THE AMERICAN JOURNAL OF GASTROENTEROLOGY
1, 2000
© 2000 by Am. Coll. of Gastroenterology
Published by Elsevier Science Inc.
Bacterial Peritonitis After Elective
Endoscopic Variceal Ligation: A Prospective Study
Otto S. Lin, M.D., Shun-Sheng Wu, M.D., Yang-Yuan Chen, M.D., and Maw-Soan Soon, M.D.
Division of Gastroenterology, ChangHua Christian Medical Center, ChangHua, Taiwan; and Division of
Gastroenterology, Stanford University Medical Center, Stanford, California
OBJECTIVE: Endoscopic variceal ligation is becoming the
therapy of choice for esophageal varices, replacing endo-
scopic variceal sclerotherapy. The latter is associated with a
5–53% incidence of port-procedural bacteremia and a
0.5–3% incidence of peritonitis, whereas the former carries
a 3– 6% risk of bacteremia. However, the incidence of
peritonitis after variceal ligation has not been well studied.
This prospective study is designed to investigate the risk of
developing bacteremia and bacterial peritonitis after elective
endoscopic variceal ligation.
METHODS: Sixty-seven patients with esophageal varices and
ascites secondary to liver cirrhosis underwent elective en-
doscopic variceal ligation. Before the procedure, ascitic
fluid was drawn under ultrasound guidance and sent for cell
counts, Gram stain, and cultures. Two to 4 days afterward,
a repeat ascitic fluid sample was sent for the same studies
whether or not the patient had symptoms or signs suggestive
of infection. Blood cultures were drawn both immediately
before and after the endoscopic ligation procedure.
RESULTS: Of 67 subjects, 11 developed asymptomatic bac-
teremia with Gram-positive commensals. However, none of
them progressed to peritonitis. Two patients who did not
have bacteremia developed mild febrile peritonitis with
Escherichia coli and were successfully treated with oral
antibiotics. No other infectious complications were noted.
CONCLUSIONS: There is a significant risk of asymptomatic
bacteremia and bacterial peritonitis after elective variceal
ligation. The peritonitis does not seem to be related to the
bacteremia, as patients who had bacteremia did not develop
peritonitis and vice versa. In addition, the involved organ-
isms were quite different. Unlike the bacteremia, postliga-
tion peritonitis may be a consequence of severe liver cir-
rhosis rather than the procedure itself. The clinical
significance of postligation bacteremia is doubtful. With
regard to peritonitis, in our opinion the use of prophylactic
antibiotics should be reserved for patients with Child’s C
class cirrhosis, a recent history of variceal bleeding, a past
history of bacterial peritonitis, or a comorbid immunosup-
pressive condition. (Am J Gastroenterol 2000;95:214 –217.
© 2000 by Am. Coll. of Gastroenterology)
Vol. 95, No.
ISSN 0002-9270/00/$20.00
PII S0002-9270(99)00746-7
INTRODUCTION
Endoscopic variceal ligation (EVL) is replacing endoscopic
variceal sclerotherapy (EVS) as the therapeutic procedure of
choice for esophageal varices at many medical institutions,
as a result of the former’s superior efficacy and lower
complication rate (1). However, infectious sequelae are of
concern for both procedures. Previous studies have shown a
5% (2) to 53% (3) incidence of bacteremia and a 0.5–3%
incidence of bacterial peritonitis after EVS (4). For EVL, the
risk of bacteremia appears to be much lower, between 3%
and 6% (5, 6). This may lead one to predict that the inci-
dence of bacterial peritonitis after EVL would be negligibly
low. However, few studies have addressed this issue. In this
report, we describe a prospective study investigating the risk
of bacterial peritonitis and its correlation with asymptomatic
bacteremia in patients who underwent elective EVL.
MATERIALS AND METHODS
Between August 1997 and December 1998, all patients
scheduled for elective endoscopic variceal ligation at a
tertiary medical center were first screened for ascites by
transcutaneous abdominal ultrasound. If ascites was present,
a sample was drawn and sent for various laboratory studies,
including complete cell count with manual differentiation of
leukocytes, protein, Gram stain, and both aerobic and an-
aerobic cultures. Two sets of blood cultures were also drawn
and sent for aerobic and anaerobic cultures.
The EVL was then performed using a standard Olympus
Gif-XQ200 video endoscope. The ligation device (Choten
Inc., Tokyo, Japan) was of the pneumatic application type
(MD-48709 with a width of 9 to 10.5 mm). Three to six
bands were applied to each patient, depending on clinical
needs. Between procedures, each endoscope was carefully
rinsed with 70% alcohol, disinfected with Metricide Plus 30
solution (Metrex Inc., Parker, CO) for 30 min, and then
wiped with a sterile cloth. Within 2 h after the procedure,
another two sets of blood cultures were drawn and sent for
cultures. The serum albumin and creatinine levels of each
subject were also measured.
Two to 4 days after the procedure, all subjects were
invited back to the hospital and a second set of ascitic fluid
sample was drawn and sent for the same studies as the
AJG – January, 2000
Table 1. Infectious Complications, Clinical Symptoms, and Laboratory Values in a Sample of 67 Patients Who Underwent Elective
Endoscopic Variceal Ligation
Number of
Complications of EVL Subjects
Neither bacteremia nor peritonitis 54
Bacteremia but no peritonitis 11
Peritonitis but no bacteremia 2 E. coli
Both bacteremia and peritonitis 0 NA
Strep p ⫽ Streptococcus pyogenes; Staph e ⫽ Staphylococcus epidermidis; E. coli ⫽ Escherichia coli; Strep v ⫽ Streptococcus viridans.
pre-EVL samples. A careful history and physical examina-
tion was also done, with particular emphasis on any clinical
signs of peritonitis.
To be eligible for the study, patients had to be aged
between 40 and 75 yr, and had to demonstrate at least grade
1 esophageal varices on endoscopy. Patients belonging to
Child-Pugh class B and class C of liver cirrhosis were
eligible. A history of having undergone EVL or EVS was
acceptable, as long as the previous procedure took place at
least 1 wk before our study and had not led to any infectious
complications. A history of previous variceal bleeding did
not preclude entry into this study; indeed, all of our subjects
had been scheduled for elective EVL because of variceal
bleeding in the past.
Any clinical or laboratory sign of infection, such as ab-
dominal pain, fever, or leukocytosis, present at the time of
the EVL was a criterion for exclusion. Other exclusion
criteria included the presence of a surgical portocaval shunt
or other interventional radiological shunts, concurrent GI
bleeding (within the past 48 h), or the current use of oral or
intravenous antibiotics (within the past 72 h). None of the
subjects received prophylactic antibiotics, either before or
after the EVL.
Informed consent was obtained from all subjects. This
protocol was approved by the Institutional Review Board at
ChangHua Christian Medical Center.
RESULTS
A total of 183 patients who underwent elective EVL were
screened, of which 68 demonstrated the presence of ascites
by ultrasound. One patient had an ascitic neutrophil count of
680 cells/ml and grew out Klebsiella pneumoniae from his
pre-EVL paracentesis fluid. This patient was excluded. The
remaining 67 patients were enrolled in the study. The mean
age of the subjects was 52 yr, and 40 (60%) of them were
men. Thirty-eight of these 67 subjects had Child-Pugh class
B liver cirrhosis, the remainder being class C. All had portal
hypertension due to hepatic cirrhosis from chronic hepatitis
B. All subjects had a past history of esophageal variceal
bleeding, occurring between 5 days and 10 months before
the EVL procedure, with a mean of 52 days.
Of 67 patients, 11 were found to be bacteremic after the
Bacterial Peritonitis After Endoscopic Variceal Ligation 215
Ascitic Mean Mean
Organisms Clinical Fluid Serum Serum
Cultured Symptoms Protein Albumin Creatinine
None None 1.08 3.1 1.4
Strep v (5 subjects) None 1.15 2.9 1.5
Strep p (3 subjects)
Staph e (3 subjects)
Fever 0.96 2.8 1.6
NA NA NA NA
procedure, as demonstrated by the post-EVL blood cultures.
Eight of these patients had Child-Pugh class B cirrhosis,
whereas the remaining three had class C cirrhosis. The
bacteremic subjects had bled, on average, 55 days before
the EVL, which was statistically similar to the 51 days for
the rest of the subjects. The mean albumin and creatinine
levels of the two groups were not significantly different. The
organisms involved were all Gram-positive oral or skin
commensals, namely Streptococcus viridans, Streptococcus
pyogenes, and Staphylococcus epidermidis (Table 1). None
of these 11 patients were symptomatic. Furthermore, none
of them developed bacterial peritonitis, as shown by the
post-EVL paracentesis. These patients were not given anti-
biotics and continued to do well.
Two other patients developed a marked increase in their
ascitic neutrophil count (to ⬎250 cells/ml) and grew Esch-
erichia coli from their ascitic fluid. However, both their
post-EVL blood cultures were negative. Their serum albu-
min and creatinine levels, as well as ascitic fluid protein and
albumin levels, were not markedly different from those of
subjects who did not develop peritonitis (see Table 1), but
both had Child-Pugh class C cirrhosis. The two subjects
with peritonitis had bled 26 and 44 days before the EVL,
respectively, requiring prolonged hospitalization in both
instances. They had mild fever beginning 1 to 2 days after
the EVL, but no other symptoms. After a 7-day course of an
oral third-generation cephalosporin and a fluoroquinolone,
both became afebrile. A follow-up paracentesis 8 days after
the EVL showed normalization of cell counts and negative
ascitic cultures in both subjects. None of the study patients
suffered any other complications from the EVL.
DISCUSSION
Endoscopic variceal sclerotherapy is an effective therapy for
esophageal varices. However, there exists a small but sig-
nificant risk of infectious complications. Previous studies
have shown that EVS is associated with a high incidence (up
to 53%) of postprocedural asymptomatic bacteremia (3) and
a notable incidence (up to 3%) of clinically apparent bac-
terial peritonitis (4, 7). Other infectious sequelae, such as
perinephric abscesses, brain abscesses, subdural empyemas,
endocarditis, and meningitis, have also been reported (4, 8).
216 Lin et al.
It is believed that endoscopic variceal ligation may rep-
resent an alternative and superior treatment to EVS for
esophageal varices. The incidence of transient bacteremia
after EVL (3– 6%) (5, 6) appears to be lower than that after
EVS (5–53%) (2, 3, 9). Furthermore, the total incidence of
all types of infectious complications after EVL (1.8%) is
lower than that for EVS (18%) (6), and a retrospective study
has implied that the rate of clinical bacterial peritonitis after
EVL may also be lower (6). However, no previous study has
prospectively assessed the rate of peritonitis after EVL.
Because the mechanism of the postprocedural bacteremia
and peritonitis is still in question, it is unclear whether or not
EVL truly possesses theoretical advantages over EVS. It is
thought that, except for rare situations where a contaminated
water supply is the culprit (10), post-EVS bacteremia is
caused by the direct inoculation of bacteria into the blood-
stream by the sclerotherapy needle, which is easily contam-
inated when being passed over the endoscope tip (6). In-
deed, bacteremic organisms found immediately after EVL
have been mostly Gram-positive skin and oropharyngeal
commensals such as Streptococcus pyogenes, Staphylococ-
cus epidermidis and aureus, and Diphtheroid species (2, 3,
9 –11). This suggests direct seeding as the mode of entry,
explaining why bacteremia after EVS is more common than
after routine diagnostic upper endoscopy (11). In addition,
one study has demonstrated a correlation between the scle-
rotherapy needle length and the incidence of bacteremia
(12).
Because EVL does not involve the direct penetration of
the esophageal mucosa with a needle, there is less oppor-
tunity for the direct introduction of bacteria. Also, EVL is
done with a protective overtube that prevents ligation bands
from picking up oropharyngeal flora on the way in (6).
Furthermore, the process of ligation itself obliterates sub-
mucosal venous channels, reducing the likelihood of sys-
temic bacteremia (5). These reasons are believed to be why
EVL is associated with less bacteremia than EVS.
However, postprocedural bacterial peritonitis, which oc-
curs after a lag of 1 to 4 days, is not necessarily related to
the bacteremia that occurs within 24 h of the procedure.
Gram-negative enteric organisms are usually recovered
from the ascitic fluid in post-EVS peritonitis (4, 7). These
pathogens are similar to those seen in spontaneous bacterial
peritonitis. There have been rare reports of post-EVS peri-
tonitis featuring the same Gram-positive organism growing
out of both blood and ascitic fluid (13), but in general the
correlation between the two is weak.
Several potential mechanisms for postprocedural perito-
nitis have been put forward. One possibility is that the
variceal ulceration occurring a few days after EVS or EVL
may serve as a portal of entry for bacteria. Some researchers
have suggested a bimodal bacteremia pattern, the first peak
appearing right after the procedure itself and the second
during the variceal ulceration 1 to 3 days later (14). The
peritonitis may be associated with the second peak, although
that does not explain why the recovered ascitic organisms
AJG – Vol. 95, No. 1, 2000
are mostly Gram-negative enterics. Alternatively, transloca-
tion of enteric organisms through the colonic wall may take
place, as in spontaneous bacterial peritonitis (4). The
chances of this occurring are heightened during episodes of
hypotension from variceal bleeding. That may be why stud-
ies have shown higher incidences of peritonitis after emer-
gency EVS than after elective EVS (4), and a higher rate of
pre-endoscopy bacteremia in emergent compared with elec-
tive settings (9). Furthermore, the incidence of peritonitis
appears to be higher in patients with more advanced liver
cirrhosis who undergo EVS (15). Hence, postprocedural
peritonitis may be associated more with the clinical condi-
tion of the patient than with the specific procedure being
performed. If that is true, then EVL should be associated
with a peritonitis incidence rate similar to that of EVS.
Our study shows a significant incidence of bacterial peri-
tonitis after EVL in cirrhotic patients undergoing elective
EVL. The rate of bacteremia is also unexpectedly high.
However, there does not seem to be a relationship between
the two as the 11 patients who had bacteremia did not
develop peritonitis, whereas the two patients who suffered
from clinical peritonitis did not have preceding bacteremia.
Furthermore, the organisms recovered from the blood were
all oral or skin commensals, whereas the peritonitis was
caused by an enteric organism, a finding consistent with that
of earlier studies on EVS.
These results support the theory that post-EVL bactere-
mia and peritonitis are not related. Although transient bac-
teremia after EVL might be relatively common, it is usually
asymptomatic and of limited clinical significance, as in our
11 patients. It is unclear whether or not clinical bacterial
peritonitis, which in our case occurred in two nonbacteremic
patients with advanced cirrhosis and a history of massive GI
bleeding, is related to the procedure. Of note, both patients
had Child-Pugh class C cirrhosis and had bled more recently
than subjects who did not develop peritonitis. Thus, we
believe that the use of prophylactic antibiotics should be
governed by the clinical condition of the patient rather than
the occurrence of the EVL procedure itself. In our opinion,
patients with Child’s C class cirrhosis, a recent history of
variceal bleeding, a past history of bacterial peritonitis, or a
comorbid immunosuppressive condition are candidates for
peritonitis prophylaxis, using various antibiotic regimens
that have been demonstrated to prevent peritonitis even in
high-risk patients (16, 17). Antibiotic prophylaxis against
bacteremia is probably not necessary except in patients with
prosthetic heart valves, surgical systemic pulmonary shunts,
or a history of endocarditis, as described in the American
Society of GI Endoscopy Guidelines on Infection Control
during Endoscopy (18).
Because of the sample size of this study, the true inci-
dence of bacteremia and peritonitis in elective EVL patients
cannot be accurately derived. However, our report repre-
sents a deliberate, prospective investigation of infectious
complications after EVL, designed to detect asymptomatic
patients as well as patients who might be mistaken for other
AJG – January, 2000
infections. Most previous reports of bacterial peritonitis
associated with EVS or EVL have been retrospective studies
that have identified only symptomatic patients. The inci-
dence of post-EVL peritonitis appears to be significant in
this study and can be further elucidated by a larger, pro-
spective trial.
Reprint requests and correspondence: Otto S. Lin, M.D., Divi-
sion of Gastroenterology, Stanford University Medical Center,
1201 Welch Road, Suite P-304, Stanford, CA 94305–5487.
Received Feb. 17, 1999; accepted Aug. 4, 1999.
REFERENCES
1. Laine L, Cook D. Endoscopic ligation compared with sclero-
therapy for treatment of esophageal variceal bleeding: A meta-
analysis. Ann Intern Med 1995;123:280 –7.
2. Camera DS, Gruber M, Barde CJ, et al. Transient bacteremia
following endoscopic sclerotherapy of esophageal varices.
Arch Intern Med 1983;143:1350 –2.
3. Sauerbruch T, Holl J, Ruckdeschel G, et al. Bacteremia asso-
ciated with endoscopic sclerotherapy of esophageal varices.
Endoscopy 1985;17:170 –2.
4. Bac DJ, de Marie S, Siersema PD, et al. Post-sclerotherapy
bacterial peritonitis: A complication of sclerotherapy or of
variceal bleeding? Am J Gastroenterol 1994;89:859 – 62.
5. Tseng CC, Green RM, Burke SK, et al. Bacteremia after
endoscopic band ligation of esophageal varices. Gastrointest
Endosc 1992;38:336 –7.
6. Lo GH, Lai KH, Shen MT, et al. A comparison of the inci-
dence of transient bacteremia and infectious sequelae after
Bacterial Peritonitis After Endoscopic Variceal Ligation 217
sclerotherapy and rubber band ligation of bleeding esophageal
varices. Gastrointest Endosc 1994;40:675–9.
7. Schembre D, Bjorkman DJ. Post-sclerotherapy bacterial peri-
tonitis. Am J Gastroenterol 1991;86:481– 6.
8. Baskin G. Prosthetic endocarditis after endoscopic variceal
sclerotherapy: A failure of antibiotic prophylaxis. Am J Gas-
troenterol 1989;84:311–2.
9. Ho H, Zuckerman MJ, Wassem C. A prospective controlled
study of the risk of bacteremia in emergency sclerotherapy of
esophageal varices. Gastroenterology 1991;101:1642– 8.
10. Brayko CM, Kozarek RA, Sanowski RA, et al. Bacteremia
during esophageal variceal sclerotherapy: Its cause and pre-
vention. Gastrointest Endosc 1985;31:10 –2.
11. Cohen LB, Korsten MA, Scherl EJ, et al. Bacteremia after
endoscopic injection sclerosis. Gastrointest Endosc 1983;29:
198 –200.
12. Snady H, Korsten MA, Waye JD. The relationship of bacte-
remia to length of injection needle in endoscopic variceal
sclerotherapy. Gastrointest Endosc 1985;31:243– 6.
13. Barnett JL, Elta G. Bacterial peritonitis following endoscopic
variceal sclerotherapy. Gastrointest Endosc 1987;33:316 –7.
14. Tam F, Chow H, Prindiville T, et al. Bacterial peritonitis
following esophageal injection sclerotherapy for variceal hem-
orrhage. Gastrointest Endosc 1990;36:131–3.
15. Tank LV, Estay R, Ovalle L, et al. Bacterial peritonitis and
sclerotherapy for variceal hemorrhage. Gastrointest Endosc
1992;38:200 (letter).
16. Soriano G, Guarner C, Tomas A, et al. Norfloxacin prevents
bacterial infection in cirrhotics with gastrointestinal hemor-
rhage. Gastroenterology 1992;103:1267–72.
17. Blaise M, Pateron D, Trinchet JC, et al. Systemic antibiotic
therapy prevents bacterial infection in cirrhotic patients with
gastrointestinal hemorrhage. Hepatology 1994;20:34 – 8.
18. Guidelines for Clinical Application. Infection control during
gastrointestinal endoscopy. ASGE Pub. no. 1033, Manchester,
1998.