The study was conducted at the UoGCSH, in central Gondar town, 747 km northwest of Addis Ababa (the capital city of Ethiopia). This hospital serves as a referral center, serving a large population of over seven million individuals residing in its catchment area and surrounding regions (Gobezie et al., 2023). It provides a wide range of diagnostic and treatment services across various medical disciplines. The hospital’s laboratory is organized into specialized sections, including dedicated units for clinical chemistry, hematology, parasitology, Mycobacterium tuberculosis culture, molecular biology, serology, viral load, and microbiology. It also has one main laboratory. The microbiology laboratory plays a vital role in bacteriological diagnostics. This laboratory is equipped to perform a range of procedures, including Gram staining, conventional culture techniques, biochemical tests, and antimicrobial susceptibility tests. The data generated within this laboratory is documented, serving dual purposes: direct patient care management and contributing to broader research initiatives.

An institution-based cross-sectional study was conducted retrospectively from January 2019 to December 2021 at the UoGCSH. The primary objective was to identify and characterize bacterial pathogens responsible for bloodstream infections and evaluate their antimicrobial susceptibility patterns. Blood samples were collected from patients with suspected bloodstream infections. These samples underwent microbiological analysis to isolate and identify the bacterial pathogens. Following isolation, antibiotic susceptibility testing was performed on the identified bacteria to determine their resistance profiles to commonly used antibiotics.

Laboratory records of patients suspected of having BSIs were gathered from the electronic database. These records contain valuable information about demographic, clinical, and blood culture results of the specific bacterial isolates associated with the infections. The study employed a comprehensive sampling approach, including all available laboratory records of patients of all age groups suspected of having BSIs. The inclusion criteria were suspicion of BSIs; complete records including demographic variables, clinical diagnosis, and blood culture results; and presentation at any of the following UoGCSH units: adult inpatient, emergency triage assessment and treatment (ETAT), emergency OPD, gynecology ward, medical ICU, medical OPD, neonate ICU, oncology ward, pediatric ICU, pediatric inpatient, surgical OPD, and surgical ward.

Blood samples were collected and processed according to standard operating procedures. Before blood drawing, the selected venipuncture sites were thoroughly cleansed with 70% alcohol followed by 2% iodine tincture. Approximately 8–10 mL of blood is routinely collected from adults, 1–4 mL from children, and 1 mL from neonates, obtained in two draws taken 30 min apart. Blood samples are then transferred to broth containing 0.025% sodium polyanetholsulfonate at a 1:10 ratio and incubated aerobically at 35–37°C for 5 days. During this 5-day incubation period, the broth media are monitored daily for bacterial growth through visual examination. Broth cultures exhibiting signs of microbial growth are Gram-stained and subsequently sub-cultured onto blood agar, chocolate agar (either in a CO2 incubator or a candle jar), and MacConkey agar plates. These plates are incubated aerobically at 35–37°C for 18–24 h. Furthermore, after 18–24 h of incubation, a blind subculture is performed for all negative blood culture bottles by streaking onto chocolate agar plates. Bacterial pathogens are identified based on macroscopic colony characteristics, including color, size, shape, and texture, as well as Gram staining. Gram-negative bacterial pathogens are identified using a series of conventional biochemical tests, including indole, urease, lysine decarboxylase, triple sugar iron agar, citrate utilization, oxidase, and sulfide indole motility tests. Conversely, Gram-positive bacteria and their hemolytic patterns are identified based on catalase and coagulase tests (Arega et al., 2018; Arega et al., 2017).

Antimicrobial susceptibility testing was performed using the Kirby-Bauer disc diffusion method. Three to five pure colonies of bacteria were selected and transferred to a tube containing 4–5 mL of sterile normal saline. The mixture was gently mixed to form a homogenous suspension. The prepared suspension was incubated at 35–37°C until the turbidity matched a 0.5 McFarland standard. A sterile cotton swab was then dipped into the prepared suspension, and excess fluid was removed by pressing it against the inside wall of the test tube. The suspension was evenly distributed on the surface of Mueller-Hinton agar for non-fastidious bacteria and on Mueller-Hinton agar supplemented with 5% sheep blood for fastidious bacteria. The inoculated plates were left at room temperature for 3–15 min before placing the antibiotic discs. The common antimicrobial agents tested, chosen based on the Clinical Laboratory Standards Institute (CLSI) M100 guidelines 2019 (Clinical and Laboratory Standards Institute, 2019), included penicillin (10 μg), cefoxitin (30 μg), gentamicin (10 μg), ampicillin (10 μg), cefotaxime (30 μg), ceftriaxone (30 μg), vancomycin (30 μg), chloramphenicol (30 μg), tobramycin (10 μg), amoxicillin-clavulanate (20/10 μg), meropenem (10 μg), amikacin (30 μg), ciprofloxacin (5 μg), piperacillin-tazobactam (100/10 μg), trimethoprim-sulfamethoxazole (1.25/23.75 μg), and ceftazidime (30 μg). The plates were then incubated at 35–37°C for 16–18 h, after which the diameters of the zones of inhibition around the disks were measured. The degree of susceptibility of the bacterial isolates was then classified as sensitive, intermediate, or resistant based on the standardized table provided by the CLSI guidelines (Clinical and Laboratory Standards Institute, 2019).

Data were extracted from the electronic health records of the UoGCSH microbiology laboratory by two experienced clinical laboratory professionals. The extraction checklist included patients’ demographic information, such as gender and age, clinical setting (Ward, OPD, ICU, etc.), and admission date. It also captured information on clinical presentations or indications, including fever, hypothermia, hypotension, increased respiratory rate, confusion, suspected pneumonia, indications of complicated urinary tract infection, suspected meningitis, suspected abscess, osteomyelitis, suspected soft tissue infection, suspected abdominal infection, severe malaria, typhoid fever, neonatal infection, indications of other severe infections, and any other relevant indications. Furthermore, the checklist encompassed information about antibiotic therapy administered within 2 weeks before sample collection for laboratory diagnosis, specifying the type of antibiotic used. Furthermore, data on culture media and microbial identification, antibiotic disks used, and susceptibility results were also collected. Bacterial isolates exhibiting intermediate resistance to antimicrobial agents were categorized as resistant strains.

Neonate: A neonate is defined as a newborn infant within the first 28 days of life (Bejitual et al., 2022).

Pediatrics: The pediatric population encompasses individuals from early infancy (above 28 days) through adolescence (up to 18 years).

Adult: Adults are defined as individuals aged 18 years and older.

Fever: A patient was defined as having a fever if their body temperature was 37.5°C to 38°C during admission. This measurement was documented by healthcare providers, alongside information gathered from patients and/or their caregivers regarding any fever episodes in the 48 h preceding admission (Egi and Morita, 2012; Mackowiak et al., 2021).

Hypothermia: Body temperature measured by a reliable method (axillary, oral, tympanic, or rectal) below 36.5°C for neonates (Alebachew Bayih et al., 2019), for pediatric 36.0°C (Nemeth et al., 2021; Rauch et al., 2021; Zhao et al., 2023), and for adults 36.0°C (Savioli et al., 2023; Haverkamp et al., 2018).

Hypotension: Defined based on age-specific systolic blood pressure (SBP) thresholds.

Multidrug resistance: This describes the capability of bacterial isolates to resist the effects of antimicrobial agents from three or more distinct categories or groups (Belew et al., 2023; Alam et al., 2011; Garoy et al., 2021; Bitew et al., 2023; Hailemariam et al., 2021).

Datasets were exported from electronic health records into Microsoft Excel spreadsheets for completeness checking and data cleaning. Subsequently, the cleaned data was transferred to STATA software (StataCorp. 2021. Stata Statistical Software: Release 17. College Station, TX: StataCorp LLC.) for coding and analysis. Descriptive statistics, including frequencies and percentages, were calculated. Furthermore, binary and multivariable logistic regression models were employed to investigate the relationship between BSI and associated factors. All variables exhibiting a p-value of ≤0.25 in the bivariable analysis were included in the multivariable logistic regression analysis. Variables with a p-value of <0.05 in the multivariable logistic regression analysis were considered to indicate statistically significant associations. The study findings are presented in textual descriptions, tables, and graphical formats.

Blood sample collection, culture media preparation, inoculation, incubation, and biochemical tests were conducted according to standard operating procedures for microbiological techniques. The sterility of the culture media was checked by randomly selecting 5% of the prepared media, which was then incubated aerobically at 35–37°C for 24 h. Furthermore, all freshly prepared blood agar, MacConkey agar, and chocolate agar media were regularly checked by inoculating 5 % of them with known standard strains of S. aureus ATCC 25923, S. epidermidis ATCC 19211, S. pneumoniae ATCC 49619, E. coli ATCC 25922, P. mirabilis ATCC 12453, and H. influenzae ATCC 47112. Microbiology experts performed culture media inoculation, colony characterization, and antibiotic susceptibility tests. The authors developed a standardized data collection form containing variables of interest. The accuracy and completeness of the collected data were then assessed using techniques such as checking for missing values and conducting duplicate checks.

Before conducting the study, ethical clearance was obtained from the University of Gondar, Institutional Review Board (Ref: VP/RTT/05/822/2024). Permission was also obtained from the hospital’s clinical director’s office and the head of the microbiology laboratory section. Patient details were removed to ensure confidentiality, and the data were analyzed anonymously. No personal identifiers were used, and the collected data was only accessible to the investigator.

This study included 4,727 patients admitted to UoCGSH between January 2019 and December 2021 who were suspected of BSIs and provided blood samples for diagnosis. Of these patients, 58.1% (2,747/4,727) were male. The study population’s age ranged from 1 day to 96 years. Neonates represented 26.3% (1,244/4,727), pediatric patients 38.5% (1,821/4,727), and adult patients 35.2% (1,662/4,727) of the total participants (Table 1).

Fever was the predominant clinical presentation among patients suspected of BSI at the UoGCSH, observed in 87.6% of cases. A significant proportion of these patients (35.5%) had received antibiotic therapy within the 2 weeks preceding their admission. An elevated respiratory rate was noted in nearly a quarter of the study participants (23.6%). In contrast, several other clinical indicators were reported less frequently, including hypothermia (2.5%), signs of confusion (2.9%), a history of fever within the preceding 48 h (2.5%), and signs of hypotension (2.5%). Specific suspected sources of infection were generally infrequent, with pneumonia being the most prevalent at 13.4%. Other suspected sites, such as complicated urinary tract infections (1.6%), soft tissue infections (1.6%), and abdominal infections (1.5%), were less common. Notably, suspicion of neonatal infection was reported in 4.5% of the neonate subgroup (Table 2).

Of the 4,727 blood specimens analyzed for microbial growth via blood culture, 17.5% (827) showed bacterial growth within the first day of incubation. After further incubation and subculture, the overall bacterial blood culture positivity rate was 14.8% (701 out of 4,727 specimens). These positive cultures comprised 63.5% Gram-negative bacteria (445 out of 701) and 36.5% Gram-positive bacteria (256 out of 701). BSIs were most frequently observed in neonates (40.8%, 286 out of 701) and pediatric patients (40.4%, 283 out of 701). The prevalence of bacterial isolates increased over the three-year study period: 3.8% in 2019, 4.4% in 2020, and 6.6% in 2021.

A total of 27 distinct bacterial isolates were identified. K. pneumoniae was the most frequently isolated pathogen, accounting for 27.1% (203 out of 750) of all isolates, followed by S. aureus at 22.0% (165 out of 750). S. viridans was also prevalent, representing 6.5% (49/750) of the isolates. Non-fermenting Gram-Negative Rods (NFGNR) constituted 4.5% (34/750) of the isolates; the specific species could not be determined using routine biochemical methods. Among the identified Coagulase-Negative Staphylococci (CoNS), only 1.1% (8/701) were considered clinically significant in the diagnosis of endocarditis. The remaining CoNS isolates were likely contaminants and thus deemed clinically irrelevant. Other, less frequently isolated bacteria, each with a single isolation, included Enterobacter aerogenes, P. multocida, S. arizonae, S. typhi, and Serratia spp. (Table 3).

Antimicrobial susceptibility test results of bacterial isolates showed significant variations in the susceptibility of Gram-positive bacteria. Specifically, CoNS isolates demonstrated 100% susceptibility to penicillin. In contrast, S. pneumoniae isolates showed only 20% susceptibility to vancomycin. S. aureus isolates, on the other hand, were predominantly resistant to both penicillin (87.6%) and oxacillin (61.1%). Comparatively, Enterococcus spp. isolates showed a significant level of resistance to vancomycin (57.1%) (Tables 4).

Among Gram-negative bacterial isolates, K. pneumoniae showed high resistance to cefotaxime (94.4%), trimethoprim-sulfamethoxazole (91.9%), ciprofloxacin (53%), and gentamicin (66.9%). In comparison, it showed lower resistance to meropenem (30.7%) and amikacin (22.4%). E. coli, the second most common isolate, also demonstrated significant resistance to ampicillin (72.2%), amoxicillin-clavulanate (84.2%), ciprofloxacin (79%), and trimethoprim-sulfamethoxazole (84.4%). However, E. coli remained largely susceptible to meropenem (7.1%). Concerningly, A. baumannii and E. cloacae displayed considerable resistance to multiple tested antibiotics, including piperacillin-tazobactam (54.1 and 60%, respectively) and, in the case of E. cloacae, cefotaxime (82.4%) (Table 5).

Multidrug resistance was a common characteristic observed across the bacterial isolates, irrespective of their Gram-staining properties. Of the 701 bacterial isolates analyzed, 89.9% (630) exhibited resistance to at least one antimicrobial agent included in the susceptibility tests. Moreover, 60.9% (427) of these isolates displayed resistance to three or more distinct classes of antimicrobial agents. Specifically, K. oxytoca (100%), K. pneumoniae (97.5%), E. coli (96.6%), Citrobacter spp. (85.7%), and Acinetobacter baumannii (81.6%) demonstrated high levels of multidrug resistance, consistent with the finding of resistance to multiple antimicrobial categories (Table 6).

The bivariable and multivariable logistic regression analyses revealed significant associations between specific demographic and clinical factors and BSI. The age groups of study participants were found to have a substantial impact, with study participants less than 1 month old (AOR = 6.8, 95, 95% CI = 2.41–19.09, p < 0.001) and those infants and toddlers aged 1 month to less than 3 years old (AOR = 2.0, 95% CI = 1.54–2.64, p < 0.001) showed significant association with BSIs. Furthermore, patients who visited the emergency OPD (AOR = 3.2, 95% CI = 1.50–6.74, p = 0.003), those admitted to the oncology ward (AOR = 3.0, 95% CI = 1.21–7.17, p = 0.017), and those in the surgical ward (AOR = 3.3, 95% CI = 1.27–8.43, p = 0.014) were identified as being at higher risk for developing BSIs (Table 7).

The study’s limitations include its single-center design, potentially restricting the generalizability of findings. The lack of molecular typing also limits insights into genetic mechanisms driving resistance and pathogen evolution. Future multicenter studies incorporating molecular epidemiology are essential to provide a more comprehensive understanding of BSI dynamics.