Risk factors for neonatal multidrug-resistant Gram-negative bacilli infection in a neonatal intensive care unit in South China: a 5-year clinical analysis Authors: Powei Tang, Xiaohong Chen, Fei Xiao, Yilin Hu, Jindou Hao, Jia Tian, Wanqu Liu, Yu Zhang, Peihui Liu Published: 2026 ABSTRACT BackgroundMultidrug-resistant (MDR) Gram-negative bacilli (MDR-GNB) infections have emerged as a public health concern. This study investigated the clinical characteristics and risk factors of MDR-GNB infections in a representative neonatal intensive care unit (NICU) in China.MethodsData on neonates with positive MDR-GNB cultures between January 2017 and May 2022 at one hospital were collected. Neonate information, including sex, premature delivery, gestational age, high-risk factors before delivery, delivery mode, invasive operation, premature rupture of membranes, routine blood test results, and outcomes, were collected.ResultsA total of 425 MDR-GNB isolates were detected among 242 neonates. Escherichia coli and Klebsiella pneumoniae were the predominant organisms identified, accounting for 78.4% and 15.3% of the 425 isolates, respectively. In addition, 217 isolates of extended-spectrum beta-lactamase (ESBL)-producing Enterobacterales were identified. Among the 242 neonates, 131 (54.1%) exhibited colonization, while 111 (45.9%) had active infections. Sixty-seven neonates had sepsis, and 55 of these developed sepsis due to infection with MDR-GNB. Low birth weight (< 1500 g) was identified as a significant risk factor for infection (p = 0.006, OR: 3.338, 95%CI: 1.418–7.858). Maternal fever before delivery (p = 0.022, OR 2.679, 95%CI 1.153–6.221) and low birth weight (< 1500 g) (p = 0.029, OR: 2.768, 95%CI: 1.107–6.918) were risk factors for sepsis.ConclusionsE. coli and K. pneumoniae were the predominant MDR-GNB strains detected in neonates in the NICU. Low birth weight and maternal fever before delivery were risk factors for sepsis among NICU patients with MDR-GNB infections.Clinical trial numberNot applicable.Supplementary InformationThe online version contains supplementary material available at 10.1186/s12887-026-07025-9. Background Infections caused by multidrug-resistant (MDR) organisms (MDROs) have emerged as a major health challenge in intensive care units (ICUs) in recent years [1]. This issue can be attributed to antibiotic abuse and MDRO-related factors, including reduced drug permeability, biofilm formation, enzymatic inactivation, target modification, and active efflux pumps [2, 3]. According to a prospective cohort study, among 1660 Gram-negative bloodstream isolates from adult patients, 30% exhibited MDR [4]. MDROs have become increasingly prevalent, resulting in heightened treatment challenges and detrimental outcomes such as sepsis, increased mortality, and prolonged hospitalizations, particularly among neonates [4–6]. Infants hospitalized in the neonatal ICU (NICU) face an elevated risk of infection due to their immature defense mechanisms. Risk factors for MDRO infection in neonates include extremely low birth weight, premature delivery, perinatal complications, prolonged hospitalization, various potential chronic diseases, and antibiotic abuse [7]. Furthermore, colonization by MDROs, especially MDR Gram-negative bacilli (MDR-GNB), is involved in the pathogenesis of sepsis among hospitalized infants [8–12]. A previous study demonstrated an 88.9% mortality rate associated with Gram-negative sepsis [13]. Nonetheless, effective treatments for MDR-GNB infections remain elusive. In addition, studies on infection with MDR-GNB in NICUs are limited, with clinical features and epidemiological patterns varying across geographical regions [6, 12, 14]. Therefore, investigating the characteristics of MDR-GNB infections in neonates will provide essential insights for developing preventive and therapeutic strategies against MDR-GNB infections and sepsis in NICUs. Hence, the primary objective of this study was to investigate the prevalence and risk factors associated with MDR-GNB infections within the NICU of one institution. Methods Patients and study design The data were collected retrospectively from all neonates with positive MDR-GNB cultures (regardless of clinical presentation) admitted to the NICU of Shenzhen Maternal and Child Health Hospital from January 2017 to May 2022. Subsequently, based on a comprehensive clinical assessment, the neonates were classified as either having infection or colonization. The neonatal department of the hospital is the only National Key Clinical Specialty in southern China and Guangdong Province’s high-level clinical key specialty. The NICU has over 110 beds and an average of 1,327 admissions annually from 2017 to 2022. This study included neonates admitted to the NICU of our hospital during the study period. Neonates with early- and late-onset infections were included. Neonates with incomplete cultures, negative culture results, or cultures that were identified without MDR-GNB were excluded. This study was approved by the ethics committee of Shenzhen Maternal and Child Health Hospital (approval No.: LLYJ2021-049-018). Informed consent from the neonates’ legal representatives was deemed unnecessary for this study, as it was retrospective and relied solely on microbiological monitoring data obtained from samples collected during ongoing medical activities. All patients admitted to the NICU routinely undergo blood culture and skin swab sampling. During the stay in the NICU, as per routine practice, rectal swabs were used to collect the post-birth secretions, and the samples were only taken immediately after birth; no further tests were conducted thereafter if the results were negative. Other cultures can be performed based on clinical conditions; for example, gastric juice cultures can be obtained in patients with nasogastric tubes, and sputum cultures can be performed in patients requiring suction or intubation. If neonates showed signs of infection, various samples, including blood, urine, feces, sputum, umbilical secretion, abdominal effusion, cerebrospinal fluid, and end blood of a peripherally inserted central catheter (PICC) catheter, or other samples (e.g., skin swab), were collected for colonization detection. MDR was defined as acquired resistance to at least one antimicrobial agent in three or more categories [15]. MDR-GNB refers to Gram-negative bacteria that are resistant to three or more classes of antibiotics, or to three or more antibiotics within a single class [16]. Carbapenem-resistant Enterobacterales (CRE) refers to Enterobacterales that are resistant to one or more carbapenem antibiotics [17]. MDR-GNB culture and drug sensitivity test results were collected for all neonates. Neonates were categorized into two groups (the colonization group and the infection group) based on their clinical presentation and signs/symptoms. In the infection group, neonates exhibited positive clinical test results and symptomatic indications of bacterial infection, including fever, respiratory failure, feeding difficulties, bradycardia, vomiting, jaundice, abdominal distention, and diarrhea [18, 19]. Conversely, the neonates displaying MDR-GNB in samples but lacking clinical symptoms were classified into the colonization group. Presently, there are no universally accepted definitive criteria for diagnosing neonatal sepsis. In this study, sepsis was operationally defined as a positive blood culture result combined with the presence of systemic inflammatory response syndrome (SIRS) [19]. The presence of SIRS was based on the data from the patient charts. SIRS was defined as either a clear mention of SIRS or the presence of fever (> 38.5 °C) or hypothermia (< 36 °C), tachycardia or bradycardia, and an elevated white blood cell count [20]. Subsequently, neonates were classified into the sepsis and non-sepsis groups depending on the presence or absence of sepsis. For neonates without symptoms of infection, such as fever, colonization was considered. For neonates with fever or other infection symptoms but not meeting the SIRS criteria or those with positive skin cultures but negative results in other cultures, the presence of infection without sepsis was considered. Microbiology Each sample was subjected to microbiological testing, and the results were reported by the microbiology unit of the Division of Laboratory of our hospital. Bacterial cultures were assessed for antimicrobial susceptibility using VITEK-2 (France) and BD-Phoenix100 (United States of America) microbial analysis systems for identifying positive samples. Drug sensitivity testing was performed using automated instrumentation, and the assessment criteria adhered to the Clinical and Laboratory Standards Institute (CLSI) recommendations. The following quality control strains were used: Escherichia coli (ATCC 25922), Klebsiella pneumoniae (ATCC 700603), and Pseudomonas aeruginosa (ATCC 27853). The testing standards recommended by the CLSI were used. Carbapenem-resistant Enterobacterales (CRE) were defined as GNB that met any of the following conditions: (1) resistance to any carbapenem drug of imipenem, meropenem, ertapenem, or doripenem; (2) for natural bacteria with low sensitivity to imipenem (such as Morganella, Proteus, and Procidencia, etc.), it was necessary to refer to the drug sensitivity results of other carbapenem antibiotics except imipenem; (3) documented carbapenemase production [21]. ESBL-producing organisms were considered to be certain bacteria that can hydrolyze extended-spectrum cephalosporins [22]. Susceptibility testing was performed using the ESBL test (VITEK-2, bioMérieux, and BD Phoenix100). Data collection The following clinical data were recorded: sex, premature delivery, gestational age, birth weight, premature rupture of membranes, high-risk factors before delivery (such as gestational diabetes, antibiotics therapy or presence of fever among mothers before delivery), delivery mode, amniotic fluid muddy, invasive operation(including invasive ventilator, PICC, umbilical artery catheterization and umbilical vein catheterization), routine blood test results (complete blood count, procalcitonin, and blood cultures), antibiotic therapies, hospitalization duration and outcomes. MDR-GNB in this study were defined as isolates exhibiting non-susceptibility to at least one agent in three or more antimicrobial categories relevant to Gram-negative organisms. The MDR-Enterobacterales included Escherichia coli, Klebsiella pneumoniae, Enterobacter aerogenes, Klebsiella oxytoca, and Proteus mirabilis, while non-fermenting Gram-negative bacilli included Pseudomonas aeruginosa and Acinetobacter baumannii [15]. GNBs are naturally resistant to glycopeptides, lipopeptides, oxazolidinones, and streptogramins, but since antibiotic susceptibility testing and microorganism identification are usually run concomitantly to save time, antibiotics are tested before the microorganism is identified as a GNB. In our hospital, empirical treatment for infections typically includes cefotaxime and piperacillin-tazobactam. Statistical analysis SPSS 24.0 for Windows (IBM, Armonk, NY) was used for data analysis. Continuous data were presented as means ± standard deviations. Continuous data were analyzed using Student’s t-test or the Mann-Whitney U test. Categorical data were characterized by number and percentage and were analyzed using the chi-squared test or Fisher’s exact test. Statistical significance was set at p < 0.05. Risk factors for infection and sepsis were analyzed using univariable and multivariable logistic regression analyses. Variables with a p-value < 0.1 in the univariable analyses were included in the multiple logistic regression analysis. In the multivariate analysis, the model’s explanatory power was assessed using the Nagelkerke R-squared. A p-value of < 0.05 was considered statistically significant. Results GNBs were detected in cultures from 793 samples obtained from 421 neonates. Among these 793 GNBs, 425 (425/793, 53.6%) organisms from 242 (242/421, 57.5%) neonates were identified as MDR-GNB. From 2017 to 2022, 190, 350, 409, 294, 415, and 427 MDROs were detected. Among them, 47 (47/190, 24.7%), 68 (68/350, 19.4%), 83 (83/409, 20.3%), 50 (50/294, 17.0%), 94 (94/415, 22.7%), and 83 (83/427, 19.4%) were MDR-GNBs. A total of 425 organisms were identified across various samples, with the distribution as follows: skin swab, 171 (171/425, 40.2%); gastric juice, 130 (130/425, 30.6%); blood, 56 (56/425, 13.2%); sputum, 50 (50/425, 11.8%); urine, nine (9/425, 2.1%); navel secretion, one (1/425, 0.2%); peripheral central venous catheterization (PICC), two (2/425, 0.5%); pharyngeal swab, three (3/425, 0.7%); peritoneal drainage fluid, two (2/425, 0.5%); eye secretion, one (1/425, 0.2%). The pathogens detected were as follows: E. coli, 333 isolates (333/425, 78.4%), K. pneumoniae, 65 isolates (65/425, 15.3%), Enterobacter aerogenes, 14 isolates (14/425, 3.3%), P. aeruginosa, three isolates (3/425, 0.7%), Klebsiella oxytoca, four isolates (4/425, 0.9%), Acinetobacter baumannii, two isolates (2/425, 0.5%), and Proteus mirabilis, four isolates (4/425, 0.9%). E. coli and K. pneumoniae were the major bacteria found in this study (Fig. 1). Among the 14 CRE specimens, 11 represented infections and were found in two neonates; nine were K. pneumoniae, and two were E. coli. (nine were obtained from sputum samples and two from pharyngeal swabs), while three specimens were considered to represent colonization or contamination. The two neonates with pharyngeal swabs were diagnosed with neonatal pneumonia, accompanied by respiratory distress, and were supported by mechanical ventilation. They only had mild fever and did not meet the diagnostic criteria for sepsis (infection + SIRS), which may be related to their very premature birth (one neonate at 22 + 2 weeks and the other at 26 + 4 weeks). The results of repeated deep suction cultures during hospitalization confirmed the diagnosis, with clear sampling times, and not just repeated cultures of a single sample. None resulted in sepsis during hospitalization. Fig. 1Distribution of detected pathogens in the study population Distribution of detected pathogens in the study population In addition, 217 (217/425, 51.1%) ESBL-producing organisms were detected among the samples: E. coli, 168 isolates (168/217, 77.4%); K. pneumoniae, 48 isolates (48/217, 22.1%); Acinetobacter baumannii, one isolate (1/217, 0.5%). From 2017 to 2022, the proportions of GNB isolates that were ESBL-producing organisms were 38.3%.(18/47), 60.3%༈41/68༉, 39.8%༈33/83༉, 44.0%༈22/50༉, 68.1%༈64/94༉, and 47.0%༈39/83༉, respectively. The drug resistance of the bacteria and CRE is shown in Table 1 and S1. The bacteria detected each year are shown in Fig. 2. The usage rates of carbapenem in the NICU and the whole hospital are shown in Figure S1. Table 1Antimicrobial resistance patterns of Gram-negative bacilli among hospitalized infantsOrganismsEscherichia coli (n = 333)Klebsiella pneumoniae (n = 65)Enterobacter aerogenes (n = 14)Acinetobacter baumannii (n = 2)Proteus mirabilis (n = 4)Pseudomonas aeruginosa (n = 3)Klebsiella oxytoca (n = 4)DR (%)DR (%)DR (%)DR (%)DR (%)DR (%)DR (%)AMK011 (16.9)00000GEN127 (28.1)26 (40.0)0001 (33.3)0TOB16 (4.8)11 (16.9)00000IMP2 (0.6)11 (16.9)1 (7.1)02 (50.0)00MEM2 (0.6)4 (6.2)00000CZO208 (62.5)52 (80.0)1 (7.1)1 (50.0)4 (100.0)04 (100.0)CAZ55 (16.5)44 (67.7)13 (81.3)003 (100.0)0CTX77 (23.1)18 (27.7)6 (42.9)1 (50.0)003 (75.0)CRO95 (28.5)34 (52.3)7 (50.0)1 (50.0)2 (50.0)*1 (25.0)FEP104 (31.2)36 (55.4)2 (14.3)1 (50.0)03 (100.0)3 (75.0)ATM102 (30.6)37 (56.9)11 (78.6)*0*0AMP286 (85.9)*0*2 (50.0)*3 (75.0)PIP164 (49.2)28 (43.1)6 (43.9)003 (100.0)3 (75.0)AMC65 (19.5)26 (40.0)2 ()*0*0CIP105 (31.5)23 (35.4)0002 (66.7)0LVX138 (41.4)17 (26.2)00000MFX54 (16.2)6 (9.2)0001 (33.3)0TCY138 (41.4)23 (35.4)000*3 (75.0)SXT198 (59.4)43 (66.2)004 (100.0)*4 (100.0)The results are shown as the n (%) of resistant isolates to each speciesDR, drug resistance; AMK, amikacin; AMP, ampicillin; AMC, amoxicillin/CA; ATM, amtreonam; CAZ, ceftazidime; CIP, ciprofloxacin; CRO, cefatriaxone; CTX, cefotaxime; CZO, cefazolin; DR, drug-resistant; FEP, cefepime; GEN, gentamicin; IMP, imipenem; LVX, levofloxacin; MEM, meropenem; MFX, moxifloxacin; PIP, piperacillin; SXT, sulfamethoxazole/trimethoprim; TCY, tetracycline; TOB, tobramycin*Natural drug-resistant Table 1Antimicrobial resistance patterns of Gram-negative bacilli among hospitalized infantsOrganismsEscherichia coli (n = 333)Klebsiella pneumoniae (n = 65)Enterobacter aerogenes (n = 14)Acinetobacter baumannii (n = 2)Proteus mirabilis (n = 4)Pseudomonas aeruginosa (n = 3)Klebsiella oxytoca (n = 4)DR (%)DR (%)DR (%)DR (%)DR (%)DR (%)DR (%)AMK011 (16.9)00000GEN127 (28.1)26 (40.0)0001 (33.3)0TOB16 (4.8)11 (16.9)00000IMP2 (0.6)11 (16.9)1 (7.1)02 (50.0)00MEM2 (0.6)4 (6.2)00000CZO208 (62.5)52 (80.0)1 (7.1)1 (50.0)4 (100.0)04 (100.0)CAZ55 (16.5)44 (67.7)13 (81.3)003 (100.0)0CTX77 (23.1)18 (27.7)6 (42.9)1 (50.0)003 (75.0)CRO95 (28.5)34 (52.3)7 (50.0)1 (50.0)2 (50.0)*1 (25.0)FEP104 (31.2)36 (55.4)2 (14.3)1 (50.0)03 (100.0)3 (75.0)ATM102 (30.6)37 (56.9)11 (78.6)*0*0AMP286 (85.9)*0*2 (50.0)*3 (75.0)PIP164 (49.2)28 (43.1)6 (43.9)003 (100.0)3 (75.0)AMC65 (19.5)26 (40.0)2 ()*0*0CIP105 (31.5)23 (35.4)0002 (66.7)0LVX138 (41.4)17 (26.2)00000MFX54 (16.2)6 (9.2)0001 (33.3)0TCY138 (41.4)23 (35.4)000*3 (75.0)SXT198 (59.4)43 (66.2)004 (100.0)*4 (100.0)The results are shown as the n (%) of resistant isolates to each speciesDR, drug resistance; AMK, amikacin; AMP, ampicillin; AMC, amoxicillin/CA; ATM, amtreonam; CAZ, ceftazidime; CIP, ciprofloxacin; CRO, cefatriaxone; CTX, cefotaxime; CZO, cefazolin; DR, drug-resistant; FEP, cefepime; GEN, gentamicin; IMP, imipenem; LVX, levofloxacin; MEM, meropenem; MFX, moxifloxacin; PIP, piperacillin; SXT, sulfamethoxazole/trimethoprim; TCY, tetracycline; TOB, tobramycin*Natural drug-resistant Antimicrobial resistance patterns of Gram-negative bacilli among hospitalized infants The results are shown as the n (%) of resistant isolates to each species DR, drug resistance; AMK, amikacin; AMP, ampicillin; AMC, amoxicillin/CA; ATM, amtreonam; CAZ, ceftazidime; CIP, ciprofloxacin; CRO, cefatriaxone; CTX, cefotaxime; CZO, cefazolin; DR, drug-resistant; FEP, cefepime; GEN, gentamicin; IMP, imipenem; LVX, levofloxacin; MEM, meropenem; MFX, moxifloxacin; PIP, piperacillin; SXT, sulfamethoxazole/trimethoprim; TCY, tetracycline; TOB, tobramycin *Natural drug-resistant Fig. 2The MDR-GNB in each year The MDR-GNB in each year Among the 242 neonates, 131 (131/242, 54.1%) were identified as showing colonization, and 111 (111/242, 45.9%) were diagnosed with infection. The clinical features of the infection and colonization of neonates are shown in Table 2. There were no significant differences in sex, premature rupture of membranes, gestational diabetes, whether mothers had a fever, cloudy amniotic fluid, type of delivery, or hospital stay between the infected and non-infected neonates. Infected neonates exhibited significantly higher proportions of premature birth (p < 0.001), birth weight < 1500 g (p < 0.001), invasive ventilator use (p = 0.001), PICC catheterization (p < 0.001), umbilical artery/vein catheterization (p < 0.001), and maternal antibiotic therapy (p = 0.001). Neonates with infection had a longer hospitalization duration (p < 0.001). Logistic regression analysis identified birth weight < 1500 g (p < 0.001, OR = 3.338, 95%CI 1.418–7.858) as a significant risk factor for infection (Table 3). The Nagelkerke R-squared was calculated to be 0.239, indicating a moderate level of fit for the model in explaining the variance in the data. Table 2Clinical features between infection and colonization groupMale sexInfection (n = 111)Colonization (n = 131) P 54 (48.6%)73 (55.7%)0.272Premature83 (74.8%)76 (58.0%)0.006Fever of mother30 (27.0%)23 (17.6%)0.076Antibiotic therapy of mother46 (41.4%)29 (22.1%)0.001Birth weight < 1500 g62 (55.9%)33 (25.2%)< 0.001Premature rupture of membranes46 (41.4%)53 (40.5%)0.877Gestational diabetes19 (17.1%)25 (19.1%)0.693Meconium-stained amniotic fluid25 (22.5%)30 (22.9%)0.944Cesarean section35 (31.5%)35 (26.7%)0.411Invasive ventilator43 (38.7%)26 (19.8%)0.001UAC/UVC49 (44.1%)29 (22.1%)< 0.001PICC59 (53.2%)35 (26.7%)< 0.001Hospitalization duration (days)39 (14, 71)11 (7, 36)< 0.001PICC, peripherally-inserted central catheter; UAC, umbilical artery catheterization; UVC, umbilical vein catheterization Table 2Clinical features between infection and colonization groupMale sexInfection (n = 111)Colonization (n = 131) P 54 (48.6%)73 (55.7%)0.272Premature83 (74.8%)76 (58.0%)0.006Fever of mother30 (27.0%)23 (17.6%)0.076Antibiotic therapy of mother46 (41.4%)29 (22.1%)0.001Birth weight < 1500 g62 (55.9%)33 (25.2%)< 0.001Premature rupture of membranes46 (41.4%)53 (40.5%)0.877Gestational diabetes19 (17.1%)25 (19.1%)0.693Meconium-stained amniotic fluid25 (22.5%)30 (22.9%)0.944Cesarean section35 (31.5%)35 (26.7%)0.411Invasive ventilator43 (38.7%)26 (19.8%)0.001UAC/UVC49 (44.1%)29 (22.1%)< 0.001PICC59 (53.2%)35 (26.7%)< 0.001Hospitalization duration (days)39 (14, 71)11 (7, 36)< 0.001PICC, peripherally-inserted central catheter; UAC, umbilical artery catheterization; UVC, umbilical vein catheterization Clinical features between infection and colonization group PICC, peripherally-inserted central catheter; UAC, umbilical artery catheterization; UVC, umbilical vein catheterization Table 3Logistic regression analysis of risk factors of infectionUnivariable logistic regressionMultivariable logistic regressionOR P OR95%CI P Premature2.1450.0070.8800.400-1.9320.750Fever of mother0.5750.0780.5650.272–1.1770.128Antibiotic therapy of mother0.4020.0010.7140.363–1.4070.330Birth weight < 1500 g0.266< 0.0013.3381.418–7.8580.006Invasive ventilator0.3920.0011.0680.378–3.0200.901UAC/UVC0.360< 0.0011.3700.423–4.4410.600PICC0.321< 0.0010.5320.208–1.3600.187Sex0.7530.272Premature rupture of membranes1.0420.877Cesarean section0.7920.411Amniotic fluid contamination1.0220.944Fetal distress0.7150.208Gestational diabetes1.1420.693CI, confidence interval; PICC, peripherally-inserted central catheter; UAC, umbilical artery catheterization; UVC, umbilical vein catheterization Table 3Logistic regression analysis of risk factors of infectionUnivariable logistic regressionMultivariable logistic regressionOR P OR95%CI P Premature2.1450.0070.8800.400-1.9320.750Fever of mother0.5750.0780.5650.272–1.1770.128Antibiotic therapy of mother0.4020.0010.7140.363–1.4070.330Birth weight < 1500 g0.266< 0.0013.3381.418–7.8580.006Invasive ventilator0.3920.0011.0680.378–3.0200.901UAC/UVC0.360< 0.0011.3700.423–4.4410.600PICC0.321< 0.0010.5320.208–1.3600.187Sex0.7530.272Premature rupture of membranes1.0420.877Cesarean section0.7920.411Amniotic fluid contamination1.0220.944Fetal distress0.7150.208Gestational diabetes1.1420.693CI, confidence interval; PICC, peripherally-inserted central catheter; UAC, umbilical artery catheterization; UVC, umbilical vein catheterization Logistic regression analysis of risk factors of infection CI, confidence interval; PICC, peripherally-inserted central catheter; UAC, umbilical artery catheterization; UVC, umbilical vein catheterization Sepsis was present in 67 (67/242, 27.7%) neonates, with 55 (55/67, 82.1%) attributed to MDR-GNB infection. There were no significant differences in clinical characteristics, such as premature rupture of membranes, gestational diabetes, cloudy amniotic fluid, or delivery type, between the neonates with or without sepsis. Neonates with premature birth (p < 0.001), invasive ventilator use (p = 0.005), PICC catheterization (p < 0.001), umbilical artery/vein catheterization (p < 0.001), maternal antibiotics therapy (p < 0.001), fever before delivery (p = 0.028), and a birth weight of < 1500 g (p < 0.001) had a higher proportion of sepsis (Table 4). In addition, neonates with sepsis experienced significantly longer hospital stays (p < 0.001). Logistic regression analysis revealed that maternal fever before delivery (p = 0.022, OR = 2.679, 95%CI 1.153–6.221) and birth weight less than 1500 g (p = 0.029, OR = 2.768, 95%CI 1.107–6.918) were risk factors for sepsis (Table S2). Table 4Clinical features between sepsis and non-sepsis groupMale sexSepsis (n = 67)Non-sepsis (n = 175) P 30 (44.8%)97 (55.4%)0.138Premature58 (86.6%)101 (57.7%)< 0.001Fever of mother21 (31.3%)32 (18.3%)0.028Antibiotic therapy of mother34 (50.7%)41 (23.4%)< 0.001Birth weight < 1500 g44 (65.7%)51 (29.1%)< 0.001Premature rupture of membranes45 (67.2%)98 (56.0%)0.114Gestational diabetes10 (14.9%)34 (19.4%)0.416Meconium-stained amniotic fluid13 (19.4%)42 (24.0%)0.445Cesarean section22 (32.8%)48 (27.4%)0.406Invasive ventilator28 (41.8%)41 (23.4%)0.005UAC/UVC35 (52.2%)43 (24.6%)< 0.001PICC42 (62.7%)52 (29.7%)< 0.001Hospitalization duration (days)45 (16, 88)12 (7, 46.5)< 0.001PICC, peripherally-inserted central catheter; UAC, umbilical artery catheterization; UVC, umbilical vein catheterization Table 4Clinical features between sepsis and non-sepsis groupMale sexSepsis (n = 67)Non-sepsis (n = 175) P 30 (44.8%)97 (55.4%)0.138Premature58 (86.6%)101 (57.7%)< 0.001Fever of mother21 (31.3%)32 (18.3%)0.028Antibiotic therapy of mother34 (50.7%)41 (23.4%)< 0.001Birth weight < 1500 g44 (65.7%)51 (29.1%)< 0.001Premature rupture of membranes45 (67.2%)98 (56.0%)0.114Gestational diabetes10 (14.9%)34 (19.4%)0.416Meconium-stained amniotic fluid13 (19.4%)42 (24.0%)0.445Cesarean section22 (32.8%)48 (27.4%)0.406Invasive ventilator28 (41.8%)41 (23.4%)0.005UAC/UVC35 (52.2%)43 (24.6%)< 0.001PICC42 (62.7%)52 (29.7%)< 0.001Hospitalization duration (days)45 (16, 88)12 (7, 46.5)< 0.001PICC, peripherally-inserted central catheter; UAC, umbilical artery catheterization; UVC, umbilical vein catheterization Clinical features between sepsis and non-sepsis group PICC, peripherally-inserted central catheter; UAC, umbilical artery catheterization; UVC, umbilical vein catheterization Fifteen neonates (15/242, 6.2%) passed away: eight (8/242, 3.3%) and seven (7/242, 2.9%) neonates in the infected and non-infected groups, respectively; the intergroup difference was not significant (p = 0.525). Nine of the neonates had sepsis, and six did not, and mortality was significantly higher among those with sepsis (p = 0.003). A birth weight of < 1500 g (p = 0.001) was a significant factor, with four of these neotates involving a gestational age of less than 28 weeks (p < 0.001). Discussion This study analyzed the prevalence and clinical characteristics of MDR-GNB infections among neonates. The findings in this study showed that E. coli and K. pneumoniae were the most common MDR-GNB detected in neonates in the NICU. Risk factors for sepsis development included low birth weight and maternal fever during delivery. This study contributes valuable clinical research insights for preventing and treating MDR-GNB infections and associated sepsis in neonates. Although Gram-positive pathogens remain the major sources of infection in the NICU [23], GNBs have exhibited increasing antibiotic resistance in medical institutions. The present study revealed that the most common pathogens among isolates were E. coli (78.4%) and K. pneumoniae (15.3%). Consistent with those findings, a previous study showed that from 2016 to 2020, GNB accounted for approximately 57.9% of organisms causing infections in Chinese children, with E. coli and K. pneumoniae being the primary pathogens [24]. In addition, Jiang et al. reported that 15% to 20% of infections in the NICU are caused by GNB, with delayed septicemia and hospital-acquired pneumonia, including ventilator-associated pneumonia, being the most significant among them [25]. Thus, monitoring these pathogenic bacteria to facilitate effective management and provide early warnings for infection control and antibiotic management is imperative. ESBL-producing organism infections represent a significant clinical challenge, especially given the higher mortality in neonates compared with adults [26]. The incidence of infection or colonization in NICU ranged from 4.7% to 59.9%, varying by region and the antibiotic therapies employed [22, 27]. Danino et al. reported that about 34.3% of ESBL-colonized mothers also had ESBL colonization in their infants, and both groups had higher antibiotic exposure, showing the significant role of mother-to-child transmission in ESBL colonization and infection [28]. In this study, the proportion of ESBL was similar to the results of the previous report, but the likelihood varied significantly over the years. Therefore, the monitoring and prevention of ESBL in mothers and infants should be strengthened. Although some reports suggested a high infection rate with CRE among infants, there are limited reports of infection with CRE in the NICU [29, 30]. According to the Antimicrobial Resistance Profiles in Children in China, about 6.8% of Enterobacterales are CRE, and these organisms, especially carbapenem-resistant P. aeruginosa (CRPA), were detected at a higher rate in neonates than in non-neonates [24]. On the other hand, a recent single-center study in China showed that the incidence of CRE infection was 0.63% to 1.96% [31], indicating that the distribution of bacteria may be associated with the regions. In the present study, only 11 (2.1%) neonates with CRE infection/colonization were found in the NICU, and none of these neonates developed sepsis, demonstrating that the incidence of infection with CRE may still not be frequent in the NICU of the study hospital. Although the infection/colonization of CRE was infrequent, it remains crucial to prevent CRE infections due to their potential for more severe primary and secondary outcomes compared to infections with carbapenem-sensitive Enterobacter among infants [29]. In neonates, colonization may raise the risk of infection. Singh et al. reported that about 8.7% of neonates experienced CRE colonization during hospitalization, and nasal gastric tube, breastfeeding, top feeding, expressed breastmilk, ventilation, and antibiotics administration were the risk factors for gut colonization [32]. Dias et al. reported that approximately 12% of neonates with colonization developed infection, and 58% of the neonates were infected by the same organisms as those colonizing the neonates [33]. In the present study, 38.7% (75/194) of the neonates with skin or gut colonization had an infection, and 20.6% (40/194) developed MDR-GNB sepsis during hospitalization. The significant proportion of neonates in the present study who were infected by the same colonizing organisms underscores the importance of preventive measures following colonization by MDR-GNB. In addition, premature birth, extremely low birth weight, and other factors, including the mother with colonization and the NICU environment, were associated with colonization [34]. Previous studies demonstrated that low birth weight and premature delivery were both risk factors for sepsis [35, 36]. In the present study, neonates whose mothers experienced fever before delivery, which is a risk factor of sepsis, exhibited a higher infection rate, indicating a significant maternal-child bond. Previous studies have consistently identified maternal fever as a significant risk factor for neonatal sepsis [37]. A study demonstrated that while low-grade fever can occur transiently during labor, maternal fever exceeding 100.4 °F necessitates further evaluation [38]. These findings corroborated our results, highlighting that premature birth, low birth weight, and maternal fever may be associated with an elevated infection rate. When the mother’s temperature surpasses 38℃, it is essential to investigate the neonate’s medical history, including medication history, prostaglandins, and epidural anesthesia, and conduct relevant tests such as complete blood count, procalcitonin, blood cultures, and fetal heart monitoring. If infection cannot be excluded, especially in pregnant women with low-birth-weight fetuses or preterm deliveries, antibiotic use is recommended. It is well-established that ESBL-producing bacteria are common in Asia, with typical MDR-GNBs being E. coli and K. pneumoniae [39, 40]. In addition, other studies have shown that low birth weight and preterm delivery are known risk factors for infections in NICU neonates [35, 36]. Nevertheless, including data from 2017 to 2022 provides a new perspective, particularly regarding the impact of the COVID-19 pandemic and the resulting shifts in antimicrobial usage patterns. Although the present study did not investigate prevention, recommendations to reduce the risk of infections in neonates should include minimizing invasive procedures (such as the duration of intravenous and arterial catheter placement) [41, 42], using maternal breast milk [43], strictly applying hand hygiene and aseptic techniques [44, 45], extending gestational age to increase birth weight when possible [46], and educating healthcare personnel [45]. Several pathogens reported here (Escherichia coli, Klebsiella pneumoniae, Enterobacter aerogenes, Pseudomonas aeruginosa, Klebsiella oxytoca, Acinetobacter baumannii, and Proteus mirabilis) can be contextualized within the 2024 WHO Bacterial Priority Pathogens List (BPPL) [47]. The BPPL is highly relevant for interpreting their clinical and public health significance, especially concerning the global threat of antimicrobial resistance. The WHO BPPL 2024 categorizes 24 bacterial pathogen-antibiotic combinations into three tiers: critical, high, and medium priority. Several of the pathogens reported here fall into these highest concern categories. The critical priority includes carbapenem-resistant Acinetobacter baumannii and carbapenem-resistant and third-generation cephalosporin-resistant Enterobacterales (including E. coli, K. pneumoniae, E. aerogenes, K. oxytoca, and Proteus mirabilis). The high priority includes carbapenem-resistant Pseudomonas aeruginosa. Carbapenem-resistant K. pneumoniae and Acinetobacter baumannii were ranked among the top priority bacteria, with K. pneumoniae receiving the highest score in the 2024 assessment. The detection of these pathogens reflects the pressing challenge of antibiotic resistance, particularly among Gram-negative bacteria, which remain a persistent global threat. The WHO BPPL 2024 guides research, development, and policy focus to combat AMR, highlighting the need for new treatments targeting these high-burden pathogens [47]. This study has some limitations. It was a single-center retrospective study that cannot avoid selection bias. The available data does not allow for a clear distinction between early-onset sepsis and healthcare-associated sepsis or between colonization and infection; only the presence of the microorganism could be confirmed. In addition, gestational age was not considered. Genetic sequencing of the bacterial community was not performed. In addition, the sample size in this study was small, limiting generalizability. Furthermore, clinical data were incomplete, and sampling was not regularly performed for all neonates, which did not reflect the situation of colonization by GNB during hospitalization. Each patient generally underwent multiple cultures from different sites and at multiple time points, but whether each sample included in the study was from surveillance or performed on a clinical basis cannot be determined based on the available data. Another limitation of this study is the lack of systematic collection of detailed symptom data. Repeated tests were not performed on the neonates regarding the GNBs that are not resistant, preventing the analysis of how the rate of MDROs increased with a longer length of stay in the unit. Future research should consider conducting a prospective multicenter study with systematic and regular sampling. Conclusions In conclusion, the MDR rate in GNB was high in the NICU, and E. coli and K. pneumoniae were the most common MDR-GNB in the NICU. Premature birth, low birth weight, and a feverish mother might be risk factors for infection or sepsis in neonates with MDR-GNB in the NICU. Further measures are needed to prevent MDR-GNB infection. Supplementary Information Supplementary Material 1. Supplementary Material 1.