Risk Factors and Outcomes for Intestinal Carriage of AmpC-Hyperproducing Enterobacteriaceae in Intensive Care Unit Patients

Abstract

In a cohort of 1,209 intensive care unit (ICU) patients, the prevalence of intestinal colonization with high-level expressed AmpC cephalosporinase-producing Enterobacteriaceae (HLAC-PE) rose steadily from 2% at admission to 30% in patients with lengths of stay (LOS) exceeding 4 weeks. In multivariate analysis, LOS was the main predictor of carriage acquisition after adjustment on antimicrobial exposure. HLAC-PE infection occurred in 15% of carriers. Carriage and infection were associated with a marked increase in carbapenem consumption.

TEXT

The burden of expanded-spectrum cephalosporin (ESC) resistance in Enterobacteriaceae now represents a daily issue for the management of antimicrobial therapy in intensive care unit (ICU) patients (1, 2). Two distinct enzymatic mechanisms may be involved, namely, plasmid-borne extended-spectrum beta-lactamases (ESBL) and high-level expressed AmpC cephalosporinase (HLAC) (2). The HLAC phenotype results from the mutation-induced derepression of a normally low-level expressed chromosomal bla_ampC gene or, more rarely, from the horizontal transfer of a plasmid-borne AmpC variant (3). HLAC-producing Enterobacteriaceae (HLAC-PE) currently account for up to 50% of the ESC-resistant Enterobacteriaceae responsible for ICU-acquired infections in settings with a high prevalence of ESBL-producing Enterobacteriaceae (ESBL-PE) (4).

The intestinal microbiota forms the main reservoir of ESC-resistant Enterobacteriaceae in ICU patients (5–7); however, the features of HLAC-PE carriage remain poorly described in this population. Our objectives were to identify the risk factors for HLAC-PE carriage and to appraise its clinical significance in terms of subsequent infections and carbapenem consumption in a large single-center cohort of critically ill patients.

This retrospective study was conducted in the 18-bed medical-surgical ICU of an 1,100-bed teaching hospital in France. During the 3-year study period (2008 to 2010), intestinal carriage of ESC-resistant Enterobacteriaceae was routinely screened by rectal swabbing at admission and at weekly intervals afterwards as part of the institutional infection control policy, with implementation of isolation measures in identified carriers. Swabs were discharged in 1 ml of 0.9% saline, and 50 μl of the suspension was subsequently plated on a selective medium for ESC-resistant Gram-negative bacilli (chromID ESBL; bioMérieux, Marcy-l'Étoile, France). After overnight incubation at 35°C, colonies were subcultured for identification (API 20E; bioMérieux) and susceptibility testing using the disk diffusion method on Mueller-Hinton agar plates (Bio-Rad, Marnes-la-Coquette, France), in accordance with the French Society of Microbiology guidelines (www.sfm-microbiologie.org). Third-generation cephalosporin-resistant Enterobacteriaceae isolates showing a synergy zone between ESC and clavulanate were categorized as ESBL-PE, while those without synergy and a ≥5-mm increase in the ESC inhibition diameter on cloxacillin-supplemented Mueller-Hinton agar (250 mg · ml⁻¹) were categorized as HLAC-PE. Isolates producing ESBL and HLAC were classified as ESBL-PE. No carbapenem-resistant Enterobacteriaceae were isolated in our ICU during the study period.

Patients with a first ICU stay of more than 48 h were included. Variables in the tables were prospectively entered by attending ICU physicians into a local database used for diagnosis and procedures coding, surveillance of ICU-acquired infections, and monitoring of antimicrobial consumption. Additional data on antibiotic exposure were retrospectively extracted from the patients' medical charts. HLAC-PE carriers were identified in the database of the bacteriology unit. Imported carriage was defined as a positive swab at ICU admission, while acquired carriage was defined as a positive surveillance swab in patients with a negative admission sample. Carriers were assumed to be colonized from the first positive swab to ICU discharge or death. The Institutional Review Board waived the requirement for informed consent (IRB 2015-02). Data are reported according to the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) statements (8). Carriers and noncarriers were compared using the Student t test or the Mann-Whitney test for continuous variables and the Fischer exact test or the χ² test for categorical variables as appropriate. All tests were two-sided, and P values of <0.05 were considered significant. Statistical analyses were carried out using R 2.15.1 (http://www.r-project.org).

A total of 1,209 patients were included (Table 1; see also Fig. S1 in the supplemental material). None of them had a history of documented HLAC-PE infection during the hospital stay prior to ICU admission or during a previous hospital stay within the year preceding ICU admission. The 24 patients (2.0%) with imported HLAC-PE carriage had more frequent histories of chronic diseases, hospitalization within the previous year, and surgery within the previous month and were more often transferred from ward than patients not colonized at admission (Table 1). Enterobacter cloacae accounted for half of the imported HLAC-PE isolates (see Table S1 in the supplemental material).

TABLE 1.

Study population and factors associated with intestinal carriage of high-level AmpC cephalosporinase-producing Enterobacteriaceae at intensive care unit admission^a

Variable	All patients (n = 1,209)	Intestinal carriage of HLAC-PE at ICU admission		P value
		Carriers (n = 24)	Noncarriers (n = 1,185)
Median age (interquartile range), yr	63 (49–75)	68 (51–75)	63 (49–74)	0.23
Male gender, no. (%)	763 (63.1)	19 (79.2)	744 (62.8)	0.13
Comorbidities, no. (%)
Cardiac	579 (47.9)	15 (62.5)	564 (47.6)	0.15
Respiratory	296 (24.5)	12 (50.0)	284 (24.0)	0.007
End-stage renal disease	23 (1.9)	0	23 (1.9)	1.0
Neurological	66 (5.5)	3 (12.5)	63 (5.3)	0.15
Liver cirrhosis	37 (3.1)	2 (8.3)	35 (2.9)	0.16
Malignancies	123 (10.2)	5 (20.8)	118 (10.0)	0.07
Diabetes mellitus	223 (18.4)	8 (33.3)	215 (18.1)	0.06
Median McCabe score (interquartile range)	0 (0–0)	0 (0–1)	0 (0–0)	0.02
Hospital stay within the previous year, no. (%)	322 (26.6)	12 (50.0)	310 (26.2)	0.02
Surgery within the previous month, no. (%)	13 (1.1)	2 (8.3)	11 (1.0)	0.02
Admission, no. (%)
Transfer	379 (31.4)	15 (62.5)	364 (30.7)	0.002
Direct	830 (68.6)	9 (37.5)	821 (69.3)
Type of ICU stay, no. (%)
Medical	1,091 (90.2)	20 (83.3)	1,071 (90.4)	0.28
Surgical	118 (9.8)	4 (16.7)	114 (9.6)
Median SAPS II at ICU admission (interquartile range)	41 (29–53)	40 (31–43)	41 (29–53)	0.54
ESBL-PE carriage at ICU admission, no. (%)	24 (2.0)	2 (8.3)	22 (1.8)	0.08
Main diagnoses of the ICU stay, no. (%)
Sepsis	648 (53.6)	12 (50.0)	636 (53.7)	0.84
Neurological	181 (15.0)	5 (20.8)	176 (14.8)	0.39
Renal	274 (22.7)	3 (12.5)	271 (22.9)	0.24
Cardiac	205 (16.9)	3 (12.5)	202 (17.0)	0.78
Intestinal/hepatic	54 (4.5)	2 (8.3)	52 (4.4)	0.29
Trauma	151 (12.5)	1 (4.2)	150 (12.6)	0.34
Use of life-sustaining therapies, no. (%)
Mechanical ventilation	806 (66.7)	16 (66.7)	790 (66.7)	1.0
Vasopressors	580 (48.0)	7 (29.2)	573 (48.3)	0.07
Renal replacement therapy	125 (10.3)	1 (4.2)	124 (10.5)	0.5

^aHLAC-PE, high-level expressed AmpC-producing Enterobacteriaceae; ICU, intensive care unit; SAPS II, simplified acute physiology score II; ESBL-PE, extended-spectrum-β-lactamase-producing Enterobacteriaceae.

Among the 1,185 patients without HLAC-PE carriage at admission (median ICU length of stay [LOS], 7 days; interquartile range, 4 to 12 days), 107 (8.9%) became colonized during the ICU stay (acquisition rate, 7.7 per 1,000 patient-days). The rate of HLAC-PE carriage increased markedly over time, exceeding 30% in patients with LOS longer than 28 days (Fig. 1). Using a proportional subdistribution hazards model in which the occurrence of death was analyzed as a competing risk (Fine and Gray method) (9), the cumulative probability of carriage acquisition rose steadily as LOS increased, reaching 20% in patients still hospitalized in the ICU after 3 weeks (see Fig. S2 in the supplemental material). Most of the acquired HLAC-PE isolates were identified as E. cloacae (40.2%), Citrobacter freundii (17.0%), and Hafnia alvei (12.5%) (see Table S1 in the supplemental material). Overall, 94.1% of HLAC-PE isolates were susceptible to cefepime, while all remained susceptible to imipenem (see Table S2 in the supplemental material).

FIG 1.

Observed rates of intestinal carriage of high-level AmpC cephalosporinase-producing Enterobacteriaceae isolates in critically ill patients according to their length of stay in the intensive care unit (days). P value for trend was <0.0001 (chi-square test). Bars indicate 95% confidence intervals.

Patients with acquired carriage were more often admitted for surgical reasons, had higher simplified acute physiology scores II (SAPS II) at admission, required more life-sustaining therapies, and had prolonged LOS compared to those with at least one collected surveillance swab and no carriage throughout the ICU stay (n = 551) (Table 2). Antimicrobial exposure was measured between admission and the first positive swab in patients with acquired carriage and during the whole ICU stay (i.e., until discharge or death) in persistent noncarriers. A total of 104 patients (97.2%) with acquired carriage and 401 patients (72.3%) without acquisition received antimicrobials (P < 0.0001), with significant variations in cephalosporins and metronidazole exposure (66.4% versus 54.1% and 30.8% versus 20.5%, respectively; P = 0.02 for the two comparisons). The clinical characteristics in Table 2 associated with acquired colonization (i.e., yielding P values of <0.20 by bivariate analysis) were entered into a multiple logistic regression model with carriage acquisition as the primary outcome. The potential impact of antimicrobial use was analyzed in the model by entering the cumulative duration of exposure for each class as a continuous variable. Potential explanatory variables were selected using a backward procedure until the best fitted model was reached according to the Akaike information criterion (10). The best model retained only ICU LOS (odds ratio [OR], 1.03 per day; 95% confidence interval [95% CI], 1.01 to 1.04; P < 0.001), the SAPS II value at admission (OR, 1.02 per point; 95% CI, 1.01 to 1.04; P = 0.02), and metronidazole use (OR, 1.09 per treatment day; 95% CI, 1.01 to 1.17; P = 0.02) as independently associated with acquired carriage (Table 3). Mechanical ventilation, although retained in this model, did not reach statistical significance (OR, 1.62; 95% CI, 0.99 to 2.14; P = 0.09). Cephalosporins use was not an independent predictor of HLAC-PE acquisition after adjustment on LOS and other covariables, including metronidazole exposure (Table 3; see also Fig. S3 in the supplemental material).

TABLE 2.

Characteristics associated with acquired carriage of high-level AmpC cephalosporinase-producing Enterobacteriaceae during the intensive care unit stay in patients not colonized at admission

Variablea	Intestinal carriage of HLAC-PE during the ICU stay		P value
	Acquisition (n = 107)	No acquisitionb (n = 551)
Type of ICU stay, no. (%)			0.03
Medical	89 (83.2)	498 (90.4)
Surgical	18 (16.8)	53 (9.6)
Median SAPS II at ICU admission (interquartile range)	49 (40–58)	42 (32–55)	0.0002
Main diagnoses of the ICU stay, no. (%)
Sepsis	72 (67.3)	333 (60.4)	0.18
Neurological	15 (14.0)	83 (15.1)	0.78
Renal	37 (34.6)	137 (24.9)	0.04
Cardiac	23 (21.5)	108 (19.6)	0.65
Intestinal/hepatic	5 (4.7)	29 (5.3)	0.80
Trauma	19 (17.7)	72 (13.1)	0.20
Antimicrobial exposure during the ICU stayc, no. (%)
Antimicrobial, any	104 (97.2)	401 (72.3)	<0.0001
Penicillins	16 (14.9)	102 (18.5)	0.38
Median duration, days (interquartile range)	3 (3–6)	5 (3–8)	0.83
Penicillin + beta-lactamase inhibitor associations	49 (45.8)	251 (45.5)	0.96
Median duration, days (interquartile range)	6 (4–9)	6 (4–8)	0.51
Third- and fourth-generations cephalosporins	71 (66.4)	298 (54.1)	0.02
Median duration, days (interquartile range)	5 (3–9)	7 (4–9)	0.19
Carbapenems	12 (11.2)	48 (8.7)	0.41
Median duration, days (interquartile range)	5 (3–7)	6 (3–9)	0.86
Aminoglycosides	21 (19.6)	120 (21.8)	0.62
Median duration, days (interquartile range)	3 (1–3)	3 (2–4)	0.97
Fluoroquinolones	0	26 (4.7)	0.02
Median duration, days (interquartile range)		5 (2–9)	NA
Metronidazole	33 (30.8)	113 (20.5)	0.02
Median duration, days (interquartile range)	7 (5–10)	6 (3–8)	0.13
Glycopeptides	15 (14.0)	76 (13.8)	0.95
Median duration, days (interquartile range)	4 (2–6)	4 (3–7)	0.28
ESBL-PE carriage, no. (%)
At ICU admission	3 (2.8)	16 (2.9)	0.95
Acquired in the ICU	6 (5.6)	15 (2.7)	0.12
Use of life-sustaining therapy during the ICU stay, no. (%)
Mechanical ventilation	96 (89.7)	409 (74.2)	<0.0001
Vasopressors	74 (69.1)	308 (55.9)	0.01
Renal replacement therapy	22 (20.6)	62 (11.2)	0.008
Median length of ICU stay, days (interquartile range)	18 (10–30)	11 (7–17)	<0.0001

^aAge, gender, comorbidities, and history of previous hospitalization or surgery did not differ between patients with acquired carriage and noncarriers (data not shown).

^bOnly patients with at least one surveillance swab collected during the ICU stay were included as comparators (527 patients not colonized at admission were discharged alive or died in the ICU before a first weekly surveillance swab was collected).

^cBetween ICU admission and the first positive rectal swab in patients with acquired HLAC-PE carriage and during the whole ICU stay in noncarriers.

TABLE 3.

Acquired carriage of high-level AmpC cephalosporinase-producing Enterobacteriaceae during the intensive care unit stay in patients not colonized at admission: results of logistic regression analyses^a

Variable	Bivariate logistic regression, unadjusted OR (95% CI)	Multivariate logistic regression, adjusted OR (95% CI)
Type of ICU stay, surgical vs medical	1.90 (1.06–3.40)
SAPS II at ICU admission, per point increase	1.02 (1.01–1.03)	1.02 (1.01–1.04)
Main diagnoses of the ICU stay
Sepsis	1.35 (0.87–2.10)
Renal	1.60 (1.03–2.29)
Antimicrobial exposure during the ICU stay
Third- and fourth-generations cephalosporins, per day	1.03 (1.00–1.07)
Metronidazole, per day	1.02 (1.01–1.03)	1.09 (1.01–1.17)
Use of life-sustaining therapy during the ICU stay
Mechanical ventilation	1.08 (1.02–1.13)	1.62 (0.99–2.14)
Vasopressors	1.77 (1.14–2.76)
Renal replacement therapy	2.04 (1.19–3.50)
Length of ICU stay, per day	1.02 (1.01–1.03)	1.03 (1.01–1.04)

^aOR, odds ratio; 65% CI, 95% confidence interval; ICU, intensive care unit; SAPS II, simplified acute physiology score II.

A total of 126 patients (10.4%) developed at least one ICU-acquired infection, mainly ventilator-associated pneumonia (see Table S3 in the supplemental material). HLAC-PE infections occurred in 20 carriers (same species isolated from carriage and clinical samples in all patients; median delay between documented carriage and infection, 3 [2 to 13] days) and 1 patient without documented intestinal colonization (15.3% versus 0.1%; P < 0.0001). Carbapenem use (including imipenem and meropenem) was measured between the first positive swab and ICU discharge or death in HLAC-PE carriers and during the whole ICU stay in noncarriers. HLAC-PE infections and HLAC-PE carriage without infection were associated with a dramatic increase in carbapenem exposure compared to a lack of carriage (174, 65, and 31 carbapenem days per 1,000 patient-days, respectively; P < 0.0001 for all comparisons) (Table 3). ICU mortality rates (overall, 19.9%) did not differ between carriers and noncarriers.

The prevalence of HLAC-PE carriage ranged from 9% to 19% in the few studies conducted in ICU patients (5–7), which is in line with the 10.8% overall rate observed in our work. However, the impact of LOS on the likelihood of carriage acquisition had not been previously investigated. Here, colonization with HLAC-PE was mostly ICU-acquired (82% of cases) and became increasingly prevalent as LOS increased—with a 3% increase in the odds of carriage for each additional day spent in the ICU—affecting up to one-third of patients with LOS longer than 28 days. A link between cephalosporin exposure and the acquisition of HLAC-PE carriage has been reported (7). Conversely, a previous work found no correlation between ICU-level volumes of beta-lactam consumption and the rate of HLAC-PE acquisition (6). Our results from a cohort heavily exposed to antimicrobials indicate that, at the patient level, LOS outweighs beta-lactam exposure—and especially cephalosporin exposure—in predicting HLAC-PE acquisition (Table 3). Of note, and contrary to ESBL-PE acquisition in ICU patients (6, 11), HLAC-PE acquisition mostly depends on mutant selection from the endogenous intestinal microbiota and not on cross-transmission (5). Antibiotic-related alterations of the anaerobic flora may ease the emergence of such mutants from subdominant, wild-type Enterobacteriaceae populations, as suggested by the marginal effect observed with metronidazole exposure.

HLAC-PE infections occurred in only 15% of carriers; however, HLAC-PE carriage was associated with a significant increase in carbapenem exposure, even in the absence of infection. Given the low rate of ESBL-PE cocolonization, this result probably reflects the local policy of empirical carbapenem use when sepsis occurs in identified HLAC-PE carriers. Indeed, taking into account a documented carriage of multidrug-resistant pathogens is advocated when choosing the empirical therapy for ICU patients with proven or suspected severe sepsis (1, 12). Yet, carbapenem overuse in patients carrying ESC-resistant Enterobacteriaceae may promote the dissemination of carbapenem-resistant Gram-negative bacilli (13, 14), including carbapenemase-producing Enterobacteriaceae in ICUs with endemicity (15). Cefepime has been recently proposed as a safe carbapenem-sparing option for the treatment of HLAC-PE infection (16, 17) and warrants further evaluation in critically ill patients to reduce carbapenem consumption in the ICU.

The single-center design represents the main limitation of this study. However, colonization pressure and cross-transmission appear to play a minor role in HLAC-PE acquisition (5), which suggests that the present results may not have been significantly impacted by our local ecology. Next, since HLAC-PE were identified using routine phenotypic methods and not molecular assays, we were unable to determine whether the HLAC phenotype resulted from the derepression of a chromosomal bla_ampC gene or from the acquisition of a plasmid-borne AmpC variant. Nevertheless, the available evidence suggests that the former mechanism accounted for most of the HLAC-PE isolates (2). Also, antimicrobial exposure was measured for each class as the cumulative number of treatment days; therefore, the respective impact of each course could not be determined for patients with multiple courses of the same class during their ICU stay. Lastly, in our hospital, patients hospitalized in non-ICU wards are not routinely screened for HLAC-PE carriage. Thus, some patients transferred from wards and classified as noncarriers at ICU admission may have been transient carriers during their hospital stay. Likewise, data on previous HLAC-PE carriage or infections may have been missing for patients hospitalized in other hospitals within the year preceding ICU admission.

In conclusion, intestinal colonization with HLAC-PE is common in critically ill patients and is mainly acquired during the ICU stay. In this cohort with high antimicrobial exposure, LOS was the main predictor of carriage acquisition. Preserving the intestinal microbiota from antibiotic-related alterations appears essential to prevent the emergence of HLAC-PE.