Extended-spectrum β-lactamase- producing gram-negative bacterial infections in severely ill COVID-19 patients admitted in a national referral hospital, Kenya

Background Bacterial infections in COVID-19 patients, especially those caused by multidrug-resistant gram-negative strains, are associated with increased morbidity, hospital stay and mortality. However, there is limited data on the epidemiology of extended-spectrum β-lactamase (ESBL)-producing bacteria in COVID-19 patients. Here, we assessed the prevalence and the factors associated with ESBL-producing gram-negative bacterial (GNB) infections among severely ill COVID-19 patients admitted in Kenyatta National Hospital (KNH), Kenya. Methods We adopted a descriptive cross-sectional study design for patients admitted between October 2021 and February 2022, purposively recruiting 120 SARS-CoV- 2 infected participants based on clinical presentation. Demographics and clinical characteristics data were collected using structured questionnaires and case report forms. Clinical samples were collected and analyzed by standard microbiological methods in the KNH Microbiology laboratory and the Centre for Microbiology Research, Kenya Medical Research Institute. Results GNB infections prevalence was 40.8%, majorly caused by ESBL—producers (67.3%) predominated by Klebsiella pneumoniae (45.5%). Generally, 73% of the ESBL producers harboured our target ESBL genes, mainly CTX-M-type (59%, 17/29) in K. pneumoniae (76.9%, 20/26). GNB harbouring TEM-type (83%, 10/12) and SHV-type (100%, 7/7) genes showed ESBLs phenotypes and inhibitor resistance, mainly involving clavulanate, but most of them remained susceptible to tazobactam (60%, 6/10). SHV-type genes carrying ESBL producers showed resistance to both cefotaxime (CTX) and ceftazidime (CAZ) (K. pneumoniae), CAZ (E. coli) or CTX (E. cloacae complex and K. pneumoniae). About 87% (20/23) of isolates encoding CTX-M-type β-lactamases displayed CTX/ceftriaxone (CRO) resistance phenotype. About 42% of isolates with CTX-M-type β-lactamases only hydrolyzed ceftazidime (CAZ). Isolates with OXA-type β-lactamases were resistant to CTX, CAZ, CRO, cefepime and aztreonam. Patients with comorbidities were 10 times more likely to have an ESBL-producing GNB infection (aOR = 9.86, 95%CI 1.30 – 74.63, p = 0.003). Conclusion We report a high prevalence of ESBL-GNB infections in severely ill COVID-19 patients, predominantly due to Klebsiella pneumoniae harbouring CTX-M type ESBL genes. The patient’s underlying comorbidities increased the risk of ESBL-producing GNB infection. In COVID-19 pandemic, enhanced systematic and continuous surveillance of ESBL-producing GNB, strict adherence to infection control measures and antimicrobial stewardship policies are warranted in the current study setting.

Due to the lack of treatment guidelines at the beginning of the COVID-19 pandemic, most patients received broad-spectrum antibiotics [8].Even though the impact of increased antibiotic use during the pandemic is still unclear, there was increased geographical distribution of carbapenemases, plasmid-encoded bacterial enzymes that hydrolyse carbapenem [8][9][10] in Latin America and the Caribbean [8].However, the impact on the epidemiology of extended-spectrum β-lactamases (ESBL), with a similar transmission mechanism to carbapenemases, is unclear.
ESBLs are a group of bacterial enzymes that hydrolyse expanded spectrum β-lactam, thus mediating resistance against penicillins and cephalosporins [11].These enzymes, produced predominantly by GNB, are worrisome because they can spread rapidly among clinical isolates through mobile genetic elements, which frequently co-harbour other non-β-lactam resistance genes, such as colistin [12,13] aminoglycosides [14], and quinolones [15].Surge in ESBL-producing bacterial infections can increase the use of carbapenems, which are among the drugs of last-resort in treatment of multidrug-resistant bacterial infections, posing a serious negative implication in clinical practice.
ESBL-production phenotype is mediated by several ESBL families, such as TEM, SHV, CTX-M, GES, PER, VEB, and BEL [11], with CTX-M-type β-lactamases mostly predominating [11].Typically, beta-lactam combined with inhibitors, such as clavulanic acid, tazobactam or sulbactam, neutralize ESBL activities.Some TEM and SHV variants are resistant to inhibitors and, similar to other ESBLs, show geographical variation based on human mobility [11,16,17].However, data on co-infections with ESBL-producing bacteria among COVID-19 patients in many developing countries, particularly in Sub-Saharan Africa, is limited.Therefore, we assessed the prevalence and risk factors for co-infection with ESBLproducing GNB among severely ill patients admitted in a Kenyan hospital unit dedicated to COVID-19 patients.

Study area, study design and data collection
We conducted this study in the Infectious Disease Unit (IDU), a ward dedicated to COVID-19 patients at Kenyatta National Hospital (KNH), Kenya, between October 2021 and February 2022.We adopted a descriptive crosssectional study design among severely ill patients with confirmed (real-time reverse transcription and quantitative polymerase chain reaction (RT-qPCR) SARS-CoV-2 infection.Selection of severely ill COVID-19 participants was based on the WHO definition of severe COVID-19 illness; defined as, a critical condition where patients exhibit significant oxygen saturation deficits, impaired oxygen exchange in the lungs, rapid and labored breathing, or extensive lung infiltrates, all of which point to a severe respiratory and medical challenge associated with COVID-19 [18,19].This study purposively recruited 120 SARS-CoV-2 infected participants based on patients' clinical presentation suggesting bacteria infection as judged by the treating clinicians, and excluded those who, through their close relatives or legally authorized representative, declined consent to participate.
Data on demographics and clinical characteristics were collected using structured questionnaires and case report forms.Blood samples were collected directly into sterile blood culture bottles (bioMérieux, Marcy l´Etoile, France), observing the standard microbiological operating procedures [20].Nasopharyngeal (NP) and oropharyngeal (OP) swabs (Sigma-Aldrich, India) and tracheal aspirate samples were collected by a licensed personnel into sterile containers, and transported in an ice box to the hospital Microbiology laboratory for immediate analysis.

Bacteria isolation and identification
Bacterial isolation followed the standard microbiological methods [21].We cultured NP/OP swabs and tracheal aspirate samples on sheep blood agar (Oxoid, United Kingdom) and MacConkey (Oxoid, United Kingdom), with an overnight incubation at 37 °C.Blood culture bottles were incubated in BACT/ALERT ® VIRTUO 3D Microbial Detection Systems (bioMérieux, Marcy l'Etoile, France), followed by sub-culture for the positive samples onto chocolate blood agar (CBA) (Oxoid, United Kingdom), sheep blood agar (Oxoid, United Kingdom) and MacConkey (Oxoid, United Kingdom).After subculture, we incubated the plates in ambient air; and 5-10% CO 2 at 37 °C overnight, followed by isolates' identification using VITEK Matrix-assisted laser desorption ionization-time-of-flight mass spectrometry (MALDI-TOF MS) (bioMérieux, Marcy l'Etoile, France).For quality control, we used Escherichia coli ATCC 8739.All the GNB isolates were transported to the Centre for Microbiology Research, Kenya Medical Research Institute (CMR-KEMRI) laboratories for further analysis.

Screening for ESBL production
We screened the isolates for ESBL production using the Double Disk Synergy Test (DDST) [22].A 0.5 McFarlandequivalent suspension of bacterial isolate were plated on Mueller-Hinton Agar (MHA) and allowed to air dry for 3 min.Antibiotic disks, including cefotaxime (30 µg), ceftazidime (30 µg), and amoxicillin/clavulanic acid (20 µg/10 µg), were added at a 30 mm radius to radius distance and incubated overnight in ambient air at 37 ℃.An inhibition zone around the cefotaxime and/or ceftazidime that increased towards the β-lactam inhibitor was considered an ESBL producer.We used K. pneumoniae ATCC 700603 and E. coli ATCC 25922 for quality control.
ESBL production was also confirmed by the Phenotypic Confirmatory Disc Diffusion Test (PCDDT) [22].Briefly, 0.5 McFarland of bacterium suspension was inoculated on MHA plate (Oxoid, United Kingdom) and allowed to air dry for 3 min.Antibiotics disks, including cefotaxime (30 µg), ceftazidime (30 µg), cefotaxime/clavulanic acid (30 µg/10 µg), and ceftazidime/clavulanic acid (30 µg/10 µg), were placed on the inoculated MHA plate, at a 30 mm (mm) radius to radius distance.The plates were incubated overnight in ambient air at 37 ℃.ESBL production was confirmed by observing an isolate with a > 5 mm-clear zone formed between the third-generation cephalosporin and the β-lactam inhibitor.Klebsiella pneumoniae ATCC 700603 and Escherichia coli ATCC 25922 were the control strains.

Detection of ESBL resistance genes
ESBL producers were PCR-screened for SHV-, TEM-, OXA-1, and CTX-M-type ESBL genes using primers in Table 1.We extracted bacterial DNA using the heat lysis method [20] and followed the PCR protocol described by Kiiru et al. [22].Briefly, 2 μl of DNA was added to 22 μl of PCR master mix (Bio-Rad Laboratories, Hercules, USA) with the target ESBL gene primers and loaded to a thermal cycler (Bio-Rad Laboratories, Hercules, USA), programmed as follows: denaturation at 95 °C for 6 min, annealing at 55 °C for 2 min, and final extension at 72 °C for 10 min.The amplification products were separated by gel electrophoresis (1.5% agar rose gel), stained with SYBR green dye and captured images by the Bio-Rad imaging system (Bio-Rad Laboratories, Hercules, USA).Klebsiella pneumoniae ATCC 700603 and Escherichia coli ATCC 25922 were the quality control organisms.
Colistin susceptibility testing was done by the Simple Disk diffusion method [25].Using Escherichia coli ATCC 25922 and P. aeruginosa ATCC 27853 as quality control strains, we placed a 10 mg colistin disk on a 0.5 McFarland-equivalent bacterium suspension plated on modified Mueller-Hinton agar 30% (5.1 g/L) (Oxoid, United Kingdom), followed by an overnight incubation at 35 ℃ in 5% CO 2 .Minimum Inhibitory Concentrations (MICs), determined by broth microdilution following CLSI guidelines [21], were used to interpret the resultant inhibition zones.We defined multidrug-resistant organisms (MDRs) by resistance to three or more antibiotic classes [26].

Data analysis and presentation
Statistical analysis was two-sided using STATA version 16.After describing continuous data in means and medians and categorical data in frequencies and percentages, we presented the data in tables and figures.Crude odds ratio (cOR) was analyzed using binary logic regression, with variables giving p-values ≤ 0.2 further computed by multiple regression analysis for adjusted odd ratio (aOR).The level of statistical significance was set at p-value ≤ 0.05, with a 95% Confidence Interval (CI), and statistically significant associations indicated in bold (Table 4).

Discussion
In this study, we screened 49 gram negative bacterial (GNB) isolates for ESBL production.Of these, 67.3% were ESBL -producers, predominated by K. pneumoniae (30.6%).Contrary to our findings, Lemenand and colleagues reported a decreasing proportion of ESBL among E. coli infections (2.9%) during the COVID-19 pandemic in France [27].The study by Lemenand et   we targeted all GNB in severely ill COVID-19 patients confirmed by real-time reverse transcription and quantitative polymerase chain reaction (RT-qPCR), and admitted in critical care unit.This could possibly explain the high prevalence of ESBL producer isolates in our study.
In the current study, the prevalence of ESBL-producing GNB infections among COVID-19 patients was higher than that reported in non-COVID-19 patients in East African (42%) and Kenya (47%) [29].In a recent study among Kenyan children at the point of hospital discharge, the prevalence of ESBL-producing E. coli was 44.3% [30].Together, these reports suggest a higher prevalence of ESBL-producing GNB in severely ill COVID-19 patients admitted in ICU in our setting.Klebsiella pneumoniae (30.6%) was the predominant ESBL producer among GNB isolates from severely ill COVID-19 patients.Even though data on ESBL-producing bacteria in COVID-19 patients is limited, in the general population, E. coli and K. pneumoniae [31][32][33][34] seems to be the most common ESBL producers.Our finding may infer similarity in ESBL-producing bacteria profiles among COVID-19 and non-COVID patients.In  [37], however, information on ESBL gene carriage among ESBL producing GNB that cause infections in COVID-19 patients is limited.Before the year 2000, SHV-and TEM-type enzymes were the most predominant ESBLs worldwide [38] but have since been outnumbered by CTX-M ESBLs in non-COVID19 patients [39,40,11,34].Therefore, our findings suggest a similar ESBL gene carriage among bacterial isolates from COVID-19 patients and the general population.Clinically, the CTX-M-producing bacterial infections are treated using carbapenems, thus promoting the spread of potentially untreatable carbapenemase-producing bacterial infections [38].
Severely ill COVID-19 patients with comorbidities were at higher risk of infection by ESBL-producing bacteria.Greco and others found that COVID-19 patients with comorbidities, such as diabetes mellitus and hypertension, were at increased risk of co-infections in Italy [54].In a multi-centre study by He and others on clinical characteristics of COVID-19 patients with clinically diagnosed bacterial co-infection, patients with cardiovascular comorbidities were more likely to have clinically diagnosed bacterial co-infection [55].In the current study, the most common comorbidities were cancer (17%), kidney disease (16%), diabetes (14.9%), hypertension (11.7%), haematological disorders (7.4%) and HIV/AIDS (6.4%).
This study has some limitations.As a single centre study, the data obtained may not be generalizable to other hospitals within our locality and therefore, a larger study is recommended to determine this epidemiology against the general patient population.Additionally, the purposive sampling may have subjected it to selection bias, and due to resource constrains, we were unable to elucidate the molecular variants of the ESBL genes detected.However, this study highlights the need for systematic and continuous surveillance of multidrug-resistant bacteria among SARS-CoV-2 infected persons in the hospital to inform AMR prevention interventions in line with national and global action plans.

Conclusion
We report a high prevalence of ESBL-GNB infections in severely ill COVID-19 patients, predominantly due to Klebsiella pneumoniae harbouring CTX-M type ESBL genes.The patient's underlying comorbidities increased the risk of ESBL-producing GNB infection.In COVID-19 pandemic, enhanced systematic and continuous surveillance of ESBL-producing GNB, strict adherence to infection control measures and antimicrobial stewardship policies are warranted in the current study setting.

Table 1
Primer combinations used for detection of ESBL-encoding genes

Table 2
Demographic and clinical characteristics of patients with GNB infections al. focused only on single bacteria, E. coli, and their data might not

Table 4
Factors associated with ESBL-producing GNB infections among COVID-19 patients admitted in KNH-IDU Statistically significant associations are indicated in bold cOR crude odds ratio, aOR adjusted odds ratio, ESBLs extended spectrum beta-lactamases, Ref Reference, SARS-CoV-2 severe acute respiratory syndrome coronavirus 2, CI confidence interval,