Risk Factors for the Acquisition of a blaNDM-5 Carbapenem-Resistant Escherichia coli in a Veterinary Hospital
Carbapenem-resistant Enterobacteriaceae (CRE) are an urgent antibiotic resistant threat. Only sporadic reports of CRE in companion animals have been described. Our objective was to identify risk factors associated with the acquisition of a blaNDM-5 CR-Escherichia coli strain as part of an outbreak investigation at a tertiary veterinary hospital in the United States. A matched case–control study was conducted among companion animals admitted during July 1, 2018, through June 30, 2019. The 15 identified blaNDM-5 CR-E coli cases were matched 1:2 with controls (culture negative for blaNDM-5 CR-E coli) based on species and number of days of hospitalization before bacterial culture sample collection. The association between exposure to various procedures and hospital services and the acquisition of blaNDM-5 CR-E. coli was assessed through conditional logistic regression. Case patients had significantly higher odds of exposure to the anesthesia service (odds ratio [OR] = 12.8, P = .017), the surgical service (OR = 4.0, P = .046), and to endotracheal intubation (OR = 10.0, P = .03). Veterinary hospitals should be aware of the potential for transmission of CRE via anesthetic and surgical procedures, especially those that require the placement of endotracheal tubes.
Introduction
The emergence of carbapenem-resistant Enterobacteriaceae (CRE) in companion animal veterinary medicine was inevitable because these extensively drug-resistant organisms have been found to colonize the gastrointestinal tract of humans and animals on all major continents.1 To date, there have been only a few sporadic reports of CRE isolated from companion animals.1–6 However, the data summarized in the review by Kock et al.1 showed that CRE from a range of animal species were found to belong to diverse bacterial clonal backgrounds. Consequently, the public health significance associated with the isolation of CRE from an animal, especially in countries such as the United States where these bacteria are rare in human health care, must be acknowledged.
The carbapenem drugs are critically important antimicrobials that are generally reserved by veterinarians and physicians for the treatment of infections caused by multidrug-resistant gram-negative bacteria.7 CRE have emerged as an important cause of human healthcare-associated infections and have become a major clinical and public health problem.7 CRE are regarded by the CDC as an urgent public health threat because they are not only carbapenem resistant but also resistant to most other antibiotic classes.
In July 2018, a strain of Escherichia coli that was subsequently shown to contain a New Delhi metallo-β-lactamase, blaNDM-5 gene, was isolated from a dog being treated at a tertiary veterinary hospital in Philadelphia.8 This isolate belonged to sequence type 167 (ST167) and contained additional antimicrobial resistance genes. A case report has since documented an outbreak of blaNDM-5 -E coli (ST 167) in companion animals in the United States.9
Since April 2018, the Philadelphia Board of Health has required the reporting of infection or colonization with CRE to the Philadelphia Department of Public Health (PDPH). A case is broadly defined as a culture yielding a bacterium in the family Enterobacteriaceae that is documented to produce a carbapenemase by means of a laboratory test. In May 2019, the microbiology laboratory at the veterinary hospital contacted PDPH to report the cases when the blaNDM-5 gene mechanism was confirmed. The veterinary hospital worked with PDPH on an outbreak investigation to determine potential sources of in-hospital transmission, review infectious disease–related protocols and procedures, and reduce the potential for both animal-to-animal and animal-to-human spread.
As part of the outbreak investigation, a case–control study was conducted to identify potential sources of blaNDM-5 CR-E coli. The goal was to identify modifiable risk factors for in-hospital acquisition of blaNDM-5 CR-E coli. Effort focused on the following three main categories of exposure: exposure to services within the hospital, exposure to procedures thought to be associated with blaNDM-5 CR-E coli acquisition, and exposure to medications including antibiotics that might put patients at risk.
Materials and Methods
Setting
The data used in this study were derived from a 24 hr tertiary care veterinary teaching hospital in Philadelphia. The hospital serves ˜35,000 patients per year and includes 12 core specialties in addition to training and research programs.
Data Set
The data used in these analyses were collected in the course of clinical activities and assessed retrospectively as part of an outbreak investigation. Data were collected from an electronic hospital information system (HIS) that is primarily used for tracking procedures and billing. The data exported included dates of admission and discharge, services used, and codes for procedures and medications used with associated dates. Service information was derived from the HIS records of hospital services associated with each procedure. Procedure codes in the data set were grouped into categories based on type of procedure (e.g., endotracheal intubation, catheterization) and medications received (e.g., glucocorticoids, antibiotics). Procedures or services in the HIS occurring after the date of the procedure code for bacterial culture were excluded from analysis. Results of testing from the in-house clinical microbiology laboratory were added to this data set to assign case status. A broad list of potential exposures (services, procedures, and medications) were considered, and only those with exposure among both cases and controls were included for evaluation.
Case Definition
A case was defined based on the criteria of the PDPH. For inclusion purposes, laboratory record review identified any animal patient from whom a CRE was recovered between July 1, 2018, and June 30, 2019. Organisms in the family Enterobacteriaceae that showed phenotypic resistance to imipenem based on Clinical and Laboratory Standards Institute interpretive guidelines10 were recovered from cryopreservation. These organisms were tested for the production of a carbapenemase using the modified carbapenem inactivation method.11
Only the first positive culture from each animal was used for statistical analysis. Controls consisted of patients who had HIS documentation of an aerobic bacterial culture submitted to the clinical microbiology laboratory during the study period and did not test positive for blaNDM-5 CR-E coli. In order to control for length of exposure to hospitalization, each case was matched to two controls by both species and number of days of hospitalization before culture.
Statistical Analyses
R (3.5.1)a was used to match cases and controls and determine exposure status using procedure code data extracted from HIS. Exact matching of cases to controls based on species and days of hospitalization before culture was performed using the program Optmach (R Software Package)a. Univariate analysis of variables was performed using conditional logistic regression in Stata (14.2)b. A two-sided α level of 0.05 was used to determine statistical significance.
Outbreak Investigation
As part of the outbreak investigation, the PDPH reviewed the hospital Infection Prevention Policy and Procedure Manual. All standard operating procedures (SOP) were reviewed with emphasis on the following areas: hand hygiene, environmental cleaning and disinfection, and medical device reprocessing. Semicritical devices were defined as devices that come into contact with but do not typically penetrate mucosal surfaces. If deficiencies were found, a corrective plan was put in place.
Results
A total of 15 patients were identified as cases between July 1, 2018, and June 30, 2019. blaNDM-5 CR-E. coli was isolated from all cases. Included in these case patients were 14 dogs and 1 cat. A total of 2637 patients served as the pool of potential eligible controls for matching. Each case patient was successfully matched by species and length of stay before culture, to two controls, resulting in a total of 30 controls. Among cases, CRE isolates were obtained from a variety of sample types: endotracheal wash (7), urine (3), lung (1), esophagostomy tube (1), bile (1), peritoneal fluid (1), and surgical incision site (1).
A variety of risk factors for acquisition of CRE were determined. Among the cases, 73% (11/15) were exposed to the anesthesia service, compared with 27% (8/30) of controls (odds ratio [OR] = 12.8, P = .017; Table 1). Exposure to the surgery service was also a significant risk factor, with 53% of cases (8/15) having this exposure compared with 20% (6/30) of controls (OR = 4.0, P = .046). Other services showed no significant difference in odds. blaNDM-5 CR-E coli case patients were also more likely to have been exposed to endotracheal intubation than control patients. Among cases, 73% (11/15) were exposed, compared with 33% (10/30) of controls (OR = 10.0, P = .03). Other procedures showed no significant difference between cases and controls. Exposure to various medications, including glucocorticoids and any antibiotics before culture, showed no significant difference between groups.
Review of the medical device reprocessing protocols by the PDPH revealed that there was no standard approach for reprocessing of endotracheal tubes between patients. A written SOP was available, but the review identified deficiencies during an audit of device reprocessing. There was an observed lack of consensus among staff regarding the standard soaking time, concentration of disinfectant to be used, or standard precleaning approaches. An unofficial practice of discarding tubes from patients with suspected respiratory infections was in place, but, in general, endotracheal tubes were disinfected and reused. It was also determined that the disinfection productc being used at the time of the review was for hard-surface disinfection and not for high-level disinfection. It was subsequently determined that only new, sterilized endotracheal tubes would be used in the facility. The new policy was implemented 50 days after the discharge of the 15th patient to be identified.
Discussion
In response to an outbreak of blaNDM-5 CR-E coli in companion animals, this study investigated potential risk factors for in-hospital acquisition. Although CRE in companion animals have been documented across the world (North America,6 Europe,2,5 Asia,3 Africa,4 and Australia12), these reports are sporadic and generally included only a few animals. The outbreak of blaNDM-5 CR-E coli in companion animals described here was large enough to identify potential risk factors for transmission in the veterinary setting. After adjusting for species and length of stay testing through matching, exposure to the anesthesia service, surgery service, and endotracheal intubation were significantly associated with acquisition of blaNDM-5 CR-E. coli. A review of the hospital Infection Prevention Policy and Procedure Manual and associated SOP’s discovered inappropriate reprocessing of endotracheal tubes.
Exposure to the anesthesia service, surgery service, and endotracheal intubation were not only statistically significant but also clinically important. Case patients had more than 10 times higher odds of being exposed to the anesthesia service and endotracheal intubation than controls. Anesthesia, endotracheal intubation, and surgery are linked (although they do not completely overlap), pointing to a potential exposure to blaNDM-5 CR-E. coli during the preparation for anesthesia. In human medicine, risk factors for infection or colonization with CRE include admission to the intensive care unit (ICU), mechanical ventilation, presence of indwelling devices, receipt of immunosuppressive drugs, and prior antimicrobial exposure.13–16 CRE are predominantly believed to be spread via healthcare worker’scontaminatedhands,17 although endoscopes have notably been documented as a source of human transmission.18 Additional studies, including environmental sampling or the use of qualitative methods may be useful in future outbreaks to identify what specificaspects of surgery or anesthesia are prone to human error or serve as critical control points for CRE infection control.
One key difference between veterinary and human healthcare settings is the common practice of reusing endotracheal tubes in veterinary medicine.19 A duodenoscope, a medical device that is commonly reprocessed when used in human medicine, was implicated in the spread of a New Delhi metallo-beta-lactamase–producing E coli in a tertiary care human hospital in northeastern Illinois.18 In that study, case status was significantly associated with a history of exposure to the duodenoscope. Although no lapses of duodenoscope reprocessing occurred, after the hospital changed its reprocessing procedures from automated high-level disinfection with ortho-phthalaldehyde to gas sterilization with ethylene oxide, no additional case patients were found. The authors concluded that the duodenoscope was a likely source of transmission. The results from this case–control analysis and the finding that endotracheal tubes were being inappropriately reprocessed supports the potential for CRE to move between animal patients during endotracheal intubation with reprocessed endotracheal tubes.
Notably, some risk factors for CRE transmission in human hospitals were not found to be significant risk factors in our study. Admission to the ICU, insertion of a medical device, and receipt of glucocorticoids were not associated with CRE, despite being risk factors in human medicine. Antibiotic use was also not found to be significantly associated with CRE acquisition. This may be because of low power from the number of patients in this outbreak, differences in companion animals and humans, or differences in practices in veterinary and human medicine. Exposure to the ICU had a high OR (6.97) and a low but not significant P value (.076), indicating that this may be a risk factor, but this study may be too underpowered to detect it.
The small size of this outbreak limited the power of the study. Matching and conditional logistic regression also were used to increase the study’s ability to detect risk factors for blaNDM-5 CR-E coli infection and colonization.20 If more patients were available, additional risk factors in this analysis may have been determined. However, even with a small sample size, the associations between the anesthesia and surgery services and endotracheal intubation with blaNDM-5 CR-E coli were strong and supported by the outbreak investigation’s review of infection prevention protocols. Misclassification of exposures or CRE infection or colonization status may also occur. Restricting controls to patients with bacterial culture results was used to reduce misclassification bias; however, the data are limited to the accuracy of the information recorded in HIS and in microbiology records. This study is also specific to exposures during an individual hospitalization that affect the risk of isolation of blaNDM-5 CR-E coli and does not include risks related to medical care outside the tertiary hospital admission, the impact of concurrent diseases, or community exposure. It was not within the scope of the study to assess the association between exposures before hospitalization and CRE. This study took place in a tertiary referral veterinary teaching hospital in Philadelphia, and the generalizability of these findings to the practice of veterinary medicine in smaller hospitals may be limited. However, CRE colonization of pets likely poses a severe risk to animal and human health.
Human healthcare settings have mitigated the risks of CRE acquisition through patient isolation, contact colonization screening, limiting the use of invasive devices, improving environmental cleaning and hand hygiene, and antibiotic stewardship.21 Our findings, and the human medical literature, suggest that veterinary hospitals should consult qualified infection control professionals and review specific infection control procedures associated with instrument reuse and reprocessing in the event that a CRE is isolated from a patient. Specifically, out of an abundance of caution, it is recommended that veterinarians should not reuse or reprocess endotracheal tubes against manufacturers’ instructions.
Reducing the spread of CRE is vital for both animal and human health. Multidrug-resistant bacteria like CRE are implicated in increasing the length of hospitalization, severity of disease, and cost of care for veterinary patients.22,23 CRE can colonize the gastrointestinal tract and remain undetected, serving as a potential source of community spread.24,25 Although CRE are still rare in veterinary medicine, evidence from human medicine points to the potential for the prevalence of these highly drug-resistant organisms to increase quickly. A study of community hospitals in the Southeastern United States recorded an over fivefold increase in the detection of CRE among human patients between 2008 and 2012.26 As CRE become more prevalent in companion animals, veterinary medicine must be prepared to take additional precautions to ensure the health of companion animals, staff, and pet owners.
Conclusion
In this outbreak, hospital risk factors for blaNDM-5 CR-Ecoli acquisition included exposure to the anesthesia service, surgical service, and endotracheal intubation. The device reprocessing policy was changed, and only new endotracheal tubes are now used. It is our expectation that the improved infection control practices will lead to a reduction in the number of blaNDM-5 CR-Ecoli–positive animals identified at our hospital. To reduce the spread of CRE in companion animals and humans, veterinarians should strengthen infection control procedures, specifically regarding the reprocessing of endotracheal tubes.
Contributor Notes
From the Department of Pathobiology, University of Pennsylvania School of Veterinary Medicine, Philadelphia, Pennsylvania.
