Prevention and Treatment of Neonatal Infections in Facility and Community Settings of Low- and Middle-Income Countries: A Descriptive Review

Globally, nearly 7 million newborns contract possible serious bacterial infections (PSBIs), and an estimated 550,000 newborns die from serious bacterial infection every year in low- and middle-income countries (LMICs) [1], compounding the economic burden for under-resourced nations and impoverished individuals. For newborns with sepsis in sub-Saharan Africa, the region with the highest neonatal mortality rate [2], 5.3–8.7 million disability-adjusted life years are lost annually, with costs ranging from USD 10 billion to USD 469 billion per year in that region [3]. Furthermore, neonatal mortality from sepsis that is attributable to antimicrobial resistance (AMR) is estimated to be between 139,000 and 318,000 annually, with a disproportionate number occurring in LMICs [4]. Among survivors of severe bacterial infection in LMICs, morbidities such as neurodevelopmental disabilities can harm quality of life beyond the neonatal period [5].

Neonatal infections may be hospital-acquired or community-acquired, with each setting exposing newborns to distinct environmental conditions and pathogenic etiologies [6]. Given these differences in neonatal infection etiology and pathogenesis, interventions to promote neonatal survival must be considered at the relevant locus of care. Although the number of women delivering children in the facility setting using skilled birth attendants has increased in regions such as Africa and Asia, this development has not translated into improved maternal or newborn health outcomes because of the inadequate quality of facility care [7, 8]. Challenges to newborn care within healthcare facilities include suboptimal infectious disease outbreak surveillance, lack of microbiology laboratory technical expertise and equipment, and variable healthcare provider compliance to basic infection prevention interventions [7]. In contrast, for certain segments of the LMIC population, home births are a necessity due to insufficient access to healthcare services, a shortage of healthcare workers, or cost [6, 9]. In some cases, community-based births are simply preferred for sociocultural reasons [6].

Neonatal intensive care unit (NICU)-acquired infections are commonly caused by gram-negative bacteria, including Klebsiella pneumoniae, responsible for bloodstream infections (BSIs) (sepsis), and Acinetobacter baumannii, responsible for respiratory tract and BSIs [10‒12]. In contrast, approximately 60% of community-acquired infections are caused by Staphylococcus aureus, Klebsiella spp., and Escherichia coli, with BSIs predominating [13].

As efforts to decrease neonatal infections largely depend on the prevention and control of AMR development and spread, judicious antibiotic use is critical in the LMIC context where there is recognized misuse and overuse [10, 14]. Antimicrobial stewardship is defined as a coherent set of actions which promote the responsible use of antimicrobials, commonly including the optimization of antimicrobial selection, dosage, timing of first dose, dose interval, duration of therapy, or route of administration [15]. Recent systematic reviews recommend the prioritization of antimicrobial stewardship programs (ASPs), concomitant with infection prevention and control interventions, to effectively and holistically address the newborn mortality burden [10, 16]. Although there is a paucity of evidence on the effectiveness of ASPs and IPC practices in mitigating newborn infection and AMR among the neonatal population in LMIC settings, ASPs are associated with a 21% reduction in antibiotic consumption in pediatric hospitals from LMIC and high-income country settings collectively [17]. In addition, handwashing with soap is associated with a 17% reduction in acute respiratory infections in patients of all ages in LMICs when compared to no handwashing [18].

Healthcare-associated colonization and infection readily occurs through the use of unsterile medical devices such as ventilators and catheters or surgical equipment [7, 19] and may prolong the newborn’s length of hospital stay. Continued exposure to the healthcare environment can be especially risky for preterm newborns (<37 weeks’ gestation) because of their weaker immune systems, immature epidermal barriers, and dysbiotic gut microbiota [20]. Effective interventions proposed for reducing healthcare-associated infections (HAIs) in LMICs include prevention of vertical multidrug-resistant bacterial transmission, maintenance of skin integrity, and promotion of a healthy gut microbiome [10]; however, robust, multisite studies using explicit and standardized definitions for infection-related outcomes are required to best evaluate these interventions [21].

Clean birth kits (CBKs) provide the items recommended for a clean delivery including but not limited to razor blades, umbilical cord ties, and alcohol swabs [22]. CBKs are a low-cost and highly scalable intervention, especially when kits are made locally instead of imported. In 2010, the median direct cost of a locally prepared CBK was USD 0.45, corresponding to an average USD 215 cost per life saved. In comparison, an imported CBK costed USD 1.34, not including product distribution or marketing, corresponding to an average USD 921 cost per life saved [23]. CBKs are also especially useful in areas where facility births are not culturally accepted or financially feasible [24]. However, because CBKs are often implemented among a variety of other IPC interventions, and components vary from kit-to-kit, evidence supporting the effectiveness of CBKs in isolation from other interventions is limited [22, 24].

Chlorhexidine, a broad-spectrum topical antiseptic, is widely employed during childbirth to cleanse the birth canal, umbilical cord, and surgical wounds and, after birth, employed in whole-body cleansing; however, there is mixed evidence supporting its use in newborns [24, 25]. A systematic review of the evidence on whole-body skin cleansing with chlorhexidine for newborns found no protective effect against neonatal mortality or sepsis but noted more randomized trial evidence was needed [25]. Another systematic review found that chlorhexidine applied to the umbilical cord reduced neonatal mortality by 34% when commenced within 24 h of birth [24]; however, debate continues over the ideal solution composition, concentration, and frequency of application for effective and safe use [10, 26, 27].

Topical emollients in the form of creams, oils, or ointments are posited to help reinforce the newborn’s delicate epidermal barrier against infection, especially among very preterm (<32 weeks’ gestation) newborns [28]. Preterm newborns have an immature epidermal barrier, and very preterm newborns do not have a fully formed a vernix caseosa, a protective biofilm imparting antibacterial and antioxidant properties to the skin, moisturizing and lubricating the skin, and providing thermal regulation [28‒30]. Among preterm and term newborns alike, loss of skin barrier integrity or function can result in atopic dermatitis, cutaneous inflammation, allergen sensitization, and increased likelihood of colonization and/or infection [28, 31, 32]. However, there is conflicting evidence on the effectiveness of prophylactic therapy with topical emollients, especially considering mode and frequency of application and emollient composition [10, 28, 31‒33], as well as safety concerns [34, 35].

Dietary probiotics are intended to reinforce the newborn’s intestinal barrier by colonizing the gut microbiome with immune system-supporting bacteria [10, 36]. In practice, there have been mixed effects largely due to variation in strains and dosages, as well as timing and frequency of administration [10, 36], in addition to challenges in trial cross-contamination between treatment groups and lack of quality control to ensure product purity and safety [20, 36]. Results of a large trial found that probiotic supplementation, compared to placebo, was associated with a 54% reduction in necrotizing enterocolitis (NEC) (Bell stage ≥ II); however, there was no difference in neonatal mortality or HAI rates [37]. In a systematic review of 23 RCTs, supplementation with probiotics was associated with a 27% reduced risk of neonatal mortality, a 20% reduced risk of HAIs, and a 54% reduced risk of NEC (Bell stage ≥ II), when compared to no supplementation [38]. Similarly, there is mixed evidence for the use of synbiotics, a combination of probiotics and prebiotics, with one trial indicating a 50% reduction in all stages of NEC after synbiotic supplementation, compared to no supplementation, though this finding did not reach statistical significance [39], and another trial revealing reduced rates of neonatal mortality, culture-positive sepsis, and NEC among newborns supplemented with synbiotics, compared to the placebo group, though the sample size was too small to be conclusive [40].

The World Health Organization (WHO) recommends treatment of PSBIs in newborns occur in healthcare facilities upon identification of signs including feeding inability, convulsions, rapid breathing (≥60 breaths per minute) or chest in-drawing signaling respiratory distress, fever (≥38°C) or low body temperature (<35.5°C), and little or no movement [41]. However, recognizing that facility-based care may be infeasible due to unaffordability, inaccessibility, or unacceptability [41], studies conducted in Pakistan and India finding nearly 85% of families refusing facility referral [42, 43], numerous trials have evaluated community-based antibiotic treatment delivered in the home or outpatient facilities by community health workers for PSBIs when simplified antibiotic regimens (predominantly or solely oral rather than injectable) are deemed appropriate [44]. Multicountry randomized controlled trial (RCT) evidence shows that community-based antibiotic delivery of simplified regimens for PSBIs is cost-effective [45, 46] and reduced neonatal mortality after day one by 17%, with sustainability of community-based PSBI management dependent on its integration within existing community-based maternal and newborn care [46]. Nevertheless, there remains concern over unnecessary antibiotic usage and the emergence of antibiotic resistance in the community [47, 48]. These interventions were reviewed in their totality for the development of evidence-based guidance on neonatal infection prevention and treatment, and their effectiveness was assessed separately based on health setting, to reduce the newborn infection-associated mortality and morbidity burden in LMICs.

A descriptive review combining different methodological approaches was conducted to synthesize evidence on the effectiveness of newborn infection prevention interventions in LMICs, which will serve as a contribution to the upcoming 2024 Lancet Global Newborn Care Series. Following an initial scoping exercise, recent and high-quality systematic reviews were identified for the topics of HAI prevention, topical emollients, and supplementation with probiotics or synbiotics. High-quality systematic reviews were also identified for the topics of CBKs, chlorhexidine cleansing, prophylactic systemic antifungal agents, and community-based antibiotic delivery for PSBIs, but these required updating. Lastly, no systematic reviews assessed strategies for reducing AMR among the neonatal population in LMICs; thus, for this topic, we conducted a de novo review. Furthermore, to provide the most suitable recommendations for real-world implementation, our reviews considered the impact of these interventions within three distinct health settings: facility, mixed, and community (Fig. 1). We also classified facility settings into primary, secondary, and tertiary facilities, as defined in an associated methods paper [49]. We report a high-level summary of methods below; however, a detailed summary of all methods can be found in the aforementioned methods paper.

To our knowledge, no evidence synthesis exists on strategies for AMR reduction among neonates in LMICs, especially with attention to the different practical considerations that diverse health settings present; therefore, we conducted a de novo systematic review. The systematic review protocol was registered in PROSPERO (CRD42023388338).

A database search was conducted across the following databases: Ovid MEDLINE, Embase, CINAHL, Global Index Medicus, and the Cochrane Central Register of Controlled Trials (CENTRAL) for studies published from 2000 to 2024. The Ovid MEDLINE database search strategy, with a final search date of June 26, 2024, is presented in online supplementary Appendix 1.1 (for all online suppl. material, see https://doi.org/10.1159/000541871). In addition, we manually searched the reference lists of relevant reviews to identify other studies to include for screening.

We included RCTs, quasi-randomized trials, observational studies, program evaluations, and implementation studies conducted in the LMIC setting. Studies eligible for inclusion implemented strategies, policies, or interventions designed to mitigate the development of AMR among preterm and term neonates in facility-based or community-based settings of LMICs. A detailed list of included primary and secondary outcomes can be found in online supplementary Appendix 2. Interventions aimed at preventing and managing AMR were first categorized as either single-component or multicomponent interventions, then classified according to their area of focus as regulation, education, or restriction interventions (see online suppl. Appendix 3, for definitions). Multicomponent interventions targeted more than one of these three areas of focus. Study quality was assessed using the Cochrane RoB-2 tool for randomized controlled trials, the Cochrane ROBINS-I tool for non-randomized studies, and the NIH tool for observational cohort and cross-sectional studies.

Random-effects meta-analyses with inverse variance weighting were conducted for pooled effect estimates reported as risk ratios (RRs) for dichotomous outcomes and mean differences (MDs) for continuous outcomes with associated 95% confidence intervals (CIs). Subgroup analyses, by study design and level of care, were performed, and sensitivity analyses excluding studies with a high risk of bias assessed the robustness of our pooled effect estimates. Of the meta-analyses conducted, we opted to report pooled effect estimates of statistical significance only; however, all additional meta-analyses can be located in online supplementary Appendix 4. Due to the heterogeneity of interventions implemented and outcomes assessed across studies, meta-analysis was not possible for most of our outcomes of interest. Furthermore, acknowledging that it may be inappropriate to meta-analyze such heterogeneous studies, we also conducted a synthesis of the evidence incorporating effect direction plots [50] and an analysis of process and implementation outcomes. The results of these quantitative and narrative syntheses are presented in further detail in a standalone paper [51].

During our scoping exercise, we identified a relevant systematic review evaluating the effectiveness of CBKs [22]; however, the final search date was in 2018 and the review required updating. We updated the evidence for the use of CBKs to prevent neonatal mortality and infection in facility and community settings of LMICs by rerunning the reported search strategy to identify eligible studies published after the last search date and synthesized the existing review’s evidence with more recently published evidence from LMICs. In the source review, PubMed, CINAHL, the Cochrane Library, and Google Scholar were searched using search terms such as “birth kit*”, “delivery kit*”, “clean delivery kit*”, “mama kit*” and “maama kit*”. This search strategy was replicated in our search for new studies, with an additional search of the EMBASE database.

We included RCTs and quasi-experimental studies that reported on women giving birth with or without the use of CBKs, regardless of their obstetric or medical characteristics, level of risk, education, or socioeconomic status, to assess the effectiveness of CBKs in improving neonatal outcomes at birth. As we were interested in the effectiveness of CBKs, we relied on trial data only, which differs from the eligibility criteria in the existing review, as they had no restriction on study design (see online suppl. Appendix 2, for a summary of our eligibility criteria).

Data only from trials in the original review were extracted and combined with data from newly found LMIC trials. All meta-analyses were conducted using Review Manager 5.4 software [52]. Subgroup analyses for updated evidence syntheses were performed where possible in accordance with the subgroups defined by the review authors. We additionally ran subgroup analyses by level of care (i.e., community, primary, secondary, or tertiary facility settings) to inform our understanding of the impact of interventions to prevent newborn infections at different levels of implementation. The Cochrane RoB-1 tool for RCTs [53] was used to assess the risk of bias for included studies.

For the chlorhexidine cleansing topic, we included five individually randomized and cluster-randomized trials from LMICs comparing application of chlorhexidine to the umbilical cord with dry cord care or usual cord care practices. These five trials were also used to inform the recently updated WHO recommendations on maternal and newborn care for a positive postnatal experience [54]. We also identified one trial for inclusion from another recent and high-quality review [26]. Although our two source reviews had a literature search date of 2020 or later, based on technical advisory group (TAG) knowledge of newly published trials of relevance, we updated this review, pooling six LMIC-based trials from the existing reviews with nine newly found LMIC-based trials. The Cochrane RoB-1 tool was used to assess the quality of included trials, and additional subgroup analyses by cleansing frequency and omphalitis severity were conducted where possible.

Our scoping exercise also identified relevant Cochrane reviews on prophylactic systemic antifungal agents [55] and community-based antibiotic delivery for PSBIs [44] published in 2015 and 2019, respectively. We updated these two systematic reviews identifying new trials using the original search syntax, inclusion criteria, and list of databases reported.

The methods for this review are reported in online supplementary Appendix 5, since we leveraged HIC evidence based on TAG recommendation.

The review on community-based antibiotic delivery for PSBIs searched CENTRAL, MEDLINE via PubMed, Embase, and CINAHL for randomized, quasi-randomized, or cluster-randomized trials on the effects of community-based delivery of antibiotics to newborns with PSBI in LMICs on primary outcomes of all-cause neonatal mortality and sepsis-specific neonatal mortality [44]. Two comparisons were evaluated: the first comparing home or basic health unit delivery of antibiotics with the standard hospital referral, and the second comparing any simplified antibiotic regimen with the standard community-delivered regimen of 7–10 days of injectable penicillin or ampicillin with an injectable aminoglycoside [44]. Data from eight trials included in the original review were combined with data from one newly identified trial, and the risk of bias in included studies was assessed using the Cochrane RoB-1 tool [56].

Recent, high-quality reviews already existed concerning HAI prevention [21], topical emollient therapy for preterm newborns [28] and term newborns [31], probiotic supplementation [20, 57, 58], and synbiotic supplementation [20, 59], and we extracted and pooled data from all studies conducted only in LMICs to produce effect estimates specific to the LMIC context. For topical emollients, probiotic supplementation, and synbiotic supplementation, we included additional studies based on expert consensus, where large and well-designed trials conducted in LMICs were known (see online supplementary Appendix 1, for the search strategies employed in as-is reviews). Subgroup analyses were replicated where possible, with additional subgroup analyses by level of care for all topics. The Cochrane RoB-1 tool was used to assess the quality of the studies included within the topical emollients, probiotics, and synbiotic reviews.

The HAI prevention topic examined hospitalized neonates in the neonatal ward or NICU receiving various interventions compared to standard of care. HAI prevention studies included RCT, controlled and noncontrolled before-after, controlled and noncontrolled interrupted time series, and cohort study designs. A summary of the eligibility criteria for all reviews is included in online supplementary Appendix 2, alongside the outcomes of interest. For this topic, the study designs, interventions, and outcomes captured for review were highly heterogeneous; therefore, we report the original review authors’ narrative synthesis. The HAI prevention review [21] used the Integrated Quality Criteria for the Review of Multiple Study Designs (ICROMS) tool to assess study quality, and studies included in that review met the minimum score and mandatory ICROMS criteria to be considered for inclusion.

For the topical emollients review, we included RCTs and quasi-experimental studies to examine the effect of oil and cream emollient applications in comparison to routine skin care on outcomes including all-cause neonatal mortality, invasive infection, and NEC in the preterm and term newborn populations.

The probiotic topic included trial data for neonates receiving enteral or oral administration of probiotic supplementation in neonates, in comparison to no supplementation or placebo feeds. In addition to subgroup analysis by level of care, we ran subgroup analyses by duration of intervention, by funding source, and by study quality based on a risk of bias assessment.

Similarly, for the synbiotic topic, trial data was included across two reviews to evaluate the effect of synbiotic supplementation in comparison to no supplementation. In addition to subgroup analysis by level of care, we ran subgroup analyses by duration of intervention, funding source, and synbiotics’ volume.

Our reviews identified 61,820 records through database searches and 388 records through other sources. After duplicates were removed, 47,560 records were screened based on title and abstract and 616 full-text studies were assessed for eligibility. Ultimately, 155 studies were included across all reviews (see online supplementary Appendix 6, for PRISMA flow diagrams detailing the selection of studies per review topic).

Table 1 presents a high-level summary of the characteristics of included studies by review topic. There were 133 studies of neonatal infection prevention in LMICs and 22 studies of treatment of suspected neonatal infections in LMICs, leveraging HIC evidence from eight studies for the review on antifungals. The de novo review of strategies to reduce AMR included 26 facility-based studies conducted in LMICs. The prevention of HAIs review analyzed LMIC data as-is from an existing review [21], including 27 facility-based studies. The updated reanalysis of LMIC-only evidence from an existing CBK review [22] identified two trials implementing birthing kits in mixed settings. The chlorhexidine review reanalyzed LMIC-only evidence from two sources, an existing systematic review [26] and a recent WHO guideline [54], identifying 16 studies in total, seven from facility settings, two from mixed settings, and seven from community settings. The emollients review reanalyzed 17 studies conducted in LMICs from two existing reviews [28, 31], with eight studies conducted in facility settings, five studies conducted in community settings, and four studies with specific outcomes reported within mixed settings. The probiotics as-is review reanalyzed LMIC data from three reviews [20, 57, 58] across 37 facility-level studies. The synbiotics as-is review reanalyzed LMIC data from two reviews [20, 59], for a total of eight studies; seven studies conducted at the facility level and one study conducted at the community level. Leveraging HIC data, the prophylactic systemic antifungals review update identified ten studies from an existing Cochrane review [55] to pool with evidence from three newly identified studies, for a total of 13 facility-based studies. As antibiotic delivery for PSBI occurred either in a mix of facility- and community-based settings or in purely home-based settings, of the nine studies sourced from a Cochrane review [44], seven studies were categorized as mixed setting and two studies were categorized as community setting (see online supplementary Appendix 7, for a detailed description of the characteristics of included studies across all interventions for the prevention and treatment of neonatal infections).

Of 103 RCTs included in this review, 37 (35.9%) were rated as high risk of bias, 19 studies (18.4%) as moderate risk of bias, 21 studies (19.4%) as low risk of bias, and 26 studies (25.2%) as unclear risk of bias. Of 18 included quasi-experimental studies, two studies (11.1%) had a serious risk of bias and 16 studies (88.8%) had a moderate risk of bias. Of seven included observational studies, five studies (71.4%) were rated as fair quality and two studies (28.6%) as good quality (see online supplementary Appendix 8, for risk of bias assessments for all included studies by study design and assessment tool used [i.e., RoB-2, ROBINS-I, and NIH]).

Table 2 presents a summary table indicating the effects of newborn infection prevention interventions when implemented in the facility setting. A summary table for the effect of HAI prevention interventions is presented separately, as results were narratively synthesized.

We identified a total of 26 studies eligible for inclusion in this systematic review, all of which occurred in a facility-based setting. Of these, eight (30.8%) studies implemented single-component antimicrobial stewardship interventions [60‒67] and 18 (69.2%) studies implemented multicomponent antimicrobial stewardship interventions [68‒85].

Among studies with single-component interventions, two (25%) studies implemented regulation interventions [60, 61], no studies implemented education interventions, and six (75%) studies implemented restriction interventions [62‒67]. Among studies with multicomponent interventions, three (16.7%) studies implemented regulation and education interventions [68‒70], one (5.6%) study implemented education and restriction interventions [74], three (16.7%) studies implemented regulation and restriction interventions [71‒73], and 11 (61.1%) studies implemented regulation, education, and restriction interventions [75‒85]. Studies assessed the effectiveness of strategies to reduce AMR before and after the interventions were introduced in the NICU.

A subgroup analysis by study design showed that implementing regulation strategies alone reduced the risk of newborns receiving at least one antimicrobial drug in the post-intervention period by 21% (95% CI: 20–23%). However, although this evidence comes from a large cohort study [61], a significant reduction in antimicrobial use was not seen overall. There was also no significant difference in neonatal sepsis/suspected sepsis before and after introduction of the intervention.

In comparison to the pre-intervention period, implementing restriction strategies alone reduced the risk of culture-positive sepsis by 32% (95% CI: 17–45%) and the proportion of newborns receiving any antibiotic by 13% (95% CI: 2–22%) post-intervention. However, there was no significant difference in neonatal mortality, neither Access nor Watch group antibiotic usage, between the pre- and post-intervention periods.

Implementing regulation and restriction interventions reduced the risk of neonatal mortality due to nosocomial BSI by 32% (95% CI: 5–51%) post-intervention, compared to pre-intervention. There were no significant subgroup differences in the risk of neonatal mortality due to nosocomial BSI by facility level.

Strategies to reduce AMR which incorporated regulation, education, and restriction reduced the risk of neonatal mortality by 27% (95% CI: 7–43%) in the post-intervention period, when compared to the pre-intervention period, an effect which did not change following a sensitivity analysis omitting a study determined to have high risk of bias [81]. Furthermore, the risk of multidrug-resistant organism infection or colonization was reduced by 29% (95% CI: 3–48%) and the risk of culture-negative sepsis was reduced by 10% (95% CI: 1–19%) post-intervention. There were no significant differences in the risk of NEC, LOS, or pneumonia before and after intervention.

Compared to pre-intervention, the risk of isolating Candida spp., Pseudomonas spp., and Klebsiella spp. from blood culture samples reduced by 85% (95% CI: 53–95%), 75% (95% CI: 59–85%), and 32% (95% CI: 12–48%) post-intervention, respectively. However, there was twice (95% CI: 1.08–3.73) the risk of isolating coagulase-negative staphylococci from blood culture during the post-intervention period. There was no significant difference in isolation of MRSA, E. coli, Acinetobacter spp., or Enterobacter spp. from blood cultures collected in the pre- and post-intervention periods. Additionally, implementing regulation, education, and restriction interventions showed no difference in mean length of hospital stay between pre- and post-intervention periods.

The risk of newborns receiving antibiotics was reduced by 29% (95% CI: 19–39%) and the risk of antibiotic therapy durations of >5 days was reduced by 64% (95% CI: 7–86%). Subgroup analysis by study design showed a 78% (95% CI: 67–85%) post-intervention risk reduction in one cohort study [77] and 42% (95% CI: 38–47%) post-intervention risk reduction in one large quasi-experimental study [78]. There was no significant difference in discontinuation of antibiotic therapy after 48 h pre- and post-intervention.

All 27 studies were conducted in facility settings of LMICs and are thus eligible for inclusion (shown in Table 3). Of these, 17 (63%) studies evaluated single interventions categorized as gastrointestinal barrier maintenance interventions, epidermal barrier maintenance interventions, antibacterial and antifungal interventions, maternal contact interventions, and device-associated infection interventions; and ten (37%) studies evaluated care bundle interventions categorized as hypothermia prevention bundle interventions, infection prevention bundle interventions, and device-associated infection prevention bundle interventions. The broad-ranging interventions in this review are specifically intended for the prevention of infection in hospitalized newborns where the primary outcome of interest was BSIs.

Interventions evaluated for gastrointestinal barrier maintenance in preterm and low birthweight (LBW, <2,500 g) newborns included bovine lactoferrin supplementation [86, 89, 98], parenteral glutamine supplementation [102], and probiotic supplementation with Lactobacillus casei, Lactobacillus acidophilus, Bacillus subtilis, and Enterococcus faecalis [101].

Oral lactoferrin supplementation of 200 mg daily, compared to placebo, resulted in no significant difference in all-cause neonatal mortality, sepsis-related mortality, or severe NEC (Bell stages 2 and 3), but significant reductions in episodes of culture-proven sepsis were found (intervention: 4.4 per 1,000 patient-days vs. control: 17.3 per 1,000 patient-days) [86]. A second study found no difference in culture-proven LOS between 200 mg/kg lactoferrin and placebo treatment groups [98]. In contrast, bovine lactoferrin supplementation dosages of 80–142 mg/kg per day, compared to placebo, significantly reduced the risk of sepsis-attributable mortality (i.e., death of neonates with culture-proven LOS) by 7.5% (p = 0.027) and reduced the risk of first episodes of culture-proven sepsis by 79% (95% CI to 96% decrease to 2% increase), though this finding did not reach statistical significance [89].

Nutritional supplementation with parenteral glutamine for a duration of >2 weeks, versus no supplementation, reduced the risk of nosocomial infection by 6%, though this finding did not reach statistical significance [102]. Probiotic therapy with L. casei, L. acidophilus, B. subtilis, and E. faecalis three times daily for the first eight days after birth, compared to placebo, resulted in no significant difference in the risk of culture-proven sepsis or the risk of NEC in hospitalized term newborns [101].

Interventions evaluated for epidermal barrier maintenance included sunflower seed oil [88, 93], Aquaphor [93, 94], and coconut oil [100]. The effect of thrice daily topical sunflower seed oil application, versus no topical prophylaxis, had no effect on sepsis-attributable neonatal mortality (specifically, sepsis-related death beyond the first two days of life) (adjusted OR = 0.72, 95% CI: 0.39–1.34); however, it significantly reduced the incidence of nosocomial infections by 54% (95% CI: 19–74%) [88]. The application of high-linoleate sunflower seed oil thrice daily for the first 14 days and subsequently twice daily until discharge resulted in significant reductions in culture-proven sepsis (adjusted IRR = 0.59, 95% CI: 0.37–0.96) compared to no intervention [93].

Compared to no intervention, topical application of Aquaphor consisting of petrolatum, mineral oil, mineral wax, and lanolin alcohol reduced the risk of nosocomial infection by 40% (95% CI: 0.35–1.03), though this finding did not reach statistical significance [93]. Similarly, a study that implemented Aquaphor once daily for the first 14 days after birth found no difference in culture-positive sepsis (intervention: 23% vs. control: 19.6%, p = 0.42) and no difference in NEC (Bell stage ≥2) (intervention: 3% vs. control: 2.1%, p = 0.67) between the treatment group and the routine skin care group [94]. Coconut oil, applied twice daily for the full neonatal period (i.e., one to 28 days after birth), reduced the risk of culture-proven sepsis by 13.8% (95% CI: 2.3–16%) compared to no intervention [100].

Antibacterial and antifungal interventions evaluated included chlorhexidine gluconate (CHG) for umbilical cord care [95] and for whole-body skin cleansing [96], and prophylactic treatment with fluconazole [99]. For umbilical cord care, 2.5% CHG application, thrice daily until 3 days after the cord separates from the newborn, significantly reduced the risk of episodes of culture-proven sepsis by 18.5%, compared to dry cord care [95]. For whole-body skin cleansing, 0.25% CHG solution, applied via wipes daily for the first 7 days of ward admission, had no significant effect on sepsis rates compared to tepid water (intervention: 3.6% vs. control: 6.9%, p = 0.195) [96]. Fluconazole prophylaxis, administered to preterm neonates with very low birthweight (VLBW, <1,500 g) within the first 3–28 days after birth, resulted in no significant difference in invasive fungal infection compared to placebo (intervention: 26.7% vs. control: 25%, p = 0.835) [99].

Maternal contact interventions evaluated included the room-in model of care (i.e., allowing parents 24/7 inhospital access to their newborns) [87], kangaroo mother care (KMC) [91, 92], and maternal massage therapy [97].

The room-in model versus the standard model (i.e., no or limited parental access to the NICU), significantly reduced the risk of all-cause neonatal mortality by 2.4% (p < 0.001) and NEC by 1.2% (p = 0.04). Additionally, there was no significant difference in nosocomial infection between room-in model of care and standard model of care groups (intervention: 9.8% vs. control: 11.2%, p = 0.41) [87].

Two studies implemented KMC interventions, one study evaluating the effectiveness of sustained KMC versus traditional management [92], the other assessing the effectiveness of intermittent skin-to-skin contact for a minimum of 1 h per day versus standard care [91]. In both studies, there was no difference in culture-proven sepsis episodes between the groups (continuous KMC intervention: 12.8% vs. control: 12.1%; intermittent KMC intervention: 3.1% vs. control: 1.6%, p = 1.00). Although a higher proportion of mild to moderate infection episodes occurred in the continuous KMC intervention group compared to the traditional management group (intervention: 6.7% vs. control: 2.8%, p = 0.019), the proportion of nosocomial infections was higher in the control group (intervention: 3.8% vs. control: 7.8%, p = 0.026) [92]. Intermittent skin-to-skin contact between mother and newborn showed no difference in NEC (grade 2 or 3) (intervention: 1.6% vs. control: 1.6%, p ≥ 0.20) compared to standard care [91]. Tactile-kinesthetic stimulation (i.e., massage therapy), compared to no intervention, significantly reduced the risk of LOS by 27.4%, but there was no effect on NEC (intervention: 0% vs. control: 2.1%, p = 1.00) [97].

Nosocomial device-associated interventions assessed changing the mode of catheterization for newborn infection prevention [90]. There was no significant difference in culture-proven sepsis among newborns with peripherally inserted central catheters versus standard peripheral intravenous catheters (intervention: 2.7% vs. control: 5.4%, p = 0.53) [90].

A hypothermia prevention bundle comprising radiant warmers set to 26°C, standardized transport teams, and a staff education program including process reviews and feedback significantly reduced the risk of neonatal mortality by 5%; however, there was no significant difference in infection incidence (post-intervention: 20.9% vs. pre-intervention: 23.3%, p = 0.53) or NEC occurrence (intervention: 9.3% vs. control: 15.1%, p = 0.34) [105].

Four studies implemented infection prevention bundles [79, 104, 106, 108]. Compared to pre-intervention, bundle implementation of IPC training, text message reminders, alcohol hand rub, enhanced environmental cleaning, and weekly bathing of neonates with 2% CHG reduced the risk of neonatal mortality by 21% (RR = 0.79, 95% CI: 0.76–0.83) post-intervention [106]. Compared to pre-intervention, implementing an IPC bundle consisting of blood culture quality improvement, provision of alcohol hand rub, infection and hand hygiene surveillance, staff education, case discussions, and infection control checklists reduced the risk of neonatal mortality by 20% (95% CI: 0.70–0.90) [79]. Compared to pre-intervention, implementing a nurses’ training package with modules on infection control and nutrition interventions resulted in no change in survival (pretraining: 80%, post-training: 78.2%) or LOS (intervention: 11.3 per 1,000 patient-days vs. control: 12.3 per 1,000 patient-days), despite improvements in provider knowledge and practices [104]. Compared to pre-intervention, introducing alcohol-based hand rub dispensers, along with staff education, daily surveillance, and quarterly feedback, reduced the risk of MRSA infections by 20% (year 2001: 2.2 infections per 1,000 patient-days vs. year 2005: 0.6 infections per 1,000 patient-days, p = 0.001), but there was no significant trend in reduction of A. baumannii infections (year 2001: 0.6 per 1,000 patient-days vs. year 2005: 0.2 per 1,000 patient-days) [108].

Compared to pre-intervention, implementing ventilator-associated pneumonia (VAP) bundle interventions of hand hygiene re-enforcement, rational waste disposal and antibiotic usage, enhanced patient isolation and ventilator disinfection, and staff education sessions significantly reduced the risk of neonatal mortality by 11.3% and reduced the VAP rate from 48.84 to 18.50 VAPs per 1,000 mechanical ventilation (MV) days [107]. Similarly, implementing a VAP bundle including active surveillance for VAP, hand hygiene protocol adherence, readiness to wean assessments, routine oral antiseptics, noninvasive ventilation, orotracheal instead of nasotracheal intubation, and proper management of ventilation circuits reduced the rate of VAP by 33% (95% CI: 0.50–0.91), corresponding to a decrease from 17.8 to 12.0 VAPs per 1,000 MV-days, when compared to the pre-intervention period [111].

Furthermore, the effect of a VAP prevention bundle and routine IPC measures combined with hand hygiene re-enforcement, sterile suctioning, ventilator circuit change if visibly soiled or malfunctioning, strict indications for intubation, daily evaluation for readiness for extubation, and sedation vacations reduced the rate of VAP by 43% (95% CI: 0.41–0.78), which corresponded to 36.4 VAP episodes per 1,000 MV-days pre-intervention and 23 VAP episodes per 1,000 MV-days post-intervention; however, there was no difference in pre- and post-intervention neonatal mortality (25% vs. 17.3%, respectively) [103].

Among bundles targeting catheter use, implementation of a bundle including hand hygiene protocols and full-barrier precautions during central venous catheter (CVC) insertion, chlorhexidine skin cleansing, avoiding the femoral site and removing unnecessary catheters, as well as group education, discussion, and feedback sessions significantly reduced the risk of culture-proven catheter-associated bloodstream infections (CA-BSI) by 12% (pre-intervention: 32% vs. post-intervention: 20%, p = 0.04), corresponding to 24.1 per 1,000 CVC days pre-intervention and 14.9 per 1,000 CVC days post-intervention [109]. Similarly, a central line-associated bloodstream infection (CLABSI) prevention bundle comprised a central line care bundle, active outcome and process surveillance, staff education, and feedback of CLABSI rates and infection control practices reduced the rate of CLABSI by 55% (95% CI: 0.33–0.63) from 21.4 cases per 1,000 central line days pre-intervention to 9.7 cases per 1,000 central line days post-intervention [110].

Five studies [95, 112‒115] assessed the effectiveness of chlorhexidine application to the umbilical cord within facility-based settings. In comparison with dry cord care, cleansing the umbilical cord with chlorhexidine significantly reduced the risk of omphalitis by 42% (95% CI: 20–58%), and the studies pooled in this meta-analysis [95, 112, 113, 115] implemented chlorhexidine as the sole intervention and not as part of a CBK; thus, this estimate represents the use of chlorhexidine in isolation from CBKs. In a subgroup analysis comparing single versus multiple chlorhexidine cord cleansing in facility settings, the test for subgroup differences did not reach statistical significance for the outcome of omphalitis. A single-study estimate exists for the outcome of neonatal mortality and can be found in online supplementary Appendix 4.1.2. For the comparison of whole-body cleansing versus water or saline, single-study estimates exist for the outcomes of neonatal mortality and BSI and can likewise be found in online supplementary Appendix 4.1.2.

For use in the preterm newborn population, application of topical ointments or creams in facility settings reduced the risk of all-cause mortality by 18% (95% CI: 2–31%) [93, 94] but had no significant effect on invasive infection, when compared to routine skin care. Comparatively, application of topical oils in facility settings reduced the risk of invasive infection by 43% (95% CI: 22–59%) [88, 93, 100, 116‒119] but had no significant effect on all-cause neonatal mortality, when compared to routine skin care. Use of topical oils also resulted in an increase in the rate of weight gain (g/kg/day) (MD = 3.42, 95% CI: 2.17–4.67) [119‒122] and change in crown-heel length (mm/week) (MD = 1.82, 95% CI: 0.20–3.44) compared to routine skin care [119‒121] (see online supplementary Appendix 4.1.3., for single-study effect estimates comparing topical ointments or creams to topical oils and comparing coconut oil to mineral oil). Finally, the combined effect of topical ointments, creams, or oils reduced the risk of all-cause neonatal mortality by 19% (95% CI: 6–30%) when compared to routine skin care.

Probiotic supplementation reduced the risk of NEC by 61% (95% CI: 51–69%), the risk of all-cause neonatal mortality by 25% (95% CI: 8–39%), and the risk of invasive infection by 19% (95% CI: 9–28%) in preterm newborns, when compared to control. There were significant subgroup differences in the risk of NEC by probiotic type. The risk of NEC was reduced by all types of probiotics; however, the greatest reduction was found for Bifidobacterium spp. by 84% (95% CI: 69–92%), and the smallest reduction was found for Saccharomyces spp. by 7% (95% CI: 55% reduction to 94% increase), though this was not statistically significant when compared to control or placebo. However, there were no significant differences in mortality or invasive infection by probiotic type.

There were significant differences in the risk of NEC by feeding type. Compared to control, the risk of NEC was reduced by 81% (95% CI: 64–90%) when preterm newborns were fed with formula, by 63% (95% CI: 38–77%) when preterm newborns were fed with human milk, and by 52% (95% CI: 36–64%) when preterm newborns were fed a mixture of human milk and formula. However, there were no significant differences in mortality or invasive infection by feeding type.

There were significant differences in the risk of NEC by duration of intervention. Compared to control, the risk of NEC was reduced for all durations of probiotic supplementation except the duration of “6 weeks or until discharge,” which increased the risk of NEC by 4% (95% CI: 0.34–3.17). The greatest reduction in NEC was found for the duration of “8 weeks or until discharge” by 89% (95% CI: 53–97%), while the smallest reduction was found for the duration of “3 weeks” by 36% (95% CI: 0.18–2.30), though this was not statistically significant. There were no significant differences in mortality or invasive infection by duration of probiotic supplementation.

There were no significant differences in the risk of NEC, all-cause neonatal mortality, or invasive infection by strain type (i.e., single strain or multiple strains), facility level (i.e., secondary or tertiary), funding source (i.e., public sector, industry, not reported, or none), or risk of bias (i.e., high risk of bias, unclear risk of bias, or low risk or bias). Probiotic supplementation had no effect on NEC, neonatal mortality, or invasive infection in extremely preterm or extremely low birthweight (ELBW, <1,000 g) newborns, when compared to control, while considering probiotic type, feeding type, or duration of intervention (see online suppl. Appendix section 4.1.4.).

Lastly, facility-based supplementation with probiotics or synbiotics (i.e., of probiotics with or without prebiotics) reduced the risk of all-cause neonatal mortality by 29% (95% CI: 15–41%) compared to control in preterm, LBW, extremely preterm, and ELBW newborns (see online supplementary Appendix 9, for a detailed breakdown of probiotic composition and administration per individual study).

Synbiotic supplementation reduced the risk of NEC by 76% (95% CI: 60–85%) and all-cause neonatal mortality until discharge by 47% (95% CI: 15–67%) in preterm or LBW newborns compared to control. Synbiotics had no significant effect on the risk of invasive infection compared to control (RR = 0.85, 95% CI: 0.58–1.26). There were, however, significant subgroup differences in the risk of invasive infection by duration of intervention, indicating a risk reduction when synbiotics were administered until discharge (RR = 0.49, 95% CI: 0.25–0.95) and when synbiotics were administered for 8 weeks or until discharge (RR = 0.62, 95% CI: 0.27–1.42), and a probable increase when synbiotics were administered for seven to 10 days (RR = 1.62, 95% CI: 0.85–3.07). There were no subgroup differences in NEC, neonatal mortality, or invasive infection by funding source (i.e., public sector, industry, not reported, or none) or by synbiotics’ volume (i.e., mixed with expressed breast milk or diluted with water) (see online supplementary Appendix 10, for a detailed breakdown of synbiotic composition and administration per individual study).

To assess the effectiveness of prophylactic systemic antifungals, we leveraged HIC evidence and reported our findings in online supplementary Appendix 5.

Table 4 presents a summary indicating the effects of newborn infection prevention interventions when implemented in mixed settings. CBK studies were classified by setting according to the location of CBK use at the time of delivery. If kits were used to deliver babies in home-based births and in healthcare facility-based births, this was classified as a mixed-setting implementation. Two such studies were included in this review [123, 124]. In the same way, topical emollient studies were classified by setting according to the locations of emollient therapy use. If emollients were initially applied to newborns in a healthcare facility with subsequent emollient applications occurring at home, this was classified as a mixed-setting implementation. Four such studies were included in this review [100, 116, 118, 125].

The use of CBKs in mixed settings reduced the risk of neonatal mortality by 27% (95% CI: 17–35%) compared to standard care [123, 124]. A single-study effect estimate comparing the risk of any omphalitis among newborns delivered with CBKs, versus newborns delivered without CBKs, can be found in online supplementary Appendix 4.2.1.

In comparison with dry cord care, cleansing the umbilical cord with chlorhexidine significantly reduced the risk of omphalitis by 55% (95% CI: 31–71%) [126, 127], and the studies pooled in this analysis implemented multiple chlorhexidine cleansing in isolation of a CBK intervention. No studies conducted in mixed settings evaluated the effect of single cleansing on omphalitis (see online supplementary Appendix 4.2.2., for single-study effect estimates for outcomes of neonatal mortality and BSI).

Topical oil use in preterm neonates was found to improve the rate of weight gain (g/kg/day) when compared to routine skin care (MD = 2.56, 95% CI: 1.47–3.66) [116, 118, 125], but no significant difference was found for crown-heel length (mm/week) or circumference (mm/week) (see online supplementary Appendix 4.2.3., for single-study estimates showing the effect of coconut oil versus mineral oil on growth outcomes).

Four studies [128‒131] compared the initiation and completion of antibiotic regimens for PSBIs in mixed settings of home-based and first-level clinic-based care with the standard hospital referral; however, results did not reach statistical significance for the outcome of all-cause neonatal mortality (see online suppl. Appendix 4.2.4.). Single-study effect estimates for the outcomes of early, late, and sepsis-specific neonatal mortality can also be found in online supplementary Appendix 4.2.4.

Pooled data from three studies [132‒134] found that there was no significant difference in all-cause neonatal mortality, treatment failure, or adverse effects when comparing home-based and first-level clinic-based antibiotic delivery of simplified antibiotic regimens, which overall requires fewer injectables, with community-based delivery of a standard, fully injectable antibiotic regimen of seven to 10 days penicillin or ampicillin with an aminoglycoside (see online suppl. Appendix 4.2.4.).

Table 5 presents a summary indicating the effects of newborn infection prevention interventions when implemented in the community setting. When the intervention was administered in home-based care, for example, this constituted a community-setting implementation. There were six community-based studies for chlorhexidine [135‒140], five community-based studies for emollients [141‒145], one community-based study for the synbiotic supplementation [146], and two domiciliary-based studies for community-based antibiotic delivery for PSBIs [147, 148].

In comparison with dry cord care, cleansing the umbilical cord with chlorhexidine significantly reduced the risk of omphalitis by 61% (95% CI: 40–74%) [135‒139, 149]. Assessing the effectiveness of chlorhexidine cord cleansing on omphalitis by severity, chlorhexidine reduced the risk of severe redness with pus by 71% (95% CI: 48–84%), moderate or severe redness with pus or severe redness alone by 45% (95% CI: 39–51%), moderate or severe redness by 33% (95% CI: 20–44%), and redness extending to skin or pus by 24% (95% CI: 2–42%) versus dry cord care. After multiple cord cleansing with chlorhexidine, the risk of omphalitis was reduced by 59% (95% CI: 38–74%) [135‒139], when compared to dry cord care, whereas the effect of single cord cleansing with chlorhexidine on omphalitis versus dry cord care did not reach statistical significance.

The risk of neonatal mortality was reduced by 12%, though this was not statistically significant (95% CI: 0.73–1.05). In a subgroup analysis comparing community-based studies which implemented chlorhexidine cleansing alone and studies which implemented chlorhexidine as a part of a birthing kit, the test for subgroup differences did not reach statistical significance for the outcome of neonatal mortality (see online suppl. Appendix 4.3.1.). For the comparison of whole-body cleansing compared to water or saline, a single-study estimate exists for the outcome of neonatal mortality (see online suppl. Appendix 4.3.1.).

Application of topical emollients to term newborns reduced the risk of atopic dermatitis by 63% (95% CI: 40–77%) compared to routine skin care [143, 144]. Single-study effect estimates comparing topical oils to routine skin care for outcomes of invasive infection, severe neurodevelopmental disability, and all-cause neonatal mortality can be found in online supplementary Appendix 4.3.2.

In a mixed population of preterm and term newborns, there was no statistically significant difference in all-cause neonatal mortality when comparing sunflower seed oil and mustard oil (intention-to-treat: RR = 0.97, 95% CI: 0.90–1.09) [142, 145]. No difference in mortality was observed in a subgroup of LBW newborns (intention-to-treat: RR = 1.05, 95% CI: 0.92–1.19), but there was a statistically significant difference observed in a subgroup of VLBW newborns (intention-to-treat: RR = 0.77, 95% CI: 0.62–0.96); however, the sample size was small [142, 145]. A single-study per protocol analysis for neonatal mortality and a single-study effect estimate for PSBIs can also be found in online supplementary Appendix 4.3.2.

In community-based settings, only one study [146] assessed the effectiveness of synbiotics on all-cause neonatal mortality and invasive infection compared to control (see online suppl. Appendix 4.3.3.).

In comparison to the standard hospital referral, two studies [147, 148] evaluated purely domiciliary-based antibiotic delivery for PSBIs and found a reduced risk in all-cause neonatal mortality by 25% (95% CI: 18–31%) (Table 5). Single-study effect estimates for the outcomes of early, late, and sepsis-specific neonatal mortality can also be found in online supplementary Appendix 4.3.4.

This descriptive review has comprehensively synthesized the best and most up-to-date evidence for newborn infection prevention and treatment interventions in LMICs (Table 6), including strategies to reduce AMR, interventions to prevent HAIs, CBKs, chlorhexidine cleansing, topical emollient therapy, probiotic supplementation, synbiotic supplementation, prophylactic systemic antifungal agents, and community-based antibiotic delivery for PSBIs, according to the level of implementation.

Among single interventions, restriction was the most promising due to its effect on culture-positive sepsis. However, implementing restriction and regulation interventions was found to reduce nosocomial BSI-attributable mortality. Although three studies implemented regulation and education interventions, data could not be meta-analyzed due to disparate outcomes or outcome metrics reported, a challenge noted by other reviews on antimicrobial stewardship [150, 151]. Implementation of regulation, education, and restriction interventions significantly reduced antibiotic use and prolonged antibiotic courses >5 days. This intervention bundle also led to reductions in bloodstream isolates of Klebsiella spp., Pseudomonas spp., and Candida spp. Only regulation, education, and restriction interventions reduced all-cause neonatal mortality and did not lengthen hospital stay, suggesting that multimodal interventions are safer and more effective to implement than single interventions. Where interventions targeting all three ASP categories are not feasible, more data are required to ascertain which combination of interventions could be just as effective, though regulation and restriction interventions seem most promising for their impact on nosocomial infection-attributable mortality. Notably, there was a high level of between-study heterogeneity, likely due to the varied actions taken to reduce AMR within each intervention category as well as differences in study designs, sample sizes, regional contexts, and compliance and adherence to the interventions. Although at least one study was conducted in each WHO region, all studies were single-site and none were from low-income countries. Only one study was conducted in a secondary care facility, only one study had an RCT design, and only one study focused exclusively on VLBW newborns, all factors that limit the generalizability of findings. In addition, there were limited data on Access, Watch, Reserve (AWaRe) [152] antibiotic usage, especially on Reserve antibiotics, as well as on multidrug-resistant pathogens isolated from cultures, and this dearth of antibiotic usage and resistance data may be indicative of the limited antibiotic utilization surveillance and laboratory capacity in many LMICs.

Among single interventions to prevent HAIs, sustained maternal contact may be beneficial and, at the very least, is not unduly harmful when compared to standard care, but more evidence is needed. Among epidermal barrier integrity interventions, sunflower seed oil was the most effective emollient type for HAI reduction when compared to no emollient, and among single interventions targeting bacterial colonization and growth, CHG applied thrice daily to the umbilical cord, versus dry cord care, was shown to reduce episodes of culture-proven sepsis. Among bundle interventions to prevent HAIs, IPC bundles, particularly those implementing IPC training, text message reminders, alcohol hand rub, enhanced environmental cleaning, and weekly bathing of neonates with 2% CHG showed the greatest neonatal mortality reduction of 21%, when compared to pre-implementation [106]. Among four IPC bundles for the prevention of HAIs, all bundles incorporated staff training and education [79, 104, 106, 108], three incorporated some form of monitoring, feedback, and reminders [79, 106, 108], and three incorporated some form of hand hygiene [79, 106, 108], while only one considered environmental cleaning [106] or laboratory culture quality improvement [79]. Among device-associated IPC bundles, all bundles reinforced hand hygiene adherence and introduced device-associated regulations; however, the bundle considering waste disposal and antibiotic usage procedures produced the only risk reduction in neonatal mortality of 11.3% [107]. Notably, this was the only bundle to include patient isolation actions, suggesting that isolation remains a challenge in the LMIC context but positively impacts newborn survival when prioritized within HAI prevention bundles. However, overall, there is more evidence supporting the implementation of device-associated IPC bundles compared to general IPC bundles, with the former originating from multisite studies representing 14 LMICs and the latter originating from single-site studies representing only four LMICs. Studies implementing device-associated IPC bundles showed consistent significant risk reductions in VAPs, CA-BSIs, and CLABSIs, with mixed effects on neonatal mortality. Our findings reflect a narrative review of evidence for HAI prevention interventions in LMICs which found bundle intervention approaches more effective than single-action interventions [10]. Notably, all studies were conducted in tertiary care facilities, and more research is needed to assess HAI prevention intervention effectiveness in secondary care facilities, which are more common in districts and towns and often less resourced. In facilities dealing with staffing shortages, high levels of staff and patient movement within hospitals and between sites may affect randomization efforts and introduce bias, a serious methodological challenge barring high-quality trial implementation in LMICs [21]. Another major limitation to assessment of AMR and HAI prevention strategies is the lack of standardized definitions for infection-related outcomes such as sepsis, making it difficult to compare findings across studies. We retained outcomes and outcome definitions as reported by original authors to underscore the diversity of terms used to denote and define sepsis. Future research should develop and use well-defined, if not standardized, outcomes and outcome definitions.

There is insufficient trial evidence to support the use of chlorhexidine for whole-body cleansing; however, we found that chlorhexidine for umbilical cord cleansing may be effective in reducing the risk of omphalitis when applied in a facility-based setting. Though most studies had small sample sizes (n < 100), they evaluated the use of chlorhexidine cord cleansing apart from any birthing kits and still identified an impact on omphalitis, which indicates that this may be a promising institutional practice in LMICs for newborns at risk of developing this condition, given that a previous review identified only one relevant trial with no cases of omphalitis in intervention nor control group [153]. Collecting further multisite, facility-based evidence is highly recommended, with additional evaluations of the optimal frequency of application and safety outcomes which predicate any justification of its use on a larger scale.

In preterm newborns, the use of topical ointments or creams had positive effects on neonatal mortality, whereas the use of topical oils reduced invasive infections and improved weight gain and crown-heel length compared to routine skin care. This corroborates the findings of a Cochrane review which found topical oils reduced invasive infection in LMICs [28]. However, where the Cochrane review suggested otherwise, topical ointments and creams may reduce the risk of all-cause neonatal mortality, though the strength of evidence was low due to imprecision and considerable heterogeneity. A WHO review of the evidence for emollients found that there were modest benefits to newborns with severe infection and no harms [154]. Although this may be a promising avenue for further research, given the substantial between-study heterogeneity and low certainty of the evidence, our analyses support emollient application based on clinical judgment, as per the WHO recommendations for preterm care [154]. Future facility-based implementation research should observe appropriate dispensing practices to reduce improper handling and avoid any risk of emollient contamination.

Probiotic supplementation had no benefit to ELBW newborns, but in preterm newborns (<32 weeks’ gestation) reduced the risk of NEC, all-cause neonatal mortality, and invasive infection, when compared to control. We found that probiotic supplementation was associated with a greater reduction in NEC and all-cause neonatal mortality than found by another recent systematic review [20]. Additionally, our results indicated choice of probiotic, feeding type, and duration of supplementation, as the most important factors in the reduction of NEC. In comparison, there was no difference in NEC between single-strain and multi-strain probiotics, contrary to some evidence suggesting multi-strain probiotics may be superior [155]. Among low risk of bias studies, the significant effect of probiotics on NEC was maintained, showing similar results as in the Cochrane review [57]. There were also no significant subgroup differences for all outcomes by funding source, indicating that the benefit of probiotics to newborn infection prevention persists regardless of funding by industries or by the public sector. Due to the consistent positive effect of probiotic supplementation on NEC, neonatal mortality, and invasive infection, we highly encourage its supplementary use alongside human milk feeding in preterm newborns. Although there is a stronger effect size among formula-fed newborns, the evidence is generated from three small studies which may be insufficiently powered to detect these outcomes. However, probiotics implementation will require strict quality control which may be a challenge in some LMICs, although large-scale trial implementation suggests this to be a surmountable challenge [38]. Furthermore, standardization of a high-quality, low-cost probiotic formulation for use in preterm newborns in facility-based settings of LMICs is required for large-scale implementation.

Synbiotic supplementation in preterm or LBW newborns reduced the risk of NEC and the risk of all-cause neonatal mortality until discharge but had no impact on invasive infection, when compared to control. These findings corroborate what was previously observed in very preterm and VLBW newborns receiving synbiotics [59]. No significant subgroup differences by volume of synbiotics or by funding source were shown for all outcomes. More evidence is needed to ascertain the optimal type, dosage, and duration of supplementation. Given that synbiotics involve the provision of a greater number of ingredients, as prebiotics and probiotics must be sourced in addition to cost of cold chain management [146], this may place a greater cost burden on low-resource facilities. Given the high cost and limited evidence of significant benefit, promoting synbiotics to prevent neonatal infection in LMICs is unwarranted at this time.

CBKs, topical emollients, and mixed home-based and first-level clinic-based antibiotic delivery for PSBIs were the only interventions employed in both facility and community settings. Use of CBKs in mixed settings showed a reduced risk of neonatal mortality, emphasizing the results of a previous systematic review [22]. Significant challenges in synthesizing trial evidence for CBKs include the observational design of most studies and the variable kit components, making direct comparisons difficult. Due to the limited trial evidence available, we pooled data from two relevant studies; notably, one trial provided standard kit components, which did not include interventions such as chlorhexidine or emollients [123], while the other trial provided an enhanced kit, which included chlorhexidine and emollients among other newborn care interventions [124]. Considering the large number of participants included in these trials, the effect estimate is robust for CBKs in mixed settings. More research is needed to elucidate whether neonatal mortality or sepsis is improved when a kit is used by a skilled birth attendant in a healthcare facility or when provided to mothers giving birth at home. Standardization of CBK components would also facilitate direct comparison between studies.

In mixed settings, where initial umbilical cord applications occurred in healthcare facilities with follow-up applications occurring at home by mothers or trained caregivers, multiple cord cleansing with chlorhexidine reduced the risk of omphalitis. Nevertheless, the strongest evidence supported purely community-based use.

Use of topical oils in preterm newborns improved the rate of weight gain but had no impact on any other growth outcomes. Another systematic review [28] showed that topical oils had a greater magnitude of impact on weight gain, with additional benefits to crown-heel length, which we did not observe, possibly owing to our LMIC-only population restriction, as we only identified one new trial [142] of relevance.

It has been shown that providers tend to treat PSBIs with broad-spectrum antibiotics at higher doses and for longer durations owing to perceived high risk of mortality in neonates with PSBI [156]. As in the Cochrane review, we found that simplified antibiotic regimens did not result in increased all-cause neonatal mortality risk. There was also no difference in treatment failure or adverse effects in neonates receiving simplified antibiotic regimens when compared to standard care (with injectable antibiotics), suggesting that it is a safe, acceptable [157], and affordable [45] treatment approach for clinics and home-based care that may contribute to improved antibiotic stewardship. Recent community-based research on simplified regimens for management of PSBIs in Pakistan [42] and Nigeria [158] similarly found low case fatality rates.

While chlorhexidine cleansing of the umbilical cord showed no impact on mortality, there was a significant benefit on the risk of omphalitis when compared to dry cord care. These results are comparable to those of a Cochrane review which also showed chlorhexidine cord cleansing reduced the risk of omphalitis by 52% but similarly showed no effect on mortality [159]. Furthermore, we found that, when compared to dry cord cleansing, multiple cleansing is more effective than single cleansing in preventing omphalitis, and chlorhexidine cleansing at any frequency was effective in preventing mild, moderate, and severe omphalitis, when compared to dry cord care. It is important to note, however, the substantial to considerable between-study heterogeneity. Although chlorhexidine cord cleansing in community settings is effective in preventing omphalitis, establishing the safety of this intervention in the long term is necessary before a case can be made for its widespread use in community settings, as is noted by another review on this topic [153]. Based on evidence from the same five trials in our meta-analysis, the WHO guidelines for a positive postnatal experience conditionally recommends the application of chlorhexidine cord cleansing only in contexts where unhygienic cord practices are known to occur [54].

Our findings also support the use of emollients for atopic dermatitis in term newborns, contrary to a recent review which found emollients in term newborns had little to no effect on atopic dermatitis [31], though their review contained HIC data where our results are LMIC-specific. Trial evidence that is adequately powered to detect a difference in mortality is needed to determine if there is a survival benefit to regular application of sunflower seed oil in VLBW newborns compared to mustard oil. Much more evidence is needed to assess the use of synbiotic supplementation in community-based settings.

Home-based antibiotic delivery was found to reduce all-cause neonatal mortality but not sepsis-specific neonatal mortality when compared to a standard hospital referral, suggesting a modest yet positive impact on newborn survival when hospital care is infeasible. One explanation for the reduction in all-cause mortality is that the study population included newborns who do not have sepsis but were in fact small for gestational age, or hypoglycemic, or had a range of other illnesses for which antibiotics are unnecessary for clinical improvement. Notably, there has been no new LMIC data published on our outcomes of interest since the Cochrane review’s [44] literature search; thus, more research is needed in this area.

Implementation is predicated on the supply of community healthcare providers such as community health workers as well as sufficient receptiveness from these providers to be trained and from the community to allow the initiation and continuation of such care within their homes [44]. Due to the complex nature of diagnosing nonspecific signs and symptoms and prescribing optimally for PSBIs, research has shown that theoretical and practical training must be ongoing, as provider classification and treatment errors decrease over time [156]. Implementation also requires the dissemination of medicines and basic medical equipment. Not only must providers be continually supplied with antibiotics, they must secure basic diagnostic tools such as thermometers, acute respiratory infection timers or secondhand clocks, weighing scales, and other decision-making aids [156].

While concerns remain over antibiotic misuse and overuse in the community, especially in the absence of cultures or susceptibility testing to guide therapy, our findings on managing PSBIs in the community setting in instances of hospital inaccessibility or refusal, as a safe and modestly effective alternative, support the WHO’s existing recommendations [41]. The most effective community-based antibiotic delivery will occur where there are robust community-based packages of strategies, substantial community engagement, and a well-coordinated, well-documented, well-supervised, and well-funded health system [158, 160].

For the reduction of AMR and prevention of HAIs in facility settings, the strongest evidence supported the implementation of multimodal interventions over single interventions. Based on current evidence, topical emollient oils are promising for the prevention of invasive infections in preterm newborns in facility settings. Probiotics can be used for the prevention of all-cause mortality, invasive infection, and NEC in preterm newborns in facility settings, but there is little evidence that probiotics are beneficial in extremely preterm or ELBW newborns. Further evidence is needed on synbiotics administered to preterm newborns in healthcare facilities, with careful consideration to the safety of study participants. While prophylactic systemic antifungal agents are effective for the prevention of invasive fungal infections in very preterm and VLBW newborns in LMICs, there is very limited data available from low-resource settings likely owing to the costs associated with implementation. For use in mixed settings, CBKs are a promising intervention to reduce neonatal mortality, but there is limited trial evidence available and varying kit composition across trials. In the community setting, umbilical cord cleansing with chlorhexidine should be done to prevent omphalitis only as an alternative to unhygienic cord applications. Topical emollients may be effective for the prevention of atopic dermatitis in term newborns, but more evidence is needed. Purely home-based antibiotic treatment for PSBIs may be effective in reducing all-cause neonatal mortality, and home- and first-level clinic-based antibiotic delivery of simplified regimens for PSBIs provides a safe alternative to hospitalization when a referral is inaccessible or unacceptable, but these necessitate extensive resources including reliable supply chains, community healthcare worker training and supervision, system-wide support, and community engagement. Our comprehensive analysis of the evidence for neonatal infection prevention and treatment interventions will support the effective design and implementation of these interventions where newborn care is delivered in LMICs.

The authors gratefully thank Li Jiang for her contributions to review topics including the prevention of HAIs, topical emollients, probiotic supplementation, and synbiotic supplementation. We also thank Nina Zhou for her input on this project.

An ethics statement is not applicable, as this review is based on published literature.

The authors have no conflicts of interest to declare.

This project was funded by the Bill & Melinda Gates Foundation (#INV-042789). The funder had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; or decision to submit the manuscript for publication.

L.H. and Z.A.B. conceived of and designed the study. R.L.H., S.R., D.S., R.Y., M.A., T.M., H.A.N., L.M., and S.H. screened articles for inclusion and conducted data extraction. R.L.H., S.R., D.S., R.Y., M.A., T.M., H.A.N., L.M., and S.H. completed data analysis. L.H., T.V., and Z.A.B. supervised the review process. R.L.H. drafted the manuscript with assistance from S.R., and M.J.S., A.D., S.E.C., D.H.H., and Z.A.B. critically reviewed the manuscript for important intellectual content. All authors approved the final version of the manuscript.

The data presented in this study are available in supplementary materials. Further inquiries can be directed to the corresponding author.