The Changing Bacteriological Profile of Early-Onset Neonatal Bacterial Infections

Introduction

Early-onset neonatal bacterial infections (EONBI) represent a significant public health concern, accounting for considerable neonatal mortality—estimated between 2% and 3% among full-term newborns,¹ and rising to 20% in preterm infants.² Clinically, EONBI presents with a wide range of nonspecific symptoms, making diagnosis challenging. The bacteriological landscape of these infections is continuously evolving, with Escherichia coli and Group B Streptococcus remaining predominant. However, there is an increasing emergence of multidrug-resistant bacterial strains,^3,4 further complicating diagnosis, treatment, and prognosis.

Methods

We conducted a retrospective study over a two-year period (from January 1, 2022, to December 31, 2023) in the neonatal unit of the Habib Bougatfa Maternity Center in Bizerte, Tunisia.

The study included all newborns (NBs) hospitalized for confirmed EONBI, defined by a positive blood culture obtained within the first 48 hours of admission.

Exclusion criteria were as follows: Newborns admitted for viral, parasitic, or fungal infections, and those with suspected EONBI whose diagnosis was later ruled out based on negative laboratory and microbiological findings. We also excluded cases where blood cultures were not obtained prior to initiation of antibiotic therapy.

Maternal and neonatal data were collected from hospital medical records and the microbiology department, then recorded in a standardized electronic data sheet.

In accordance with hospital protocol and based on national guidelines by the Tunisian Association of Neonatal Medicine,³ newborns were categorized upon admission into three groups based on clinical presentation and risk factors for EONBI, as outlined in Appendix 1.

Blood cultures were systematically performed before starting empirical antibiotic therapy. The classification of pathogens as sensitive or resistant was based on microbiological features and antibiogram results, compared against the empiric antibiotics recommended by the unit’s protocol. Cases of Stenotrophomonas maltophilia were only considered confirmed if two consecutive blood cultures yielded the same pathogen.

Data were analyzed using SPSS software version 11.5. A p-value of <0.05 was considered statistically significant for all tests performed.

Results

Between 2022 and 2023, a total of 5,342 live births were recorded, with 2,248 newborns (NBs) admitted to the neonatal unit. Among these, 135 cases of early-onset neonatal bacterial infections (EONBI) were microbiologically confirmed, resulting in an annual incidence of 62.8 per 1,000 hospitalizations and an overall prevalence of 25.27 per 1,000 live births.

The mean maternal age was 32 ± 5 years (range: 18 to 45 years), with the most common age group being 29–36 years. A low socioeconomic status was reported in 56.3% of mothers. A previous cesarean delivery was noted in 19.84% (n = 26). The most frequent gynecologic-obstetric history included spontaneous miscarriages (12.21%) and early pregnancy losses (8.39%). A previous history of neonatal hospitalization for confirmed EONBI was identified in 12.21% of cases (n = 16); when excluding primigravida patients, this rate rose to 17.98%. In line with these maternal characteristics and obstetric histories, screening for EONBI risk factors was routinely performed. Urine cultures were obtained during pregnancy in 83.97% of cases (n = 110), and postnatally in 9.92% (n = 13). Among all samples, 23.57% (n = 29) were positive, with the majority (62.07%, n = 18) collected during the third trimester. Escherichia coli was the most frequently isolated pathogen (43.75%). Antibiotic therapy was initiated in 81.25% of these cases, and the treatment was appropriate to the antibiogram in 84.62% of instances.

Vaginal swabs were obtained during pregnancy in 49.61% of cases (n = 65), and postpartum in 3.05% (n = 4). Among these, 39.13% (n = 27) were positive, with most samples collected during the third trimester (65.22%), of which 40% yielded a positive result. Group B Streptococcus was the most frequently identified microorganism (33.33%). These findings highlight the role of late pregnancy vaginal colonization as a potential risk factor for neonatal infection.

In close relation to these infectious screening results, premature rupture of membranes (PROM) was observed in 42.75% of mothers (n = 56), with prolonged rupture exceeding 12 hours in 32.59% (n = 44) of newborns—more commonly among preterm infants (61.35%). Notably, antibiotic prophylaxis in cases of prolonged PROM was deemed appropriate in 56.82% of instances, reflecting partial adherence to preventive strategies.

Moreover, several additional infectious conditions during pregnancy were identified, including intrauterine infection (15.27%), maternal sepsis (2.29%), and other localized infections (pulmonary, ENT, digestive, or cutaneous) in 21.37% of cases. These were often associated with medical-obstetric complications such as gestational diabetes (33.33%), gestational hypertension (10.37%), and threatened preterm labor (7.41%), which may have contributed to neonatal vulnerability.

Hospitalization during pregnancy, outside of labor, was recorded in 42.96% of cases (n = 58), most frequently in the third trimester (34.07%). The mean hospital stay was 6.29 ± 6.62 days (range: 2–33), with PROM (27.58%) being the most common cause for admission, followed by threatened preterm labor (17.24%).

Delivery took place in the hospital (inborn) in the majority of cases (81.48%), while 16.30% were born in other healthcare facilities (outborn), and 2.22% at home. Cesarean section was the predominant mode of delivery (63.7%). Regarding amniotic fluid characteristics, it was clear in 85.19% of cases, meconium-stained in 0.74%, and discolored in 14.07%. Oligohydramnios was observed in 28.89% of deliveries. Neonatal resuscitation at birth was necessary in 31.85% of cases, underscoring the perinatal risk context.

The mean gestational age was 36 weeks + 5 days ± 15.3 days (range: 28+4 to 41+2), and mean birth weight was 2,787 ± 850 grams (range: 920–5100 g). The majority of newborns were admitted directly from the delivery or operating room (71.11%), while others came from postpartum wards (17.04%), emergency departments (2.22%), or were transferred from other hospitals (14.07%). The mean age at admission was 5.28 ± 7.36 days (range: 1 hour to 63 hours of life), with neonatal respiratory distress being the most frequent reason for admission (63.07%).

The clinical presentation of the newborns was diverse and often evolved during hospitalization. At birth, 34.08% (n = 46) of newborns were asymptomatic. Based on initial risk stratification, 18.52% (n = 25) and 15.56% (n = 21) were classified into Groups A and B, respectively, while symptomatic newborns accounted for 65.92% (n = 89) and were assigned to Group C. Among those initially categorized in Groups A and B, more than half (54.34%, n = 25) developed symptoms during their hospital stay and were subsequently reclassified into Group C. Newborns who remained asymptomatic throughout hospitalization represented 15.56% of cases, with severe IUGR, low birth weight, and prematurity being the most common reasons for admission. Among symptomatic infants, clinical presentations were polymorphic, non-specific, and often progressive, with respiratory and gastrointestinal symptoms being the most commonly reported (77.78% and 39.26%, respectively).

The bacteriological profile revealed that 14 distinct organisms were isolated from blood cultures ( Table 1), with a strong predominance of Gram-negative bacilli (96.3%). The most commonly identified pathogens were Serratia marcescens (37.78%) and Escherichia coli (27.41%). Resistant strains were found in 46.66% of cases (n = 63), while 53.33% (n = 72) remained susceptible to first-line empirical antibiotics.

Several predictive factors were identified in association with infections caused by resistant EONBI pathogens, as summarized in Table 2.

Consequently, management and prognosis differed significantly between resistant and sensitive EONBI cases. Second- and third-line antibiotics were more frequently required in resistant EONBI cases (p = 0.04 and p = 0.049, respectively). These infections were significantly associated with longer durations of antibiotic therapy (11.02 ± 3.57 vs. 8.97 ± 3.21 days; p = 0.0006) and hospitalization (12.8 ± 7.7 vs. 10.3 ± 5 days; p = 0.027).

Resistant cases were also more frequently associated with complications such as healthcare-associated infections, transfer to intensive care, and increased mortality, although these associations did not reach statistical significance (19% vs. 11.1%; p = 0.195).

Discussion

The prevalence of early-onset neonatal bacterial infections (EONBI) shows considerable temporal and spatial variability. It is estimated to range between 0.7 and 1 per 1,000 live births in the United States according to the perinatal network,¹ and between 1 and 4 per 1,000 live births in France according to HAS.⁵ Among published studies, the highest prevalence was reported by Devred et al. at 62.6 per 1,000 live births,⁶ while the lowest, 1 per 1,000 live births, was found in France by Mteyrek.⁷ The prevalence observed in our study is considered high compared to previous reports, which may be explained by the low socioeconomic status of the population, irregular gynecological follow-up, and the lack of systematic screening for urinary and genital infections during pregnancy.

Regarding the bacteriological profile and antimicrobial resistance, HAS reports that the most frequently isolated pathogens in EONBI are Escherichia coli and Group B Streptococcus.⁵ According to the perinatal network, Group B Streptococcus accounts for 40 to 50% of cases, Escherichia coli for 10 to 15%, and other bacteria for 30 to 40%.¹ A 2007 English study by Aujard et al. found that Group B Streptococcus represented 31% of isolates, coagulase-negative staphylococci 22%, followed by Escherichia coli and other streptococci at 9% each.^6,7 Our results are consistent with those from Kloula’s study at CMNT,³ but differ from Aujard et al.,⁸ who observed an increase in Staphylococcus aureus frequency. In our study, the bacterial profile was marked by a decrease in Group B Streptococcus and Gram-positive cocci, replaced by Gram-negative bacilli (96.3%), the emergence of unusual pathogens such as Serratia and Stenotrophomonas maltophilia, and an increase in resistance rates.

Our antibiogram data showed a global resistance rate of 46.66%. Resistant strains were mainly Serratia marcescens (57.14% of resistant isolates), followed by Escherichia coli resistant to third-generation cephalosporins (12.7%), Stenotrophomonas maltophilia (7.96%), and Pseudomonas aeruginosa (4.76%), among others. Escherichia coli resistance to third-generation cephalosporins was observed at 52% and 8% in the studies of Chemsi et al.⁹ and Stoll et al.,¹⁰ respectively. Resistance among Gram-negative bacilli was estimated at 40% in Kemeze et al.¹¹ and Chiabi.¹² Kloula described a new bacteriological profile including 50% ampicillin resistance in Escherichia coli, and resistance of Klebsiella oxytoca and Stenotrophomonas maltophilia to third-generation cephalosporins at 11.6% and 64.7%, respectively. Staphylococcus epidermidis resistance was noted in 25% of cases.³ Over time, resistance rates seem to progressively increase. Differences among studies reflect varying healthcare practices, neonatal unit types, healthcare staff qualifications, and local microbial reservoirs.

Several predictive factors for resistant EONBI pathogens were identified. Maternal blood group O positive was more frequently associated with sensitive strains (p = 0.012, OR = 0.409, 95% CI). Literature suggests that blood group O offers relative protection against severe infections such as SARS-CoV-2^13,16 and may also protect against resistant EONBI pathogens due to the presence of anti-B isoagglutinins in group O serum, which reduce susceptibility to Gram-negative enteric bacteria with B-type specificity,¹⁴ the predominant bacteria in our study (96.3%). Other infectious maternal conditions, mainly respiratory, ENT, or cutaneous, are common during pregnancy because physiological changes increase vulnerability.¹⁵

In studies by Ben Hamida et al. and Kloula, where resistant pathogens were frequent, maternal antibiotic exposure in the week before delivery was found in 61.1% and 12.2% of cases, respectively,^3,17 suggesting a link between maternal infections, antenatal antibiotic use, and resistance. In our study, maternal infections during pregnancy were associated with a higher risk of resistant bacteria (OR = 3.073, 95% CI [0.950–9.941], p = 0.003). The sometimes excessive or prolonged use of antibiotics in this vulnerable population likely contributes to resistance emergence.

Given the presence of resistant EONBI, evaluating the impact of hospitalizations during pregnancy is justified. Hospitalization during the third trimester was strongly associated with resistant pathogens (OR = 9.669, 95% CI [3.516–26.593], p < 0.001). Similarly, hospital stays of five days or longer correlated with resistant germs (OR = 0.845, 95% CI [0.717–0.996], p = 0.044). Kloula’s study reported hospitalization in the week before delivery in 18.29% of cases, mainly for premature rupture of membranes.³ A hospital stay longer than 48 hours is a major criterion for nosocomial infections¹⁸ and a key factor in the emergence of resistance.

In conclusion, EONBI represents a significant public health issue. The hospital environment during delivery and the neonatal period, often marked by resistant pathogens, influences infection risk, clinical presentation, and prognosis. Therefore, combating emerging antibiotic resistance, a serious neonatal threat, must be a global health priority.

Ethical approval

The study protocol was reviewed and approved by the Ethics Committee of Bougatfa Hospital (09/2022). Given the retrospective and anonymized nature of the data, the requirement for individual informed consent was waived.

Informed consent

Patient consent was waived due to the retrospective nature of the study and the use of anonymized data.