Highlight
Four key takeaways from the iNeo 2025 cohort analysis:
– Survival to neonatal discharge for infants born at 22–23 weeks varied markedly across 11 neonatal networks (adjusted ranges 9%–64% at 22 weeks and 16%–80% at 23 weeks).
– Major morbidities (severe intraventricular hemorrhage [grade 3–4 PVH], cystic periventricular leukomalacia, treated retinopathy of prematurity, bronchopulmonary dysplasia, necrotizing enterocolitis) also showed substantial between‑network differences.
– Much of the variation likely reflects differences in perinatal and neonatal practice, systems (including inborn vs outborn care), and local attitudes toward provision of intensive care at the edge of viability.
– Identifying and studying these differences creates opportunity for targeted research, benchmarking, and policy work to improve outcomes and to inform parental counseling.
Background and clinical context
Infants born at 22–23 weeks’ gestation occupy the “periviable” boundary between expected nonviability and potential survival with intensive neonatal care. Advances in neonatal resuscitation and supportive care have increased efforts to provide postnatal intensive care for these infants in many regions, yet survival and long‑term neurodevelopmental outcomes remain variable and often poor. Accurate, contemporary outcome data across jurisdictions are essential for clinicians advising families, for health systems planning and benchmarking, and for designing trials and quality improvement initiatives.
The International Network for Evaluation of Outcomes (iNeo) investigated outcomes for infants at 22–23 weeks across multiple national and regional neonatal networks to characterize between‑network differences and identify targets for further study.
Study design
This was an international registry‑based cohort study including infants born at 22 to 23 weeks’ gestation between January 1, 2015, and December 31, 2021, who were admitted for neonatal intensive care and had no major congenital anomalies. Data came from 11 neonatal consortia representing 12 countries or regions participating in iNeo. Analysis was restricted to infants admitted to NICU (thereby excluding stillbirths and infants not offered or not receiving active NICU care), and outcomes were adjusted for relevant clinical case mix.
Primary outcomes included survival to NICU discharge, major neonatal morbidities (severe periventricular hemorrhage [PVH] or periventricular leukomalacia [PVL], treated retinopathy of prematurity [ROP], bronchopulmonary dysplasia [BPD], and necrotizing enterocolitis [NEC]), and survival without major morbidities. The authors compared adjusted probabilities of outcomes across networks and reported standardized incidence ratios compared with a reference population comprising all other networks.
Key findings
Population and baseline data
– Total cohort: 5,019 neonates admitted to NICU; 846 born at 22 weeks and 4,173 at 23 weeks.
– Demographics: 53% male (2,641/5,017), and 23% of mothers were >35 years among those with recorded maternal age (1,084 of 4,636).
Perinatal practice variation
Marked variation existed across networks in several perinatal practices for infants at 22 and 23 weeks, respectively:
– Any antenatal corticosteroids: ranges 18%–83% at 22 weeks, 57%–98% at 23 weeks.
– Cesarean birth: 0%–42% at 22 weeks, 5%–73% at 23 weeks.
– Outborn status (born outside the tertiary center): 0%–26% at 22 weeks, 0%–22% at 23 weeks.
These differences reflect differing policies, referral systems, and thresholds for active perinatal interventions.
Survival to discharge
After adjustment, survival to NICU discharge varied widely across networks:
– 22 weeks: adjusted survival ranged with 95% confidence intervals spanning 9%–64% (P < .001 for between‑network differences).
– 23 weeks: adjusted survival ranged across networks from 16%–80% (P < .001).
Major neonatal morbidities
Significant between‑network variation was also observed in severe brain injury, treated ROP, BPD, and NEC:
– Severe PVH (grade 3–4) or PVL: adjusted network ranges 24%–65% at 22 weeks and 18%–56% at 23 weeks (P < .001).
– Survival without severe PVH or PVL: 7%–53% at 22 weeks and 9%–69% at 23 weeks (P < .001).
– Treated ROP among survivors: 32%–57% at 22 weeks (P = .008) and 16%–48% at 23 weeks (P < .001).
– BPD among survivors (reported for 23 weeks): 64%–88% (P < .001).
– NEC among 23‑week infants: 6%–28% (P < .001).
Standardized incidence ratios comparing each network against the reference population revealed that some networks had significantly higher or lower-than-expected rates of survival and specific morbidities.
Interpretation of the magnitude and statistical significance
These are clinically and statistically meaningful differences. For example, a network with a 16% adjusted survival at 23 weeks versus one with 80% reflects profound differences in either patient selection, decision-making thresholds for initiating/providing intensive care, or differences in systems of perinatal and neonatal care. Similarly, a two- to threefold spread in severe brain injury and treated ROP among survivors suggests that both antenatal and neonatal practices and resource availability materially influence morbidity.
Expert commentary and implications
What drives international variation?
Several non‑mutually exclusive explanations likely account for the observed variation:
– Selection and denominators: This analysis included only infants admitted to NICU; regional differences in whether infants at 22–23 weeks are offered resuscitation or NICU admission will affect both numbers and outcomes.
– Antenatal practice differences: Use of antenatal corticosteroids, tocolysis, magnesium sulfate for neuroprotection, and decisions to perform cesarean deliveries at extreme prematurity differ by region and influence survival and morbidity.
– Systems issues: Rates of inborn vs outborn delivery, regionalization of perinatal care, transport capabilities, and staffing levels impact early stabilization and outcomes.
– Neonatal care practices: Variability in respiratory management (e.g., early CPAP vs invasive ventilation), surfactant strategies, parenteral nutrition, infection prevention, ROP screening and treatment thresholds, and BPD prevention bundles.
– Cultural and ethical attitudes: Local attitudes toward life‑sustaining interventions at the edge of viability influence both the initiation and continuation of intensive care.
Clinical and counseling consequences
These findings matter for clinicians who counsel expectant parents. Counseling should incorporate local real‑world outcomes while acknowledging that survival and morbidity vary even among similar populations. Transparent presentation of local network outcomes and the factors that drive them (or limit comparability) will support shared decision-making.
Opportunities for quality improvement and research
– Benchmarking: Network‑level data enable benchmarking and identification of outlier centers or systems for targeted improvement.
– Comparative effectiveness research: The observed natural experiment—differences in practice across networks—can guide observational comparative studies to identify practices associated with better outcomes.
– Standardized data and follow-up: Harmonizing definitions and collecting long‑term neurodevelopmental outcomes are essential to understand the meaningfulness of short‑term survival differences.
– Ethical and policy research: Work is needed to align resource allocation, parental preferences, and societal values about care at the limit of viability.
Study limitations
Key limitations that affect inference and generalizability include:
– Selection bias: The cohort excludes stillbirths and infants not admitted to NICU; networks with more conservative approaches may have lower admission rates and therefore different denominators.
– Residual confounding: Although adjusted analyses were performed, unmeasured differences (socioeconomic factors, antenatal risk, center case mix) may explain some variation.
– Lack of long‑term outcomes: The study reports survival to discharge and in‑hospital morbidities, but neurodevelopmental outcomes at age 2 or later—critical for counseling—are not included.
– Data harmonization: Despite iNeo’s efforts, differences in data recording, coding, and clinical practice definitions across countries may introduce measurement variability.
Conclusions and clinical takeaways
This large international cohort study demonstrates substantial variation in survival and major neonatal morbidities among infants born at 22–23 weeks who were admitted to NICU across 11 networks. The findings underscore that outcomes at the limits of viability are not fixed biologic constants but are sensitive to clinical practice, systems of care, and local policies. Moving forward, stakeholders should prioritize harmonized outcome measurement (including long‑term neurodevelopment), comparative effectiveness studies of modifiable practices, transparent benchmarking for quality improvement, and careful, individualized parental counseling grounded in local data.
Funding and clinicaltrials.gov
Funding and trial registration details are provided in the original publication: Isayama T, Norman M, Kusuda S, et al.; International Network for Evaluation of Outcomes (iNeo) Investigators. Outcomes of Preterm Infants Born at 22 to 23 Weeks’ Gestation. JAMA Pediatr. 2025;179(11):1183–1193. For specific funding sources and acknowledgments, consult the article and its erratum.
References
1. Isayama T, Norman M, Kusuda S, Reichman B, Lehtonen L, Lui K, Adams M, Vento Torres M, Filippi L, Battin M, Guinsburg R, Modi N, Håkansson S, Klinger G, de Almeida MF, Helenius K, Bassler D, Su YC, Shah PS; International Network for Evaluation of Outcomes (iNeo) Investigators. Outcomes of Preterm Infants Born at 22 to 23 Weeks’ Gestation in 11 International Neonatal Networks. JAMA Pediatr. 2025 Nov 1;179(11):1183-1193. doi: 10.1001/jamapediatrics.2025.2958. Erratum in: JAMA Pediatr. 2025 Nov 1;179(11):1247. PMID: 40853670; PMCID: PMC12379133.
(Readers are encouraged to consult additional guideline documents and region-specific outcome reports for local practice context.)
Research and practice priorities
– Standardize denominators and reporting (include births, live births, NICU admissions) to improve comparability.
– Collect and report long‑term neurodevelopmental outcomes for infants born at 22–23 weeks.
– Conduct comparative effectiveness and implementation studies focused on antenatal steroids, delivery location optimization (inborn care), respiratory management bundles, and ROP/BPD prevention strategies.
– Foster international collaborations for data sharing, consensus definitions, and ethical frameworks to guide care at the limits of viability.
