Hot Takes

In 493 Danish newborns with probable culture-negative early-onset sepsis, a recovery- and CRP-guided strategy reduced median antibiotic exposure from 6.8 to 2.8 days. Infection-related readmission was 2/246 versus 1/247, so the prespecified 4-percentage-point noninferiority margin was met—but the safety estimate rests on only three events in a selected, closely followed population.

A single-center neonatal chest-radiograph model reached an AUC of 0.975 in an internal held-out set, but sensitivity fell to 87.6% at its fixed threshold and 52 of 165 positive calls were false. With no external validation, reader comparison, workflow test, or patient outcome, this is a strong development signal—not a deployment result.

In 20 preterm infants, an external depth camera and the impedance monitor were both quality-valid during 40.4% of non-intervention time; either was valid during 75.3%. Agreement inside the paired intervals was close (MAE 1.44 breaths/min), but tuned quality thresholds, an imperfect low-rate comparator, one-site design, and no apnea or alarm testing keep this in feasibility territory.

In 156 selected infants born before 28 weeks, major brain injury was associated with a 1.45-point lower Burdjalov score at 32 weeks, and the retrospective within-cohort AUC was 0.886. But 118 of 274 screened infants were excluded, the reference group was mostly grade II IVH, and no long-term outcome was measured. This is group separation, not a bedside prognosis.

In 316 preterm infants, 72-hour NIPPV failure occurred in 27/158 with a long, narrow cannula versus 23/158 with short binasal prongs. The 95% CI crossed the 10-point noninferiority margin by 0.56 points, while any nasal trauma fell from 44.9% to 12.0%. Gentler is credible; equivalent efficacy is not established.

A 10-year referral-center review found 62 preoperative deaths and 43 deaths among 52 operated infants within 14 days. The median birth-to-diagnosis interval was 5 days, but 16 records lacked survival outcomes and nonrandom timing cannot prove that earlier surgery alone changes survival.

In one 390-infant retrospective cohort, hemangioma was documented before discharge in 17/92 darbepoetin-exposed infants versus 18/262 unexposed. The crude OR of 3.07 fell to 1.83 (95% CI 0.63-5.30) after adjustment; nonrandom treatment, maturity imbalance, and discharge-only ascertainment prevent a causal risk estimate.

In 441 cooled term and near-term infants, phosphate around 6 hours was higher among non-survivors and showed an AUC of 0.90 for in-hospital mortality. The cutoff remains hypothesis-generating: it came from one retrospective cohort, phosphate was available in 394 infants, and the endpoint included redirection-of-care deaths.

In one Italian NICU, 121 pressure-injury events were recorded in 83 neonates across 49,134 patient-days. Event density fell abruptly in January 2020 (IRR 0.378, 95% CI 0.148-0.963), while 113/121 events were device-related. The study cannot separate safer care from changes in case mix, device exposure, or reporting.

In 102 preterm infants, 10.26-µL phenylephrine/tropicamide microdrops were noninferior to 26.58-µL standard drops for 45-minute pupil diameter. All examinations were completed, but this single-center trial was too small to establish rare-harm safety.

In a 65-fetus retrospective cohort, 44 survived to age 1. Among 20 isolated survivors assessed at age 3, 11 had Language-Social DQ below 85. The uncontrolled protocol and selected follow-up sample make this a surveillance signal, not proof that treatment or timing caused the outcomes.

In 180 preterm infants, cortisol fell 0.20 µg/dL in a mother-newborn unit and rose 0.43 in conventional NICU care. Bed availability set the groups, while KMC exposure, feeding and heel-stick position also changed. This is a bundle signal, not an isolated zero-separation effect.

Across 35 Botswana facilities, targeted birth testing reached 2,627 of 8,844 HIV-exposed newborns, confirmed 14 infections, and started treatment at a median 2.7 days. Yet 56 initial positives did not confirm, 15.4% of birth-negative infants lacked documented 6-week testing, and low-risk newborns were not tested at birth. The cohort supports a confirmatory diagnostic pathway—not a true programme-sensitivity estimate or a simpler-prophylaxis rule.

In a prospective single-center cohort of 168 term neonates meeting sepsis criteria, 23 died within 28 days. FIBTEM clot firmness was lower in non-survivors, but adding it to nSOFA plus INR raised model AUC only from 0.862 to 0.878; the 0.016 gain was not statistically significant (DeLong P=.455). ROTEM described a hypocoagulable severity phenotype, not a validated stand-alone mortality tool.

In one retrospective cohort of 286 outborn infants born before 32 weeks, IVH was present in 19.9% on day 1. Mode, duration and measured transport intervals were not associated with IVH; perinatal inflammation and markers of early instability were. The operational focus may need to start before departure—but the data do not prove the ride is harmless.

In 28 NICU infants, a perfused skin patch detected glucose in 224 of 243 skin-perfusate samples. The headline blood-glucose fit came from only 17 paired samples in 15 infants (weighted R² 0.92); lactate fit was R² 0.01. Promising biosensing, but no agreement analysis, threshold accuracy, or clinical outcome makes this a feasibility signal—not a blood-test replacement.

In 102 extremely preterm or extremely low-birth-weight infants, complete tibial ultrasound measurements had inter-rater ICCs of 0.84 at 1 month and 0.94 at 36 weeks' PMA. Yet 12 of 98 infants available at 1 month had no reading, often because of instability or device limits. Repeatability is promising; diagnostic accuracy remains untested.

BLOOM randomized 105 mothers after birth before 34 weeks. Brewer's yeast was an adjusted 94 mL/day above placebo at day 7, but the 95% CI spanned -51 to +239; beta-glucan was -25 mL/day (95% CI -173 to +123). The trial is informative, but neither supplement has a proven milk-volume benefit.

FAERS contained 130 neonatal reports and no exposed denominator. Its extreme infusion-site signal is a prompt for validation, not an estimate of how often harm occurs or proof that acyclovir caused it.

In 104,451 newborns, 782 had initial free carnitine below 10 µmol/L and 735 normalized on repeat. A four-factor preanalytic model reached AUC 0.75, but the study did not report molecular confirmation or follow screen-negative infants—so it is a quality signal, not a PCD diagnostic shortcut.

In a 350-bag single-center validation, gravimetric checking found six failures missed by residual-volume review. Yet total weight cannot confirm each ingredient, no known-error challenge or patient outcome was tested, and contradictory timing and detection claims block an efficiency verdict.

A single-center case series reported discharge survival in 11 of 35 resuscitated infants labeled 21 weeks. Resuscitation followed parent request and a favorable postnatal phenotype, the number of all liveborn infants was not reported, and dating plus follow-up inconsistencies prevent this from defining a new viability threshold.

In 142 matched ventilated preterm infants, BPD occurred in 13/50 fentanyl-exposed versus 5/92 unexposed infants; adjusted OR 5.49 (95% CI 1.72 to 17.54). But surfactant-treated RDS was 88.0% versus 43.5%, so confounding and reverse causation make the proposed 3-day threshold a hypothesis, not a stop rule.

Dubai's citywide program diagnosed 32 of 62 NICU referrals and returned results in a median 3.4 days. The implementation signal is strong, but a selected NICU/PICU cohort and historical comparator do not prove better outcomes from routine sequencing in every unit.

In 504 analyzed newborns with suspected or proven late-onset sepsis, procalcitonin guidance reduced median antibiotic exposure from 10 to 8 days. The trial randomized 511 of 2,162 planned participants, so uncommon harms remain uncertain despite meeting its 3-point mortality noninferiority margin.

Across 16 hospitals and 139,300 neonates, ALERT was linked to lower adjusted odds of fresh stillbirth or death within 24 hours (aOR 0.78, 95% CI 0.65–0.94). Unadjusted comparison and intervention period totals were 3.1% in both, and a full-uptake sensitivity analysis attenuated the estimate to 0.96.

In 300 Ugandan pregnancies, GBS6 antibody responses and infant transfer looked similar with well-controlled HIV. Protection and uncommon harms remain untested.

Death or neurodevelopmental disability at 18 to 24 months occurred in 82/113 infants assigned to occlusive wrap and 80/127 assigned to control (OR 1.55, 95% CI 0.90 to 2.69). Because 116/356 infants had no known primary outcome, disproportionately from the wrap group, this report cannot establish equivalent long-term outcomes or determine whether there is harm.

A Saudi external validation found the NICHD BPD estimator's composite-outcome AUC was 0.762 on day 14 and 0.642 on day 28, with changing case sets across time. Moderate discrimination without reported calibration or impact analysis is not a bedside treatment threshold.

In a 128-infant prospective cohort, adjusted Bayley-4 outcomes at 18–24 months did not differ after transient hypoglycemia was identified and managed using AAP thresholds. Selection, attrition, small severe subgroups, and a short follow-up horizon keep this from proving long-term safety.

In 130 survivors randomized at postnatal days 6 to 14, neurodevelopmental impairment was 10/66 (15%) without ibuprofen and 14/64 (22%) with it (P = .371). With 22% power for that comparison, the follow-up supports uncertainty, not equivalence.

In 55 preterm infants, 87%-88% of analyzed day-to-day cardiac-output changes met a trend benchmark after small shifts were excluded.

PETITE-DTG met pharmacokinetic targets in 41 term neonates. Sixteen mothers found the oral film usable, but routine-care and preterm evidence are still missing.

In 151 VLBW infants, 6 versus 2 mg/kg/day produced higher 4-week ferritin while prespecified microbiome measures and short-term clinical outcomes did not differ. A two-week, single-center biologic-endpoint trial is not a routine dosing rule.

A 107,505-infant US cohort found Bifidobacterium or Lactobacillus in blood cultures of 8/19,990 probiotic-exposed infants versus 3/87,515 unexposed infants—an absolute excess of 0.04 percentage points. Without strain typing or a causal effectiveness analysis, the study measures a rare risk but cannot settle the NEC trade-off.

In 242 randomized preterm infants, the MIND bundle changed speech-sound EEG processing near discharge. At 2 years, raw language means were nearly identical even though adjusted models favored MIND—making the neural signal clearer than the clinical claim.

In 87 extremely preterm infants receiving rescue vasopressin, median mean blood pressure rose from 29 to 39 mmHg by 24 hours. With no comparator, shrinking follow-up and 60 deaths before discharge, the study describes physiologic response—not treatment benefit.

After matching, late-onset invasive fungal infection occurred in 2 of 442 fluconazole recipients and 1 of 442 controls. Low baseline risk matters, but three events cannot prove no benefit.

In 121 initially stable newborns, a score of 9 had 89.3% sensitivity and 88.2% specificity for oxygen use within 48 hours. The same cohort set and tested the cutoff.

Staff saw heart rate within 60 seconds in 35/51 cases, but 11/62 recordings were unavailable. An FDA Class 3 cap-label recall adds a separate check.

In 74 low-risk term neonates, a 10-hour schedule did not show a statistically significant difference in average 24-hour bilirubin decline, but bilirubin at 24 hours was 24 micromol/L higher and the trial was not designed for equivalence or outpatient safety.

In 138 analyzed preterm infants, ventilation by 72 hours occurred in 14.5% with NDT-InSurE and 18.8% with LISA. The confidence interval clears a 10-point boundary, but the paper calls 20 points prespecified while the protocol describes 10—and 13 randomized infants were not analyzed.

Across 1.56 million live births, VACTERL was recorded in 2.3 per 10,000 IVF births and 1.2 per 10,000 naturally conceived births. Adjusted odds were higher, but the absolute excess was about 1.3 per 10,000—and this observational study cannot establish causation.

Among 513 children who had infant cardiac surgery and returned for Bayley-III testing, earlier gestation tracked lower adjusted cognitive and motor scores. Only 38% of 1,350 eligible children returned, so the signal is clinically important but not an individual forecast.

Among 74,423 very preterm infants, a crude 9.1-point gap in death or severe IVH narrowed to 0.3–1.8 points after intended-mode weighting. The breech signal remains observational.

In 32 selected pregnancies with early anhydramnios from fetal kidney failure, 19 of 29 live-born infants reached 14 days with dialysis access. Fourteen survived to discharge after a median 4.7 months, and every live birth was preterm.

In 1,283 transfused VLBW infants, median pretransfusion hemoglobin and red-cell storage solution were associated with a composite of severe morbidity or death. This prospective cohort supplies testable signals, not a threshold or storage-solution rule.

In 997 high-risk infants, first-season adverse events were numerically similar with clesrovimab and palivizumab. The 276-child second-season cohort had no concurrent comparator, so comparative efficacy was not tested.

In a 138-transport Vietnam pilot, 51.4% met the broad unsafe definition and 31.9% met a stricter one. The gap was mild hypothermia, and the study cannot separate transport effects from surgery and illness.

A 147-neonate Malawi trial found lower adjusted skin bacterial load with 1% chlorhexidine than standard care. Clinical benefit remains untested.

In 45,754 catheter episodes, phlebitis was 13.4% with splint alone and 3.8–4.2% with glue. The retrospective analysis did not cluster repeat lines.

A 282-person cohort linked sodium channel blockers by 1 month to earlier seizure offset; LOF KCNQ2-DEE impairment odds were lower after adjustment.

QFT conversion was 5.3% with VPM1002 and 4.3% with BCG. Only 334 of 632 planned events accrued, leaving infant TB disease protection unresolved.

At 15 minutes, SpO₂/FiO₂ ratios were 271 versus 269 and PPV was 10/28 in both arms; pneumothorax occurred in 0/28 versus 3/28. PHILODENDROOM establishes that 5-versus-8 cmH₂O randomization is feasible—not which starting pressure is better.

In 313 infants with moderate or severe HIE, death or moderate/severe developmental deficit occurred in 34.1% with erythropoietin and 28.7% with placebo (RR 1.19, 95% CI 0.84–1.68). Together with HEAL, PAEAN weakens the efficacy case while leaving rare-harm uncertainty.

Across five Japanese NICUs, LED guidance placed every tube in the stomach, but only 81 of 101 were centrally positioned before X-ray adjustment. The device is promising; this study does not yet validate replacing radiography or pH checks.

Across 125 sessions in 46 infants, skin-to-skin care met the study’s thermal non-inferiority criterion. A wide margin, strict readiness checks and two tube dislodgements keep this a protocol-feasibility signal.

In a 45,624-infant Chinese cohort, caffeine was not associated with overall Bell stage IIA-or-higher NEC after adjustment, while first doses within 72 hours tracked with less surgical NEC. Baseline imbalance, missing dose data, and a reporting mismatch make that timing signal observational, not a protocol.

In a single-center randomized study, the restricted-fluid arm averaged 36 versus 57 hours of NIPPV, but discharge was not clearly earlier and six post-randomization exclusions plus retrospective registration keep this exploratory.

The 2024 MOH pathway means one routine specimen for most newborns, two for infants under 37 weeks or under 1500 g without day-one transfusion, and four when day-one red-cell transfusion changes the schedule.

In a South Korean claims cohort of 2,303,567 first-year survivors, 7,656 had recorded neonatal TPN. The primary weighted analysis found no association with later metabolic disease, while an exploratory analysis found T2DM HR 2.14 (95% CI 1.34-3.41) after at least 3 recorded days. Residual illness imbalance and missing dose data make that duration a risk marker, not a causal threshold.

In SILDI-SAFE's 122-infant safety population, protocol-defined hypotension occurred in 1/92 sildenafil recipients and 0/30 placebo recipients. Exposure was overwhelmingly enteral, efficacy analyses were exploratory, and three registry-reported deaths in sildenafil cohorts require explanation rather than causal spin.

A single-centre randomized trial found fewer deaths, severe IVH, or BPD with intact-cord resuscitation, but its public registry snapshot does not match the paper's primary outcome or target sample.

In a single-center randomized trial of 95 lumbar punctures, static ultrasound left first-attempt nontraumatic success at 50% in both groups and added eight minutes on average. The operator signal is a training hypothesis, not a deployment verdict.

A 136-infant pilot RCT found 3.20 g/kg/day higher adjusted weight gain to discharge with a plant-derived amino acid fortifier. The products also differed in energy, protein, fat, and micronutrients; the gap vanished by 40 weeks PMA, and safety estimates were too imprecise for equivalence.

Among 1,289 infants with classifiable primary-composite data, 72 hours of cerebral oximetry-guided care produced nearly the same 2-year death-or-disability rate as usual care. Incomplete follow-up and sparse Bayley testing limit certainty about smaller effects.

In 100 Swedish infants with severe prethreshold ROP, type 1 ROP occurred in 20% with dexamethasone eye drops and 38% with saline, but the adjusted primary analysis crossed the null (OR 0.44, 95% CI 0.17–1.12; P=.08).

A five-year nationwide surveillance cohort covering 2.18 million NICU patient-days found lower BSI and CLABSI rates in Korea, yet persistent between-hospital variation and missing patient-level acuity data limit benchmarking and causal claims.

In the 23-hospital OPTimize NOW cluster-crossover trial, symptom-based opioid dosing shortened adjusted time to medical readiness by 2.3 days among 383 infants managed with Eat, Sleep, Console, while actual length of stay and treatment initiation were not significantly different.

In a prospective single-center study of 150 suspected-sepsis events, presepsin separated culture-confirmed sepsis poorly at the initial assessment (AUC 0.62, 95% CI 0.48–0.76). Its cohort-derived cutoff traded 73% sensitivity for only 42% specificity; CRP discriminated better here, but neither marker can carry the sepsis decision alone.

A blinded crossover trial in 80 caffeine-treated preterm infants found no reduction in desaturation or bradycardia events when vanilla or strawberry scent was applied inside nCPAP masks. Unmeasured odor delivery keeps the null result specific to this method.

A 104-infant pilot shows that selective echocardiography-guided early PDA treatment can be studied with high protocol adherence—and signals less pulmonary haemorrhage—but its clinical estimates are too fragile to support protocol adoption.

A new randomized study sharply reduced reported glucose excursions in preterm infants—but found no significant neurodevelopmental benefit at two years.

The largest trial yet found automated FiO₂ control safe and effective at saturation targeting—but no significant reduction in death, NEC, BPD, or severe ROP.

A new cohort reports lower adjusted BPD odds with SMOF—but the raw result, historical design, safety signals, and inconsistent effect estimates demand restraint.

A preview of focused online sessions built around neonatal education, shared practice, and moderated discussion.

Nothing dramatic changed. The team simply paused long enough to name the priorities and decide who was carrying each next step.

A parent asked what would happen next. The answer revealed that the room was holding three different versions of the afternoon.

The transfer was difficult. The moment that mattered most was the pause when every voice became part of one plan.