Extended-spectrum β-lactamase (ESBL)-producing organisms pose a great challenge to epidemiologists, infection control practitioners and physicians1,2. The dissemination of these broad spectrum resistant bacteria is difficult to control and the therapeutic options for severe infections are limited1,3,4.
The emergence of ESBL-producers bacteria is increasing in both hospitals and community5. This resistant mechanism is most found in Klebsiella pneumoniae in hospital and Escherichia coli in community as well5,6.
There are several reasons for the increasing prevalence of these organisms in hospitals4,7, such as, the selective pressure of antimicrobials overuse, the use of invasive devices, cross-transmission between patients, hospital cross-infection, and the increase in prevalence of community origin6,8,9,10.
Infection by these organisms is associated with higher mortality rates11,12. Carbapenems use is associated with lower mortality in patients with serious infections8,13. Although other agents may be used in non-severe infected patients, this use must be viewed with caution14,15,16,17
We conduct a case control-study in a cardiac center in Brazil to identify risk factors for ESBL-producing organisms identification.
A matched case-double control study was conducted at Instituto de Cardiologia, a 250-bed hospital for cardiology patients in southern Brazil. Instituto de Cardiologia attends adult and pediatric, surgical and clinical cardiology patients. In addition, the hospital has a cardiac transplant service, and three intensive care units (ICUs), which account for 16% of institution’s beds.
Adult (age ≥ 18 years) inpatients were selected from the entire hospital irrespective of unit. Case patients were those with identification of ESBL-producing bacteria at any site, after 48 hours of admission. For this analysis, we included two control groups: Group 1 was composed of patients from hospital units but not with ESBL-producing organisms; and patients with non ESBL-producing composed Group 2 Klebsiella spp., Escherichia coli, or Proteus mirabilis. Controls were matched in terms of age (± 3 years), date of sample identification (± 3 days), gender, and hospital ward.
From January 2008 to December 2009 all ESBL-producing organisms were included (patient cases). Controls were selected in a rate 1:2 (Group 1): 1 (Group 2).
All samples were processed at the microbiology laboratory at Instituto de Cardiologia. Detection of ESBL-producing bacteria was made according to the National Committee for Clinical Laboratory and Standards guidelines18. The susceptibility to antibiotics used agar disc diffusion method tested. For detection ESBL-producing strains, Double-disc synergy test was used. Bacteria considered as multidrug resistant organism (MDRO) were as follows: methicillin resistant Staphylococcus aureus (MRSA), ESBL-producing Escherichia coli, Klebsiella spp., Proteus mirabilis, carbapenems resistant Acinetobacter baumanii and Pseudomonas aeruginosa, vancomycin resistant enterococci (VRE).
We reviewed data from patients’ medical charts. Antibiotic use was measured for 24 months of study period, for the entire hospital. Data from prescribed drugs, such as, carbapenems, fluoroquinolones, cephalosporin’s, piperacillin/tazobactam, vancomycin, ampicillin+sulbactam and oxacillin, metronidazole, clindamycin, sulfametoxazole+trimetoprrim were reviewed. Patient comorbidities, invasive devices use and surgical procedures, ICU admission, previous MDRO identification in the previous 90 days, previous hospital stay during the last year, and other than ESBL-producing MDRO identified in patient ward were also reviewed.
Previous antibiotic use was at least 48h of inpatient use in the current admission. Time-at-risk is defined as the duration of time between admission and the detection of the antibiotic-resistant organism on culture for cases; as the number of days between admission and detection of the susceptible organism on culture, for non-resistant enterobactereacea infected controls; and the time between admission and discharge for non-infected control patients. Antibiotic consumption counted a number of defined daily doses (DDD), expressed as DDD per 100 patient-days. Central venous catheter (CVC), urinary catheterization, and mechanical ventilation were considered as invasive devices.
A descriptive analysis of the variables collected from each patient was performed. The chi-squared test or Fisher’s exact test were used for univariate analysis of selected categorical variables. All odds ratios on univariate analysis were controlled for time at risk exposure. Associations were considered statistically significant when P value was ≤.05. Multivariate analysis, along with 95% confidence intervals (CI) and Odds ratios were calculated using the Logistic regression model. We divided the analysis in three models. The first model included the risk factors other than specific antimicrobials that were statistically significant on univariate analysis; the second model included the antimicrobials with statistical significance on univariate analysis and days on antimicrobial use. The final model included the variables that were statistically significant in the first two models. All analysis was corrected for time at risk. All collected data was stored in Excel® 2000 version and analyzed using SPSS® 18.0 program.
The research and ethics committee of Fundação Universitária de Cardiologia (Brasil) approved the study and waived the need for informed consent because of the nature of the study.
Patient characteristics are shown in Table 1. Through the study period, we included fifty-eight case patients. Most of them infected with Klebsiella spp. (65.5%; N=38). The rest of patients were infected with Proteus spp. (25.9%; N=15) and E. coli (8.6%; N=5). Control
Group 2, were composed by 70 patients with identification of non ESBL-producing Klebsiella spp. (85.7%; N=60), Proteus spp. (10.0%; N=7), and E. coli (4.3%; N=3). One hundred and twenty patients were included in control Group 1.
Most patients were at general ward (55.2%; N=137); 26.6% (N=66) at post-surgical ward; 15.7% (N=39) at ICU; and 2.4% (N=6) were at emergency department. From case patients, most specimens were from urinary tract (39.6%; N=23), respiratory tract (32.8%; N=19), and surgical wound (12.1%; N=7). From control group 2 sites of specimen identification were as follows: respiratory tract (45.7%; N=32), urine (31.4%; N=22), blood culture (10.0%; N=7), and surgical wound (7.1%; N=5).
From January 2008 to December 2009 the mean consumption of antibiotics (in DDD/100 patient-days) were as follows: fourth-generation cephalosporins (5.9), third-generation cephalosporins (0.3), second-generation cephalosporins (2.0), first-generation cephalosporins (4.1), oxacillin (5.7), ampicillin+sulbactam (4.4), piperacillin+tazobactam (2.2), quinolones (2.1), vancomycin (1.4), and carbapenems (0.8).
Table 2 shows the multivariate analysis of statistically significant variables on univariate model, for both control groups.
Antimicrobials, associated with ESBL-producing organism, on univariate analysis of control Group 1, controlled for time at risk, were: piperacillin+tazobactam (OR 3.14, IC 95% 1.33-7.39; P<0.01), first-generation cephalosporins (OR 0.35, IC 95%, 0.16-0.73; P<0.01), second-generation cephalosporins (OR 4.99, IC 95% 1.44-17.27; P=0.01), fourth-generation cephalosporins (OR 3.71, IC 95%, 1.65-8.33; P<0.01), metronidazole (OR 11.14, IC 95%, 1.24-100.39; P=0.03). From the analysis of control group 2: second-generation of cephalosporins (OR 3.97, IC 95%, 1.00-15.46; P=0.05), and fourth-generation cephalosporins (OR 3.06; IC 95%, 1.24-7.55; P=0.01). Multivariate analysis of specific antimicrobial use is shown in Table 3.
The table 4 illustrates the final multivariate model, where specific antimicrobial drugs, with statistical significance on previous multivariate model and length of antimicrobial use are tested with significant statistical variables related to patient severity, previous presence of MDRO, or MDRO in patient ward.
To answer the question “What are the risk factors for acquiring ESBL-producers pathogen among hospitalized patients?” Our study concluded that the second-generation cephalosporins, metronidazole, days on antibiotic use and previous MDRO other than ESBL were independently associated with ESBL-producer. For the question “What are the risk factors for developing ESBL-producers pathogen among patients with non-ESBL Klebsiella spp. E. coli, and Proteus spp.?” The answer was the second- and fourth generation cephalosporins, previous identification of a MDRO other than ESBL, and, interestingly, the presence of other MDRO in ward were associated with protection form ESBL-producers identification.
In our setting where there is a high consumption of cephalosporins, especially fourth-generation cephalosporins, but a low consumption of third-generation cephalosporin and quinolones, use of second- and fourth-generation cephalosporins were associated with ESBL production. Besides the use of piperacillin+tazobactam was not associated with ESBL production. The hospital antimicrobial stewardship program (ASP) recommends the use of fourth- and second- instead of the third-generation cephalosporins and quinolones, because the associated risk of resistance linked to these drugs19,20. Besides the use of penicilins and piperacillin+tazobactam are stimulated, the carbapenems are the drugs reserved for infection by ESBL-producers21,22.
A variety of antimicrobial classes and antimicrobial drugs have been associated with ESBL-production in bacteria23,24. Cephalosporins, especially third-generation cephalosporins, fourth-generation cephalosporins, the quinolones, piperacillin+tazobactam, vancomycin, penicilins and beta-lactamase-inhibitors penicilins, and gentamicin all have been linked to ESBL-producing bugs25. Although a few studies have assessed important methodological principles like those that control group selection derived from appropriate sampling of the base population, time at risk and comorbid illnesses for studies on ESBL resistance.
Behar et al, evaluated risk factors for ESBL resistance in Klebsiella pneumoniae in four different groups of comparisons, with two types of controls: non-resistant Klebsiella pneumoniae and patients without infection derived from the same source of cases26. They found the cephalosporins use as a consistent risk factor. Besides, time at risk and CVC use were also identified as risk factors.
Furthermore, Odds ratios were slightly higher considering antimicrobial use in the comparison between case and control Group 2, as stated by others. Furthermore, the cephalosporins as risk factors confirms the risk associated with ESBL-producers, however, our study confirms that second- and fourth-generation cephalosporins can also be related to ESBL-resistance in a setting of low use of other cephalosporins. In addition, this is the first study to link metronidazole use to ESBL-producers. Interestingly, this association was only seen in control group 1. We speculate that this might be related to the suppression of intestinal anaerobic flora and the rise of resistant enterobactereacea. The use of other antimicrobial with activity against anaerobes like meropenem, ampicillin+sulbactam, and piperacillin+tazobactam were not related to resistance probably because enterobactereacea suppression flora too20.
A few institutions implemented the use of piperacillin+tazobactam intended to reduce ESBL-producers. Other independently factors associated with ESBL were: previous MDRO isolation and days on antibiotic use. Interestingly, the presence of any MDRO in case patient ward was a protector factor for identification of ESBL-resistance, a group from Baltimore suggests that patient-to-patient transmission is not an important cause of the acquisition of ESBL-producing27. Although, we have not studied the association between adherence to contact precautions and resistance, we suppose that with the identification of a MDRO other than ESBL-producers in the ward, the implementation of contact precautions protected other patients from ESBL-producing bacteria. In our hospital, most general wards were composed of three to five patients dividing the same room.
Control-group selection in antimicrobial resistant studies is of great importance26. As recommended, we choose to include controls selected from the same unit of cases patients that were in hospital in a period near to the case patient resistant bacteria identification. Besides, using two control groups we try to control for the selection bias that arises when only antimicrobial-susceptible organisms are used as control patients28. Although we could not include in the final analysis a severity score for our patients, the inclusion of comorbities that were statistically related to the outcome on univariate analysis and the adjustments for time-at-risk on multivariate analysis could reduce bias to this confounding. Patients admitted for longer periods are more likely to receive a greater number of antibiotics20. In addition, time-at-risk correlates with illness severity.
CONCLUSION
Our study has some limitations: our patients are from a specific cardiac center, which caused difficult generalization of results. We could not study the mechanisms of resistance in our patients. Our patients’ selection was based on clinical specimens collected as decision from the attending physician. We did not systematically collected surveillance samples, this way we could not ascertain about previous colonization.
In order to review our antimicrobial policy we aim to investigate which specific antimicrobial or other risk factor was associated with ESBL-producing bacteria. Although cefepime and cefuroxime have been used for preserving third-generation cephalosporins, they were also associated with ESBL-resistance29. The implementation of contact precautions and a strict adherence to this measure contributes to the control of resistance in our setting30.
ACKNOWLEDGEMENTS:
Authors acknowledge the immense help received from the scholars whose articles are cited and included in references of this manuscript. The authors are also grateful to authors/editors/publishers of all those articles, journals and books from where the literature for this article has been reviewed and discussed.
FUNDING:
This study was conducted without financial support, study resources came from the authors.
CONFLICT OF INTEREST:
The authors declare that they have no competing interests.
IJCRR 
This work is licensed under a Creative Commons Attribution-NonCommercial 4.0 International License
