Risk Factors and Outcomes of Bacteremia Caused by Carbapenem Resistant Enterobacterales Compared to Carbapenem Susceptible Enterobacterales


 BackgroundDue to shrinking therapeutic options, infections due to Carbapenem-resistant Enterobacterales (CRE) are an urgent threat in healthcare systems. We compared the risk factors and outcomes of bacteremia secondary to CRE with bacteremia secondary to carbapenem susceptible Enterobacterales (CSE).MethodsWe conducted a retrospective cross-sectional study on patients admitted to a tertiary care hospital in Karachi, Pakistan between 2013 and 2016. Patients with CRE bacteremia were matched to those with CSE bacteremia while excluding those with polymicrobial cultures.ResultsA total of 131 patients were enrolled (65 CRE and 66 CSE) with the mean age of 51.8 years and 57.1 years in CRE and CSE groups respectively. Compared with CSE, CRE bacteremia was more likely to occur in patients with Diabetes Mellitus or those with a tracheostomy (P = 0.002 and 0.014, respectively). The most common source of CRE bacteremia was central line associated (24.6% of all cases) as opposed to urinary tract infections in those with CSE bacteremia (62.1% of all cases). Fewer patients with CRE bacteremia received appropriate antibiotics (72.3% vs. 81.8%). Mortality was significantly higher in the CRE group (41.5% vs. 12.1%, P = 0.001) even when adjusted for the severity of illness using the PITT-bacteremia score. Increased mortality was also associated with central venous catheterization in both groups. The median length of hospital stay was longer in patients with CRE (8 vs. 6 days, P = 0.021)ConclusionCRE bacteremia was associated with central lines and led to significantly higher mortality and length of stay.


Background
Infections from antibiotic-resistant organisms occur in 2.8 million individuals and result in 35,000 deaths annually in the United States alone (1). Globally, these infections result in an estimated 700,000 deaths annually-a number expected to rise to 10 million by 2050 if no action is taken to counter antimicrobial resistance (AMR) (2). The global economic loss from AMR is projected to be between USD 1 trillion to USD 3.4 trillion annually by 2030, with a greater impact on low-income countries (3).
Carbapenems have been considered one of the last treatment options for infections caused by multidrugresistant pathogens (4). In recent years, increased carbapenem use coupled with poor infection prevention practices has led to an increase in the global rates of Carbapenem-Resistant Enterobacterales (CRE) (5).
CRE are typically resistant to other beta-lactams and often resistant to other antimicrobial classes, signi cantly limiting treatment options. Additionally, infections with CRE result in higher morbidity when compared to non-CRE strains(6-8). This had led the Centers for Disease Control and Prevention to classify CRE as an "urgent threat".
Carbapenemase production is one of the more concerning mechanisms of carbapenem resistance in Enterobacterales. Carbapenemase-encoding genes on mobile genetic elements can rapidly transmit resistance to other bacteria, and infections from CP-CRE result in higher mortality compared to those with non-CP CRE (9). The multi-national PANORAMA study from low and middle-income countries found most CRE to be Carbapenemase producers (93.75%) (7) in these regions. In contrast, carbapenemase production accounts for only 30% of carbapenem resistance in the US (1). Despite this evidence of geographic variation in the more virulent CP-CRE strains, data from low and middle-income countries on surveillance and outcomes from CRE infections remains scarce to date(10).
We, therefore, aimed to compare the risk factors for bacteremia caused by CRE compared to Carbapenem Susceptible Enterobacterales (CSE) strains and subsequently assess outcomes of bacteremia caused by CRE in an area of high CRE endemicity in a low-middle income country.

Methods
We conducted a retrospective cross-sectional study at a 700 bedded tertiary care teaching hospital in the south of Pakistan. All adult patients admitted between January 2013 and July 2016 with bacteremia due to Enterobacterales were included. For every case of CRE bacteremia identi ed, a CSE control was randomly selected from the same month. We excluded patients with polymicrobial bacteremia.
Clinical data was collected from patients' medical records using a structured proforma. Data was analyzed using the SPSS 19.0. The baseline parameters of patients with CRE and the CSE bacteremia were compared. Comparison of differences in proportions was assessed by using the Chi-square test or Fisher exact test where appropriate. In univariate analyses, a comparison between CRE and CSE was done for each variable of interest. Multivariable analysis was conducted to identify the factors associated with poor outcomes. All p-values were two-sided and considered statistically signi cant if < 0.05. Enterobacterales showing decreased susceptibility to carbapenems (diameter for imipenem greater than 19mm, meropenem greater than 19mm). On the other hand, CSE was also identi ed via blood culture and susceptibility reports which showed Enterobacterales sensitive to any antibiotic of the carbapenem family.
Appropriate antibiotic regimens were de ned as one containing at least one agent to which the organism was susceptible. For the CRE group, the antibiotic regimen considered appropriate in our setting was primarily a meropenem-based regimen with or without Colistin. Patients receiving appropriate antibiotic  The Susceptibility pattern of the CRE and CSE organisms to various antibiotics is displayed in Fig. 1

Discussion
This study was designed to investigate the risk factors and outcomes of bacteremia caused by Carbapenem-resistant (CRE) Enterobacterales compared to Carbapenem Susceptible Enterobacterales (CSE). We found that infections with CRE were more severe than those with CSE, with a higher Pitt bacteremia score, and led to signi cantly higher mortality, higher rates of ICU admission, and longer median length of hospital stay. Moreover, CRE infections were most commonly nosocomial and associated with central line placement.
Nearly a third of CSE were resistant to 3rd generation cephalosporins, possibly re ecting a high endemicity of Extended-Spectrum Beta-Lactamase producers in the community. The increased prevalence of ESBL producing species in India and South Asia in general have been hypothesized to have driven the use of Carbapenems leading to selection pressure for the emergence of Carbapenem Resistance among Enterobacterales (12)(13)(14). While overall most infections were community-acquired, the frequency of nosocomial infections (37.4%) exceeded that reported in other resource-limited settings (5.7%-19.1%) (15). Additionally, we observed a higher proportion of nosocomial infections with CRE (52.3%) than that observed in a multi-center study in resource-limited settings (40%) (7). These variations can be due to differences in compliance with infection control, antibiotic stewardship, and endemicity of antimicrobial resistance, and highlight the relevance of individual institutional data to guide clinical practice.
Outcomes and disease severity in patients with CRE infection remained worse compared to those with CSE infection, even though there was no statistically signi cant difference in the initiation of timely and appropriate antibiotic therapy between the groups. Existing literature in both developed and resourcelimited settings has consistently reported similar outcomes for infections (7,(16)(17)(18)(19)(20)(21)(22) with CRE Similar to the ndings of a meta-analysis(22), we observed a higher odds for mortality for CRE bacteremia compared with CSE bacteremia on multivariate analysis. Similarly, compared to the multi-center PANORAMA study that reported a 35% inpatient mortality for CRE bacteremia in resource-limited settings, we found a higher mortality rate of 45.8% at our center (7). However, mortality in our study was slightly higher than reported in the PANORAMA study. Mortality due to bacteremia is in uenced by several factors, including patient characteristics, source control, and appropriate antibiotic therapy. In our study, patients with CRE had a higher PITT bacteremia score (Pitt score 4 or greater in 40% vs 20%). This may be re ective of the higher proportion of CLABSI which is associated with more severe disease. Source control also affects mortality; however, we were unable to document source control in our patients. Finally, while there was no difference in the time to initiation of appropriate therapy in both groups, most patients received colistin (71.2% vs 52%), due to limited availability of alternative agents in Pakistan.
Colistin may have contributed to the morbidity given the higher rates of adverse effects and possible suboptimal e cacy (23,24). Furthermore, due to resource constraints, we were unable to conduct genetic analysis and were therefore unable to assess the proportion of carbapenemase-producing Enterobacterales in our cohort, which are known to cause a 3 -fold higher mortality compared to noncarbapenemase producing CRE(9).
Our study is limited by its retrospective design and restriction to a single, private tertiary care hospital in Karachi-limiting generalizability to public sector hospitals in the region. Additionally, although by using equal numbers of CRE and CSE infections we aimed to draw comparisons in outcomes and risk factors between the two groups, we were unable to estimate the incidence of the CRE and CSE infections during the study period.

Conclusion
Nevertheless, our study highlights that differences exist in the epidemiology and outcomes of infections with CRE and CSE within the developing world, emphasizing the relevance of exploring local data. Future studies are needed across multiple centers in Pakistan where antibiotic stewardship and quality assurance for antibiotic procurement is inadequate (25,26 The study was exempted by the Ethics and Review Committee (IRB) of Aga Khan University Hospital prior to its start (4461-Med-ERC-16). The study was exempted from patient consent as it only involved retrospective chart reviews with subsequent deidenti cation of data to ensure con dentiality.

Consent for Publications
Not Applicable

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

Con ict of Interest
The authors declare that they have no competing interests  Antibiotic Susceptibility Pattern (%)