Antimicrobial resistance profiles and associated factors of Acinetobacter and Pseudomonas aeruginosa nosocomial infection among patients admitted at Dessie comprehensive specialized Hospital, North-East Ethiopia. A cross-sectional study

Introduction Hospital admitted patients are at increased risk of nosocomial infections (NIs) with multi-drug resistant (MDR) pathogens which are prevalent in the hospital environment. Pseudomonas aeruginosa (P. aeruginosa) and Acinetobacter baumannii (A. baumannii) are common causes of NIs worldwide. The objective of this study is to determine antimicrobial resistance profiles and associated factors of Acinetobacter spp and P. aeruginosa NIs among hospitalized patients. Methods A cross-sectional study was conducted at Dessie comprehensive specialized hospital, North-East Ethiopia, from February 1 to April 30, 2020. A total of 254 patients who were suspected of the bloodstream, urinary tract, or surgical site nosocomial infections were enrolled consecutively. Socio-demographic and other variables of interest were collected using a structured questionnaire. Specimens were collected and processed following standard microbiological procedures. Antimicrobial susceptibility was determined using the Kirby-Bauer disk diffusion method following Clinical and Laboratory Standards Institute guidelines. Data were analyzed with SPSS version 23 and p-value < 0.05 was considered statistically significant. Results Overall, 13% of patients had nosocomial Acinetobacter spp and/or P. aeruginosa infections. The culture positivity rate was 16(6.3%) for Acinetobacter spp and 18(7.1%) for P. aeruginosa. Patients admitted in the surgical ward (Adjusted odds ratio (AOR):10.66;95% confidence interval (CI):1.22–93.23), pediatric ward (AOR:14.37;95%CI:1.4–148.5), intensive care unit (AOR:41.93;95%CI:4.7–374.7) and orthopedics (AOR:52.21;95%CI:7.5–365) were significantly at risk to develop NIs compared to patients admitted in the medical ward. Patients who took more than two antimicrobial types at admission were 94% (AOR:0.06; 95% CI:0.004–0.84) times more protected from NIs compared to those who did not take any antimicrobial. About 81% of Acinetobacter spp and 83% of P. aeruginosa isolates were MDR. Amikacin and meropenem showed promising activity against Acinetobacter spp and P. aeruginosa isolates. Conclusion The high prevalence of MDR Acinetobacter spp and P. aeruginosa nosocomial isolates enforce treating of patients with NIs based on antimicrobial susceptibility testing results.


Introduction
Nosocomial infection (NI) is defined as healthcare-associated infection (HCAI) developing at least 48-72 hr after admission or up to 3 days after discharge and was not incubating on admission [1]. According to the world health organization (WHO) fact sheet report, NI is a major global health problem affecting a hundred million patients annually [2] and rapidly increasing especially in developing countries [3]. The point prevalence of IN ranged from 5.7% to 19.1% and 3.5% to 12% in Low-and Middle-Income Countries (LMICs) and developed nations, respectively [2]. In Ethiopia, according to a recent meta-analysis, HCAI affected 16.96% of patients. Surgical site infection, urinary tract infection, bloodstream infection, and respiratory tract infection were the commonest types of HCAI in the country [4]. Besides associated mortality and morbidity, NI increases the occurrence of antimicrobial resistance [5], hospitalization times, and healthcare cost for the patients as well as the health care systems [6].
Bacterial pathogens could be acquired either exogenously or endogenously and are transmitted through direct or indirect contact between patients, healthcare workers, visitors, and contaminated objects or different hospital environmental sources. Risk factors of NIs are so diverse and depend on infection sites. The use of invasive devices [7,8] and prior use of broadspectrum antimicrobials increase the risk of infection [7]. Longer hospitalization [9], intravascular catheterization [10], age of the patient [7,11], gender, surgery after admission, urinary catheterization [8], and type of hospital settings [11] were also identified as significant risk factors for nosocomial infections.
Epidemiological studies revealed that different species of both gram-positive and gram-negative bacteria are associated with NIs [12][13][14][15]. Among gram-negative pathogens, A. baumannii and P. aeruginosa have got significant importance associated with NI [12]. Recently in Nepal, A. baumannii (44.0%), and P. aeruginosa (40.1%) were the most common gram-negative bacteria isolated among hospitalized patients [2]. According to a recent meta-analysis report, the pooled proportion of A. baumannii associated NIs was 15.3% with the highest burden (20.9%) in the intensive care unit (ICU). Authors conclude that A. baumannii infection accounted for a significant risk to the health of hospitalized patients in Europe, Eastern Mediterranean regions, and Africa [16].
Both P. aeruginosa and Acinetobacter spp. are ubiquitous in hospital environments and frequently associated with NIs that are difficult to treat. This is because these organisms possess inherent drug resistance mechanisms (e.g. constitutive expression of AmpC beta-lactamases and efflux pumps, low permeability of the outer membrane, etc.). They are also gifted to acquire additional resistance mechanisms to multiple classes of antimicrobials (e.g. beta-lactams, aminoglycosides, and fluoroquinolones) through horizontal gene transfer [2,17,18]. Such behavior is responsible to resist multiple antimicrobials at the same time and limits the choice of antibiotics for the treatment of infections due to these microorganisms.
Infection due to multi-drug resistant (MDR) bacteria is challenging for treatment and resulted in poor clinical outcomes, prolonged hospitalization, and elevated health care costs [19,20]. MDR Acinetobacter species and P. aeruginosa are recognized in the WHO lists as dangerous pathogens. The WHO has also ranked carbapenem-resistant A. baumannii as a critical priority pathogen that needs drug research and development [21]. Infections due to these microorganisms are a global health threat and are considered as the number one critical priority pathogen for which new therapeutics are urgently required [22]. In Ethiopia, scholars reported that more than half of isolated gram-negative rods were extended-spectrum beta-lactamase (ESBL) producers and 25% of isolates were resistant to carbapenem [13]. Motbainor et al [23] and Alemayehu et al [14] recently identified that all nosocomial isolates of A. baumannii and P. aeruginosa were MDR [23], and infection due to these bacteria is alarming in the country since it is difficult and impossible to manage with currently available antibiotics.
Despite the global increasing incidences of NIs associated with Acinetobacter spp. and P. aeruginosa, and the increasing frequency of antibiotic-resistant strains, there is limited data about the nosocomial importance of these bacterial species in Ethiopia [23]. In most clinical settings of the country, clinicians lack real and up-to-date information about the nosocomial importance of these organisms. This is mainly due to poor laboratory facilities and infrastructure to isolate these species and screening of drug-resistant strains. And also, factors associated with NIs due to Acinetobacter spp. and P. aeruginosa are not well identified in Ethiopia.
Moreover, the type of bacterial species colonizing the hospital environment changes over time and varies among hospitals within the country and different locations in the same hospital. It is also known that bacterial antibiotic resistance is a dynamic process so that resistance patterns observed in the past might not reflect the current situation. As a result, investigation of the causative agents and their current antimicrobial susceptibility profile is essential to optimize the management and reduction of the rate of NIs. The aim of this study is, therefore, to determine the prevalence and antimicrobial resistance pattern, and to identify factors associated with Acinetobacter spp. and P. aeruginosa infection among admitted patients suspected of nosocomial infection at Dessie comprehensive specialized hospital, Ethiopia.

Patient selection
The source population was all hospital admitted patients during the study period. Patients who were admitted for more than 48 hrs and developed clinical evidence of nosocomial wound (surgical site) infections, bloodstream infection, and urinary tract infection were included in the study. The criteria of the European Centre for Disease Prevention and Control were used to define NIs and to select eligible study participants [24]. Patients who had either purulent drainage, pain, localized swelling, redness, or heat in the skin, subcutaneous tissue, deep soft tissue, organ or spaces, and one positive culture for Acinetobacter spp and P. aeruginosa after 48 hr of operation were considered as nosocomial surgical site infection. Patients who had either fever (> 38˚C), urgency, frequency, dysuria, or suprapubic tenderness with no other recognized cause but has �10 2 CFU/milliliters (ml) and � 10 5 CFU/ml of urine culture for catheterized and non-catheterized patients, respectively after 48 hr of admission were considered as nosocomial urinary tract infection. On the other hand, patients who had either fever (> 38˚C), chills, or hypotension and one positive blood culture for A. baumannii and P. aeruginosa after 48 hr of admission were considered as nosocomial bloodstream infection and included as a study participants. Moreover, attending internists and surgeons were consulted for their decision if we were not able to select based on the above criteria. However, those patients who were unable to give clinical samples due to different conditions were excluded from the study.

Sample size and sampling procedures
The minimum sample size was determined using the single population proportion formula: N = z 2 p (1-p)/ d 2 , where N = minimum sample size; Z = standard normal distribution value at 95% CI which is 1.96; P = 8.4% prevalence of A. baumannii and P. aeruginosa infection taken from the previous study [23]; d = the margin of error taken as 4%. Accordingly, the estimated sample size was 185. However, 254 admitted patients who were clinically suspected for the nosocomial bloodstream, wound, and urinary tract infections were enrolled consecutively during the study period.
Data and specimen collection. Patients admitted in different wards including ICU were followed prospectively and assessed for the development of the surgical site, urinary tract, and bloodstream NI by internists and surgeons as per the European Centre for Disease Prevention and Control criteria [24]. Information on socio-demographic variables as well as potential risk factors of IN was collected from each patient by face-to-face interview using a structured questionnaire. For children, the respective guardian/caregiver was interviewed. Clinical data related to chronic diseases, hospitalization, admission ward type, and previous antimicrobial taking history were collected by reviewing the patient's medical record and consulting the attending physician and surgeon. Clinical specimens such as urine, blood, and wound swabs were collected as soon as NI was suspected.
Blood sample collection and processing. Venous blood samples of 10 ml, 5ml, and 2ml were aseptically collected from adults, children, and neonates, respectively. The samples were collected in duplicate for each patient from a different site within a 30-minute difference [25]. The collected blood samples were inoculated directly to 5-10 ml Tryptic Soya broth medium (Oxoid, England) and transported immediately to Amhara public health institute's microbiology laboratory, Dessie branch. The inoculated broth was incubated overnight aerobically at 35-37˚C and inspected for bacterial growth daily. Gram stain was done from blood culture bottles that showed growth and sub-cultured onto blood agar (BA) and MacConkey (MAC) agar (Oxoid, England) plates. Inoculated plates were incubated aerobically at 35-37˚C and examined for bacterial growth after 24 hr. However, those blood culture bottles which did not show growth were continuously monitored for the potential growth of pathogens until 7 days, and plates with no growth after 7 days of incubation were reported as negative [25].
Wound sample collection and processing. Wound/pus specimens were collected aseptically by sterile cotton swabs dipped in normal saline using Levine's technique [26]. All collected wound specimens were labeled and transported to Amhara public health institute's microbiology laboratory within 30 minutes by placing the swab into the sterile test tubes having 0.5 ml of sterile normal saline solution. The specimen was inoculated on blood and Mac-Conkey agar plate and incubated aerobically at 35-37˚C for 24-48 hrs. All positive cultures were identified by colony characteristics on the respective media [26,27].
Urine sample collection and processing. Urinary tract infection suspected patients (noncatheterized) were instructed to collect 10 ml midstream clean-catch urine samples using a sterile wide-mouth container. The same amount of urine sample was transferred to a sterile container after cleansing the outlet of a catheter of catheterized patients. The collected urine was immediately transported to Amhara public health institute's microbiology laboratory and inoculated on blood and MacConkey agar plate using the calibrated loop that measurers about 1μL. All inoculated agar plates were incubated aerobically at 35-37˚C for 24-48 hr and inspected for bacterial growth. Colonies on blood agar were counted using a colony counter and checked for significant bacteriuria. Cultures from catheterized and non-catheterized patients that grew �10 2 CFU/ml and 10 5 CFU/ ml, respectively were taken as significant bacteriuria and processed further [28].

Identification of bacterial isolates
For heterogeneous colonies, sub-culturing of individual distinct colonies was performed to ensure pure cultures. All suspected P. aeruginosa and Acinetobacter isolates were identified phenotypically by manual standard microbiological methods based on colonial morphology, pigmentation of the colony, and cell morphology. Further characterization of the isolates was performed using biochemical tests including catalase, oxidase, urease, indole, citrate utilization, lysine decarboxylation, motility, glucose, and lactose fermentation tests [25].

Antimicrobial susceptibility test
A modified Kirby Bauer disk diffusion method was used to test each isolate for in vitro antimicrobial susceptibility pattern based on the 2020 Clinical and Laboratory Standards Institute (CLSI) criteria [29]. About 3-5 freshly grown pure colonies were taken and a homogeneous inoculums suspension was prepared using sterile normal saline and adjusted to 0.5 McFarland standard's turbidity. The sterile cotton swab was dipped into the prepared suspension, rotated several times, and swabbed over the entire surface of the Mueller Hinton agar plates. Antimicrobial impregnated paper disks were placed on the inoculated plates. Both Acinetobacter spp. and P. aeruginosa isolates were tested against ceftazidime (30μg), ciprofloxacin (5μg), gentamicin (10μg), meropenem (10μg), piperacillin-tazobactam (100/10μg) and amikacin (30μg).
Additionally, Acinetobacter isolates were tested against cefotaxime (30μg) and trimethoprimsulfamethoxazole (1.25/23.75μg), while P. aeruginosa isolates were tested against aztreonam (30μg). P. aeruginosa isolate inoculated plates with respective antimicrobials were incubated aerobically at 35˚C for 16 to 18 hr, whereas the incubation period was 20-24hr for Acinetobacter isolates. The zone of inhibition was measured by calibrated ruler and interpreted as sensitive, intermediate, or resistant based on the CLSI guidelines [29]. However, for the description of the MDR profile, intermediate results were categorized as resistant. An isolate was considered MDR if it is resistant to at least one agent in three or more antimicrobial categories [30].

Quality assurance
The questionnaire was pre-tested for its quality and validity before data collection. Data collectors were trained on the data collection procedure and interview techniques. The standard operating procedure was strictly adhered and quality control measures were implemented throughout the whole laboratory process. Culture media was prepared according to the manufacturers' instructions. The sterility of prepared culture media was checked by incubating 5% of the batch at 35-37˚C overnight before using it. Sample collection as well as bacterial isolation, identification, and antimicrobial susceptibility testing were performed in strict aseptic conditions. Media performance and potency of antimicrobial discs were tested using American Type Culture Collection (ATCC) standard reference strains (P. aeruginosa ATCC27853 and E. coli ATCC 25922). A 0.5% McFarland turbidity standard was used to standardize bacterial inoculums suspension for the antimicrobial susceptibility testing.

Data analysis
Collected data were checked for their completeness and entered into Epi-Data version 3, then exported to Statistical Package for Social Sciences (SPSS) version 23 for further data cleaning and analysis. Descriptive statistics were computed and presented using words and tables. Bivariable and multivariable analyses were computed to identify factors independently influencing the occurrence of dependent variables. The odds ratio and 95% confidence interval were calculated to measure the strength of the association. A P-value of <0.05 was considered statistically significant.
Ethics approval and consent to participate. The protocol was ethically approved by Wollo University, College of Medicine, and Health science ethical review committee. To access the data, permission was obtained from Dessie Comprehensive Specialized Hospital. We followed all chains of command to 'get support from legally authorized representatives for data collection. After clarifying the objective of the study, written informed consent was obtained from adult study participants; moreover, the ascent was obtained from guardians or caregivers of study participants whose age is below 18 years. Confidentiality of the result was maintained anonymously and used only for the study purpose. Positive findings were communicated to the attending clinician for appropriate treatment.

Socio-demographic and clinical characteristics of study participants
A total of 254 patients with clinical evidence of nosocomial infection were enrolled in this study. The majority of the study participants were males, 145 (57.1%). The age of the study participants ranged from 0 to 90 years with a mean age of 29.8 years with a standard deviation of ± 24.3. The majority (35%) of participants were found in the age group of 31-60 years. One hundred fifty-seven (61.8%) study participants were living in rural settings. About 33% of study participants were illiterate in educational status. Similarly, 25.95% of study participants were children under age to have specific occupations followed by farmers, 60(23.62) ( Table 1).
The majority of study participants were from medical wards (33.9%) followed by ICU (22.8%). About 3/4 th of study participants had no history of previous admission and 68.5% of participants do not have an underlying chronic disease. However, 72.44% of participants had taken different types of antimicrobials during admission and 178(70.07%) participants utilized intravascular catheters during their medication as shown in Table 1.

Rate of Acinetobacter spp. and P. aeruginosa nosocomial infection
From a total of 254 collected clinical samples, Acinetobacter spp. and P.aeruginosa were isolated from wound/pus and blood. All of the urine specimens did not show significant bacterial growth (colony counts less than 10 2 or 10 5 CFU for catheterized and non-catheterized patients, respectively). A total of 34 bacterial isolates were isolated with an overall culture positivity rate of 13% (33/254). The proportion of bloodstream infection and wound infection was 13.33% (10/75) and 23.23% (23/99), respectively. From the total of 34 isolates, 70.6% were isolated from patients who developed wounds, with an equal number of Acinetobacter spp. and P. aeruginosa. The proportion of isolated P. aeruginosa (17.6%) was more among patients with bloodstream infection compared to Acinetobacter spp. (11.8%). All Acinetobacter spp. and about 78% of P. aeruginosa isolates were from patients with intravenous catheterization as shown in Table 2.

Factors associated with nosocomial infection
Exploring independent risk factors of NIs showed that type of admission ward, number of antimicrobials taken at admission, and having chronic diseases/infections were independent predictors of NI due to either of Acinetobacter spp. and/or P. were statically significantly at risk of NI compared to admitted patients who had no underlying chronic disease. On the other hand, patients who took three and more antimicrobial types during admission were 94% times more protected from NI as compared to patients who did not take any antimicrobial as shown in Table 3.

Antimicrobial susceptibility profiles
The antimicrobial susceptibility profiles of the isolated Acinetobacter spp. and P. aeruginosa are shown in Table 4. Of the total isolates, 85.3% and 73.5% were resistant to ceftazidime and piperacillin-tazobactam respectively, whereas 11.8% of isolates were resistant to amikacin. Acinetobacter spp. showed more resistance rate to cefotaxime (93.8%), piperacillin-tazobactam, and ceftazidime (87.5% for each), but the least resistance rate was observed for amikacin (12.5%) followed by ciprofloxacin (18.8%). Similarly, P. aeruginosa isolates were more resistant to ceftazidime (83.3%) and aztreonam (77.8%); however, 88.9% and 83.3% of P. aeruginosa isolates were susceptible to amikacin and meropenem, respectively. Multidrug-resistance profiles of isolates. The overall MDR rate of isolated bacteria was 82.35%. About 81% of Acinetobacter spp. and 83% of P. aeruginosa isolates showed MDR. Among MDR isolates one and four isolates of each Acinetobacter spp and P. aeruginosa were resistant against antimicrobials from 6 and 5 different categories, respectively (Table 5).

Discussion
Even though NI is a global health threat, it is alarmingly increasing in low-income countries like Ethiopia with a national prevalence rate of 16.96% [4]. Different bacterial species are associated with NI [11,13,31,32], and the high rate of drug-resistant nosocomial bacteria poses a significant threat to the infected patients, communities, and health care providers [13]. A. baumannii and P. aeruginosa are known to develop resistance to most antibiotics. NIs due to MDR A. baumannii and P. aeruginosa is a critical health problem to hospitalized patients [23].

PLOS ONE
observed high prevalence of combined Acinetobacter and P. aeruginosa associated NI might be due to the commonly observed overcrowding of patients, poor infrastructure of the hospital, and poor implementation of infection control measures especially hand hygiene practices and decontamination of the hospital environment. The rate of NIs was significantly higher among patients with underlying chronic diseases like HIV and diabetes as shown in Table 3. This can be explained because patients with chronic underlying diseases, besides immune suppression, could visit health facilities frequently and they are potentially at risk to be contaminated with bacteria circulating in the health facilities as a result develop nosocomial infection even with low virulent pathogens. This is evidenced from a previous study that revealed Acinetobacter spp NIs was significantly associated with immunosuppression [41]. Our result also showed that patients admitted in the surgical ward, pediatric, ICU, and orthopedics were significantly at risk to develop NIs (Table 3) which is consistent with a previous study conducted in France [42]. Acinetobacter-associated NIs is found significantly associated with admission to the ICU [41]. A recent systematic review and meta-analysis revealed that the rate of NIs was more in the ICU followed by the pediatric ward, surgical ward, and obstetrics

PLOS ONE
ward [4]. This might be linked with poor infection prevention practices observed in some health facilities [43] and the high rate of exposure of patients to nosocomial pathogens from the hospital environment including health care professionals. But, our study hospital's infection prevention practice level is not well known and needs immediate investigation. Patients who took more than three antimicrobial types at the time of admission were 94% times more protective from nosocomial infection as compared to patients not taking antimicrobials at all. In contrast, Uwingabiye et.al indicated that previous exposure to antibiotic polytherapy was a risk for ICU acquired A. baumannii infection but the classes of taken antibiotics are not specified [37]. Playford et.al also showed that previous exposure of patients to antibiotics is a significant risk factor for A. baumannii infection [44]. On the other hand, taking active antipseudomonal antibiotics is described as protective [45], but taking inactive antibiotics has been identified as a risk for NIs [46]. In contrast, Hoang et.al found no significant association with inactive antibiotic support and P. aeruginosa infection, but there is an association with colonization [42]. In our result, the observed protective role of previous antibiotic polytherapy might be due to the incorporation of active antibiotics and their synergistic effect. However, we are not able to describe the types of antibiotics taken by patients and this could be a limitation of this study. Thus, the role of number and type of antibiotic therapy for the prevention of NI warranted further investigation.
In the present study, resistance was observed to different antibiotics. Acinetobacter isolates were highly resistant to cefotaxime (93.8%) and piperacillin-tazobactam (87.5%). Similarly, a 90% resistance rate was observed against piperacillin-tazobactam in Vietnam [47], China [48], and Turkey [49]. In contrast, a low resistance rate was reported against piperacillin-tazobactam in Uganda (55%) [36] and Iran (24%) [34]. Moreover, 87.5% of isolated Acinetobacter species was resistant to ceftazidime which is in agreement with previous reports from Ethiopia (77.8%) [23], Uganda (80%) [38], Vietnam (90%) [47], China (92.6) and Turkey (86.7%) [49]. But, our result is higher than reports from Iran (15%) [34] and Uganda (45%) [36]. Another recent study conducted in Iran also reported a higher prevalence (96.6%) of ceftazidime resistant A. baumannii [50]. Similarly, 83.3% and 77.8% of P. aeruginosa isolates were resistant to ceftazidime and aztreonam respectively which is comparable to previous reports with a ceftazidime resistance rate of 100% in Ethiopia [23], 69% in Uganda [36], 77.7% in India [39] and 72.4% in Vietnam [47]. In contrast, less number of aztreonam (40%) and ceftazidime (50%) resistant P. aeruginosa isolates were reported in Uganda [36] and Iran [34] respectively. The observed differences might be due to variations in resistance screening methods and antimicrobial prescription policy and frequency. Furthermore, the variation could be due to genetic alteration of isolates caused by unnecessarily prescribed antibiotics in different countries around the world [51]. In the current study, the highest level of resistance against cefotaxime, ceftazidime, piperacillin-tazobactam, and aztreonam might be linked with excessive and inappropriate use of these antibiotics in the study area since there is no clear antibiotic policy and controlling mechanisms of antimicrobial usage in Ethiopia.
In this study, MDR was observed in 28 isolates (82.4%), which is extensively high. The proportion of MDR isolates of Acinetobacter spp. and P. aeruginosa was 81.3% and 83.3% respectively. Previously in Ethiopia, scholars reported that all nosocomial isolates of A. baumannii and P. aeruginosa were MDR [14,23]. The currently observed proportion of MDR P. aeruginosa isolates was in agreement with previous reports from Uganda (81%) [36] and India (84.7%) [39], but higher than Egypt (52%) [40] and Northeast of Iran (16.5%) [50]. And also the proportion of MDR isolates of Acinetobacter spp. was higher than reports from Uganda (62%) [36] but comparable to reports from northeast Iran (74.75%) [50]. Differences in the selected panel of antimicrobial agents could be the possible explanations for variations in the prevalence of MDR between the present study and other similar studies carried out elsewhere. Moreover, the observed high rate of MDR isolates could be associated with their biological natures. They have been known to express different antimicrobial resisting mechanisms including the production of aminoglycoside modifying enzymes, ESBLs, carbapenemase, topoisomerases as well as outer membrane proteins as penicillin-binding proteins [2,17,18].
Our study revealed a low level of amikacin-resistant Acinetobacter (12.5%) and P. aeruginosa (11.1) isolates. Previously in Iran [34] and Uganda [36], a low proportion of amikacinresistant Acinetobacter isolates (<25%) were reported. In contrast, Ö zdemir et.al in Turkey reported that Acinetobacter isolates resistant to amikacin was 86% [49]. In Egypt, Mahmoud et.al found amikacin is effective against nosocomial P. aeruginosa isolates [40], but more proportion of amikacin-resistant isolates were reported in Ugandan (60%) [38] and Vietnam (65.5%) [47]. In the current study, meropenem is most effective against P. aeruginosa isolates with a resistance rate of 16.7%, which is supported by a study conducted by Sharma et.al [33]. However, 45.5% and 86% of previous isolates were resistant to meropenem in Ethiopia [23] and Vietnam [47] respectively. About 40% of Acinetobacter isolates were resistant to meropenem. Our result is in line with previous reports in Uganda (28-50%) [36,38] and Turkey (53.3%) [49], but lower than previous reports from Vietnam (90%) [47] and China (93.6%) [48]. The currently observed relative effectiveness of meropenem and amikacin against Acinetobacter and P. aeruginosa might be due to the low level of prescription practice of these antimicrobials since they are used as the last treatment option for serious infections and comparatively more expensive in Ethiopia. Infections caused by MDR Acinetobacter and P. aeruginosa isolates could be treated with either meropenem or amikacin and more strongly using a combination of these antimicrobials. Recent literature also showed promising activity of carbapenem subclass of β-lactams against drug-resistant Acinetobacter spp and P. aeruginosa [52].
There are some limitations of the current study. This study assesses the point prevalence of NI in which a single sample is collected at a point in time. Further, patients with a problem of the respiratory system were not included due to patient quarantine and fear of coronavirus transmission. Secondly, being a cross-sectional study, hospital-acquired infections that arose after discharge were not detected due to lack of follow-up. These could underestimate the prevalence of NI as well as the isolation rate of investigated organisms. Third, we could not conduct a molecular analysis to identify bacterial species especially Acinetobacter species and this is a significant weakness of the study. Finally, the sample size for the study was humble and did not have adequate power to detect smaller effect sizes. Thus, these results should be interpreted and utilized cautiously.

Conclusions
Our results showed an overall prevalence of 13.3% nosocomial infection with either Acinetobacter spp. or P. aeruginosa. Underlying chronic diseases such as HIV and diabetes were significant associated factors for nosocomial infection. Admission in the surgical, Paediatric, ICU, and Orthopaedics wards were a significant factor for nosocomial Acinetobacter species and P. aeruginosa infection. Even though, amikacin and meropenem showed promising activity against Acinetobacter spp. and P. aeruginosa isolates, the proportion of MDR isolates of Acinetobacter spp and P. aeruginosa was 81.3% and 83.3% respectively. It is also concluded that treatment of NIs should be guided by antimicrobial susceptibility testing.