1) Citrobacter bacteremia most commonly occurred in patients with malignancies (48.9%) or hepatobiliary stones (22.2%). The primary sites of infection were commonly the abdominal cavity (51.1%), urinary tract (20%), and lung (11.1%). 2) Polymicrobial bacteremia was diagnosed in 33.3% of patients and the source was often intraabdominal (60% of polymicrobial cases). 3) Prior treatment with third-generation cephalosporins was significantly associated with developing multidrug resistant isolates. The mortality rate of citrobacter bacteremia was 17.8%.

1. 543
Bacteremia Due to Citrobacter Species: Significance of Primary Intraabdominal
Infection
Chiang-Ching Shih,* Yee-Chun Chen,
Shan-Chwen Chang, Kwen-Tay Luh,
and Wei-Chuan Hsieh
From the Section ofInfectious Disease, Department of Internal
Medicine and the Department ofLaboratory Medicine, National Taiwan
University Hospital, Taipei, Taiwan. Republic of China
From 1982 to 1994, 45 patients (1.22 episodes per 10,000 discharged patients) were treated for
citrobacter bacteremia at National Taiwan University Hospital (Taipei). All patients had at least one
underlying disease. Citrobacter bacteremia most commonly occurred in patients with malignancies
(48.9%) or hepatobiliary stones (22.2%). Intraabdominal tumors comprised the majority (59.1%) of
malignancies. Bacteremia commonly originated from sites such as the abdominal cavity (51.1%), urinary
tract (20%), and lung (11.1%). Polymicrobial bacteremia was diagnosed in 15 patients (33.3%);for nine
(60%) of these patients, the source of the infection was intraabdominal. Prior treatment with a third-
generation cephalosporin was significantly associated (P < .01) with the development of multidrug
resistance among the isolates. The mortality associated with citrobacter bacteremia was 17.8%. Poor
prognostic factors included pneumonia, altered mental status on presentation, hypothermia, oliguria,
septic shock, deterioration in mental status, hyperbilirubinemia, azotemia, and thrombocytopenia. Com-
bination therapy, as compared with other regimens, improved the outcome of citrobacter bacteremia.
Citrobacter species are aerobic, gram-negative bacilli com-
monly found in water, soil, food, and the intestinal tracts of
animals and humans [I]. These organisms cause a wide spec-
trum ofinfections in the urinary tract, respiratory tract, wounds,
bone, peritoneum, endocardium, meninges, and intestines [1].
Citrobacter bacteremia is a rare infection; we are aware of
only two reported series in the English-language literature
[2, 3]. Therefore, little is known about citrobacter bacteremia
in terms of incidence, associated underlying diseases, primary
sites of infection, and outcome. Although differences between
Citrobacter freundii and Citrobacter divers us in terms of anti-
microbial susceptibility have been cited [3-6], that these differ-
ences exist when these organisms are the cause of bacteremia
is unclear. Citrobacter has been reported to be frequently asso-
ciated with polymicrobial bacteremia [3], but there are no data
that explain this phenomenon.
After the the third-generation cephalosporins were intro-
duced, multidrug resistant strains of Enterobacter emerged as
a cause of bacteremia [7], but no data are available on Citro-
bacter species. We review our experience with citrobacter bac-
teremia over a B-year period and compare it with that pre-
viously reported in the literature [2, 3].
Patients and Methods
We reviewed the hospital records from 1 January 1982
through 31 December 1994 for all patients whose blood cul-
Received 18 January 1996; revised 26 March 1996.
*Current address: Koo Foundation Sun Yat-Sen Cancer Center, Taipei,
Taiwan, Republic of China.
Reprints or correspondence: Dr. Shan-Chwen Chang, Department ofIntemal
Medicine, National Taiwan University Hospital, No.7, Chung-Shan South
Road, Taipei, Taiwan, Republic of China.
Clinical Infectious Diseases 1996;23:543-9
© 1996 by The University of Chicago. All rights reserved.
1058--4838/96/2303-0020$02.00
tures yielded Citrobacter species at the National Taiwan Uni-
versity Hospital (Taipei), a major teaching hospital in Taiwan;
this hospital had 1,200 beds before 1991 and 1,500 beds after
1991. The blood culture medium used to grow Citrobacter
had been changed from trypticase soy broth containing sodium
polyanetholesulfonate and modified Lombard-Dowell broth to
the Bactec 6A broth and the Bactec 7A broth (Becton Dickin-
son, Sparks, MD) in 1987. Citrobacter species were identified
according to standard laboratory methods. Antimicrobial sus-
ceptibility tests were done by means of the Kirby-Bauer disk
diffusion method [8, 9]. Multidrug-resistance was defined as
resistance in vitro to the extended-spectrum penicillins (ticarcil-
lin, ticarcillinlclavulanate, and piperacillin) and the third-gener-
ation cephalosporins (cefotaxime, ceftazidime, ceftizoxime,
ceftriaxone, and cefoperazone) [7].
A patient was considered to have citrobacter bacteremia
when this organism was isolated from blood cultures on at least
one occasion. The primary site of infection was determined on
the basis of a clinical picture that was consistent with the
laboratory data and/or by a culture of tissue that was positive
for a Citrobacter species. If none of these findings was present,
the origin of the bacteremia was deemed unknown. Bacteremia
was defined as nosocomial if infections were acquired during
treatment at the study hospital, at other hospitals before transfer
to the study hospital, or during out-patient clinic visits or emer-
gency room visits. Otherwise, the bacteremia was considered
to be community acquired.
The empirical treatment was considered appropriate if all
organisms cultured were found to be susceptible to the drug(s)
during in vitro susceptibility testing. Treatment was defined as
delayed if no appropriate treatment was begun within 48 hours
after blood for cultures was drawn. Septic shock was defined
as the septic syndrome, with a systolic blood pressure of <90
mm Hg or a drop in the mean arterial pressure of >40 mm
Hg from the baseline, in the absence of other causes of hypoten-

2. 544 Shih et al. em 1996;23 (September)
sion. Outcome was evaluated at discharge or 1 month after
treatment was started. Death was considered bacteremia-related
if patients died within 10 days of the report of positive culture
results. Patients who died of other conditions and had obvious
initial clinical responses after receiving antibiotic treatment
were excluded from the study.
Statistical analysis was performed with use of the X2
test
and a two-tailed Fisher's exact test. The odds ratios and 95%
confidence intervals were calculated at the same time. The logit
estimators used a correction of 0.5 in every cell ofthose labels
that contained a zero.
Results
From 1982 through 1994, 56 blood cultures for 45 patients
yielded Citrobacter species, and these isolates accounted for
0.55% of the 10,263 blood isolates recovered at our hospital
during this period. Twenty-two patients (48.9%) were male,
and 23 patients (51.1%) were female. The ages ranged from 2
days to 86 years (median age, 60 years). Eight patients (17.8%)
were < 18 years of age.
Of the 45 episodes of bacteremia, 21 occurred during the
first 6.5 years of the study period (1.28 episodes per 10,000
patients discharged) and the remaining 24 episodes occurred
in the second period (1.18 episodes per 10,000 patients dis-
charged). No significant change in the incidence of citrobacter
bacteremia was found during this 13-year period. Throughout
the entire study period, recovery of Citrobacter isolates from
blood appeared to be random, since the isolates were not clus-
tered by season or in outbreaks. Twenty-three episodes (51.1%)
were community acquired, and 22 (48.9%) were nosocomial.
Underlying diseases. The patients' underlying diseases are
listed in table 1. Of those who had a malignancy, five had
leukemia, and 16had solid tumors (including 13 of intraabdom-
inal origin). The number of cases of citrobacter bacteremia per
10,000 newly registered cancer patients in our hospital was
49.5 for those with acute leukemia, 37.8 for those with chronic
leukemia, 6.3 for those with solid tumors, and 19.9 for those
with intraabdominal tumors (table 1). Eight patients had re-
ceived chemotherapy, and five of them were neutropenic (neu-
trophil count, <500/mm3
) when citrobacter bacteremia devel-
oped.
Primary sites of infection. The primary sites of infection
and the numbers ofpatients who presented with each are listed
in table 1. Intraabdominal infections (23 patients) included
hepatobiliary tree infections (19, including three with liver ab-
scesses), peritonitis (three), and perianal abscess (one). Ofthese
23 patients, 20 (87%) had underlying intraabdominal lesions
including hepatobiliary stones (10 patients), malignancy with
biliary obstruction (nine), and pancreatitis (one). Citrobacter
bacteremia of intraabdominal origin was often associated with
underlying intraabdominal pathology.
Eight (88.9%) of nine patients with urinary tract infectionshad
urinary tract abnormalities. These abnormalities included neuro-
genicbladder, retroperitonealfibrosisand hydronephrosis, chronic
cystitis, cervical cancer and hydronephrosis, rectal cancer and
hydronephrosis, rectal cancer with a uroanal fistula, transitional
cell carcinoma, and benign prostate hypertrophy with urethral
stricture. The bacteremias that originated from the urinary tract
were also frequently associated with local lesions.
Five patients had pneumonia, which was associated (in order
of frequency) with prematurity, congestive heart failure,
lymphoma, colon cancer, and congenital heart disease with
asplenism.
Ofthree patients who had soft tissue infections and/or wound
infections, one had cellulitis at the irradiated site of nasopha-
ryngeal carcinoma, and one had a skin ulcer caused by extrava-
sation with vincristine. The other patient had chronic lympho-
cytic leukemia with cellulitis of the leg but no obvious
cutaneous breakdown.
One patient had citrobacter bacteremia secondary to gouty
arthritis, and one premature neonate had citrobacter meningitis.
Nineteen of 42 patients with identified origins of infection
had Citrobacter species isolated from the primary sites ofinfec-
tion. The specimens included bile (four patients), liver abscess
(two), gall bladder pus (two), discharge from peritoneal cavity
(one), perianal abscess (one), urine (five), lung tissue (one),
sputum (one), CSF (one), and wound discharge (one) (table 1).
Three (6.7%) of the 45 patients presented with signs of
citrobacter sepsis that had no identifiable origin. Two of these
patients were children; one had acute lymphocytic leukemia,
and the other had chemotherapy-induced leukopenia. The third
patient, a 56-year-old male, had idiopathic segmental axial
dystonia and underwent intermittent urinary catheterization;
however, the results of his urinalysis were normal.
Fifteen patients for whom a hepatobiliary origin of infection
was documented had community-acquired bacteremia. Six of
nine patients with urinary tract infections acquired the infec-
tions in the community; the other three had nosocomial infec-
tions, and all ofthem had undergone urinary tract manipulation
in the hospital.
Initial clinical manifestations ofbacteremia. Fever was the
most common initial manifestion of bacteremia in these patients.
Thirty-nine (86.7%) of the 45 patients had fever, two had hypo-
thermia, and four had normal temperatures. Thirty-five febrile
patients (77.8%) had chills. Fifteen patients (33.3%) were hypo-
tensive. Eight (18.2%) of 44 patients had altered mental status,
and 13 (32.5%) of 40 had oliguria. Five (12.5%) of 40 had
cough, and 26 (61.9%) of 42 had abdominal pain, ileus, and/
or gastrointestinal bleeding; of these patients, 12 had jaundice.
Twenty-eight (65.1%) of 43 patients had leukocytosis (WBC
count, > 1O,000/mm3
) , and five (11.4%) of 44 had neutropenia
(neutrophil count, <500/mm3
) due to prior chemotherapy for
cancer.
Complications. Among the 45 patients, the most frequent
complication was septic shock, which was present in 15
(33.3%) of the patients. Liver dysfunction was present in 15
(33.3%) of the patients; respiratory failure, in 13 (28.9%); dete-
riorated mental status, in 12 (26.7%); renal dysfunction, in 11
(24.4%); and thrombocytopenia, in 7 (15.6%).

3. em 1996;23 (September) Citrobacter Bacteremia 545
Table 1. Summary of data from 45 cases of citrobacter bacteremia in Taiwan.
No. of patients with indicated primary site of infection
Lower
Intraabdominal Urinary respiratory Total no. of cases per 10,000
Variable tissues tract tract Others* Unknown Total newly registered patients
Underlying disease
Malignancy 10 2 3 2 22 5.27
Solid tumor 9 5 I I a 6 6.3
Intraabdominal tumor 9 3 1 a 0 13 19.9
Hematological tumor I 0 1 2 2 6 26.3
Acute leukemia I a 0 I 2 4 49.5
Chronic leukemia 0 0 0 I 0 I 37.S
Hepatobiliary stone 10 0 0 0 0 10
Heart disease 2 I 2 0 1 6
Diabetes mellitus 1 4 0 0 0 S
Plaee of acquisition
Community 1St 6 I I 0 23
Hospital 8 3 4 4 3 22
Etiologic organism
C.jreundii 11 3 1 2 1R
C. diversus 2 3 0 1 7
Citrobacter species plus other organisms 9 2 3 0 15
NOTE. The number of patients with culture-proven primary sites of infection were as follows: intraabdominal tissues, 10 of 23; urinary tract,S of 9; lower
respiratory tract, 2 of 5; others, 2 of 5; and unknown, 0 of 3.
*Includes bone and joint infection (n = 1), CNS infection (1), and wound and soft-tissue infection (3).
I All IS patients' infections originated in the hepatobiliary tract.
Bacteriology. Ofthe 45 episodes ofcitrobacter bacteremia,
18 (40%) were due to Cifrcundii, and seven (15.6%) were due
to C diversus. The Citrobacter isolates were not identified to
the species level in the remaining cases. C freundii was a
more frequent cause ofbacteremia than was C diversus among
patients with infections of intraabdominal origin (table 1).
Citrobacter was isolated in association with other bacteria
(most frequently gram-negative bacilli) from 15 of the 45 patients.
Other concomitantly isolated bacteria included Escherichia coli
(six patients), Klebsiella pneumoniae(four), Bacteroides species
(three), Enterococcus species (three), Aeromonas hydrophila
(two), Morganella morganii (two), and Proteus vulgaris (one).
Polymicrobial infection was more frequent in patients with an
intraabdominal origin of infection (nine patients), a community-
acquired infection (nine), or a malignancy (eight, including six
with intraabdominal malignancies).
Antimicrobial susceptibilities. Antimicrobial susceptibility
patterns were analyzed for 44 strains (these data were not avail-
able for one strain). Resistance to ampicillin, cefazolin, and
cefamandole was common. However, almost all of the strains
tested were susceptible to gentamicin (table 2).
The percentage of C freundii isolates that were resistant to
ampicillin, cefazolin, cefamandole, and cefotaxime was higher
than the percentage of C diversus isolates that were resistant
to these drugs.
Treatment with first- or second-generation cephalosporins be-
fore the onset of bacteremia resulted in an increase in the rate of
resistance to ampicillin, cefazolin, and cefamandole but not to
cefotaxime. On the other hand, pretreatment with third-generation
cephalosporins resulted in an increase in the rate of resistance to
cefotaxime. Multidrug resistance was found in five strains and
was associated significantly (P < .01) with pretreatment with
third-generation cephalosporins. All four multidrug-resistant
strains that were tested were susceptible to ciprofloxacin.
Treatment. Thirty-five patients received appropriate antibi-
otic treatment. Four patients did not receive appropriate treat-
ment within 48 hours of the onset of bacteremia, and five did
not receive any effective medical treatment. Two of these nine
patients died of bacteremia. The appropriateness of the treat-
ment one patient received could not be judged because the
drug susceptibilities of his isolates were not determined.
The 45 patients received one or more of the following anti-
biotics: penicillins (two patients [4.4%]); first- or second-
generation cephalosporins (25 [55.6%]); third-generation ceph-
alosporins (17 [37.8%]); fluoroquinolones (two [4.4%]);
nitrofurantoin (one [2.2%J);and aminoglycosides (20 [44.4%D.
Eighteen patients received combination therapy with an amino-
glycoside and a ,B-lactam. Of the 18 patients who received
combination therapy, only one (5.6%) died, whereas five
(45.5%) of II patients who received monotherapy with a third-
generation cephalosporin died; thus combination therapy was
significantly superior to monotherapy with a third-generation
cephalosporin (OR = 0.07; 95% CI = 0.01-0.73; P = .018).
When compared with all other single-agent regimens, combina-
tion therapy was found to be more protective, although this
difference was not significant (P = .11) (table 3).

4. 546 Shih et al. em 1996; 23 (September)
Table 2. Rates of antimicrobial resistance and significance of factors influencing the antimicrobial susceptibilities of Citrobacter species
causing bacteremia in patients in Taiwan.
No. of indicated species No. of resistant isolates/no.
tested/no. of resistant strains of isolates from patients
(%) pretreated with a first- or
No. of resistant strains/no. second-generation
Antimicrobial agent of strains tested (%) C.freundii C. diversus OR* 95% CI* cephalosporin (%)
Ampicillin 33/43 (76.7) 16/18 (88.9) 5/7 (71.1) 3.2 0.35-28.95 10/10 (100)
Cefazolin 26/44 (59.1) 13/18 (72.2) 1/7 (l4.3)t 15.6 1.48-164.38 7110 (70)
Cefamandole 8/33 (24.4) 8118 (44.4) 1/7 (14.3) 4.48 0.48-48.46 5/7 (71.4)
Cefotaxime 7/41 (17.1) 5/17 (29.2) 0/7 6.58 0.32-142.86 2/10 (20)
Ciprofloxacin 0117
Imipenem 0/13
Gentamicin 2/43 (4.7) 0117 0/7 0/9
Multiple agents 5/41 (12.1) 3117 (17.7) 0/7 3.62 0.16-76.92 2/10 (20)
* Determined by means of the X2
test and two-tailed Fisher's exact test.
t p = .01-.05.
t P = .05-0.1.
§ P = .001-.01.
Surgical procedures or drainage were performed in 10 pa-
tients.
Outcome. Fifteen (33.3%) of the 45 patients died. Six of
these 15patients died ofcauses other than bacteremia, although
they responded well to treatment of bacteremia. One other
patient died of hepatic failure that was associated with gastric
cancer, and the bacteremia probably contributed to his death.
Eight (17.8%) of the 45 patients died of bacteremia. Table 3
lists potential risk factors for death due to citrobacter bacter-
emia. The initial manifestations that were significant risk fac-
tors included pneumonia, altered mental status, hypothermia,
and oliguria, and complications during the course of the illness
that were significant included septic shock, further deterioration
in mental status, hyperbilirubinemia, hypercreatinemia, and
thrombocytopenia. Polymicrobial bacteremia and alcoholism
were also associated with an increase in mortality, but this
increase was not significant statistically.
Factors such as old age, cold weather, place of acquisition,
the primary site and/or manifestation of the infection (except
pneumonia), antibiotic resistance, the initial presence of hypo-
tension, the leukocyte count, chemotherapy, previous invasive
procedure, pretreatment, delayed treatment, or no treatment
did not have significant influence on mortality. Appropriate
treatment did not result in lower mortality. Although infection
that originated in the urinary tract was associated with lower
mortality, the difference was not significant. Surgical interven-
tion and combination therapy were associated with a protective
effect.
Discussion
Citrobacter species have been reported as a cause of many
kinds of human infections [1, 6, 10-13], but bacteremia due
to these organisms remains uncommon. The incidence of Citro-
bacter bacteremia among our patients was similar to that re-
ported by Drelichman and Band [3].
The urinary tract was the leading site of citrobacter infection
in many previous reports [1-3, 14], including the two that
described citrobacter bacteremia [2, 3]. However, in our series,
intraabdominal tissues (mainly in hepatobiliary system) were
the most common primary sites of infection in bacteremic pa-
tients. The reason that such sites predominated in our series
was that a large portion of our patients had hepatobiliary stones
(22.2%) and intraabdominal malignancies (28.8%). Hepatobili-
ary infection was the most frequent (82.6% ofpatients) intraab-
dominal infection due to Citrobacter species, which is consis-
tent with the findings of Lew et al. [15]. We emphasize that
enterococci, E. coli, and anaerobes still predominate among
patients with the pancreatic and hepatobiliary cancer and in-
traabdominal abscesses [16], and antimicrobial coverage for
these organisms should be considered first.
According to previous reports [1-3, 14], most citrobacter
infections have been hospital acquired. In our study, about one-
half of the cases (51.1%) were community acquired, a finding
that may be due to the predominance of cases hepatobiliary
infection (19) in our study. One large-scale study [17] showed
that hospital-acquired cases ofbacteremia predominated among
patients with infections that originated from any site other than
the biliary tree and reproductive tract. Fifteen (78.9%) of the
19 patients with hepatobiliary infection in our study had com-
munity-acquired bacteremia.
In one previous report of citrobacter bacteremia in patients
with cancer [2], those with acute leukemia accounted for the
highest number with bacteremia due to Citrobacter species
alone (this number was 20 times higher than the number of
patients with solid tumors and citrobacter bacteremia). Al-
though patients with acute leukemia still had the highest rate
of citrobacter bacteremia in our study, those with tumors of

5. cm 1996;23 (September) Citrobacter Bacteremia 547
Table 2. (Continued)
No. of resistant isolates/no.
of isolates from patients No. of resistant isolates/no. No. of resistant isolates/no.
without pretreatment with of isolates from patients of isolates from patients
a first- or second- pretreated with a third- without pretreatment with
generation cephalosporin generation cephalosporin a third-generation
(%) OR* 95% CI* (%) cephalosporin (%) OR* 95% CI*
23/33 (69.7); 9.35 0.5-166.67 6/6 (l00) 27/37 (73.0) 4.98 0.26-10
19/34 (35.9) 1.84 0.41-8.36 6/6 (l00) 20/38 (52.6)t 11.76 0.62-250
7/26 (26.9)t 6.79 1.06-43.36 4/5 (80) 8/28 (28.6); 10 0.96-103.78
5/31 (l6.1) 1.3 0.21-8.03 4/6 (66.7) 3/35 (8.6)§ 21.33 2.69-168.94
2/34 (5.9) 0.68 0.03-45.51 0/6 2/37 (5.4) 1.09 0.05-25.48
3/31 (9.7) 2.06 0.33-16.47 3/6 (50) 2/35 (5.7); 16.5 1.93-140.85
the abdominal cavity were also found to have a high incidence
(10.7 cases per 10,000 newly registered patients) of the infec-
tion. Furthermore, the incidence among such patients was even
higher ifcases of polymicrobial bacteremia were included (19.9
cases per 10,000 newly registered patients). We emphasize the
importance of underlying intraabdominal tumors in the devel-
opment of Citrobacter bacteremia among our patients, since
this observation has not been made previously.
The incidence (33.3%) of bacteremia due to Citrobacter,
in combination with other organisms, in our hospital is similar
to that (35%-46.1 %) observed by other investigators [3, 15]
and is higher than that for bacteremias due to all organisms
(6%-17.8%) or bacteremias due to gram-negative organisms
(4%-25%) [17-23]. Isolation of Citrobacter as a part of a
mixed infection in the abdominal cavity was unexpectedly
common (nine patients) in our study. These cases presumably
represented the introduction of Citrobacter species that were
already present in the patients' gastrointestinal tracts; this
phenomenon has been mentioned in previous reports [1, 3].
One other important finding in our study was that administra-
tion of a third-generation cephalosporin within 14 days before
the onset of citrobacter bacteremia had a significant influence
on the selection of cefotaxime-resistant strains (P = .005) and
multidrug-resistant strains (P = .017). This finding confirmed
the fact that multidrug-resistant organisms may emerge more
rapidly when the third-generation cephalosporins are used rou-
tinely, as has been predicted by other investigators [7]. The
difference between C. freundii and C. diversus in terms of
susceptibility to the cephalosporins has been noted since the
1970s [2, 4-6] and has been confirmed again in this study of
strains that cause bacteremia.
Combination therapy had a protective effect in our study.
This benefit of combination therapy has been proposed by other
authors [2, 7]. Samonis et al. [24] reported that imipenem,
amikacin, and the new fluoroquinolones had good activity
against Citrobacter species. Our results are compatible with
their findings. Because the data are limited, we suggest the
use of combination therapy for initial empirical treatment of
citrobacter bacteremia, and the fluoroquinolones can be used
for the treatment of episodes due to multidrug-resistant strains.
However, further studies are needed to support this recommen-
dation.
The overall mortality associated with citrobacter bacter-
emia was 33.3% in our series; this percentage is lower than
that (48.3%) reported by Drelichman et al. in 1985 [3]. The
mortality associated with citrobacter bacteremia is similar to
that for bacteremia due to Klebsiella species (37%) [25],
Enterobacter species (20%) [7], Proteus mirabilis (29.0%)
[21], and bacteremias due to gram-negative organisms (25%)
[4] but higher than that associated with E. coli bacteremia
(10%) [22]. In previous reports [2, 7, 19,21-23,25-27],
many risk factors including the two extremes ofage, pneumo-
nia, sources of bacteremia other than the urinary tract, alco-
holism, diabetes mellitus, congestive heart failure, infection
with a multidrug-resistant strain, inappropriate treatment,
respiratory tract infection, polymicrobial bacteremia, nosoco-
mial infection, chemotherapy-induced neutropenia, leu-
kocytosis, septic shock, azotemia, hyperbilirubinemia, and
thrombocytopenia have all been significantly associated with
death due to gram-negative bacteremia. In our series pneumo-
nia, altered mental status, oliguria, septic shock, deterioration
in mental status, azotemia, hyperbilirubinemia, and thrombo-
cytopenia were found to have a significant influence on mor-
tality. The fact that other factors were not significant was
probably due to the smaller number of cases in our study.
In conclusion, citrobacter bacteremia is uncommon and
usually develops in patients with underlying diseases.
In our series, about one-half of cases were community

6. 548 Shih et al. em 1996;23 (September)
Table 3. Risk factors for death due to citrobacter bacteremia.
No. of patients who died! No. of patients who died!
no. of patients with risk no. of patients without
Risk factor factor (%) risk factor (%) aRt CIt P value
Enrollment during first 6.5 years of study* 6/21 (28.6) 2/24 (8.3) 4.40 0.78-24.81 NS
Occurrence of bacteremia during months
of November-February 3/14 (21.4) 5/31 (16.1) 1.42 0.29-6.99 NS
Age ;0.65 years 2/15 (13.3) 6/30 (20) 0.62 0.11-3.50 NS
Male sex 5/22 (22.7) 3/23 (13.0) 1.96 0.41-9.43 NS
Presence of underlying condition
Malignancy 3/22 (13.6) 5/23 (21.7) 0.57 0.12-2.73 NS
Hematologic tumor 1/6 (16.7) 2/16 (12.5) 1.4 0.10-19.01 NS
Other conditions
Intraabdominal lesions 2/11 (18.2) 6/34 (17.6) 1.04 0.18-6.07 NS
Diabetes mellitus 0/5 8/40 (20) 0.35 0.02-6.94 NS
Heart disease 1/6 (16.7) 7/39 (17.9) 0.91 0.09-9.10 NS
Delayed admission 1/11 (9.1) 7/34 (20.6) 0.39 0.04-3.54 NS
Chemotherapy 0/8 8/37 (21.6) 0.2 0.01-3.91 NS
Steroid therapy 2/7 (28.6) 6/38 (15.8) 2.13 0.33-13.67 NS
Alcoholism 2/3 (66.7) 6/42 (14.3) 12.33 0.96-158.08 NS
Invasive procedure 1/15 (6.7) 7/30 (23.3) 0.23 0.03-2.11 NS
Prior antibiotic treatment 3/17 (17.6) 5/28 (17.9) 0.99 0.20-4.78 NS
Polymicrobial bacteremia 4/15 (26.7) 4/30 (13.3) 2.36 0.50-11.19 NS
Nosocomial acquisition of bacteremia 4/22 (18.1) 4/23 (17.4) 1.05 0.21-4.76 NS
Primary site of bacteremia
Lung 3/5 (60) 5/40 (12.5) 10.5 1.39-79.13 .03
Intraabdomina1 site 4/23 (17.4) 4/22 (18.2) 0.95 0.21-4.37 NS
Urinary tract 0/9 8/36 (22.2) 0.18 0.01-3.36 NS
Multidrug resistance 1/5 (20) 7/40 (17.5) 1.18 0.11-12.21 NS
Initial clinical manifestation
Hypotension 4/15 (26.7) 4/30 (13.3) 2.36 0.50-11.19 NS
Altered mental status 4/8 (50) 4/36 (11.1) 8 1.41-45.23 .03
Body temperature
<37°C 2/2 (100) 6/43 (14) 28.85 1.24-672.13 .03
>38SC 6/39 (15.4) 2/4 (50) 0.18 0.02-1.55 NS
Oliguria 4/13 (30.8) 1/27 (3.7) 11.56 1.14-117.44 .03
Jaundice 4/12 (33.3) 4/33 (12.1) 3.63 0.74-17.81 NS
Leukocytosis (WBC count, > 10,000/mm3
) 6/28 (21.4) 2/15 (13.3) 1.77 0.31-10.11 NS
Neutropenia (neutrophil count, <500/nun3
) 0/5 8/39 (20.5) 0.34 0.02-6.71 NS
Complications
Septic shock 7/15 (46.7) 1/30 (3.3) 25.38 2.71-237.58 .0009
Deterioration in mental status 7/12 (58.3) 1/33 (3.0) 44.8 4.50-445.75 .0001
Bilirubin level > 1 mg/dL 6/15 (40) 1/29 (3.4) 18.67 1.97-176.45 .003
Increase in creatinine level of more than twofold 7/11 (63.6) 1/34 (2.9) 57.75 5.57-598.44 .00005
Platelet count, < 100,000/nun3
5/7 (71.4) 1/36 (2.8) 87.5 6.65-1151.19 .00013
Treatment
None or delayed 2/9 (22.2) 5/35 (14.3) 1.71 0.27-10.74 NS
Delayed 1/4 (25) 5/35 (14.3) 2 0.17-23.25 NS
Surgical/invasive procedure 1/10 (10) 7/35 (20) 0.44 0.05-4.12 NS
Combination therapy 1/18 (5.6) 7/27 (25.9) 0.17 0.02-1.51 .11
NOTE. NS = not significant.
* Twenty-one of 45 episodes of citrobacter bacteremia occurred during this period.
t Statistical analysis by means of the X2
test with two-tailed Fisher's exact test.
acquired. An intraabdominal site, rather than the urinary
tract, was the leading primary site of citrobacter infection.
Malignancy, especially in the intraabdominal organs, and
hepatobiliary stones were the two most predominant under-
lying diseases.
Citrobacter species were more often isolated in our cases of
polymicrobial bacteremia than were other gram-negative bacilli
because of the predominance of primary infections at contami-
nated sites, especially the abdominal cavity. When a patient
presents with citrobacter bacteremia, a thorough search for an
intraabdominal lesion should be made. Multidrug resistance
among Citrobacter species was found to be associated with
administration of a third-generation cephalosporin before the
onset of bacteremia.

7. cm 1996;23 (September) Citrobacter Bacteremia 549
Because the data are limited, we suggest that ciprofloxacin
be considered the drug of choice for bacteremia due to these
strains. Combination therapy with a ,B-Iactarn and an aminogly-
coside are suggested as the initial empirical treatment because
this combination was associated with a lower mortality rate in
our study. Septic shock with organ failure was the most im-
portant poor prognostic factor, and the need for good supportive
care for patients with this complication cannot be overempha-
sized.
Acknowledgment
The authors thank Professor Andrew T. F. Huang of Duke Uni-
versity (Durham, NC) for reviewing the manuscript.
References
I. Lipsky BA, Hook EW III, Smith AA, Plorde J1. Citrobacter infections in
humans: experience at the Seattle Veterans Administration Medical
Center and a review of the literature. Rev Infect Dis 1980; 2:746-60.
2. Samonis G, Anaissie E, Elting L, Bodey GP. Review ofCitrobacter bacter-
emia in cancer patients over a sixteen-year period. Eur J Clin Microbiol
Infect Dis 1991; 10:479-85.
3. Drelichman V, Band JD. Bacteremias due to Citrobacter diversus and
Citrobacter freundii: incidence, risk factors, and clinical outcome. Arch
Intern Med 1985; 145:1808-10.
4. Hodges GR, Degener CE, Barnes WG. Clinical significance of Citrobacter
isolates. Am J Clin Patho11978; 70:37 -40.
5. Holmes B, King A, Phillips I, Lapage SP. Sensitivity of Citrobacter freun-
dii and Citrobacter koseri to cephalosporins and penicillins. J Clin
Pathol 1974;27:729-33.
6. Madrazo A, Geiger J, Lauter CB. Citrobacter diversus at Grace Hospital,
Detroit, Michigan. Am J Med Sci 1975;270:497-501.
7. Chow JW, Fine MJ, Shlaes DM, et al. Enterobacter bacteremia: clinical
features and emergence of antibiotic resistance during therapy. Ann
Intern Med 1991; 115:585-90.
8. Bauer AW, Kirby WMM, Sherris JC, Turck M. Antibiotic susceptibility
testing by a standardized single disk method. Am J Clin Pathol 1966;
45:493-6.
9. National Committee for Clinical Laboratory Standards. Performance stan-
dard for antimicrobial disk susceptibility tests: approved standard.
NCCLS document no. M2-A3. Villanova, Pennsylvania: National Com-
mittee for Clinical Laboratory Standards, 1984.
10. Madrazo A, Henderson MD, Baker L, Vaitkevicius VK, Lauter CB. Mas-
sive empyema due to Citrobacter diversus. Chest 1975; 68:104-6.
11. Roselle GA, Watanakunakorn C. Polymicrobial bacteremia. JAMA 1979;
242:2411-3.
12. Parry MF, Hutchinson JH, Brown NA, Wu C-H, Estreller L. Gram-nega-
tive sepsis in neonates: a nursery outbreak due to hand carriage of
Citrobacter diversus. Pediatrics 1980;65:1105-9.
13. Grant MD, Horowitz HI, Lorian V. Gangrenous ulcer and septicemia due
to Citrobacter. N Engl J Med 1969;280:1286-7.
14. Jones SR, Ragsdale AR, Kutscher E, Sanford JP. Clinical and bacteriologic
observations and a recently recognized species of Enterobacteriaceae,
Citrobacter diversus. J Infect Dis 1973; 128:563-5.
15. Lew PD, Baker AS, Kunz LJ, Moellering RC Jr. Intra-abdominal Citro-
bacter infections: association with biliary or upper gastrointestinal
source. Surgery 1984;95:398-403.
16. Bartlett JG. Intra-abdominal sepsis. Med Clin North Am 1995; 79:599-
617.
17. Kreger BE, Craven DE, Carling PC, McCabe WR. Gram-negative bacter-
emia. III. Reassessment of etiology, epidemiology, and ecology in 612
patients. Am J Med 1980;68:332-43.
18. Kiani D, Quinn EL, Burch KH, Madhavan T, Saravolatz LD, Neblett TR.
The increasing importance of polymicrobial bacteremia. JAMA 1979;
242: 1044-7.
19. Elting LS, Bodey GP, Fainstein V. Polymicrobial septicemia in the cancer
patient. Medicine (Baltimore) 1986;65:218-25.
20. DuPont HL, Spink WW. Infections due to gram-negative organisms: an
analysis of 860 patients with bacteremia at the University of Minnesota
medical center, 1958-1966. Medicine (Baltimore) 1969;48:307-32.
21. Watanakunakorn C, Perni SC. Proteus mirabilis bacteremia: a review of
176 cases during 1980-1992. Scand J Infect Dis 1994; 26:361-7.
22. Skerk V, Bobinac E, Popovic-Uroic T, SchonwaldS. IstrazivanjeKlinicko-
laboratorijskih parametara bakterijemije i sepse uzrokovane Escheri-
chiom coli. Lijec Vjesn 1993; 115:85-9.
23. Weinstein MP, Murphy JR, Reller LB, Lichtenstein KA. The clinical
significance of positive blood cultures: a comprehensive analysis of 500
episodes of bacteremia and fungemia in adults. II. Clinical observations,
with special reference to factors influencing prognosis. Rev Infect Dis
1983; 5:54-70.
24. Samonis G, Ho DH, Gooch GF, Rolston KVI, Bodey GP. In vitro suscepti-
bility of Citrobacter species to various antimicrobial agents. Antirnicrob
Agents Chemother 1987;31:829-30.
25. Watanakunakorn C, Jura J. Klebsiella bacteremia: a review of 196 episodes
during a decade (1980-1989). Scand J Infect Dis 1991;23:399-405.
26. Rello J, Quintana E, Mirelis B, Gurgui M, Net A, Prats G. Polymicrobial
bacteremia in critically ill patients. Intensive Care Med 1993; 19:22-5.
27. Ashkenazi S, Leibovici L, Samra Z, Konisberger H, Drucker M. Risk
factors for mortality due to bacteremia and fungemia in childhood. Clin
Infect Dis 1992; 14:949-51.