piatkat

This was a strong session with three speakers who looked at AI from three angles. Geraldine Boylan covered what it can already do, Sampsa Vanhatalo covered what we should ask of it, and Petra Hüppi covered what imaging can tell us about mechanisms. Two of the three opened by saying “I’m not an AI expert”, which suited a session that cared more about clinical questions than about technology. 1. AI for neonatal seizure detection — Geraldine Boylan (Cork, Ireland)The case. A term baby had a markedly discontinuous EEG from about 3 hours of age. Seizures followed in the middle of the night between Friday and Saturday. Full video-EEG was eventually set up and a neurologist confirmed seizures remotely. By the time the first drug was given, the baby was in status epilepticus and needed several antiseizure medications, including midazolam. On review, the first seizure had occurred at about 7.5 hours. It was low-amplitude and easy to dismiss as movement artefact. Could earlier detection have lowered the seizure burden and the number of drugs this baby needed? Take-home messages Neonatal seizures are mostly invisible. Many are subclinical from the start, and electroclinical uncoupling after ASMs hides even more. EEG is the only reliable tool. The background warns you. In large cohorts, a severely abnormal or discontinuous early EEG identifies babies at higher seizure risk. Be extra vigilant in these babies. The first seizure is a red flag. Seizures can escalate quickly in frequency, spread and amplitude. Burden matters, and so does time to treatment. Higher total and peak seizure burden are linked to poorer cognitive and language outcomes at 18 months and to abnormal outcome in HIE. The number of minutes of seizures before first treatment matters, and earlier treatment reduces burden. There is only one real-time RCT so far. The ANSeR trial covered 8 European NICUs and used an algorithm from around 2014. Algorithm support increased the proportion of seizure hours detected from about 45% to about 66%. Algorithms have improved a lot. Open datasets such as the Helsinki dataset and much better models have given recent algorithms what she called expert-equivalent performance. They detect both seizure events and seizure duration, so they can estimate burden. They also work from a single channel up to a full montage, which helps with artefacts. Ask harder questions about metrics. AUC alone is not enough, because seizure data are heavily imbalanced and most of the recording is not seizure. Ask for sensitivity, specificity, PPV, NPV, false detections, agreement on seizure burden with experts, and inter-rater agreement. Testing must be on truly unseen data, with no patient overlap and data from many centres and devices. Two algorithms can have similar AUCs while one produces far more false alarms. Jurassic Park analogy. The raptors learned to open doors with lever handles. Show them a round knob and they’re stuck. An algorithm trained on one centre’s data behaves the same way. What comes next: regulatory approval, clinical translation, and treatment protocols ready to act on the first detected seizure, plus clinical trials. Technical accuracy alone does not establish clinical benefit. 2. AI-based monitoring of cortical recovery and maturation in neurocritical care — Sampsa Vanhatalo (Helsinki, Finland)Take-home messages AI should not just emulate the expert. If it only reproduces what the EEG reader already says, it adds little. It should give clinicians more than that. Break down the bedside question. “How is the brain?” covers different time frames: what has happened, what is happening now, what will happen in the next hours, and how things evolve over hours to days and over weeks. Each needs a different approach and different underlying questions, such as suppression, encephalopathy grade, seizures, sleep state and maturity. Ground truth depends on the question. For seizures, background and sleep, models can learn from EEG experts. For maturation and prognosis, the ground truth must come from clinical information outside the EEG. Validation happens at two levels. The first is scientific validation for publication. The second is pragmatic: whether clinicians perceive the tool as valuable at the bedside. This matters most for adoption, but it is hard to measure and regulators don’t recognise it. One solution his group uses is a cloud platform where centres run the algorithms on their own data without sharing it. Background as a continuum. The brain is not “discontinuous” or “continuous” in discrete steps. Their model converts background into a continuous 0–100 score, which gives an objective multi-day recovery trend. Several studies link it to outcome. EEG-based “brain age”. Models trained on large preterm EEG datasets can estimate functional brain age and plot individual maturational trajectories. A preterm-age version was presented later at the meeting. An open question for clinicians: what error is acceptable, and what change in such a score is clinically meaningful? Engineers cannot answer this. The neonatal market is small, and getting from idea to certified product takes years and millions. Many algorithms are openly available for academic research, so use them and collaborate. 3. New horizons in neonatal brain imaging — Petra Hüppi (Geneva, Switzerland)Take-home messages MRI still helps with counselling. In the case of “Hugo”, born at 26 weeks with severe IVH, MRI showed blood in the ventricles but intact parenchyma and metabolism. His parents chose to continue care. Faster scanning. MR fingerprinting, accelerated with machine learning, can produce several quantitative maps (T1, T2, synthetic contrasts, water fraction) in about 5 minutes, compared with roughly an hour of conventional sequences. Grey and white matter mature out of step in preterms. In healthy infants, T1 maturation of white matter and cortex track each other. In preterm infants, the cortex lags behind its white-matter connections. Low-field portable MRI (64 mT) makes imaging possible in low-resource settings, where most preterm babies are born. One example showed effects of maternal anaemia on infant brain growth, especially in the basal ganglia and corpus callosum. Socioeconomic factors weigh heavily. ML-based outcome prediction from MRI had only modest predictive value, and much of that depended on socioeconomic status. Preterm infants at term-equivalent age have reduced cortical folding, with fewer, shorter and shallower sulci. Clinical factors (birth-weight z-score, IUGR, multiple pregnancy, prolonged ventilation) explained about 3% of the folding variation. Maternal education explained 7–18%. Imaging can test interventions. In NICU music intervention studies, the music group had the most mature cortical folding. Interestingly, babies in an open bay did better than those in single rooms without parents present. Preterm birth affects specific networks. These include frontal, temporal, limbic and paralimbic connectivity, thalamic connectivity, the “rich club” hubs, and the salience, auditory and sensory networks. Variability in the BOLD signal between 33 and 40 weeks matches gene expression patterns in brain tissue at the same ages, which links imaging to biology. The mother’s singing and instrumental music engage different networks, so interventions can target specific networks. What’s coming next: glymphatic (brain waste-clearance) imaging and iron/myelin mapping at 7 T, and imaging of individual cortical layers. Overall take-homeThe algorithms are getting good. Seizure detection in particular is close to ready. What’s missing now is clinical: protocols that act on alarms, clear statements of which errors and changes matter, validation against real bedside questions, and trials showing patient benefit. Clinicians need to drive this. Disclosure: This is an AI-assisted summary of the session, generated from a recording made by our team in attendance. We believe it is as accurate as our own notes would be, but please refer to the speakers’ original work before applying anything in clinical practice.

This was a packed NEC session. It moved from the microbiome to the bench, then to future trials, a national probiotic cohort, and the brain. One message ran through every talk: NEC is probably not one disease. Until we treat it as a group of subtypes, prediction, prevention and trials will keep struggling. 1. The influence of the microbiome on development of NECDaphne Klerk (Netherlands) Daphne gave a clear tour of 15 years of NEC microbiome research. The early 16S era (2013–2017): Studies were small, but the pattern was consistent. Dysbiosis seemed to precede NEC, with a bloom of Gram-negatives (Gammaproteobacteria/Enterobacteriaceae, the family Klebsiella belongs to). The proposed mechanism was LPS → TLR4 → excessive inflammation. Barbara Warner’s Lancet study (46 NEC cases, 3 US centres) found that differences emerged only after the first month of life. The signal was strongest in infants born before 27 weeks. Christopher Stewart’s Newcastle group found that a Bifidobacterium-dominant community was associated with healthy outcomes. Bifidobacteria are good at digesting HMOs. The shotgun metagenomics era (2016 onwards): This method has strain-level and functional resolution, but it is more expensive and computationally harder. Some studies linked particular E. coli strains to worse outcomes. One study found a Klebsiella signal only about 2 days before diagnosis. It came with adhesion genes and faster replication rates. The study she called the most important so far: This was a prospective, longitudinal metagenomic study of 236 preterm infants in 3 NICUs, 48 of whom developed NEC. A NEC-associated pattern (more Klebsiella, lower diversity) appeared only in late-onset NEC (>40 days). No taxon, pathway, growth rate or virulence gene separated pre-NEC samples from controls. Medication was the dominant force shaping the preterm gut microbiome. It outweighed gestational age and mode of delivery. Why the field is stuck: Studies are mostly single-centre and underpowered. They are confounded by medication and feeding. Geography also matters, because location affects the microbiome. And sites may not mean the same disease when they say “NEC”. The gut virome is an emerging area. There are hints of early- versus late-onset differences. Probiotics are most likely protective. They are also further evidence that microbes are involved. But NEC incidence doesn’t fall to zero, and the optimal strain combination is unknown. Her own cohort included 150 infants born <30 weeks or <1000 g, of whom 11 developed NEC. Routine probiotics were introduced partway through the study. Microbial composition separated clearly by probiotic exposure. Probiotics also changed gut function: bile acid profiles shifted in a way that matched the genetic profile of the supplemented strains. No overall composition or diversity signal predicted NEC. Only a few species, including Klebsiella pneumoniae, were associated with risk. Q&A, on postbiotics: Postbiotics gained interest after the FDA warning on live probiotics about three years ago. They are probably safer, but she expects less impact than from live probiotics, which act in the gut and produce active metabolites. The postbiotic literature is also fragmented into single-centre and mouse studies. She would rather the field focus on identifying the best probiotic strains. Take-home messages: There is no universal pre-NEC microbial signature. Research should shift from composition to function, and to how the microbiome interacts with the developing immune system. Sites need to agree on NEC subtypes before running meaningful multicentre studies. 2. NEO-O-100: Dose-dependent effects of dexamethasone on epithelial injury in human intestinal epithelial cellsParvesh Mohan Garg (USA) Rationale: Steroids are standard anti-inflammatories in adult IBD, but we give none in NEC. Garg’s group tested how dexamethasone affects intestinal epithelium injured with LPS, using human fetal small-intestinal epithelial cells. Cell viability: LPS killed most of the cells. Low-dose dexamethasone preserved viability. High-dose dexamethasone was itself cytotoxic. The effect was U-shaped. Scratch (wound-healing) assay: Dexamethasone did not impair epithelial migration or wound closure. Griess assay (nitrite as a marker of inflammatory stress): Low-dose dexamethasone brought nitrite production back to baseline. This held at both LPS concentrations, including the higher one. Clinical follow-up (unpublished, in preparation): In their surgical NEC cohort, there was about a 20% mortality difference between infants who did and did not receive hydrocortisone around surgery. Take-home message: In vitro, low-dose steroids protect injured gut epithelium and high doses harm it. This is hypothesis-generating only. Dosing translation and clinical studies are the next step. 3. Future studies in the field of NECNicholas Embleton (UK) Nick gave a broad, forward-looking talk. Disclosures: He is the PI of an industry-sponsored enteral insulin trial. He has advised a US physician being sued for allegedly causing NEC by using a fortifier. Context: NEC has been described for over 200 years. He cited Michael Obladen’s “150 years of fruitless search for the cause” and doubts a single cause will ever be found. More children in Europe and the developed world die from NEC than from leukaemia. Yet academic neonatology gets a fraction of paediatric oncology funding. Challenges: Definitions and heterogeneity. NEC in a 24-week infant on day 10 is not the same disease as NEC in a 28-week infant on full feeds at day 30. He described two peaks: around days 10–12 before full feeds, and around 4–5 weeks in fully fed infants. Animal models mostly rely on hypoxia-ischaemia, which is probably not the main mechanism in preterm infants. Biomarkers have been slow to develop. Outcome definitions are soft: Stage 2 NEC varies between units. “Required surgery” is not the same as “received surgery”. “Death from NEC” is often unclear, since an infant may die with NEC rather than from it. Opportunities across four stages: The at-risk infant (prediction and prevention): Multi-omics will matter (metabolomics, proteomics, transcriptomics), not the microbiome alone. “I was absolutely convinced we would find the microbiome signal for NEC… I don’t think we have any chance.” The infant developing NEC (diagnosis): Integrate physiology, vital signs, imaging, the electronic patient record and bedside omics, with longitudinal and clinically relevant AI. Urine may be an underused sample: it is non-invasive and easier than stool, which lags hours to days behind the disease. A combination of I-FABP, L-FABP and TFF3 looks promising, as do tight-junction proteins. Volatile organic compounds and calprotectin in stool are also candidates. Confirmed NEC (treatment): Surgical choices: timing, drain versus laparotomy, stoma versus anastomosis. Remote ischaemic conditioning (Toronto/Southampton collaborations), now heading to clinical trials. TLR4 blockade: HMOs such as 2'-FL, short-chain fatty acid metabolites, and the Hackam group’s Compound 34. After NEC (the “NEC exposome”): Everything the infant goes through can affect brain outcomes: pain, loss of kangaroo care and closeness, interrupted breast milk, anaesthetics and excipients, metabolic disruption, long lines and parenteral nutrition. Human milk: Fresh mother’s milk remains one of the most effective interventions. Donor milk may vary in protective potential, for example by its HMO and IgA patterns. Formula regulates about 75 components, compared with the vast bioactive repertoire of breast milk. Fresh breast milk contains roughly 10⁴–10⁷ cells/mL, many of them stem cells that can traffic to peripheral organs. That is potentially billions of cells per day, but only in fresh, warm milk. Q&A: Most trial-ready intervention? Milk-derived stem cells. He also stressed supporting mothers to keep lactating while their infant is sick and nil by mouth, and building on oral (buccal) colostrum to give cells before full feeds restart. Abdominal X-rays? He is not optimistic. We have all seen florid NEC with an unremarkable film, and “pneumatosis” in infants who did fine. He sees more promise in intestinal ultrasound and NIRS. Call to action: Join the ESPR NEC Special Interest Group, and involve parent organisations in designing research. 4. NEO-O-101: Probiotics and risks of death, NEC and sepsis in very preterm infantsStefan Johansson (Sweden) Disclosure: He is co-founder of the company supplying the probiotics used in Sweden. Setting: After the 2020 ESPGHAN position paper, the Swedish Neonatal Society recommended probiotics for infants born at 28–31 weeks. Probiotics were to start in the first days of life and continue to 34 weeks PMA. Uptake rose quickly to about 65%. Design: Swedish Neonatal Quality Register, 2017–2024. About 4,600 very preterm infants, of whom about one third were exposed. Composite outcomes, treating death as a competing risk. Poisson regression adjusted for birth year and key covariates, clustered by birth hospital. Sensitivity analyses with propensity-score matching and inverse probability weighting gave consistent results. Results (probiotics vs none): Outcome Exposed Unexposed Death and/or NEC 0.8% 3.4% (aRR 0.28) Death 0.3% 1.4% NEC stage ≥2 0.6% ~2.5% Surgical NEC 1 case in ~1,500 infants — A mediation analysis suggested much of the mortality reduction ran through less NEC. There was no association with sepsis. A national NICU survey found no reported cases of probiotic sepsis. The register doesn’t capture this outcome, and rare events cannot be excluded. Limitations: The design is observational. The baseline risk was low; all infants were initially fed human milk. Outcome misclassification and immortal time bias are possible. The results should not be extrapolated to extremely preterm infants. Q&A (France), on blood cultures: Sweden uses a single combined paediatric bottle, without separate anaerobic culture. The Karolinska lab has shown the strains are culturable and asks clinicians to note probiotic use on the request. In the Netherlands, aerobic and anaerobic cultures are routine. The questioner warned that few studies have used proper anaerobic culture, so “no probiotic sepsis” should be interpreted carefully. Johansson agreed: the risk is low, but it is real with any strain. Take-home message: In real-world national data, routine probiotics in 28–31-week infants were associated with a large reduction in death and/or NEC. Surgical NEC almost disappeared, and there was no sepsis signal. 5. Long-term outcome of NEC survivorsAlexandre Lapillonne (France) Lapillonne covered NEC and spontaneous intestinal perforation (SIP) together, since the literature often mixes them. EPIPAGE-2 (French national cohort): 95 infants with NEC, 38 with SIP, and more than 3,000 controls; 85% were followed to 2 years. Death or disability: NEC 35%, SIP about 46%, controls 14%. The adjusted odds were about 2.4× for NEC and 3.1× for SIP. Most of the excess was in-hospital death. Developmental delay at 2 years (ASQ): SIP 71%, NEC about 50%, controls 44%. The difference was significant only for SIP. The domains affected differed: SIP mainly affected gross motor skills (41% vs 13% in controls). NEC mainly affected problem solving (26% vs 12%). Unpublished two-centre SIP study (n = 29, perforation at median day 5): More than half died or had severe brain lesions. About two-thirds of brain lesions appeared or progressed after the perforation. GA, birth weight, antenatal steroids, blood culture, platelets and CRP did not predict lesions. The white cell count did. Median WBC was about 5 with normal ultrasound, 8.8 with IVH and 32 with white matter damage; each doubling of WBC was associated with about 6× higher odds of white matter damage. His hypothesis is that SIP has two phenotypes: a low-inflammation one that largely spares the brain, and a high-inflammation one with white matter damage. Literature (four meta-analyses over 20 years, plus large cohorts): NEC raises the risk of neurodevelopmental impairment by about 1.4–1.9×, with no improvement over two decades. The risk applies to 29–32-week infants too, not only to the extremely preterm. Vision and hearing impairment are about 4× more likely; CP, motor delay and cognitive delay about 2×. Epilepsy and behaviour do not appear to be affected. Medical NEC carries little extra risk. Surgical NEC carries an odds ratio of about 1.8–2.9. In the ELGAN cohort, surgical NEC followed by late-onset sepsis carried the highest risk on brain-structure measures, roughly 8–9×. The IQ deficit persists to school age (6 and 9 years), mainly after surgical NEC. Mechanism: inflammation as the gut–brain mediator. The injured gut sends danger signals (TLR4). These spread through the circulation and reach the blood–brain interface. Microglia are activated. Pre-oligodendrocytes fail to mature, which leads to diffuse hypomyelination (maturation arrest rather than cell death). Secondary disturbances follow in axonal and cortical development. Experimental evidence supports this sequence. In TLR4-deficient mice, NEC does not cause brain injury. The severity of neuroinflammation tracks NEC severity, and inflammation alone can reproduce the lesions. Repeated inflammatory hits during the window of pre-oligodendrocyte vulnerability (chorioamnionitis, NEC, SIP, late-onset sepsis) add up. In practice, there is no tested neuroprotective intervention after NEC. Three levers remain: Prevent NEC. Limit second hits, especially late-onset sepsis. Consider laparotomy over peritoneal drain. Identify high-risk infants and provide structured neurodevelopmental follow-up. Take-home messages: NEC is one of the strongest acquired predictors of impairment, with about 2× the risk overall. The risk is mostly driven by surgical NEC. SIP is not benign. White matter is the main target, and about a third of infants also have cerebellar injury. Disclosure: This is an AI-assisted summary of the session, generated from a recording made by our team in attendance. We believe it is as accurate as our own notes would be, but please refer to the speakers’ original work before applying anything in clinical practice.

This was a short, interactive session from EBNEO (Evidence-Based Neonatology) by Dr Kiran Shrivastava and Prof Gautham Suresh on low-value care. It drew lively input from an audience working in very different health systems: India, the US, the UK, Sweden and the Netherlands. What is low-value care?There is no single agreed definition. A working one is: any intervention where the evidence suggests little or no benefit to the patient. It can take three forms: Ineffective: it works in some subgroups but not in everyone we use it on. Inefficient: it works, but costs too much, comes at the wrong intensity, or takes too long. Unwanted: it is technically useful but doesn’t meet the family’s needs. The example given was term-equivalent MRI for all preterm babies. It produces images, but at that moment it tells parents little about the future. What they often want is a conversation about what it means and what follow-up looks like. The speaker was equally clear about what low-value care is not. It is not cost-cutting or rationing. It is not withholding care we know works. And it is not the opposite of caring. Most of it sits in a grey zone, and avoiding it means more thinking about each individual baby, not less. Examples from the roomRoutine admission bloods and routine blood gases in stable babies “because it’s the protocol” Antibiotics started for virtually every NICU admission, plus routine X-rays and blood cultures Automatic NICU admission for every baby under 36 weeks, even when transitioning well, which separates mother and baby Continuous saturation monitoring in bronchiolitis, which prolongs hospital stay Escalation of care, such as intubation, when it isn’t clearly needed Choosing Wisely: the 2015 neonatal listThe Choosing Wisely campaign started about 15 years ago. Each specialty society names its own low-value practices. The 2015 newborn medicine recommendations were: No routine anti-reflux medication for symptomatic GER, or for apnoea/desaturations in preterm infants No antibiotics beyond 48 h in initially asymptomatic infants without evidence of bacterial infection No routine daily chest X-rays without a clinical indication No routine pneumograms before discharge No routine term-equivalent or discharge MRI in preterm infants This isn’t a rich-country problemThe Lancet Right Care series (2017) found overuse and underuse side by side in every setting, high- and low-resource, sometimes within the same hospital or even the same patient. Data from India were also cited showing more deaths attributable to poor-quality care than to no access to care at all. Why do we keep doing it?It isn’t ignorance. We attend conferences and we know the evidence. The speaker grouped the drivers into four categories: Behavioural. It is hard to stand back and do nothing, especially when families want action. We also tend to overestimate pre-test probability. And there is an asymmetry: the harm of overtesting is invisible and statistical, while the one missed diagnosis “has our name on it” in the M&M meeting or in litigation. In other words, defensive medicine. Social. Practice spreads person to person, shaped by where we trained, by hierarchy, and by who is on service this week. The speaker described “mindlines” (Gabbay & le May): researchers followed GPs for two years and found they relied on colleagues rather than guidelines. Mindlines are collectively reinforced, internalised, tacit guidelines that nobody questions or traces back to their source. Systemic. Order sets with pre-ticked boxes, nursing routines taught at induction, and guidelines that nobody actually uses. Economic. Fee-for-service systems reward more tests and longer stays. The speaker also summarised what sustains a practice as evidence, eminence and economics, with evidence counting for least. The vertebroplasty story illustrated this. After two negative trials, the orthopaedic societies advised against the procedure and orthopaedic surgeons stopped. Interventional radiology issued no such guidance, so their practice didn’t change. Voices from the floor on money: India: families receive a daily itemised bill, and clinicians discuss finances with parents almost every day. US: clinicians rarely know what anything costs. One suggestion was that showing the running bill in the electronic record could change ordering behaviour. In one hospital, whole-genome sequencing now needs a genetics consult first, but CBCs, X-rays and MRIs can still be ordered freely. UK: almost no pushback on investigations, which may itself entrench routine imaging. Netherlands: funding thresholds can create an incentive to keep a baby a few days longer. What can we actually do?Ideas from the audience: Make costs visible to clinicians, for tests and for materials. Know what you are looking for before you order. Don’t do the echo first and then work out the question. Teach trainees the same way. Use EHR “hard stops”, such as “Do you really want this? Why?” Build a culture where juniors, nurses and all staff can question why we do things. Nurses are often the ones who say “enough”, including when it is time for comfort care. Revisit unit practice and guidelines every couple of years. What the evidence says works: ❌ Education alone has little measurable effect, and gains fade over time. ✅ Audit and feedback, ideally delivered by respected local leaders (eminence works both ways). ✅ Accountability at the point of ordering: state the indication, as in antibiotic stewardship time-outs, and extend that to other tests. ✅ Peer comparison and collaborative QI with benchmarking against your region or network. ✅ Change the default: no automatic order sets; choose tests baby by baby. And AI?The room split. Optimists see AI as a fast evidence assistant for learners, used under senior supervision, and as a help in explaining risk to parents. Sceptics warned of “cognitive surrender”, where we stop thinking and let the tool decide, which would make overuse worse. Take-home messagesLow-value care exists everywhere, in every resource setting. It survives on structure, habit and culture, not ignorance. Before any test, ask: What am I expecting, and what will I do with the result? Education alone won’t fix it. Audit and feedback, accountability, peer comparison and better defaults can. Questioning a practice is not the opposite of caring. It is caring.

Ann-Marie Brennan gave a clear tour of the human milk fortification evidence, built around three questions: which fortifier, how to fortify, and when to start. She ended with the Irish national experience of integrating fortification with parenteral nutrition. Why fortify at all Unfortified human milk at 150–180 ml/kg/d provides only about 2–2.3 g/kg/d of protein. Even at very high volumes it covers only around 50–60% of the lower ESPGHAN limit for calcium and phosphorus. Multinutrient fortification (macronutrients plus minerals) improves in-hospital growth without increasing adverse effects. ESPGHAN 2022 recommends a multinutrient fortifier. Individualized strategies (adjustable or targeted) may be appropriate. There is insufficient evidence for routine human milk-derived fortifier. Fortification should start at enteral volumes of 40–100 ml/kg/d. HOW: standard vs enhanced vs individualized Retrospective modelling study (2026), infants <33 weeks: with standard fortification, only 19% of maternal milk samples and 2% of donor milk samples reached 4 g/kg/d of protein, and none reached 4.5. Enhanced fortification brought all samples to 4 g/kg/d, and 74% of maternal and 35% of donor samples to 4.5. Donor milk clearly lags behind. Note that this study estimated delivery, not clinical outcomes. Double-blind RCT, n=130, 24–30 weeks: protein and fat supplementation guided by point-of-care milk analysis was compared with standard care, where supplements were added when clinically indicated. Both arms met or exceeded recommendations. There was no difference in weight, length or head growth, in body composition, or in brain MRI indices. Her reading: milk analysis adds little where standard care already reaches targets. That does not mean individualization never works. Meta-analysis of individualized vs standardized fortification (10 RCTs, 454 infants): individualized fortification increased weight gain by +2.87 g/kg/d, with gains in length and head growth, fat-free mass (+121 g) and fat mass (+69 g), all at low certainty. At 18 months there was no difference in weight, but only 69 infants contributed. The gains are short-term, not long-term. WHICH: human vs bovine-derived fortifier AAP, BAPM and ESPGHAN do not recommend human milk-derived fortifier over bovine-derived fortifier. Jensen et al. RCT, n=228, 22–27 weeks, all on an exclusive human milk diet (maternal or donor): the only difference between arms was the fortifier. The composite primary outcome occurred in 35.7% with human milk-based fortifier vs 34.5% with bovine-based fortifier (OR 1.05). The trial was not powered for NEC alone. Systematic review and meta-analysis of 9 RCTs: the authors usefully separated trials by design. Fortifier alone, same base diet: no significant NEC reduction (RR 0.83, wide CI, very low certainty). Fortifier and base diet both differed (human fortifier plus donor milk vs bovine fortifier plus preterm formula): NEC risk was lower. The same pattern held for mortality, ROP and BPD: any benefit appears only when the base diet changes. 👉 Take-home: the NEC benefit belongs to the human milk diet, not to the source of the fortifier. WHEN: timing Recommendations vary: ESPGHAN: 40–100 ml/kg/d. European Milk Bank Association: 50–80 ml/kg/d in infants <1800 g. AAP: no threshold. Early fortification RCT: very early fortification was compared with fortification at about 120 ml/kg/d. Early fortification reduced weight z-score loss in the first two postnatal weeks and improved linear and head growth (secondary outcomes). It did not reduce postnatal growth failure or change body composition at 36 weeks (the primary outcome). Swedish national cohorts, about a decade apart: fortification in the first week of life rose from 3.6% to 27%. NEC also rose (1.7% → 5.8%), but the authors could not attribute this to early fortification. Survey of 15 countries and 185 NICUs: large variation in fortifier practice. Overall: early fortification looks feasible and safe, with no increase in NEC in the trials, and growth results are encouraging. Practice still varies widely. The overlooked transition phase: PN + enteral nutrition Brennan’s key message was that early fortification usually happens while the infant is still on parenteral nutrition. We have precise targets for exclusive PN and for exclusive enteral nutrition, but no international targets or upper limits for combined intakes. Faster advances combined with earlier fortification could push infants into excessive intakes without anyone noticing. The Cork / Irish solution: the “PremSmart” system A standardized system integrating parenteral and enteral nutrition: an aqueous standardized PN plus a standardized lipid infusion, with SPN dosed according to milk feed volume. Below 80 ml/kg/d of feeds parenteral targets apply; above 80, the higher enteral targets apply. Fortification starts at 80 ml/kg/d. The SPN was designed by modelling data from 59 infants. SPN complemented feeds well up to 80 ml/kg/d; beyond that it overshot. Clinical evaluation in 49 infants: Intakes were consistently within target, with no excessive intakes, and actual intakes closely matched modelled ones. The historical nutrient deficits during the transition phase disappeared. Early postnatal weight loss was significantly reduced, and better growth was sustained to term-corrected age. PremSmart is now national. It accounts for about 95% of all preterm PN used in Ireland, and about 95% of staff use the protocol. A national survey (105 responses) found that the most frequent local adaptation was the timing of fortifier introduction. Some units hold back because of NEC experiences (“people are scared after a few babies who get NEC”). Even robust data doesn’t automatically overcome NEC anxiety. Q&A highlights Duration: in Cork, fortification continues until discharge (anywhere from about 35 to 40 weeks), and to term-corrected age if the infant is still in the unit. Beyond growth, the fortifier supplies minerals (Ca, P, iron), so the unit gives no additional iron supplementation. They do not fortify beyond 40 weeks’ corrected age because of the calcium load and the risk of mineral disturbances. Discharge at about 36 weeks: decided case by case. Poor growth: some expressed breast milk feeds are fortified until term-corrected age. Good growth: fortification is stopped shortly before discharge to confirm that growth holds. Comment from a Swedish colleague: many units treat infants as “on PN” until they switch to “on enteral” and ignore the combination. Similar Swedish work also showed improved intakes and growth. 🎯 Take-home messages Which: with the base diet held constant, the source of the fortifier makes no clear difference. The benefit comes from the human milk diet. How: individualized or targeted fortification adds little when standard practice already reaches targets. Short-term growth gains in meta-analyses don’t persist to 18 months. When: early fortification appears feasible and safe, but recommended thresholds range widely (40–100 ml/kg/d). The gap: we need targets and upper limits for the PN + enteral transition phase, especially as feeds advance faster and fortification starts earlier. Disclosure: This is an AI-assisted summary of the session, generated from a recording made by our team in attendance. We believe it is as accurate as our own notes would be, but please refer to the speakers’ original work before applying anything in clinical practice.

Health and socioeconomic characteristics underlying educational attainment of children born preterm (James Boardman, UK)Background In the UK, 27% of children live in poverty after housing costs, which is about 4.3 million children. That figure has risen since 2010. The EU’s at-risk-of-poverty-or-social-exclusion figures are similar on average, ranging from about 12% in Slovenia to about 35% in Bulgaria. In earlier work, his group used parent reports at age 2 to define four developmental clusters: typical, communication, neuromotor, and multiple problems. Neighbourhood deprivation at birth was among the top six predictors for every cluster. For the communication cluster it was the strongest predictor, ahead of all medical factors. This study It is a new linkage between the National Neonatal Research Database (NNRD) and England’s National Pupil Database. The pupil database covers state-funded schools, which educate about 93% of children. This was a collaboration with Cheryl Battersby (Imperial College London). About 16,000 very preterm children were followed with six binary teacher-assessed outcomes from age 5 to 7: school readiness at 5 (Early Years Foundation Stage Profile) phonics screening at 6 reading, writing, maths and science at 7 Key findings 55.9% of very preterm children were not ready for school at 5. That is about 21 percentage points worse than the general population. Gaps in the other domains were 12 to 23 percentage points. There was a steep gestational age gradient. Almost 90% of 23-weekers were not school-ready, compared with just under 50% at 31 weeks. For 23–26 weeks, the most common pattern (29.5%) was not reaching the expected level in any domain between ages 5 and 7. The difficulties affect the whole classroom, not one subject. Deprivation and gestational age were additive. The curves ran in parallel and there was no statistical interaction. Deprivation hurts equally at every gestation. Mutually adjusted model: Risk factors: 23–25 weeks: OR about 2–3 most deprived neighbourhood: 20–40% higher risk free school meal eligibility (a close proxy for family income): OR 1.72 maternal smoking maternal diabetes summer birth: strong effect at every age (in England, summer-born children are the youngest in their school year) postnatal steroids. Given the era, these are mainly rescue steroids for severe lung disease or extubation, not low-dose prophylactic hydrocortisone. NEC, BPD, sepsis, ROP IVH grade 3–4 and PVL any nutrition other than exclusive breast milk Protective factors: higher birth weight female sex antenatal steroids No added risk: IVH grade 1–2 Disadvantages that faded: English as an additional language affected school readiness only and was gone by ages 6–7. 💡 Free school meal eligibility carried roughly the same risk as grade 3–4 IVH. Far more children are affected by the first than by the second. Take-home messages Prematurity and social disadvantage are a double burden. In no sense does prematurity override the social determinants. Only gestational age, sex and season of birth are fixed. The rest can potentially be changed: neighbourhood deprivation and family income smoking in pregnancy judicious corticosteroid use, both antenatal and postnatal fewer brain injuries, cases of BPD, NEC and ROP better nutrition, including breast milk There may be a role for precision support before school entry. One example is boys born into deprivation in the summer months. This needs further research. Q&A Person-centred or intersectional analyses? These are being considered in a deeply phenotyped cohort. Genetics? Boardman disputed that genetic risk for these outcomes is high in this population. ADHD and autism? New funding will map the age-2 clusters onto education. The communication cluster may include children later diagnosed with autism or ADHD. 🔹 Early growth trajectories and body composition in very preterm children: critical windows for fat-free mass accretion (Anna le Clercq, NL)Why it matters Body composition, especially fat-free mass (FFM), may matter more than weight gain alone. FFM is linked to better neurodevelopment. Body composition is rarely measured in follow-up clinics, so a child can look fine on weight while lean mass accretion lags behind. Study 1,402 children born at less than 30 weeks (median GA about 28 weeks, mean birth weight z-score about 0). Longitudinal weight z-scores from birth to 5.5 years. Two approaches: Data-driven: latent class trajectories. Hypothesis-driven: low FFM at 5.5 years (z-score < −1) compared with normal or high FFM, with growth looked at retrospectively. Findings Two trajectories were found: stable and early reduced growth. Birth z-scores were similar. The groups diverged during the first ~2 years and ran in parallel afterwards. The reduced-growth group had lower FFM at 5.5 years. The hypothesis-driven approach showed the same pattern. Children with low FFM had a larger early postnatal fall in weight z-score and a weaker recovery. The difference came from weight-gain velocity in the first two years. Fat mass % did not differ. Both groups were below the population mean, so the better weight gain reflects lean tissue, not excess fat. GA, birth weight z-score and BPD did not modify these differences. Take-home messages The first ~2 years after birth are a window to watch for later body composition. Whether nutritional or other interventions in this window can improve body composition, and in turn long-term health, is still to be tested. Q&A Nutrition data were available at 2 years corrected age for about half the cohort and showed no difference. Clinical decisions such as adding nutritional support could not be captured, and may have been made for the slower-growing children. Which trajectory is better if weight at 5.5 years ends up similar? The authors hypothesise the higher-FFM group will do better neurodevelopmentally, but the cohort is too small to test IQ. There were no adverse cardiometabolic signs at 5.5 years. Follow-up continues. 🔹 Immune cross-talk between mother and baby – basis for longevity (Anke Diemert, Germany)A maternal–fetal medicine view. The fetal immune system develops in layers The fetal immune system is not simply “immature.” It builds up in overlapping waves of cell lineages from the yolk sac, liver and bone marrow. Some lineages exist only before birth. These cells shape immunity, barrier tissues and neurodevelopment. Preterm birth interrupts this process, and the consequences can be long-lasting. Chronic maternal immune activation Stressors in modern pregnancies, such as high BMI, smoking, infections and lifestyle, drive chronic maternal immune activation. For the mother, this raises the risk of complications such as preeclampsia and hypertension. For the child, it reprograms the developing immune system. Examples given: antenatal betamethasone affecting immune cell development gestational diabetes and excess weight gain leading to a pro-inflammatory state in cord blood preeclampsia associated with lower neutrophil counts and impaired neutrophil function Mode of birth also changes microbiome seeding. Windows of opportunity Vaccination in pregnancy Appropriate gestational age at birth Optimising how antibodies are transferred Reducing harmful exposures Maternal vaccination Maternal vaccination bridges the newborn’s susceptibility gap. Placental IgG transfer through the FcRn receptor is active, so cord IgG can reach about 120–150% of maternal levels. Some vaccines protect mostly the mother (for example influenza), which also helps prevent iatrogenic preterm delivery. Others are aimed mainly at the baby: pertussis and RSV. New data from her cohort Mothers vary a lot in how efficiently they transfer IgG to the baby, even after correcting for GA. Each mother’s transfer efficiency stays the same in her next pregnancy. This individual variation is not currently taken into account. Turning transfer off FcRn antagonists block transfer of harmful IgG to the fetus. They are being tested in randomised double-blind trials for: HDFN (haemolytic disease of the fetus and newborn) FNAIT (fetal and neonatal alloimmune thrombocytopenia) Take-home message: Immune programming for lifelong health starts before birth. Learning how to modulate mother–baby immune cross-talk is a frontier for both obstetrics and neonatology. Disclosure: This is an AI-assisted summary of the session, generated from a recording made by our team in attendance. We believe it is as accurate as our own notes would be, but please refer to the speakers’ original work before applying anything in clinical practice.

A nice detail first: the chairs were a mentor–mentee pair from the society’s mentoring programme. The programme pairs established and early-career researchers to co-write papers and co-chair sessions, and it is always looking for new mentors. This summary covers three abstracts. 1. Cerebral sinovenous thrombosis (CSVT) in preterm infants: risk factors and imaging patternsBregje van Oldenmark (Netherlands) – NEO-O-012, European multicentre case-control study Why it matters: Preterm CSVT is probably more common than we think. In a German nationwide study, about one-third of neonatal CSVT cases were preterm. One single centre that routinely scans all infants born before 29 weeks found 4.4%. Despite this, preterm CSVT is poorly characterised. Earlier outcome studies could not separate the effect of CSVT from the effect of prematurity, because they had no matched controls. The study: A matched case-control study run with European centres expert in cranial ultrasound (CUS). The presenter described it as the largest preterm CSVT cohort to date. Findings: Who and when: Median gestational age around 30 weeks. Median day of diagnosis was after the second week of life. How it was found: More than half of cases were found on routine CUS, reflecting how many cases had no symptoms. Imaging: The transverse sinus was most often involved, and 60% of infants had more than one sinus affected. Periventricular white matter injury was a common accompanying finding. Risk factors: Perinatal and delivery-room factors did not differ between cases and controls; the differences were all postnatal. Cases more often had: invasive ventilation or inotropic support sepsis or meningitis surgery After multivariable analysis, cardiorespiratory support and neonatal inflammation remained independent risk factors. Severe illness score: 64% of cases had at least one item in a composite “severe neonatal illness” score. Each additional item raised the odds about eightfold. Treatment: About two-thirds of infants received anticoagulation (LMWH). Treatment started a median of 1 day after diagnosis and lasted about 7 weeks. No haemorrhage progression or bleeding complications were seen after treatment, including in infants with pre-existing intracranial haemorrhage. Outcome: Mortality was significantly higher in the CSVT group. Among 15 survivors followed up, 4 had neurodevelopmental impairment (3 mild) and 1 developed epilepsy. Two illustrative cases: Good outcome: An asymptomatic 29-weeker had a thrombus in the superior sagittal sinus found on routine CUS at day 14. The infant was treated with LMWH, the sinus had fully recanalised by term-equivalent age, and development was normal at follow-up. Poor outcome: A 26-weeker had sudden deterioration at 5 weeks, with apnoeas and motor agitation. Imaging showed extensive deep venous CSVT with bilateral IVH and white matter injury. Care was redirected. Detection is improving: Case numbers rose in the second half of the study period, as did the share first detected on CUS. About 80% of cases came from only three centres. This suggests we are getting better at spotting CSVT, and that earlier cases were probably missed. A prospective, low-effort worldwide registry is being built to estimate the true incidence. Q&A: Treatment practice varies between centres. Some anticoagulate infants below 3 weeks of age and some do not. Non-treatment was usually explained by very severe lesions or extreme prematurity. 2. Brain injury and short-term outcomes after single fetal demise in monochorionic twinsMathies Rondagh (Netherlands) – NEO-O-013, international multicentre retrospective study Background: After the death of one monochorionic twin, the surviving co-twin can acutely exsanguinate into the low-pressure circulation of the dead twin. This happens through placental anastomoses, mainly arterio-arterial (AA) and veno-venous (VV). The result can be acute fetal anaemia, hypoperfusion and hypoxic-ischaemic brain injury. The study: 13 fetal therapy centres, covering 2000–2025. 395 monochorionic pregnancies with single fetal demise (sIUFD) and available neuroimaging (fetal or neonatal ultrasound and MRI). Excluded: demise after laser for TTTS, since the coagulated placenta carries no exsanguination risk selective feticide TAPS, which has only tiny anastomoses major congenital anomalies Findings: Median gestational age at sIUFD: 22.7 weeks. Severe brain injury in 18% (about one in five) of surviving co-twins. Co-twin demise occurred in 10% of pregnancies. Termination occurred in 9%, mostly because of severe brain injury. When MCA-PSV was measured, fetal anaemia was found in 43%, and 12% received intrauterine transfusion. The injury pattern was consistent with acute hypoxic-ischaemic injury after exsanguination. Non-cystic white matter injury was the most common finding, followed by ischaemic infarction and cystic PVL. Timing of injury: 82% antenatal, 5% perinatal, 10% postnatal, and 2 cases undetermined. Independent risk factors (multivariable): each additional week of gestation at sIUFD fetal anaemia within 2 weeks of sIUFD The proposed explanation is that AA and VV anastomoses grow with gestation, so a later demise allows a larger, faster transfusion into the dead twin. Interval from demise to birth: An interval under 4 weeks was associated with more neonatal death, severe brain injury and neonatal morbidity. Confounding by indication is likely. The median interval was 11.6 weeks, with large variation between centres. Q&A: Injury was called “postnatal” when fetal MRI 4–6 weeks after the demise was normal and later imaging evolved accordingly. The authors propose a “triple hit” model: the underlying monochorionic complication (sFGR or TTTS) the acute exsanguination itself prematurity This makes causality hard to establish. Earlier fetal MRI looks promising for counselling and decision-making, and is a planned future study. For now, the authors support current guidance of fetal MRI 4–6 weeks after demise. 3. Validating the Kidokoro MRI score and a simplified version for predicting neurodevelopment in very preterm infantsCarmen Rodriguez Barrios (Spain) – NEO-O-014 Background: The Kidokoro score (2013) grades brain abnormality on term-equivalent MRI in very preterm infants. It covers four domains: cerebral white matter, cortical grey matter, deep grey matter and cerebellum. It is predictive, but many items make it cumbersome in routine practice. The study: Single-centre cohort of very preterm infants with term-equivalent MRI. 260 had complete scores and 180 had Bayley assessment at 2 years corrected age. Mean gestational age about 29 weeks; mean birth weight about 1180 g. 24% had an adverse outcome, defined by a threshold on any Bayley domain. Original score: Internal consistency was high (Cronbach’s α), and no single item improved α when removed. Exploratory factor analysis showed one dominant factor, supporting a single global severity construct. Discrimination per Bayley domain was moderate (AUC 0.68–0.79). Simplified score: Selected by all-possible-subsets regression. It uses gestational age plus five MRI features. Items are weighted by prognostic importance; for example, cystic lesions score 0/2/4/6/8, while corpus callosum thinning scores 0/1/2. Outcome by category: Category Adverse outcome Mild ~18% Moderate ~36% Severe 73% Each additional point raised the odds of adverse outcome by about 23%, and higher scores were linked to worse scores in all three Bayley domains. Comparison with the original score: Model AUC Simplified score 0.73 Original score 0.69 Original score + gestational age 0.72 None of these differences was significant, so simplifying the score did not cost discrimination. Validation: Internal (split-sample and 5-fold cross-validation). Inter-rater reliability was excellent. Next steps: Test–retest reliability and external multicentre validation. Disclosure: This is an AI-assisted summary of the session, generated from a recording made by our team in attendance. We believe it is as accurate as our own notes would be, but please refer to the speakers’ original work before applying anything in clinical practice.

NB: We are grateful to the speakers for openly sharing their own unit protocols and day-to-day practices. With so little evidence for babies at 22–23 weeks, describing in detail “how we do it” takes generosity and a certain vulnerability. We hope readers will treat these protocols as a starting point for discussion and shared learning, not as material for judgement. Questions, comparisons with your own practice, and different approaches are all very welcome in the thread below. This European School of Neonatology workshop covered the babies at the very edge of viability, 22–23 weeks. It had three parts: fluid balance, skin care, and how a whole country (the UK) is learning together about 22-weekers. Two units with long experience (Uppsala and Cologne) gave very practical bedside detail. 1. Fluids (and a little nutrition) at 22–23 weeksNima Naseh, Uppsala, Sweden Physiology first Total body water is about 85% of body weight before 24 weeks. The early postnatal weight loss is mostly a physiological contraction of the extracellular water. It is larger the more immature the baby is. Too much sodium early on disturbs this transition. Of all the ways these babies lose water, only two matter clinically: skin (transepidermal) and urine. Kidneys Immature kidneys lose a lot of sodium, and glucose appears in the urine. Infants born at 22–23 weeks still have raised urinary albumin at term. AKI is inversely related to gestational age. It was about three times more common at 22–24 weeks than at 25–27 weeks. Urine output is high in week 1 and falls during week 2. Skin Transepidermal water loss (TEWL) is very high at birth and is not linearly related to gestational age. The barrier matures quickly. By about one week, the differences between gestational ages are much smaller. High humidity reduces TEWL, and the effect is biggest in the most immature babies. But there is a trade-off: In animal data, dry conditions sped up barrier recovery. In an RCT of infants at 23–27 weeks randomised to 50% vs 75% humidity from day 7, TEWL fell more slowly at the higher humidity. So high humidity lowers immediate losses but may slow barrier maturation. Insensible losses differ between cohorts largely because humidity practice differs (Uppsala uses about 85%). What to expect Weight: across studies, the nadir is around day 3–4 (to day 5), with up to about 15% loss. Birth weight is usually regained at about 2 weeks. In one study, more liberal fluid intake did not change the weight trajectory. At 22 weeks, the median weight loss was about 17%. Sodium: it rises over the first days, and hypernatraemia is common around day 3. Hypernatraemia, the weight nadir and hyperglycaemia tend to cluster at about the same time. Hidden sodium With fixed-volume flushes and infusions, the smallest babies get far more sodium per kg. In a 500 g baby, these alone can give about 6–7 mmol/kg/day. In a 2024 study, saline flushes made up more than 25% of day-1 sodium intake in the smallest infants. Serum sodium can look normal despite a high sodium intake. High intake in the first week has been associated with worse outcomes. The Uppsala approach Starting fluid volume depends on humidity. It is calculated to allow about 3–4% weight loss per day, and it counts all fluids: drugs, flushes and line infusions. This keeps practice standardised and makes it comparable between centres. Sodium, worked example: Urinary losses are about 8 mmol/kg/day. Allowing for the expected extracellular contraction, they give back about 4 mmol/kg/day, which deliberately leaves a negative balance. In their data, median total sodium intake in week 1, from all sources, was just under 5 mmol/kg/day. Monitoring: Weigh every 12 h, then daily once weight, sodium and urine output have stabilised (around day 6–7). Check sodium every 6–12 h, then daily. Expected urine output is roughly 4–5 mL/kg/h. Targets: Maximum weight loss of 10–15%, with the nadir on day 3–5. Slow recovery, back to birth weight at about 2 weeks. Sodium below 150 mmol/L. In practice, about 95% of their 22-weekers go above this. Even with a cut-off of 155 mmol/L, they found no link to worse outcomes. Why “slow” matters: Weight loss by day 3 has a U-shaped association with mortality and severe brain injury: both too much and too little are bad. Faster regain of birth weight was linked to poorer developmental scores. Later on: Losses continue and sodium needs go up. The decision to start supplementing is based on weight and sodium trends together. If the baby is well and growing, they supplement “blind” rather than add extra blood tests just to check sodium. 2. Skin care at 22–23 weeks – two viewsCologne (Eva Heine, pre-recorded)At 22–23 weeks, the skin is essentially in functional organ failure: it is thin and translucent, with almost no stratum corneum. The task is to support it, protect it and connect through it. Warm and humid They start at 85–90% humidity. Very high humidity carries an infection risk; they showed a case of cutaneous aspergillosis that began in the diaper area. They use the skin (servo) mode of new-generation incubators. Humidity then adjusts automatically and sometimes drops to about 70% while temperature stays in range. She was clear that small studies have not recommended skin mode, and that this is experience-based, not evidence-based. Protect the barrier Minimal handling. No adhesives where avoidable. Sterile water or saline only; no antiseptics, which can cause burns. No ECG electrodes; pulse oximetry only. Connect – skin-to-skin is protection, not a threat Evidence that SSC is safe: A Swedish register study (~2,500 infants) found early SSC was not linked to more IVH or sepsis. EPIPAGE-2 (~3,400 infants) found no association with late-onset sepsis. Evidence that SSC helps: SSC may reduce pain. Attachment is modifiable: maternal sensitivity predicted attachment security, not prematurity. The gap between evidence and practice: In Sweden, SSC started at a median of about 88 h. Only about 5% of infants had SSC on day 1. DAISY RCT (25–30 weeks): 60 min of SSC vs 5 min of visual contact in the delivery room, after stabilisation. There were no adverse events. Parent–infant interaction at 6 months was better. Parental mental health improved in the short term; long-term effects are unclear. Delivery-room SSC at 22–23 weeks It is feasible when the whole team treats it as the goal, not the exception. Their checklist: Prepare the parents in advance. Secure lines and airway before the transfer. Preheat the room. Do a two-person transfer, with a neonatologist and the bedside nurse present. Monitor continuously; keep the room dark and quiet; give the parents privacy. Foil over the baby, not between mother and baby. Prewarm the mother’s skin. Use a hat, heated and humidified gas, warm towels and warm packs. Monitor temperature. Aim for at least 30, ideally 60, minutes undisturbed. Transfer to the ward with the baby still on the mother’s chest. Uppsala (Victoria Karlsson)She was honest that there is no evidence base, so she shared the Uppsala protocol. Their NICU has about 20 babies at 22–23 weeks each year, aims for parental presence 24/7, and has a parent bed at every care space. Key concept: by the second week of life, TEWL in extremely preterm infants approaches term values. The dry, flaky skin seen after a week or two is normal maturation, not “skin breakdown”, so there is nothing to fix. Care environment by week Week Environment 1 85% humidity in the incubator; SSC introduced at the end of the week 2 50% humidity; SSC unrestricted 3 Out of the incubator to an open bed with radiant heat and a heated mattress, then into the parents’ bed on a heated mattress Practical details Lines and tubes: Babies start in a plastic bag, which is removed after the first X-ray. A hydrocolloid goes under all tape, including for the ETT and feeding tube. Umbilical lines are sutured, with no tape on the skin. Saturation probes are non-adhesive. Vitamin K is given IV. Day 1: Cares every 6 h, with a parent present. No other interventions or examinations. All blood samples from umbilical lines; no capillary sampling. Diaper left open; no servo skin probe. No skin-breaking procedures in the first two weeks: the UAC stays until the end of week 1 and the UVC until the end of week 2. After that, sampling is venous, never by heel prick, ideally during SSC. Parents are key: they change the very first diaper. The incubator hood is raised from day 1 to allow closeness. Timing of SSC at the end of week 1 is guided by the weight and sodium trends from talk 1. Hygiene: No sterile gloves or linen; short nails, no nail polish or jewellery. Staff wear gloves for diaper changes; parents do not. Sterile gloves are used only for IV procedures. Cleaning: water only in the early weeks, pressing a wet compress rather than wiping. Disinfection: chlorhexidine in the first two weeks, chlorhexidine with alcohol later. If the skin is damaged: Cavilon (silicone-based barrier cream) for diaper-area redness or frequent stools, because the aim is dry skin even at 85% humidity. An absorbent dressing for exudative skin folds; it heals within a day or two. Pain relief. A recent paper (45 infants <26 weeks, 20 of them at 22–23 weeks) showed that early SSC maintained body temperature. Evidence is coming: large multicentre humidity RCTs are ongoing in the US and in Australia. 3. Improving care at 22 weeks through national collaboration: the UK approachCheryl Battersby, Imperial College London / BAPM A very different talk: not “what we do at the bedside” but how a country with modest 22-week outcomes is learning systematically. Context and caution Single-centre results from places like Cologne, Uppsala, Iowa and Japan are impressive. Those centres work in very different systems: centralised care, outreach teams, and (in Japan) almost all births in tertiary centres. Don’t copy them without accounting for your own system. The UK situation Many units each see very few 22-weekers; the median is about one per unit. Up to 40% of births are transferred. Survival is about 2–3 in 10. The NHS structure makes national, population-based learning possible. The UK journey 2019: the BAPM framework moved from no active care at 22 weeks to a risk-based assessment. National data (BMJ Medicine) then showed a threefold increase in active care. Survival did not improve, partly because babies with very unfavourable risk profiles were being admitted. Framework revision: Chorioamnionitis and prolonged rupture of membranes were added as very unfavourable features. Where the chance of a poor outcome is >90%, comfort-focused care is advised. The wording changed from parents “choosing” management to professionals explaining the recommended management, so parents don’t carry the burden of an impossible decision. Babies born outside a tertiary centre without antenatal steroids should not be transferred postnatally (>90% of them died). Her personal view: a country should not start offering care at 22 weeks until outcomes at 23 weeks are good. A new BAPM 22-week group, with representatives from every network plus parents, dietitians and psychologists. It runs national “roadshows” so units can share experience. Recurring themes from the roadshows Perinatal optimisation and birth in the right place (with maternity colleagues). Capacity and transfer responsibility. Counselling support for non-tertiary units. Very intense first hours. Equipment not sized for these babies. Wide practice variation and reluctance to write guidelines without evidence. A strong need for psychological support for staff as well as families: when most 22-weekers die, moral injury is real. The UK 22-week study A national observational study (England, Wales, Scotland, Northern Ireland) in nearly all tertiary centres since last June. Run without individual consent: parents receive an information sheet in five languages, with parent-group support. Data collected daily for the first 7 days, then weekly, in REDCap. Definitions are aligned with the US Tiny Baby Collaborative. Led on the ground by trainees. About 100 babies so far. She also showed very preliminary, exploratory results. These are not reported here; we’ll wait for the published data. Overall take-home: Care at 22–23 weeks is still largely experience-based. The units with the best results share an attention to detail: counting every millilitre and millimole, protecting the skin, and putting parents at the centre. The way forward is pooled data and the RCTs that are now underway. Disclosure: This is an AI-assisted summary of the session, generated from a recording made by our team in attendance. We believe it is as accurate as our own notes would be, but please refer to the speakers’ original work before applying anything in clinical practice.

Speaker: Daniel Martínez Garcia (Switzerland), humanitarian paediatrician with more than 20 years at Médecins Sans Frontières (MSF). He has worked in over 20 settings, including earthquakes and, above all, armed conflict. In her introduction, the chair, who first worked with him on her own first MSF mission in 2017, described him as “the lighthouse in Geneva” for field paediatricians. He opened by noting how rarely child health in war reaches a plenary stage. He spoke in a personal capacity and declared no conflicts of interest. The big picture: a short, fragile golden age For about 300,000 years, 40–50% of children died before adulthood, across every civilisation studied. Within the last 200 years, under 0.07% of human history, global child mortality has fallen below 4%. Vaccination, antibiotics and the industrial-era demographic transition drove this. “Childhood” as a protected category is also a recent social construction. Paediatrics as a profession exists because children started to survive. The goal now is for them to thrive, not only survive. Country curves for Nigeria, DRC, Ethiopia, Somalia, South Sudan, Rwanda, Afghanistan and Sudan show steep declines interrupted by sharp spikes back to pre-industrial mortality during wars. These gains can reverse very quickly, as seen in Syria, Libya, Venezuela and Ukraine. Children in conflict: the numbers About 1 in 5 children live in conflict zones, and around 50 million are displaced by conflict and violence. Their risk of dying before age 5 is about 3× higher. Roughly half of global under-5 mortality occurs in fragile or conflict settings. India, which is large but stable, is the main outlier. Indirect deaths outnumber direct trauma deaths about 3:1. Children die of diarrhoea without clean water, of pneumonia when there is no electricity and therefore no oxygen, and of vaccine-preventable diseases when systems collapse. Excess infant mortality persists for more than 8 years after the fighting stops. State-led sieges that restrict food and medical supplies to civilians have returned, citing Ukraine, Sudan and the Palestinian territories. They violate international humanitarian law and drive severe acute malnutrition. Chronic disease, disability and mental health are often overlooked, even by humanitarian organisations (“we come as cowboys for emergency care”). Most crises last years. Take-home for neonatology: the perinatal period is the highest-yield intervention About 47% of under-5 deaths occur in the neonatal period, more than 2 million deaths a year, concentrated in conflict settings. In conflict settings, the most cost-effective technical intervention is to protect the mother–newborn dyad: safe delivery care, essential newborn care and no mother–baby separation. Where full newborn care isn’t feasible, he proposed a minimum package: water and sanitation food access routine vaccination shelter, protection and safety education, which is usually not classed as an emergency service and is often forgotten Oxygen saves lives. In one MSF clinic, introducing paediatric intensive care and then reliable oxygen brought respiratory-infection mortality close to zero. He added a personal example: his daughter’s bronchiolitis needed only oxygen for three days, which would have been impossible without electricity. Ebola in DRC: a very large current outbreak, with more than 9,000 cases reported, is receiving little attention. In the cohort shown, 24 of 30 children died, a mortality of 60–80%. Contributing factors include biology, the absence of paediatric therapeutic research for a disease that rarely affects rich countries, and the harm of separating children from caregivers. Children are not passive victims Children have agency, diversity and voice, and they can tell us what they need. He warned that “resilience” can mask the long-term consequences of trauma, particularly in the first 1,000 days. He contrasted an anonymous, de-identified fundraising image of a war victim with a 17-year-old girl from El Fasher, Sudan. She chose to share her name, her story and a poem about her lost childhood and her hope that other children in her country will no longer live in fear. Dignity-respecting representation strengthens empathy. Calls to action for paediatricians Equal empathy for all children. He pointed out the gap in Switzerland: thousands of Ukrainian children were received, versus about 100 Palestinians granted refugee status. He called it appalling. Protect healthcare. Health workers and facilities used to be hit by mistake. In the last 20 years they have become open targets. Don’t take part in medical age assessment for legal status. There is no scientific evidence base. Clinicians are being instrumentalised to remove protection from young people, and nationality affects how they are treated. Professionalise humanitarian paediatrics. Treat it as a career path rather than “tourism”, go to learn from local health workers, and know your context. Tropical paediatrics is not a substitute. Treat refugee and migrant children in your own clinic as individuals. Don’t treat them as threats to be screened for every disease. Listen to what they need. Bear witness. Speaking out about rights violations is medical advocacy, not activism, and he urged clinicians not to hide behind neutrality. This applies within your own institution even if you never deploy. A ceasefire is the most cost-effective child-health intervention of all. Closing message: The era of children surviving and thriving is a blink in human history, with no guarantee it will last. It also shows what collective action can achieve. Whatever you do may seem insignificant, but it is important that you do it. Disclosure: This is an AI-assisted summary of the session, generated from a recording made by our team in attendance. We believe it is as accurate as our own notes would be, but please refer to the speakers’ original work before applying anything in clinical practice.

Sepsis doesn't wait for a convenient hour, so neither will we. This Friday, 30 October, 08:00–09:00, the Newborn Sepsis Society launches in Europe at CEPAS 2026 in Lyon. Yes, it's early. We promise it's worth the alarm. 📍 Room: St Clair 2 New society, same goal: fewer newborns harmed by sepsis, and more of us working on it together. What to expect: 🔹 Why a new society? Eleanor Molloy, Johan Gyllensvärd and Joanna Seliga-Siwecka (short answer: because babies deserve better, long answer: come and see) 🔹 The problem: where are we in 2026? Joseph Ting 🔹 The future: diagnostics and therapies. Eric Giannoni (evidence included, crystal ball optional) 🔹 Open discussion: where do we go from here? You, with the microphone Families have a seat at our table from day one, not as an afterthought. Mandy Daly will open the session with a message from our parent and public involvement committee. Already more than 100 members, and we've saved you a chair. 💻 Can't make it to Lyon? Join us at newbornsepsissociety.org See you at 8:00 sharp (coffee optional, curiosity required).

Everybody says conferences are the place to network. But how do you network when you are just starting out? Or when you rarely get to go to conferences? Or when you are coming back after a break, and everyone seems to know each other already? At 99nicu, we are clinicians and scientists at almost every stage of our careers. And, sad to say, we know that feeling quite well. Many of us do more than clinical work: research, teaching, building communities like this one. Yet most of us have stood alone with a coffee, wondering who to talk to. Awkward? Yes. We also know that “networking” can feel like a transaction. Waiting around the corner for a chance to talk to that one person. Asking questions in front of a poster just to start a conversation. Or holding back, because you feel you don’t have much to offer yet. That’s not what we are here for. We are here for real connections, where nobody needs a reason to say hello. Why a treasure hunt? We wanted to bring something a little unusual to a very scientific conference: a treasure hunt. Yes, just like the ones from childhood birthday parties or beloved team-building days. Why? Because we are used to fast social media. One line, one hashtag, a few likes and a comment, and we get our small dopamine boost. Then we move on to the next shiny thing. But what if we tried to engage here and now instead? At a conference like CEPAS, nobody can be everywhere or meet everyone. So we invite you to slow down, look around, and bring us what caught your attention. Take 3 pictures: 📸 A conclusion slide from a lecture or session (when allowed). Why did it catch your eye? Will it change your practice? Or the way you look at your own research? 📸 A poster (when allowed). Do you see a chance to work together? Is it close to your own topic, or something completely different? 📸 A selfie with someone you’ve just met. Who are they? What stage of their career are they in? Do you share research interests? Or are they first-timers too, just looking for a selfie partner for the Treasure Hunt? You don’t have to post your pictures anywhere (unless you want to!). Just bring them to the 99nicu booth in the Meet & Connect Zone, show them to us and tell us about your experience. The first 50 people will get a 10% discount voucher for the upcoming 99nicu conference and precious 99nicu pin (they are awesome, we are very proud of them). While you’re there, charge your phone, try our pen-and-paper puzzles, or just sit down and talk with us. We’ll also introduce the Treasure Hunt in a short session on Thursday 29 October, 10:10–10:25. Less scrolling, more talking. See you in Lyon! -- Kat from 99nicu

Hello everyone, CEPAS 2026 is getting close. The congress takes place in Lyon, 28–31 October, and 99nicu will be there - in the middle of things! Where to find us We have a home base in the Meet & Connect Zone, in the Early Career Professionals' Space. Come by, say hello, ask questions, or just sit down for a moment between sessions. If you have ever posted here, or only read quietly for years, you are equally welcome. Our two sessions 🗓️ Thursday 29 October, 10:10–10:25 — a short introduction to 99nicu, and the start of our Treasure Hunt. Small, fun, and a good excuse to talk to people you have never met. 🗓️ Saturday 31 October, 12:45–13:15 — our Anniversary Session. We begin with an interactive clinical quiz, then open the floor for discussion. Twenty years of this community, and the best part is still the conversation. One practical reminder The regular registration rate ends on 30 September. From 1 October, late rates apply. For example, the member fee goes from €550 to €685, and the early career / nursing rate from €315 to €390. So if you are planning to come, better book your spot this week. 👉 Registration: https://www.cepas.org/registration See you in Lyon 👋

The interactive scientific programme for CEPAS 2026 is now online, and if you haven't explored it yet — check out https://www.cepas.org/scientific-programme. Built on more than 15 years of collaboration between the European Academy of Paediatrics, the European Society for Paediatric Research, and GFCNI, the CEPAS programme brings together science, clinical practice, advocacy and paediatric training in one place. Sessions, topics, networking opportunities — it's all there, and it's a lot. Which brings us to a question we've been thinking about. With a programme this rich, how do you actually navigate a conference? Do you plan every session in advance or follow your instincts on the day? Do you wander beyond neonatology or stick to what you know? Paper programme or app — or both, just in case? We're genuinely curious how the 99nicu community does it. We've put together a quick poll — it takes two minutes, the results are live, and we'll be sharing them at our session at CEPAS in October. Because let's be honest - everyone has their own system. Or complete lack of one... Share your approach in the comments too — we want to know if the 99nicu community is a meticulous-colour-coded-schedule crowd or a go-where-the-wind-takes-you crowd. Probably both. Probably at the same time. Hope to see you in Lyon! - Kat from 99nicu.org

Hi @Mariana Oliveira , I'd like to just ask for a practical clarification. I've read this case, and now I'm wondering- this mother seems to be interested in a transition to direct breastfeeding. While pumping and bottle/tube feeding we can control amounts of MOM intake, however with direct breastfeeding, especially at home, parents have much less control of the amount of consumed milk. What would be your practical advice for the mother while preparing them for a home discharge- aim to breastfeed before medication dose (as much as infant wants, without controlling the amount?), and offer formula if infant demands feeds at peak plasma concentration moments? Would you recommend that mother pumps-and-dumps in between the feeds to sustain the lactation (hoping when infant grows, the medication levels will become less problematic and she will be able to breastfeed until a desired timepoint, eg. 1 or 2 yrs of age?). I'm just thinking of this case in terms of long term planning and supporting lactation beyond NICU and immediate post-discharge.

Such an important topic! Having a dear colleague, a brilliant scientist with a PhD, who was born very preterm in the 90's as one of the triplets, where one of his siblings died in the NICU, and another one is living with a severe disability, I'm always reminded: you need to survive in good overall health to be able to tell your story. And I'm glad that, although it feels a bit like a pushback, your ethics committee is actively trying to be the voice of the ones who don't get to tell their stories.

Conference in Brazil? Packing ASAP!