postoperatively. In theory, this group should reflect those
patients without a leak after cholecystectomy.
In summary, evaluation of JP drain fluid bilirubin or JP
drain output/serum bilirubin concentration ratio would
seem accurate in patients with documented leaks as dem-
onstrated by ERCP. Postoperative or trauma patients with
an elevated JP drain output bilirubin or a ratio greater
than 5 along with a high or persistent JP drain output
should be considered at high risk of a bile leak. Even
though such patients with an elevated JP drain output af-
ter cholecystectomy could proceed directly to a therapeu-
tic procedure, referring physicians often order a HIDA
scan to confirm the diagnosis. JP drain fluid bilirubin as-
sessment is quicker and less costly if a confirmatory test
before intervention is desired.
REFERENCES
1. Katsinelos P, Kountouras J, Paroutoglou G, et al. The role of endo-
scopic treatment in postoperative bile leaks. Hepatogastroenterology
2006;53:166-70.
2. Ahmad F, Saunders RN, Lloyd GM, et al. An algorithm for the manage-
ment of bile leak following laparoscopic cholecystectomy. Ann R Coll
Surg Engl 2007;89:51-6.
3. Brugge WR, Rosenberg DJ, Alavi A. Diagnosis of postoperative bile
leaks. Am J Gastroenterol 1994;89:2178-83.
4. Perkins A, Hilson A, Hall J. Global shortage of medical isotopes
threatens nuclear medicine services. BMJ 2008;337:a1577.
5. Sharif K, Pimpalwar AP, John P, et al. Benefit of early diagnosis and
preemptive treatment of biliary tract complications after major blunt
liver trauma in children. J Pediatr Surg 2002;37:1287-92.
6. Cogbill TH, Moore EE, Jurkovich GJ, et al. Severe hepatic trauma:
a multi center experience with 1335 liver injuries. J Trauma 1998;28:
1422-38.
7. Asenio JA, Demetriades D, Chaahwan S, et al. Approach to the man-
agement of complex hepatic injuries. J Trauma 2000;48:66-9.
8. Shinhar S, Nobel M, Shimonov M, et al. Technetium-99 m-HIDA scintig-
raphy vs ERCP in demonstrating bile leaks post cholecystectomy.
J Nucl Med 1998;39:1802-4.
9. Peters JH, Ollila D, Nichols KE, et al. Diagnosis and management of bile
leaks following laparoscopic cholecystectomy. Surg Laparosc Endosc
1994;4:163-70.
10. Kaldor A, Akopian G, Recabaren J, et al. Utility of liver function tests
after laparoscopic cholecystectomy. Am Surg 2006;27:1238-40.
11. Halevy A, Gold-Deutch R, Negri M, et al. Are elevated liver enzymes
and bilirubin levels significant after laparoscopic cholecystectomy in
the absence of bile duct injury? Ann Surg 1994;219:362-4.
Volume 71, No. 1 : 2010 GASTROINTESTINAL ENDOSCOPY 103
JP bilirubin for bile leak diagnosis Darwin et al

12. Bjorkman D, Carr-Locke D, Lichtenstein D, et al. Postsurgical bile leaks: 18. Truedson H. Cholecystectomy with intraperitoneal drain. Presence and
endoscopic obliteration of the transpapillary pressure gradient is significance of conjugated bilirubin in the drainage fluid. Acta Chir
enough. Am J Gastroenterol 1995;90:2128-33. Scand 1983;149:179-85.
13. Bridges A, Wilcox CM, Varadarajulu S. Endoscopic management of
traumatic bile leaks. Gastrointest Endosc 2007;65:1081-5.
14. Hedenborg G, Jonsson G, Norman A, et al. Bile constituents in ascites
fluid. Scand J Clin Lab Invest 1988;48:543-52. Received January 25, 2008. Accepted August 23, 2009.
15. Elis A, Meisel S, Tisher T, et al. Ascitic fluid to serum bilirubin concen-
Current affiliations : Division of Gastroenterology-Hepatology (P.D., E.G.),
tration ratio for the classification of transudates or exudates. Am J Gas-
University of Maryland Medical System, Baltimore, Maryland, Division of
troenterol 1998;93:401-3.
Gastroenterology (L.U.), George Washington University, Washington, DC,
16. Runyon BA. Ascitic fluid bilirubin concentration as a key to choleper-
USA.
itoneum. J Clin Gastroenterol 1987;9:543-5.
17. Torzilli G, Olivari N, Del Fabbro D, et al. Bilirubin level fluctuation in Reprint requests: Peter Darwin, MD, Division of Gastroenterology-
drain discharge after hepatectomies justifies long-term drain mainte- Hepatology, University of Maryland Medical System, 22 South Greene
nance. Hepatogastroenterology 2005;52:1206-10. Street, N3W62, Baltimore, MD 21201.

104 GASTROINTESTINAL ENDOSCOPY Volume 71, No. 1 : 2010 www.giejournal.org