Epidemiology of Carbapenem-resistant Enterobacteriaceae in Egyptian intensive care units using National Healthcare–associated Infections Surveillance Data, 2011–2017

Objective To describe the epidemiology of carbapenem-resistant Enterobacteriaceae (CRE) healthcare-associated infections (HAI) in Egyptian hospitals reporting to the national HAI surveillance system. Methods Design: Descriptive analysis of CRE HAIs and retrospective observational cohort study using national HAI surveillance data. Setting: Egyptian hospitals participating in the HAI surveillance system. The patient population included patients admitted to the intensive care unit (ICU) in participating hospitals. Enterobacteriaceae HAI cases were Klebsiella, Escherichia coli, and Enterobacter isolates from blood, urine, wound or respiratory specimen collected on or after day 3 of ICU admission. CRE HAI cases were those resistant to at least one carbapenem. For CRE HAI cases reported during 2011–2017, a hospital-level and patient-level analysis were conducted using only the first CRE isolate by pathogen and specimen type for each patient. For facility, microbiology, and clinical characteristics, frequencies and means were calculated among CRE HAI cases and compared with carbapenem-susceptible Enterobacteriaceae HAI cases through univariate and multivariate logistic regression using STATA 13. Results There were 1598 Enterobacteriaceae HAI cases, of which 871 (54.1%) were carbapenem resistant. The multivariate regression analysis demonstrated that carbapenem resistance was associated with specimen type, pathogen, location prior to admission, and length of ICU stay. Between 2011 and 2017, there was an increase in the proportion of Enterobacteriaceae HAI cases due to CRE (p-value = 0.003) and the incidence of CRE HAIs (p-value = 0.09). Conclusions This analysis demonstrated a high and increasing burden of CRE in Egyptian hospitals, highlighting the importance of enhancing infection prevention and control (IPC) programs and antimicrobial stewardship activities and guiding the implementation of targeted IPC measures to contain CRE in Egyptian ICU’s .


Background
Antimicrobial resistance (AMR) is being increasingly recognized as a global health security threat that requires integrated action across government sectors and society as a whole [1]. Carbapenem-resistant Enterobacteriaceae (CRE) are especially concerning pathogens due to their resistance to last resort antibiotics [2][3][4], high morbidity and mortality, and the high potential for their resistance to spread via mobile genetic elements [5,6].
CRE are often associated with healthcare transmission, as demonstrated in the United States, where more than 9000 healthcare-associated infections (HAI) are caused by CRE each year [7,8]. Healthcare-related risk factors associated with CRE infection include prolonged hospital stay, presence of invasive medical devices, admission to an intensive care unit (ICU), and previous exposure to antimicrobials [9][10][11][12]. Data on these risk factors are useful to guide CRE prevention and control efforts, but most data describing CRE epidemiology are reported from high resource settings [13].
The healthcare system in Egypt includes a network of secondary and tertiary healthcare facilities in the public, university, or private sector in 27 geographic regions or governorates. Egypt is one of the first countries in World Health Organization's (WHO) Eastern Mediterranean Region (EMR) to develop a prospective, standardized national HAI surveillance system. Established in May 2011, Egypt's HAI surveillance system aims to estimate HAI prevalence and incidence, establish national benchmarks, and describe HAI-causing pathogens in order to inform prevention activities [14,15]. Using data from Egypt's national HAI surveillance program, we described HAIs caused by CRE to examine burden, trends, and risk factors associated with CRE HAIs in ICU patients compared to those with carbapenem-susceptible Enterobacteriaceae (CSE) HAIs.

Egyptian HAI surveillance system
Egypt's HAI surveillance system was implemented in a phased approach, with 310 ICUs in 72 hospitals across 25 governorates collecting data between 2011 to 2017. Hospitals were selected to participate in the surveillance system based on the presence of hospital management support, well-trained infection prevention and control (IPC) teams including link IPC nurses in ICUs, adequate microbiology laboratory capacity (i.e., ability to conduct pathogen identification and antibiotic susceptibility testing [AST]), and data entry capacity. The surveillance system focuses on the four most common types of HAIs in ICU patients as identified by data from the first year of Egypt's HAI surveillance program [14,15]: bloodstream infections (BSI), urinary tract infections (UTI), pneumonia (PNA), and surgical site infections (SSI). The methodology of the national HAI surveillance program, including HAI definitions, have been described in earlier publications [14,15].

Definitions
HAI surveillance definitions used for BSI, UTI, PNA, and SSI were derived from the Centers for Disease Control and Prevention's (CDC) 2012 National Healthcare Safety Network (NHSN) [16], with minor adaptations for the primary BSI case definition. Cases of HAI with a culture growing an Enterobacteriaceae from a blood, urine, respiratory tract, or surgical site were included in this analysis. CRE cases were HAIs with Enterobacteriaceae isolates resistant to at least one of the following carbapenems: imipenem, meropenem, ertapenem [17]. CSE cases were HAIs with Enterobacteriaceae isolates not resistant to any of these carbapenems. This analysis was focused on the most common Enterobacteriaceae pathogens according to existing AMR data from Egypt, and included Klebsiella spp (only K. pneumoniae and K. oxytoca aggregated, but not other rare species), Escherichia coli and Enterobacter spp [18].

Microbiological testing
On a monthly basis, all isolates associated with signs or symptoms of infection were sent to the AMR reference laboratory for confirmatory identification and AST for quality control purposes. Antimicrobial susceptibility testing was performed using disk diffusion (Becton Dickinson, USA) according to Clinical and Laboratory Standards Institute guidelines (CLSI) [19]. Only microbiologic confirmatory testing data from the AMR reference lab were used for this analysis to ensure accurate laboratory results.

Electronic data collection, entry, and reporting
Egypt's HAI surveillance program uses electronic data collection methods and automated analysis, thereby reducing the workload on hospital HAI surveillance coordinators (SC) and ensuring timeliness of reporting and feedback. To identify cases, the hospital SC screens ICU patients for new clinical signs and symptoms suggestive of infection (fever, crackles, cough, etc.) at least 3 days per week by reviewing medical records, interviewing physicians, and reviewing diagnostic test results (microbiology results or radiology reports). When microbiology results are not available, the physician requests appropriate specimens to be collected for microbiology testing. When a patient is suspected of a HAI based on signs, symptoms, or diagnostic test results, patient information is entered into a standardized case report form installed on the electronic device. The device automatically analyzes the entered data to determine whether the patient meets the HAI case definition and if so, specifies the type of infection. Denominator data (i.e., patient-days, central line-days, urinary catheter-days, ventilator-days) are entered daily on a standardized denominator reporting form installed on the device. The clinical and epidemiological HAI data are uploaded weekly to a secured web-based surveillance application [15] . HAI microbiological data produced by the AMR reference laboratory are also uploaded to the web application, where the data are merged with the HAI clinical and epidemiological data. The web application has built in data quality checks and analytic tools for immediate data cleaning and analysis, which allows hospital teams to generate automated individualized facility reports.

Data variables
The following patient data were available for each case: patient demographics (i.e., age, sex), hospital and ICU type as defined by NHSN [20,21], admission and discharge date, length of ICU stay prior to specimen collection, location prior to ICU admission, symptoms that met the case definition, presence of invasive devices, associated surgical procedures in last 90 days, radiological results, and patient outcome (i.e. death, discharged). Relevant microbiology data were also collected including specimen type, specimen collection date, pathogen, and AST results.

Statistical analysis
Data on CRE and CSE cases identified between May 2011 and December 2017 were included in this analysis. An isolate-level analysis of all Enterobacteriaceae isolates from CRE and CSE cases, including multiple isolates from an individual case, was conducted to calculate the proportion of CRE isolates among all Enterobacteriaceae isolates, stratified by pathogen. A hospital-level analysis was conducted by calculating the proportion of hospitals with ≥1 CRE positive specimen among hospitals conducting HAI surveillance, stratifying by hospital type and size.
For the patient-level analysis, patients were counted as being a CRE case if they had any sample positive for CRE; however, patients with multiple Enterobacteriaceae isolates from the same admission, but from different specimen sites or different dates were counted only once. For patients with multiple isolates, only the first CRE isolate was included in the analysis. Frequencies of categorical variables were stratified by CRE status and compared between CRE and CSE cases using χ2 test or 2-tailed Fisher Exact test. For continuous variables, means with standard deviations (or medians and interquartile ranges if distributions were skewed) were stratified by CRE status and compared between CRE and CSE cases using Student t test or Mann-Whitney test, dependent on the validity of normality assumption.
Univariate analysis was performed using logistic regression to calculate odds ratios, 95% confidence intervals (CI), and p-values to determine the strength of the association between these variables and CRE status. The reference category was assigned based on the category with the highest frequency of Enterobacteriaceae HAI cases, except in the analysis of ICU category, where Surgical Critical Care was used as the reference because the neonatal intensive care unit (NICU) was considered too specialized to serve as the reference. Variables with chisquare test p-value < 0.10 in the univariate analysis were then included into a multivariable logistic regression model, using a forward stepwise approach to identify risk factors associated with CRE status. Variables were kept in the final model if the p-value for the likelihood ratio test was < 0.05. All analysis were performed using STATA 13 [22].
Incidence rates for CRE cases were calculated as the proportion of CRE cases per 10,000 patient days [20]. For the trend analysis, χ2 test was used to determine statistical significance of the trend in CRE case incidence and proportion of CRE cases among all Enterobacteriaceae cases.

Results
A total of 3836 Enterobacteriaceae isolates from 3109 patients were reported to Egypt's HAI surveillance system from 2011 to 2017. For the isolate-level analysis, 1105 (47.9%) of the 2306 Enterobacteriaceae isolates submitted to the AMR reference laboratory were CRE [ Table 1]. When stratified by pathogen, approximately half of Klebsiella (n = 929, 53.7%) and Enterobacter (n = 54, 43.5%) isolates were CR, while a smaller percentage of Escherichia coli isolates (n = 122, 27.1%) were CRE. The overall incidence of HAI due to CRE was 3.7 per 10,000 patient-days. Among the 72 hospitals performing HAI surveillance, 46 (63.9%) reported at least one CRE isolate during the study period, but this percentage varied by hospital type and hospital size ( Table 2).
CRE cases were less likely to have underwent a surgical procedure (OR = 0.72, 95% CI = 0.59-0.88) or had a urinary catheter during the hospital admission (OR = 0.52, 95% CI = 0.42-0.64), but there was no significant difference between CRE and CSE cases with respect to mechanical ventilation or central lines. Mortality was also higher among CRE cases than CSE cases (OR = 1.46, 95% CI = 1.2-1.79). When adjusted for patient and specimen characteristics, the multi-variate analysis showed that hospitalization immediately prior to ICU admission (OR = 1.38; 95% CI: 1.18-1.76; p-value 0.008) and a longer ICU stay prior to specimen collection (OR = 1.08; 95% CI: 1.14-1.36; p-value 0.03) remained significantly associated with carbapenem resistance, while infection with E. coli (OR = 0.22; 95% CI: 0.17-0.3; p-value < 0.001) and identification in a wound specimen (OR = 0.67; 95%CI: 0.5-0.89; p-value 0.01) were associated with not having carbapenem resistance. The association between carbapenem resistance and specific hospital or ICU types did not remain significant in the multivariable analysis.
There was an overall increase in the proportion of cases that were CRE (17.6 to 54.6%, p = 0.003), which remained when stratified by pathogen (Fig. 1). Although not statistically significant, the incidence of carbapenem resistance cases for all pathogens increased between 2011 and 2013, followed by a decline between 2013 and 2015. Since 2015, the incidence overall and for each pathogen has again been increasing (Fig. 2).

Discussion
Our analysis found that carbapenem resistance is widespread and the prevalence is increasing in Egypt. We found that more than half of hospitals (64%) had at least one CRE isolate and half (47.9%) of Enterobacteriaceae isolates were CRE, which is higher than estimates reported from other Arab, African, or Asian countries [23][24][25][26]. The incidence of CRE HAI (3.7/10,000 patientdays) is also much higher than the overall incidence of    [27][28][29]. The severity of the CRE problem in Egypt emphasizes the importance of healthcare facilities' to implement CRE prevention and control efforts. The Centers for Disease Control and Prevention (CDC) and the WHO have developed guidance for healthcare facilities on strategies proven to be effective for limiting CRE transmission, including IPC, antimicrobial stewardship, and CRE surveillance [29,30]. Data from this analysis can be used to guide how and where these strategies can be most efficiently implemented in Egypt.
Potential causes of the high prevalence of CREs in several hospitals in Egypt might be due to the limitation in implementing stewardship programs and IPC measures. IPC activities are a critical part of preventing healthcare-related CRE transmission and include practices such as hand hygiene, minimizing device use, environmental cleaning, and isolation through contact precautions and improve cohorting of patients or staff. Multivariate analysis did not identify specific  Previous global studies have found that CRE is associated with increased length of ICU stay, undergoing surgical procedures, and use of medical devices, specifically mechanical ventilation and central venous catheters [9,[33][34][35]. Our multivariable analysis did not find that carbapenem resistance was significantly associated with exposure to medical devices or surgical procedures. This may be explained by the very high frequency of invasive devices used in Egyptian ICUs, among both CRE and CSE cases. However, this finding should not discount the importance of CRE prevention strategies focused on reducing healthcare exposures, such as device utilization.
We identified a relative decrease in CRE incidence and proportion of carbapenem resistance in 2014-2015. The introduction of an analytic tool in 2013 may have contributed to the reduction in CRE incidence by providing more timely data feedback to hospitals for guiding IPC interventions. The high frequency of CR found in blood and the increased odds of carbapenem resistance among blood may reflect culturing practices, where blood cultures may only be ordered for very ill patients with resistant infections refractory to treatment.
The ability to adequately contain CRE requires sufficient microbiology laboratory testing to ensure accurate CRE detection and timely notification of laboratory results. In this surveillance system, some of the best laboratories in the country are included; the burden and epidemiology in other facilities, particularly those without reliable susceptibility testing, is largely unknown. Although global guidance recommends active surveillance by conducting CRE screening of patients in some situations with an outbreak or ongoing high prevalence of CRE, this practice is not routine in Egypt, and further efforts to implement such activities are likely needed.
The limited number of studies published about CRE in Egypt have involved few facilities and focused mainly on identifying genetic resistance mechanisms, rather than epidemiological risk factors [36][37][38]. This study uses data from many healthcare facilities across Egypt, thereby strengthening evidence about epidemiological risk factors associated with CRE. While there are different types and sizes of healthcare facilities included, they are a convenience sample and do not constitute a representative sample of the country.
Detection of CRE in this network was also limited by inconsistent microbiology and surveillance capacity at hospital laboratories. Some isolates were not tested for carbapenem susceptibility at the hospital laboratory or were not sent to the AMR reference laboratory for confirmatory testing, resulting in data for these samples not being included in the data analysis. Because active screening for CRE was not being performed, the variable propensity for clinicians to test patients also likely impacted CRE detection. The trend analysis does not take into account changes at the national, facility, or ward level which impact CRE detection, including changes in specimen collection and testing.
The scope of this surveillance system is limited to ICU locations since ICUs are expected to have both the highest risk for transmission and the most vulnerable population. Therefore, this data cannot be used to draw any conclusions about CRE burden or risk factors in non-ICU wards or the community. This surveillance system did not collect data on clinical variables which prior studies have found to be associated with CRE infections, such as patients' comorbid medical conditions or prior antibiotic exposure.

Conclusions
This study shows that CRE is prevalent and increasing in Egyptian hospitals, suggesting the presence of selective pressures and healthcare transmission. Future implementation of evidence-based IPC strategies to prevent CRE transmission, strengthening microbiology capacity and molecular characterization in addition to including antibiotic stewardship programs, are needed to reduce the burden of CRE and optimize patient treatment strategies. CRE surveillance should be strengthened to better track the incidence and prevalence of this pathogen and define the impact of interventions on the burden of this serious, emerging threat.

Disclaimer
The views expressed in this article are those of the authors and do not necessarily reflect the official policy or position of the Department of the Navy, Department of Defense, U.S. Government, the Centers for Disease Control and Prevention, the Egyptian Ministry of Health and Population, or the Egyptian Universities.
Authors' contributions SK participated in study design, overview the data collection, performed the analysis, wrote the paper. ML review the analysis, wrote the manuscript and oversight manuscript preparation. GI, MA, SG, AH, SH, JE, MZ, HR and GK participated in data collection. OH overview the data collection and wrote the manuscript. MT conceived and designed the study, review the analysis, wrote the paper and oversight manuscript preparation. All authors contributed and commented on the manuscript and approved the publication of this version and is accountable for the accuracy and integrity of the work.

Authors information
Sara Kotb is an epidemiologist in the Division of Global Health Protection, U.S. Centers for Disease Control and Prevention at the Cairo, Egypt Country Office with research interests in healthcare-associated infection surveillance and antimicrobial resistance.

Availability of data and materials
The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request.

Consent for publication
Not applicable.