Carbapenem-resistant Acinetobacter baumannii and Klebsiella pneumoniae across a hospital system: impact of post-acute care facilities on dissemination

Abstract

Background

Resistance to carbapenems among Acinetobacter baumannii and Klebsiella pneumoniae presents a serious therapeutic and infection control challenge. We describe the epidemiology and genetic basis of carbapenem resistance in A. baumannii and K. pneumoniae in a six-hospital healthcare system in Northeast Ohio.

Methods

Clinical isolates of A. baumannii and K. pneumoniae distributed across the healthcare system were collected from April 2007 to April 2008. Antimicrobial susceptibility testing was performed followed by molecular analysis of carbapenemase genes. Genetic relatedness of isolates was established with repetitive sequence-based PCR (rep-PCR), multilocus PCR followed by electrospray ionization mass spectrometry (PCR/ESI-MS) and PFGE. Clinical characteristics and outcomes of patients were reviewed.

Results

Among 39 isolates of A. baumannii, two predominant genotypes related to European clone II were found. Eighteen isolates contained bla_OXA-23, and four isolates possessed bla_OXA-24/40. Among 29 K. pneumoniae isolates with decreased susceptibility to carbapenems, two distinct genotypes containing bla_KPC-2 or bla_KPC-3 were found. Patients with carbapenem-resistant A. baumannii and K. pneumoniae were elderly, possessed multiple co-morbidities, were frequently admitted from and discharged to post-acute care facilities, and experienced prolonged hospital stays (up to 25 days) with a high mortality rate (up to 35%).

Conclusion

In this outbreak of carbapenem-resistant A. baumannii and K. pneumoniae across a healthcare system, we illustrate the important role post-acute care facilities play in the dissemination of multidrug-resistant phenotypes.

Introduction

The emergence of bacterial resistance to carbapenems is the focus of significant attention as it heralds an age of absent effective antibiotic options.^1,2 Carbapenem-resistant Acinetobacter baumannii and Klebsiella pneumoniae are notorious because of their increasing prevalence and their ability to cause outbreaks.^3,4 Carbapenem-resistant organisms affect hospitalized patients who are often debilitated, severely ill, experience high mortality rates and prolonged hospitalizations.^5–7 Furthermore, infections with carbapenem-resistant organisms may require patients to receive additional treatment in long-term care facilities (LTCFs) or long-term acute care hospitals (LTACHs), and impose extraordinary diagnostic, therapeutic and infection-control efforts that add significantly to the cost of healthcare.^8,9

The outbreak of A. baumannii associated with United States military operations in Iraq generated special interest in this organism.^10,11 Outside of the military experience, carbapenem-resistant A. baumannii became endemic in New York hospitals for close to a decade and outbreaks of carbapenem-resistant A. baumannii also occurred in hospitals in Pittsburgh and Chicago. Carbapenem resistance among these outbreak isolates was attributed to the expression of class D β-lactamases such as OXA-23, OXA-24/40 and OXA-58, which inactivate carbapenems.^12–14 At the same time, hospitals in New York City reported an increased incidence of carbapenem-resistant K. pneumoniae.¹⁵ Resistance in this instance was mediated by KPC (K. pneumoniae carbapenemase), a class A β-lactamase with the ability to hydrolyse carbapenems and extended-spectrum β-lactams and not affected by the commercially available β-lactamase inhibitors (i.e. tazobactam, sulbactam and clavulanate).¹⁶

Recently, a dominant clone of carbapenem-resistant K. pneumoniae harbouring KPC was detected at multiple medical centres in the Eastern USA.¹⁷ As suggested by the most recent (2006–2007) summary of data reported to the National Healthcare Safety Network at the CDC, carbapenem-resistant organisms are becoming more widespread in US hospitals than previously anticipated.¹⁸

Carbapenem-resistant A. baumannii and K. pneumoniae are also detected outside of acute care hospital settings in LTCFs and LTACHs.^19–22 This new development poses additional difficulties for the control and treatment of carbapenem-resistant organisms and brings attention to their impact and transmission dynamics at the level of healthcare systems consisting of acute care hospitals, LTACHs and LTCFs. Here, we describe the clinical and molecular epidemiology of outbreaks of carbapenem-resistant A. baumannii and K. pneumoniae affecting a healthcare system in Northeast Ohio and illustrate the impact of this emerging problem.

Methods

Setting

University Hospitals (UH) is an integrated healthcare system that consists of outpatient centres, acute care and post-acute care facilities (PACFs) located in Northeast Ohio and serves a population of more than two million. UH comprises University Hospitals Case Medical Center (UHCMC—Hospital A), a 1032 bed tertiary medical centre with 65 intensive care unit (ICU) beds, active solid organ and bone marrow transplant programmes and associated full-service paediatric hospital and women's hospital, as well as a 50 bed short-term skilled nursing facility. UH also includes five community hospitals (ranging from 25 to 225 beds—Hospitals B–F) and a facility combining assisted living, long-term skilled nursing, and long-term acute care services, serving as both an LTCF and an LTACH. UH performs >4.5 million outpatient procedures and nearly 63 000 inpatient discharges annually. The UH system is served by a central microbiology laboratory located at UHCMC that processes ∼150 000 bacterial cultures annually. At the end of 2007, the increasingly frequent occurrence of multidrug-resistant (MDR) A. baumannii and K. pneumoniae was noted (including carbapenem-resistant isolates), prompting this retrospective observational investigation of the clinical and molecular epidemiology of these organisms in the UH healthcare system.

Bacterial isolates and antimicrobial susceptibility testing

Clinical isolates of A. baumannii that displayed an MDR profile and K. pneumoniae with decreased susceptibility to carbapenems (ertapenem MIC ≥4 mg/L, or imipenem or meropenem MIC ≥8 mg/L) identified in the microbiology laboratory at UHCMC between April 2007 and April 2008 were studied. Herein we define MDR as resistance to antibiotics from three or more different classes to which these organisms are expected to be susceptible (i.e. aminoglycosides, quinolones, extended-spectrum cephalosporins, β-lactam/β-lactamase inhibitor combinations, carbapenems and tigecycline). Only the initial isolate recovered from each patient was included in the study.

Bacterial identification to the species level and routine antimicrobial susceptibility testing (AST) were done with the MicroScan^® system (Siemens Healthcare Diagnostics, Sacramento, CA, USA). AST results were interpreted according to breakpoints defined by the CLSI.²³ Additional testing to determine MICs of ertapenem (in the case of K. pneumoniae), doripenem, imipenem, meropenem, tigecycline and colistin was done with the use of Etest strips (AB bioMérieux, Sweden). Susceptibility was determined according to CLSI criteria in the case of imipenem, meropenem (MICs ≤4 mg/L), ertapenem (MIC ≤2 mg/L) and colistin (MIC ≤2 mg/L; non-Enterobacteriaceae breakpoint). Criteria set forth by the USA Food and Drug Administration (FDA) to determine susceptibility to tigecycline among Enterobacteriaceae (MIC ≤2 mg/L), were applied to both A. baumannii and K. pneumoniae. Similarly, FDA criteria were applied to determine susceptibility to doripenem (MIC ≤1 mg/L for A. baumannii and MIC ≤0.5 mg/L for K. pneumoniae). In the case of K. pneumoniae, clinical isolates with decreased susceptibility to carbapenems were screened for the presence of carbapenemases using the modified Hodge test.²⁴

PCR analyses for detection of β-lactamase genes associated with carbapenem resistance

PCR was performed to detect genes coding for β-lactamases associated with carbapenem resistance. In the case of A. baumannii, bla_OXA-23, bla_OXA-24/40, bla_OXA-58, bla_KPC, bla_IMP and bla_VIM were investigated. Additionally, PCR was carried out to detect intrinsic β-lactamase genes found in A. baumannii: bla_OXA-51/69 and bla_ADC. These analyses were performed as previously described by Hujer et al.¹¹ In the case of K. pneumoniae, PCR and DNA sequence analysis for bla_KPC genes were performed using primers and conditions previously reported.¹⁷

Molecular genotyping and determination of genetic relatedness

In order to establish their genetic relatedness, clinical isolates of A. baumannii and K. pneumoniae were typed using the automated rep-PCR-based strain typing system (DiversiLab™; bioMérieux, Athens, GA, USA). Extraction of DNA was performed using the UltraClean™ Microbial DNA Isolation Kit (Mo Bio Laboratories, Inc., Carlsbad, CA, USA). Rep-PCR was performed using the DiversiLab Acinetobacter™ and Klebsiella™ kits, respectively. DNA fragment separation and detection was done using the Agilent^® 2100 Bioanalyzer (Agilent Technologies, Santa Clara, CA, USA) and results were analysed and interpreted with the DiversiLab™ web-based software using the Kullback–Leibler method, as previously validated for K. pneumoniae and A. baumannii; isolates that clustered >95% were considered related.^17,25,26

Genotyping by PFGE was performed in 39 clinical isolates of A. baumannii and 29 clinical isolates of K. pneumoniae, according to previously described methods.²⁷ Genomic DNA was digested using the enzymes ApaI (for A. baumannii) and XbaI (for K. pneumoniae). The restriction fragments were separated by PFGE using a temperature-controlled CHEF DR III system (Bio-Rad Laboratories, Hercules, CA, USA). After electrophoresis, the gels were stained with ethidium bromide and photographed under UV light. Differences in the band patterns determined by visual inspection were used to establish the genetic relatedness of the isolates, following recommendations advanced by Tenover et al.²⁸

PCR coupled with electrospray ionization mass spectrometry (PCR/ESI-MS), a form of multilocus sequence typing (MLST), was performed in order to further establish the genetic relatedness of the clinical isolates of A. baumannii. The Acinetobacter Genotyping Kit was used on the Ibis T5000 Biosensor (Ibis Biosciences, Inc.). The primers contained in this kit are designed to yield amplicons with unique mass signatures from six highly conserved genes (efp, trpE, adk, mutY, fumC and ppa) that are then measured by high-performance MS and distinguished by base composition analysis. Results of base composition of each isolate are matched with a pre-existing database to yield unique sequence types (STs) identifying groups of strains.²⁹

Chart review

The medical records of patients in whom carbapenem-resistant A. baumannii or K. pneumoniae was isolated were reviewed, after approval from the Institutional Review Board. The following demographic and clinical data were obtained: age, sex, date and location prior to admission and source of isolate. Criteria set by the National Nosocomial Infections Surveillance System were used to determine whether isolation of A. baumannii or K. pneumoniae indicated colonization or infection.³⁰ Healthcare acquisition was defined as detection of colonization or infection at least 72 h after arrival at the medical facility. Co-morbidity scores (age adjusted) were determined according to the Charlson weighted index.³¹ The following predisposing conditions were investigated: endotracheal intubation or tracheostomy, intravascular catheterization, bladder catheterization, percutaneous endoscopic gastrostomy (PEG), chronic haemodialysis, previous hospitalization and antibiotic treatment prior to isolation of A. baumannii or K. pneumoniae. Additionally, records were examined to establish the duration of hospitalization, disposition and isolation of other MDR organisms, such as Pseudomonas aeruginosa, vancomycin-resistant Enterococcus spp. (VRE) and methicillin-resistant Staphylococcus aureus (MRSA). Outcome (dead/alive) was recorded; death was attributed to infection when it occurred during the acute phase of the infection or when the patient was still receiving treatment. Follow-up of patients to determine mortality was carried out 28 days after onset of infection.

Results

Clinical isolates and antimicrobial susceptibility testing

Between April 2007 and April 2008, 54 single-patient isolates of A. baumannii, 39 of which demonstrated an MDR phenotype, were found distributed across the healthcare system, including the LTACH (n = 7), two of the community hospitals (n = 7) and the tertiary care centre (n = 25). MDR A. baumannii strains were first isolated in the hospital system's PACF in April 2007, while dissemination into the hospitals occurred after August 2007. In the same period of time, 29 single-patient isolates of K. pneumoniae with decreased susceptibility to carbapenems were also identified. K. pneumoniae were present at the tertiary care hospital (n = 16), two community hospitals (n = 8) and the LTACH (n = 5).

Results of AST (Table 1) revealed that resistance to imipenem and meropenem was present in 85% of the A. baumannii isolates and 90% of the K. pneumoniae strains, while resistance to doripenem approached 100% for A. baumannii and K. pneumoniae. However, the MIC distributions of all the carbapenems were similar (MIC₅₀ and MIC₉₀ >32 mg/L, range 0.75–>32 mg/L) and the discrepancy in susceptibility rates likely reflects the lower resistant breakpoint for doripenem. The modified Hodge test suggested the presence of a carbapenemase in all of the K. pneumoniae isolates. Antimicrobial alternatives to carbapenems that demonstrated in vitro activity included, in the case of A. baumannii, ampicillin/sulbactam (33%) and amikacin (18%). MICs of tigecycline were mostly in the intermediate range (MIC₅₀ and MIC₉₀ 4 mg/L, range 1.5–6 mg/L). Colistin retained activity against A. baumannii (MIC₅₀ 0.5 mg/L and MIC₉₀ 1 mg/L, range 0.25–4 mg/L), except for one colistin-resistant isolate (MIC 4 mg/L), which was also non-susceptible to tigecycline (MIC 3 mg/L) and was only susceptible to amikacin. In the case of K. pneumoniae, amikacin and gentamicin offered some in vitro activity (45% and 31%, respectively). MICs of tigecycline were mostly in the susceptible range (MIC₅₀ 1.5 mg/L and MIC₉₀ 2 mg/L, range 0.75–4 mg/L). Colistin remained active against most isolates of K. pneumoniae (MIC₅₀ 0.38 mg/L and MIC₉₀ 0.75 mg/L, range 0.25–6 mg/L), although there were two instances of resistance to colistin. The two colistin-resistant isolates (MICs of 4 and 6 mg/L, respectively) were only susceptible to tigecycline (MIC 1.5 mg/L) and, in one case, to amikacin and, in the other, to gentamicin.

Table 1.

Antimicrobial susceptibility testing among A. baumannii and K. pneumoniae isolates from UH

	Number of susceptible isolates (%)
	MicroScan							Etest
Organism	CAZ	FEP	TZP	AMK	GEN	CIP	SAM	TGCa	CST	DORa	IPM	MEM	ETP
A. baumannii (n = 39)	3 (8)	3 (8)	n/a	7 (18)	0 (0)	0 (0)	13 (33)	3 (8)	38 (98)	1 (2.5)	6 (15.4)	6 (15.4)	n/a
K. pneumoniae (n = 29)	0 (0)	0 (0)	0 (0)	9 (31)	13 (45)	0 (0)	0 (0)	27 (93)	27 (93)	0 (0)	3 (10)	3 (10)	3 (10)

CAZ, ceftazidime; FEP, cefepime; TZP, piperacillin/tazobactam; AMK, amikacin; GEN, gentamicin; CIP, ciprofloxacin; SAM, ampicillin/sulbactam; TGC, tigecycline; CST, colistin; DOR, doripenem; IPM, imipenem; MEM, meropenem; ETP, ertapenem.

^aSusceptibility determined according to FDA criteria for doripenem and tigecycline. For all other antibiotics, CLSI criteria were used.

Molecular genotyping

Analysis of PFGE patterns from A. baumannii revealed five main groups (Figure 1a). The largest group (PFGE type A) comprised 26/39 isolates. There were three additional groups (types B, C and D) comprising seven, three and two isolates, respectively. The remaining isolate possessed a unique band pattern indicating a different strain type (E). Based on PCR/ESI-MS, PFGE type A corresponded to ST10 (n = 24) and ST86 (n = 1) (one isolate was not typeable by PCR/ESI-MS). PFGE type B corresponded to ST12 (n = 6) and ST1 (n = 1) and type C to ST54 (n = 3). The two isolates belonging to PFGE type D corresponded to two different STs (ST51 and ST68) that are very closely related. Type E corresponded to a unique ST27. With regard to rep-PCR, it did not distinguish between PFGE types A and B and ST10 and ST12, although rep-PCR also identified two groups among the predominant strain types (Figure 1b). Of note, both ST10 and ST12 are identified as part of European clone II;²⁹ PFGE type A from this study is similar to representative isolates of European clone II.³² Rep-PCR differentiated PFGE type C/ST54, PFGE type D/ST51–ST68 and PFGE type E/ST27. Isolates belonging to the predominant strains were found in different components of the healthcare system, including the LTACH.

Figure 1.

(a) PFGE of ApaI macrorestriction of representative A. baumannii isolates, demonstrating PFGE type A (Ab21, Ab30, Ab31, Ab33, Ab37, Ab40, Ab41 and Ab53), type B (Ab20, Ab22 and Ab39), type C (Ab69, Ab76 and Ab79), type D (Ab51 and Ab83) and type E (Ab52). (b) Dendrogram describing percentage similarity and band patterns of A. baumannii by rep-PCR, interpreted using the Kullback–Leibler method. Similarity within group 1 (1–13) and within group 2 (14–29) is >95%. Similarity between the two groups is >90%. Also in this figure, ST by PCR/ESI-MS, PFGE types, carbapenemase genes detected and distribution by location in hospital ward, medical intensive care unit (MICU), cardiac intensive care unit (CICU), surgical intensive care unit (SICU) and long-term acute care hospital (LTACH) are shown.

Among carbapenem-resistant K. pneumoniae, PFGE demonstrated two groups, types A (n = 17) and B (n = 10) (Figure 2a). There was one isolate with a unique band pattern (Kp57) and an isolate that did not yield an adequate restriction pattern despite multiple attempts (Kp08). Rep-PCR identified two distinct groups, 1 and 2, that corresponded to PFGE types A and B, with the exception of two isolates (Kp45 and Kp59) (Figure 2b). When analysed with rep-PCR, isolate Kp57 also appeared unrelated. We noted that rep-PCR groups 1 and 2 appear related to strains from the Eastern USA and Florida previously typed with this methodology (data not shown).^17,19 Both predominant types were found throughout the healthcare system, including the LTACH. Gentamicin was more frequently active against isolates from group 1 than group 2 (12/17 versus 0/10 susceptible isolates), whereas amikacin was more frequently active against isolates from group 2 than group 1 (5/10 versus 2/17 susceptible isolates) (Figure 2a and 2b).

Figure 2.

(a) PFGE of XbaI macrorestriction of selected K. pneumoniae isolates, demonstrating PFGE type A (Kp01, Kp03, Kp04, Kp05, Kp07, Kp23, Kp24, Kp27 and Kp47) and type B (Kp17, Kp18, Kp22, Kp32, Kp40, Kp42, Kp48 and Kp54). L corresponds to lambda ladder. (b) Dendrogram describing percentage similarity and band patterns of K. pneumoniae by rep-PCR, interpreted using the Kullback–Leibler method. Similarity within group 1 (1–17) and within group 2 (19–28) is ∼95%. Similarity between the two groups is ∼85%. Also in this figure, the type of bla_KPC detected, PFGE types, results of susceptibility testing against amikacin (AMK) and gentamicin (GEN) and location in hospital ward, medical intensive care unit (MICU), surgical intensive care unit (SICU) and long-term acute care hospital (LTACH).

Detection of β-lactamase genes associated with carbapenem resistance

With respect to the detection of β-lactamase genes associated with carbapenem resistance in A. baumannii, PCR revealed that bla_OXA-23 was present in 17 isolates of carbapenem-resistant A. baumannii belonging to PFGE types A and B. Four isolates (three belonging to PFGE type C and the fourth to type D) contained bla_OXA-24/40 (Figure 1b). β-Lactamase genes universally carried by A. baumannii, bla_ADC and bla_OXA-51/69, were present. PCR did not detect bla_OXA58, bla_KPC, bla_VIM and bla_IMP among these isolates.

All K. pneumoniae isolates were confirmed to harbour bla_KPC by PCR. Sequencing revealed that bla_KPC-2 was present in 16/17 isolates that belonged to PFGE type A, whereas bla_KPC-3 was present in 9/10 isolates from type B. Additionally, one strain from type B (Kp22) possessed the novel KPC-7 (deposited in GenBank under accession number EU729727).

Clinical characteristics and outcomes

Table 2 summarizes the clinical characteristics of 37 patients in whom carbapenem-resistant A. baumannii and 28 patients in whom carbapenem-resistant K. pneumoniae were isolated (three charts were not available). Our cohort consists of elderly patients (mean age is 64.6 and 63 years, respectively), with significant co-morbidities (as indicated by a Charlson index of 6). Patients were frequently admitted from PACFs (51.3% and 75%, respectively), had been frequently hospitalized within the previous year (75%), and exposed to carbapenems and other broad-spectrum antimicrobials.

Table 2.

Characteristics of patients with carbapenem-resistant A. baumannii and K. pneumoniae at UH

Characteristics	A. baumannii	K. pneumoniae
No. of cases reviewed	37a	28a
Colonized, n (%)	20 (54)	15 (53.6)
Infected, n (%)	17 (46)	13 (46.4)
Female sex, n (%)	19 (51.3)	20 (71.4)
Age, mean (range)	64.6 (12–93)	63 (19–90)
Charlson co-morbidity index, age adjusted mean (range)	6 (0–12)	6.5 (0–10)
In ICU at time of isolation, n (%)	23 (62.2)	13 (46.4)
Admission from PACF, n (%)	19 (51.3)	21 (75)
Days of hospitalization before isolation, mean (range)	14.5 (0–84)	6.4 (0–40)
Source of isolate, n (%)
urine	5 (13.5)	22 (78.6)
blood	6 (16.2)	7 (25)
sputum	23 (62.2)	3 (10.7)
wound	7 (18.9)	2 (7.1)
Predisposing factors, n (%)b
bladder catheter	15 (40.5)	16 (57.1)
intravascular catheter	15 (40.5)	13 (46.4)
intubation/tracheostomy	22 (59.5)	7 (25)
percutaneous endoscopic gastrostomy	17 (45.9)	7 (25)
treatment with β-lactam/β-lactamase combination	16 (43.2)	11 (39)
treatment with fluoroquinolones	11 (29.7)	10 (35.7)
treatment with carbapenem	12 (32.4)	11 (39)
Previous hospitalization (<1 year), n (%)	31 (83.8)	27 (96.4)
Isolation of other MDR bacteria, n (%)	14 (37.8)	7 (25)
Discharged to post-acute care facility, n (%)	18 (48.6)	17 (60.7)
Duration of hospital stay, mean days (range)	24.6 (1–128)	16 (5–56)
Overall in-hospital mortality, n (%)	10 (27)	6 (21.4)
Mortality attributable to infection, n (%)	6/17 (35.3)	4/13 (30)

^aThree charts could not be located.

^bSeveral patients had more than one predisposing factor.

The most common source of carbapenem-resistant A. baumannii was the respiratory tract, and many patients had a tracheostomy or endotracheal tube. In contrast, the most frequent source of carbapenem-resistant K. pneumoniae was the urinary tract (78.6%), and a bladder catheter was the most common predisposing condition. The average length of hospital stay for patients with A. baumannii was 24.6 days, and 16 days for patients with K. pneumoniae. Six deaths were attributable to A. baumannii infection and four to K. pneumoniae infection, corresponding to 35.3% and 30% crude mortality rates, respectively. Of those who survived, 48.6% and 60.7% were discharged to PACFs.

Tables 3 and 4 provide the details and outcomes of patients infected with carbapenem-resistant A. baumannii (n = 17) and K. pneumoniae (n = 13). Pneumonia was the most common syndrome caused by A. baumannii, while urinary tract and bloodstream infections were most common among patients infected with K. pneumoniae. Three patients with A. baumannii bloodstream infection died receiving inappropriate empirical antimicrobial therapy; three deaths occurred despite treatment with colistin, combined with rifampicin and meropenem or tigecycline. Two patients with bloodstream infection caused by carbapenem-resistant K. pneumoniae resistant to colistin died while receiving colistin in combination with rifampicin or amikacin; two additional deaths occurred in patients infected with colistin-susceptible isolates, despite treatment with this antimicrobial agent. Of note, persistent colonization of the urinary and respiratory tracts with carbapenem-resistant organisms occurred frequently. The characteristics and outcomes of patients did not vary according to the A. baumannii and K. pneumoniae genotype isolated.

Table 3.

Clinical characteristics and outcomes of patients infected with carbapenem-resistant A. baumannii at UH

Isolate no.	Age	Sex	Reason for admission	Underlying condition	Charlson index	Type of infection	Discharge/outcome day 28	Comment
Ab01	73	F	ischaemic bowel	rheumatoid arthritis	5	UTI	deceased	died from underlying disease
Ab02	68	F	pancreatitis	DM2	4	VAP	deceased	died from underlying disease
Ab06	73	M	fever and delirium	OSA	4	HAP	PACF/alive
Ab07	42	M	upper GI bleeding	cirrhosis	3	BSI (line)	deceased	died from underlying disease + infection
Ab08	49	M	CAP/ARDS	none	0	VAP/SSI	PACF/alive	persistent airway colonization
Ab13	77	F	abdominal pain	COPD/myasthenia gravis/respiratory failure	6	VAP/BSI	PACF/alive
Ab22	50	M	respiratory failure	DM2/dementia	3	HAP	PACF/unknown
Ab33	60	M	respiratory failure	cholecystectomy/ myasthenia gravis	7	IAI	deceased	died from underlying disease + infection
Ab39	65	F	respiratory failure	lung transplant	3	VAP	deceased	died from underlying disease + infection
Ab41	76	F	abdominal cellulitis	COPD/stroke	5	SSI	PACF/alive	persistent airway colonization
Ab52	50	M	Fever	quadriplegia	4	BSI	PACF/alive
Ab69	66	M	respiratory failure	COPD/heart failure	8	VAP	PACF/alive
Ab79	66	F	hypoglycaemia	DM2/heart failure/CKD	11	BSI	deceased	died from underlying disease + infection
Ab81	82	F	respiratory failure	DM2/stroke/COPD	9	VAP	PACF/alive
Ab84	50	M	sepsis/SSI	cirrhosis	4	SSI: Fournier's gangrene/BSI	deceased	died from underlying disease + infection
Ab88	71	F	sepsis	CKD/breast cancer	8	VAP/BSI	deceased	died from underlying disease + infection
Ab90	73	M	respiratory failure	DM2/CAD	8	BSI (line)	PACF/alive

GI, gastrointestinal; CAP/ARDS, community-acquired pneumonia/acute respiratory distress syndrome; SSI, skin and soft tissue infection; DM2, diabetes mellitus type 2; OSA, obstructive sleep apnoea; COPD, chronic obstructive pulmonary disease; CKD, chronic kidney disease; CAD, coronary artery disease; UTI, urinary tract infection; VAP, ventilator-associated pneumonia; HAP, hospital-acquired pneumonia; BSI, bloodstream infection; IAI, intra-abdominal infection.

Table 4.

Clinical characteristics and outcomes of patients infected with carbapenem-resistant K. pneumoniae at UH

Isolate no.	Age	Sex	Reason for admission	Underlying condition	Charlson index	Type of infection	Discharge/outcome day 28	Comment
Kp24	68	M	PEA	convalescing from CAP/ARDS	3	HAP	PACF (day 26)	persistent sputum colonization, colistin resistance
Kp42	76	F	fever and delirium	DM2, heart failure and lung cancer	10	UTI	PACF/unknown
Kp54	54	F	hepatic encephalopathy	DM2, cirrhosis, CKD	8	UTI	PACF/alive	persistent urinary colonization, renal failure on amikacin
Kp27	78	F	hyperkalaemia	DM2, heart failure, CKD	9	UTI/HAP	PACF/alive	persistent urinary colonization, colistin resistance
Kp33	82	F	haematemesis	CAD, heart failure	7	UTI	PACF/alive	airway colonization with A. baumannii
Kp05	77	F	cardiogenic pulmonary oedema	CAD, aortic stenosis	4	BSI 2ry to UTI vs line	PACF/alive
Kp23	76	M	respiratory failure	heart failure and stroke	6	BSI 2ry to line	PACF/alive	persistent urinary colonization
Kp57	41	M	vomit and abdominal fistula	Crohn's disease	0	BSI 2ry to line	PACF/alive	recurred with urinary colonization and BSI
Kp47	54	M	renal and liver transplants	cirrhosis, CKD	6	BSI 2ry to line	home/alive	previously colonized
Kp08	50	M	fever and respiratory failure	renal transplant	5	acute/chronic pyelonephritis	deceased	died from underlying disease + infection
Kp40	54	M	pericardial effusion	DM1, heart failure, ESRD	8	BSI 2ry to line	deceased	died from underlying disease + infection
Kp28	64	F	delirium and respiratory failure	COPD, lupus, ESRD, heart failure	10	BSI 2ry to line	deceased	died from underlying disease + infection; colistin resistance (A. baumannii HAP)
Kp45	39	F	respiratory and renal failure and seizures	COPD, spina bifida and VP shunt	2	BSI 2ry to line	deceased	died from underlying disease + infection; colistin resistance (P. aeruginosa/A. baumannii co-infection)

CAP/ARDS, community-acquired pneumonia/acute respiratory distress syndrome; DM1, diabetes mellitus type 1; DM2, diabetes mellitus type 2; COPD, chronic obstructive pulmonary disease; CKD, chronic kidney disease; PEA, pulseless electrical activity; CAD, coronary artery disease; ESRD, end-stage renal disease; VP, ventriculoperitoneal; UTI, urinary tract infection; HAP, hospital-acquired pneumonia; BSI, bloodstream infection; 2ry, secondary.

Discussion

This report describes the dissemination of genetically related strains of carbapenem-resistant A. baumannii and K. pneumoniae in a healthcare system in Northeast Ohio. These findings highlight the evolving role of PACFs (LTCFs and LTACHs) in spreading carbapenem-resistant organisms and illustrate several important challenges that this poses to contemporary healthcare systems.

Our study draws attention to the role of PACFs in the spread of carbapenem-resistant organisms. Our analysis revealed that >50% of the carbapenem-resistant isolates were obtained from patients admitted from PACFs. Many of these patients were of advanced age, were receiving intensive care at the time of isolation of the carbapenem-resistant organisms, and ∼80% were hospitalized during the previous year. Moreover, a large proportion of patients with carbapenem-resistant organisms were discharged to PACFs. Likely, these trends will continue to rise as the population ages, medical interventions continue to increase in complexity and PACFs become more important components of the healthcare system.³³ LTCFs have long been recognized as ‘reservoirs’ of MDR organisms.^34–36 LTACHs, where antibiotic use is considerable and patients often come from ICUs requiring mechanical ventilation and invasive devices, add a new and heightened dimension to this problem.⁸ Therefore, the dissemination of carbapenem-resistant pathogens in these settings is worrisome.^13,19–22

These trends indicate that coordination and centralization of infection control efforts between acute-care hospitals and PACFs is required. Comprehensive and multifaceted infection control programmes, or ‘bundles’ consisting of improved detection and isolation, personnel education, hand hygiene promotion and environmental cleaning, effectively control carbapenem-resistant A. baumannii and K. pneumoniae in the acute care setting.^37–40 Implementing such measures in PACFs is even more challenging and necessary.⁴¹

Next, we describe the clinical impact of carbapenem-resistant organisms across a healthcare system. We found that patients with carbapenem-resistant A. baumannii and K. pneumoniae have prolonged hospital stays and high crude and attributable mortality rates. These findings, although limited by the descriptive nature of this study, are consistent with those of other studies.^5,42,43 Of note, several of the deaths attributed to infection due to carbapenem-resistant A. baumannii and K. pneumoniae in this report occurred in patients who did not receive adequate antimicrobial therapy, either because they died before carbapenem resistance was identified or because of resistance to colistin. The extent to which delaying adequate antimicrobial therapy in the face of MDR organisms negatively impacts survival is unclear,^5,44 and discerning the clinical impact of carbapenem-resistant A. baumannii and K. pneumoniae in a population with severe underlying illness remains problematic.^45,46

This hospital system-wide analysis provided the opportunity to evaluate different typing methods in the analysis of outbreaks caused by carbapenem-resistant organisms. In this instance, typing of A. baumannii by three different methods (i.e. rep-PCR, PFGE and PCR/ESI-MS) suggested an important role for cross-transmission in this multi-institutional outbreak. Based on our results, isolates that appear unrelated by rep-PCR are likely to be unrelated by PFGE and PCR/ESI-MS. In contrast, strains that seem identical by rep-PCR may be different by PFGE or represent unique STs by PCR/ESI-MS. Therefore, an initial approach based on rep-PCR to establish clonality among A. baumannii during an outbreak investigation is useful, but may necessitate confirmation with additional typing methods such as PFGE or PCR/ESI-MS. On the other hand, rep-PCR typing of carbapenem-resistant K. pneumoniae matched the results obtained with PFGE. This supports earlier findings that rep-PCR may be a valid tool in the evaluation of carbapenem-resistant K. pneumoniae outbreaks.¹⁹ The role of PCR/ESI-MS in the investigation of the molecular epidemiology of K. pneumoniae is still evolving.

Lastly, we analysed the β-lactamase genes that determined carbapenem resistance in isolates from this outbreak. The predominant strain types of carbapenem-resistant A. baumannii harboured bla_OXA-23, whereas only a few isolates harboured bla_OXA-24/40. This corresponds with other genetic descriptions of carbapenem-resistant A. baumannii outbreaks in the USA^13,14,47 and contrasts with other locations where metallo-β-lactamases are prevalent.⁴⁸ Of note, carbapenemases were not detected in 15 isolates displaying carbapenem resistance, except that they possessed the intrinsic bla_OXA-51/69. We did not investigate the presence of ISAba1 in the promoter region of the intrinsic bla_OXA-51/69, which in some studies has been associated with carbapenem resistance.⁴⁹ However, other studies do not support this correlation and suggest the involvement of mechanisms different from carbapenemases.^32,48,50

Our observations of this outbreak do not permit us to obtain insights about the relationship between genetic characteristics and clinical impact of carbapenem-resistant organisms. In contrast, a particular strain of MDR A. baumannii (defined by PFGE and MLST) from a hospital in the Southwestern USA was associated with increased rates of bacteraemia and mortality compared with other strains of A. baumannii within the same outbreak.⁵¹ In another study, patients from a military treatment facility affected with carbapenem-resistant A. baumannii experienced longer hospital stays and more frequently required mechanical ventilation if their strains harboured bla_OXA-23.⁵² In our report, there were two predominant strains of carbapenem-resistant K. pneumoniae that were differentiated by the type of carbapenemase they contained (bla_KPC-2 or bla_KPC-3) and their susceptibilities to aminoglycosides. However, clinical or epidemiological correlates to these differences were not evident. Moreover, similar findings have not been reported among KPC-producing K. pneumoniae circulating elsewhere in the USA.^17,19

In conclusion, we provide a snapshot of the clinical impact, molecular epidemiology and mechanisms of resistance of carbapenem-resistant A. baumannii and K. pneumoniae in a healthcare system in Northeast Ohio. Significant mortality occurred among elderly and debilitated patients. Carbapenem resistance among A. baumannii and K. pneumoniae was mediated predominantly by the production of OXA-23 and KPC carbapenemases, respectively. Our evidence suggests that horizontal transmission of carbapenem-resistant organisms occurred throughout the hospital system and PACFs played a central role in the dissemination of these isolates. Our findings emphasize that LTCFs and LTACHs represent an important setting in which to study carbapenem-resistant organisms. Compartmentalizing infection control efforts in the acute care and long-term care environments may be inadequate; it is important to realize that patients, pathogens and antibiotic selective pressures move across boundaries. These considerations should come to the forefront when approaching the issue of antibiotic resistance and infection control in the context of healthcare systems.

Funding

This work was supported by the Wyeth Fellowship in Antimicrobial Resistance (F. P.) and grants from the Steris Corporation (F. P. and R. A. B.), the Veterans Affairs Merit Review Program (R. A. B.), the National Institute of Allergy and Infectious Diseases at the National Institutes of Health (grant RO1-AI063517 to R. A. B.) and the Geriatric Research, Education and Clinical Center VISN 10 (R. A. B.).

Transparency declarations

R. A. B., R. A. S., M. R. J. and A. J. R. receive research and speaking invites from various pharmaceutical companies. None of these poses a conflict of interest with the present work. D. J. E. owns stock and is an employee of Abbott Molecular. Other authors: none to declare.