Incidence and Risk Factors for Recurrence of NDM- and KPC-Producing Enterobacterales after Achieving Clearance: A Retrospective Study

Abstract

Background

We aimed to determine the incidence and risk factors of recurrence after clearance of carbapenemase-producing Enterobacterales (CPE). We also sought to investigate whether there were differences in recurrence according to carbapenemase type.

Materials and Methods

We conducted a retrospective case-control study (From January 2022 to December 2023) at a tertiary referral hospital in Korea. Recurrence was defined as positive culture or PCR test results for CPE with the same genotype as the first CPE isolation within 1 year after clearance.

Results

Among 1,503 patients with New Delhi metallo-β-lactamase (NDM)- or Klebsiella pneumoniae carbapenemase (KPC)-producing Enterobacterales in whom clearance status could be assessed, 569 (37.8%) achieved clearance. Clearance was observed in 349 of 749 patients (46.6%) with NDM-producing isolates and 220 of 754 patients (29.2%) with KPC-producing isolates (P<0.001). Recurrence occurred in 65/569 (11.4%) patients after clearance - 34/349 (9.7%) in NDM group and 31/220 (14.1%) in the KPC group (P=0.112). Multivariate analysis identified central venous catheter dwelling after clearance as an independent risk factors for CPE recurrence (adjusted odds ratio [aOR], 2.27; 95% confidence interval [CI], 1.16–4.43; P=0.016).

Conclusion

Recurrence occurred in approximately one out of ten patients after clearance within one year, and did not differ significantly by carbapenemase type. Post-clearance indwelling devices, particularly central venous catheters, were associated with recurrence, underscoring the need for targeted infection control strategies after documented clearance.

Graphical Abstract

Introduction

The timing of discontinuation of isolation and contact precautions for patients with carbapenem-resistant Enterobacterales (CRE) is controversial. According to the Centers for Disease Control and Prevention, there is insufficient data to make a general recommendation for the timing of discontinuation of isolation for patients with CRE. However, they recommend waiting for at least 3–6 months since the last positive culture or screen [1]. Expert guidance from the Society for Healthcare Epidemiology of America recommends that acute care hospitals maintain contact precautions for CRE carriers indefinitely [2]. In Israel, an individual with carbapenemase-producing Enterobacterales (CPE) carrier status can only lose their status if they have at least three consecutive negative tests, including two screening rectal swab cultures and one PCR test, beginning at least 3 months after the last positive culture and taken over a period of at least 7 days [3].

In Korea, three consecutive negative rectal swab cultures obtained at least 3 days apart are recommended before discontinuing contact precautions for patients colonised with CRE [4]. However, there have been reports of the recurrence of CRE isolation after three consecutive negative surveillance cultures [5,6]. There is a risk of transmission from patients who are roomed with multiple individuals after discontinuing isolation and experience a subsequent recurrence of CRE. Therefore, post-clearance recurrence is an important concern for CRE control.

However, prolonged contact precautions and single-room isolation are associated with high costs, increased resource utilisation, patient safety issues and the burden of emotional stress on the patient [7,8]. Additionally, the number of patients with CRE is increasing, leading to a shortage of single isolation rooms and resource constraints. Thus, identifying risk factors for recurrence among patients who have achieved clearance will help to reduce the risk of transmission and the amount of money spent on single-room isolation.

Previous studies conducted at our institution have shown that Klebsiella pneumoniae carbapenemase (KPC)-producing Enterobacterales are less likely to achieve clearance than New Delhi metallo-β-lactamase (NDM)-producing isolates [9], are more prone to progress from colonization to infection, and are associated with poorer clinical outcomes once infection occurs [10,11]. However, data on whether recurrence rates after clearance differ according to the carbapenemase type remain limited.

This study aimed to determine the incidence of recurrence and identify risk factors for recurrence among patients who achieved CPE clearance and establish whether there were differences in the recurrence rate according to the carbapenemase type (NDM or KPC).

Materials and Methods

1. Settings and definitions

We conducted a retrospective case-control study at a 2,736-bed tertiary referral hospital in Seoul, Korea, of patients with NDM- and KPC-producing Enterobacterales isolated from January 2022 to December 2023, and they were followed up for 1 year. Cases were defined as patients who experienced recurrence of CRE, and controls were those who did not.

Active surveillance for CPE was implemented as follows: In the intensive care units (ICUs), surveillance cultures are performed upon admission and then weekly during the ICU stay. In the hematology ward, cultures are obtained at admission and every three weeks during hospitalization. In the liver transplant unit and the gastrointestinal medicine wards (liver disease section), all patients are screened upon admission. In the biliary medical/surgical wards, screening is conducted for high-risk patients, such as those transferred from other hospitals or with a prior history of admission to our institution. In addition, when a patient with CPE is identified, contact screening is performed for all patients who shared the same room during the exposure period.

Contact precautions (gown and gloves) and single-room isolation were implemented for patients colonized or infected with CRE. Weekly follow-up surveillance cultures from rectal swabs were performed during the hospital stay until three consecutive negative cultures. Single-room isolation was discontinued when three weekly consecutive rectal swabs tested negative. Thereafter, monthly follow-up samples were cultured for inpatients, and contact precautions were eventually discontinued after two consecutive negative cultures. Patients who were discharged prior to clearance and readmitted were screened at 2-day intervals, and contact precautions were discontinued if two consecutive negative results were obtained.

Clinical data were collected from patients’ electronic medical records, including demographic information, pre-existing medical conditions, microbiological data at the time of first CPE isolation, antibiotic exposure, length of stay in the intensive care unit, and presence of indwelling devices at the time after clearance. Clearance was defined as three consecutive negative cultures for patients during hospitalization, and two consecutive negative cultures for those who were readmitted after discharge. Recurrence was defined as a positive culture from a rectal swab or clinical specimen with the same genotype as the first CPE isolation or a positive PCR test with the same genotype as the first CPE isolation within 1 year after clearance. The follow-up period was defined as the time from clearance to either CPE recurrence, in-hospital death, or one year, whichever occurred first.

2. Ethics statement

The study protocol was approved by the Institutional Review Board at Asan Medical Center (IRB No. 2024-0742). Informed consent was waived by the ethics committee because of its retrospective nature, no intervention was involved, and no patient-identifying information is included.

3. Microbiology

All samples were cultured on ChromID CARBA agar (bioMérieux, Marcy I’Etoile, France) at 35°C for 24 h. Colonies suggestive of CRE were selected and subcultured on McConkey agar plates at 35°C for 24 h. The isolates were then identified using a MALDI-TOF MS system (Bruker Daltonics, Billerica, MA, USA) according to the manufacturer’s instructions. Antimicrobial susceptibility testing was performed using the Kirby–Bauer disk diffusion method. Carbapenemases were detected using NG-Test Carba 5 (NG Biotech, Guipry, France) following the manufacturer’s protocol.

4. Statistical analysis

Student t-test or Mann-Whitney U test were used to compare differences between continuous variables, and Pearson chi-squared test or Fisher exact test were used to compare differences between categorical variables, as appropriate. Variables with a P-value <0.100 in univariate analysis were considered as candidates for multivariable analysis. Among these, a multivariable logistic regression model was constructed using variables selected a priori based on clinical relevance and biological plausibility. Four variables (hematologic malignancy, ICU stay after clearance, dwelling of a central venous catheter after clearance, and carbapenem use after clearance) were simultaneously included in the primary model.

Sensitivity analyses were conducted to assess the robustness of the primary findings and to address potential concerns regarding model instability and correlated exposures. Specifically, dwelling of a Foley catheter after clearance was added to evaluate the unexpected inverse association observed in univariate analysis; any antibiotic use before CPE acquisition was included to examine the potential impact of correlated antibiotic exposure variables; and Kluyvera species was added to assess the influence of a rare pathogen with a small number of events on model stability.

All multivariable and sensitivity analyses were performed using the enter method. We used the Kaplan–Meier method stratified by carbapenemase type (KPC and NDM) to analyse the time to develop recurrence within 365 days after clearance. A two-tailed p-value < 0.05 was considered statistically significant. All statistical analyses were performed using SPSS (version 24.0, IBM Corp, Armonk, NY, USA).

Results

We identified a total of 1,878 patients with NDM- or KPC-producing Enterobacterales during the study period, including 889 and 989 patients with NDM- and KPC-producing Enterobacterales, respectively. Of these, 1,503 (80.0%) underwent the full clearance process, either during initial hospitalization or upon readmission. Overall, 569 of 1,503 (37.8%) patients achieved clearance, including 349/749 (46.6%) with NDM-producing Enterobacterales and 220/754 (29.2%) with KPC-producing Enterobacterales (P<0.001).

The median time to clearance of NDM- and KPC-producing Enterobacterales was 68 (interquartile range [IQR], 30–137) days and 95 (IQR, 40-170) days, respectively (P=0.013). During the one-year period following CPE identification, patients underwent a mean of 14 CPE tests (range, 1–160). After clearance, a mean of 7 tests were performed within one year (range, 1–65). When stratified by carbapenemase type, patients with NDM underwent a median of 5 tests (IQR, 2–10), and those with KPC underwent a median of 4 tests (IQR, 2–10) during the year following clearance (P=0.287).

Recurrence was experienced in 65/569 (11.4%) patients who achieved clearance, including 34/349 (9.7%) who achieved clearance of NDM-producing Enterobacterales and 31/220 (14.1%) who achieved clearance of KPC-producing Enterobacterales (P=0.112). The median follow-up time was 48 days (IQR, 21–70). The median time to recurrence after clearance was 38 days (IQR, 17-52) and 59 days (IQR, 35–100) for NDM- and KPC-producing Enterobacterales, respectively (P=0.012). Kaplan–Meier curves comparing the time to recurrence after clearance between NDM- and KPC-producing Enterobacterales showed no significant difference between the two groups (P-value by log-rank test = 0.160) (Fig. 1).

Figure 1. Kaplan–Meier survival estimates of recurrence after clearance in patients with NDM-producing Enterobacterales compared with patients with KPC-producing Enterobacterales.

NDM, New Delhi metallo-β-lactamase; KPC, Klebsiella pneumoniae carbapenemase.

The recurrence rate among the patients who had three consecutive negative cultures was 8.9% (10/112). Among patients who had two consecutive negative cultures, the recurrence rate was 10.9% (50/457) (P=0.522). Among patients who experienced recurrence after initial clearance, 12 (18.5%) subsequently achieved clearance again, with a median interval of 151 days (IQR, 105–224 days) from recolonization to re-clearance.

We compared the clinical and microbiological characteristics of patients with (n=65) and without recurrence (n=504) to identify the risk factors for recurrence among the 569 patients who achieved clearance of NDM- or KPC-producing Enterobacterales (Table 1). Compared with patients without recurrence, those with recurrence had a higher prevalence of hematologic malignancy (40.0% vs. 25.0%, P=0.018). Any antibiotic use within 3 months before CPE acquisition was less frequent in the recurrence group than in the non-recurrence group (84.6% vs. 93.1%, P=0.028); however, carbapenem use within 3 months before CPE acquisition was more common among patients with recurrence (43.6% vs. 26.2%, P=0.032). After clearance, patients with recurrence more frequently had a central venous catheter in place (76.9% vs. 53.6%, P<0.001), whereas foley catheter dwelling was less common in the recurrence group (32.3% vs. 53.4%, P=0.002).

Table 1. Clinical and microbiological characteristics of patients with and without CPE recurrence.

Characteristic				Recurrence (n=65)	Non-recurrence (n=504)	P-value
Variables at the time of CPE acquisition
	Age, median (IQR)			62 (51–68)	63 (53–70)	0.182
	Male			28 (43.1)	176 (34.9)	0.246
	Underlying disease or condition
		Diabetes mellitus		16 (24.6)	146 (29.0)	0.595
		Hematologic malignancy		26 (40.0)	126 (25.0)	0.018
		Solid cancer		28 (43.1)	243 (48.2)	0.526
		Chemotherapy within the past 3 months		22 (33.8)	159 (31.5)	0.816
		Surgery within the past 3 months		19 (29.2)	182 (36.1)	0.337
		End-stage renal disease		4 (6.2)	34 (6.7)	>0.999
		Neutropenia within the past month (ANC <500/µL)		35 (53.8)	255 (50.6)	0.720
		Any antibiotic use within 3 months before CPE acquisition		55 (84.6)	469 (93.1)	0.029
		Carbapenem use within 3 months before CPE acquisition (vs. other antibiotics)		24 (43.6)	123 (26.2)	0.032
After clearance
	Any antibiotic use			53 (81.5)	374 (74.2)	0.229
	Carbapenem use			25 (38.8)	133 (26.4)	0.055
	ICU stay			5 (7.7)	14 (2.8)	0.054
	Drain
		Central venous catheter		50 (76.9)	270 (53.6)	<0.001
		Foley catheter		21 (32.3)	269 (53.4)	0.002
		JP drain		8 (12.3)	62 (12.3)	>0.999
Days between first CPE isolation and clearance, median (IQR)				58 (29–150)	77 (35–149)	0.300
Microbiological variables at the time of CPE acquisition
	Organisms by carbapenemase
		KPC		31 (47.7)	189 (37.5)	0.146
			Klebsiella pneumoniae	27 (87.1)	172 (91.0)	0.498
			Klebsiella spp. other than K. pneumoniae	0	2 (1.1)	>0.999
			Escherichia coli	2 (6.5)	12 (6.3)	>0.999
			Citrobacter spp.	1 (3.2)	5 (2.6)	>0.999
		NDM		34 (52.3)	315 (62.5)	0.146
			K. pneumoniae	11 (32.4)	111 (35.2)	0.880
			Klebsiella spp. other than K. pneumoniae	4 (11.8)	43 (13.7)	>0.999
			Citrobacter spp.	10 (29.4)	107 (34.0)	0.726
			Escherichia coli	5 (14.7)	58 (18.4)	0.757
			Kluyvera spp.	5 (14.7)	5 (1.6)	0.008a
			Raoultella ornithinolytica	2 (5.9)	8 (2.5)	0.247
			Enterobacter spp.	0	7 (2.2)	>0.999
Source of initial CPE isolation						0.598
	Screening			57 (87.7)	448 (88.9)	0.943
	Blood			3 (4.6)	8 (1.6)	0.118
	Other			5 (7.7)	48 (9.5)	0.804
In-hospital mortality				10 (15.4)	75 (14.9)	0.852

^aP-value was adjusted for multiple comparisons using Bonferroni correction.

CPE, carbapenemase-producing Enterobacterales; IQR, interquartile range; ANC, absolute neutrophil count; ICU, intensive care units; JP, Jackson–Pratt; KPC, Klebsiella pneumoniae carbapenemase; NDM, New Delhi metallo-β-lactamase.

Regarding microbiologic characteristics, Kluyvera species were more frequently identified at the initial CPE acquisition among patients who subsequently experienced recurrence (14.7% vs. 1.6%, P=0.008). However, the type of carbapenemase was not statistically significant between the recurrence and non-recurrence groups (KPC in the recurrence group, 47.7%; KPC in the non-recurrence group, 37.5%; P=0.136). In-hospital mortality did not differ according to recurrence status.

Multivariate analysis identified central venous catheter dwelling after clearance (adjusted odds ratio [aOR], 2.27; 95% confidence interval [CI], 1.16–4.43; P=0.016) as an independent risk factor for CPE recurrence (Table 2). To evaluate the robustness of the findings, sensitivity analyses were performed by separately adding dwelling of a Foley catheter after clearance (Supplementary Table 1), any antibiotic use before CPE acquisition (Supplementary Table 2), and Kluyvera species (Supplementary Table 3) to the primary model. In all sensitivity analyses, dwelling of a central venous catheter after clearance remained significantly associated with recurrence, with consistent direction and magnitude of effect estimates.

Table 2. Univariate and multivaria analysis of risk factors for CPE recurrence among patients who achieved clearance (n=569).

Variable	Univariate analysis, OR (95% CI)	Multivariate analysisa
		aOR (95% CI)	P-value
Hematologic malignancy	2.00 (1.17–3.42)	1.43 (0.79–2.57)	0.237
Any antibiotic use before CPE acquisition	0.42 (0.20–0.90)
Carbapenem use before CPE acquisition	1.81 (1.05–3.12)
ICU stay after clearance	2.92 (1.02–8.38)	2.10 (0.70–6.31)	0.187
Dwelling of central venous catheter after clearance	2.89 (1.58–5.28)	2.27 (1.16–4.43)	0.016
Dwelling of Foley catheter after clearance	0.42 (0.24–0.72)
Carbapenem use after clearance	1.84 (1.02–2.98)	1.20 (0.67–2.14)	0.539
Kluyvera species	8.32 (2.34–29.56)

^aVariables were selected a priori based on clinical relevance and biological plausibility. Four variables (hematologic malignancy, ICU stay after clearance, dwelling of a central venous catheter after clearance, and carbapenem use after clearance) were included in the primary multivariable logistic regression model.

CPE, carbapenemase-producing Enterobacterales; OR, odds ratio; CI, confidence interval; aOR, adjusted odds ratio; ICU, intensive care units.

Baseline characteristics differed modestly according to carbapenemase genotype (Supplementary Table 4). Patients with NDM-producing Enterobacterales were slightly older and less likely to be male than those with KPC-producing isolates. Solid cancer was more frequent in the NDM group, whereas prior antibiotic exposure, particularly carbapenem use before CPE acquisition, was more common in the KPC group. After clearance, any antibiotic use was more frequent among patients with NDM-producing isolates.

Microbiologically, K. pneumoniae predominated in the KPC group, whereas the NDM group showed greater species diversity, including higher proportions of non–K. pneumoniae Klebsiella species, Escherichia coli, and Citrobacter spp. (all Bonferroni-adjusted P < 0.007). The source of initial CPE isolation also differed between groups, with screening cultures more common in the NDM group. However, recurrence and in-hospital mortality did not differ significantly according to carbapenemase genotype.

Discussion

In our study, approximately 40.0% of patients with NDM- or KPC-producing Enterobacterales achieved clearance, and patients harbouring NDM-producing Enterobacterales achieved clearance more frequently than those harbouring KPC-producing Enterobacterales. Among the patients who achieved clearance, 11.4% subsequently experienced recurrence. However, the recurrence rate did not differ significantly between the two carbapenemase types. Central venous catheter dwelling after clearance was an independent risk factor for CPE recurrence.

Consistent with our previous institutional study [9], patients harboring NDM-producing organisms achieved clearance significantly more frequently than those harboring KPC-producing organisms. Several potential explanations may be considered. These include intrinsic differences in bacterial characteristics, such as the predominance of K. pneumoniae in KPC versus non–K. pneumoniae species in NDM; the virulence and host fitness of prevalent KPC-producing K. pneumoniae strains in Korea (e.g., ST307 or ST11) [12,13]; and the possibility that NDM-producing organisms may be present in lower intestinal colonization burdens. Further studies, including investigations of microbiome differences between patients colonized with NDM- and KPC-producing organisms, are warranted to better elucidate these mechanisms.

The recurrence rate of CRE has been reported to range from 7.9% to 33.3% in patients who achieved clearance [3,5,6,14,15,16,17]. However, the definition of clearance varied across studies, including negative test results at readmission, a single negative test, two consecutive negative cultures, three consecutive negative cultures and negative results from two cultures and one PCR test. The recurrence rate in our study (11.4%) is similar to Israeli studies (7.9%–13.0%) [3,15,17]. Although Lewis et al. reported 15.0% recurrence after two consecutive negative cultures and 25.0% recurrence after three consecutive negative cultures [5], their sample size was small. Thus, among patients who have achieved clearance, the likelihood of recurrence may not be high, particularly among those who do not have any risk factors. From an infection control perspective, once patients meet the clearance criteria (e.g., three consecutive negative cultures at ≥3-day intervals), contact precautions may be safely discontinued, particularly among those who do not have any risk factors, and standard precautions resumed for the remainder of the hospital stay. Follow-up testing could still be considered, particularly in high-risk patients or upon readmission.

Data comparing recurrence rate after clearance between different carbapenemase types are scarce. We found no difference in the recurrence rate between NDM- and KPC-producing Enterobacterales. Israeli [17] and Korean [6] studies also reported that there was no difference in the recurrence rate according to the carbapenemase type. Although more data are required, this suggests that type of carbapenemase may not be associated with recurrence.

Antibiotic exposure, shorter time between last positive CRE isolation and clearance, recurrent admission, presence of a foreign body, presence of diabetes mellitus, central venous catheterisation, K. pneumoniae and a positive sputum culture have been reported as risk factors for CRE recurrence [6,15,16]. Consistent with these findings, our study also identified the presence of a central venous catheter after clearance as an independent risk factor for recurrence. In contrast, dwelling of a Foley catheter after clearance was inversely associated with recurrence in univariate analysis; however, this association was not robust and may reflect residual or unmeasured confounding, including differences in patient characteristics or competing clinical risks. Similarly, antibiotic use prior to CPE acquisition showed an inverse association with recurrence. Selection bias cannot be excluded, as patients with substantial antibiotic exposure may have experienced early mortality or achieved sustained clearance before being at risk for recurrence. Colonization or infection with specific bacterial species, including Kluyvera spp., appeared to influence recurrence in exploratory analyses; however, given the small number of cases, this observation should be interpreted cautiously and warrants confirmation in larger studies.

Our study has several limitations. First, due to the observational nature of the study, the intensity and duration of follow-up CPE assessment was varied across patients, as testing was conducted as a part of routine clinical care. As we did not systematically perform follow-up CPE surveillance tests to assess recurrence, which may have led to misclassification of some patients as having no recurrence. This could result in underestimation of the recurrence rate, and may also limit the validity of our analysis of risk factors for recurrence due to potential misclassification bias. Second, we defined recurrence as the isolation of the same carbapenemase genotype within 1 year after clearance. However, due to the lack of whole genome sequencing data for isolates at the time of initial detection and recurrence, we were unable to distinguish between new acquisition and persistence (recrudescence). Additionally, cases where the same carbapenemase genotype was isolated more than 1 year after clearance may indicate a new acquisition. In a previous study, whole genome sequencing analysis of different isolates (interval between specimens >1 year) cultured from the same patients revealed that three of four patients showed new acquisition rather than persistent colonisation [18]. Third, since our analysis focused only on KPC and NDM, which are the most prevalent carbapenemases in South Korea, the results may not be generalizable to settings with different carbapenemase distributions or to other geographic regions.

In conclusion, patients harboring NDM-producing Enterobacterales achieved clearance more frequently than patients harboring KPC-producing Enterobacterales. Among patients who achieved clearance, 11.4% subsequently experienced recurrence. Given the relatively low recurrence rate, discontinuation of contact precautions in patients with clearance may be considered, particularly for patients without risk factors and in resource-limited settings.

SUPPLEMENTARY MATERIALS

Supplementary Table 1

Multivariable logistic regression model including dwelling of Foley catheter after clearance (sensitivity analysis)

Supplementary Table 2

Multivariable logistic regression model including any antibiotic use before CPE acquisition (sensitivity analysis)

Supplementary Table 3

Multivariable logistic regression model including Kluyvera species (sensitivity analysis)

Supplementary Table 4

Clinical and microbiological characteristics of patients according to carbapenemase genotype