By Anais Cottage-stone · 25 August 2026
On Thursday, 23rd July 2026, NeoNET AFRICA held the second seminar of its 2026 series, From Detection to Prevention: Surveillance and IPC in Neonatal Sepsis. Chaired by Dr Felicity Fitzgerald (Imperial College London, UK, and The Health Research Unit Zimbabwe), the session followed a single line of argument across three talks: you cannot prevent what you are not measuring, and measurement only helps if it is simple enough to sustain.
Watch the full seminar recording here.
Dr Christina Obiero (KEMRI-Wellcome Trust Research Programme, Kenya) opened with lessons from setting up neonatal sepsis surveillance across African hospitals through SNIP-Africa.
The case for surveillance is straightforward. Empiric antibiotic choices vary widely between sites, and the resistance profile of the organisms actually causing neonatal sepsis differs from what those regimens assume. Without local data, treatment guidance is guesswork. The surveillance objective is correspondingly broad: to build the capacity and the methods to track neonatal bloodstream infection, pathogen and susceptibility profiles, the burden and trend of resistance, colonisation prevalence and antibiotic prescribing across neonatal units in sub-Saharan Africa.
What makes her talk useful is the honesty about what building such a system involves. It is not one problem but four. Clinically, it needs standard case definitions, enrolment, follow-up, outcome ascertainment and consent. In the laboratory, pathogen identification, susceptibility testing and quality assurance. For data, case report forms and electronic capture, validation, protection and reporting. And underpinning all of it, governance: ethics approvals, data sharing agreements, standard operating procedures, site monitoring and resources.
The design principles were deliberately modest, and that is the point. Collect simple, routinely available data at three moments, namely admission, clinical deterioration and exit. Start with a small defined population that can expand later. Prioritise completeness of a minimal dataset over breadth. Feed key metrics back to sites through a dashboard so the people collecting data see something useful in return. And build it so it can continue after the funded project ends.
Her lessons echoed that framing: standardise methods across sites, strengthen clinical and laboratory capacity together, collaborate across disciplines, learn from surveillance studies that came before, treat monitoring as continuous, and embed surveillance within routine care so it informs patient care and infection prevention rather than sitting alongside them. Local ownership, she argued, is what determines whether any of it survives.
Dr Victoria Nakibuuka Kirabira (Nsambya Hospital, Uganda) presented results from implementing an infection prevention bundle in two Kampala neonatal units, a 10-bed unit at Kibuli Muslim Hospital and a 25-bed unit at Naguru (China-Uganda Friendship) Hospital, the latter a regional referral hospital handling around 9,000 deliveries a year.
The bundle was unglamorous and concrete. Training and capacity building. Daily cleaning of surfaces, incubators, cots and high-touch areas, weekly comprehensive scrubbing of the whole unit, monthly fumigation, a one baby per cot policy, and designated isolation areas. Standard operating procedures covering feeding utensils and catheter and device care. And an antimicrobial stewardship component requiring septic screening before starting antibiotics for suspected late sepsis, stopping antibiotics at five days if cultures were negative, continuing beyond seven days only for culture-confirmed sepsis with sensitivity-guided therapy, pre-authorisation for third and fourth line agents, and regular audits with feedback to prescribers.
Comparing 99 neonates before the intervention with 455 afterwards, late-onset sepsis fell from 82.8% to 71.9%, an absolute reduction of 11 percentage points. Klebsiella pneumoniae infection fell from 8.1% to 3.3%, and all-cause mortality from 16.2% to 9.0%. Blood culture utilisation and length of stay both decreased.
The resistance data underline why the Klebsiella result matters. Across 87 confirmed isolates, K. pneumoniae was the most common at 26.4%, and 82.6% of those isolates were resistant to gentamicin, though 91.3% remained sensitive to amikacin. These are organisms that first-line empiric therapy frequently fails to cover.
Professor Angela Dramowski (NeoNET AFRICA and Stellenbosch University, South Africa) closed with the NeoIPC toolkit, which exists to solve the problem the first two talks kept running into.
Africa records around a million neonatal deaths a year, and neonatal survival has improved far less there than elsewhere. Healthcare-associated bloodstream infections are the most common infection type in neonatal units in low and middle income countries, with reported mortality ranging from 20% to 80% and substantial neurodevelopmental disability among survivors. Yet before NeoIPC there were no global guidelines or tools for neonatal infection surveillance, no agreed criteria to make data comparable between countries, no reference data against which a unit could benchmark itself, and methods too complex to be practical in the settings that most needed them.
The NeoIPC surveillance network now spans 28 partner sites across Europe and Africa. Participation is free and open to any neonatal unit. Sites use a toolkit of data sheets and a secure online submission platform to collect data prospectively during routine care, available at neoipc.org/surveillance/resources.
Professor Dramowski then showed what the toolkit produces, using a two year pilot at Tygerberg Hospital covering 340 very low birthweight neonates, 21% of all such admissions, across 15,571 patient surveillance days. Just over a third, 119 of 340, developed at least one healthcare-associated infection. Healthcare-associated sepsis occurred at 8.2 events per 1,000 patient days, necrotising enterocolitis at 2.1 and healthcare-associated pneumonia at 0.1.
Three findings deserve attention. First, only 21% of primary bloodstream infection cases had a pathogen identified, which is a stark reminder of how much neonatal sepsis is treated blind. Second, among the organisms that were identified, two thirds of Enterobacterales were third-generation cephalosporin resistant, a fifth were carbapenem resistant, and 71% of S. aureus isolates were methicillin resistant. Third, 60% of these neonates received at least one antibiotic, amounting to 2,242 antibiotic days across the cohort, with just over half of prescriptions falling into the WHO Watch category.
The toolkit also captures protective factors, which is where the picture is more encouraging. Kangaroo mother care was applied on 46% of patient days and expressed breast milk on 75%, reaching 87% and 98% of babies respectively. No probiotics were used.
The three talks fit together as a sequence. Dr Obiero showed what it takes to build surveillance that lasts, Dr Nakibuuka showed what a well-implemented prevention bundle can shift once you can measure it, and Professor Dramowski offered a free, standardised way for any neonatal unit to start measuring. The gap between detection and prevention is narrower than it looks, provided the counting is simple enough that busy units keep doing it.
