Keywords Bloodstream - colonization - enteric - infections
Introduction
Bloodstream infections (BSI) with multidrug-resistant (MDR) bacteria are associated
with a higher morbidity and mortality.[1 ] Our institution has previously reported a rising incidence in BSI caused by MDR
bacteria, a high prevalence of rectal colonization by MDR and extended-spectrum β-lactamase-producing
organisms (ESBLO), and a high conversion rate from sensitive bacteria to resistant
strains over time in pediatric patients with cancer. This informed the decision of
the pediatric hematolymphoid disease management group to adopt the “de-escalation”
strategy of the Expert Group of the 4th European Conference of Infections in Leukemia
(ECIL-4) guidelines for empirical antibiotics in the management of patients with febrile
neutropenia.[2 ]
The objectives of this audit were to determine the correlation between enteric colonization
with MDR bacteria and BSI and to evaluate the association between colonization with
MDR bacteria and intensive care unit (ICU) admission and mortality from infections.
The audit was approved by the Institutional Ethics Committee of the Tata Memorial
Hospital with approval number P. No 3325/2019.
Design/Methods
This audit was conducted in the pediatric oncology unit of a large tertiary cancer
care center in western India, delivering services to patients from all over India
and other countries in Asia and Africa.
Antibiotic Policy
The unit utilized the “de-escalation” policy of the ECIL-4 guidelines and prospectively
applied it to patients with hematolymphoid malignancies who developed febrile neutropenia.
The policy is to start with a combination of colistin and carbapenem on the first
sign of fever in neutropenic patients who were known to be colonized with MDR. The
higher end of the colistin maximum dose recommended for children was used at 150,000
IU/kg, divided every 6 hours, and modified when indicated by creatinine clearance.
If the patients subsequently improved, and blood culture did not grow any organism
or showed an organism with a different sensitivity, the antibiotics would be rapidly
“de-escalated” to a more standard first-line combination of use in the unit.
Other Measures to Reduce Colonization
Other strategies employed by our unit to limit colonization included segregation and
cohorting of MDR colonized patients during admissions and use of best infection control
policies, including handwash, gowning, and masks when dealing with these patients.
Data prospectively collected from January 2015 to July 2016 was examined. Rectal swabs
were obtained by introducing a sterile swab stick into the rectum of patients registered
with hematolymphoid malignancies (profile in [Table 1 ]) on day of presentation, or within 7 days if missed on day 1. Repeat rectal swabs
within 15 days were taken to track in cases of “no isolate” on the initial swab. This
occurred sometimes as patients had received antibiotics before referral to our center.
Others were repeated during febrile neutropenia episodes and admission to evaluate
change in flora in previously non-MDR colonized patients who did not show improvement
on empirical antibiotics or to evaluate the persistence of MDR strains (this was done
only for MDR-positive patients).
Table 1
Profile of patients, baseline rectal swabs, and BSI (1094 patients)
Variable
n (%)
Abbreviations: ALL, acute lymphoblastic leukemia; AML, acute myeloid leukemia; CML,
chronic myeloid leukemia; E. coli , Escherichia coli ; ESBL, extended-spectrum β-lactamase; ESBLO, Extended-spectrum β-lactamase producing
organism; HL, Hodgkin lymphoma; K. pneumoniae , Klebsiella pneumoniae ; MDR, multidrug resistant; MDS, myelodysplastic syndromes; MRSA, methicillin-resistant
S. aureus ; NHL, non-HL; S. aureus , Staphylococcus aureus ; VRE, vancomycin-resistant enterococci.
Male
743 (68)
Female
351 (32)
Male:female ratio
2.1:1
Median age (range)
7 years (6 months–15 years)
ALL
688 (62.9)
AML
157 (14.4)
NHL
124 (11.4)
HL
102 (9.3)
MDS
10 (0.9)
CML blast/accelerated phase
13 (1.2)
Total number of patients colonized with two different bacteria at presentation
191
Total number of patients colonized with three different bacteria at presentation
35
Total number of organisms isolated at presentation
997
Number of patients with E. coli
680
enteric colonization
(62% of total number of patients)
MDR E. coli
134 (12.2)
ESBL E. coli
330 (30)
Number of patients with K. pneumoniae enteric colonization
285
(26% of total number of patients)
MDR K. pneumoniae
97 (8.7)
ESBL K. pneumoniae
129 (11.8)
Number of patients colonized by MDR at presentation
231 (21)
Number of patients colonized by ESBLO at presentation
495 (41.9)
Total number of febrile neutropenic episodes
1461
Total blood culture positivity rate
298/1461 (20.4)
Blood culture positivity rate in colonized patients
253/736 (34.4)
Blood cultures were taken during febrile episodes from peripheral veins and central
lines when present. Records of all cultures were examined. Sensitivity patterns were
studied using the Kirby–Bauer’s disc diffusion method. Clinical and Laboratory Standards
Institute (CLSI) guidelines were used to select the antibiotic disc. A limited panel
was used to determine ESBLO, carbapenem-resistant Enterobacteriaceae (CRE), and vancomycin-resistant
enterococci (VRE).
ESBL production was confirmed by CLSI recommendations using cephalosporin-clavulanate
combination disks. A difference of ≥ 5 mm between zone diameter of either of the cephalosporin
discs and their respective cephalosporin-clavulanate disc was taken to be phenotypic
confirmation of ESBL production. We used cefotaxime (30 μg), ceftazidime (30 μg),
and ceftazidime/clavulanic acid (30 μg/10 μg) disks for ESBL determination. Carbapenem
resistance was reported as per the CLSI guidelines. Vancomycin resistance was confirmed
by minimal inhibitory concentrations with the “E” test. In an event when more than
one organism with differing sensitivity profile was isolated from the culture, the
worse sensitivity pattern was taken into account for analysis.
Definition of Resistance
MDR was defined as nonsusceptibility to at least a carbapenem, polymyxin (colistin),
and an antipseudomonal cephalosporin (ceftazidime).[3 ] Nonsusceptibility to a third-generation cephalosporin or in combination with a clavulanate
but not any of the agents mentioned in defining MDR delineated ESBLO.[4 ] ESBLO intermediate strains were considered to be resistant.
Statistical Methods Used
We analyzed the data of children with hematolymphoid malignancies who developed BSI
from January 2015 to July 2016. We correlated their blood culture results during the
episodes of febrile neutropenia with their baseline rectal colonization before starting
therapy. Finally, we determined the outcome of the febrile neutropenia episodes in
terms of ICU transfer and death. Simple descriptive statistics were used to present
the demographic, clinical characteristics, details of baseline rectal colonization,
and BSI of the patients. Paired data were analyzed to test strength of the association
of MDR colonization with MDR BSI and outcome parameters such as ICU admissions and
mortality using a two-sided Pearson Chi-square. Statistical analysis was conducted
using SPSS version 24.0, and significance level was set at p < 0.05 with a confidence interval of 95%.
Results
In the 18-month study, 1094 patients with hematolymphoid malignancies were registered
for treatment and had baseline rectal swab taken. The profile of bacterial isolates
from rectal swab cultures is represented in [Table 1 ]. The incidence of rectal swab colonization at baseline was 67.3% from 1094 patients.
Escherichia coli was isolated in 680 patients, 19.7% of patients at baseline were colonized with MDR
E. coli and 30% were ESBLO. Klebsiella pneumoniae was isolated from 285 patients, 8.7% of patients were colonized with MDR K. pneumonia , while 11.8% were ESBLO. VRE was isolated in 32 patients. The profile of these patients
is summarized in [Table 1 ].
During the study, 298 patients had positive blood cultures; of these, 253 had rectal
colonization at baseline. These are summarized in [Table 2 ].
Table 2
Rectal colonization status and blood infection
Isolate
Sensitivity
n (%)
Abbreviations: BSI, bloodstream infection; ESBLO, Extended-spectrum β-lactamase-producing
organism; MDR, multidrug resistant
BSI (positive cultures) (n = 298)
ESBLO
44 (14.8)
MDR
84 (28.2)
Sensitive
170 (57)
Rectal swab colonization in those with positive blood culture (n = 253)
ESBLO
95 (31.9)
MDR
90 (30.2)
Sensitive
68 (22.8)
No growth
45 (15.1)
The majority (57%) of organisms causing BSI in our cohort were sensitive bacteria,
while 43% were either MDR or ESBLO.
Further analysis of the 253 patients with rectal colonization who developed BSI and
their outcomes is illustrated in a Sankey diagram ([Fig. 1 ]).
Fig. 1 Illustration of baseline rectal colonization and the development of blood stream
infection (BSI) and outcome of infection. ESBLO: Extended-spectrum β-lactamase-producing
organisms, MDR: multidrug resistant organisms.
At baseline, most of the patients (62.1%) with confirmed BSI were colonized with either
ESBLO (31.9%) or MDR (30.2%). Forty-five patients (15.1%) were not colonized at baseline.
We assessed the relationship between rectal colonization at baseline with confirmed
BSI. For patients colonized with either MDR or ESBLO, the majority 116 (62.7%) developed
BSI caused by either MDR or ESBLO, while the remaining 37.3% of them developed BSI
caused by sensitive bacteria. In contrast, among patients without baseline colonization
or colonized by sensitive bacteria, only 12 (10.6%) of them developed BSI caused by
either MDR or ESBLO, while most (89.4%) developed BSI caused by sensitive organisms.
These differences were statistically significant (p < 0.001). In other words, the baseline rectal swab had a sensitivity and a specificity
of 90.6% and 59.4%, respectively, in predicting BSI with either MDR or ESBLO ([Table 3 ]).
Table 3
Correlation of rectal colonization at baseline by MDR organisms and ESBLO with BSI
caused by MDR organisms and ESBLO
Rectal colonization at baseline (n = 298) MDR or ESBLO
n (%)
BSI with MDR or ESBLO, n (%)
p (χ2 )
Abbreviations: BSI, blood stream infection ESBLO, Extended-spectrum β-lactamase-producing
organism; MDR, multidrug-resistant.
Yes
No
Either
185 (62.1)
116 (62.7)
69 (37.3)
< 0.001 (77.661)
Neither
113 (37.9)
12 (10.6)
101 (89.4)
Rectal colonization at baseline (
n
= 185)
n
BSI
p
( χ2
)
ESBLO
MDR
Sensitive
ESBLO
95
24 (25.3)
34 (35.8)
37 (38.9)
0.049 (6.032)
MDR
90
12 (13.3)
46 (51.1)
32 (35.6)
Comparing MDR to ESBLO rectal colonization, the likelihood of patients colonized with
ESBLO to develop sepsis due to ESBLO was 25.3%, while that of those colonized with
MDR to develop MDR sepsis was 51.1% (p = 0.049) ([Table 3 ]).
Patients with either MDR or ESBLO colonization were more likely to be transferred
to the ICU (49.7%) as compared with those without baseline colonization or colonized
by sensitive bacteria (1.8%), as shown in [Table 4 ].
Table 4
Rectal colonization with MDR and ESBLO and likelihood of ICU admission
Rectal colonization at baseline (n = 298) MDR and ESBL
n (%)
ICU admission, n (%)
No ICU admissions, n (%)
p (χ2 )
Abbreviations: ESBL, extended-spectrum β-lactamase; ESBLO, extended-spectrum β-lactamase-producing
organism; ICU, intensive care unit; MDR, multidrug-resistant.
Either
185 (62.1)
92 (49.7)
93 (50.3)
< 0.001
Neither
113 (37.3)
2 (1.8)
111 (98.2) (74.725)
ESBL
95 (51.4)
44 (46.3)
51 (53.7)
0.34
MDR
90 (48.6)
48 (53.3)
42 (46.7) (0.910)
Patients colonized with MDR or ESBLO had an equal chance of ICU transfer ([Table 4 ]), but mortality was higher in the MDR group as compared with the ESBLO group ([Table 5 ]).
Table 5
Rectal colonization with multidrug resistant and extended spectrum β-lactamase producing
organism and likelihood of mortality
Rectal colonization at baseline (n =298) MDR and ESBLO
n ( %)
Dead, n ( %)
Alive, n ( %)
p ( χ2 )
Abbreviations: ESBLO, extended spectrum β-lactamase producing organisms; MDR, multidrug
resistant.
Either
185 (62.1)
46 (24.9)
139 (75.1)
0.001 (10.382)
Neither
113 (37.9)
11 (9.7)
102 (90.3)
ESBLO
95 (51.4)
17 (17.9)
78 (82.1)
0.024 (5.078)
MDR
90 (48.6)
29 (32.2)
61 (67.8)
Death was also more likely in those colonized with MDR or ESBLO at baseline compared
with those who were not ([Table 5 ]).
Discussion
Recent reports from our institution showed the high incidence of both MDR and ESBLO
in BSI compared with what is being reported in Western literature.[5 ]
[6 ]
[7 ] Another study from our institute also reported an alarming incidence of MDR bacteria
in rectal surveillance cultures of pediatric patients, 58.4% and 20.2% for ESBL and
CRE, respectively, thus further strengthening the call for institutions to review
antibiotic usage policy.[8 ]
Antibiotic stewardship, institutional antibiotic usage, and infection prevention and
control have gained prominence in recent times as the world is faced with an increasing
spate of drug-resistant organisms without any appreciable increase in the development
of new antimicrobials to combat this pandemic.[2 ] The focus has now shifted to the judicious use of antimicrobials with emphasis on
early commencement of appropriate empirical antimicrobials while awaiting culture
reports in immunosuppressed patients. It has been demonstrated in reports that organisms
invading the bloodstream following commencement of cancer treatment with its attendant
effect on breaching the integrity of the gastrointestinal mucosal barrier has led
to translocation of pathogens colonizing the gut into the bloodstream consequently,
leading to sepsis.[2 ]
[9 ]
[10 ]
[11 ]Enteric colonization by drug-resistant pathogens has therefore been named in ECIL-4
guidelines as the most important risk factor for infection by MDR bacteria, with special
emphasis on ESBLO, CRE, resistant nonfermenters, methicillin-resistant Staphylococcus aureus (MRSA) with mic ≥ 2 mg/L and VRE.
Several other reports and guidelines such as the Indian Council of Medical Research
(ICMR) have informed the need to have surveillance cultures to guide the choice of
empirical antibiotics for the management of patients with febrile neutropenia.[2 ]
[12 ]
[13 ]
[14 ]
[15 ]
[16 ]However, some do not recommend this strategy, owing to contradictory or inconclusive
evidence to support enteric colonization as a source of clinical infections in the
immunocompromised host. In this context, a retrospective review of 794 allogeneic
hematopoietic transplant patients had found only 12 patients developing MDR Pseudomonas
infections, out of which only about half had prior colonization, indicating that there
was another source for the infection. Fluoroquinolone prophylaxis was however administered
to 83% of this cohort, which may have influenced the outcome.[17 ] Although a few studies are inconclusive on the utility of surveillance cultures,
these have major limitations such as inadequate sample size and prior utilization
of antimicrobial prophylaxis. There is however a consistent pattern of increased morbidity,
higher overall mortality, and increased cost of care in patients who develop infections
caused by drug-resistant pathogens without timely choice of appropriate empirical
antimicrobials.[8 ]
[10 ]
[18 ]
The correlation between gut colonization of resistant pathogens and their subsequent
implication in BSI in a pediatric hematolymphoid cohort is being reported in a low-middle
income country, with a high prevalence of drug-resistant pathogens ([Tables 2 ],[ 3 ] and [Fig. 1 ]), that has utilized the ECIL-4 guidelines.
In an observational prospective cohort of 126 adult patients with hematolymphoid malignancies
in Mexico, it was found that colonization with ESBL-E. coli increased the risk of BSI by the same strain, had a shorter time to death, longer
hospital stay, and higher infection-related costs.[10 ]This was corroborated in our study as patients with either MDR or ESBLO colonization
were more likely to be transferred to the ICU (49.7%) compared with those without
baseline colonization or colonized by sensitive bacteria (1.8%), and this was associated
with a higher mortality, especially in the patient with MDR colonization ([Fig. 1 ] and [Tables 4 ] and [5 ]). This was found in an earlier study among pediatric patients in our institution
and strengthened the call for early suspicion of sepsis and appropriate upfront use
of high-grade antibiotics in this group of immunosuppressed patients with MDR colonization
to curtail these adverse outcomes.
VRE colonization did not lead to an increased incidence of BSI by the same suggestion
that our center can continue to withhold initial coverage for VRE in those colonized
with this pathogen, as the literature is still inconclusive with regard to initial
coverage for VRE due to its low pathogenicity.[19 ]
[20 ]
Conclusions
We conclude that a high prevalence of MDR colonization at presentation significantly
results in MDR BSI, which further results in a significant increase in ICU admissions
and mortality. This would justify the use of a “de-escalation” antibiotic policy as
defined by ECIL4 recommendations. Whether such a strategy has been successful in impacting
outcomes, would need further study.
