Antibiotic Abuse and Antimicrobial Resistance in Hospital Environment: A Retrospective Observational Comparative Study

Background and Objectives: Antimicrobial resistance represents a serious problem, and it may be life-threatening in the case of severe hospital-acquired infections (HAI). Antibiotic abuse and multidrug resistance (MDR) have significantly increased this burden in the last decades. The aim of this study was to investigate the distribution and susceptibility rates of five selected bacterial species (E. coli, K. pneumoniae, P. aeruginosa, S. aureus and E. faecium) in two healthcare settings located in the Apulia region (Italy). Materials and Methods: Setting n.1 was a university hospital and setting n.2 was a research institute working on oncological patients. All the enrolled patients were diagnosed for bacterial HAI. The observation period was between August and September 2021. Clinical samples were obtained from several biological sources, in different hospital wards. Bacterial identification and susceptibility were tested by using the software VITEC 2 Single system. Results: In this study, a higher incidence of multi-drug-resistant K. pneumoniae was reported (42,2% in setting n.1 and 50% in setting n.2), with respect to the Italian 2019 statistics report (30.3%). All the isolates of E. faecium and S. aureus were susceptible to linezolid. All the bacterial isolates of P. aeruginosa and most of K. pneumoniae were susceptible to ceftazidime–avibactam. Amikacin and nitrofurantoin represented a good option for treating E. coli infections. Multidrug-resistant (MDR) P. aeruginosa, methicillin-resistant S. aureus (MRSA) and vancomycin-resistant E. faecium (VRE) had a lower incidence in the clinical setting, with respect to E. coli and K. pneumoniae. Conclusions: The data obtained in this study can support clinicians towards a rational and safe use of antibiotics for treating the infections caused by these resistant strains, to enhance the overall efficacy of the current antibiotic protocols used in the main healthcare environments.


Introduction
Antimicrobial resistance is considered a worldwide impacting burden, affecting the patients of critical hospital wards, such as Intensive Care Units (ICU). Hospitalized patients have been demonstrated to have an increased risk to develop infections due to exposure to several invasive devices (mechanical ventilation, urinary tract catheters) [1][2][3][4] and to other related conditions. Careful clinical surveillance, together with the monitoring of the wellknown bacterial strains responsible for inducing HAI, may help clinicians to choose the appropriate antibiotic therapies. The aim of this study was to investigate the distribution and susceptibility rates of five selected bacterial species in two healthcare settings located in the Apulia region (Southern Italy).
Bacterial infections have impacted humans throughout the centuries, until the discovery of antibiotics, which have revolutionized the treatment of infectious diseases. Because of their ability to survive in different environments, bacteria can increasingly face antibacterial treatments over time by means of different adaptative strategies [5][6][7][8][9]. They are able to modify the quaternary structure of specific target proteins, to substitute a metabolic pathway by synthesizing alternative biomolecules and to produce enzymes able to inactivate antibiotics; this is also possible through the camouflage of their structure, for example, behind a proteoglycan capsule [10][11][12][13].
A common bacterial weapon against penicillin is the beta-lactamase enzyme, which alters the beta-lactamic structure, thus maintaining the building of the bacterial wall and creating the local conditions to promote several diseases [14][15][16][17][18]. Bacteria are also able to synthesize effective isoforms of the beta-lactamase enzyme; the extended spectrum beta lactamase (ESBL) and the ESBL carbapenemase give bacteria resistance towards third-generation cephalosporins and carbapenamase-class antibiotics, respectively. These antibiotics are widely used in several nosocomial infections [19,20].
The spread of antibiotic-resistant bacterial strains is a severe issue for the healthcare systems of several countries [21,22]; it can be considered as an effect of antibiotics abuse [23][24][25]. Antibiotic-resistant bacteria are able to rapidly disseminate within the human body by transferring a ring of DNA to other species or strains [26][27][28].
According to the European Center for Disease Prevention and Control (ECDC), the most common and clinically relevant bacterial species in European hospitals include E. coli, P. aeruginosa, K. pneumoniae, S. aureus and E. faecium.
Common nosocomial infections involve the soft tissues, the urinary tract, the gastrointestinal organs and the respiratory apparatus. It is interesting to highlight that about 33% of patients receive antibiotics during their stay in hospital and about 6% undergo a hospital-acquired infection (ECDC) [29]. The spread of antibiotic resistance in Intensive Care Units (ICU), mainly due to the use of broad-spectrum antibiotics, has become a particular problem for clinicians and patients [29,30].
Among hospital-acquired infections in Europe, 41% of S. aureus infections are methicillinresistant (MRSA), 24% of E. coli infections are resistant to cephalosporins (ESBL), 18% of E. faecium infections are resistant to vancomycin (VRE) and 32% of P. aeruginosa infections are resistant to carbapenems (ESBL-carba). Multidrug resistance (MDR) for P. aeruginosa is defined as a resistance to three different classes of antibiotics: beta-lactams (penicillintazobactam, cephalosporins or carbapenems), aminoglycosides and fluoroquinolones.

Materials and Methods
In this study, the antibiotic resistance of two hospital settings was compared. Setting n.1 is a big University hospital, A.O.U.C. Policlinico in Bari, Southern Italy. Setting n.2 is a Cancer Research Institute, IRCCS Istituto Tumori "Giovanni Paolo II" located in Bari, Southern Italy.
Both of the settings are located in the macro-region Apulia, Italy. Based on the findings of ECDC, five of the most common pathogens in clinical settings have been chosen to build up this survey: E. coli, K. pneumoniae, P. aeruginosa, S. aureus and E. faecium. The inclusion criteria have been described below.
Within setting n.1, clinical samples were obtained by four different departments (Neonatology, Infectious Disease, Intensive Care Unit and Internal Medicine). Within setting n.2, clinical samples were obtained by all departments (surgery, oncology, hematology, interventistic oncology).
The most suitable clinical samples and materials to correlate with a potential clinical infection were blood, urine and respiratory cultures, rectal swabs and samples extracted from devices implanted into the patient's body, such as venous catheters. All patients were included, regardless of age, but multiple samples of the same date for the same patient with the same result were considered only once in the report.
From 30 August to 30 September 2021 (only one month, to easily build a research framework), clinical samples were collected retrospectively from the software VITEC 2 Single system (BioMérieux, Inc, Hazelwood, Mo, USA). The bacteria were identified by using VITEC MS (MALDI system, BioMérieux, Grassina, Italy). MALDI-TOF-MS uses the software Mass-Up, distributed under license GPLv3 [31]. The antibiograms were achieved starting from standard dilution in physiological solution to 0.53-0.67 density; then, the samples were cultured overnight, with antibiogram cards for determination of Minimal inhibitory concentration (MIC) values and interpreted to Eucast European Committee On Antimicrobial Susceptibility Testing (EUCAST) clinical breakpoints for susceptibilty. ESBL phenotype was defined as resistance to cefotaxime or ceftazidime and inhibition by ESBL inhibitor such as tazobactam. Finally, we have demonstrated the MIC results of bacterial strains examined by using ETEST (BioMérieux, Italy) [31].
In setting n.2, among the six isolated bacteria, only two (33.3%), deriving both from urine samples, showed antibiotic resistance to cephalosporins and penicillin in the oncology department, showing no other resistance patterns.
No cases of resistant E. faecium strains were observed in setting n.2.
No cases of S. aureus resistance were observed in setting n.2.
No cases of resistant P. aeruginosa strains were observed in setting n.2.

K. pneumoniae
In setting n. Data from setting n.2 showed four bacterial isolates of MDR K. pneumoniae infections from a blood culture and three rectal swabs, with an incidence rate of 50% (4/8). K. pneumoniae cases overlap with data from setting n.1 in terms of multidrug resistance to antibiotics; in particular, all strains were resistant to cephalosporins, fluoroquinolones, carbapenems and aminoglycosides.

Discussion
The results of the surveillance on antimicrobial resistance obtained in this study conducted in Apulia are in agreement with the 2019 national report. According to ECDC, the proportion of resistant bacterial isolates of each species was higher than the European average for hospital-acquired infections, except for MRSA, which was similar (40-41%) [12][13][14][15][16].
The different trend principally refers to the higher incidence of multidrug-resistant K. pneumoniae, recognized by clinicians as a very difficult challenge [32]. In detail, the Italian 2019 statistics report that 30.3% of K. pneumoniae was MDR, while in this study the rate was closer to 50% both in setting n.1 (42.2%) and in setting n.2 (50%). In the report study conducted in setting n.2, it is evident there is a correlation between MDR K. pneumoniae infections and low immunity defense in hematologic patients [33,34]. In setting n.1, extensively resistant strains of K. pneumoniae, susceptible to only one or two antibiotics such as colistin or amikacin, were not rare. The combination of MDR K. pneumoniae with the few treatment options and its prevalence in respiratory and urine cultures still represent a great safety problem for clinicians [15][16][17][18].
Within the data collected, the species here investigated were shown to promote infections in different biological sites: E. faecium was prevalently isolated from venous catheter cultures, P. aeruginosa from respiratory cultures, S. aureus from wound cultures and K. pneumoniae and E. coli were mainly isolated from urine cultures (and rectal swabs for setting n.2) [7,12,19,22].
In setting n.1, Internal Medicine wards produced more cultures, probably due to the number of patients. Predomination of K. pneumoniae and P. aeruginosa was evident in ICU and in respiratory cultures with respect to other species. The cultures from Neonatology included a larger number of S. aureus and E. coli infections.
All isolates of E. faecium and S. aureus were susceptible to linezolid, which could be an alternative to vancomycin as empirical treatment for resistant Gram-positive pathogens (30% of VRE). S. aureus strains were also sensitive to teicoplanin, tigecycline, vancomycin and daptomycin. No resistant strains of S. aureus to trimethoprim-sulfamethoxazole were observed.
All isolates of P. aeruginosa and most of K. pneumoniae were susceptible to ceftazidimeavibactam. Therefore, this association represents a potent weapon in the treatment of resistant Gram-negative infections. E. coli strains were not tested for ceftazidime-avibactam, having many other treatment options, including carbapenems [33,34].
The results also indicate that amikacin could be a good choice for treating E. coli infections, with only 4.6% resistance compared to 18.6% gentamycin resistance.
Nitrofurantoin remains an excellent treatment option for uncomplicated cystitis caused by E. coli, even in hospital settings; in fact, no resistant strains were observed in this study.

Conclusions
In this study, patients with MDR bacterial infections were selected across various hospital wards. Samples collected from these medical departments gave us a panorama of which kinds of bacterial strains must undergo continuous surveillance. Oncological and hematological, Intensive Care Unit and Neonatology patients have an especially high risk of infections with bacterial resistance to most common antibiotics. In these patients, the most proper empirical antibiotic therapy must be applied with the aim to enhance efficacy and decrease forms of resistance. Data obtained from the current studies can address the rational use of antibiotics by clinicians by avoiding the inappropriate use of non-active antibiotics.
In conclusion, compared to E. coli and K. pneumoniae, MDR P. aeruginosa, MRSA S. aureus and VRE E. faecium certainly present a lower incidence in the clinical setting (no cases in setting n.2). The effective management of these resistant strains and a correct antibiotic therapy based on the resistance's epidemiology represent the most potent clinical approach to enhance the efficiency of antibiotics and to reduce bacteria-associated mortality.

Future Insights
Healthcare-associated infections (HAIs) are one of the most impacting causes of preventable death and disability within the hospitalized population.
Recently, several strategies have been proposed to face this challenge, such as strong prevention or the support of computer-based analyses. The EU's call for projects has also promoted the development of innovative artificial intelligence (AI) solutions to prevent infections inside clinical departments. In particular, an interesting ongoing project (LAO-COONTE project, by Energent S.p.A.) has the objective to develop specific use cases, where data can be used by machine and deep learning models to evaluate the likelihood of infection in clinical departments, in an Italian clinical setting. This approach is really promising; in fact, nowadays, AI acts an important role in different fields from smart manufacturing to the Internet of things, human-computer interaction and medical scenarios, of course. The attention of the scientific community and industry to the AI field is related to the excellent performance achieved in recent years by the so-called artificial neural networks, in particular the deep architectures, in various fields such as text, images and audio [31][32][33]. Nosocomial infections (NIs) are even more preventable, as they represent a biological and social cost for hospitalized patients. The growing availability of computerized patient records in hospitals allows for the improvement of data storage with traditional machine learning methods, which have been shown to outperform deep learning's performance when applied to tabular data.
The objective of the future is to understand how to prevent the causes underlying NIs and to increase safety procedures once patients have been admitted to hospitals.

Conflicts of Interest:
The authors declare no conflict of interest.