Neonatal Sepsis in Low- and Middle-income Countries – Where Are We Now?

Global neonatal deaths have declined substantially in the last two decades, but neonates (<28 days of age) now contribute to almost half of all deaths in children under-five according to a World Health Organization (WHO) report.¹ The pace of neonatal mortality reduction is slowest in low- and middle-income countries (LMIC), with neonatal mortality rates increasing to unprecedented levels in some countries. For example, in Zimbabwe, neonatal mortality rates (NMR) have increased from 24 to 37 deaths/1000 live births from 2006 to 2024, compared to the current NMRs of 1-3/1000 live births in Europe and North America.² At least 60 African and Asian countries are unlikely to achieve the Sustainable Development Goal target of under 12 neonatal deaths/1000 live births by 2030.³ In this review, we discuss key challenges and potential solutions for neonatal sepsis in LMIC (Table 1).

Table 1. Key challenges and potential solutions for neonatal sepsis in low- and middle-income countries.

Research area	Key challenges	Potential solutions
Neonatal sepsis burden and impact	Neonatal sepsis remains a leading cause of morbidity and death in LMIC. The lack of standardized surveillance definitions, sepsis diagnostics and harmonized reporting contributes to underestimation of neonatal sepsis burden and impact in LMIC. Sepsis-attributable neonatal mortality rates in LMIC range from 10-29%, although data on long-term impacts of neonatal sepsis are lacking.	Develop standardized neonatal sepsis surveillance definitions, including definitions for culture-negative sepsis. Facilitate aggregation of clinical and laboratory sepsis data globally by using standard variables, identification of the source population type i.e. neonates, and reduce barriers to data sharing. Embed and report clinical follow-up data to track the cost, clinical impact and long-term neuro-developmental outcomes.
Neonatal sepsis diagnostics	Diagnostic confirmation of neonatal sepsis is challenging in LMIC owing to lack access to microbiology laboratories and accurate, affordable point-of-care tests. The utility of blood cultures in neonatal asepsis diagnosis in LMIC is limited by the high costs, low sensitivity, slow laboratory turnaround times and high contamination rates.	The WHO recently released a Target Product Profile for neonatal sepsis diagnosis, which could include a point of care test in isolation or in combination with clinical features and risk factors. The potential to use artificial intelligence-enabled diagnostic algorithms for neonatal sepsis may have widespread utility in LMIC.
Pathogens, AMR and infection classification and empiric antibiotic therapy	Gram negative bacterial and AMR pathogens predominate as causes of neonatal sepsis in LMIC. AMR rates vary by setting and pathogen: aminoglycosides (42-69%), 3rd generation cephalosporins (59-84%), and carbapenems (13-82%). The historical classification into early-and late-onset neonatal sepsis, implying maternal versus healthcare-acquired transmission is outdated, and may lead to inappropriate empiric antibiotic selection. Empiric antibiotic treatment guidelines for neonatal sepsis are inadequate for current pathogen and AMR profiles in LMIC.	Improved access to microbiology services and laboratory strengthening is crucial to improve representation of neonatal sepsis data from LMIC. More clinical trials of new and repurposed antibiotics should be conducted in LMIC neonates with sepsis to enhance the evidence base for informing guidelines. Improved tools for developing empiric antibiotic treatment regimens for neonatal sepsis should be used i.e. WISCAs. Research on antibiotic stewardship interventions targeting hospitalized neonates in LMIC should be prioritized.
Prevention of neonatal sepsis	Suboptimal intrapartum and postnatal infection control practices contribute to the high rates of HAI in LMIC neonatal units. The evidence base for neonatal sepsis prevention interventions in LMIC is limited.	Bundled and multi-modal interventions targeting device-associated infections should be prioritized. Probiotic supplementation for preterm neonates is a promising intervention. For prevention of community-onset neonatal sepsis, use of clean birth kits, and chlorhexidine cord cleaning are evidence-based interventions.

Neonatal sepsis epidemiology and antimicrobial resistance in LMIC

Morbidity and mortality from neonatal sepsis remains high in LMIC, with an estimated 3 million cases and 570 000 neonatal deaths globally per year.⁴ Approximately a quarter of all neonatal deaths are caused by sepsis⁵, with estimates of sepsis-attributable neonatal mortality rates varying from 10 to 29% in LMIC.^4,6–7 Conservative estimates from sub-Saharan Africa report a loss of up to 8 million disability adjusted life years and an economic burden of over $400 billion annually due to neonatal sepsis.⁸ There are, however, no data available that accurately quantify and cost the long-term neurodevelopmental morbidity experienced by survivors of neonatal sepsis in LMIC.

In a recent systematic review of neonatal sepsis epidemiology globally, Gram-negative bacterial pathogens predominated as causes of early-onset (days 0-2) and late-onset (days 3-27) sepsis.⁹ The traditional division of neonatal sepsis by timing of onset, reflecting maternal versus healthcare-acquired transmission, is increasingly contested, especially in LMIC settings.^9,10 Klebsiella pneumoniae is the dominant neonatal sepsis pathogen overall in LMIC, while Group B Streptococcus and Escherichia coli account for 70% of early-onset neonatal sepsis cases.⁹ Neonatal Gram-negative sepsis is inextricably linked to the global problem of antimicrobial resistance (AMR) as Gram-negative bacteria exhibit high levels of resistance to multiple antibiotic classes. The association between Gram-negative bacteria and AMR compounds problems of providing effective antibiotic therapy for neonatal sepsis and contributes to increasing sepsis-attributable neonatal mortality rates. There is considerable heterogeneity in the pathogen spectrum and AMR profiles by postnatal age, place of infection onset (community versus healthcare facility), level of care (district/central, ward/intensive care) and region or country.¹¹

Data on the proportion of neonatal sepsis episodes caused by AMR pathogens are limited and mainly derived from tertiary hospitals, underscoring the need to develop robust surveillance and reporting systems in all LMIC and at all levels of care. Several large neonatal sepsis studies (2011-2024) including NeoOBS⁶, DENIS⁷, BARNARDS¹², BabyGERMS¹³ and CHAMPS¹⁴ have confirmed the dominance of AMR Gram-negative bacterial pathogens in LMIC (see Figure, Supplemental Digital Content 1). Reported AMR rates for commonly used neonatal antibiotic regimens vary by facility type and region, including 42-69% resistance to aminoglycosides, 59-84% resistance to 3rd generation cephalosporins, and 13-82% resistance to carbapenems.^6,7,12–14 Given the substantial AMR profile of neonatal pathogens in LMIC, most empiric antibiotic treatment guidelines recommending penicillin plus gentamicin or cephalosporin-based regimens are likely to be ineffective (discordant) owing to bug-drug mismatch.

Encouragingly, the 2024 update to the Global Research on Antimicrobial Resistance (GRAM) Project, reported decreases in AMR-related deaths in neonates and children (1990-2021), through reductions in Streptococcus pneumoniae incidence from vaccination and bacterial enteric pathogens through improved water sanitation and hygiene (WASH) practices.¹⁵ However, in 2021, global AMR-attributable mortality increased considerably for some typically healthcare-associated neonatal pathogens including K. pneumoniae and Acinetobacter baumannii. Together with E. coli, these common neonatal pathogens producing extended-spectrum β-lactamases and carbapenemases, severely limit antibiotic treatment options for neonates with sepsis in LMIC.

Challenges in neonatal sepsis surveillance and diagnosis in LMIC

The burden and impact of neonatal sepsis and AMR in LMIC may be substantially underestimated, as most hospitals (particularly in Africa) have no, limited or intermittent availability of blood cultures, laboratory consumables and other essential laboratory tests required to diagnose neonatal sepsis.¹⁶ Many deaths currently ascribed to prematurity in LMIC, may be directly or indirectly attributable to infection, particularly in infants frequently excluded from surveillance programs e.g. extremely low birth weight babies (ELBW) < 1000g, and infants still hospitalized beyond day 28 of life. With enhanced molecular pathogen detection at a large South African neonatal unit, infection was shown to be the immediate or underlying cause of mortality in 57% of neonates.¹⁷ Very preterm infants and those still hospitalized beyond 28 days of life are important sentinel populations to include in neonatal infection surveillance, as they have immature innate and acquired immunity, high rates of invasive device use and exposure to multiple courses of antibiotic therapy during extended hospital stays of up to 3 months.

Other obstacles to accurate estimation of neonatal sepsis and AMR burden in LMIC are the lack of standardized neonatal sepsis surveillance definitions and the failure to aggregate and harmonize clinical and laboratory data globally. Although global AMR and antibiotic use surveillance reports exist for common infectious syndromes and selected bacterial pathogens, they do not currently report data separated by population type i.e. neonatal, pediatric, and adult sepsis.¹⁸ Similarly, neonatal databases seldom include extensive neonatal sepsis-related information and definitions that would allow LMICs to report clinical sepsis and antibiotic use data. Other important omissions in many LMIC neonatal databases include lack of data on sepsis-related prolongation of hospital stay, repeated courses and duration of antibiotic use, infection-related complications, neurodevelopmental disability, laboratory diagnostic use and excess hospital costs.

Although neonatal sepsis data from LMIC are increasingly published, the ability to collect data at all facilities and conduct pooled, and therefore more powerful analyses is hampered by: lack of protocols and centralized data servers to harmonize neonatal sepsis data; difficulty linking laboratory and clinical data owing to lack of a single patient identifier; and difficulty accessing data due to complex data sharing agreements and custodian permissions.¹⁹ For community-onset neonatal sepsis, the introduction of the WHO’s presumed serious bacterial infection (pSBI) criteria has facilitated standardized reporting.²⁰ For diagnosis of neonatal sepsis in hospitalized patients, clinical prediction scores have been developed e.g. the NeoOBS sepsis severity and recovery scores⁶ and the NeoHoP²¹ score for healthcare-associated infection.

Even with the use of standardized clinical tools, confirming a diagnosis of neonatal sepsis in LMICs remains challenging. Neonatal sepsis often presents with non-specific symptoms and signs, requiring additional laboratory-based tests to improve diagnostic accuracy. However, many LMIC healthcare facilities lack accurate, affordable, and point-of-care sepsis diagnostics, with underutilization or limited access to diagnostic microbiology laboratory services. In a large study at 61 hospitals in four African countries, 70% of neonates received broad-spectrum antibiotic therapy but only 6% of neonates had a blood culture specimen submitted. These data suggest major underutilization of available laboratory services, and likely underestimation of neonatal sepsis rates.¹⁶ Even in settings where blood cultures are routinely submitted in neonates with suspected sepsis, challenges include the high cost and low sensitivity of blood cultures, excessive contamination rates (>30% in some studies), and slow turnaround times for traditional microbiology processing. Failure to submit blood cultures may contribute to excess sepsis deaths, particularly for neonates with AMR/fungal sepsis where there is delay in switching to an effective antimicrobial agent. Similarly, the lack of robust clinical and laboratory surveillance systems in most LMIC hospitals results in delayed recognition and action to contain outbreaks, with consequently large numbers of affected neonates and high mortality experienced.²²

Challenges in neonatal sepsis treatment in LMIC

Selecting treatment for neonatal sepsis in LMICs is complex, given limited access to rapid molecular diagnostic tests²³, unaffordability of new antibiotic agents and frequent mismatch of empiric antibiotic guidelines to the AMR and pathogen profile in LMIC neonatal units. Furthermore, neonates who receive discordant empiric antibiotic therapy have been shown to have a three-fold increased risk of sepsis-attributable death²⁴, highlighting the need for prompt access to effective empiric antibiotics in settings where Gram-negative and AMR pathogens predominate. Another potential consequence of the lack of local data on prevalent neonatal pathogens and AMR profiles, is the indiscriminate use of very broad-spectrum, prolonged courses of empiric antibiotic therapy in LMIC neonates with suspected sepsis.

The slow pace of antibiotic discovery and development and underrepresentation of neonates in antibiotic clinical trials is a further problem that limits the evidence base to inform sepsis guidelines.²⁵ In the last 25 years, only a handful of clinical trials of antibiotics were conducted in the neonatal population to inform dosing recommendations for this unique population. This problem is most apparent for data informing treatment of carbapenem-resistant infections in neonates, where pharmacokinetic and safety data for old and new antibiotics are extremely limited. To ensure equitable access to new antibiotics for vulnerable neonates in LMIC, future studies should include research on improved tolerability, safety, ease of dosing, administration and storage, affordability, and enhanced spectrum of action, particularly targeting AMR Gram-negative pathogens. Additional research gaps include the lack of guidance on implementation of antibiotic stewardship interventions targeting hospitalized neonates in LMIC, which is essential to ensuring the long-term effectiveness of new antibiotic agents and regimens.²⁶

Challenges in neonatal sepsis prevention in LMIC

Robust infection prevention and control (IPC) programs are essential to ensuring patient safety in neonatal units and when implemented effectively, may contribute to reduced antibiotic use. Hospitalized neonates, especially sick and preterm infants, are highly vulnerable to development of sepsis owing to deficits in acquired and innate immunity, prolonged hospital stay, high rates of invasive device use and frequent antibiotic exposure. Other factors driving higher infection risks in LMIC include suboptimal intrapartum and postnatal infection control practices, and lack of access to optimized water sanitation and hygiene (WASH) in community and healthcare facility settings.²⁷

The evidence base for IPC and neonatal sepsis prevention interventions in LMIC is limited but supports the role of bundled and multi-modal interventions targeting device-associated infections, use of clean birth kits, and probiotic supplementation for preterm neonates.²⁸ For community-based IPC, a recent evidence synthesis identified chlorhexidine cord cleansing for omphalitis prevention as an effective practice in settings with unhygienic cord care.²⁹ Community-based antibiotic delivery for pSBI was also noted to be a safe alternative to hospitalization in LMICs where admission is not possible or is declined.

As a population vulnerable to infection, neonatal units and clinicians in LMIC hospitals are acutely aware of the need to prioritize IPC activities. Despite the awareness of the need for implementing IPC best practice in neonatal units, many LMIC hospitals struggle with infrastructure issues, lack of isolation space and personal protective equipment, overcrowding, understaffing, and weak organizational leadership in IPC.²⁷ A recent systematic review of neonatal IPC implementation determinants neonatal units, found that partnership and collaborations, institutional support for IPC, and empowerment of healthcare professionals, and parents were important facilitators of IPC best practices.³⁰ Frequently cited barriers were human resource related (staff shortages, lack of specialized IPC staff and training opportunities in IPC), physical and financial resource related (consumables, equipment, physical space limitations and IPC budget limitations).

Conclusion

Despite the importance of neonatal sepsis as a leading contributor to morbidity and mortality in LMIC, substantial knowledge gaps remain regarding its epidemiology, diagnosis, treatment and prevention. Many of these knowledge and practice gaps could be addressed by removing barriers to neonatal sepsis data sharing, development of standardized definitions and key variable lists to facilitate data harmonization and pooled analyses. Accurate, affordable, and accessible diagnostics for neonatal sepsis together with laboratory strengthening interventions are key to generating representative pathogen and AMR profile data to inform guidelines for empiric antibiotic treatment. Advocacy efforts are also crucial to enhance antibiotic access for neonatal sepsis in LMIC and promote inclusion of neonates as key populations in clinical trials for new antibiotics. Intensified research specifically focused on LMIC neonates with sepsis is urgently needed to expand the evidence base for improved surveillance, IPC, and AMS interventions.