Secondary bacterial infections and antimicrobial resistance in COVID-19: comparative evaluation of pre-pandemic and pandemic-era, a retrospective single center study

Purpose In this study, we aimed to evaluate the epidemiology and antimicrobial resistance (AMR) patterns of bacterial pathogens in COVID-19 patients and to compare the results with control groups from the pre-pandemic and pandemic era. Methods Microbiological database records of all the COVID-19 diagnosed patients in the Ege University Hospital between March 15, 2020, and June 15, 2020, evaluated retrospectively. Patients who acquired secondary bacterial infections (SBIs) and bacterial co-infections were analyzed. Etiology and AMR data of the bacterial infections were collected. Results were also compared to control groups from pre-pandemic and pandemic era data. Results In total, 4859 positive culture results from 3532 patients were analyzed. Fifty-two (3.59%) patients had 78 SBIs and 38 (2.62%) patients had 45 bacterial co-infections among 1447 COVID-19 patients. 22/85 (25.88%) patients died who had bacterial infections. The respiratory culture-positive sample rate was 39.02% among all culture-positive samples in the COVID-19 group. There was a significant decrease in extended-spectrum beta-lactamase (ESBL)-producing Enterobacterales (8.94%) compared to samples from the pre-pandemic (20.76%) and pandemic era (20.74%) (p = 0.001 for both comparisons). Interestingly, Acinetobacter baumannii was the main pathogen in the respiratory infections of COVID-19 patients (9.76%) and the rate was significantly higher than pre-pandemic (3.49%, p < 0.002) and pandemic era control groups (3.11%, p < 0.001). Conclusion Due to the low frequency of SBIs reported during the ongoing pandemic, a more careful and targeted antimicrobial prescription should be taken. While patients with COVID-19 had lower levels of ESBL-producing Enterobacterales, the frequency of multidrug-resistant (MDR) A. baumannii is higher. Supplementary Information The online version contains supplementary material available at 10.1186/s12941-021-00454-7.

annually due to multidrug-resistant (MDR) (resistant to more than three or more antimicrobial categories) infections [2]. It is widely accepted that antimicrobial surveillance is crucial for tackling AMR globally [3]. The current pandemic, a consequence of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) is associated with high mortality, morbidity, and healthcare-related costs [4]. Recent data shows that more than 175 million people have so far been infected and 3.7 million of them died [5]. Secondary bacterial infections (SBIs) result in higher mortality rates in patients with COVID-19 [6]. As previously reported in recent pandemics [7,8], viral infections can also promote the development of bacterial invasive respiratory infections by impairing the immune response, enhancing the destruction of cells and tissue.
In the pandemic era, deaths and hospitalizations have been reported across all age groups, with a higher frequency in the elderly (> 60 years) patients [9]. SBIs have previously been reported in patients with COVID-19 [6,10,11]. It was recently shown that SBIs were observed in 15% of COVID-19 patients, while the SBIs were associated with 50% of all deaths [12]. Therefore, it is crucial to consider SBIs in order to improve the outcome of COVID-19. Consequently, patients with COVID-19 are treated with prophylactic broad-spectrum antibiotics in order to prevent and treat possible SBIs [13][14][15][16]. Recently, other approaches including phage therapeutics have been suggested as possible non-antibiotic treatment options [17].
Hygiene procedures such as hand hygiene, usage of disinfectants, personal protective equipment (PPE) are significantly changed due to the COVID-19 pandemic [18].
In environments where personnel and personal protective equipment (PPE) are insufficient, hygienic conditions may deteriorate, as well as increased use of PPE in interventions may prevent the increase of SBIs and the spread of common resistant bacterial strains. A previous study from Italy has suggested that lack of PPE, and lack of healthcare professionals associated with increased risk of spreading the carbapenem-resistant Klebsiella pneumoniae in the intensive care units (ICUs) [19].
It is important to evaluate the etiology and resistance patterns of SBIs as well as to compare the data with other patients before and during the pandemic. There is scarce data on SBI etiology and antimicrobial susceptibility data [10,16]. However, hitherto published studies have not analyzed the changes in AMR in patients with COVID-19 and controls. In addition, there is no published data from developing countries on SBIs where high AMR is prevalent. The data is crucial for establishing effective antibiotic therapy as well as avoiding unnecessary treatment [20].
In this study, we aimed to evaluate the etiologies of SBIs and co-infections and AMR profiles of the bacterial pathogens causing these infections in patients with COVID-19. Also, our secondary aim was to reveal the differences in the AMR between patients with COVID-19 and other patients from the pre-pandemic and pandemic era.

Study design
This study was performed at Ege University Hospital, Izmir, with a total of 2426 patient beds. The study covers the data between December 15, 2019, and June 15, 2020, which can be considered as the preceding three months, and the first three months of the COVID-19 pandemic in Turkey.

Study population COVID-19 group
During the pandemic peak for SARS-CoV-2, real-time reverse transcription PCR (RT-PCR) assay positive and negative patients with symptoms consistent with COVID-19 were included. The RT-PCR assay negative patients had radiological imaging findings meeting the Ministry of Health's probable case [21] criteria for COVID-19 and therefore were included in the COVID-19 patient group (Additional file 2). SBIs are defined as infections occurring two days or more after patients were admitted to the hospital. Infections occurring not more than two days of hospitalization were defined as co-infections.

Control group
This study included two control groups, i.e., one prepandemic era group with clinical microbiology culture results registered between December 15, 2019-March 15, 2020, and one pandemic era group with microbiology culture results registered between March 15, 2020-June 15, 2020, with no clinical and/or laboratory diagnosis of COVID-19 ( Fig. 1).

Bacterial culture
The investigation of bacterial pathogens in clinical samples was conducted with standard procedures as requested by the attending physician and was evaluated by microbiologists (Additional file 1). Identification and antimicrobial susceptibility testing of the isolates were performed using MALDI-TOF MS (Matrix-Assisted Laser Desorption/Ionization-Time of Flight Mass Spectrometry) (BioMérieux, France) and VITEK 2 (Bio-Mérieux, France) automated systems, respectively. For viridans group streptococci, Haemophilus spp., Corynebacterium striatum, Streptococcus pneumoniae, and Stenotrophomonas maltophilia, the international Kirby Bauer disk diffusion method was used. All the susceptibility results were evaluated in accordance with the EUCAST criteria [22]. When imipenem or meropenem resistance was detected in Enterobacterales strains with the automated system, the result was confirmed by gradient test (BioMérieux, France). Antibiotic MIC values were confirmed by gradient test when isolates were determined resistant to vancomycin, teicoplanin, linezolid, tigecycline with VITEK 2 automated system. The VITEK 2 cefoxitin screen test was used to detect MRSA strains. Only one isolate per patient was studied.

Data collection
Blood, respiratory tract, urinary tract, and other samples such as gastrointestinal tract, tissue, normally sterile fluids cultures sent from patients were evaluated and susceptibility test results were collected from the microbiology laboratory database.

Statistical analysis
Statistical analyses were performed using Microsoft Excel 2016 and IBM SPSS Statistics Version 18.0 (IBM Corp, Armonk, NY). Cross tables were created for categorical variables and chi-square analysis was performed. Categorical variables were shown as numbers and %, numerical variables as median (min., max.). A. baumannii changes between different groups were analyzed with two-tailed Fisher's exact test. Other comparisons (e.g., etiological changes, multidrug-resistant bacteria changes) were done with two-tailed Pearson's chisquared test. A p value < 0.05 was considered statistically significant.

Ethical approval
This study has been approved by the Ege University Medical Research Ethics Committee (20-9T/75) and Turkish Ministry of Health.

Results
In total, 4859 culture-positive samples from 3532 patients were studied. The samples were collected between December 15, 2019, and June 15, 2020. The prepandemic era control group consists of 3034 samples from 2143 patients and the pandemic era control group included 1702 samples from 1304 patients. 1447 COVID-19 diagnosed patients' data were evaluated separately, and 85 of them had 123 bacterial infections.
In total, 78/123 (63.41%) bacterial strains were obtained two days after hospitalization of the patients and described as hospital-acquired infections while 45/123 (36.59%) were community-acquired infections. The most common bacteria was A. baumannii (10, 9.76%) among all respiratory tract samples ( Table 2).  outcomes. Patients receiving invasive mechanical ventilation in other SARS and MERS epidemics developed secondary infections and had higher mortality [24]. In the present study, 2.62% of patients with COVID-19 had bacterial co-infections whereas 3.59% of them had secondary bacterial infections. In total, 22/85 (25.88%) COVID-19 patients with bacterial infections died. Recent studies from Turkey showed that the overall mortality rate in patients with COVID-19 is 4.5% [25,26]. Therefore, it is reasonable to suggest that bacterial infections are related to higher mortality rates in patients with COVID-19. UTI was the most common infection type (45.5%) and followed by RTI (39.02%) which was significantly higher than the two control groups. Previous studies analysing SBIs in patients with COVID-19 are contradictory. He et al. showed that 50% of the patients with COVID-19 had a SBI or carried bacterial pathogens [27]. However, in a recent meta-analysis, it was reported that the SBI rate in COVID-19 patients was between 4.7 and 19.5% and was associated with an increased risk of severe course or fatal outcomes [28,29]. The underlying reason for low bacterial infection rates in the present study is not known. It might be related to several factors including the severity of the disease in patients included prior antimicrobial therapy or stringent local hygiene protocols applied during the pandemic era [30]. The  The most common pathogens in the COVID-19 study group were E. coli (22,17.89%), K. pneumonieae (15,12.2%), A. baumannii (12,9.76%), and S. aureus (11,8.94%). Detection of E. coli (22,17.89%) was significantly lower in patients with COVID-19, compared to the pandemic era control group (447, 26.26%) (p = 0.04). In contrast, detection of A. baumannii in patients with COVID-19 was higher than in the two control groups.

Discussion
In lower respiratory tract infections, the most detected pathogens were Gram-negative bacteria (32, 66.66%), following by Gram-positive bacteria (16,33.33%). In Gramnegative bacteria, the most common isolated strains were A. baumannii (10/32, 31.25%), followed by P. aeruginosa (5/32, 15.63%). Among Gram-positive bacteria, S. aureus (7/16, 43.75%) was the most common isolated strain. Although the distribution of Gram-negative and Grampositive bacteria were similar, a higher A. baumannii occurrence as observed in our COVID-19 study group and needs further evaluation with comprehensive clinical data. A recent study from China showed that the most common bacterial pathogens isolated from respiratory tract samples were Gram-negative bacteria (26, 65%), following by Gram-positive bacteria (14,34.99%). In that study, the most common bacterial pathogens encompassed K. pneumoniae (n = 11), E. faecium (n = 9), followed by A. baumannii (n = 8) [6]. The underlying reason for discrepant results between the present study and the previous report might be associated with differences in colonisation of bacteria types between centers, patients' clinical profile as well as administration of prophylactic antibiotics.
ESBL-producing Enterobacterales were significantly lower in COVID-19 compared to pre-pandemic (p = 0.001) and pandemic era control group (p = 0.002). In a recent study which is reported from Egypt, Gramnegative isolates were mostly ESBL-or carbapenemase-producers which differs from our study [31]. The distinction may be due to the difference in the drug resistance profile between countries and/or local hygiene measures taken for patients with COVID-19. MDR A. baumannii was the most common bacteria (9.76%) in respiratory tract samples (39.02%) from COVID-19 patients, and this rate was significantly higher than that in the pre-pandemic (3.49%) (p = 0.002) and pandemic (3.11%) (p = 0.001) era control groups. In another study from New Jersey, when COVID-19 cases surge, increased carbapenem-resistant A. baumannii (CRAB) counts were reported. CRABs were mostly observed among COVID-19 patients admitted to ICUs and receiving ventilation therapy [32]. In both cases, the increase in intervention needed for COVID-19 patients may have led to a deterioration in hygiene conditions and an increase in the spread of MDR A. baumannii.
The present study has some limitations. First, not all COVID-19 patients were confirmed by PCR to be SARS-CoV-2 positive, and the comprehensive clinical data and disease severity of these patients were not studied. Secondly, a limited number of patients were available in our single-center study. However, we followed the official guidelines for the diagnosis of COVID-19, and it is highly unlikely that the patients with typical COVID-19 radiological findings had other infections. The lack of clinical severity data might be important, but the study focuses on the description of the prevalence of microorganisms and AST results in patients with COVID-19 in general. In addition, the sample size covers a period of 3 months in the pandemic era. The study has the following strengths: First, unlike other studies, we compared both the prevalence of bacterial infections and antimicrobial resistance patterns in patients with COVID-19 and control groups. Also, this study is the first comparative study presented from Turkey which is a developing country with a general high MDR bacteria profile.

Conclusion
In conclusion, the present study shows that SBIs and bacterial coinfections were low in COVID-19 patients; but when present, it causes severe outcomes and is