2. surgical drainage first for patients with liver abscesses Ͼ5 cm.7
There is no consensus in the literature. Patients with liver
abscesses in Taiwan tend to be elderly, cryptogenic in origin,
and with complex underlying diseases.2,9
We must search for
specific guides for the most effective surgical treatment for
these patients.
The Acute Physiology and Chronic Health Evaluation II
(APACHE II) score can reflect the severity of some dis-
eases, such as acute pancreatitis or intra-abdominal ab-
scesses, and is well correlated with the associated mortality
rate.9–11
We also found that patients with APACHE II
scores Ն15 had significantly higher mortality rates.12
This
study aimed to delineate whether early aggressive hepatic
resection (HR) for patients with pyogenic liver abscesses
and APACHE II scores Ն15 would achieve better out-
comes.
Patients and Methods
From January 2001 to April 2006, 432 patients with
pyogenic liver abscesses were admitted into 2 medical cen-
ters: Tri-Service General Hospital and Yeezen General Hos-
pital, Taipei, Taiwan. Of these, 92 patients had APACHE II
scores Ն15. No liver transplant patients were included in
this study. A retrospective comparative study of non-
matched groups of patients defined by eventual intervention
(HR, percutaneous drainage [PD], or PD ϩ HR) was per-
formed.
Medical records were reviewed retrospectively to docu-
ment the results of the following: imaging studies (eg,
ultrasonography or computed tomography); bacteriologic
studies of aspirated pus; abscess patterns, origins, locations,
and sizes; underlying diseases; APACHE II scores11
; treat-
ments; infectious and procedure-related complications13
;
procedure failure and double-treatment rates; and hospital-
ization outcomes. PD was performed by the radiologists
using pigtail catheters under sonoguidance and fluoroguid-
ance. Most HRs were performed with the wedge resection
pattern, including 3 patients who underwent bilateral resec-
tion. Two patients underwent right lobectomies, and 1 pa-
tient underwent left lobectomy.
The therapeutic approaches of our patients are summa-
rized in Figure 1. During hospitalization, we tried to control
patients’ blood sugar levels within normal ranges and elim-
inate other possible infectious causes. All patients were
given empirical antibiotic therapy initially and were then
directed according to the culture results. If the abscesses
originated in the biliary tract, endoscopic retrograde cholan-
giopancreatography with papillotomy and stone removal
and/or endoscopic nasal biliary drainage were performed to
relieve the common bile duct obstruction. If these treat-
ments did not successfully cure the abscesses, PD or HR
was arranged. PD was applied first except in patients with
nonliquefied abscesses or abscess size Ͻ3 cm. HR was
recommended for patients who failed PD, or those with
suspect malignancy, biliary tract infection, abscess rupture–
induced peritonitis.
In-hospital mortality was defined as death caused by this
disease during hospitalization. Hospitalization outcomes in-
cluded durations of hospital stay, antibiotic use, and persis-
tent symptoms after intervention. All clinical data and
APACHE II scores were collected on days 1 and 2 before
PD or HR (Table 1). Failure of treatment was defined as
death or need for additional procedures related to the liver
abscess. Double treatment was defined as need for repeated
procedures related to progression or persistence of sepsis or
recurrence of liver abscess. Length of hospital stay was
defined as number of days from admission to discharge.
Figure 1 The scheme of treatment for pyogenic liver abscess.
The number in the parentheses expressed the number of mortality.
Table 1 Characteristics of patients with pyogenic liver
abscesses treated by PD, by PD ϩ HR, or by HR
Characteristics
PD and PD ϩ
HR (n ϭ 65) HR (n ϭ 16) P
APACHE II score 17.7 Ϯ 3.2 17.4 Ϯ 2.3 .7653
Age (y) 67.2 Ϯ 13.4 65.7 Ϯ 14.4 .6856
Men 38 10 .99
Women 27 6
DM history 41 10 .99
KP infection 48 11
Single bacterium 45 11 .99
Multiple bacteria 20 5
Biliary origin 13 8 .08
Cryptogenic origin 52 8 .37
Single abscess 40 6 .4653
Multiple abscesses 25 10 .3357
Size of abscess (cm) 17.2 Ϯ 2.8 6.4 Ϯ 3.6 .3719
Gas forming 14 1 .295
Single lobe 51 13 .99
Bilateral 14 3
DM history ϭ history of DM; KP infection ϭ Klebsiella pneumoniae
infection.
347H.-F. Hsieh et al. Liver abscess and APACHE II score
3. Length of hospital stay after drainage or surgery was de-
fined as number of days after intervention to day of dis-
charge. Duration of antibiotic use was defined as number of
days of antibiotic use during hospitalization; duration of
antibiotic use after drainage or surgery was defined like-
wise. Duration of persistent symptoms after intervention
was defined as number of days after intervention until pa-
tient body temperature decreased to Յ37.5°C.
Statistical analysis
Continuous numeric data are expressed in the tables as
means Ϯ SDs. The difference between the 3 patient groups
(see Results section) was analyzed by 1-way analysis of
variance (ANOVA). Student’s t and Fisher’s exact tests
were used to analyze differences between pairs of patient
groups or procedures. P Ͼ .05 was accepted as significant.
Results
Our study initially included 92 patients with APACHE II
scores Ն15. Eleven patients receiving conservative treatments
were excluded, and 9 of these died within 2 days. Eighty-one
patients were enrolled in the study and were allocated to 2
groups: (1) 65 patients who underwent PD, including 19 pa-
tients who underwent PD followed by HR (PD ϩ HR) and (2)
16 patients who underwent HR. Sixty-five patients underwent
drainage procedures, and 35 underwent surgery.
The characteristics of the 2 patient groups are listed in
Table 1. Mean APACHE II scores of the PD and PD ϩ HR
groups and the HR group were 17.7 and 17.4, respectively.
Mean ages of the 2 groups were 67.2 and 65.7 years, and
mean abscess sizes were 7.2 and 6.4 cm, respectively. There
were no significant differences in patient characteristics
between groups.
Table 2 lists infection-induced septic symptoms and com-
plications in the 2 groups. These symptoms included pleural
effusions, metabolic acidosis or alkalosis, acute respiratory
distress syndrome, acute renal failure, thrombocytopenia, dif-
fuse intravascular coagulopathy, multiple organ failure, septic
shock, and infections spreading to other organs. Distributions
of infection-related complications were not statistically differ-
ent between groups. The underlying diseases and conditions of
the 2 groups are listed in Table 3. These included advanced age
(Ͼ75 years), hemodialysis before hospitalization, end-stage
liver disease, chronic obstructive pulmonary disease, heart fail-
ure, biliary tract stones, history of intra-abdominal cancer,
upper gastrointestinal tract bleeding, stroke, long-term bed-
ridden state, and pancreatitis. No statistically significant differ-
ences in the distributions of underlying diseases were found
between the 2 groups.
To evaluate the effect of salvage HR on liver abscess
treatment, comparisons were made between the 3 groups (PD,
PD ϩ HR, and HR) for mortality rate, length of hospital stay,
antibiotic use, persistence of symptoms, and procedure-related
complications, all of which are listed in Table 4. Mortality rates
differed significantly between the 3 groups (14 of 46 for PD, 2
of 19 for PD ϩ HR, and 1 of 16 for HR; P ϭ .038) as well as
the lengths of hospital stay (39.6, 27.2, and 24.1 days respec-
tively; P ϭ .0453). Length of hospital stay after PD was
significantly longer than after HR (36.8 vs 19.5 days; P ϭ
.0296) but was similar to that in the PD ϩ HR group. Lengths
of hospital stay in the PD ϩ HR and HR-only groups were not
significantly different.
Duration of antibiotic use after PD was significantly
longer than after HR (29.4 vs 15.7 days; P ϭ .0082). There
were no significant differences between the PD and
PD ϩ HR groups or between the PD ϩ HR and HR-only
Table 2 Spectrum of infection-induced septic syndrome
and complications
Syndromes and
complications
PD and PD ϩ
HR (n ϭ 65) HR (n ϭ 16) P
Pleural effusion 25 5 .7918
Metabolic acidosis 17 4 .99
Metabolic alkalosis 3 1 .99
ARDS 17 5 .7696
Acute renal failure 7 2 .99
Thrombocytopenia 5 4 .1064
DIC 17 5 .7696
Infection metastasis 11 2 .99
MOF 12 0 .1186
Septic shock 27 6 .99
Ventilator support 15 3 .99
No complications 1 0 .99
1 complication 7 2 .99
Ն2 complications 57 14
ARDS ϭ acute respiratory distress syndrome; DIC ϭ diffuse intra-
vascular coagulopathy; MOF ϭ multiple organ failure.
Table 3 Underlying diseases or conditions
Disease or condition
PD and PD ϩ
HR (n ϭ 65) HR (n ϭ 16) P
Age Ͼ75 y 10 3 .7219
HD 10 2 .99
ESLD 5 2 .6282
COPD 2 0 .99
Heart failure 13 0 .1153
Biliary stones 13 5 .5253
History of intra-
abdominal cancer
4 0 .99
UGI bleeding 6 4 .2178
Stroke 11 0 .2014
Long-term bedridden 7 1 .99
Pancreatitis 1 0 .99
No underlying disease 1 0 .99
1 disease 16 5 .7632
Ն2 diseases 48 11 .99
HD ϭ hemodialysis; ESLD ϭ end-stage liver disease; COPD ϭ
chronic obstructive pulmonary disease; UGI ϭ upper gastrointestinal
tract.
348 The American Journal of Surgery, Vol 196, No 3, September 2008
4. groups regarding duration of antibiotic use. No significant
differences in duration of persistent symptoms between the
3 groups were found. Regarding procedure-related compli-
cations, there were significant differences between PD and
HR procedures in failure (33 of 65 vs 3 of 35, P ϭ .004) and
double-treatment rates (32 of 65 vs 0 of 35, P ϭ .0002).
Comments
According to the results of this retrospective study, pa-
tients undergoing HR for severe liver abscesses with
APACHE II scores Ն15 had better clinical outcomes than if
they only underwent PD. These included lower mortality,
procedure failure, and double-treatment rates, shorter hos-
pital stay, and shorter duration of antibiotic use.
According to the classic study by Ochsner et al in
1938,14
open surgical drainage or HR were the only recom-
mended curative treatments for patients with liver ab-
scesses. These were performed widely in the 1970s and
early 1980s.4,14–16
The mortality rate of patients with pyo-
genic liver abscesses was approximately 40% until the
1980s. Because of improvements in the availability of an-
tibiotic drugs and US-guided techniques for evaluating liver
abscesses, as well as the contention that surgical treatment
would manipulate the liver and cause life-threatening bac-
teremia,8,17
the treatment of patients with liver abscesses
shifted to a nonsurgical approach in the late 1980s. Surgical
treatment was only indicated in those patients who devel-
oped complications, such as perforation, peritonitis, and
drainage tube failure, or when their conditions worsened.18
Hepatic resection of liver abscess was deemed a salvage
therapy after medical or drainage tube failures.19
Mischinger et al used the APACHE II score as a phys-
iologic measure of illness severity.18
For patients with lower
APACHE II scores, it was recommended that US-guided
percutaneous treatment could be done initially to lower
surgical risks.18
However, some investigators noted that
patients with more complex abscesses developed abdominal
pathologies; therefore, they recommended that patients with
high APACHE II scores might be served better by surgical
exploration.9,10,20
In addition, the results of our previous
study12
confirmed that patients with pyogenic liver ab-
scesses and high APACHE II scores (Ն15) at admission had
higher risk of in-hospital mortality.
With the advancement of modern surgical techniques,
HR is no longer a dangerous procedure.21,22
With this in
mind, the role of HR in patients with severe pyogenic liver
abscesses should be reconsidered. HR could achieve total
eradication of pathogens from the abscess region and in-
flamed liver tissue. It could achieve a better outcome for
patients with severe liver abscesses and infectious compli-
cations. In fact, many patients with pyogenic liver abscesses
have shown dramatic improvements after surgical interven-
tion. Serious infectious complications, such as renal failure,
respiratory failure, or disturbed consciousness, improved
within 2 days after HR. In our series, HR for patients with
liver abscesses proved an effective treatment, with a lower
failure rate, as well as a decreased need for repeated pro-
cedures, than PD. Durations of hospital stay and antibiotic
use after HR were shorter than after PD. Mortality rate after
HR was also significantly lower than after PD. Our results
are similar to those of Herman et al,23
as well those of as
other studies,7,23,24
confirming the improved results of sur-
gical drainage for patients with liver abscesses.
Because of the geographic distribution and characteris-
tics of our patient population, most etiologies of our liver
abscesses were cryptogenic. Our results and treatment al-
gorithm are consistent with other groups in Taiwan.2,9,17
In general, the PD procedure produced fewer complica-
tions in patients with single, liquefied liver abscesses and
without septic syndromes. In patients with less severe liver
Table 4 Comparisons between mortality rates, durations of hospital stay and antibiotic use, persistence of symptoms, and
procedure-related complications
Rates, symptoms, and complications PD PD ϩ HR HR P a
Mortality rate 14 of 46 2 of 19 1 of 16 .038
Hospital stay (d) 39.6 Ϯ 29.2 27.2 Ϯ 15.5 24.1 Ϯ 7.3 .0453
After PD 36.8 Ϯ 29.3 26.2 Ϯ 15.4 .1713 .0296
After surgery 19.4 Ϯ 13.5 19.5 Ϯ 7.0 .9768
Antibiotic use (d)
After PD 29.4 Ϯ 18.7 21.0 Ϯ 15.5 .1188 .0082
After surgery 14.2 Ϯ 12.8 15.7 Ϯ 5.6 .6774
Persistence of symptoms (d)
After PD 18.6 Ϯ 19.9 14.2 Ϯ 8.9 .3926 .0631
After surgery 7.4 Ϯ 6.6 8.3 Ϯ 8.5 .7174
Procedure-related complications
Failure 33 of 65 3 of 35 .004
Double treatment 32 of 65 0 of 35 .0002
Hospital stay ϭ time from admission to discharge; hospital stay after PD ϭ time from first day of drainage to day of discharge; hospital stay after
surgery ϭ time from surgery to day of discharge.
a
Comparison between PD and surgery groups by Student t test.
349H.-F. Hsieh et al. Liver abscess and APACHE II score
5. abscesses, a combination of PD and antibiotic treatment
could achieve the therapeutic goal easily. However, in pa-
tients with more severe liver abscesses, the PD procedure
alone could not eradicate inflammatory hepatic cells and
necrotic tissue. As a result, the infectious symptoms might
progress to sepsis or death. Since the late 1980s, surgical
drainage of liver abscesses has been considered a salvage
therapy because of high mortality and complication rates. In
this study, the complication rates of PD and HR were
equivalent. Therefore, the role of surgical treatment for liver
abscesses, especially in severe cases, should be re-evalu-
ated. Tan et al concluded that when a liver abscess is Ͼ5 cm
in diameter, surgical drainage should be a first-line
strategy.7
However, high APACHE II score, rather than
abscess size, is a risk factor for mortality in such patients.12
Severe liver abscess will be indicated by APACHE II scores
Ն15, and such patients would have no time to complete the
whole treatment course of advanced broad-spectrum antibi-
otics or US-guided tube drainage. In these patients, HR is
beneficial because it can completely remove necrotic tissue
and inflammatory hepatic cells. In our series, patients with
severe pyogenic liver abscesses and APACHE II scores
Ն15 treated by HR had a significantly lower mortality rate
than patients treated by PD (P ϭ .038).
In patients who underwent PD followed by HR for treat-
ment of severe liver abscesses had longer hospital stays,
more prolonged antibiotic use, and higher complication
rates. Differences between the PD ϩ HR group and those
who underwent only 1 procedure were significant (Table 4).
For clinical application, we recommend that patients with
severely pyogenic liver abscesses (APACHE II scores Ն15)
should receive HR as a first-line treatment.
The treatment strategy in this study was for PD to be
applied first. Patients who failed PD or those with biliary
tract infection, peritonitis, or suspect malignancy were rec-
ommended to undergo HR. Therefore, there may have been
some patient selection bias, but the clinical characteristics
showed no statistical differences between treatment modal-
ities. In conclusion, patients with pyogenic liver abscesses
and APACHE II scores Ն15, or those with increasing
APACHE II scores after tube drainage, should be treated
with surgical intervention to decrease risk of death.
References
1. Jepsen P, Vilstrup H, Schonheyder HC, et al. A nationwide study of
the incidence and 30-day mortality rate of pyogenic liver abscesses in
Denmark, 1977–2002. Aliment Pharmacol Ther 2005;21:1185–8.
2. Wang JH, Liu YC, Lee SSJ, et al. Primary liver abscess due to
Klebsiella pneumoniae in Taiwan. Clin Infect Dis 1998;26:1434–8.
3. Cheng HP, Siu LK, Chang FY. Extended-spectrum cephalosporin
compared to cefazolin for treatment of Klebsiella pneumoniae-caused
liver abscess. Antimicrob Agent Chemother 2003;47:2088–92.
4. Chu KM, Fan ST, Lai ECS, et al. Pyogenic liver abscess: an audit of
experience over the past decade. Arch Surg 1996;131:148–52.
5. Chou FF, Sheen-Chen SM, Chen YS, et al. Prognostic factors for
pyogenic abscess of the liver. J Am Coll Surg 1994;179:727–32.
6. Kandel G, Marcon NE. Pyogenic liver abscess: new concepts of an old
disease. Am J Gastroenterol 1984;79:217–26.
7. Tan YM, Chung YF, Chow KH, et al. An appraisal of surgical and
percutaneous drainage for pyogenic liver abscesses larger than 5 cm.
Ann Surg 2005;241:485–90.
8. Huang CJ, Pitt HA, Lipsett PA, et al. Pyogenic hepatic abscess: change
trends over 42 years. Ann Surg 1996;223:600–9.
9. Chou FF, Sheen-Chen SM, Chen YS, et al. Single and multiple
pyogenic liver abscesses: clinical course, etiology and results of treat-
ment. World J Surg 1997;21:384–9.
10. Levison MA, Zeigler D, Michigan D. Correlation of APACHE II
score, drainage technique and outcome in postoperative intra-abdom-
inal abscess. Surg Gynecol Obset 1991;172:89–94.
11. Knaus WA, Draper EA, Wagner DP, et al. APACHE II: a severity of
disease classification. Crit Care Med 1985;13:818–29.
12. Hsieh CB, Tzao C, Yu CY, et al. APACHE II score predicts hospital
mortality in patients with pyogenic liver abscess. Dig Liver Dis 2006;
38:498–502.
13. Dindo D, Demartines N, Clavien PA. Classification of surgical com-
plications: a new proposal with evaluation in a cohort of 6336 patients
and results of a survey. Ann Surg 2004;240:205–13.
14. Ochsner A, DeBakey M, Murray S. Pyogenic abscess of the liver: II.
An analysis of forty-seven cases with review of the literature. Am J
Surg 1938;40:292–319.
15. Pitt HA, Zuidema GD. Factors influencing mortality in the treatment of
pyogenic hepatic abscess. Surg Gynecol Obstet 1975;140:228–34.
16. Balasegaram M. Management of hepatic abscess. Curr Probl Surg
1981;18:282–340.
17. Chang FY, Chou MY. Comparison of pyogenic liver abscesses caused
by Klebsiella pneumoniae and non-K. pneumoniae pathogens. J For-
mos Med Assoc 1995;94:232–7.
18. Mischinger HJ, Hauser H, Rabl H, et al. Pyogenic liver abscess:
studies of therapy and analysis of risk factors. World J Surg 1994;18:
852–8.
19. Klatchko BA, Schwartz SI. Diagnostic and therapeutic approaches to
pyogenic abscess of the liver. Surg Gynecol Obstet 1989;168:332–6.
20. van Sonnenberg E, Wittich GR, Goodacre BW, et al. Percutaneous
abscess drainage: update. World J Surg 2001;25:362–72.
21. Imamura H, Seyama Y, Kokudo N, et al. One thousand fifty-six
hepatectomies without mortality in 8 years. Arch Surg 2003;138:
1198–1206.
22. Fan ST, Lo CM, Liu CL, et al. Hepatectomy for hepatocellular carci-
noma: toward zero hospital deaths. Ann Surg 1999;229:322–30.
23. Herman P, Pugliese V, Montagnini AL, et al. Pyogenic liver abscess:
the role of surgical treatment. Int Surg 1997;82:98–101.
24. Farges O, Leese T, Bismuth H. Pyogenic liver abscess: an improve-
ment in prognosis. Br J Surg 1988;75:862–5.
350 The American Journal of Surgery, Vol 196, No 3, September 2008