Journal
of Hospital
Septicaemia the Hospital
Injection
(1992)
22, 185-l
in paediatric intensive-care patients at for Sick Children, Great Ormond Street H. Holzel
Department
95
of Microbiology,
Accepted
and M. de Saxe
Hospital for Street, London,
Sick UK
Children,
for publication
6 July
1992
Great
Ormond
Summary:
A review of nosocomial septicaemia in paediatric intensive care in a tertiary referral setting was undertaken for a 33-month period (1988-90). This involved six units: Cardiothoracic surgery; Neonatal surgery; general medical; Renal dialysis/transplant; Haematology/Oncology and Infectious disease/Immunology. The latter two units undertake bone marrow transplantation. During the study period, 10 719 admissions were made to these areas and 624 episodes of septicaemia were documented in 464 children. The frequency of septicaemia per 100 admissions ranged from 1.5 in the Renal Transplant Unit to 17.3 in the Haematology/Oncology unit. Over 60% of all septicaemic episodes occurred in children in the Haematology/Oncology and Cardiac Units. Gram-positive organisms were responsible for 66% of episodes, Gram-negative organisms for 17% and fungi for 3%. Polymicrobial episodes accounted for 13%. Coagulase-negative staphylococci were the most frequent isolates overall (43% of episodes in pure culture, and a further 6% in combination with other organisms). Staphylococcus aureus was associated with 10% of episodes, Enterobacteriaceae with 9% and Pseudomonas spp. 6% among which environmental pseudomonads predominated. Anaerobes and Haemophilus injluenzae were each isolated in less than 1% of episodes. Keywords: lococci.
Septicaemia;
paediatric
intensive
care;
coagulase-negative
staphy-
Introduction
The Hospital for Sick Children, Great Ormond Street, is a tertiary paediatric referral centre consisting of 347 beds arranged in 22 wards. Intensive care is provided in three formal areas: a Cardiothoracic surgical unit, a Neonatal surgical unit, and a General medical unit. Intensive therapy is also provided in the Haematology/Oncology unit, the Renal dialysis/Transplant unit and the combined Infectious disease and Immunology unit. Bone marrow transplantation is undertaken in both the Haematology and Infectious disease units. Nosocomial septicaemia is a well-recognized problem in paediatric intensive care lP4but although blood-stream infection has been extensively documented in neonates&’ and in children with haematological Correspondence 019%6701/92/110185+
to: Dr H. Holzel. 11 SOS.OOjO
Q 1992 The Hos,xtal
185
Infection
Socmy
186
H. Holzel
and M. de Saxe
malignancies or solid tumours S-12there are few publications relating to this problem in other specialist paediatric groups undergoing intensive care.‘z4 We therefore present our findings in a study of septicaemia occurring in patients on the six intensive care units at the Hospital for Sick Children, over a 33-month period, from 1 January 1988 to 30 September 1990, during which time 10 719 admissions were made to these areas. Patients
and
methods
Patients These were in-patients on the six units undertaking intensive care or intensive therapy in our hospital. At the onset of a febrile episode, blood cultures were taken from each available lumen of central venous catheters, from other indwelling arterial or venous catheters and from a peripheral vein. Microbiology Blood cultures were processed in the diagnostic laboratory using a non-radiometric automated system (Bat Tee NR 730, Becton Dickenson). Isolation and identification of organisms were carried out by standard bacteriological methods. Isolates were tested for susceptibility to appropriate ranges of antibiotics by disc diffusion using a rotating Stokes’ technique. Species of coagulase-negative staphylococci (CNS) were determined according to the nomenclature of Kloos and Schleiferi3 using a multipoint inoculation technique. l4 Isolates of the same species of CNS were considered to belong to different strains if they differed in resistance to two or more unrelated antibiotics. Analysis of results All positive blood cultures from children in the six intensive care areas were reviewed for the 33-month period, from 1 January 1988 to 30 September 1990. Those taken 72 h or more after patient admission were considered to indicate nosocomially-acquired septicaemia and were included in the analysis. Dejinitions A treatment period in our hospital is considered to be the administration of 10 days’ intravenous (iv) antibiotics for the treatment of serious sepsis. A clinically-significant septicaemic episode was defined as the isolation of a clinically and microbiologically appropriate organism (or organisms), from the blood of a pyrexial child, in whom no other cause of pyrexia was identified, and who subsequently received antibiotic therapy directed against the particular isolate(s). A single episode was defined as the isolation of the same organism(s) within the lo-day treatment period. In a few instances when cultures
Septicaemia
in paediatric
187
intensive-care
yielded the same organism in ongoing sepsis outside the lo-day was considered to be a continuation of the same episode. A distinct episode was defined as:
period, this
(a) The isolation of the same organism(s) outside the lo-day treatment period, if negative blood cultures intervened. (No attempt was made to distinguish between relapse and reinfection in these episodes.) (b) The isolation of different organism(s) more than 24 h after the initial positive blood culture. A polymicrobial episode was the isolation of more than one species either (a) in the initial positive blood culture, or (b) within 24 h of the initial positive culture, or (c) where the same combination of organisms was found on more than one occasion. Exclusions We excluded isolates of CNS, streptococci or Corynebacterium spp. obtained on a single occasion from peripheral blood cultures in children in whom corresponding blood cultures from intravascular catheters failed to yield the organism, or in patients in whom these devices were absent. Results
During the 33 months of the study, there were 10 719 admissions to the six units providing intensive care and therapy. A total of 624 episodes of septicaemia were documented in 464 children. The number of episodes in individual units, and their frequency, expressed as episodes of septicaemia per 100 admissions, are given in Table I. Sixty-one per cent of all episodes occurred on two units, the Haematology/Oncology unit (36%) and the Cardiothoracic surgical unit (25%). In contrast, only 7% occurred on the Renal dialysis/transplant unit. The frequency of septicaemia (per 100 admissions) was also highest on the
Table
I. Frequency Sick
of septicaemia in six intensive-care Children, Great Ormond Street
Unit
Admissions
Haematology/Oncology Neonatal surgery General medical surgery Cardiothoracic Infectious disease Renal dialysis/transplant All
units
1315 655 739 3452 1635 2923 10719
areas at the Hospital (1988-90)
Septicaemic episodes
Septicaemia per 100 admissions
227 75 67 154 56 45
17.3 11.5 9.1 4.5 3.4 1.5
624
5.8
for
188
H. Holzel
and M. de Saxe
Haematology/Oncology unit (17.3) and lowest (1.5) on the Renal dialysis/transplant unit. The range and distribution of organisms responsible for episodes of septicaemia in the different units are shown in Figure 1. Gram-positive bacteria were isolated in pure culture in 66% of the 624 episodes and ranged from 52% of episodes in the Infectious disease unit to 75% in the Neonatal surgical unit. Gram-negative organisms accounted for only 17% of episodes overall and were most common in the Infectious disease unit (29%). Polymicrobial episodes accounted for 13 % of septicaemic episodes. A small number of episodes were caused by anaerobes, Bacteroides fragilis or Clostridium spp. (in three and two patients respectively). Candida albicans was responsible for 15 episodes, Candida spp. for three episodes and Trichosporon beiglii for one episode in a patient with an underlying diagnosis of systemic lupus erythematosus. Candida albicans was also associated with 6% of the 80 polymicrobial episodes in our study. The organisms causing Gram-positive and Gram-negative septicaemia are presented in Tables II and III. Coagulase-negative staphylococci (CNS), were the predominant isolates. They were isolated in pure culture from 43% of episodes and from a further 6% in combination with other organisms. During the study period, the proportion of septicaemias due to CNS ranged from 68% in the Neonatal surgical unit to 32% in the Infectious disease unit (Table II). Identification of the individual strains of CNS showed that 85 % were Staphylococcus epidermidis; the remaining 15% comprised Staphylococcus haemolyticus, Staphylococcus hominis and Staphylococcus warneri (IO%, 3 % and 2% respectively). Staphylococcus aureus was found in pure culture in 10% of episodes. Streptococci caused 8% and enterococci 3% of episodes overall, and were most common in the Haematology/Oncology unit (19’/,). A range of other Gram-positive organisms were responsible for 15 episodes in patients on the Haematology and Cardiac units and included: Bacillus spp., Corynebacterium spp., micrococci and Rhodococcus sp. (six, five and three episodes and one episode respectively). During the period studied, patients on the six units experienced 107 episodes of Gram-negative septicaemia (17% of the total) (Table III) and Gram-negative organisms were associated with a further 6% of polymicrobial episodes. Gram-negative organisms in pure culture were most common as a cause of septicaemia in the Infectious disease unit (29%) and least common in the Haematology/Oncology unit (14%). Enterobacteriaceae were responsible for 54% of these episodes and Escherichia coli together with organisms of the Enterobacterlcitrobacter group predominated in all units. Pseudomonads were responsible for 35 (6%) of septicaemic episodes overall and caused 33 % of Gram-negative septicaemias. They were most common in the Haematology/Oncology unit (12 episodes) and the Cardiothoracic surgical unit (nine episodes). Pseudomonads of environmental origin were associated with 63% of the
Figure Great
unit
ICU
ICU
0
1. Organisms associated Ormond Street, London
Hoemotology/Oncology
surglcol
surgical
medico1
General
Neonatal
unit
disease
infectious
ICU
unit
dialysis/transplant
Cardiothorocic
Renal
50
with 624 episodes 1 January 1988-30
25 100
(12%)
( I I %)
Number
of septicaemia in 464 September 1990.
75
N=75
N = 67
(9%)
( 7 %)
N = 56
N = 45
150
receiving
of septicoemic
children
125
intensive
episodes
175
N = 154 (25%)
care
200
Streptococci
a
Gram-negatives
Other
q 0
at the
Hospital
225
for
250
N : 227
Polymlcrobial
organisms
Other
q
ta
Sick
Children,
‘/o)
aureus
(36
negative
Gram-positives
Staphylococcus
Coagulosestaphylococci
q
4
isolated
II.
Gram-positive
* Micrococci,
Number
Corynebacterium
of episodes
spp., Bacillus
Gram-positive organisms Coagulase-negative staphylococci Staphylococcus aureus All streptococci: S. sanguis/salivarius/mitis S. pneumoniaelothers Enterococci Other Gram-positive organisms* Polymicrobial (including Gram-positive organisms)
All
Organisms
Table
spp. and
organisms
Rhodococcus
sp. caused
2 67
3, 5, 6 episodes
75
5
13 154
:, 0
: 2
: 0
2
;
and
1 episode
227
12
:: 12 3 4 6
70 33
45
56
16
:,
i 2
52 32 11
Infectious disease
areas
respectively.
13
Fl
:,
58 40 11 7
Renal dialysis/ transplant
of septicaemia
in
in six intensive-care
episodes Haematology /Oncology
in children
% septicaemic General medical
episodes
72 58 4
Neonatal surgery
septicaemic
75 68 4
with
:: 10 5 4
Cardiothoracic surgery
associated
624
10
;
2; 10 8 7 *
All units
s
St k (D r
F m e s r E
isolated
III.
Gram-negative
sp.
spp.
spp. 0 0
1
154 presented
2(3) 75
16(7) 227
0 1
5(11) 45
: 0
2(4) 56
: 1
; 5
1 0
i
-__
-
624 were
W6)
z
3: 13 22 7
23 31 3
107(17)
All units
organ,sms
16(28)
Infectious disease
areas
as actual numbers of episodes wth which particular
3(2) 67
i
lw3)
:,
0 0
:
t
2 1
3 0 3
1: 7
4 2
2 4
9(20)
in Renal dialysis/ transplant
episodes
in six intensive-care
:
1
10 5
33(14)
____
of septicaemic Haematology /Oncology
in children
i
1 3
lO(15)
(%) General medical
episodes
z 2
i
5 4
4 9 i
12(16)
No.
septicaemic
Neonatal surgery
with
27(18)
-____
associated
Cardiothoracic surgery
organisms
As the total number of Gram-negative episodes was low, figures are associated. Percentages are given in parentheses.
sp. and Capnocytophaga
of episodes
* Agrobacterium
Number
Gram-negative organisms Enterobacteria: Eschevichia coli Klebsiella/Enterobacter/Citrobacter Proteus sp./Morganella sp. Salmonella typhimurium Pseudomonads: P. aeruginosa Other pseudomonads Alcaligenes/Acinetobacter/Moraxella Haemophilus injluenzae Others* Polymicrobial (including Gram-negative organisms)
All
Organisms
Table
c z
d
B e.
E
Et v ;i 6’
b
B L. Y w tj-
F w d c
192
H. Holzel
and M. de Saxe
episodes of pseudomonas septicaemia, whereas Pseudomonas aeruginosa was associated with only 37%. Five immuno-suppressed patients with tunnelled central venous lines had severe, ongoing problems due to P. aeruginosa (four patients) and Xanthomonas maltophilia (one patient). Almost a quarter of septicaemic children experienced recurrent episodes (i.e. 102 of 424), and 3% (44 of 424) suffered three or more episodes (Table IV). These recurrent episodes were most common in patients on the Haematology/Oncology, Infectious disease and General intensive care units and, in virtually all cases, affected immunodeficient patients who required repeated hospital admission and protracted hospital stay. Different bacterial species were associated with recurrent episodes in 62 children, different strains of CNS in 28 children, and recurrences due to the same organism occurred in only 12 of the 102 children (CNS in nine; S. aweus in two patients and Enterobacter sp. in one patient). Discussion
The study carried out at Great Ormond Street is unusual in that we review data on septicaemia in over 10 000 patients admitted to six specialist paediatric intensive care/therapy units. Information regarding the frequency of blood-stream infections in different groups of paediatric patients is usually restricted to studies of individual specialist units, e.g. Neonatal, Cardiothoracic surgical, Haematology/Oncology or General intensive care. In addition, most reports have been concerned with a wide range of nosocomial infections and have not provided detailed data on of nosocomial septicaemia, i.e. that septicaemia. 1,2,4*6Our definition occurring 72 h or more after admission, excludes the majority of community-acquired septicaemias.15 Blood-stream infection rates as high as 33 per 100 admissions have been reported from adult medical and surgical intensive care units, although much lower frequencies are more usual.‘s4 The rate for blood-stream infection in the Paediatric intensive care unit (PICU) at the Johns Hopkins
Table
IV.
Recurrent
Unit
Haematology/Oncology Infectious disease General medical Renal dialysis/transplant Neonatal surgery Cardiothoracic surgery All units
septicaemic
episodes Septicaemic children
in children
receiving
% experiencing more than episode
141 53; 36 64 136 464
;; 17 :z
122
one
intensive
care
% experiencing more than three episodes 6 9 :, 0 0 3
Septicaemia
in paediatric
intensive-care
193
Hospital over a 4-year period (1981-4) was 1.7 per 100 admissions and ranged from 0 to 4.5 per 100 admissions in the individual referring specialities. Among 19 admissions from unspecified referring specialities, septicaemia occurred in three (15.8%).’ In our study, the frequency of septicaemia expressed as episodes per 100 admissions was low in the Cardiothoracic unit (4.5) and high in the Haematology/Oncology unit (17.3), although patients in both units experienced large numbers of septicaemic episodes (136 and 141 respectively). This reflects the difference in patient populations in the two units; i.e. the large number of short-stay, immuno-competent patients admitted for cardiac surgery, in contrast to the immunocompromised children on the Haematology/Oncology unit who frequently require repeated hospital admission and protracted hospital stay. The frequency of septicaemia in our Neonatal surgical unit- 11.5 per 100 admissions-was in keeping with the findings of other workers.‘,4,6 Gram-positive organisms are now the most frequent cause of septicaemia in children undergoing intensive care in our hospital. Coagulase-negative staphylococci (predominantly S. epidermidis) are the most frequently isolated organisms; a finding in common with those of other paediatric and adult centres.‘Aa’6 Septicaemia due to CNS has been well documented in neonates undergoing intensive care,‘,‘,’ and in our Neonatal surgical intensive care unit, accounted for 68% of episodes. Staphylococcus aweus was associated with lo-12% of septicaemia in all units, with the exception of the Neonatal surgical and General medical units. These findings agree with data from a large study of nosocomial infection at the Hospital for Sick Children, Toronto (HSC-Toronto). 4 A high proportion of septicaemic episodes in patients on our Haematology/Oncology unit were associated with oral streptococci (12%). This raised the possibility of invasion during periods of mucositis in children undergoing intensive chemotherapy,’ and emphasized the need to review mouth care. Gram-negative organisms were identified in only 17% of our 624 septicaemic episodes and Enterobacteriaceae accounted for 9%; findings which agree with results from HSC-Toronto.4 Pseudomonas spp., in our patients, were responsible for 35 (6%) of septicaemic episodes overall. Pseudomonads of environmental origin were associated with 63% of the 35 episodes whereas P. aeruginosa was associated with only 37%. It was therefore of note, that in four of five immuno-suppressed patients with tunnelled central venous lines and severe ongoing pseudomonas septicaemia, P. aeruginosa was the causative organism. A matter of great concern was the unexpectedly high proportion of Gram-negative episodes in children on our Cardiothoracic surgical unit involving pseudomonads and Enterobacteriaceae (18%). In reviewing unit practice, it seemed likely that the use of femoral arterial and venous catheters for periods of more than 5 days was likely to have been an important contributory factor. Although our Neonatal surgical unit has a
H. Holzel
194
and M. de Saxe
particular interest in complex congenital abnormalities of the gastro-intestinal tract, septicaemia due to enteric organisms accounted for only 16% of episodes on this unit. In our study, Candida spp. accounted for 18 septicaemic episodes (3%) and C. albicans was responsible for 1.5 of these episodes. Similar data from the HSC-Toronto documented Candida spp. in 5% of cases of septicaemia. Anaerobic organisms were responsible for only five of the 624 episodes in our patients (two children on the Haematology/Oncology unit and three on the Renal dialysis/transplant unit) and Huemophilus influenxae was associated with five episodes in patients on the General medical, Cardiothoracic and Infectious disease units. In contrast, both anaerobes and H. infkenzae were absent from the list of pathogens associated with nosocomial bacteraemia in the Paediatric Intensive Care Unit in Toronto.3 The major risk factors in our paediatric intensive-care patients are those which are already well recognized, i.e. frequent patient handling, use of intravenous catheters, immunosuppression, prolonged hospital stay and frequent hospital readmission.‘~~9~“,16 Our data were derived from an ongoing study primarily designed to investigate the problem of severe sepsis caused by CNS in a wide paediatric setting. It confirmed the predominant role of CNS as the major cause of septicaemia in all six of our intensive care areas. The information has proved extremely useful in highlighting particular problems pertaining to the individual specialist units. It has also allowed us to guide and support measures of infection control and to prioritize the issues which require further prospective detailed investigation. We are grateful Great Ormond Health Research
to the staff of the Microbiology Department, Street, London, for their continued support Appeal Trust for the financial aid for our
Hospital for Sick Children, of this project and to the Child research programme.
References 1. Merritt 2.
WT, Stephens M. Nosocomial infections in the pediatric Rogers MC, Ed. Text Book of Pediatric Intensive Care. Baltimore: 1987; 2: 75.5-785. Allen U, Ford-Jones EL. Nosocomial infections in the pediatric
J Infect Control
3. Milliken 4. 5. 6. 7. 8.
intensive Williams patient:
care unit. In: and Wilkins an update.
Am
1990; 18: 176-193.
J, Tait GA, Ford-Jones EL, Mindorff CM, Gold R, Mullins G. Nosocomial infections in a pediatric intensive care unit. Crit Care Med 1988; 16: 233-237. Ford-Jones EL, Mindorff CM, Langley JM et al. Epidemiologic study of 4684 hospital-acquired infections in pediatric patients. Pediutr Infect DisJ 1989; 8: 668-675. Leonard EM, van Saene HKF, Shears P, Walker J. Pathogenesis of colonisation and infection in a neonatal surgical unit. Crit Care Med 1990; 18: 264269. Brown RB, Hosmer D, Chen HC et al. A comparison of infections in different ICUs within the same hospital. Crit Care Med 1985; 13: 472-476. Freeman J, Platt R, Sidebottom DG, Leclair JM, Epstein MF, Goldman DA. Coagulase-negative staphylococcal bacteremia in the changing neonatal intensive care unit population. JAMA 1987; 258: 2548-2552. Eshali H, Ringertz S, NystrGm S, Faxelius G. Septicaemia with coagulase negative staphylococci in a neonatal intensive care unit. Acta Paediatr Stand 1989; Suppl. 360: 127-134.
Septicaemia
in paediatric
intensive-care
195
9. Morrison VA, Peterson BA, Bloomfield CD. Nosocomial septicemia in the cancer patient: the influence of central venous access devices, neutropenia, and type of malignancy. Med Pediatr Oncol 1990; 18: 209-216. IO. Shapiro D, Wald ER, Nelson KA, Spiegelman KN. Broviac catheter-related bacteremia in oncology patients. Am J Dis Child 1982; 136: 679-681. 11. Wade JC, Schimpff SC, Newman KA, Wiernik PH. StuphyZococcus epidermidis: an increasing cause of infection in patients with granulocytopenia. Ann Intern Med 1982; 97: 503-50s. 12. Friedman LE, Brown AE, Miller DR, Armstrong D. Staphylococcus epidermidis septicemia in children with leukemia and lymphoma. Am J Dis Child 1984; 138: 715-719. 13. Kloos WE, Schleifer KH. Simplified scheme for routine identification of human Staphylococcus species. J Clin Microbial 1976; 1: 82-88. 14. Geary C, Stevens M, Sneath PHA, Mitchell CJ. Construction of a database to identify Staphylococcus species. 7 Clin Path01 1989; 42: 289-294. 15. Bryan CS, Reynolds KL, Derrick CW. Patterns of bacteremia in pediatrics practice: factors affecting mortalitv rates. Pediatr Infect Dis 1984: 3: 312-316. 16. Martin MA, P?aller MA, Wenzel RP. Coagulase-negative staphylococcal bacteremia: mortality and hospital stay. Ann Intern Med 1989; 110: 9-16.
of Hospital
Septicaemia the Hospital
Injection
(1992)
22, 185-l
in paediatric intensive-care patients at for Sick Children, Great Ormond Street H. Holzel
Department
95
of Microbiology,
Accepted
and M. de Saxe
Hospital for Street, London,
Sick UK
Children,
for publication
6 July
1992
Great
Ormond
Summary:
A review of nosocomial septicaemia in paediatric intensive care in a tertiary referral setting was undertaken for a 33-month period (1988-90). This involved six units: Cardiothoracic surgery; Neonatal surgery; general medical; Renal dialysis/transplant; Haematology/Oncology and Infectious disease/Immunology. The latter two units undertake bone marrow transplantation. During the study period, 10 719 admissions were made to these areas and 624 episodes of septicaemia were documented in 464 children. The frequency of septicaemia per 100 admissions ranged from 1.5 in the Renal Transplant Unit to 17.3 in the Haematology/Oncology unit. Over 60% of all septicaemic episodes occurred in children in the Haematology/Oncology and Cardiac Units. Gram-positive organisms were responsible for 66% of episodes, Gram-negative organisms for 17% and fungi for 3%. Polymicrobial episodes accounted for 13%. Coagulase-negative staphylococci were the most frequent isolates overall (43% of episodes in pure culture, and a further 6% in combination with other organisms). Staphylococcus aureus was associated with 10% of episodes, Enterobacteriaceae with 9% and Pseudomonas spp. 6% among which environmental pseudomonads predominated. Anaerobes and Haemophilus injluenzae were each isolated in less than 1% of episodes. Keywords: lococci.
Septicaemia;
paediatric
intensive
care;
coagulase-negative
staphy-
Introduction
The Hospital for Sick Children, Great Ormond Street, is a tertiary paediatric referral centre consisting of 347 beds arranged in 22 wards. Intensive care is provided in three formal areas: a Cardiothoracic surgical unit, a Neonatal surgical unit, and a General medical unit. Intensive therapy is also provided in the Haematology/Oncology unit, the Renal dialysis/Transplant unit and the combined Infectious disease and Immunology unit. Bone marrow transplantation is undertaken in both the Haematology and Infectious disease units. Nosocomial septicaemia is a well-recognized problem in paediatric intensive care lP4but although blood-stream infection has been extensively documented in neonates&’ and in children with haematological Correspondence 019%6701/92/110185+
to: Dr H. Holzel. 11 SOS.OOjO
Q 1992 The Hos,xtal
185
Infection
Socmy
186
H. Holzel
and M. de Saxe
malignancies or solid tumours S-12there are few publications relating to this problem in other specialist paediatric groups undergoing intensive care.‘z4 We therefore present our findings in a study of septicaemia occurring in patients on the six intensive care units at the Hospital for Sick Children, over a 33-month period, from 1 January 1988 to 30 September 1990, during which time 10 719 admissions were made to these areas. Patients
and
methods
Patients These were in-patients on the six units undertaking intensive care or intensive therapy in our hospital. At the onset of a febrile episode, blood cultures were taken from each available lumen of central venous catheters, from other indwelling arterial or venous catheters and from a peripheral vein. Microbiology Blood cultures were processed in the diagnostic laboratory using a non-radiometric automated system (Bat Tee NR 730, Becton Dickenson). Isolation and identification of organisms were carried out by standard bacteriological methods. Isolates were tested for susceptibility to appropriate ranges of antibiotics by disc diffusion using a rotating Stokes’ technique. Species of coagulase-negative staphylococci (CNS) were determined according to the nomenclature of Kloos and Schleiferi3 using a multipoint inoculation technique. l4 Isolates of the same species of CNS were considered to belong to different strains if they differed in resistance to two or more unrelated antibiotics. Analysis of results All positive blood cultures from children in the six intensive care areas were reviewed for the 33-month period, from 1 January 1988 to 30 September 1990. Those taken 72 h or more after patient admission were considered to indicate nosocomially-acquired septicaemia and were included in the analysis. Dejinitions A treatment period in our hospital is considered to be the administration of 10 days’ intravenous (iv) antibiotics for the treatment of serious sepsis. A clinically-significant septicaemic episode was defined as the isolation of a clinically and microbiologically appropriate organism (or organisms), from the blood of a pyrexial child, in whom no other cause of pyrexia was identified, and who subsequently received antibiotic therapy directed against the particular isolate(s). A single episode was defined as the isolation of the same organism(s) within the lo-day treatment period. In a few instances when cultures
Septicaemia
in paediatric
187
intensive-care
yielded the same organism in ongoing sepsis outside the lo-day was considered to be a continuation of the same episode. A distinct episode was defined as:
period, this
(a) The isolation of the same organism(s) outside the lo-day treatment period, if negative blood cultures intervened. (No attempt was made to distinguish between relapse and reinfection in these episodes.) (b) The isolation of different organism(s) more than 24 h after the initial positive blood culture. A polymicrobial episode was the isolation of more than one species either (a) in the initial positive blood culture, or (b) within 24 h of the initial positive culture, or (c) where the same combination of organisms was found on more than one occasion. Exclusions We excluded isolates of CNS, streptococci or Corynebacterium spp. obtained on a single occasion from peripheral blood cultures in children in whom corresponding blood cultures from intravascular catheters failed to yield the organism, or in patients in whom these devices were absent. Results
During the 33 months of the study, there were 10 719 admissions to the six units providing intensive care and therapy. A total of 624 episodes of septicaemia were documented in 464 children. The number of episodes in individual units, and their frequency, expressed as episodes of septicaemia per 100 admissions, are given in Table I. Sixty-one per cent of all episodes occurred on two units, the Haematology/Oncology unit (36%) and the Cardiothoracic surgical unit (25%). In contrast, only 7% occurred on the Renal dialysis/transplant unit. The frequency of septicaemia (per 100 admissions) was also highest on the
Table
I. Frequency Sick
of septicaemia in six intensive-care Children, Great Ormond Street
Unit
Admissions
Haematology/Oncology Neonatal surgery General medical surgery Cardiothoracic Infectious disease Renal dialysis/transplant All
units
1315 655 739 3452 1635 2923 10719
areas at the Hospital (1988-90)
Septicaemic episodes
Septicaemia per 100 admissions
227 75 67 154 56 45
17.3 11.5 9.1 4.5 3.4 1.5
624
5.8
for
188
H. Holzel
and M. de Saxe
Haematology/Oncology unit (17.3) and lowest (1.5) on the Renal dialysis/transplant unit. The range and distribution of organisms responsible for episodes of septicaemia in the different units are shown in Figure 1. Gram-positive bacteria were isolated in pure culture in 66% of the 624 episodes and ranged from 52% of episodes in the Infectious disease unit to 75% in the Neonatal surgical unit. Gram-negative organisms accounted for only 17% of episodes overall and were most common in the Infectious disease unit (29%). Polymicrobial episodes accounted for 13 % of septicaemic episodes. A small number of episodes were caused by anaerobes, Bacteroides fragilis or Clostridium spp. (in three and two patients respectively). Candida albicans was responsible for 15 episodes, Candida spp. for three episodes and Trichosporon beiglii for one episode in a patient with an underlying diagnosis of systemic lupus erythematosus. Candida albicans was also associated with 6% of the 80 polymicrobial episodes in our study. The organisms causing Gram-positive and Gram-negative septicaemia are presented in Tables II and III. Coagulase-negative staphylococci (CNS), were the predominant isolates. They were isolated in pure culture from 43% of episodes and from a further 6% in combination with other organisms. During the study period, the proportion of septicaemias due to CNS ranged from 68% in the Neonatal surgical unit to 32% in the Infectious disease unit (Table II). Identification of the individual strains of CNS showed that 85 % were Staphylococcus epidermidis; the remaining 15% comprised Staphylococcus haemolyticus, Staphylococcus hominis and Staphylococcus warneri (IO%, 3 % and 2% respectively). Staphylococcus aureus was found in pure culture in 10% of episodes. Streptococci caused 8% and enterococci 3% of episodes overall, and were most common in the Haematology/Oncology unit (19’/,). A range of other Gram-positive organisms were responsible for 15 episodes in patients on the Haematology and Cardiac units and included: Bacillus spp., Corynebacterium spp., micrococci and Rhodococcus sp. (six, five and three episodes and one episode respectively). During the period studied, patients on the six units experienced 107 episodes of Gram-negative septicaemia (17% of the total) (Table III) and Gram-negative organisms were associated with a further 6% of polymicrobial episodes. Gram-negative organisms in pure culture were most common as a cause of septicaemia in the Infectious disease unit (29%) and least common in the Haematology/Oncology unit (14%). Enterobacteriaceae were responsible for 54% of these episodes and Escherichia coli together with organisms of the Enterobacterlcitrobacter group predominated in all units. Pseudomonads were responsible for 35 (6%) of septicaemic episodes overall and caused 33 % of Gram-negative septicaemias. They were most common in the Haematology/Oncology unit (12 episodes) and the Cardiothoracic surgical unit (nine episodes). Pseudomonads of environmental origin were associated with 63% of the
Figure Great
unit
ICU
ICU
0
1. Organisms associated Ormond Street, London
Hoemotology/Oncology
surglcol
surgical
medico1
General
Neonatal
unit
disease
infectious
ICU
unit
dialysis/transplant
Cardiothorocic
Renal
50
with 624 episodes 1 January 1988-30
25 100
(12%)
( I I %)
Number
of septicaemia in 464 September 1990.
75
N=75
N = 67
(9%)
( 7 %)
N = 56
N = 45
150
receiving
of septicoemic
children
125
intensive
episodes
175
N = 154 (25%)
care
200
Streptococci
a
Gram-negatives
Other
q 0
at the
Hospital
225
for
250
N : 227
Polymlcrobial
organisms
Other
q
ta
Sick
Children,
‘/o)
aureus
(36
negative
Gram-positives
Staphylococcus
Coagulosestaphylococci
q
4
isolated
II.
Gram-positive
* Micrococci,
Number
Corynebacterium
of episodes
spp., Bacillus
Gram-positive organisms Coagulase-negative staphylococci Staphylococcus aureus All streptococci: S. sanguis/salivarius/mitis S. pneumoniaelothers Enterococci Other Gram-positive organisms* Polymicrobial (including Gram-positive organisms)
All
Organisms
Table
spp. and
organisms
Rhodococcus
sp. caused
2 67
3, 5, 6 episodes
75
5
13 154
:, 0
: 2
: 0
2
;
and
1 episode
227
12
:: 12 3 4 6
70 33
45
56
16
:,
i 2
52 32 11
Infectious disease
areas
respectively.
13
Fl
:,
58 40 11 7
Renal dialysis/ transplant
of septicaemia
in
in six intensive-care
episodes Haematology /Oncology
in children
% septicaemic General medical
episodes
72 58 4
Neonatal surgery
septicaemic
75 68 4
with
:: 10 5 4
Cardiothoracic surgery
associated
624
10
;
2; 10 8 7 *
All units
s
St k (D r
F m e s r E
isolated
III.
Gram-negative
sp.
spp.
spp. 0 0
1
154 presented
2(3) 75
16(7) 227
0 1
5(11) 45
: 0
2(4) 56
: 1
; 5
1 0
i
-__
-
624 were
W6)
z
3: 13 22 7
23 31 3
107(17)
All units
organ,sms
16(28)
Infectious disease
areas
as actual numbers of episodes wth which particular
3(2) 67
i
lw3)
:,
0 0
:
t
2 1
3 0 3
1: 7
4 2
2 4
9(20)
in Renal dialysis/ transplant
episodes
in six intensive-care
:
1
10 5
33(14)
____
of septicaemic Haematology /Oncology
in children
i
1 3
lO(15)
(%) General medical
episodes
z 2
i
5 4
4 9 i
12(16)
No.
septicaemic
Neonatal surgery
with
27(18)
-____
associated
Cardiothoracic surgery
organisms
As the total number of Gram-negative episodes was low, figures are associated. Percentages are given in parentheses.
sp. and Capnocytophaga
of episodes
* Agrobacterium
Number
Gram-negative organisms Enterobacteria: Eschevichia coli Klebsiella/Enterobacter/Citrobacter Proteus sp./Morganella sp. Salmonella typhimurium Pseudomonads: P. aeruginosa Other pseudomonads Alcaligenes/Acinetobacter/Moraxella Haemophilus injluenzae Others* Polymicrobial (including Gram-negative organisms)
All
Organisms
Table
c z
d
B e.
E
Et v ;i 6’
b
B L. Y w tj-
F w d c
192
H. Holzel
and M. de Saxe
episodes of pseudomonas septicaemia, whereas Pseudomonas aeruginosa was associated with only 37%. Five immuno-suppressed patients with tunnelled central venous lines had severe, ongoing problems due to P. aeruginosa (four patients) and Xanthomonas maltophilia (one patient). Almost a quarter of septicaemic children experienced recurrent episodes (i.e. 102 of 424), and 3% (44 of 424) suffered three or more episodes (Table IV). These recurrent episodes were most common in patients on the Haematology/Oncology, Infectious disease and General intensive care units and, in virtually all cases, affected immunodeficient patients who required repeated hospital admission and protracted hospital stay. Different bacterial species were associated with recurrent episodes in 62 children, different strains of CNS in 28 children, and recurrences due to the same organism occurred in only 12 of the 102 children (CNS in nine; S. aweus in two patients and Enterobacter sp. in one patient). Discussion
The study carried out at Great Ormond Street is unusual in that we review data on septicaemia in over 10 000 patients admitted to six specialist paediatric intensive care/therapy units. Information regarding the frequency of blood-stream infections in different groups of paediatric patients is usually restricted to studies of individual specialist units, e.g. Neonatal, Cardiothoracic surgical, Haematology/Oncology or General intensive care. In addition, most reports have been concerned with a wide range of nosocomial infections and have not provided detailed data on of nosocomial septicaemia, i.e. that septicaemia. 1,2,4*6Our definition occurring 72 h or more after admission, excludes the majority of community-acquired septicaemias.15 Blood-stream infection rates as high as 33 per 100 admissions have been reported from adult medical and surgical intensive care units, although much lower frequencies are more usual.‘s4 The rate for blood-stream infection in the Paediatric intensive care unit (PICU) at the Johns Hopkins
Table
IV.
Recurrent
Unit
Haematology/Oncology Infectious disease General medical Renal dialysis/transplant Neonatal surgery Cardiothoracic surgery All units
septicaemic
episodes Septicaemic children
in children
receiving
% experiencing more than episode
141 53; 36 64 136 464
;; 17 :z
122
one
intensive
care
% experiencing more than three episodes 6 9 :, 0 0 3
Septicaemia
in paediatric
intensive-care
193
Hospital over a 4-year period (1981-4) was 1.7 per 100 admissions and ranged from 0 to 4.5 per 100 admissions in the individual referring specialities. Among 19 admissions from unspecified referring specialities, septicaemia occurred in three (15.8%).’ In our study, the frequency of septicaemia expressed as episodes per 100 admissions was low in the Cardiothoracic unit (4.5) and high in the Haematology/Oncology unit (17.3), although patients in both units experienced large numbers of septicaemic episodes (136 and 141 respectively). This reflects the difference in patient populations in the two units; i.e. the large number of short-stay, immuno-competent patients admitted for cardiac surgery, in contrast to the immunocompromised children on the Haematology/Oncology unit who frequently require repeated hospital admission and protracted hospital stay. The frequency of septicaemia in our Neonatal surgical unit- 11.5 per 100 admissions-was in keeping with the findings of other workers.‘,4,6 Gram-positive organisms are now the most frequent cause of septicaemia in children undergoing intensive care in our hospital. Coagulase-negative staphylococci (predominantly S. epidermidis) are the most frequently isolated organisms; a finding in common with those of other paediatric and adult centres.‘Aa’6 Septicaemia due to CNS has been well documented in neonates undergoing intensive care,‘,‘,’ and in our Neonatal surgical intensive care unit, accounted for 68% of episodes. Staphylococcus aweus was associated with lo-12% of septicaemia in all units, with the exception of the Neonatal surgical and General medical units. These findings agree with data from a large study of nosocomial infection at the Hospital for Sick Children, Toronto (HSC-Toronto). 4 A high proportion of septicaemic episodes in patients on our Haematology/Oncology unit were associated with oral streptococci (12%). This raised the possibility of invasion during periods of mucositis in children undergoing intensive chemotherapy,’ and emphasized the need to review mouth care. Gram-negative organisms were identified in only 17% of our 624 septicaemic episodes and Enterobacteriaceae accounted for 9%; findings which agree with results from HSC-Toronto.4 Pseudomonas spp., in our patients, were responsible for 35 (6%) of septicaemic episodes overall. Pseudomonads of environmental origin were associated with 63% of the 35 episodes whereas P. aeruginosa was associated with only 37%. It was therefore of note, that in four of five immuno-suppressed patients with tunnelled central venous lines and severe ongoing pseudomonas septicaemia, P. aeruginosa was the causative organism. A matter of great concern was the unexpectedly high proportion of Gram-negative episodes in children on our Cardiothoracic surgical unit involving pseudomonads and Enterobacteriaceae (18%). In reviewing unit practice, it seemed likely that the use of femoral arterial and venous catheters for periods of more than 5 days was likely to have been an important contributory factor. Although our Neonatal surgical unit has a
H. Holzel
194
and M. de Saxe
particular interest in complex congenital abnormalities of the gastro-intestinal tract, septicaemia due to enteric organisms accounted for only 16% of episodes on this unit. In our study, Candida spp. accounted for 18 septicaemic episodes (3%) and C. albicans was responsible for 1.5 of these episodes. Similar data from the HSC-Toronto documented Candida spp. in 5% of cases of septicaemia. Anaerobic organisms were responsible for only five of the 624 episodes in our patients (two children on the Haematology/Oncology unit and three on the Renal dialysis/transplant unit) and Huemophilus influenxae was associated with five episodes in patients on the General medical, Cardiothoracic and Infectious disease units. In contrast, both anaerobes and H. infkenzae were absent from the list of pathogens associated with nosocomial bacteraemia in the Paediatric Intensive Care Unit in Toronto.3 The major risk factors in our paediatric intensive-care patients are those which are already well recognized, i.e. frequent patient handling, use of intravenous catheters, immunosuppression, prolonged hospital stay and frequent hospital readmission.‘~~9~“,16 Our data were derived from an ongoing study primarily designed to investigate the problem of severe sepsis caused by CNS in a wide paediatric setting. It confirmed the predominant role of CNS as the major cause of septicaemia in all six of our intensive care areas. The information has proved extremely useful in highlighting particular problems pertaining to the individual specialist units. It has also allowed us to guide and support measures of infection control and to prioritize the issues which require further prospective detailed investigation. We are grateful Great Ormond Health Research
to the staff of the Microbiology Department, Street, London, for their continued support Appeal Trust for the financial aid for our
Hospital for Sick Children, of this project and to the Child research programme.
References 1. Merritt 2.
WT, Stephens M. Nosocomial infections in the pediatric Rogers MC, Ed. Text Book of Pediatric Intensive Care. Baltimore: 1987; 2: 75.5-785. Allen U, Ford-Jones EL. Nosocomial infections in the pediatric
J Infect Control
3. Milliken 4. 5. 6. 7. 8.
intensive Williams patient:
care unit. In: and Wilkins an update.
Am
1990; 18: 176-193.
J, Tait GA, Ford-Jones EL, Mindorff CM, Gold R, Mullins G. Nosocomial infections in a pediatric intensive care unit. Crit Care Med 1988; 16: 233-237. Ford-Jones EL, Mindorff CM, Langley JM et al. Epidemiologic study of 4684 hospital-acquired infections in pediatric patients. Pediutr Infect DisJ 1989; 8: 668-675. Leonard EM, van Saene HKF, Shears P, Walker J. Pathogenesis of colonisation and infection in a neonatal surgical unit. Crit Care Med 1990; 18: 264269. Brown RB, Hosmer D, Chen HC et al. A comparison of infections in different ICUs within the same hospital. Crit Care Med 1985; 13: 472-476. Freeman J, Platt R, Sidebottom DG, Leclair JM, Epstein MF, Goldman DA. Coagulase-negative staphylococcal bacteremia in the changing neonatal intensive care unit population. JAMA 1987; 258: 2548-2552. Eshali H, Ringertz S, NystrGm S, Faxelius G. Septicaemia with coagulase negative staphylococci in a neonatal intensive care unit. Acta Paediatr Stand 1989; Suppl. 360: 127-134.
Septicaemia
in paediatric
intensive-care
195
9. Morrison VA, Peterson BA, Bloomfield CD. Nosocomial septicemia in the cancer patient: the influence of central venous access devices, neutropenia, and type of malignancy. Med Pediatr Oncol 1990; 18: 209-216. IO. Shapiro D, Wald ER, Nelson KA, Spiegelman KN. Broviac catheter-related bacteremia in oncology patients. Am J Dis Child 1982; 136: 679-681. 11. Wade JC, Schimpff SC, Newman KA, Wiernik PH. StuphyZococcus epidermidis: an increasing cause of infection in patients with granulocytopenia. Ann Intern Med 1982; 97: 503-50s. 12. Friedman LE, Brown AE, Miller DR, Armstrong D. Staphylococcus epidermidis septicemia in children with leukemia and lymphoma. Am J Dis Child 1984; 138: 715-719. 13. Kloos WE, Schleifer KH. Simplified scheme for routine identification of human Staphylococcus species. J Clin Microbial 1976; 1: 82-88. 14. Geary C, Stevens M, Sneath PHA, Mitchell CJ. Construction of a database to identify Staphylococcus species. 7 Clin Path01 1989; 42: 289-294. 15. Bryan CS, Reynolds KL, Derrick CW. Patterns of bacteremia in pediatrics practice: factors affecting mortalitv rates. Pediatr Infect Dis 1984: 3: 312-316. 16. Martin MA, P?aller MA, Wenzel RP. Coagulase-negative staphylococcal bacteremia: mortality and hospital stay. Ann Intern Med 1989; 110: 9-16.
