Objective: Provide a progress report updating our long-term quality improvement collaboration focused on major morbidity reduction in extremely premature infants 23-27 weeks.

Methods: 10 Vermont Oxford Network (VON) neonatal intensive care units (NICUs) (the POD) sustained a structured alliance: (A) face-to-face meetings, site visits and teleconferences, (B) transparent process and outcomes sharing, (C) utilisation of evidence-based potentially better practice toolkits, (D) family integration and (E) benchmarking via a composite mortality-morbidity score (Benefit Metric). Morbidity-specific toolkits were employed variably by each NICU according to local priorities. The eight major VON morbidities and the risk-adjusted Benefit Metric were compared in two epochs 2010-2013 versus 2014-2018.

Results: 5888 infants, mean (SD) gestational age 25.8 (1.4) weeks, were tracked. The POD Benefit Metric significantly improved (p=0.03) and remained superior to the aggregate VON both epochs (p<0.001). Four POD morbidities significantly improved through 2018 – chronic lung disease (48%-40%), discharge weight <10th percentile (32%-22%), any late infection (19%-17%) and periventricular leukomalacia (4%-2%). In epoch 2, 34% of survivors had none of the eight major morbidities, while 36% had just one. Mortality did not change.

Conclusions: Inter-NICU collaboration, process and outcomes sharing and potentially better practice toolkits sustain improvement in 23-27 week morbidity rates, notably chronic lung disease, extrauterine growth restriction and the lowest zero-or-one major morbidity rate reported by a quality improvement collaboration. Unrevealed biological and cultural variables affect morbidity rates, countless remain unmeasured, thus duplication to other quality improvement groups is challenging. Understanding intensive care as innumerable interactions and constant flux that defy convenient linear constructs is fundamental.