Risk Factors for Extended-Spectrum Beta-Lactamase-Producing Enterobacteriaceae Carriage upon Pediatric Intensive Care Unit Admission

To the Editor:

The incidence of infections due to ESBL-producing organisms is increasing in both children and adults [1]. This increase is particularly concerning as ESBL-producing organisms are frequently resistant to antibiotics used in empiric sepsis regimens, such as ceftriaxone, cefepime, and piperacillin-tazobactam [2]. Additionally, plasmids carrying genes encoding ESBLs often harbor additional resistance mechanisms reducing the activity of aminoglycosides and fluoroquinolones [1,2]. Thus, children with signs and symptoms of severe infections ultimately found to have infections with ESBL-producing organisms may not receive the most appropriate empiric therapy, increasing the likelihood of poor outcomes [3].

Early recognition of ESBL colonization may be important as colonization with ESBLs has been associated with subsequent, invasive infections [4]. Children warranting pediatric intensive care unit (PICU) admission are at particularly high risk for serious infections. An understanding of carriers of ESBL-producing organisms may play an important role in guiding appropriate empiric treatment.

We characterized risk factors for ESBL colonization among children admitted to the PICU by conducting a case-control study among patients admitted to The Johns Hopkins Hospital 40-bed tertiary care PICU in Baltimore, Maryland. Rectal swabs were obtained from all children admitted to the unit between July 2014 and January 2015. Rectal swabs were inoculated into T-soy broth containing a 30 μg ceftriaxone disk and incubated at 37°C. Within 48 hours of inoculation, 100 μL broth samples with visible turbidity were plated on MacConkey agar with a 30 μg ceftriaxone disk and incubated at 37°C overnight. All recovered isolates within the zone of inhibition (Inhibition zone < 23 mm) underwent routine identification and antimicrobial susceptibility testing using the BD Phoenix™ Automated System (BD Diagnostics, Sparks, Maryland).

Genomic DNA was extracted from isolates with ceftriaxone MICs of ≥2 μg/mL using the DNeasy Blood & Tissue Kit (Qiagen, Germantown, Maryland) and identification of β-lactamase-encoding genes was assessed utilizing the Check-MDR CT103XL kit microarray-based assay (Check-Points, Wageningen, Netherlands). Multilocus Sequence Typing (MLST) was used for amplification and sequencing of eight housekeeping genes for E. coli and seven housekeeping genes of K. pneumoniae (http://www.pasteur.fr/mlst).

Cases were defined as children whose admission surveillance culture grew an ESBL-producing organism. Each case was matched to three controls using a random number generator.Potential risk factors for colonization were collected on all patients. Data were extracted from all available inpatient and outpatient medical records from facilities within the Johns Hopkins Health System, and from medical records of children who received care at institutions within the Epic Care Everywhere Network, a secure exchange that contains patient medical information from a large number of inpatient and outpatient healthcare networks throughout the United States. This study was approved by the Johns Hopkins University School of Medicine Institutional Review Board with a waiver of informed consent.

Baseline characteristics of cases and controls were compared using Chi-square or Fisher’s exact test for categorical variables and the Wilcoxon rank-sum test or Student’s t-test for continuous variables. P-values ≤0.05 were considered significant. All analyses were performed using Stata, version 13 (StataCorp, College Station, Texas).

Eight hundred fifty-four rectal swabs from unique patients were obtained over the study period. Twenty-four children were colonized with ESBLs (2.8%). bla_CTX-M genes were identified in all 21 E. coli isolates, 1 of 2 K. pneumoniae isolates, and 1 of 1 E. cloacae isolates. A bla_SHV-12 gene was identified in the second K. pneumoniae isolate. 17 of the 24 (71%) ESBLs contained bla_{CTX-M-15-iike} genes. The predominant circulating clonal strain was ESBL-producing E. coli ST131, identified in 63% of isolates.

The 24 ESBL-positive cases were matched to 72 ESBL-negative controls (Table). Within the previous six months, cases were more likely to have had previous ESBL colonization or infection (17% vs. 1%; p=0.01) or to have been hospitalized in a high ESBL burden foreign country (17% vs. 1%; p=0.01), including China (1), India (1), Qatar (1), and Saudi Arabia (2). Cases were more likely to have received recently chemotherapy (OR = 4.6; 95% CI 0.9 – 22.3) or a hematopoietic stem cell transplantation (OR=10.1; 95% CI 1.0 – 102.7). Nine cases (38%) developed invasive infections with ESBL-producing organisms on a subsequent clinical culture (4 during the hospital admission and 5 within the subsequent 6 months). No control patients developed subsequent ESBL infections.

Table.

Comparison of children colonized and not colonized with ESBL-producing Enterobacteriaceae on admission to a pediatric intensive care unit.

	ESBL-positive (n=24)	ESBL-negative (n=72)	OR (95% CI)	p- value
Age, years (median, interquartile range)	6.0 (3.0–18.0)	8.0 (3.0–15.0)	1.0 (1.0– 1.1)	0.40
Male	14 (58.3)	50 (69.4)	0.6 (0.2 – 1.8)	0.32
Race
White	16 (66.7)	37 (51.3)	0.8 (0.3 – 2.2)	0.24
Black	4 (16.7)	24 (33.3)	0.4 (0.09 – 1.4)	0.19
Asian	2 (8.3)	2 (2.8)	3.2 (0.2 – 45.6)	0.26
Latino	2 (8.3)	4 (5.6)	1.5 (0.1 – 11.6)	0.64
PRISM score (median, IQR)	4.0 (2.0 – 7.0)	4.0 (2.0 – 7.0)	1.0 (0.9 – 1.2)	0.80
Days from hospital admission to PICU admission (median, IQR)	0.0 (0.0 – 4.0)	0.0 (0.0 – 17.0)	1.4 (1.0 – 2.0)	0.11
Reason for PICU admission
Planned surgical procedure	11 (45.8)	32 (44.4)	1.1 (0.4 – 2.9)	0.90
Respiratory failure	2 (8.3)	12 (16.7)	0.4 (0.05 – 2.3)	0.51
Trauma	0 (0.0)	13 (18.0)	--	0.02
Sepsis	5 (20.8)	2 (2.8)	9.2 (1.3 – 100.8)	0.01
Metabolic derangements	1 (4.2)	4 (5.6)	0.7 (0.01 – 8.0)	1.00
Other1	5 (20.8)	8(11.1)	2.1 (0.5 – 8.3)	0.23
Pre-existing conditions
Previously healthy	3 (12.5)	28 (38.9)	0.2 (0.04 – 0.9)	0.02
Congenital heart disease	1 (4.2)	9 (12.5)	0.3 (0.01 – 2.4)	0.44
Chemotherapy within the previous 6 months2	4 (16.7)	3 (4.2)	4.6 (0.95 – 22.3)	0.05
Hematopoietic stem-cell transplantation (HSCT) within the previous 12 months	3 (12.5)	1 (1.4)	10.1 (1.00 – 102.6)	0.03
Chronic steroid use or immunotherapy	1 (4.2)	2 (2.8)	1.5 (0.02 – 30.4)	1.00
Other3	0 (0.0)	2 (2.8)	--	0.41
Healthcare exposures in the 6 months prior to current PICU admission
Hospital admission	12 (50.0)	29 (40.3)	1.5 (0.5 – 4.2)	0.40
PICU admission	2 (8.3)	2 (2.8)	3.2	0.26
			(0.2 – 45.6)
Hospitalization in a foreign country	4 (16.7)	1 (1.4)	14.2 (1.3 – 708.6)	0.01
ESBL colonization or infection	4 (16.7)	1 (1.4)	14.2 (1.3 – 708.6)	0.01
Number of hospital days4, mean (SD)	11.8 (19.5)	11.3 (20.9)	1.0 (1.0 – 1.0)	0.78
Number of antibiotic days4, mean (SD)	4.5 (7.2)	11.0 (34.7)	1.0 (1.0 – 1.0)	0.47
Antibiotic use in the 6 months prior to current PICU admission
Any antibiotic	9 (37.5)	24 (33.3)	1.2 (0.4 – 3.4)	0.71
Penicillin	6 (25.0)	12 (16.7)	1.7 (0.4 – 5.6)	0.36
Cephalosporins	6 (25.0)	17 (23.6)	1.1 (0.3 – 3.4)	0.89
Carbapenems	3 (12.5)	1 (1.4)	10.1 (0.7 – 539.6)	0.05
Fluoroquinolones	2 (8.3)	1 (1.4)	6.4 (0.3 – 385.9)	0.15
Aminoglycosides	3 (12.5)	5 (6.9)	1.9 (0.3 – 10.7)	0.39

All values represent n (%), unless otherwise noted.

¹Other reasons for PICU admission included: intractable seizures (3), altered mental status (2), burns (2), myocarditis (1), chemotherapy induction (1), anemia (1), cardiac arrest (1), allergic reactions (2)

²Patients who received both chemotherapy and HSCT only categorized as HSCT; ³Excludes patients receiving immunotherapy for SOT, HSCT, or chemotherapy

³Other pre-existing conditions include end-stage renal disease on hemodialysis (1) and solid organ transplantation (1)

⁴Calculation of the mean included only patients with the specific exposure.

Our findings suggest that targeted screening of high-risk patients may be a reasonable consideration to identify ESBL colonization. Identifying children colonized with ESBL-producing organisms may be particularly relevant to help guide empiric antibiotic therapy as 40% of children colonized with ESBL-producing organisms at the time of PICU admission went on to develop invasive ESBL infections.

ESBL-producing bacteria have increasingly been identified in the community, which appears to be driven by clonal expansion of E. coli ST131 and person-to-person transmission, sometimes in the absence of significant health care exposure [5-8], Our results indicate that previously healthy children are still at low risk for ESBL colonization. However, hospitalization in a foreign country was a strong predictor of ESBL colonization. This finding reflects that there are significant regional differences in ESBL prevalence, with a disproportionate burden in India, East and Southeast Asia, and the Middle East [9-10].

This is a single-center study. Our findings must be repeated in a larger and more diverse setting. Additionally, although we completed a thorough review of inpatient and outpatient records from a number of health care facilities in the state of Maryland, there may still have been missing data. However, this is expected to be similar for both cases and controls.

As the incidence of ESBL increases and they contribute to considerable morbidity and mortality, it is imperative to develop systems to identify children who are most at risk of infections caused by ESBL-producing organisms. Our findings suggest that in addition to reviewing prior culture histories to identify children with previous ESBL colonization or infection, targeted screening of children who received medical care abroad in high risk countries, who recently received chemotherapy, or who recently underwent hematopoietic stem cell transplantations may be another strategy to help identify those most at risk for ESBL colonization.