Journal
of Hospital
Infection
(1994)
27, 299-305
Klebsiella bacteraemia: National University K. H. Lee, Department
K. P. Hui,
of Medicine, Accepted
a report of 101 cases Hospital, Singapore W. C. Tan
National
University
for publication
from
and T. K. Lim Hospital,
21 February
Singapore
0511
1994
Summary:
One hundred and one cases of Klebsiella bacteraemia from the National University Hospital, Singapore, were reviewed retrospectively. There were 54 (53.5%) males and 47 (46.5%) females. Mean (&SE) age was 54 (j12.4) years. Overall mortality was 26%. Nosocomial infections accounted for 20%. Underlying diabetes mellitus and malignancy were present in 36 and 26% respectively. The source of the bacteraemia was not known in 33% of cases, 17% had liver abscess, 29% had urinary tract infections, 9% had pneumonia, 10% had an abscess separate from the liver, and 3% had biliarv sepsis. Elevated alkaline phosphatase (> 100 U-‘) aas seen in all cases of iiver abscess (sensitivity loo%, specificity 27%). Nonsurvivors had a significantly lower platelet count than survivors (104 f 25 X lo”/1 vs. 176 f 15 X log/l, unpaired t-test P
Klebsiella
bacteraemia;
source;
thrombocytopenia
Introduction The genus Klebsiella is composed of a group of Gram-negative bacilli, with a prominent polysaccharide capsule that is important in virulence. Klebsiella is historically the second most commonly isolated Gram-negative organism (6-66%) in blood cultures after Escherichia coli, and is associated with the second highest mortality after Pseudomonas aeruginosa.“” We have retrospectively reviewed 101 cases of Klebsiella bacteraemia in Singapore. Methods One hundred and one cases of Klebsiella bacteraemia admitted to the National University Hospital, Singapore, were studied retrospectively over Correspondence to: Kang Lee, Fellow, Division of Critical XLledical Center, 200 Lothmp Street, Pittsburgh, PA 15213. This study was supported by NUS grant GR6105.
299
Care
bledicine,
University
of Pittsburgh
300
K. H.
Lee
et al.
a 2-year period (1 January 1990 to 31 December 1991). These cases were identified from blood culture logbooks. Klebsiella bacteraemia comprised 14% of all positive blood cultures over this period. Blood cultures were performed with the radiometric blood culture analyser (BACTEC). Antimicrobial susceptibility tests were done according to the standard disc diffusion method. Klebsiella was identified and speciated according to standard methods. Unfortunately not all the species were identified, and we have therefore considered all of them as Klebsiella spp. for the purpose of this analysis. Nosocomial infection was defined as an infection acquired after 72 h in hospital. The source of infection was determined by cultures or positive imaging results or both. For instance, liver abscesses were diagnosed by imaging alone, thus diagnosis was not dependent on positive cultures from the abscess, while urinary infection was always accompanied by positive urine cultures. Pneumonia was determined by typical radiological changes. Biliary sepsis was confirmed from direct positive cultures. Antibiotic sensitivities and resistance were recorded, along with the age, sex, underlying illnesses, haemoglobin concentration, total white count, lowest platelet count, urea, creatinine and liver function tests. Unfortunately, prior antibiotic usage could not be determined. Results are expressed as mean f SE of mean. Data were analysed using a statistics package StatView 5 12 + (1986) f rom Brainpower Inc., California. Student’s t-test and x2 testing were used as indicated. Results One hundred and thirty-nine patients were identified during the study period. Only 101 of their notes could be recovered (73%). There were 54 males (53*5%), and 47 females (46-S%), with a mean age of 53 ( f 2.4) years. None of these patients were alcoholics. Overall mortality was 26%. The laboratory results of haemoglobin, total white count, lowest platelet count, urea and creatinine are shown in Table I. Twenty per cent of infections were hospital-acquired, and there was a tendency for nosocomial infection to be associated with a higher mortality rate (45%) compared with community-acquired infection (21 O/o), though the difference was not significant. Diabetes mellitus was present in 36% of cases, and malignancy
Table
I.
Laboratory
Variables Haemoglobin (g dl-‘) Total white count (lo9 1-l) Lowest platelet count (IO9 1-l) Urea (mm01 1-l) Creatinine (pm01 1-l)
findings in 101 patients Mean
(SE
11 14.4 158 11.5 141
of mean) (0.3) (1.7) (13) (1.0) (17)
Klebsiella
bacteraemia
in
in 27%. Seventy-nine per cent of patients 38°C at the time of blood culture. Klebsieh of the overall group, with 6.9% Klebsiella
301
Singapore
had a temperature greater than pneumoniae constituted 48.5% oxytoca, and 44.6% Klebsiella
SPP.
Non survivors had significantly lower platelet counts along with a significantly higher percentage of thrombocytopenic ( < 150 X 10” l- ‘) patients, plus a significantly higher incidence of disseminated intravascular coagulation (DIC) (Table II). However, thrombocytopenia was not found significantly more often in DIC-positive patients. Neither the presence of malignancy or diabetes mellitus predicted mortality. Associated sites of infection are listed in Table III. Only 21% of patients with urinary tract infection had an indwelling urinary catheter. Sputum was positive only in one of nine (11 O/o) patients with pneumonia. The most common identifiable source of nosocomial infection was the urinary tract (35%). Pneumonia was community-acquired in 80% of cases, and had the highest mortality rate (50%). Liver abscess was found to be multiple in nine of 17 (53%) cases, with 56% localized only in the right lobe, 31% in both lobes and 12.5% on the left. Two patients had radiological evidence of pneumonia that had spread haematogenously. All the patients with liver abscess had elevated alkaline phosphatase (388 U ll’ f 56 U ll’), and only one patient (6%) was jaundiced. However, an elevated alkaline phosphatase ( > 100 U l- ‘) had only a positive predictive value of 27% for the presence of liver abscess, even though it was 100% sensitive (specificity 27%). Seventy-one per cent
Table
I I. Haematological
values
SUrViVOrS
Lowest platelet Thromhocytopenia (<150X 10’1 DIG* * Disseminated
count ‘)
176
intravascular
and non surzGors
survivors
Statistics
Non
(15)X
11% 21%
in surzk’ors
lO”l-’
mortality mortality
104 (25) x 10” I ’
P~0.05
(unpaired
37% 64%
P < 0.01 P-c 0.01
(x’ (x’
mortality mortality
coagulation.
Table
III. focus
Site
of infection
Unknown Urinary tract Liver abscess Lung Other abscess Biliary
Site of of infection
primary
% 32.7 28.7 16.8 8.9 9.9 3.0
test) test)
t-test)
K. H. Lee et al.
302
and 67% of these had positive cultures for had a drainage procedure, Klebsiella in the aspirated pus. Only one of the 12 (8%) had a mixed growth of organisms. Mortality was 18% and diabetes mellitus was present in 41%) and malignancy in 12% of these patients. Fifty-seven per cent had chills, while 43% had right upper quadrant pain. The in-vitro antibiotic sensitivities are shown in Figure 1. All the isolates were found to be sensitive. Gentamicin (78% (n = 19) tested with imipenem sensitive), cotrimoxazole (78%) and ceftriaxone (75%) were the next most effective antibiotics based on in-vitro testing. There was 94% resistance to ampicillin. Discussion This is the first survey of Klebsiella bacteraemia in Singapore. There was a high frequency of associated diabetes mellitus and underlying malignancy. A site of infection was demonstrated in 67% of cases. The most common site was the urinary tract, whether the infection was acquired from the community or within the hospital. Liver abscess was the next most common site. One therefore should look carefully at the urine as a primary site of infection in cases where Klebsiella bacteraemia is present. In a survey of cancer patients, Bodey et ~1.‘~ found a source of the Klebsiella bacteraemia in only 49% of cases. Their most common source was the lung (16%), followed by the genitourinary tract (14%), and the gastrointestinal tract (12%). Several surveys have found the urinary tract to be the primary focus of infection. 5,11,17The incidence of urinary infection varies considerably among the surveys. Liver abscess as a source or
g; 8 9 .m 7i $
90 80 70 60 ‘50 40 30
20 10 0
Figure
1. In-vitro
antibiotic
resistance
to
Klebsiella
spp.
Klebsiella
bacteraemia
in Singapore
303
complication of Klebsiella bacteraemia was not reported in these series. Our series is unique in its high percentage of liver abscess associated with Klebsiella bacteraemia, although we were unable to ascertain whether the liver abscesses were secondary to the bacteraemia, or the inciting event. Only 12% ( nme of 75) of pyogenic liver abscesses were caused by . Klebsiella-Enterobacter spp. in a series from the Mayo Clinic’s all of which also had positive blood cultures. They found multiple abscesses in 48% of cases. Their mortality rate was 40% for the whole group (compared with our rate of 18% for liver abscesses in our series), and they had a lower incidence of diabetes mellitus (16 vs. 41% in our series) in their patients. We did not observe any septic endophthalmitis which Cheng et a1.19have reported to be a feature of Klebsiella liver abscesses in Taiwan. The alkaline phosphatase was always elevated in those cases with liver abscesses but jaundice was not usually present. It would be worth searching for liver abscesses when the alkaline phosphatase is elevated. Klebsiella pneumonias are uncommon especially if they are acquired from the community. Eighty per cent of the pneumonias in our series were community-acquired, and the associated mortality rate for pneumonia was the highest (50%). Bacteraemic Klebsiella pneumonias form only 3% or less of the total of community-acquired pneumonias.20 If sputum culture alone is taken as proof of infection, the true incidence may be over-reported as this may merely reflect colonization in the oropharynx. Another explanation may be prior usage of antibiotics in the community, but unfortunately we have no data on this issue. Diabetes mellitus was the most commonly (36%) associated condition. This is much higher than the reported prevalence (< 5%) of diabetes mellitus in Singapore,21 and higher than the rate of 21% seen in a recent study of adult community-acquired pneumonia in our hospital.22 The cause for this association may be related to impaired granulocyte phagocytic functions found in diabetic patients.23 Nosocomial Klebsiella bacteraemia was associated with a higher mortality though this was not statistically significant. This may be a true finding, although one cannot discount the possibility of delayed diagnosis with the consequent label of a nosocomial infection, and the associated delay in instituting appropriate antibiotics, which may have then caused the observed increase in mortality. Thrombocytopenia appeared to be an adverse prognostic factor. This has not been previously described, although we have reported that thrombocytopenia was an independent predictor of mortality in a group of septic patients in a medical intensive care unit.24 DIC status was also an important prognostic factor but did not predict thrombocytopenia. The explanation for this finding remains speculative. The presence of thrombocytopenia may reflect the severity of sepsis and hence, mortality. In their survey of cancer patients, Bodey et al. described a poorer prognosis in patients with bronchogenic carcinoma and a lower response rate in patients
K. H. Lee et al.
304
with pneumonia.16 We did not find any increased mortality in patients with cancer. Others have described appropriate antibiotic therapy, shock, azotemia, neutropenia and temperature response as factors affecting mortality.5-9 Antibiotic sensitivities that were tested demonstrated that 78% of isolates were sensitive to gentamicin. Watanakunakorn and Jura17 demonstrated an even higher susceptibility to gentamicin (98%) for K. pneumoniae. Imipenem was not tested against sufficient isolates in this study to be able to interpret its activity. Of all the third generation cephalosporins tested, Klebsiella was most susceptible to ceftriaxone (75%). We are grateful to our research nurse, Ms. Tan Sok Har, and to the Consultants in the Departments of Medicine, Surgery and Paediatrics who allowed us to include their patients in this study. We also thank the Department of Microbiology for their logbook on blood culture results.
References 1. EORTIC Antimicrobial Therapy Project: Ceftazidime combined with a short or long course of amikacin for empirical therapy of Gram-negative bacteraemia in cancer patients with granulocytopenia. N Engl J Med 1987; 317: 1692-1698. 2. Klatersky J, Glauser MP, Schimpff SC et al. Prospective randomized comparison of three antibiotic regimens for empirical therapy of suspected bacteraemia infection in febrile granulocytopenic patients. Antimicrob Agents Chemother 1986; 29: 263-270. 3. Weinstein MP, Murphy JR, Reller LB et al. The clinical significance of positive blood culture: Comprehensive analysis of 500 episodes of bacteraemia and fungemia in adults. II. Clinical observations, with special reference to factors influencing prognosis. Rev Infect Dis 1983; 6: 54. 4. Spengler RF, Greenough WB III, Stolley PD. A descriptive study of nosocomial bacteraemia at The Johns Hopkins Hospital, 1968-1974. Johns Hopkins Med J 1978;
142: 77. 5. Kreger BE, Craven DE, Carling P et al. Gram-negative bacteraemia. III. Reassessment of etiology, epidemiology and ecology in 612 patients. Am J Med 1980; 68: 332-343. 6. Singer C, Kaplan M, Armstrong D. Bacteraemia and fungemia complicating neoplastic disease. A study of 364 cases. Am J Med 1977; 62: 731-742. 7. McHenry MC, Gavan TL, Hawk WA et al. Gram-negative bacteraemia: variable clinical course and useful prognostic factors. Cleve Clin Q 1975; 42: 15. 8. Myerowitz RL, Medeiros AA, O’Brien TF. Recent experience with bacillaemia due to Gram-negative organisms. J Infect Dis 1971; 127: 239. RE, Hood AF, Hood CE et al. Factors affecting mortality of Gram-negative rod 9. Bryant bacteraemia. Arch Intern Med 1971; 127: 120-l 28. 10. DuPont HL, Spink WW. Infections due to Gram-negative organisms: An analysis of 860 patients with bacteraemia at the University of Minnesota Medical Center, 1958-1966. Medicine Baltimore 1969; 48: 307. 11. Altemeier WA, Todd JC, Inge WW. Gram-negative septicaemia: A growing threat. Ann Surg 1967; 166: 53&542. 12. de la Torre MG, Romero-Vivas J, Martinez-Beltran J et al. Klebsiella bacteraemia: An analysis of 100 episodes. Rev Infect Dis 1985; 7: 143-150. 13. Montgomerie J. Epidemiology of Klebsiella and hospital-associated infections. Rev Infect Dis 1979; 1: 736-753. 14. Lim TK, Chan TB. A clinical evaluation of positive blood cultures and bacteraemia. Singapore Med J 1983; 24: 128-l 34. 15. Young LS. Gram-negative sepsis. In: Mandell GL, Gordon Douglas R, Benett JE, Eds. Principles and Practice of Infectious Diseases, 3rd Edn, 1990: 61 l-637. 16. Bodey GP, Elting LS, Rodriguez S, Hernandez M, Klebsiella Bacteraemia: A lo-year review in a cancer institution. Cancer 1989; 64: 2368-2376.
Klebsiella 17. 18. 19.
20. 21. 22.
23. 24.
bacteraemia
in
Singapore
305
Watanakunakorn C, Jura J. Klebsiella: A review of 196 episodes during a decade (1980-1989). ScnndJ Znfect Dis 1991; 23: 399405. Lazarchick I. de Souza e Silva NA. Washington IA. Pvoeenic Liver Abscess. Mnvo Clin Proc 1973; 48: 349-355. Cheng DL, Liu YC, Yen MY, Liu CY, Wang RS. Septic metastatic lesions of pyogenic liver abscess. Their association with Klebsiella bneumonin bacteraemia in diabetic patients. Arch Intern Med 1991; 151: 1557-1559. I Carpenter JL. Klebsiella pulmonary infections: occurrence at one medical center and review. Ret Znfect Dis 1990; 12: 672-682. Thai ,4C, Yeo PPB, Lun KC et al. Changing prevalence of diabetes mellitus in Singapore over a ten \-ear period. 7 Med Assoc Thailand 1987; 70: 63-67. ._ _ Hui KP, Chin NK, Chow K et nl. Prospective study of the aetiology of adult community acquired bacterial pneumonia needing hospitalization in Singapore. Silzgnpore MedJ (in press). Repine JE, Cla\vson CC, Goetz FC. Bactericidal function of neutrophils from patients with acute bacterial infections and from diabetics. J Infect Dis 1980; 142: 8699874. Lee KH, Hui KP, Tan ESH, Lim TK, Chan TB, Tan Vv;C. Thromhocytopaenia in sepsis: A predictor of mortality. Singapore IWed J 1993; 34: 245246. I/
Y
-
.Y