Neonatal Sepsis Resistance Across Hospital Settings

Evidence

Three recent hospital studies examined bacterial isolates from newborns, but they did not ask exactly the same question. One Indonesian study analyzed multiple specimen types from neonates with confirmed bacterial infections. A study at Dilla University Referral Hospital in southern Ethiopia enrolled neonates with suspected sepsis and tested blood cultures. A third retrospective study examined culture-proven sepsis episodes in one tertiary neonatal intensive care unit. Reading them together reveals both a recurring resistance burden and major differences in which organisms were most visible.

The Indonesian cross-sectional study included 223 isolates collected from January 2022 through December 2024 at a tertiary referral hospital in Semarang. Investigators used Gram staining, colony morphology, and the automated VITEK2 Compact system for identification and susceptibility testing. They defined multidrug resistance as nonsusceptibility to at least one agent in three or more antimicrobial classes. Overall, 148 of 223 isolates, or 66.4%, met that definition.

Resistance varied sharply by specimen. Among 55 blood isolates, 46 (83.6%) were multidrug resistant. The reported proportions were 91.7% in sputum, 63.1% in feces, and 50.9% in urine. Across the major organisms, multidrug resistance ranged from 45.6% in Escherichia coli to 71.9% in Klebsiella pneumoniae; all 17 Pseudomonas aeruginosa isolates were classified as multidrug resistant. Gram-negative isolates had a higher multidrug-resistant proportion than Gram-positive isolates (62.7% versus 24.6%, p<0.001). The study did not include neonatal outcomes or detailed clinical characteristics, so these microbiology results cannot show how resistance affected prognosis. Its mixture of blood, respiratory, urine, and fecal samples also means that not every isolate should be interpreted as bloodstream sepsis; some non-sterile-site findings may reflect colonization rather than invasive infection.

The Ethiopian cross-sectional study used a narrower sampling frame. From August to October 2023, researchers systematically enrolled 280 neonates with suspected sepsis and analyzed venous blood using standard microbiology and Kirby-Bauer disk diffusion susceptibility testing. Blood cultures were positive in 88 neonates, a prevalence of 31.4% (95% confidence interval 25.7%–36.8%). E. coli accounted for 27 positive cultures (30.7%) and K. pneumoniae for 18 (20.5%). Overall, 70.2% of isolates were multidrug resistant. The abstract reported 80%–100% resistance among Gram-negative and Gram-positive bacteria to ampicillin, amoxicillin-clavulanate, ceftriaxone, and ceftazidime. Because only the abstract was ingested, those reported aggregate results support this comparison, but finer organism-by-drug conclusions would exceed the available text.

The tertiary-NICU cohort offers a contrasting organism profile. It retrospectively reviewed admissions from 2014 through 2019 and included 190 infants with 212 culture-proven sepsis episodes after excluding contaminated cultures and incomplete records. The infants were predominantly preterm: mean gestational age was 31.13 weeks and mean birth weight was 1,613.61 grams. Late-onset sepsis accounted for 183 episodes (86.3%). Gram-positive organisms made up 73.1% of isolates, led by coagulase-negative staphylococci at 64.6% of episodes, while Gram-negative organisms accounted for 23.1% and fungi for 3.8%. Seven of 16 K. pneumoniae isolates were extended-spectrum beta-lactamase positive, and two were meropenem resistant. Sepsis-related mortality was 7.9%, with episode-based mortality higher in Gram-negative than Gram-positive infections. These cohort details are also abstract-level evidence and cannot establish why its distribution differed from the other hospitals.

Analysis — Local ecology shapes the resistance signal

The cross-study convergence is the presence of substantial resistance within neonatal care, not a single universal pathogen hierarchy. The Indonesian mixed-specimen study and Ethiopian blood-culture study both foregrounded Gram-negative organisms and reported multidrug resistance near two thirds or higher. The tertiary-NICU cohort instead found a Gram-positive majority dominated by coagulase-negative staphylococci, while still identifying resistant K. pneumoniae and greater mortality among Gram-negative episodes. One plausible analysis is that the signal seen by surveillance depends on who is sampled, which body sites are cultured, how contamination is excluded, and whether the unit cares for many preterm infants with late-onset infection. That interpretation is consistent with the different designs, but the studies did not directly test design effects or compare hospitals under one protocol. Their agreement therefore supports attention to local microbiology as a research variable; it does not justify transferring one hospital's organism distribution, resistance percentages, or clinical approach to another. A valuable next research step would be prospective multicenter surveillance using harmonized case definitions, specimen rules, susceptibility standards, and outcome collection.

Limitations

All three studies were observational and hospital-specific. The Indonesian study was descriptive and single-center, combined sterile and non-sterile specimen types, lacked neonatal risk factors and outcomes, and used phenotypic testing without molecular confirmation of resistance mechanisms. Its unusually high percentages for small organism groups, including 17 P. aeruginosa isolates, are numerically fragile and should not be treated as population estimates.

The Ethiopian study covered only three months, and its associations with factors such as prolonged hospital stay or delivery characteristics came from a cross-sectional analysis, which cannot establish direction or causality. The tertiary-NICU cohort was retrospective, covered 2014–2019, and predominantly represented preterm, low-birth-weight infants and late-onset episodes. Even with stated contamination exclusions, coagulase-negative staphylococci can be difficult to interpret across differing blood-culture practices.

Two of the three sources were available to this briefing only as PubMed abstracts. That limits assessment of subgroup definitions, missing data, laboratory quality controls, regression choices, and full numerical tables. Differences among VITEK2 testing, Kirby-Bauer disk diffusion, study periods, enrollment criteria, and sample types prevent a direct ranking of hospitals. The studies document resistance patterns and associations; they do not test an intervention, establish that resistance caused an outcome, or support individualized treatment decisions.