Epidemiology of Carbapenem-Resistant Klebsiella pneumoniae in a Network of Long-Term Acute Care Hospitals

Abstract

Background.

The rapid emergence of carbapenem-resistant Klebsiella pneumoniae (CRKP) represents a major public health threat, including in the long-term acute care hospital (LTACH) setting. Our objective in this study was to describe the epidemiologic characteristics of CRKP in a network of US LTACHs.

Methods.

An observational study was performed among 64 LTACHs from January 2014 to March 2015. Clinical cultures were included, with the first CRKP isolate recovered from each patient per study quarter evaluated. LTACH and geographic area–based CRKP prevalence and clinical and microbiologic characteristics were described.

Results.

A total of 3846 K. pneumoniae cultures were identified, with an overall carbapenem resistance rate of 24.6%. There were significant differences in CRKP rates across geographic regions, with the highest in the West (42.2%). Of 946 CRKP isolates, 507 (53.6%) were from a respiratory source, 350 (37.0%) from a urinary source, and 9 (9.4%) from blood. Among 821 unique patients with CRKP colonization or infection, the median age was 73 years. There was a high prevalence of respiratory failure (39.8%) and the presence of a central venous catheter (50.9%) or tracheostomy (64.8%). Resistance rates of CRKP isolates were high for amikacin (59.2%) and fluoroquinolones (>97%). The resistance rate to colistin/polymyxin B was 16.1%.

Conclusions.

Nearly 25% of K. pneumoniae clinical isolates in a US network of LTACHs were CRKP. Expansion of national surveillance efforts and improved communication among LTACHs and acute care hospitals will be critical for reducing the continued emergence of CRKP across the healthcare continuum.

Carbapenem-resistant Enterobacteriaceae (CRE) are an urgent public health threat [1]. Infections due to CRE are associated with high mortality rates and limited antibiotic treatment options [2, 3]. In recent years, the prevalence of CRE, most commonly carbapenem-resistant Klebsiella pneumoniae (CRKP), has dramatically increased both in the United States and worldwide [4–7]. In the United States, CRKP have been identified in nearly every state [8], with much of this rapid and widespread dissemination attributed, in part, to emergence and clonal expansion of a dominant K. pneumoniae carbapenemase (KPC)–producing strain ST258 [9].

The epidemiology of CRKP in acute care hospitals has been well described in the literature, including in a recent large, population-based surveillance study of CRE among patients with healthcare exposures in 7 US states [10]. Notably, the majority of identified isolates were CRKP, and more than half of patients were ultimately discharged to long-term acute care hospitals (LTACHs). Also, recent studies have highlighted the increasing importance of LTACHs as reservoirs of CRKP, with up to 9-fold higher prevalence rates of colonization compared to those in acute care hospitals [11–14].

LTACHs care for a chronically, critically ill patient population that is characterized by prolonged durations of stay, significant comorbidities, and high rates of intensive care needs such as ventilator weaning, indwelling device use, advanced wound care, and intravenous antibiotic administration [15]. In addition, the prevalence of multidrug-resistant organisms, including CRKP, is amplified in LTACHs due to the convergence of complex, high-risk transfers from multiple acute care hospitals to a single facility.

Understanding the evolving epidemiology of CRKP in LTACHs in the United States, particularly in the nonoutbreak setting, is critical to informing national and regional infection prevention and surveillance efforts. In addition, quantification of the burden of CRKP in the LTACH setting will help elucidate transmission dynamics between acute care hospitals and LTACHs and also inform strategies for reducing the emergence of these multidrug-resistant organisms across the continuum of care. However, previous studies on the epidemiology of CRE in LTACHs have been limited to single LTACHs or geographic areas [11–14] and evaluation in an outbreak setting [12], and none were designed to perform surveillance on a larger, national basis [11–14].

Therefore, we performed a surveillance study to describe the prevalence and epidemiologic characteristics of colonization and infection due to CRKP in a large, geographically widespread network of LTACHs in the United States.

METHODS

Study Design

An epidemiologic, LTACH-based surveillance study of CRKP was performed within the Kindred Healthcare network of LTACHs from 1 January 2014 to 31 March 2015. Kindred Healthcare is the largest provider of post-acute care in the United States, including LTACHs, skilled nursing facilities, rehabilitation facilities, and home care services. A total of 64 geographically widespread LTACHs contributed data to this study. Geographic locations of the participating LTACHs are shown in Table 1. The study was reviewed and approved by the University of Pennsylvania Institutional Review Board.

Study Population

All LTACH patients with a clinical culture (blood, respiratory, or urine) with K. pneumoniae were included. The first clinical culture with K. pneumoniae was included for each unique patient per calendar quarter; therefore, each patient could contribute multiple K. pneumoniae cultures to this study but only 1 unique culture episode per quarter. This method was selected in order to more uniformly assess rates of carbapenem resistance among K. pneumoniae isolates, as the number of cultures obtained per patient could vary widely depending on provider ordering.

A CRKP case was defined using the updated 2015 Centers for Disease Control and Prevention (CDC) definition [16]. Because confirmatory carbapenemase testing was not performed during the study period, an isolate was designated as CRKP if it demonstrated at least 1 of the following: a minimum inhibitory concentration (MIC) of ≥4 µg/mL to meropenem and/or imipenem or an MIC to ertapenem of ≥2 µg/mL, in accordance with current Clinical and Laboratory Standard Institute guidelines [17]. Notably, all Kindred LTACHs have a standardized infection control policy for strict contact precautions for patients colonized or infected with carbapenem-resistant Enterobacteriaceae, including CRKP.

Data Collection

Data were extracted using centralized queries from ProTouch, Kindred Healthcare’s universal clinical and medical record system used for all participating LTACHs. Baseline information on demographics (eg, age, gender) and comorbidities (eg, chronic kidney disease, respiratory failure) for all patients with CRKP cultures were obtained. Data on indwelling devices were also collected, including the presence of a central venous catheter, urinary catheter, tracheostomy, and gastrostomy tube at the time of the clinical culture.

Results of antibiotic susceptibility testing for all CRKP isolates were also obtained, including for colistin or polymyxin B and tigecycline, if available. CRKP isolates at the study LTACHs were tested for susceptibility to tigecycline and colistin or polymyxin B using Etest (bioMérieux, Durham, North Carolina), with isolates with an MIC of ≤2 µg/mL determined to be susceptible, in accordance with European Committee on Antimicrobial Susceptibility Testing criteria [18].

Definitions for infection for urine and respiratory cultures were adapted from CDC National Healthcare Safety Network (NHSN) surveillance definitions during the study period [4], using available microbiologic, laboratory, and radiographic data. A urine culture was determined to be representative of infection if 1 of the following criteria were met: a positive urine culture of ≥10⁵ colony-forming units (CFU)/mL with no more than 2 species of organism and either a positive urinalysis or peripheral leukocytosis (≥10000 white blood cells [WBC]/μL) within 7 days of specimen collection or a positive urine culture of ≥10³ and <10⁵ CFU/mL with no more than 2 species of organism and a positive urinalysis. A positive urinalysis was demonstrated by either the presence of ≥10 WBC/mm³ or a positive leukocyte esterase and/or nitrite. A respiratory culture was determined to be representative of infection if a chest X-ray was performed within 7 days of the culture and included an interpretation of infiltrate, consolidation, or pneumonia, and if a peripheral leukocytosis (≥10000 WBC/μL) was present within 7 days of specimen collection.

Statistical Analyses

CRKP resistance rates were determined for the entire study population, as well as separately for each LTACH, state, and US census region. Changes in CRKP prevalence across study quarter were also assessed. Descriptive analyses were performed to summarize isolate and patient information, including demographics, comorbidities, presence of indwelling devices, and antibiotic resistance rates. For all calculations, a 2-tailed P value of <0.05 was considered significant. All analyses were performed using Stata 14.0 (StataCorp, College Station, Texas).

RESULTS

Carbapenem Resistance Rates

During the 15-month study period, 3846 unique quarterly K. pneumoniae clinical cultures were identified (Table 1). These cultures were identified in 3470 patients across the 64 study LTACHs. Among the 3846 K. pneumoniae isolates, 946 were carbapenem resistant, for an overall carbapenem resistance rate of 24.6%. CRKP rates varied significantly across quarter as follows: quarter 1, 2014: 199 (27.3%); quarter 2, 2014: 191 (24.8%); quarter 3, 2014: 249 (30.9%); quarter 4, 2014: 148 (19.4%); and quarter 1, 2015: 159 (20.4%; P < .001, test for trend).

The greatest numbers of K. pneumoniae isolates were from California (n = 1540; 40.0%), followed by Florida (n = 686; 17.8%) and Texas (n = 498; 13.0%). These were the 3 states with the most LTACHs represented, with 13 facilities in California, 14 in Texas, and 9 in Florida (Table 1). Notably, CRKP rates varied significantly across the 16 states (Figure 1). California had the highest rate of CRKP, with 701 of 1540 K. pneumoniae isolates (45.5%) demonstrating carbapenem resistance. High rates of carbapenem resistance among K. pneumoniae isolates were also present in South Carolina (35.7%), Kentucky (22.1%), and Indiana (20.0%). Of the 16 states, only Colorado had no CRKP isolates, with all 49 isolates demonstrating susceptibility to carbapenems.

Similarly, there were significant differences in CRKP rates across US census regions as follows: Northeast (n = 21; 9.9%), Midwest (n = 30; 7.3%), South (n = 189; 12.2%), and West (n = 706; 42.2%). For the West census region, which had the highest proportion of CRKP, rates varied significantly across the study quarters: quarter 1, 2014: 158 (48.5%); quarter 2, 2014: 135 (40.4%); quarter 3, 2014: 181 (49.5%); quarter 4: 106 (32.5%); and quarter 1, 2015: 126 (39.0%; P < .001, test for trend).

Source Information

The majority (n = 507; 53.6%) of the 946 CRKP isolates were from a respiratory source. Of these, 299 (58.9%) were determined to be representative of infection using the previously described laboratory and radiographic criteria. A urinary source was the second most common (n = 350; 37.0%). Of these, 173 (49.4%) were determined to be representative of infection. Finally, there were 89 (9.4%) blood cultures during the study period. Notably, more than half of these episodes (50.6%) occurred in patients with a central venous catheter.

Study Population

There were 946 unique quarterly CRKP isolates during the study period, representing 821 patients. Demographic and clinical characteristics of patients with CRKP clinical cultures are shown in Table 2. The median age of patients was 73 years (interquartile range [IQR], 63–82) and 416 (50.7%) were female. The median length of stay prior to CRKP culture positivity was 21 days (IQR, 7–40). Comorbidities were common in patients with CRKP, including respiratory failure (39.8%), advanced stage decubitus ulcers (27.5%), and severe or end stage kidney disease (22.0%). A diagnosis of acute kidney injury prior to CRKP isolation was also common (n = 325; 42.0%). In addition, the majority of patients (n = 532; 64.8%) had a tracheostomy present. Rates of central venous catheter and urinary catheter use were also greater than 50%.

Antibiotic Susceptibility Profiles of CRKP Isolates

Antibiotic susceptibility results for the CRKP isolates are shown in Table 3. Nearly all of the isolates (> 97%) demonstrated resistance to levofloxacin, ciprofloxacin, gentamicin, and tobramycin. Overall, resistance rates to amikacin were high (59.2%). Among tested isolates, tigecycline resistance rates were low (0.7%). Of 690 isolates that underwent colistin or polymyxin B susceptibility testing, 111 (16.1%) demonstrated resistance.

DISCUSSION

During the 15-month surveillance period, nearly 25% of K. pneumoniae isolates from 64 geographically widespread US LTACHs demonstrated resistance to carbapenems. Carbapenem resistance rates varied significantly by both geographic region and state, with the highest prevalence observed in the West census region and California, respectively. The LTACH patients colonized or infected with CRKP were of older age and had a high rate of comorbidities and indwelling device use. Resistance rates to broad-spectrum gram-negative agents were high in CRKP isolates, including a particularly high rate of resistance to colistin/polymyxin B of approximately 16%.

The results of our study are notable for an overall high rate of carbapenem resistance in LTACH patients colonized or infected with K. pneumoniae. The resistance rate of 25% is of major concern and is substantially higher than the rate of 10.4% previously reported in a large NHSN surveillance study of mostly intensive care unit–associated infections from acute care hospitals [19]. Notably, there was significant geographic variation, with the highest rate of CRKP occurring in LTACHs in the West census region, with the majority located in California. This geographic variation is likely reflective of the overall emergence and dissemination of CRE, particularly K. pneumoniae, across regions of the United States. Interestingly, the rate of CRKP was considerably higher in the South at 12% overall, a geographic area that has not previously been reported to exhibit high rates of carbapenem resistance. However, given the lack of systematic surveillance requirements for CRE in US healthcare facilities at present, it is likely that there is significant underreporting of CRKP prevalence from acute care hospitals in certain regions.

Along these lines, it is clear that the epidemiology of CRKP in the United States is rapidly evolving, and enhanced surveillance and reporting are needed across the healthcare continuum [20]. However, most efforts to date have focused on acute care hospitals, despite the increasing importance of LTACHs in the regional epidemiology of these multidrug-resistant organisms. Data on regional prevalence rates of carbapenem resistance are critical for development of coordinated approaches to CRKP control, including improved interfacility communication among LTACHs and acute care hospitals, as well as engagement of local and regional health departments. Our study is the largest to date to assess the epidemiology of CRKP in LTACHs and is further strengthened by surveillance in a nonoutbreak setting and in a network of LTACHs from the largest provider of post-acute care in the United States. As such, these results serve as a comprehensive baseline of the burden of CRKP in US LTACHs.

Notably, the majority of CRKP isolates were detected more than 72 hours after LTACH admission, with a median of 3 weeks of LTACH stay prior to isolation. However, given the lack of active surveillance admission cultures (ie, perirectal or rectal) performed for detection of gastrointestinal colonization or data on cultures from transferring acute care hospitals, it is unclear if these cultures represented acquisition in the LTACH vs colonization or infection present on admission. Interestingly, in a recent study that assessed CRKP in a university-affiliated LTACH that had a policy of screening blood, urine, and respiratory cultures on admission for all patients, approximately 40% of CRKP isolates were detected within 72 hours of arrival to the LTACH [14]. Thus, active surveillance screening with prompt implementation of contact precautions may be a particularly important infection control strategy in LTACHs and, as suggested by the CDC [16], should be considered for facilities with high local or regional prevalence of CRE. In addition, while both colonization pressure (ie, from colonized or infected patients transferred into LTACHs) as well as patient acquisition within the LTACH are likely important in the amplification of CRE in the LTACH setting, further studies are needed to elucidate the relative contributions of these modalities to better inform infection prevention strategies.

LTACH patients who were colonized or infected with CRKP were generally of older age and demonstrated high rates of certain comorbidities including renal disease, advanced stage wounds, and respiratory failure. In addition, rates of indwelling device use, including central venous catheters and tracheostomies, were high in the study population. Ventilation-dependent respiratory failure is the leading diagnosis on admission to LTACHs [15], and a significant proportion of LTACH patients require intensive care unit level of care. These results underscore the distinct epidemiology of LTACH patients as a critically ill, at-risk population for colonization and infection with CRKP and the importance of interventions targeted toward this complex healthcare setting. In particular, while significant attention and resources have been directed toward evaluation of optimal practices for infection prevention in the acute care hospital setting, further research is urgently needed to define these practices in LTACHs. For example, a recent study demonstrated that a bundled intervention, which included the use of daily chlorhexidine bathing and regular active surveillance, reduced CRE colonization and infection rates in 4 LTACHs, even in the setting of significant colonization pressure [21]. Further studies to evaluate these interventions, including in the context of available resources and personnel in the LTACH setting, are needed.

We also demonstrated high rates of resistance of CRKP isolates to commonly used broad-spectrum gram-negative antibiotic classes on susceptibility testing. The overall resistance rate to amikacin was considerably higher among tested isolates, and nearly one and a half fold greater than that reported previously in a US surveillance study of isolates obtained from patients with healthcare exposure, most commonly acute care hospitalization [10]. In addition, nearly all isolates demonstrated resistance to the fluoroquinolones ciprofloxacin and levofloxacin. The finding of a resistance rate of nearly 700 tested isolates to colistin or polymyxin B of 16% is of significant concern and considerably higher than that reported in a large-scale surveillance study of acute care hospitals [22]. Similarly, while a 15% resistance rate was reported in the previously mentioned US surveillance study performed in 7 states [10], only 12 isolates were tested. The polymyxins are typically the last-line agents used for treatment of infections due to CRKP and are often critical for combination therapy regimens. These results emphasize the importance of healthcare setting–specific antibiotic susceptibility rates for antibiotic stewardship efforts and guidance of empiric treatment of patients at high risk for infections due to CRKP. In addition, increased resources and funding should be directed toward enhanced monitoring of colistin/polymyxin B resistance rates on a national basis.

There are several potential limitations in this study. First, given that rectal cultures for detection of asymptomatic gastrointestinal colonization were not routinely performed, our results likely underestimated the true burden of CRKP colonization. Second, confirmatory carbapenemase testing was not routinely performed on isolates, so the relative proportion of carbapenemase-producing vs noncarbapenemase-producing isolates is unknown. However, given that all study LTACHs tested susceptibility to meropenem and/or imipenem (vs ertapenem), the proportion of the former was likely high. Third, given the lack of available medical record data for ascertainment of signs and symptoms, our definitions for infection likely overestimated the proportion of urinary and respiratory isolates indicative of infection, particularly in cases of asymptomatic bacteriuria. Fourth, while data from a large number of LTACHs were captured in this study, the majority of isolates were from 3 states, and certain high-prevalence regions (eg, the New York metropolitan area) were not included.

In conclusion, there was a prevalence of carbapenem resistance of nearly 25% among K. pneumoniae isolates from this large, LTACH-based surveillance study in the United States.

Given the chronically, critically ill population, with convergence of at-risk patients from multiple facilities, future studies of optimal infection prevention strategies are urgently needed for this setting. In addition, expansion of national surveillance efforts and improved communication among LTACHs and acute care hospitals will be critical for the reduction of the continued emergence and dissemination of CRKP across the healthcare continuum.

Notes

Acknowledgments. We gratefully acknowledge Kindred Healthcare for their assistance in collecting data used in this study.

Disclaimer. The funding agencies had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; or preparation, review, or approval of the manuscript.

Financial support. This work was supported by the National Institutes of Health (K01-AI103028 to J. H. H.) and, in part, by the Centers for Disease Control and Prevention Epicenters Program (U54-CK000163 to E. L.).

Potential conflicts of interest. E. J. C. G. has served on advisory boards and a speakers’ bureau for Merck and has received grant support from Cerexa, the Medicines Company, and Astella All other authors: No reported conflicts. All authors have submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest. Conflicts that the editors consider relevant to the content of the manuscript have been disclosed.

References

Author notes