Bacteraemia at the University Hospital of the West Indies - a report of 222 cases

Journal of Infection (1985)

Bacteraemia

I0, 126-142

at the University Hospital of the West Indies a r e p o r t o f 222 c a s e s D. E. M a c f a r l a n e a n d V. R. N a r l a

Department of Microbiology, University of the West Indies, Mona, Kingston, Jamaica Accepted for publication 18 October 1984 Summary A total of 222 cases of septicaemia was recorded at the University Hospital of the West Indies between June 1982 and June 1983. This gave an overall incidence of I6.I per IOOOadmissions. The 233 bacterial strains isolated comprised IOO Gram-positive and 133 Gram-negative organisms with Klebsiella pneumoniae, Streptococcus pneumoniae and Staphylococcus aureus being the most frequent. Highest rates of septicaemia were recorded in patients less than I year and over 5o years of age. Septicaemia caused by Gram-positive organisms was predominantly a disease of children whereas that caused by Gram-negative organisms arose more often in neonates and in patients over 5o years of age. A predisposing factor was noted in lO4 patients of whom 42 had neoplastic disease. The most frequently identified initial sites of infection were the respiratory tract, the gastro-intestinal tract and the meninges. Most blood stream infections were community-acquired, three quarters of all septicaemic patients being admitted to the departments of medicine or paediatrics. There were i I cases of polymicrobial septicaemia caused predominantly by Gram-negative organisms in patients with underlying disease. Appropriate antimicrobial drugs were administered to 57 ~o of septicaemic patients whereas 17 % received superfluous antimicrobial therapy. In those patients who received inappropriate antimicrobial therapy there was a marked increase in mortality. Forty of 61 deaths were attributed to septicaemia. Mortality from septicaemia caused by Gram-negative organisms was 21 ~o compared with 13 ~o for that caused by Gram-positive organisms. The organisms associated with the highest case fatality rates were Escherichia coli, 53 ~o ; Enterobacter sp., 27% ; and fl-haemolytic streptococci 24 ~o. There were no deaths from septicaemia caused by Haemophilus influenzae, Salmonella sp. or Serratia sp. The highest mortality rates were associated with neoplastic disease, diabetes, polymicrobial septicaemia, urinary tract infections and old age. Introduction M o s t recent publications on bacteraemia have concentrated on infections caused by G r a m - n e g a t i v e organisms 1-6 on selected populations 7-1°, or on particular organisms, n-13 T h e r e have been few general studies of septicaemia in a representative hospital population which included a significant n u m b e r of paediatric patients. M o s t of the present information on septicaemia is based on studies c o n d u c t e d in major teaching or research institutions in the more developed countries. Extrapolation of this data to hospitals in developing countries m a y be misleading because of the different effect of factors such as u n d e r l y i n g disease, endemic pathogens, the population f r o m which patients are derived and antimicrobial prescribing practices. T h e present study was done o163-4453/85/o2oI26+o7 $o2.oo/o ~ 1985 The British Society for the Study of Infection 5-2

Bacteraemia in the West Indies

I27

in order to assess the effect of these and other factors on the incidence of bacteraemia and on the mortality from bloodstream infections in the University Hospital of the West Indies (U.H.W.I.). Materials and Methods

T h e hospital charts for all patients with positive blood cultures were reviewed for the period June I 9 8 2 - June I983 and information relevant to this study was transferred to prepared record sheets. Each positive blood culture was evaluated so as to determine whether it reflected bacteraemia or contamination. T h e evaluation was based on the number of positive cultures, the identity of the organisms isolated, the results of cultures from other sites, clinical signs and symptoms, as well as total and differential white blood cell counts. T h e course of each episode of bacteraemia was followed until its conclusion. Bacteriology

Blood was collected by venepuncture after preparation of the skin with iodine and alcohol. Each Io-2o ml blood sample was divided between two blood culture bottles containing respectively, 50 ml brain-heart-infusion broth and 5o ml thioglycollate broth. After I8-24 h incubation at 37°C, a sample from each bottle was examined by means of the Gram stain and another sample subcultured on to blood, 'chocolate' and MacConkey agar. Incubation of bottles continued for 7 days with daily macroscopic examination and terminal subculture. Organisms isolated were identified by means of standard biochemical tests or by the API zoE system. C o m m u n i t y or n o s o c o m i a l b a c t e r a e m i a

Bacteraemias which developed within 3 days after admission, or before admission to hospital were designated as community-acquired; those with later onset were considered to be hospital-acquired. P r e d i s p o s i n g factors

Patients who were compromised by underlying disease or had other recognised factors predisposing to septicaemia were identified. T h e significance of these risk factors was assessed in terms of overall numbers of episodes and increased frequency of bacteraemia. Focus o f infection

An attempt was made to identify a focus of infection in each case. This was based on cultural results or on clinical evidence. T h e source of bacteraemia was categorised under one of the following general headings; respiratory tract, urinary tract, gastrointestinal tract, cardiovascular system, skin and soft tissue, bone and joint, and meninges. A n t i m i c r o b i a l therapy

Antimicrobial therapy was assessed before and after results of blood cultures were available. Antibiotic therapy was considered appropriate if it conformed to the general guidelines laid down in a standard text 14 and if the organism

128

D. E. M A C F A R L A N E

A N D V. R. N A R L A

recovered from the blood was sensitive to the antimicrobial drugs prescribed. Inappropriate therapy was defined as failure to prescribe antibiotics, wrong first choice, wrong dosage or failure to make an appropriate response to blood culture and sensitivity reports. Superfluous therapy was defined as the administration of unnecessary antimicrobial drugs in addition to an appropriate first choice, or continuation of unnecessary drugs on receipt of sensitivity test results derived from positive blood cultures. Results During the i 2 - m o n t h period of this study there were I3 688 admissions to the U . H . W . I . These included 222 cases of bacteraemia giving an overall rate of I6.I per Iooo admissions. T h e 233 bacterial strains recovered from blood cultures included IOO Gram-positive and 133 Gram-negative organisms. Those most frequently isolated were Klebsiella pneumoniae, Streptococcus pneumoniae and Staphylococcus aureus. T h e relationship between organisms recovered from the blood, age of the patient, focus of infection, predisposing factors and mortality is summarised in Tables I - I I I . Septieaemia according to causative organism Staphylococcal bacteraemia was primarily a disease of children with I9 of 27 patients being under 20 years of age. T h e r e was an increased incidence in neonates and in patients over 5o years of age. Skin and soft tissue infections accounted for Io of the 2o identified sources of septicaemia. T h e r e was no significant association with underlying disease and only I of 27 such patients died. Staphylococcus epidermidis was recovered from the blood of five patients, four with malignancy and one with systemic lupus erythematosus (SLE). One patient with acute lymphoblastic leukaemia died. Septicaemia caused by Streptococcus pyogenes (Group A streptococci) was more prevalent in adults, 8 of I3 such patients being over 3o years of age. It was not found in neonates or infants. T h e r e was a strong association with underlying disease particularly malignancy; two of three deaths were in patients with neoplastic disease. Streptococcus agalactiae (Group B streptococci) was recovered from the blood of nine neonates, one infant and two adults; two of three deaths were in neonates. T h e respiratory tract and meninges were the main sources of infection. Other streptococci (6 of Group D, 3 of Group G, 2 of Group C and 2 microaerophilic streptococci) were responsible for I3 cases of septicaemia. T h e r e was no association with any particular age group or with underlying disease, and no predominant source of infection. One patient with endocarditis caused by Group G streptococci died. Streptococcus pneumoniae septicaemia was predominantly a disease of infants and young children, with 18 of 27 patients being under 5 years of age and only two being over 50 years of age. This organism, which was invariably associated with respiratory tract infection and meningitis, was responsible for five deaths. One third of such patients had underlying disease. Haemophilus influenzae septicaemia was almost exclusively a disease of infants and children under 5 years of age, with I5 of I6 patients being in this age group. T h e r e were IO patients with meningitis, three with respiratory tract

Bacteraemia in the West Indies

129

e,'t

A

.2

,.o~

~I

~

~t

~ L

- ~ - e ~ ,p

%)

p,.

,e¢

0

~3

~1

"*~

~ t ~ . ~ l

r~ ~ ' ' ~ ~-

t'-.

".0

'~

09

~

I ~

~ I '~'*

tl

~ ° " * L-

~3

,--'-x t"-

o~ (',,1 %) $.

t~

o

Z e~ fJ

~LB~ I ',o

~OmN

o

~OOF~

13 0

D.E. MACFARLANE

AND

V. R. N A R L A

,~o~£~S~.~o~o

~i

~=~o~~

~ ~,~

o~I

I I~1

s ~,

0

a~

0

I I ~ I,~1

'O

,-1

~III

~I~I ,.0

~

~

~

¢q

el

I~

~

O0

I ~

c~

S

~.~°

~uu

~.00

~ o o ©

o

~rao~6

~

¢q

Bacteraemia in the West Indies

~"~ ~

~'~ ~

o

~[~

~'~ 0~ ~

~

~

~

~3

C~ O

z~

e~

I t ~

I I t I I t ~I

C~

~D

o~

I32

D.E. MACFARLANE

A N D V. R. N A R L A

infections, two with osteomyelitis, two with underlying disease but no deaths caused by this organism. Klebsiella pneumoniae, the most frequent organism isolated, was recovered from the blood of 39 patients including 8 neonates and 14 patients over 5o years of age. T h e gastrointestinal tract and urinary tract were the main sources of infection, 54 % of patients having identifiable risk factors, particularly neoplastic disease or post-operative infections. T h e IO deaths from klebsiella septicaemia included 4 patients with malignancy and I neonate. Escheriehia coli was cultured from the blood of 19 patients, 9 of whom were more than 5o years of age. There was a strong association between E. coli septicaemia and underlying disease. There were five patients with malignancy, three with diabetes and two with SLE. T h e urinary tract was the most frequently identified source of the blood-stream infection. Mortality from E. coli septicaemia was 53 %. Enterobacter septicaemia was diagnosed in I I patients, 2 with post-operative infections and 5 with underlying diseases. A similar pattern was observed in acinetobacter septicaemia with three of seven infections in patients with malignancy, 2 in patients with post-operative infections and 2 infections in premature infants. Four of six serratia septicaemias were in patients with post-operative infections. O f the seven additional strains of Enterobacteriaceae recovered from blood cultures, three were Proteus mirabilis while Aeromonas sp., Arizona sp., Alcaligenes sp. and Providenda sp. accounted for the remaining four. All these organisms were cultured from the blood of adult patients, particularly those over 5o years of age, with malignancy, post-operative infections or some other predisposing factor. Salmonella typhi was isolated from the blood of 14 children and young adults with typhoid fever. There was no association with underlying disease. Other Salmonella sp. accounted for five additional cases of septicaemia. Four of these, S. in/antis, S. enteritidis, S. ohio and S. heidelberg were isolated from children with sickle-cell disease. A multi-resistant strain of S. ohio was recovered from the blood of a I-year-old boy who was being treated with chloramphenicol for H. influenzae meningitis. T h e source of infection was an infected drip site. There were no deaths from salmonella septicaemia. Anaerobes were the causative organisms in four patients with bacteraemia. T w o patients with neoplastic disease had Baeteroides/ragilis septicaemia. An anaerobic streptococcus was cultured from the blood of one patient with S L E and a fourth patient with nephrotic syndrome was infested with both E. coli and B. fragilis. All four patients died. Neisseria spp. were isolated from the blood of two patients; N. gonorrhoeae from a patient with gonococcal arthritis and N. meningitidis from a patient with meningococcal meningitis. A single fatal case of Listeria monocytogenes septicaemia was recorded in a 24-year-old woman with SLE. S e p t i c a e m i a according to age There was a high incidence of bacteraemia patients being children under I year of age; minimum of 16 patients in the 21-3o years a slight increase in the 31-4o years age group

in neonates and infants, with 54 thereafter numbers declined to a age group (Table I). There was and a marked increase to 58 cases

Bacteraemia in the West Indies

133

Table IV Incidence of septicaemia in relation to predisposing factors Predisposing factor SLE Malignancy Malnutrition Sickle-cell disease Diabetes Prematurity

Total patients

Septicaemic patients

Incidence of septicaemia per IOOO

27 465 85 199 377 561

8 42 7 ii 12 9

296 90"3 82 55"2 3 i-8 16

in patients over 5o years of age. Septicaemia caused by Gram-positive organisms was primarily a disease of children with 53 patients being under io years of age. Thereafter the incidence decreased to 5 cases in the zI-3O years age group followed by an increase to i I cases in the 31-4o years age group and 14 cases in patients over 5o years of age. Septicaemia caused by Gram-negative organisms was predominantly a disease of the elderly with 43 cases in patients over 5o years of age. T h e r e was also an increased incidence in neonates. T h r e e organisms, S. aureus, S. agalactiae and K. pneumoniae accounted for 22 of 26 cases of neonatal septicaemia. Haemophilus influenzae and S. pneumoniae tended to predominate in the 1-5 year age group, Salmonella sp. was the most frequent isolate in the 5-IO years age group while S. aureus was an important cause of septicaemia in children of all ages. T h e r e was no significant association between any particular organism and septicaemia in the age group lO-5o years, but in patients over 50 years of age Gram-negative organisms accounted for 76 % of all bacteraemias.

Predisposing factors Predisposing factors were identified in 47 % of 222 patients with septicaemia (Tables II & IV). Malignancy was diagnosed in 19 % of all septicaemic patients and was the most significant risk factor in terms of numbers and mortality. Streptococcus sp. and coliforms accounted for most infections in patients with neoplastic disease. T h e spectrum of organisms isolated from the blood of diabetics and of patients with SLE was similar, but mortality was significantly higher in diabetic patients (Table II). Salmonella sp. and S. pneumoniae accounted for septicaemia in eight of eleven patients with sickle cell disease. Gram-positive organisms were responsible for septicaemia in six of seven malnourished patients. In all 13 patients with post-operative septicaemia Gram-negative organisms were responsible as they were in three of nine premature infants with septicaemia. T h e r e was a significant association between predisposing factors and septicaemia caused by organisms such as S. epidermidis, S. pyogenes, E. coli, K. pneumoniae and other Enterobacteriaceae. Conversely, bacteraemia caused by S. aureus, S. agalactiae, S. pneumoniae, H. influenzae and S. typhi was not normally associated with underlying disease.

I34

D.E. MACFARLANE

A N D V. R. N A R L A

Focus o f infection

A source of infection was identified in 68 % of 222 patients with septicaemia (Table III). T h e respiratory tract was the most frequently identified source in 35 patients, followed by the gastrointestinal tract in 32, meninges in 28, urinary tract in 22, skin and soft tissue in 2I, bone and joints in 9 and the cardiovascular system in 3. Streptococcus pneumoniae was responsible for 54 % of respiratory tract-associated septicaemias, H. influenzae and S. pneumoniae for 68 °/o of meningitis-associated septicaemias, Salmonella.sp. for 47 % of gastrointestinal tract-associated septicemias, Enterobacteriaceae for 73 % of urinary tractassociated septicaemias and S. aureus and fl-haemolytic streptococci for 7I % of skin and soft tissue-associated septicaemias. A source of infection was identified in 97% of S. pneumoniae septicaemias, 94% of H. influenzae septicaemias, 89 % of salmonella septicaemias and 74 % of S. aureus septicaemias compared with 57 % for Klebsiella and 5o % for other Enterobacteriaceae. Mortality was highest at 27 ~o for urinary tract-associated septicaemias followed by respiratory tract-associated 23 ~o, gastrointestinal tract-associated 16 % and meningitis-associated x4%. T h e r e were no deaths from bone and jointassociated septicaemias but a 5 % mortality from skin and soft tissue-associated septicaemias. Place o f acquisition

Approximately two-thirds of the cases of bacteraemia in this study were acquired in the c o m m u n i t y (Table V). All cases of S. pneumoniae and H. influenzae septicaemia were community-acquired as were most salmonella septicaemias. Infections in neonates contributed to a significant proportion of the 4I % hospital-acquired S. aureus and 34 % hospital-acquired streptococcal septicaemias. T h e Enterobacteriaceae, with the exception of E. coli and Salmonella sp. accounted for most hospital-acquired septicaemias. T h r e e quarters of all septicaemic episodes arose in the departments of Paediatrics and Medicine, the remainder being distributed among the departments of General Surgery (42), Obstetrics and Gynaecology (3), Orthopaedics (3), Ophthalmology (2), Dermatology (2) and Ear, Nose and T h r o a t Surgery (2). C o m m u n i t y - a c q u i r e d organisms such as H. influenzae, S. pneumoniae and Salmonella sp. caused septicaemia almost exclusively in patients being cared for by the medical and paediatric services. Staphylococcus aureus and Streptococcus sp. were recovered mostly from medical and paediatric patients whereas the Enterobacteriaceae (excluding Salmonella sp.) were most often isolated from the blood of medical and surgical patients. P o l y m i c r o b i a l infections Sixteen of 22 organisms isolated from II polymicrobial infections were m e m b e r s of the Enterobacteriaceae, with Klebsiella sp. and Enterobacter sp. being the most frequent isolates. All ~I patients with polymicrobial septicaemia had some predisposing factor; 4 had neoplastic disease, 2 had post-operative infections, 2 had diabetes, one had SLE, one was a four-year-old child with 6o % burns and one was a pre-term infant. Eight of I I patients were over 40 years of age. T h e case mortality rate was 45 %.

I35

B a c t e r a e m i a in the W e s t Indies

Table V S e p t i c a e m i a according to organism a n d place o f acquisition Organism Staphylococcus aureus Staphylococcus epidermidis Streptococcus pneumoniae Streptococcus sp. Haemophilus influenzae Klebsiella sp. Escherichia coli Salmonella sp. Other

Enterobacteriaceae Others Total (%)

Hospital- Communityacquired acquired Medicine Paediatrics Surgery

Others

II 5

16 o

6 3

17 o

4 2

o o

o

29

9

20

0

O

I3

25

I2

I8

5

3

o

16

o

15

I

o

2,I 6 2

I8

IO

8

I8

3

13

II

2

5

I

17

7

I2

o

o

24

7

5

IO

6

4 86 (37)

6 I47 (63)

io

7 75 (32)

2 99 (43)

I 46 (2o)

o I3 (5)

Antimicrobial therapy Antimicrobial therapy was considered appropriate in 5 7 % of 222 patients. Superfluous antibiotics were administered to 17 % o f patients, with gentamicin being the m a i n offender, followed by penicillin, clindamycin, co-trimoxazole and metronidazole. Empirical first-choice therapy was considered inappropriate in 13 % of patients, and there was no response, or an inappropriate response, to positive blood culture reports in 7"6 % of patients. Eleven of I2 patients who received incorrect dosage of antimicrobial drugs were being treated with approximately half the r e c o m m e n d e d dose of aminoglycosides. T h r e e patients with septicaemia did not receive any antimicrobial therapy. T h e mortality rate for patients receiving appropriate or superfluous therapy was 12"7 %. Mortality increased to 2 1 % for wrong first choice, 29% for wrong response, 41"6% for w r o n g dose and lOO% for three patients who did not receive any antimicrobial drug. Mortality T h e r e were 61 deaths among 222 bacteraemic patients giving an overall mortality rate of 27-4%. F o r t y deaths were attributed to septicaemia. T h e remaining 21 patients died o f other causes although septicaemia m a y have been a contributing factor. T h e mortality rate for septicaemia caused by G r a m positive organisms was I3 % c o m p a r e d with 2I % for septicaemia caused by G r a m - n e g a t i v e organisms and 27 % for septicaemia caused by the EnterobacterJaceae (Table I). Organisms associated with the highest case mortality rates were E. coli ( 5 3 % ) , Enterobacter sp. (27%), Klebsiella sp. (230/0), fl-haemolytic streptococci (24 % ) and S. p n e u m o n i a e (i 7"2 % ). T w o organisms, E. coli and K . p n e u m o n i a e , accounted for 48 % o f all septicaemic deaths. T h e r e were no deaths f r o m H . influenzae, salmonella or serratia septicaemia, and only I of 27 patients with S. aureus septicaemia died.

I36

D. E. M A C F A R L A N E

A N D V. R. N A R L A

Mortality according to age was less than 5 % in children under Io years of age, and more than 3o % in patients over 4o years of age (Table I). T h e comparatively high mortality rates of 17"6 ~o in the I I-2o-year-old age group and 43"7 % in the 21-3o-year-old age group were caused by a high incidence of malignancy-related deaths. Twenty-seven of 4o patients who died from septicaemia had underlying disease, in particular malignancy and diabetes (Table I I). T h e most frequent identified source of infection in fatal septicaemia was the urinary tract, followed by the respiratory tract, the gastrointestinal tract, and the meninges (Table III). T h e mortality rate among I8I patients with community-acquired bacteraemia was I9.2% compared with I3"5% among 81 patients with nosocomial bacteraemia. Discussion

T h e incidence of septicaemia has varied considerably in reports from various institutions. 1~-2° Rates of 3"4 and 5 per IOOO admissions have been recorded for community hospitals, 18, 19 compared with 28 per ~ooo admissions for a large city hospital. 16 T h e increased incidence of septicaemia at large city or teaching hospitals has been attributed to a greater n u m b e r of patients with serious underlying disease or other risk factors. In the present study the incidence of septicaemia was similar to the I6"5 per IOOO admissions reported at the University of Colorado HospitaP ~but higher than the 6"25 per Iooo admissions recorded at a London teaching hospital) ° Large city or teaching hospitals have been associated with an increased incidence of nosocomial bacteraemias. Although U.H.W.I. is a teaching and referral centre, most bacteraemias were community-acquired, the percentage of nosocomial bacteraemias being similar to the figures reported by Scheckler and colleagues for a community hospital. 18T h e size of the paediatric population and the endemicity of certain bacterial pathogens in the community may have been significant in this respect. At U.H.W.I., community-acquired organisms such as S. pneumoniae, H. influenzae and Salmonella sp. accounted for 65 % of paediatric septicaemias and 32 % of all blood-stream infections. In a study by Weinstein and colleagues, which excluded paediatric patients, 65 % of septicaemias were hospital-acquired, S. pneumoniae, H. influenzae and Salmonella sp. accounting for I o % of bacteraemiasJ 5 These authors also reported a predominance of nosocomia! S. aureus bacteraemias, whereas at U.H.W.I. most S. aureus infections were community-acquired as was the case in Scheckler's study of a community hospital.18 Previous authors have noted a low incidence of community-acquired bacteraemias caused by members of the Enterobacteriaceae with the exception of E. coli. TM 16 A similar pattern was observed at U.H.W.I. where 66 % of E. coli bacteraemias were communityacquired compared with 35 % for other members of the Enterobacteriaceae. E. coli is the major pathogen in community-acquired urinary tract infections and most E. coli bacteraemias are urinary tract associated, therefore the preponderance of community-acquired E. coli bacteraemias is not surprising. Varying rates of bacteraemia among the various hospital services have been recorded in previous reports with medical services having the highest overall rate.15, 16, 21 In the present study, the paediatric department was the source of

Bacteraemia in the West Indies

I37

most cases of bacteraemia. T h e incidences of septicaemia in the departments of medicine and general surgery were 34"5 per Iooo and I4 per Iooo respectively. These rates are similar to the corresponding rates of 36"4 per Iooo and I6"4 per IOOO at Boston City Hospital 16 but higher than those recorded elsewhere.iS, 21.22 T h e incidence ofo.6 per Iooo admissions for the departments of obstetrics and gynaecology was considerably lower than the rates of 5 and 7 per Iooo admissions recorded in two previous studies. 7. 1~ T h e r e were marked differences in the frequency with which particular organisms caused septicaemia among the different hospital services. H. influenzae occurred almost exclusively in paediatric services, whereas Salmonella sp. were confined to medicine and paediatrics. T h e low incidence of these organisms in other studies makes comparison difficult. Staphylococcus aureus, S. pneumoniae and Streptococcus sp. occurred most frequently in the paediatric services, whereas Weinstein and colleagues recovered these organisms mostly from medical and obstetric patients. 15T h e high incidence ofE. colibacteraemias in the medical services can probably be attributed to the frequency of community-acquired E. coli infections, compared with other members of the Enterobacteriaceae, which were mainly hospital-acquired and found on medical and surgical units. A source of infection was identified in 63 % of septicaemias at U.H.W.I. which is similar to the figures of 62 % and 69 % reported elsewhere. TM 2', T h e respiratory tract was the main source of infection, and this is in agreement with the findings of Weinstein and colleagues25 Other authors, however, have identified the genito-urinary tract as the major source of infection in bloodstream infections.i, 19, 21, 22 Septicaemia associated with the gastrointestinal tract, the central nervous system or skin and soft tissue was more frequent at U.H.W.I. than other hospitals 1, 15.21, 22 possibly because of a larger paediatric population, and a greater n u m b e r of salmonella infections. T h e figure of 4 % for bone and joint associated septicaemias is similar to the 3"4 % recorded by Scheckler at a community hospital. 18 T h e frequency with which particular organisms are recovered from blood cultures at different institutions will depend on several factors, including the type of institution and the patient population. Haemophilus influenzae, a community-acquired pathogen, and an important cause of septicaemia in young children, was recovered from 7 % blood cultures at U.H.W.I. compared with 39% at a children's hospital 8 and I % in a study which effectively excluded paediatric patients. 15 Salmonella sp. are endemic in Jamaica and were recovered from 8 % of blood cultures, whereas in areas with a low incidence of salmonella infections, these organisms are not a significant cause of septicaernia25-2° Pseudomonas was recovered from 1% of blood cultures at U.H.W.I., which is less than the 3'3% and 3"7% recorded in two previous studies16, 18 and considerably less than the 7"2 ~o reported by Weinstein and colleagues25 Pseudomonas septicaemia is primarily a hospital-acquired infection which has been associated with malignancy, instrumentation of the urinary tract, immunosuppressive therapy, indwelling venous and arterial catheters, tracheostomy and the use of mechanical ventilators. 2~ It is, therefore, likely to be more common in institutions with sophisticated high technology and having a high proportion of severely ill patients. A previous report has identified infections of the genito-urinary tract in obstetric and gynaecological

I38

D. E. M A C F A R L A N E

A N D V. R. N A R L A

patients as an important source of anaerobic septicaemia. 15 T h e low overall incidence of septicaemia in the departments of obstetrics and gynaecology at U . H . W . I . may be responsible for the corresponding low incidence of anaerobic septicaemia. In most recent studies, Gram-negative organisms have been isolated more often from blood cultures than have Gram-positive organisms. Although E. coli, S. aureus, Klebsiella sp. and S. pneumoniae have been identified as the commonest causes of septicaemia, E. coli and Klebsiella sp. have accounted for more than half the Gram-negative blood culture isolates:l~, 16,19, 21, 22 Other organisms commonly associated with septicaemia have included Proteus sp., Enterobacter sp., Serratia sp. and Pseudomonas sp.1, 2, 21 Similar findings pertained in the present study apart from a lower incidence of E. coli and pseudomonas septicaemias. T h e frequency of bacteraemia according to age followed the same general pattern reported previously with higher rates in children and patients over 50 years of age. At U . H . W . I . 44"6 % of septicaemias were in children under Io years of age and 2 3 ~o in patients over 5o years of age compared with 2o and 55 % respectively at Boston City Hospital. Patients in these two age groups, however, accounted for three-quarters of all septicaemias at both hospitals. 16 S. agalactiae, S. aureus and Enterobacteriaceae were responsible for all cases of neonatal septicaemia. These findings are similar to those reported elsewhere.9, 2~ Escherichia coli has been identified as an important cause of neonatal septicaemia and meningitis 1°, 25 but this organism was not recovered from neonatal blood cultures during the I2 months of this study. T h e high incidence of klebsiella septicaemia was the result of an outbreak in the intensive care nursery with 6 cases and I death. Haemophilus influenzae and S. pneumoniae tend to predominate in most studies of bacteraemia in children although the relative incidence of these two organisms may vary. s In a recent study of bacteraemia in children, H. influenzae was isolated twice as often as S. pneumoniae and these two organisms accounted for 57 % of childhood septicaemias. 26 At U . H . W . I . , S. pneumoniae was recovered more often than H. influenzae and these two organisms accounted for 6o 0/0 of positive blood cultures in the age group I month to 5 years. An increased incidence of septicaemia, particularly that caused by Gramnegative organisms in patients over 5o years of age, has been noted by previous authors. This is confirmed in the present study where 2 7 ~o of septicaemias were in patients over 5o years of age and 7 o % of these were caused by Gram-negative organisms. 1-4' 27 Various predisposing factors have been associated with an increased incidence ofbacteraemia including malignancy, diabetes mellitus, corticosteroid therapy, administration of antimetabolites and surgery. 1,1~,2s At U . H . W . I . a predisposing factor was identified in 47 % of patients, I9 % having some form of malignancy. Weinstein and colleagues identified a predisposing factor in 76 % of predominantly hospital-acquired bacteraemias in adults, I 6 % with neoplastic disease. 15 In a review of nosocomial bacteraemia, Spengler and colleagues recorded a I9"1% incidence of malignancy in bacteraemic patients. ~2 In both these studies, the mortality for bacteraemic patients with malignancy was 4 2 % which is similar to the 40 % recorded at U . H . W . I . Klebsiella sp., E. coli, Pseudomonas sp. and S. aureus have been identified as the most frequent

Bacteraemia in the West Indies

I39

isolates in bacteraemic patients with malignancy. 29 Klebsiella sp. and E. coli were the causative agents in 40% of malignancy-related bacteraemias at U.H.W.I., but S. aureus was not isolated from any and Pseudomonas sp. was isolated only twice. In a recent review of septicaemia at two large city hospitals, 9% of predominantly adult septicaemic patients had diabetes. T h e case mortality rate was 30 % if other predisposing factors were excluded, but 53 % in diabetic patients requiring surgery, with renal failure or on corticosteroid therapy. 15 At U.H.W.I., 5"4 % of bacteraemic patients had diabetes and the case mortality rate was 42 %. I f patients under IO years of age are excluded, however the incidence increases to 9"9 %, which is similar to the figure reported above for adult patients. T h e importance of SLE as a predisposing factor is probably associated with the use of corticosteroids or other immunosuppressive agents in these patients although defects in the immune system may be a contributing factor. All eight such patients in the present study had received steroid therapy before the onset of bacteraemia. T h e incidence of septicaemia in this group of patients was 18"5 times the hospital average which is much higher than in any other group of patients with underlying disease. T h e prevalence of sickle-cell disease is I per 32o in Jamaica and previous reports have detected a predisposition to Salmonella and pneumococcal septicaemia in patients with this disease2 °, 31 This is confirmed by the present study in which in eight of eleven patients with sickle-cell disease septicaemia was caused by these two organisms. Problems associated with incorrect use of antimicrobial drugs in hospital patients are well recognised. 32-3a Increased mortality associated with wrong antimicrobial therapy is not surprising and has been noted by several authors. 6,15, 19, 35 A reluctance to discontinue additional antibiotics which were considered superfluous on the basis of blood culture reports was a significant feature of the present study. Moreover it is interesting to note that Kreger and colleagues have previously failed to demonstrate any advantage in combination therapy in the treatment of septicaemia. 6 T h e same authors also observed a high mortality in septicaemic patients treated with gentamicin. A similar high mortality was associated with the use of subtherapeutic doses of gentamicin at U.H.W.I. T h e 4"5 % incidence of polymicrobial bacteraemia in the present study is similar to the 5 % reported by Warren and colleagues~6 and the 6"7 % reported by Kotin 37 but m u c h lower than the I8 % recorded in two other studies. 11, x6 Bodey and colleagues have shown a strong association between polymicrobial septicaemia and acute leukaemia. T h e incidence of these infections may therefore depend on the patient population. T h e same author also noted a predominance of Gram-negative organisms in polymicrobial infections28 At U.H.W.I. only 2 of r r patients with polymicrobial bacteraemia had leukaemia. All I i patients, however, had some predisposing factor and I6 of the 22 organisms isolated were members of the Enterobacteriaceae. T h e high case mortality rate of 45 % for this group of patients was consistent with previous reports. 1-3 Mortality rates for septicaemia have varied from a low of I3 % to a high of 57 % .3. a9 These differences may be attributed to the inconsistent effect of various factors known to affect case mortality rates. In a study by Weinstein

I40

D. E. MACFARLANE AND V. R. NARLA

and colleagues, 15 the overall mortality in bacteraemic patients was 42~o. Examination of the patient population, however, reveals an excess of hospitalacquired versus community-acquired bacteraemias, a high proportion of patients over 5o years of age as well as a significant incidence of neoplastic disease, diabetes, pseudomonas bacteraemia and polymicrobial infections, all of which would contribute to increased mortality. T h e present study consisted of predominantly community-acquired infections with a high proportion of paediatric patients, a low incidence of pseudomonas bacteraemias and polymicrobial infections and an overall mortality rate of 27 % . T h e higher mortality rate for septicaemia caused by Gram-negative as opposed to Gram-positive organisms at U.H.W.I. is consistent with several other studies TM 16.22 but at variance with the finding of Setia and colleagues who reported a 4 2 % mortality rate for septicaemia caused by Gram-positive organisms at a community hospital. 19 T h e same author also reported an increased mortality from community-acquired versus nosocomial bacteraemia, as was the case at U.H.W.I. Most other studies, however, have recorded substantially higher mortality in hospital-acquired bacteraemia. ~, 6, ~, 16.39 Deaths from staphylococcal bacteraemia have decreased markedly from 71% in the i93o's 1~to 9 % in i98o. 1~T h e figure of 3"7 % in the present study would seem to confirm this trend. In contrast, the mortality from pneumococcal bacteraemia of I7"2 % is not m u c h reduced from the 2 3 % recorded by M c G o w a n and colleagues at Boston City Hospital in i947 .16 Mortality from streptococcal bacteraemia has been reported to be 23 0/O13 which is similar to the 24 0/0 recorded at U.H.W.I. In a n u m b e r of recent studies, E. coli has been the most frequent Gram-negative organism isolated by blood culture followed by K. pneumoniae. Case mortality rates for these two organisms were I 3 - I 9 ~o and 24-38 ~o respectively. 1' 2.19 At U.H.W.I., this situation was reversed with K. pneumoniae being recovered more often from blood cultures than E. coli, case mortality rates being 23 and 53 0/0 respectively. T h e reason for the exceptionally high mortality from E. coli bacteraemia was not apparent. Deaths from septicaemia have been shown to increase with age, 3,~5,89 possibly because of a corresponding increase in age-related predisposing factors. Not all predisposing factors, however, are found in elderly patients and in the present study, a 60 0/0 incidence of malignancy-related deaths resulted in a sharply increased mortality in the Io-3o-years age group. Apart from this anomaly, case mortality rates in relation to age were in agreement with previous studies, there being increased rates in neonates and elderly patients. ~2 In any study of septicaemia, the incidence, case mortality rate and prevalence of particular pathogens will depend on various factors. These include the age distribution of the hospital population, the type of institution, the proportion of patients in different hospital units, endemicity and virulence of particular organisms in the hospital or community, the nature and frequency of predisposing factors as well as antimicrobial prescribing practices. T h e influence of such variables should be taken into account when comparisons are made among institutions. Extrapolation of data from one hospital to another is likely to be misleading. Differences in predisposing factors and in the incidence of septicaemia in relation to age, organisms and site of the initial infection may be significant in the care of patients. T r e a t m e n t should if possible always be based on local data.

B a c t e r a e m i a in the W e s t Indies

141

References

1. Kreger BE, Craven DE, Carling PC, McCabe WR. Gram-negative bacteremia. I I I . Reassessment of etiology, epidemiology and ecology in 612 patients. Am J Med I98O; 68: 332-342. 2. Bryan CS, Reynolds K L , Brenner ER. Analysis of 1,186 episodes of gram negative bacteremia in non-university hospitals: the effect of antimicrobial therapy. Rev Infect Dis 1983; 5: 629-638. 3. Dupont H L , Spink W W . Infections due to gram negative organisms: an analysis of 86o patients with bacteremia at the University of Minnesota Medical Center I958-1966. Medicine (Baltimore) 1969; 48: 3o7-332. 4. Myerowitz RL, Medeiros AA, O'Brien T F . Recent experience with bacillemia due to gram negative organisms. J Infect Dis 1971; r24: 239-246. 5. McCabe WR, Wolff S M, Bennett JV. Incidence of gram negative bacteremia. N EngtJ Med I975; 292: I I I - I 1 5 . 6. Kreger BE, Craven DE, McCabe WR. Gram-negative bacteremia. IV. Re-evaluation of clinical features and treatment in 612 patients. Am J Med 198o; 68: 344-3557. Ledger WJ, Norman M, Gec, C, Lewis W. Bacteremia on an obstetric-gynecologic service. Am J Obstet Gynecol 1975; I 2 I : 2o5-212. 8. Winchester PD, T o d d JK, Roe MH. Bacteremia in hospitalised children. Am J Dis Child 1977; x3x: 753-758. 9. Bermet R, Erikson M, Zetterstrom R. Increasing incidence of neonatal septicemia: causative organisms and predisposing risk factors. Acta Paediatr Scand 1981 ; 70: 2o7--21o. IO. McCracken G H , Shinefield HR. Changes in the pattern of neonatal septicemia and meningitis. Am J Dis Child 1966; I x2 : 33-39. x x. Curtin JA, Petersdorf RG, Bennet I L Jr. Pseudomonas bacteremia: a review of 91 cases. Ann Intern Med 1961; 54: IO77-IO8I. I2. Wilson R, Hamburger M. Fifteen years experience with staphylococcus septicemia in a large city hospital: analysis of 55 cases in the Cincinnati General Hospital i94o-I954 . A m J Med I957; 22: 437-457I3. D u m a RJ, Weinberg AN, Medrek T F , Kunz LJ. Streptococcal infections: a bacteriologic and clinical study of streptococcal bacteremia. Medicine (Baltimore) I969; 48: 87-I27. I4. Garrod LP, Lambert HP, O ' G r a d y F. Antibiotic and chemotherapy, 5th ed. Edinburgh: Churchill Livingstone, 198I. 15. Weinstein MP, Reller LB, M u r p h y JR, Lichtenstein KA. The clinical significance of positive blood cultures: a comprehensive analysis of 5oo episodes of bacteremia and fungemia in adults. Rev Infect Dis 1983; 5: 35-7o. i6. M c G o w a n JE Jr, Barnes M W , Finland M. Bacteremia at Boston City Hospital: occurrence and mortality during I2 selected years (1935-1972) with special reference to hospital acquired cases. J Infect Dis I975; I32: 316-335. I7. Williams G T , Howang E T , Shaw EJ, Tobaqchali S. Bacteremia in a London teaching hospital. Lancet I976; ii: I 2 9 I - I 2 9 3 . 18. Scheckler WE. Septicemia in a community hospital, 197o through 1973. J A M A 1977; 237: I938-x94x. I9. Setia U, Gross PA. Bacteremia in a community hospital, spectrum and mortality. Arch Intern Ailed 1977; I 3 7 : 1 6 9 8 - I 7 or. 2o. Abseynndere RL, Bradley JM, Chipping P, Rogers BT, Noone P. Bacteraemia in the Royal Free Hospital i972-6. J Infect 1979; x : I27-I38. 2I. M c G o v e r n JE Jr, Parrott PL, Duty VP. Nosocomial bacteremia: potential for prevention of procedure related cases. J A M A I977; 237:2727-2729 • 22. Spengler RF, Greenough W B II1, Stoley PD. A descriptive study of nosocomial bacteremias at the Johns Hopkins Hospital i968-i974 . Johns Hopkins M e d J I978; x42: 77-84. 23. Flick MA, Cluff LE. Pseudomonas bacteremia: review of lO8 cases. Am J Med I976; 6o:

501--508. 24. Pichichero ME, T o d d JK. Detection of neonatal bacteremia. J Pediatr 1979; 94: 958-96o. 25. Freedman R M , Ingram DL, Gross I, Ehrenkranz RA, Warshaw JB, Baltimore RS. A half century of neonatal sepsis at Yale 1928-I978. Am J Dis Child 1981; 135 : r4o-I44.

14 2

D. E. M A C F A R L A N E AND V. R. NARLA

26. Johnston RB, Sell SH. Septicemia in infants and children. Pediatrics 1964; 34: 473-47927. Young SEJ. Bacteraemia 1975-198o: a survey of cases reported to the PHLS Communicable Disease Surveillance Centre. J Infect 1982; 5: I9-26. 28. Freid MA, Vosti KL. The importance of underlying disease in patients with gram-negative bacteremia. Arch Intern Med 1968; 121: 418-423. 29. Singer C, Kaplan MH, Armstrong D. Bacteremia and fungemia complicating neoplastic disease: a study of 364 cases. A m J Med 1977; 62:731-742. 3o. Stevens MCG, Hayes RJ, Sergeant GR. Body shape in young children with homozygous sickle cell disease. Pediatrics 1983; 71: 61o-614. 31. Bassett-Connor E. Bacterial infection and sickle cell anemia. Medicine 1971 ; 5o: 97-1 ix. 32. Kumin CM, Tupasi T, Craig WA. Use of antibiotics: a brief exposition of the problem and some tentative solutions. Ann Intern Med 1973; 79: 555-56o. 33- Scheckler WE, Bennett JV. Antibiotic usage in seven community hospitals, f f A M A 197o; 213: 264-267. 34. Maki DG, Schuna AA. A study of antimicrobial misuse in a university hospital. A m ff Med Sci 1978; 275: 271-282. 35. Bryant RE, Hood AF, Hood CE, Koenig MG. Factors affecting mortality of gram negative rod bacteremia. Arch Intern Med 1971; 127: 12o-128. 36. Warren M, Herrick WW. Analysis of 134 cases of bacteremia. A m ff Med Sci 1916; 151 : 556--561. 37' Kotin P. Techniques and interpretation of routine blood cultures: observations in 5,ooo consecutive patients, f f A M A I952; 149: 1273-1282. 38. Bodey GP, Nies BA, Freireich EJ. Multiple organism septicemia in acute leukemia. Arch Intern Med 1965; I I 6 : 266-272. 39. Hodgin UG, Sanford JP. Gram negative bacteremia: an analysis of IOO patients. A m J Med 1965; 39:952-960.