Microbial Predominance and Antimicrobial Resistance in a Tertiary Hospital: A Six-year Retrospective study from Outpatients and Patients Visiting the Emergency Department

Objective: To assessing the characteristics of of microbial species and the antimicrobial resistance in a Tertiary Hospital with 49 outpatient clinics and emergency department in Northwestern China, of six years. Methods: A retrospective study was conducted using HIS database of a tertiary hospital between the full-year period of 2013 and 2018. Antimicrobial susceptibility tests were conducted by automated systems and/or the Kary-Bauer disc diffusion method. Data were analyzed using the WHONET 5.6 software. The Cochran–Armitage test was used to study the trends over the period. Results: A total of 19,028 specimens were submitted for the laboratory tests of microbiology. Among 49 units, only Emergency Department and Kidney Transplantation Clinic with the number of submission showed a signicant increase annually (P<0.001). A total of 3,849 non-repetitive isolates were identied, covering more than 200 species, of which gram-positive cocci accounted for 46.4% and gram-negative bacilli 45.3%. The methicillin-resistant rates of S. aureus and S. epidermidis were 25.1% and 74.6%, respectively. The isolates of 60.9% of E. coli and 33.5% of K. pneumonia contained extended spectrum β lactamases. Moreover, there were no Staphylococci and Enterococci resistant to linezolid, vancomycin and tigecycline. In addition, the percentages of E. coli, K. pneumonia and P. aeruginosa isolates resistant to carbapenems were low (0.0%, 1.1% and 18.7%, respectively). Conclusion: Vancomycin, linezolid and tigecycline are among the most effective treatment for outpatients with gram-positive infection. Carbapenems are among the most effective for gram-negative infection. There is no signicant annual increase of common multidrug resistances.


Introduction
The increase of using the broad-spectrum antimicrobial treatments leads to microbial resistant to these treatments in which they were originally sensitive [1]. The increased antimicrobial resistance (AMR) limits treatment options and is considered as a global challenge to public health with increased morbidity, mortality and healthcare cost [1]. To accurately estimate the challenge, the surveillance of AMR pro ling at local, regional and national levels has been employed to understand the global trends on the type of predominant pathogens and on their respective resistance pro ling [2]. In China, the surveillance of AMR is established at the hospital, provincial and national levels. For example, CHINET is one of wellestablished national surveillance networks for AMR [3][4][5][6]. This network recently reported that the frequency of gram-negative bacilli among clinical isolates was over twice higher than that of grampositive cocci [3,4]. Moreover, it is important to recognize that China is a large country with remarkable varieties of microbial occurrences and AMR pro ling. For example, Xiao et al. reported that the percentage of MRSA (Methicillin resistant Staphylococcus aureus) in North Central and South Central China is higher than other regions [7]. The geographic variety is likely associated with the regional differences in socioeconomic development [7]. Therefore, the microbial predominance and AMR pro ling in a speci c region and province is not always consistent with the trend at a national level. Indeed, the surveillance of AMR at the hospital and provincial levels is valuable to complement to the national surveillance.
Moreover, the updated AMR pro ling at hospital and regional levels is particularly important for the physicians to determine adequate empiric antimicrobial therapy in clinical practice. Therefore, in this study, we analyzed the data from a six-year surveillance of AMR at a tertiary hospital and regional level and attempted to provide insights in both surveillance and clinical practice.
The routine surveillance is mainly derived from hospital-acquired infections and most specimens were taken from inpatient service with severe infections [8]. In contrast, community-acquired infections taken from outpatient service are always under-reported [2,8]. Therefore, in our study, we tried to bridge the gap and to study the microbial predominance and AMR pro ling in outpatient clinics and emergency department in a speci c region of Northwestern China and we compared our results with the ndings from CHINET.

Data Source:
This study was conducted at the First A liated Hospital of Xi'an Jiaotong University, which is located in Northwest China and it is a tertiary class A hospital. This hospital provides medical and surgical care to the residents of Shaanxi Province (total population 37.33 million). An institutional review board approved this project (No: XJTU1AF2017LSK-83) and written informed consent was not obtained since the laboratory tests for microbiology are part of standard care and the record about patient's privacy was excluded prior to analysis. Archived laboratory data between the full-year period of 2013 and 2018 were retrieved from HIS database of the hospital for analysis. A total of 19,028 specimens were collected from 49 outpatient clinics and emergency department. A full list of these clinics was provided in Table 1. The types of the specimens include urine, blood, prostatic uid, sputum, pleural effusion and ascites, stool, dialysate, pus, secretion, drainage, cerebrospinal uid and bronchoalveolar lavage uid.

Isolate identi cation:
The clinical specimens were processed based on the recommended microbiology procedures as we previous described [9][10][11][12]. Species identi cation when available was con rmed on colony morphology, conventional biochemical reactions and/or use of an automated system (bioMerieux, Marcy l'Etoile, France).

Statistical analyses:
Data about the isolates and the susceptibility tests were analyzed using the WHONET 5.6 software provided by World Health Organization. The Cochran-Armitage test was used to study the trends of the specimen numbers submitted from different outpatient clinics, different specimen types and the antibiotic resistance percentage over time. This statistical analysis was performed using R 3.6.1 (R Foundation, Vienna, Austria) and a two-sided P-value less than or equal to 0.05 was considered as statistical signi cance.

Results
The specimens submitted between different outpatient clinics and emergency department. From 2013 to 2018, there were 19,028 specimens submitted to the clinical microbiology laboratory from 49 outpatient units. The total number of specimens increased from 2,303 in 2013 to 5,378 in 2018. As we can see from Table 1, the number of specimens submitted from each department showed an increased tendency over the time. The top three units with the largest number of submission were the Emergency Department (30.7%), Urology (18.9%) and Nephrology (16.0%). Moreover, only the Emergency Department and Kidney Transplantation Clinic showed a signi cant increase on the percentage of specimens submitted annually (P < 0.001). Notably, the data from Breast Surgery Clinic were incomplete and therefore they were not included in the analysis. There is no signi cant difference on annual percentage of specimen submission from the Peritoneal Dialysis Clinic. In contrast, all other departments showed a signi cant decrease on annual percentage of specimen submission (Table 1). The different types of specimens submitted for the laboratory tests of microbiology. As showed in Table 2, the number of specimens submitted annually among different types of specimens was increased, consistent with the increase of total number of specimen submission each year in Table 1. Sputum, whole blood and pus showed a signi cant increase on the percentage of specimen submission annually (P < 0.001). In contrast, urine, prostatic uid and other types showed a signi cant decrease on    Table 3 shows the number and percentage (%) of bacterial isolates from the outpatient specimens annually between 2013 and 2018. The percentage of S. aureus (P = 0.002) and E. cloacae (P = 0.033) showed a signi cant increase annually, while S. epidermidis (P < 0.001) and S. haemolyticus (P < 0.001) showed a decrease. Candida albicans (78 isolates) and Aspergillus fumigatus (24 isolates) are the major fungi isolated, followed by Aspergillus avus (9 isolates), Genus of Mucor and Fusarium (9 isolates) and other Candida species (30 isolates). Other rare bacterial isolates included Corynebacterium (109 isolates), Lactobacillus (36 isolates), anaerobic bacteria (13 isolates) as well as Brucella, Eikenella, Actinomyces, Nocardia and Nontuberculous Mycobacterium. Among Corynebacterium, Kroppenstedt Corynebacterium (23 isolates) and Ribbone Corynebacterium (10 isolates) are predominant.  The antimicrobial resistance pro le of the common pathogens. 1) Staphylococcus spp. Our study showed that all of S. aureus isolates in our community are resistance to penicillin (Table 4). However, many S. aureus strains, while resistant to penicillin, remain susceptible to penicillinase-stable penicillins, such as oxacillin. Strains that are resistant to oxacillin and methicillin are historically termed methicillin-resistant S. aureus (MRSA) [13]. Our study showed that the percentage of MRSA isolates is 25.1% during the studied period (Table 4) without signi cant changes annually (Table 6). In contrast, CHINET reported that the prevalence of MRSA was decreased from 69.0% in 2005 to 35.2% in 2017 [3]. The difference between CHINET and our study is likely due to the prevalence of MRSA in community acquired infections is lower over the period of our study, suggesting the in ltration of MRSA isolates from hospitals to community is very limited.  ** indicates that the result is not available. As showed in Table 4, less than 30% of S. aureus isolates were resistance to trimethoprim/sulfamethoxazole (29.5%), gentamicin (23.3%), cipro oxacin (17.6%), levo oxacin (17.6%) and moxi oxacin (6.6%), indicating a minor resistance to uoroquinolones and aminoglycosides.

Discussion
In this study we reported the microbial predominance and AMR pro ling in a tertiary hospital using a sixyear retrospective data from outpatient clinics and emergency department within a speci c region of Northwestern China. We will discuss our ndings in the following ve aspects.
1. The medical value to study the distribution of specimen submitted between different outpatient clinics and emergency department. Based on our literature search, there is little publication to compare the number and percentage of specimens submitted to the clinical microbiology laboratory between different outpatient clinics. Different from other countries, Chinese patients can directly visit the specialty services without the need of the referral from the primary care. Our study showed a trend of increased number on specimen submission in each clinic (Table 1), which is likely associated with the increased population in the communities. The Emergency Department, Urology and Nephrology were among the top three clinics that submitted specimens for microbiological studies. Urinary tract infection has been considered the major reason for outpatients to visit Urology and Nephrology [15] and is also one of the major concerns for emergency room [16]. More important, our study revealed that only two departments including the Emergency Department and Kidney Transplantation Clinic showed a signi cant increase on the percentage of specimen submitted annually (P < 0.001), which indicates that these two departments, particularly the Emergency Department [17], are in the front line of infection control. The increased importance of the Emergency Department is likely associated with enhanced medical functions of the emergency service; there is a trend for the Emergency Department to handle more previous inpatient cares. The increased importance of Kidney Transplantation Department indicated that Kidney Transplantation is becoming a mainstream medical service and repeated infection after kidney transplantation is an urgent medical issue in outpatient clinics [18].
2. Relatively lower ratio of gram-negative isolates in our community acquired infections. Among 3,849 non-repetitive isolates, the ratio of gram-negative vs gram-positive is nearly 1:1 (45.3% vs 46.4%) in our study. In contrast, CHINET report the ratio is nearly 2:1 with approximately 70% of gram-negative bacteria [3,4]. Since CHINET data was largely derived from inpatient service, the relatively lower ratio of gramnegative isolates in our community acquired infections is likely due to the difference between outpatient and inpatient service. The studies from US hospitals report that gram-negative bacteria are more common in cases of ventilator-associated pneumonia and urinary tract infection [19] and are a dominant type of infection at intensive care units [20]. These results suggest a likely association between the higher ratio of gram-negative infection and the inpatient experiencing invasive medical devices and surgical procedure. Indeed, the gram-negative bacteria contain highly e cient mechanism of antibiotic drug resistance [21].
By comparing Tables 4 and 5 in our study, gram-negative bacteria, particularly E. coli, appear more resistant to the antibiotics and require more complicated therapeutic regimen. Overall, relatively lower ratio of gram-negative isolates is likely a general feature in outpatient service.
3. Vancomycin, linezolid and tigecycline are among the most effective treatment for patients with grampositive infection in our community. Vancomycin, linezolid and tigecycline have been used to treat multidrug resistance bacteria in the community [22]. In our study, Staphylococci and Enterococci (Table 4) are 100% sensitive to vancomycin and linezolid, which is in general consistent with the ndings from CHINET. For example, CHINET surveillance in 2018 showed that Staphylococci isolated from blood, urine, lower respiratory tract and cerebrospinal uid were 100% sensitive to vancomycin and linezolid except less than 1% of Staphylococci isolated from blood resist to linezolid [4]. Moreover, less than 5% of E. faecium and less than 1% of E. faecalis were vancomycin-resistant in CHINET surveillance from 2005 to 2017 [3]. These ndings indicate that vancomycin and linezolid remains the most effective treatment for gram-positive infection for both hospital and community settings.
In our study, Staphylococci and Enterococci (Table 4) are also 100% sensitive to tigecycline, an expanded broad-spectrum intravenous glycylcycline antibiotic. This nding supports tigecycline as a reserve for multidrug resistance bacterial infection in community settings.
4. Carbapenems are among the most effective treatment for patients with gram-negative infection in our community. Table 5 showed that higher resistance of gram-negative pathogens, particularly E. coli, to the cephalosporins and quinolones in this study. To overcome the clinical challenge, carbapenems such as imipenem and meropenem have been reserved for these multidrug-resistant bacterial infections [23]. Table 5 also showed that the percentages of E. coli, K. pneumonia and P. aeruginosa isolates resistant to imipenem were 0.0%, 1.1% and 18.7%, respectively. Moreover, the annual trend of resistant percentage of E. coli and K. pneumonia to carbapenems is not signi cant increased (Table 6). These ndings support that carbapenems as a reserve for multidrug resistance bacterial infection in community settings.
5. There is no signi cant increase of multidrug-resistant bacteria observed in our community settings.
Tables 4 and 5 showed higher resistance percentage between certain bacteria and antibiotics, such as methicillin-resistant gram-positive pathogens (MRSA and MRSE) and extended spectrum β lactamases (ESBLs)-resistant gram-negative pathogens. By using ceftriaxone and oxacillin, Table 6 showed the annual change of the resistance rate of multidrug-resistant bacteria in recent years. Should has a signi cant increase of resistance rate been observed, it would require an immediate intervention to investigate the reason. Our results suggest that there is no signi cant increase of multidrug-resistant bacteria, indicating a reasonable prescription of antibiotics in our outpatient and emergency services and a successful physician education on the prevention of antibiotic resistance.
Our study has several limitations. For example, this was a single-center retrospective observational study. We must be cautious when translating our results to other hospitals. However, the quality of clinical sampling procedure and techniques were in general better controlled in a single center than multiple centers. They also allow us to obtain a representative specimen in a large number of clinical departments throughout the hospital and, when repeated periodically, show the main trends. Moreover, a single-center study when appropriately carried out is useful and inexpensive surveilling tool to re ect the regional situation of prevailing microorganisms and their resistance to antimicrobials. Second, outpatient surveillance is an important tool to study the community-acquired infection. We must realize that not all infection of outpatient and patients visiting emergency department derive from the community. This is the reason we only used the term of "community settings" in this article since the patients directly came from the free-living community to seek the medical care in our outpatient and emergency services.

Ethics Approval and Consent to Participants
This project was approved by the Institutional Review Board of the First A liated Hospital of Xi'an Jiaotong Unviersity (No: XJTU1AF2017LSK-83) but the written informed consent can not obtained since the laboratory tests for microbiology are part of standard care and the record about patient's privacy was removed prior to data analysis.