Healthcare-Associated Bloodstream Infections Due to Multidrug-Resistant Acinetobacter baumannii in COVID-19 Intensive Care Unit: A Single-Center Retrospective Study

Healthcare-associated infections are an emerging cause of morbidity and mortality in COVID-19 intensive care units (ICUs) worldwide, especially those caused by multidrug-resistant (MDR) pathogens. The objectives of this study were to assess the incidence of bloodstream infections (BSIs) among critically ill COVID-19 patients and to analyze the characteristics of healthcare-associated BSIs due to MDR Acinetobacter baumannii in an COVID-19 ICU. A single-center retrospective study was conducted at a tertiary hospital during a 5-month period. The detection of carbapenemase genes was performed by PCR and genetic relatedness by pulsed-field gel electrophoresis (PFGE) and multilocus-sequence typing. A total of 193 episodes were registered in 176 COVID-19 ICU patients, with an incidence of 25/1000 patient-days at risk. A. baumannii was the most common etiological agent (40.3%), with a resistance to carbapenems of 100%. The blaOXA-23 gene was detected in ST2 isolates while the blaOXA-24 was ST636-specific. PFGE revealed a homogeneous genetic background of the isolates. The clonal spread of OXA-23-positive A. baumannii is responsible for the high prevalence of MDR A. baumannii BSIs in our COVID-19 ICU. Further surveillance of resistance trends and mechanisms is needed along with changes in behavior to improve the implementation of infection control and the rational use of antibiotics.


Introduction
For almost three years now, the coronavirus disease 2019 (COVID-19) pandemic caused by severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) has been a leading health problem in countries worldwide. By the beginning of 2023, more than 750 million cases of COVID-19 and over 6.5 million deaths had been reported worldwide [1]. Although most people infected with SARS-CoV-2 develop a mild to moderate disease, some progress to life-threatening pneumonia that warrants hospitalization with up to approximately 16% requiring admission to an intensive care unit (ICU) [2,3]. It is known that ICU patients are susceptible to healthcare-associated infections (HAIs), especially ICU-acquired bloodstream

Setting
The Dubrava University Hospital is one of the largest hospitals in the Republic of Croatia, with a capacity of 650 patient beds. More than 25,000 patients are treated in the inpatient departments annually, while more than 1,500,000 health services are provided to more than 400,000 patients in the outpatient department. In March 2020, the hospital was converted into a primary respiratory intensive care center (PRIC) for the treatment of patients with severe COVID-19. Prior to the COVID-19 pandemic, the hospital had an ICU capacity of 35 beds in three ICU subunits. In a short period of time, that number gradually increased to 85 ICU beds distributed in six ICU subunits. Due to organizational changes, the microbiology laboratory was closed and microbiological diagnosis was performed at another hospital.

Definitions and Data Collection
COVID-19 diagnosis was defined by the typical clinical presentation in the presence of a positive real-time polymerase chain reaction (RT-PCR) for SARS-CoV-2 on a nasopharyngeal swab. An ICU-acquired BSI was defined by at least one positive blood culture for bacterial or fungal microorganisms in a patient with systemic signs of infection, diagnosed >48 h after ICU admission. The isolation of the same microorganism from the bloodstream within 14 days was not considered a novel episode [17]. For coagulase-negative staphylococci (CoNS) and other common skin contaminants, at least two sets of blood cultures positive for the same pathogen were necessary to define a clinically significant BSI [18]. A polymicrobial bloodstream infection was defined by the isolation of two or more microorganisms from a single blood culture set, or from different blood culture sets obtained within 48 h [19,20].
Patients' data were retrospectively extracted from computerized hospital databases. The following information was reviewed: demographic parameters (age in years and gender), comorbid conditions (cardiomyopathy, peripheral artery disease, chronic obstructive pulmonary disease (COPD), chronic renal insufficiency, diabetes mellitus, and neurologic disease), COVID-19 characteristics at ICU admission (the duration of the disease prior to admission, the presence of bilateral pneumonia, PaO 2 /FiO 2 ) and outcome, including fatal outcome in the ICU. The severity of the clinical condition was determined using Sequential Organ Failure Assessment (SOFA), Acute Physiology and Chronic Health Evaluation II (APACHE II), and Simplified Acute Physiology Score II (SAPS II) scores calculated at the time of ICU admission. Furthermore, we analyzed the immunomodulatory and antimicrobial therapy that patients received for COVID-19 treatment, the duration of antimicrobial therapy prior to BSI, as well as the targeted therapy for A. baumannii BSI.

Laboratory Procedures and Statistical Analysis
In general, two sets of blood cultures were taken whenever clinically indicated. In patients with central venous catheters, one set was drawn from a central venous catheter and one from a peripheral venipuncture site. A blood culture set consisted of one aerobic and one anaerobic bottle. Blood cultures were collected aseptically and then inoculated in an automated BacT/Alert system (bioMérieux, Marcy l'Étoile, France). Blood cultures were continuously processed on a 24 h per day, 7 days per week schedule and a preliminary report based on a blood culture Gram-stain was reported to the responsible clinician on a "24/7" basis. Identification of blood culture isolates was based on routine laboratory techniques, which included Gram-staining, biochemical tests, and an automated VITEK-2 system. Antimicrobial susceptibility testing was performed by a disk diffusion test for ampicillin-sulbactam, imipenem, meropenem, gentamicin, amikacin, ciprofloxacin, and trimethoprim-sulfamethoxazole, and by the microbroth dilution method for the detection of the minimal inhibitory concentration (MIC) for colistin (Mikrolatest ® MIC Colistin, Erba Lachema s.r.o., Brno, Czech Republic) and for confirmation, if necessary, an additional test was performed (MIC-Strip Colistin, Bruker Daltonics GmbH & Co., KG Bremen, Germany). The interpretation of antimicrobial susceptibility was performed according to the European Committee on Antimicrobial Susceptibility Testing (EUCAST), except for ampicillin-sulbactam and A. baumannii, in which case the interpretation according to the Clinical Laboratory Standards Institute (CLSI) was applied [21,22]. Susceptibility testing for cefiderocol was not performed since it was not available in Croatia during the studied period.

Molecular Epidemiology
Pulsed-field gel electrophoresis (PFGE) of ApaI digested genomic DNA was used in order to assess the genetic relatedness of all the A. baumannii isolates included in this study. PFGE was performed using the CHEF-III System (Bio-Rad Laboratories, Hercules, CA, USA) according to the previously described protocol [26]. PFGE band pattern similarity calculation and clustering were performed using the dice coefficient (1.5% tolerance) and unweighted pair group method with arithmetic mean (BioNumerics v7.6, Sint-Martens-Latem, Belgium). The threshold of ≥85% similarity was used for the genetic cluster assignment. Multilocus sequence typing (MLST) was performed on representative isolates according to the "Pasteur scheme" protocol described on the A. baumannii MLST web site (https://pubmlst.org/primers-used-mlst-acinetobacter-baumannii-complex-pasteurscheme; accessed on 1 March 2021) [27]. All the isolates were assigned to International Clone Lineages (IC) using the sequence-based typing (SBT) of bla OXA-51 [28].

Statistical Analysis
The incidence rate of ICU-acquired BSIs in the study population was calculated as the number of BSI episodes per 1000 patient-days at risk (defined as the cumulative days of stay elapsed from 48 h after ICU admission to death or discharge from the ICU). The demographic and clinical data are recorded as absolute numbers, percentages, or median values-all calculated using Microsoft Excel (2016, Redmond, WA, USA).

Results
During the study period, a total of 193 episodes of ICU-acquired BSIs were registered in 176 patients, with an incidence rate of 25 episodes per 1000 patient-days at risk. A significant proportion (73/193, 37%) of the ICU-acquired BSIs were polymicrobial.
The most common etiological agents of healthcare-associated BSIs in 176 COVID-19 patients (193 episodes) in the COVID-19 intensive care unit at the Dubrava University Hospital during the second wave of COVID-19 pandemic are shown in Figure 1.
Among the S. aureus isolates, 20/24 (83%) were methicillin-resistant S. aureus (MRSA). Other Gram-positive bacteria that were isolated were Corynebacterium species, Listeria monocytogenes and group G beta-hemolytic streptococci, each in a percentage of less than 2%. Candida species accounted for 1% of all isolates.
The most common etiological agents of healthcare-associated BSIs in 176 COVID-19 patients (193 episodes) in the COVID-19 intensive care unit at the Dubrava University Hospital during the second wave of COVID-19 pandemic are shown in Figure 1.

Characteristics of Patients
Out of a total of 176 patients, 106 (60.2%) were infected with MDR A. baumannii. The characteristics of the 106 patients with healthcare-associated bloodstream infections due to MDR A. baumannii in COVID-19 ICU are summarized in Table 1
The median SOFA score was 4, the APACHE II score was 13, and the SAPS II score was 30.5. At the ICU admission, 99 (93.4%) of the patients had bilateral pneumonia with a median PaO 2 /FiO 2 of 67.9.
Regarding the treatment of severe COVID-19, 65 (61.3%) of the patients received immunomodulatory therapy, mainly dexamethasone or methylprednisolone, while 21 (19.8%) patients received antivirals, mainly remdesivir. A significant number (99/106, 93.4%) of the patients had antibiotic exposure in the 30 days preceding a positive blood culture. They received at least one dose of the following agents: ceftriaxone, meropenem, piperacillin/tazobactam, imipenem, vancomycin, colistin, azithromycin, cefepime, linezolid, amoxicillin/clavulanic acid, ampicillin/sulbactam, clindamycin, amikacin, gentamicin, intravenous Fosfomycin, or metronidazole. The most prescribed antibiotics were ceftriaxone (in 57/106, 53.8%), meropenem (in 24/106, 22.6%), piperacillin/tazobactam (in 18/106, 16.9%), and amoxicillin/clavulanic acid (in 12/106, 11.3%). Sixty (56.6%) patients had received more than one antibiotic prior to a BSI. Subsequently, antibiotics were changed based on the blood culture report and clinical improvement. For the targeted treatment of a BSI, 47 (44.3%) of the patients received colistin as monotherapy, while in 29 (27.3%) of the patients, colistin was combined with another antibiotic, depending on the result of the antimicrobial susceptibility testing. The antibiotics most frequently combined with colistin were meropenem, ampicillin/sulbactam, and linezolid. Furthermore, nine patients received more than one antibiotic in addition to colistin. Only six (5.7%) patients received therapy that did not include colistin: they were treated with ampicillin/sulbactam in monotherapy or in combination with linezolid for presumed polymicrobial infection.
Only 15 (14.2%) of the patients were discharged from the hospital while 91 (85.8%) had a fatal outcome during their treatment in the ICU.

Antimicrobial Resistance Pattern
The antimicrobial resistance pattern of the MDR A. baumannii isolates is shown in Table 2. All the A. baumannii isolates were resistant to carbapenems (imipenem and meropenem) and to ciprofloxacin. Resistance to gentamicin was registered in 111 (99.1%) of the patients, to amikacin in 100 (89.3%), and to trimethoprim-sulfamethoxazole in 43 (38.4%). There was a low incidence of resistance to ampicillin/sulbactam (11/112, 9.8%) and colistin (4/105, 3.8%).

Carbapenem Resistance Determinants
The oxacillinase gene bla OXA-51 was detected in all the isolates included in this study. The detection of bla OXA-23 and bla OXA-24 was sequence-type-specific, resulting in bla OXA-23 gene detection in all ST2 isolates (83%), while bla OXA-24 was detected only in ST636 (17%). The presence of class B metallo-β-lactamases was not detected in any of the A. baumannii isolates included in this study. The ISAba1 genetic element was detected in all the ST2 isolates upstream of the bla OXA-23 gene.

Molecular Epidemiology Results
PFGE analysis revealed the homogeneous genetic background of the isolate. Isolates were delineated according to the three main genetic clusters (A-C) (≥85% similarity) and eight singleton PFGE profiles. Most of the isolates analyzed belonged to cluster A (n = 45, 63%), followed by cluster B (n = 12, 17%), and cluster C (n = 7, 10%), respectively. Genetic clusters A and C corresponded to ST2, while cluster B corresponded to ST636 by MLST. All isolates were designated as international clone 2 (IC2).

Hospital Environment and Infection Control
We encountered several problems that the hospital was facing at the time. In a short period, due to the COVID-19 pandemic and the need for additional ICU beds, the number of ICU beds increased from 35 in three ICU subunits to 85 in six ICU subunits (pop-up ICUs). An additional problem was the need to transfer patients with pre-existing MDR infections from other institutions (e.g., surgical, neurological, internal medicine, and hematological ICUs) to our hospital due to COVID-19. ICU beds were placed too close to each other, and the environment was insufficiently cleaned and disinfected at that time. The overcrowding of the ICUs made the isolation of patients with MDR pathogens very difficult. Due to the labor shortage and overload of work in the ICUs, healthcare workers were transferred from other hospitals and departments to our ICUs. The correct use of personal protective equipment has been a particular challenge. Hand hygiene was inadequate; there was the widespread use of double or triple gloving and the decontamination of gloves with alcohol.

Discussion
In the present study, we found an incidence rate of 25/1000 patient-days at risk for healthcare-associated BSIs in COVID-19 ICU, with a high prevalence of resistant Gramnegative microorganisms as causative agents, particularly MDR A. baumannii (40.3%). In addition to the high incidence of A. baumannii, the other prevalent Gram-negatives were ESBL and OXA-48 carbapenemase producing K. pneumoniae, while high percentages of MRSA and VRE were observed among the Gram-positives. In comparison to the data found in this study, during the same period prior to the COVID-19 pandemic (from 1 October 2019 to 28 February 2020), the incidence rate of healthcare-associated BSIs was much lower (5/1000 patient-days at risk) while neither OXA-48 producing Enterobacteriales nor MRSA and VRE isolates were confirmed among healthcare-associated BSIs in our hospital (unpublished work).
Similar observations were described in previous reports, with a prevalence of Gramnegative microorganisms of up to 82.8% [29][30][31][32]. Conversely, in a study by Giacobbe et al. [33] and a few other studies [20,34], an increased prevalence of Gram-positive microorganisms with a predominance of Enterococcus spp. and S. aureus was reported. This disparity in prevalence may be a consequence of different patient settings, duration of hospital stays, or local diversity in endemically present pathogens as well as local antimicrobial policies. Furthermore, in the aforementioned studies, data were collected from COVID-19 wards within a multipurpose hospital, while the Dubrava Clinical Hospital was exclusively a COVID-19 center during the study period.
A possible explanation for the unexpectedly high frequency of BSIs due to MDR A. baumannii could be the excessive usage of broad-spectrum antimicrobial agents as initial empirical therapy in patients with severe COVID-19. In our study, 93.4% of the patients received initial empirical antibiotic therapy prior to the BSI episode. Furthermore, 56.6% of our patients received a combination of two or more antibiotics upon admission to the ICU.
In a recent systematic review by Chedid et al., the rate of antibiotic use among COVID-19 hospitalized patients was 74%, mainly in an empirical setting. However, this study also included patients with mild and moderate COVID-19, not only ICU patients, as in our study [35]. He et al. reported a significant positive association between the use of a combination of two or more antibiotics in cases of COVID-19 and nosocomial infection [36]. It is possible that the excessive usage of antibiotics contributed to the outbreak of MDR A.
Another additional factor contributing to the high incidence of healthcare-associated BSIs due to MDR A. baumannii could be the endemical presence of MDR A. baumannii strains in Croatian hospitals. According to the Croatian Committee for Antibiotic Resistance Surveillance of the Croatian Academy of Medical Science, the rate of carbapenem resistance in A. baumannii isolates exceeds 90% in many Croatian hospitals. Resistance to carbapenems in A. baumannii has rapidly increased from less than 10% in 2008 to more than 60% in 2011, reaching 90% in 2020 in Croatia [10]. Similar situations were observed in other neighboring countries, such as Italy, Hungary, Serbia, Romania, Bulgaria, and Greece where carbapenem resistance exceeds 50% [15,40,41].
Regarding the antimicrobial treatment of MDR A. baumannii infections, successful options are limited due to the high rates of resistance to first-line antibiotics, such as betalactam antibiotics, carbapenems, and fluoroquinolones. Therefore, second-line agents, including polymyxins (colistin) and tetracycline derivatives (minocycline and tigecycline), are recommended [42,43]. Furthermore, a combination of ampicillin-sulbactam was also found to be effective against severe infections caused by MDR A. baumannii [44]. Even a colistinglycopeptide combination has been shown to be synergistic against MDR Gram-negatives, especially A. baumannii, and was shown to be a protective factor for mortality if administered for ≥5 days [45]. All our patients received either colistin or ampicillin-sulbactam alone or together immediately after a clinical microbiologist reported A. baumannii as a causative agent of BSI.
Nevertheless, the findings of our study revealed a disturbingly high mortality rate (85.8%) among COVID-19 patients with MDR A. baumannii BSI. A study conducted in China on a similar group of patients, although without COVID-19, reported a mortality rate of 69.4% [46]. In addition, other studies have reported 21-50% mortality in COVID-19 patients with associated BSI [20,33,34]. On the other hand, in a recent study by Russo et al., the 30-day mortality among COVID-19 ICU patients with associated MDR A. baumanni infection was 81% [6]. Falagas et al. also found infection or colonization with MDR A. baumannii to be associated with higher mortality [47]. The data above are in concordance with our findings, suggesting that clinicians should be particularly alert in the case of BSI caused by A. baumannii during the course of COVID-19 pneumonia, which is associated with high mortality.
In our study, we found that resistance to carbapenems is based on the production of carbapenem-hydrolyzing class D oxacillinases: bla OXA-23 and bla OXA-24 . Furthermore, an ISAba1 genetic element was detected in all the ST2 isolates upstream of bla OXA-23 . These mechanisms of resistance were previously documented worldwide [14][15][16] and seem to be the most prevalent in our BSI isolates. In Croatia, carbapenem resistance in A. baumannii associated with the upregulation of the bla OXA-51 gene by ISAba1 was previously reported [48].
In the present study, the MLST results correlated with PFGE clustering. The homogenous genetic background of the analyzed isolate collection was dominated by a single clone, indicating that the outbreaks were mediated through the spread of the ST2 CRAB lineage during the study timeline in our hospital. Various studies have reported the worldwide distribution of A. baumannii ST2/ICII with the bla OXA-23 gene [14,49], while ST636 bla OXA24 was less frequently reported [15,50]. MDR IC2 has been a well-established endemic clone in Croatia since 2009 [51]. The first clinical isolates of A. baumannii with reduced sensitivity to carbapenems were registered from 2002 to 2007. A study presenting the genotyping results of those isolates demonstrated the presence of the European clone 1 (later IC1) [48]. In 2009, there was an epidemic spread of a new clone belonging to IC2, which is the predominant clone in most Croatian hospitals today [16,52,53].
The abrupt increase in the numbers of critically ill COVID-19 patients admitted to our ICU during the study period led to the neglect of usual infection control practices that had been previously well established. Focus on protecting healthcare workers from the new and relatively unknown COVID-19 has sometimes obscured the importance of personal protective equipment as part of basic infection control measures. The data obtained from the literature also suggest that the COVID-19 pandemic was associated with a less effective implementation of infection control measures among healthcare workers [54,55]. At the beginning of the pandemic, healthcare workers were mainly focused on self-protection rather than the prevention of the cross-transmission of MDR pathogens among patients [55]. Another study conducted in Hong Kong during an outbreak of severe acute respiratory syndrome (SARS) observed an increase in MRSA acquisition in the ICU, suggesting that the excessive usage of gloves may have contributed to reduced compliance with hand hygiene among healthcare workers [56]. It is, therefore, possible that the inadequate implementation of infection control measures played a significant role in the episodes of cross-contamination that eventually increased the number of BSIs in our ICU during the COVID-19 pandemic.
Our study has several limitations. It is a single-center study conducted in a setting with a high prevalence of MDR pathogens, thus limiting the generalizability of our results. Other mechanisms of resistance, apart from carbapenemases, such as porin loss or the overexpression of efflux pumps that could contribute to carbapenem resistance, were not analyzed in this study. The lack of comparative data makes it difficult to ascertain the extent to which organizational changes contributed to the increase in the frequency of BSIs in our COVID-19 ICU.

Conclusions
The results of our study suggest that the clonal spread of OXA-23 positive A. baumannii was responsible for the high prevalence of MDR A. baumannii BSIs in the antibiotic-rich COVID-19 ICU environment in our hospital. Further investigations are needed to determine whether the early prescribing of antibiotics in cases of severe COVID-19 is justified, with emphasis on the potential spread of MDR pathogens and secondary healthcare-associated BSIs. In the future, surveillance of resistance trends and mechanisms is needed to streamline research and innovations in the development of new antimicrobials. From our perspective, effective class D beta-lactamase inhibitors are urgently needed to combat severe MDR A. baumannii infections. Moreover, a major focus should be on education, as well as behavioral changes in the area of antimicrobial stewardship and infection prevention and control.