All newborns admitted to the neonatal intensive care unit (NICU) during the study period were eligible. Newborns discharged against medical advice or transferred to another healthcare facility were excluded from the analysis, as were those with incomplete medical records.

The study database did not contain a separate list of eligible and excluded newborns. Therefore, exclusions could only be described based on the available analytical data. This point was taken into account when interpreting a possible selection bias.

Data were extracted from patients’ medical records and treatment sheets, as well as from the medical biology department’s database. Variables included sociodemographic profile, reasons for hospitalization, infectious risk, laboratory data, and blood culture results in newborns. Regarding prescribed antibiotics, in addition to consumption data, dosage, duration of treatment, and any changes or interruptions in treatment after obtaining microbiological results were recorded. Missing laboratory data were considered as such and were not imputed. For the quality assessment of prescriptions, the absence of blood cultures limited the analysis of therapeutic adjustments guided by microbiological results; these cases were interpreted separately from those for which a blood culture result was available but not used for therapeutic adjustment.

In this study, early neonatal bacterial infection was classified as probable or microbiologically confirmed and defined as occurring within the first 72 hours of life. A confirmed early neonatal bacterial infection was defined as a clinically consistent infection with a positive blood culture validated by the medical biology department. A probable early neonatal bacterial infection was defined by the presence of compatible maternal or perinatal risk factors and/or neonatal clinical signs, with or without elevated inflammatory biomarkers, in the absence of microbiological confirmation.

Two clinical risk assessment tools were used:

The Houenouscore [7] (Table 1), developed for screening suspected maternal-fetal infections, whereby neonates classified with moderate-to-high risk were deemed eligible for antibiotic therapy.The RIB-MFscore [8] (Table 2), employed to inform the decision to initiate antimicrobial treatment.

Table 1.Houenou infectious score [7].

Interpretation: 1 ≤ Score 1 ≤ 2: Low risk; 3 ≤ Score 2 ≤ 4: Moderate risk; Score 3 ≥ 5: High risk. *APGAR: Appearance, Pulse, Grimace, Activity, Respiration.

Table 2.Classification of patients according to the maternofetal bacterial infectious risk (RIB-MF) before paraclinical investigation [8].

The investigators did not directly participate in blood sample collection, blood culture processing, or microbiological identification. In accordance with standard institutional practice, neonatal blood cultures are obtained by peripheral venipuncture or, when clinically indicated, by umbilical catheter after antisepsis. It is recommended to collect at least 1 mL of blood, and up to 2 mL when possible, particularly in newborns who have previously received antibiotics. Blood cultures are incubated for at least 5 days. Positive blood culture results and bacterial identification were validated by medical biologists according to standard laboratory procedures. However, detailed information on the actual blood volume collected, the number of blood culture bottles obtained per neonate, the antiseptic agent used, and adherence to recommended sampling procedures was not systematically available in the study database. The specific criteria used to distinguish contamination from true bloodstream infection, particularly for coagulase-negative staphylococci, were also unavailable and could not be independently reassessed by the investigators.

To assess antibiotic use pressure, we computed the penetration rate (number of antibiotic treatment days per 1000 days of hospitalization) [9], applying the following formula:

where: PI = Penetration index; Q = Quantity of antibiotic consumed (in grams); P = Number of patients-day in hospital; DDD = Defined Daily Dose.

The primary criterion was the proportion of antibiotic prescriptions deemed appropriate according to a composite quality criterion. This criterion was predefined according to the neonatology department’s internal protocol, aligned, where applicable, with french neonatology guidelines. Five aspects were evaluated:

the justification for the indication; the appropriate choice of antibiotic; the dosage within the recommended range (mg/kg/day); the appropriate duration of treatment; and reassessment or adjustment based on blood culture results and clinical progress.

Prescriptions were classified as appropriate when all five predefined criteria were met. Inappropriate prescriptions were further categorized as debatable or unacceptable. A prescription was considered debatable when the indication and antibiotic choice were appropriate but one or more deviations concerned dosage, treatment duration, or treatment reassessment based on microbiological findings. A prescription was considered unacceptable when the indication for antibiotic therapy was not justified, the antibiotic choice was inappropriate for the clinical condition, an unnecessary antibiotic combination was prescribed, or the treatment regimen posed a potential risk to patient safety.

Data were analyzed descriptively. Qualitative variables were summarized as counts and percentages. For the main proportions relevant to the study objectives, 95% confidence intervals (95% CIs) were calculated using the Wilson score method. Given the single-center design, limited sample size, and incompleteness of several laboratory variables, no multivariable analysis was performed.

For the purposes of our study, an administrative procedure was followed to obtain approval from the Medical and Scientific Directorate of the Angré University Hospital Center for access to patient records. Data were collected confidentially and with patient anonymity maintained.

A total of 139 neonates were included in the study. The majority of patients (95%) were admitted to the neonatal unit within the first 24 hours of life. A male predominance was observed, with a sex ratio of 1.57. Nearly half of the neonates (49.6%) were born at term, and 52.5% were delivered by cesarean section. The main reasons for hospitalization were an APGAR score < 6 (54.7%), respiratory distress (46.8%), and prematurity (23.7%).

Laboratory tests were performed in 84.2% of newborns (117/139), including 21 C-reactive protein (CRP) measurements and 96 procalcitonin measurements. Elevated inflammatory markers were detected in 41% of newborns (48/117), with 8 positive CRP results and 40 elevated procalcitonin levels.

The proportion of neonates classified as requiring antibiotic therapy differed markedly between the Houenou score and the RIB-MF classification (Table 3).

Early neonatal bacterial infection was ultimately the most frequent discharge diagnosis (114 cases).

Table 3.Distribution of neonates according to infection risk.

Among 114 newborns for whom blood cultures were requested, results were available for 63 (55.3%; 95% CI: 46.1 - 64.1). The reasons why blood cultures were not performed or completed in the remaining cases were not systematically recorded in the study database. Of the 63 newborns for whom blood culture results were available, 34 had positive cultures (54.0%; 95% CI: 41.8 - 65.7). Microbiological confirmation was obtained in 34 of the 114 newborns treated for early neonatal bacterial infection (29.8%; 95% CI: 22.2 - 38.8). Positive blood culture results and bacterial identification were validated by medical biologists according to routine laboratory procedures. Gram-positive cocci, mainly staphylococci, were the most frequently identified microorganisms (Table 4).

Table 4.Distribution of identified microorganisms.

Antibiotics comprised 53.4% of the medications utilized in neonatal therapy. The general profile of drug consumption is presented in Figure 1. Three principal antibiotic classes constituted the majority of prescriptions, as summarized in Table 5.

Table 5.Antibiotic consumption by family and molecule.

Figure 1. Drug consumption profile. *Anatomical Therapeutic Chemical; **Number of defined daily doses per 1000 hospital days.

Figure 2 outlines the antibiotic prescribing strategy, which included a systematic re-evaluation of the treatment after 72 hours.

Deviations from guidelines were identified, primarily related to treatment adjustment according to blood culture results, dosage and treatment duration (Table 6). In total, 34.2% of all prescriptions were classified as inappropriate.

D: Day.

Figure 2. Antibiotic prescribing strategy.

Overall, 75 of 114 antibiotic prescriptions were classified as appropriate, whereas 39 prescriptions were classified as inappropriate (34.2%; 95% CI: 26.1 - 43.3). The main deviations from the predefined appropriateness criteria were unjustified continuation of antibiotic therapy after blood culture results became available, non-compliant dosage, and non-compliant treatment duration (Table 6)

Table 6.Pharmaceutical assessment of antibiotic prescriptions.

A male predominance was observed in our study, consistent with existing epidemiological data from African settings. This finding is consistent with the documented increased susceptibility of male neonates to NBI in settings such as Bamako (57%) and Yaoundé (63%) [10][11], although geographical variations exist, as evidenced by the sex ratio of 0.8 observed in Maradi [12].

The early admission of hospitalized newborns corroborates the early onset of NBI in our cohort, with most admissions occurring within the first 24 hours of life. This profile aligns with recent Malian observations [13] and fits within current definitions of early-onset neonatal infections [14].

Analysis of perinatal factors revealed a balanced distribution between cesarean sections and vaginal deliveries, reflecting variability in obstetric practices. Disparities exist in the literature: Kamaye [15] reported 61.5% cesarean sections, while Coulibaly etal. [13] in Bamako observed a predominance of vaginal deliveries (84.6%).

The finding that prematurity was not a major risk determinant in our cohort, in contrast to reports by Djidi [10] and Cohen etal. [14], underscores the heterogeneity of risk profiles across different settings.

The observed discrepancy between the Houenou and RIB-MF scores in assessing the risk of neonatal infection significantly impacted antibiotic prescribing. This discrepancy raises important questions regarding the standardization of risk assessment tools in African settings, where most studies rely on conventional maternal and neonatal clinical risk factors and medical history [15][16]. In our study, the RIB-MF score identified all newborns who were ultimately treated for probable early neonatal infection, while the Houenou score classified a smaller subgroup as being at moderate or high risk. This discrepancy suggests that the RIB-MF score may be more sensitive in identifying at-risk newborns and thus reducing the likelihood of undertreatment; however, its use could also increase antibiotic exposure in settings where microbiological confirmation is limited. The validity, sensitivity, and specificity of these two tools have not been fully established outside the contexts in which they were initially developed; therefore, their results should be interpreted as clinical decision-support tools rather than definitive diagnostic tests. These results underscore the need for locally conducted validation studies comparing the performance of clinical risk scores to standardized microbiological and clinical reference criteria. However, because the RIB-MF classification contributed to the initial therapeutic decision, the concordance between this score and the final diagnosis of probable infection should not be interpreted as independent evidence of diagnostic accuracy.

Our diagnostic approach systematically combined blood cultures with inflammatory biomarkers. Although the diagnostic performance of C-reactive protein (CRP) and procalcitonin (PCT) in the early stages of infection remains limited, serial measurement of these biomarkers is a well-established strategy to support decisions regarding the early discontinuation of antibiotic therapy [14]. The microbiological profile observed in our study, characterized by a predominance of Gram-positive cocci, particularly staphylococci, is consistent with data reported in West Africa [13][15]. Although positive blood culture results have been validated as microbiologically significant by medical biologists in routine laboratory practice, the predominance of coagulase-negative staphylococci warrants cautious interpretation, as the specific criteria used to distinguish contamination from true infection were not available for independent reassessment. By contrast, our results differ from those reported in the European context, particularly in France. European data indicate that maternal-fetal infections are predominantly caused by Group B Streptococcus and Escherichiacoli [14]. These differences may be attributed to variations in local epidemiology, screening practices, and socioeconomic factors. Additional comparative and multicenter studies are warranted to better understand the underlying factors responsible for these regional differences and to optimize prevention strategies accordingly.

The restriction to four main pharmacological classes reflects the persistent difficulties encountered in neonatal pharmacology in Africa, where marketing authorization for pediatric drugs remains limited. The predominance of antimicrobial use reflects the high prevalence of infections in our setting.

Therapeutic protocols aligned with practices in other Ivorian university hospitals [16] but differed from West African regimens favoring ceftriaxone [13][15] First-line treatment was based on the combination of cefotaxime and gentamicin, while the combination of ofloxacin and amikacin was used as second-line therapy.

The progressive phasing out of ceftriaxone in Côte d’Ivoire was driven by safety imperatives (calcium precipitation and risk of fetal erythroblastosis), illustrating how practices evolve based on pharmacological data [17]. In contrast, the divergence from French guidelines, which recommend amoxicillin-gentamicin combination [18], might be explained by local epidemiological specificities.

The high consumption of third-generation cephalosporins raises legitimate concerns about the emergence of bacterial resistance. This study argues for the implementation of antimicrobial resistance surveillance programs, the adaptation of therapeutic protocols to local epidemiology, the strengthening of diagnostic capacities and the continuing education of healthcare personnel.

Approximately one-third of antibiotic prescriptions did not meet the predefined appropriateness criteria. This finding is broadly consistent with the observations reported by Goulet etal. [19], while also reflecting context-specific prescribing challenges. The main deviations concerned treatment adjustment in response to blood culture findings, dosage issues, and treatment duration.

According to Maury etal., optimal antibiotic prescribing requires the simultaneous application of a justified indication, an appropriate drug, an adequate dosage, and adjustment based on microbiological results [20]. However, non-compliance should not be interpreted solely as individual prescriber behavior. Several systemic constraints can explain the observed discrepancies: delayed blood culture results, insufficient laboratory capacity, difficulty in obtaining neonatal blood samples, socioeconomic barriers limiting access to diagnostic tests, clinical instability of newborns, and fear of undertreatment in cases of potentially severe sepsis. These constraints may incentivize clinicians to continue antibiotic therapy even when available microbiological data do not fully justify continued treatment. Therefore, an effective antimicrobial stewardship intervention in this context should combine prescriber training with diagnostic management, improved turnaround time for laboratory tests, a standardized 48- to 72-hour review, pharmacy-supported auditing and feedback, and local resistance monitoring.

This study has several strengths, including its prospective design, its focus on a neonatal unit in a resource-constrained setting, and its combined assessment of infection risk, antibiotic consumption, and prescription quality. However, several limitations should be acknowledged. The study was conducted at a single center over a short three-month period, limiting the generalizability of the findings and precluding the assessment of seasonal variations. Blood culture results were unavailable for a substantial proportion of neonates for whom they had been requested, and the reasons for these missing results were not systematically recorded. Consequently, the impact of missing microbiological data on treatment reassessment and prescription appropriateness could not be formally quantified. In addition, because the specific criteria used to distinguish contamination from true bloodstream infection were unavailable for independent reassessment, microbiological findings, particularly those involving coagulase-negative staphylococci, should be interpreted with caution. Finally, the use of DDD-based indicators is limited in neonates because DDDs are based on adult dosing assumptions and do not reflect weight-based neonatal exposure. The DDD approach was retained as an aggregate measure of antibiotic consumption rather than as an indicator of individual neonatal exposure. Days of therapy would provide a more appropriate complementary indicator in future studies.