Etiology and antimicrobial resistance patterns of sepsis in infants 0–59 days old in Jimma, Ethiopia: a longitudinal study

Abstract

Introduction

In Ethiopia, empiric treatment for sepsis or possible serious bacterial infection (PSBI) in infants 0–59 days old is based on World Health Organization (WHO) guidelines. We aimed to assess the etiology, antimicrobial susceptibility and outcomes of empiric treatment in PSBI in Jimma, Ethiopia and created local antibiograms to assess empiric treatment guidelines in this setting.

Methods

We prospectively enrolled 363 infants 0–59 days of age admitted to Jimma University Hospital with PSBI over 12-months. Following institutional practice, blood and cerebrospinal fluid (CSF) cultures were collected; positive cultures were identified and evaluated for antibiotic susceptibility. We compared microbiologic results to the WHO guideline based empiric treatment selections at the hospital and evaluated the clinical outcomes at discharge and 30-days of age.

Results

Of 279 patients who had blood cultures obtained, 212(76.0%) were positive, yielding 216 isolates. Four CSF cultures were positive and were included in analysis of blood cultures due to the smaller number of isolates. The most common isolates were Klebsiella (31.8%), coagulase-negative Staphylococci (24.6%), and Staphylococcus aureus (11.6%). Of Klebsiella species, 87% were resistant to at least one commonly utilized antibiotic and 82% were resistant to first-line empiric antimicrobials. In-hospital mortality was 12.3% and it was highest (41%) in participants with Klebsiella. At 30-days of age, mortality for infants with positive culture was 6.6%.

Conclusions

Isolates from PSBI showed high rates of antibiotic resistance to first- and second-line antimicrobials. In this setting, the WHO empiric treatment guidelines inadequately treat infants admitted with PSBI, particularly those with Klebsiella. To provide the most effective care for PSBI in 0–59 days old infants, institutionally used guidelines should be customized to reflect local epidemiology and resistance patterns.

Introduction

In 2000, the United Nations prioritized decreasing under-five mortality in the Millennium Development Goals (MDGs), recognizing the critical role of developing local health system capacity in improving childhood morbidity and mortality [1]. Moreover, the sustainable development Goals (SDGs) and specifically, SDG 3.2 set the target of ending preventable death of newborns and children under 5 years of age, all countries aiming to reduce neonatal mortality to at least as low as 12 per 1000 live births and under 5 mortality to at least as low as 25 per 1000 live births by 2030 [2]. Even though Ethiopia was one of the few countries to achieve the MDG of a 67% decrease in under-five year old mortality, gains in neonatal mortality lagged substantially behind other age groups; contrary to the SDG target set, the neonatal mortality is in an increasing trend (increased from 29 per 1000 live births in 2016 to 33 per 1000 live births in 2019) [1–3]. Globally, neonatal sepsis accounts for one-third of neonatal deaths and annually, there are estimated over 3 million infants 0-59days old dying, over 500,000 of them being due to possible serious bacterial infections (PSBI) [4].

In infants 0–59 days old, symptoms and signs of serious bacterial infections like sepsis and meningitis are usually non-specific and overlapping, making distinction between the different spectrum of presentations difficult for health care providers. PSBI is defined as the presence of one or more of the below in infants 0–59 days old: fast breathing (respiratory rate of ≥ 60 per minute), movement only when stimulated, not feeding well on observation, temperature greater than or equal to 38 °C or less than 35.5 °C, severe chest in-drawing, convulsions, unable to feed at all, cyanosis, grunting, hypoxemia, lethargy/unconciousness [4–6].

The WHO recommendation for the management of PSBI in infants 0-59 days old is hospitalization and treatment with at least 7-days course of antibiotics (with a combination of injectable penicillin and gentamicin). If staphylococcal infection is suspected, a combination of cloxacillin and gentamicin is recommended [4–6]. However, inpatient treatment options are not accessible, acceptable or affordable to many infants and families in many of the developing countries. These challenges leave such infants without treatment and hence leading to unnecessary preventable deaths which is against SDG of ending preventable neonatal mortality [2, 5]. In order to overcome these challenges and improve access to the treatment of PSBI in infants 0–59 days old, WHO issued guidance in 2015 [5] which was updated in 2024 [6] for outpatient treatment of PSBI to be used when referral for inpatient treatment is not possible or refused in infants 0-59 days old [5]. The guidance simplifies antibiotic selection for infants 0–59 days of age with PSBI in the absence of clinical microbiological data. Even if different clinical trials have been conducted on these simplified treatment of PSBI in infants 0–59 days old and demonstrated the efficacy and safety of these treatments [7, 8], its implementation in real life has not been conducted. Hence, supported by the WHO, between 2016 and 2017, Jimma and Mekelle Universities, institutions in different regions of Ethiopia, implemented the simplified management of PSBI in infants 0–59 days old when referral is not possible or refused at primary health care unit level using the 2015 WHO treatment guidelines. The implementation studies conducted in the two settings have demonstrated that the simplified treatment was acceptable and feasible from the care givers and health care workers perspectives [9, 10]. The studies used for the development of the guidelines, including the African Neonatal Sepsis Trial (AFRINEST) [7, 8] multicenter trial, have focused almost exclusively on clinical outcomes, including treatment failure and mortality. However, these studies have not assessed the clinical microbiology and antimicrobial susceptibilities of PSBI in outpatient settings, so it remains unclear if the simplified antibiotic regimens recommended in the guidance are appropriate.

Notably, the guidelines are standardized across the globe. Studies from Ethiopia and other low- and middle-income countries (LMICs) have found that the bacterial species responsible for neonatal infection differ from those in high-income countries (HICs), with higher rates of Staphylococcus aureus, Escherichia coli, and Klebsiella species and lower rates of Streptococcus agalactiae or “Group B Strep” in LMICs [11–15]. Despite this discrepancy, in most of these studies, the antibiotics administered for PSBI in the inpatient setting in LMICs were similar to those administered in HICs: ampicillin or penicillin with gentamicin as first-line therapy and a 2nd or 3rd generation cephalosporin as second-line treatment. Some studies included oxacillin or cloxacillin as a first-line treatment, but this was uncommon. Moreover, several studies in Ethiopia and other LMICs have shown high rates of resistance to first- and second-line antimicrobials [11, 14, 15], and a recent review estimated resistance to both first- and second-line antimicrobials to be as high as 43–44% [16]. Taken together, the effectiveness of the guidelines is likely to differ by setting, and likely, not be as effective in LMICs.

Blood culture is the most important diagnostic test for identifying neonatal sepsis, allowing microbial identification and antibiotic susceptibility testing, which guide antibiotic selection. Unfortunately, due to the cost of blood culture, supply chain challenges, and laboratory capacity, many therapeutic decisions in LMICs including the study setting are made empirically using WHO or local guidelines [4]. The most common empiric treatment selections for PSBI were ampicillin and gentamicin, plus cloxacillin or a cephalosporin if there was concern for skin, bone, or joint infection. We aimed to compare the standard of care antibiotics provided per existing protocols to the antibiotic susceptibilities of the isolated microorganisms and develop local antibiogram to ascertain whether the existing WHO guidelines are appropriate for the treatment of PSBI in 0–59 days old infants admitted to Jimma University Hospital from March 2019 to March 2020.

Methods

Study design and setting

We did prospective, observational longitudinal study in infants 0–59 days old admitted with sepsis/PSBI to Jimma University Hospital. Jimma University Hospital is one of the tertiary level hospitals in Ethiopia; it is located in the Southwest part of the country serving around 20 million people in the sub-region. Diagnosis of sepsis was made by the treating physicians and the study just used the diagnosis made by the treating physicians. The WHO guideline is used for the diagnosis and treatment of sepsis in 0–59 days old infants in the setting.

Participants

All patients 0–59 days of age who were admitted to the Pediatric ward or Neonatal Intensive Care Unit (NICU) at Jimma University Hospital with the diagnosis of sepsis were eligible for enrollment. We excluded the infants with presumed hospital acquired infections.

The study was conducted between March 21, 2019 and March 23, 2020. However, we were unable to enroll participants from December 5, 2019 to January 11, 2020, while blood culture supplies were delayed in customs.

Participants’ enrollment and data collection

All patient samples were collected as part of routine diagnostics. Blood and cerebrospinal fluid (CSF) cultures were obtained from participants at the discretion of the treating physician as soon as possible after the diagnosis of sepsis was made by the reating physicians. Even if it was important to repeat blood culture for those not responding to treatment, we didn’t do that because of the scarcity of blood culture bottles. Obtaining consent and collecting study related data were carried out later on not to delay initiation of treatment. Patients were enrolled into the study only after securing written informed consent from parent/caregiver. Trained study nurses read the consent form to the parent/caregiver in their preferred language and offered sufficient time for questions.

Microbiologic analyses

Blood cultures were incubated using Benton-Dickinson BACTEC™ system. Positive blood cultures and all CSF cultures were inoculated onto blood, MacConkey, and chocolate agar plates (Oxoid England), incubated for 48 h at 35.0 °C in a CO₂ incubator, and identified using biochemical testing. Antibiotic susceptibilities were determined using Kirby-Bauer disc diffusion and classified as sensitive, intermediate, or resistant by Clinical and Laboratory Standards Institute (CLSI) breakpoints, version 2017 [17].

Clinical and outcome data collection

The study team abstracted data pertaining to the hospital course and microbiology results from participants’ medical charts. Data was entered into a secure, online study database in REDCap. Microbiologic results were provided to the treating physicians for subsequent use. Participants with positive blood and/or CSF cultures completed a 30-day follow up survey when returning to the hospital for a follow-up visit or via telephone call when an in-person visit was not feasible. Due to COVID-19 restrictions, telephone follow-up was required after March 23, 2020. In order to minimize/avoid lost-to-follow up, we tried to get the parents’/guardians’ phone number/s. If they don’t have, we tried to get their neighbors’ and/relatives’ numbers so that we can use all the possible options to collect the follow up data. If parents/guardians were unreachable on the first call, we made repeated attempts to reach out to them by using all the phone numbers they gave us. We declared lost-to-follow up if they were unreachable by 60 days of age (1 month after their follow up due date).

Outcomes

The primary microbiological outcomes of interest were culture positivity, culture isolate(s), and antimicrobial susceptibilities. Clinical outcomes were patient outcomes at hospital discharge and 30-day follow-up.

Data analyses

Statistical analyses were conducted using SAS (SAS Institute Inc., Cary NC), version 9.4 or higher. Descriptive statistics like frequency and mean and some tests of association like Chi square analyses were also computed to see association between the dependent and independent variables. A p-value of < 0.05 was used to define statistical significance. The number and frequency of culture positivity, culture isolate(s), antimicrobial susceptibilities, and antibiotic exposure prior to admission are summarized as narratives and presented as tables and figure. Data was de-identified prior to analysis and submission for publication.

Ethics

The study and methods were reviewed and approved by the University of Wisconsin-Madison Minimal-Risk Institutional Review Board, (protocol #2018 − 1076), and the Jimma University Institutional Review Board (Ref. No. IHRPGD/513/2018). Written informed consent was obtained from the parents/guardians of the infants.

Results

We prospectively enrolled a total of 363 0–59 days old infants, 217 (59.8%) were male, 338 (93.6%) were > 72 h old at admission, and 203 (56.2%) were evaluated at another site prior to admission; the remaining once were either referred from the labor and delivery room of Jimma University Hospital or they came directly to Jimma University Hospital without having visited another health facility. Of those evaluated elsewhere and transferred, 61.2% were evaluated at their local hospital; the remainder were evaluated at local clinics and health centers. The most frequently identified symptoms on presentation were poor feeding 150 (42.1%), difficulty breathing 148 (41.6%), and fever 121 (34.0%). A quarter of the participants 87 (24.0%) received antibiotics prior to admission to the hospital; majority of them receiving Ampicillin 75 (86.2%) and gentamicin 69 (79.3%). Similarly, upon initial admission to the hospital, majority of the participants received the same antibiotics, Ampicillin 304 (86.1%) and gentamicin 339 (96.0%). The mean (SD) duration of antibiotic treatment in the hospital was 9.6 (6.7) days; 95%CI (2, 20.5) (Table 1).

Table 1.

Demographic and clinical characteristics of 0–59 days old infants admitted with sepsis/PSBI to Jimma University Hospital, 2019/2020

Variable		Number (n)	Percent (%)
Gender (N = 363)	Male	217	59.8
	Female	146	40.2
Age at admission (N = 361)	< 72 h	23	6.4
	≥ 72 h	338	93.6
Evaluated elsewhere prior to admission to Jimma University Hospital (N = 361)	Clinic Health center Another Hospital No Unknown	12 66 123 157 3	3.3 18.3 34.1 43.5 0.8
Symptoms at admission (N = 356)	Poor feeding	150	42.1
	Difficulty breathing	148	41.6
	Fever	121	34.0
	Vomiting	60	16.9
	Cough	53	14.9
	Fast breathing	39	11.0
	Irritability	18	5.1
	Diarrhea	10	2.8
	Seizure	6	1.7
	Bulging fontanel	2	0.6
	Chest in-drawing	1	0.3
Antibiotics prior to admission (N = 363)	Yes	87	24.0
	No	98	27.0
	Unknown	178	49.0
Type of antibiotics received prior to admission (N = 87)	Ampicillin	75	86.2
	Amoxicillin	3	3.4
	Gentamicin	69	79.3
	Ceftriaxone	9	10.3
	Other	3	3.4
	Unknown	4	4.6
Initial antibiotics given on admission	Ampicillin	304	86.1
	Gentamicin	339	96.0
	Ceftriaxone	52	14.7
	Vancomycin	5	1.4
	Ceftazidime	3	0.8
	Metronidazole	3	0.8
	Ciprofloxacin	1	0.3
	Other	36	10.2
Antibiotics changed during course of hospitalization (N = 320)	Yes	128	40.0
	No	192	60.0

Over three fourth 279(76.9%) of enrolled patients had blood cultures done; over three fourth 212 (76.0%) of these blood cultures were positive, yielding 216 isolates. Four out of 41 CSF cultures were positive and given the low number of positive CSF cultures, we elected to include and analyze them together with positive blood cultures as “invasive bacterial infections.” The three most common isolates were Klebsiella species, Coagulase-negative Staphylococci (CoNS), and S. aureus. These organisms were cultured from over 40.0% of enrolled patients and accounted for almost 70.0% of all positive cultures (Table 2). Six organisms were isolated from 4 CSF samples; CoNS [2], S. pyogenes [1], Citrobacter [1], K. oxytoca [1] and 1 organism which was not identified. The organisms identified in the early onset group (< 72 h of age) did not differ significantly from the late onset group (≥ 72 h of age), however there were only 18 patients and 11 isolates in the early onset group. Due to the small sample size, these were not analyzed separately, and were analyzed with the late onset group. Receipt of antibiotics prior to being cultured did not significantly alter the likelihood of that culture growing out an organism; culture positivity rate was 71.3% in infants who received antibiotics before culture whereas it was 77.9% in those who didn’t receive antibiotics before culture (p = 0.24).

Table 2.

Organisms isolated (n = 224) from 0–59 days old infants admitted with sepsis/PSBI to Jimma University Hospital, 2019/2020

Organism	Frequency	Percent
Klebsiella spp	74	33
CoNS	55	24.5
Staphylococcus aureus	26	11.6
Other enterobacterales*	33	14.7
Citrobacter spp	16	7.1
Streptococcus spp	9	4.0
Pseudomonas aeruginosa	6	2.7
Enterococcus spp	5	2.2

*E.coli (10), Enterobacter spp (18), acinetobacter spp (3), Providencia spp (2) CoNS: Coagulase Negative Staphylococcus

The microorganisms identified in this study demonstrated very low rates of susceptibilities to the commonly used first-line and second-line antibiotics (Table 3; Fig. 1). Of the 74 isolates of Klebsiella species, 70 had susceptibilities available, with low rates of susceptibilities to cephalosporins and aminoglycoside. A substantial number of gram-negative isolates were resistant to carbapenems; 37.0% of Klebsiella isolates were resistant to imipenem and 17.0% of the isolates were resistant to meropenem. The first-line and second line empiric antibiotics recommended by the WHO and used in Ethiopia are ampicillin and gentamicin and ceftriaxone and gentamicin respectively, and majority of Klebsiella isolates were resistant to these antimicrobials (Table 3).

Table 3.

Antibiotic susceptibility of most commonly isolated organisms from 0–59 days old infants admitted with sepsis/PSBI to Jimma University Hospital, 2019/2020

Antibiotic	Kliebsella		CoNS		S. aureus
	S/R + S	S%	S/R + S	S%	S/R + S	S%
Cefixime	10/35	28.6
Cefotaxime	1/14	7.1
Ceftazidime	8/37	21.6
Ceftriaxone	8/40	20.0
Cefuroxime	3/25	12.0
Ciprofloxacin	18/52	34.6	24/41	58.5	6/15	40.0
Gentamicin	21/50	42.0
Imipenem	7/11	63.6
Meropenem	49/59	83.1
TMP-SMX			2/24	8.3	7/17	41.2
Clindamycin			24/40	60.0	11/18	61.1
Oxacillin			11/30	36.7	11/12	91.7
Vancomycin			7/11	63.6	2/5	40.0

TMP-SMX: Trimethoprim-sulfamethoxazole; CoNS: Coagulase Negative Staphylococcus

Fig. 1.

Selected Antibiotic Susceptibility Proportions for All Organisms isolated from 0–59 days old infants admitted with sepsis/PSBI to Jimma University Hospital, 2019/2020, Stratified by Gram Stain

There were 55 patients whose cultures grew CoNS species, which were not further speciated and we didn’t try to further differentiate whether these were contaminants or true infections. Of CoNS isolates, 36.6% were susceptible to oxacillin and 63.6% were susceptible to vancomycin. With low rates of empiric vancomycin use, most patients who grew this organism were started on inappropriate empiric therapy. In our study, we defined appropriate empiric therapy when the susceptibility pattern of the organism was found to be compatible with the antibiotics they were prescribed empirically.

There were 26 subjects whose cultures grew S. aureus. Of those with susceptibility data, 91.7% were susceptible to oxacillin and 40.0% were susceptible to vancomycin; S. aureus was not tested for resistance to penicillin/ampicillin per laboratory protocols. Given these resistance rates, at least 98% of these patients were started on inappropriate empiric therapy, as only 1.4% of patients received vancomycin on admission (Table 2) and even the vancomycin susceptibility rates were low (Table 3).

Overall, 123 (44.0%) patients who had cultures collected had a gram-negative organism isolated from their cultures (2 patients had 2 gram-negative organisms isolated) and 93 (34.4%) patients had a gram-positive organism isolated (2 patients had 2 gram-positive organisms isolated). Based on the empiric antibiotic selections, 69/123 (56%) of those growing a gram-negative organism might have been on inappropriate empiric therapy and 24/93 (25.8%) of those growing a gram-positive organism might have been on inappropriate therapy. Therefore, of 216 patients with a positive culture (212 blood cultures and 4 CSF cultures), up to 93 (44.0%) would be categorized as treated with inappropriate empiric therapy, based on our available susceptibility data.

Majority of the infants were discharged alive 297(86.7%) whereas the in-hospital mortality rate was 12.3%. Majority of those discharged alive have improved at the time of discharge 272(91.6%). At 30 days of age, majority of those who were discharged alive were still alive 241(83.7%) and in a good health condition 233(97.9%). None of the infants were re-admitted before 30days of age (Table 4).

Table 4.

Outcomes of 0–59 days old infants admitted with sepsis/PSBI to Jimma University Hospital, 2019/2020

Variable		Frequency	Percent (%)
Outcome at discharge (N = 342)	Alive	297	86.8
	Deceased	42	12.3
	Unknown	3	0.9
If alive, conditions at discharge (N = 297)	Improved	272	91.6
	Same	7	2.4
	Worse	1	0.3
	Unknown	17	5.7
Outcome at 30 day follow up (N = 288)	Alive	241	83.7
	Deceased	19	6.6
	Unknown	28	9.7
Condition at 30 day follow up (N = 238)	Back to normal	233	97.9
	Improved, but not yet back to normal	5	2.1
Feeding at 30 day follow up (N = 241)	Feeding well	237	98.3
	Feeding poorly	4	1.7
Activity at 30 day follow up (N = 239)	Moves well	237	99.2
	Moves little	2	0.8
Growth at 30 day follow up (N = 241)	Weight gain	234	97.1
	Weight same	5	2.1
	Unsure	2	0.8

Of those with Klebsiella, 18 (24.0%) died while in hospital; 17/18 (94.0%) had isolates that were resistant to the first-line therapy, while only 1/18 (6%) had an organism that was susceptible to first-line therapy. Among the infants with Klebsiella isolates that were resistant to first-line therapy, mortality was 28.8% (17/59) compared to 7.7% (1/13; p = 0.16) for those susceptible to first-line therapy. The likelihood of in-hospital mortality was higher if cultures grew Klebsiella, which accounted for 31.8% of isolates and 41% of mortality.

Follow-up of those infants who had a positive culture showed mortality rate of 6.6% at 30 days of age. Outcome at follow up was available for 46 subjects from whom Klebsiella was isolated; there was no mortality in subjects with organisms for which no resistance was detected (0/7) vs. 10.0% (4/39) in subjects with organisms resistant to at least one of the antibiotics (p = 0.999).

Discussion

The culture positivity rate (76.0%) in this study was significantly higher than other studies (17.7% in the NeOBS study and 6.8% in the BARNANDS study) [18, 19]. This significant difference might be due to the difference in the study settings (single center and single country in our study versus multicenter, multiple countries and continents in the other two studies. Moreover, the other studies had the criteria they used to consider the isolates as contaminants and discarded some of the isolates which were categorized as contaminants which we didn’t.

While global guidelines such as the WHO recommendations for empiric treatment of PSBI [5] provide a valubable framework for standardizing care, our findings underscore the importance of incorporating local microbiological data into local treatment protocols. Accurate, uptodate local antibiograms and access to real-time culture results could significantly enhance the appropriateness of antibiotic selection for infants with PSBI or sepsis. Tailoring WHO recommendations to reflect local microbiological epidemiology and antimicrobial susceptibility patterns would likely improve treatment efficacy, reduce the emergence of antimicrobial resistance, preserve the utility of existing antibiotics, lower healthcare costs and reduce neonatal mortality.

A key objective of this study was to assess the effectiveness of the empiric antibiotics recommended by the WHO commonly used in low and middle income contry setting. Our study showed that antibiotics exposure (majority receiving Ampicillin and gentamicin) prior to blood culture collection did not significantly reduce culture positivity rates. This could possibly suggest the resistance of the isolated microorganisms to the first line antibiotics the participants received prior to blood culture collection. It could also possibly indicate the quality of culture media used. The organisms isolated exhibited high resistance rates to ampicillin and cephalosporins and moderate resistance rates to gentamicin, suggesting substantial resistance in the community to first-line antibiotics.

Given that the predominant pathogens in this population were Klebsiella spp., CoNS, and S. aureus, our findings suggest that the current WHO empiric treatment guideline used at Jimma University Hospital may be inadequate for a substantial portion of 0–59 days old infants with sepsis/PSBI. Ampicillin demonstrated minimal efficacy, with resistance rates between 92 and 100%. Although gentamicin showed better activity, it is insufficient as monotherapy due to limited coverage of gram-positive organisms. Similarly, second-line agents such as advanced generation cephalosporins exhibited resistance rates approaching 80–90%. Alarmingly, carbapenems also showed higher than expected resistance rates.

Vancomycin resistance among S. aureus isolates was high in our sample (3/5), though only 5 isolates were tested. Eventhough vancomycin-intermediate and vancomycin-resistant S. aureus (VISA/VRSA) remain rare globally [20], a recent systematic review done on the prevalence of VRSA in Ethiopia has demonstrated a pooled prevalence of as high as 47.7% in certain parts of the country [21]. Our susceptibility testing relied on the Kirby-Bauer disc diffusion method, which, while commonly used, is less reliable than MIC testing. Notably, we observed concerning rates of vancomycin-resistance among CoNS isolates, which may reflect hetero-resistance-a phenomenon previously reported in other countries in the region [22]. It could also be indicative of the limitations of using disc diffusion methods in assessing vancomycin susceptibility.

Interpreting the high prevalence of CoNS and resistance rates among gram-positive organisms posed a challenge. Although CoNS are often considered contaminants, they can be significant pathogens in very premature and low birthweight infants [23]. Not all CoNS species carry the same pathogenic potential; for instance, S. lugdunensis is associated with invasive infections such as osteomyelitis, endocarditis, and bacteremias [24], whereas S. epidermidis is typically linked with central line associated infections [25], devices not used in the study setting during the study period. Future efforts should prioritize improved infection control practices, species-level identification of CoNS, and enhanced training in sterile collection techniques to better distinguish between contamination and true infection.

The strength of the study includes the use of prospectively collected data and the use of automated Bactec system for the isolation of bacteria. This study has several limitations. It is a single center study with small sample size, affecting generalizability to the wider population. Clinical data collection relied on caregiver interviews and hospital records’ reviews, leading to some missing information. Enrollment was also interrupted at times, as detailed in the methods. Furthermore, susceptibility testing was on local laboratory practices and could not be independently validated. Some of the CoNS isolated might be contaminants than true infections and we didn’t attempt to differentiate between the two. Additionally, some of the outcome measurements were assessed over the phone which is prone to subjectivity.

Despite these limitations, our findings highlight the critical need for roboust, locally generated microbiological data to inform empiric treatment strategies. High-quality microbiology services are essential for monitoring resistance patterns rates in LMICs, yet the cost and infrastructure required for routine blood cultures remain significant barriers. Global initiatives are working to develop empiric guidelines tailored to regions with high prevalence of extended-spectrum-beta-lactamase (ESBL) and carbapenem-resistant organisms [26]. For example, the BARNARDS study, conducted across seven LMICs, found reduced mortality with empiric regimens that include amikacin and ceftazidime [19]. However, such guidelines must be continually updated based on surveillance data. To provide high quality care for neonates with PSBI, empiric antibiotics protocolos must be grounded in local antimicrobial resistance patterns whenever possible.

Conclusion

Throughout the course of this study, Klebsiella, CoNS, and S. aureus, were identified as the predominant pathogens responsible for the majority of microbiologically confirmed serious bacterial infections in infants 0–59 days old. These isolates exhibited high rate of resistance to first- and second-line antibiotics, as well as moderate rates of resistance to carbapenems. These findings raise significant concerns about the adequacy of empiric antibiotics regimens currently recommended by WHO guidelines for treating PSBI in infants 0–59 days old in this setting. To ensure the most effective care for infants with PSBI, treatment guidelines must be adapted to reflect local antimicrobial resistance patterns. We recommend that the Ethiopian Ministry of Health consider the following actions:

1. Designate sentinel hospitals with the capacity to perform culture and antimicrobial susceptibility testing.

2. Support the development of local antibiograms at these sites to inform national or regional PSBI treatment guidelines.

3. Implement regular updates to empiric treatment protocols based on evolving microbiologic and resistance data.

Relying solely on global guidelines like the WHO guidance without incorporating local data risks inappropriate empiric therapy, poorer clinical outcomes, and increased preventable neonatal and infant mortality. Strengthening microbiological surveillance and tailoring treatment recommendations accordingly are essential steps toward improving neonatal health outcomes in LMICs.

Author contributions

DG, AA, DS and JH secured funding. DG, AA, DS, JH, JC, MB, MS and MG developed the proposal. DG, MB, MS and MG acquired the data. DG and JC analyzed the data. DG and MB wrote the initial draft of the manuscript. All authors reviewed and approved the manuscript.

Data availability

No datasets were generated or analysed during the current study.

Declarations

Competing interests

James H. Conway: research funding from US Centers for Disease Control to study viral transmission in Wisconsin schools; scientific consulting agreements with: Pfizer, Merck, Moderna and GSK vaccine programs – no relation between those projects and this work. All other authors have no any conflict of interest related to the work.