Neonatal sepsis deaths in LMICs are Gram-negative, and mostly Klebsiella

Design

prospective, multicentre, multicountry observational post-mortem surveillance with minimally invasive tissue sampling and panel adjudication

Population

2,609 neonatal deaths (aged <28 days) at 11 sites in Ethiopia, Kenya, Mali, Mozambique, Sierra Leone, South Africa and Bangladesh, 2016–2023

Primary outcome

pathogen-specific causes of infection-related neonatal death

Effect

infection in the causal pathway in 44.0%; Gram-negatives 74.1% of infection deaths; Klebsiella pneumoniae 41.7%, Acinetobacter baumannii 25.7%, E. coli 10.4%, Group B Streptococcus 5.7%

Cause of death in neonates in low- and middle-income settings is usually assigned clinically, which means the pathogen is usually unknown. CHAMPS did it properly: 2,609 neonatal deaths from December 2016 to December 2023 across sites in Ethiopia, Kenya, Mali, Mozambique, Sierra Leone, South Africa and Bangladesh, with minimally invasive tissue sampling — blood, cerebrospinal fluid, rectal, nasopharyngeal and oropharyngeal swabs, and brain, lung and liver biopsies — worked up by culture, PCR array and histopathology, with a multidisciplinary panel adjudicating each death.

Infection lay somewhere in the causal pathway in 1,147 deaths (44.0%) and was the underlying cause in 432 (16.6%). Gram-negative bacteria were attributed to 850 (74.1%) of the infection deaths. Klebsiella pneumoniae appeared in 478 (41.7%), Acinetobacter baumannii in 295 (25.7%), Escherichia coli in 119 (10.4%) and Group B Streptococcus in only 65 (5.7%). Polymicrobial infection was found in 31.4%.

The standard empirical neonatal sepsis regimen — ampicillin or penicillin with gentamicin — is built around Group B Streptococcus and E. coli. That is not what is killing these babies. Klebsiella and Acinetobacter in this setting are frequently resistant to gentamicin and often to third-generation cephalosporins, and the study's authors say directly that empirical treatment needs review. The other headline number is that over 80% of these deaths were judged avertable under current or improved facility conditions.

• Know your own unit's neonatal isolates and resistance pattern — the empirical regimen should follow that, not a textbook derived from high-income settings.
• Acinetobacter at a quarter of infection deaths is a nosocomial signal: it points at lines, ventilators, hand hygiene and cohorting, not at community-acquired sepsis.
• A third of these deaths were polymicrobial, which single-organism blood culture thinking systematically misses.
• Indian neonatal units report a very similar Klebsiella-dominant picture; this is the closest large post-mortem dataset to Indian practice that exists.
• Infection prevention in the unit does more here than antibiotic choice — 80% avertable is a statement about care systems.

Why it matters

The organisms behind these deaths are not the ones first-line empirical therapy was designed to cover.

Don't overread it

This was observational post-mortem surveillance at high-mortality sites; it describes which pathogens are implicated in deaths, not which antibiotic regimen would prevent them.

The statistics, in plain English

These are attributions by an adjudication panel, not laboratory-confirmed causation in each case, and highly sensitive PCR on post-mortem tissue will detect organisms that were colonising rather than killing — that is what the adjudication process exists to filter. The 44% figure counts infection anywhere in the causal chain, which is a deliberately wide net; the 16.6% underlying-cause figure is the conservative one. Note also that these are deaths, not all sepsis, so the pathogen mix is weighted toward the organisms that kill rather than those most commonly isolated.

Read the rest in the app

You have read your two free briefings this month. The app carries all 27 specialties, every morning, free — and this finding is waiting in it.

Get Daily Doctor — free

Scan to keep reading on your phone. No account needed to start.