Pediatrics

Pediatric ERP rollout alone did not shorten LOS; patient-level adherence did.

In this stepped-wedge trial of 597 adolescents undergoing elective GI surgery, overall ERP implementation did not reduce postoperative length of stay across phases. Patients receiving at least 13 ERP elements had a shorter median LOS by 1.14 days and fewer complications (adjusted OR 0.48). ERP adherence increased with implementation supports and correlated with integration into order sets and site culture. Implication: focus on achieving high patient-level fidelity to ERP elements rather than assuming site implementation alone will change outcomes. 1

Intrapartum 2 g azithromycin did not change 2-year neurodevelopment after birth asphyxia.

This multicountry follow-up of A-PLUS enrolled 403 children with birth-asphyxia exposure and found no difference in BSID-III cognitive scores at 24±1 months (mean difference 0.29; 95% CI -1.77 to 2.34). No significant differences were seen in language, motor scores, or ASQ-3 domain totals between azithromycin and placebo groups. Regional subgroup analyses similarly showed no neurodevelopmental benefit from intrapartum azithromycin. Implication: do not use intrapartum single-dose azithromycin to prevent neurodevelopmental impairment after birth asphyxia. 2

Pediatric emergency LLM ensembles show sociodemographic-driven recommendation differences.

An ensemble of 10 LLMs produced over 3.7 million outputs across 1,000 cases and recommended more urgent interventions for adversity-labeled and Black unhoused children. Examples included higher probabilities for urgent interventions (+10.5 pp), additional investigations (+14.1 pp), and suspected maltreatment (+26.6 pp) without clinical justification. Models were sensitive to caregiver and child sociodemographic labels, amplifying intersectional effects compared with physician-derived ground truth. Implication: deploy LLM decision support only with explicit safeguards and monitoring for sociodemographic bias. 3