Carbapenem-resistant Enterobacteriaceae (CRE) and, in particular, carbapenemase-producing (CP-) CRE, are a growing concern due to their multidrug-resistance and propensity to spread within healthcare facilities.1 The Centers for Disease Control and Prevention has designated CRE an ‘urgent threat’2 due to an estimated incidence rate of 2.93 per 100,000 population, limited treatment options,2 and a 40%–50% mortality rate.1 Aggressive containment strategies could prevent 1,600 cases of CRE in 1 state over a 3-year period.2 However, little research has been conducted looking at facility-level risk factors. The Department of Veterans’ Affairs (VA) medical centers (VAMC) provide an opportunity to evaluate CRE facility-level risk factors in the largest integrated healthcare system in the United States.
Methods
This ecological study included 136 VAMCs from January 1, 2017, to December 31, 2017, with antibiotic data from January 1, 2016, through December 31, 2016. Facilities were limited to VAMCs with at least 2 of the following care settings: inpatient, outpatient, residential, and/or institutional extended care.
The VA Corporate Data Warehouse (CDW) was used to identify CRE/CP-CRE cases and to collect microbiology, antibiotic, and facility characteristic data. The CDW is a continuously updated relational database of VA clinical and administrative data.3 CRE cases included Escherichia coli, Klebsiella pneumoniae/oxytoca, and Enterobacter spp resistant to imipenem, meropenem, and/or doripenem. CP-CRE was defined as a positive diagnostic test for carbapenemase production (eg, PCR). Specialty care units were identified if corresponding clinical programs were present at a VAMC. VAMCs were classified by complexity: levels 1a–c were high complexity and levels 2–3 were low complexity. Complexity was based on patient characteristics, clinical programs, and teaching programs. Geographic region was based on US Census region, with Puerto Rico grouped in the Southern region. A facility was affiliated with an academic medical center if it had a training program for health professionals. Antibiotic rates were calculated by total days a patient was prescribed an antibiotic class divided by unique patients. Facility average length of stay was defined as total length of stay divided by cumulative admissions. Intensive care unit (ICU) admission was defined as cumulative ICU admissions for a facility. International Classification of Disease, Tenth Revision (ICD-10) procedure codes were used to identify inpatient surgeries and outpatient procedures (Current Procedural Terminology (CPT)/Healthcare Common Procedure Coding System (HCPCS)). Facility CRE rate was calculated as the number of cases divided by the total unique patients.
Bivariate analyses were conducted using the χ2 or Fisher exact test and the Mann-Whitney U test to assess the association of facility characteristics and the presence or absence of CRE. A P value <.05 was considered significant. Analyses were conducted using STATA version 14.2 software (StataCorp, College Station, TX). ArcGIS software (Redlands, CA) was used to geographically display data.
Results
In total, 801 CRE cases were identified among 7,100,299 unique patients treated in 136 VAMCs (11.3 CRE cases per 100,000 patients), and 97 VAMCs (71%) had at least 1 CRE case. Among CRE cases, 13% were E. coli, 26% Enterobacter spp, and 61% Klebsiella pneumoniae or K. oxytoca. Overall, 324 CRE (40%) were CP-CRE. Table 1 describes facility characteristics and comparisons of facilities with and without CRE. A higher proportion of facilities with CRE were high complexity, urban, and were located in the Southern region. Most facilities were affiliated with an academic center, and had spinal cord injury, interventional radiology, radiation oncology units, and invasive cardiac catheterization labs. Most facilities performed neurological and cardiac surgeries. Moreover, increased numbers of ICU admissions, transplants, inpatient surgeries, outpatient procedures, and increased days of carbapenems, as well as higher use of fourth-generation cephalosporins, and quinolones, were associated with facilities with CRE.
Table 1. Frequency of Facility Characteristics and Comparison of Facilities With and Without CRE, 2017
Note. CRE, carbapenem-resistant Enterobacteriaceae; CP-CRE, carbapenemase-producing CRE.
a 1 facility missing.
b 21 facilities missing.
We detected significant geographic variation in CRE. The 5 states and territories with the highest numbers of CRE cases were Puerto Rico (n = 224), California (n = 78), New York (n = 76), Texas (n = 59), and Florida (n = 49). The Northeastern, Midwestern, and Western census regions had states with no cases of CRE, whereas all states in the Southern region had at least 1 CRE case (Fig. 1A). The rate of CRE across the United States ranged from 0 to 35.3 cases per 10,000 patients. The highest rates were seen in Puerto Rico, New Jersey (3.48 per 10,000 patients), New York (2.39 per 10,000 patients), Washington, DC (1.89 per 10,000 patients), and Wyoming (1.59 per 10,000 patients) (Fig. 1B).
Fig. 1. Geographical distribution of CRE and CP-CRE in the United States, 2017. (A) Number of CRE observations. (B) Rate of CRE per 10,000 patients.
Discussion
We identified CRE cases in almost three-fourths of VAMCs. Of 801 CRE cases detected, 324 (40%) were CP-CRE. The distribution of organisms identified as CRE in our study was consistent with those of previous studies, with ranges of 2%–15% of E. coli, 1%–49% of Enterobacter species, and 44%–91% of Klebsiella pneumonia/oxytoca.Reference Kelly, Mathema and Larson4
Facility characteristics associated with higher rates of CRE included urban location, high complexity, presence of specific specialty units, greater ICU admissions, and higher numbers of surgeries. In other studies, high-complexity urban facilities are also more likely to treat patients with complex health conditions compared to rural, less complex locations.Reference Ray, Lin, Weinstein and Trick5 These patients are often asymptomatic CRE carriers and can mediate transmission to other medically complex patients.Reference Park, Liu, Furuya and Larson6 Furthermore, higher complexity facilities not only have long-term-care specialty units that increase patient retention, they also have short-term therapeutics in acute care that causes high patient turnover, creating an opportunity for CRE to spread.Reference Ray, Lin, Weinstein and Trick5
We detected significant geographic variation in CRE (Fig. 1A, 1B), with states/territories with the highest CRE cases and rates being in areas with major cities. This finding may relate to greater international travel and population migration in these cities, which could contribute to increased CRE.Reference Schwartz and Morris7 In contrast to our results, outside the VA, the greatest proportion of CRE cases were observed in the western United States,Reference Gupta, Ye, Olesky, Lawrence, Murray and Yu8 which could be due to methodological differences in region categorization or different patient characteristics between VA and non-VA facilities. Recognizing variations in regional CRE epidemiology can help hospitals prevent future outbreaks. The interconnective system of healthcare in the VA increases the risk of importing CRE, resulting in greater opportunities for CRE to spread.Reference Ray, Lin, Weinstein and Trick5 Failure to control CRE could lead to wider spread due to the ability for CRE to transmit resistance to other gram-negative bacilli.Reference Kelly, Mathema and Larson4
Finally, facilities with CRE were associated with increased days of carbapenems, fourth-generation cephalosporins, and quinolones, which is similar to patient-level studies.Reference Patel, Huprikar, Factor, Jenkins and Calfee9 The increasing prevalence of MDR organisms has led to limited therapeutic options and has increased the use of broad-spectrum antibiotics. This antibiotic selective pressure could create an association between broad-spectrum antibiotic use and increased CRE.Reference Xu, Gu and Huang10 This finding stresses the importance of monitoring broad-spectrum antimicrobial use to aid the prevention of CRE.
Our study has several limitations. It was an ecological study, with the possibility of ecological fallacy; however, our results were similar to those of patient-level studies.Reference Ray, Lin, Weinstein and Trick5,Reference Patel, Huprikar, Factor, Jenkins and Calfee9 Secondly, the sample size was small and precluded the ability to conduct multivariate analyses; however, 136 VAMCs is the largest number of integrated healthcare centers in the United States. Furthermore, CRE/CP-CRE testing is not universal across VAMCs, therefore cases may have been underrepresented and may not reflect the true proportion of CRE/CP-CRE. Furthermore, because we used a laboratory-based definition for CRE, we were not able to differentiate between CRE colonization and infection, or between clinical and surveillance specimens, both of which likely influenced facility-level CRE prevalence rates and may have attributed more cases to facilities that perform active surveillance for colonization.
In conclusion, almost three-fourths of VAMCs had at least 1 CRE case, with significant variation across geographic regions. Infection control strategies targeted to facility characteristics are critical to preventing the spread of CRE. Understanding local patterns of CRE can help facilities determine where to focus control measures for better prevention of future CRE.
Acknowledgments
The views expressed in this article are those of the authors and do not necessarily reflect the position or policy of the Department of Veterans’ Affairs or the US government.
Financial support
This work was supported by the Department of Veterans’ Affairs, Veterans’ Health Administration, Office of Research and Development, Health Services Research and Development Quality Enhancement Research Initiative (QUE 15-269).
Conflicts of interest
All authors report no conflicts of interest or financial disclosures relevant to this article.
