Highlights
– In a national Japanese inpatient database spanning July 2010–March 2022, 129,375 children (≤14 years) received invasive mechanical ventilation; 63% were cared for in general wards rather than ICUs.
– After 1:1 propensity score matching (15,760 pairs), in-hospital mortality was higher for children treated on general wards compared with ICUs (6.4% vs. 4.1%; OR 1.49; 95% CI 1.35–1.65; p < 0.001).
– Nonoperative patients disproportionately received ward-based ventilation (78%), whereas the ICU cohort had a high proportion of postoperative patients (56%), 70% of whom had cardiac surgery—underscoring case-mix differences and system-level triage patterns.
Background
In high-income countries, invasive mechanical ventilation for children is commonly delivered in pediatric intensive care units (PICUs) staffed with pediatric intensivists, specialized nursing, respiratory therapists, and multidisciplinary support. Concentrated resources and expertise in ICUs are thought to improve monitoring, early recognition of deterioration, ventilator management, sedation and weaning strategies, and complication prevention.
Japan differs from many Western systems in that a substantial number of mechanically ventilated children are managed on general wards. This study by Sakurai et al. utilized a national inpatient administrative database to characterize who receives ventilation on wards versus ICUs in Japan and to compare in-hospital outcomes between the two settings. The work addresses important health-systems questions: are outcomes equivalent when children are ventilated on wards, and should care be centralized to ICUs?
Study design and methods
This was a retrospective cohort analysis of a national inpatient database in Japan covering July 2010 through March 2022. Inclusion criteria were children aged 14 years or younger who received invasive mechanical ventilation during hospitalization. The primary exposure was location of care while mechanically ventilated—general ward versus intensive care unit. Key patient characteristics (demographics, operative status, diagnoses) were compared between groups.
To address confounding by indication and case mix, the investigators performed 1:1 propensity score matching. Matched pairs were used to compare the primary outcome of in-hospital mortality between ward- and ICU-treated patients. The study reported absolute and relative effect sizes, with odds ratios and 95% confidence intervals.
Key results
From July 2010 to March 2022, 129,375 mechanically ventilated children were identified. Major descriptive findings:
- Overall, 63% of ventilated children were treated in general wards.
- Nonoperative patients: 78% were managed on general wards.
- Postoperative patients: only 17% were treated on general wards, with 56% of ICU patients being postoperative; of these postoperative ICU patients, 70% underwent cardiac surgery.
The propensity score matching process produced 15,760 matched pairs of ward-treated and ICU-treated patients. In this matched cohort, in-hospital mortality was significantly higher for patients treated on general wards compared with ICUs:
In-hospital mortality — general ward: 6.4% vs ICU: 4.1% (odds ratio 1.49; 95% CI 1.35–1.65; p < 0.001).
These results indicate an approximate 50% relative increase in odds of death associated with ward-based care for mechanically ventilated children after accounting for measured confounders.
Interpretation and clinical relevance
The study presents compelling population-level evidence that mechanically ventilated children in Japan are frequently managed outside ICUs and that ward-based management is associated with higher in-hospital mortality after propensity matching. Several interpretations merit consideration:
1. Association versus causation
While propensity matching helps balance measured covariates, residual confounding may persist. Administrative databases often lack granular physiologic severity scores (e.g., Pediatric Index of Mortality, PRISM), detailed ventilator settings, timing of intubation, or complications occurring before transfer. Unmeasured severity differences or treatment preferences could influence both location of care and outcome.
2. Case-mix and triage patterns
The ICU cohort included a high proportion of postoperative patients—especially children after cardiac surgery—who are typically triaged to ICU care by protocol. Conversely, nonoperative patients were disproportionately managed on wards. This suggests structural triage rules or resource constraints underpinning location decisions and highlights the complexity of comparing outcomes across settings.
3. Structural and process differences
ICUs are designed for continuous monitoring, higher nurse-to-patient ratios, specialized ventilator expertise, and rapid response to deterioration. Ward environments may lack these attributes, contributing plausibly to worse outcomes. The observed mortality difference is biologically and operationally plausible, supporting the hypothesis that critical care structures matter for ventilated children.
4. External validity
The findings are based on comprehensive national data from Japan between 2010 and 2022 and likely reflect Japanese hospital organization, ICU bed distribution, and perioperative pathways. Generalizability to other countries depends on differences in pediatric critical care capacity, staffing models, and practice patterns.
Expert commentary and contextual considerations
This study raises important policy and clinical questions: should mechanically ventilated children be centralized to ICUs, and if so, how can systems adapt? Key considerations include:
- Workforce and capacity: Expanding PICU capacity and training pediatric intensivists and specialized nursing are long-term solutions but require substantial investment.
- Regionalization and transfer networks: Concentrating high-risk pediatric care in centers with comprehensive ICU resources improves outcomes in other domains (e.g., trauma, neonatal care) and may be appropriate for children expected to require prolonged or complex ventilation.
- Intermediate solutions: Use of high-dependency units, dedicated pediatric step-up beds, rapid response teams, standardized ventilator protocols on wards, and tele-ICU support could mitigate risk where immediate centralization is infeasible.
- Postoperative pathways: The very high proportion of postoperative (particularly cardiac) patients in ICUs underlines the role of standardized ICU admission criteria where protocolized care is accepted practice.
Limitations of the study
Key limitations influence interpretation and planning for further research:
- Administrative data lack physiological severity scores and granular process measures (e.g., time to intubation, ventilator management details, sedation practices), which limits adjustment for baseline illness severity.
- Potential misclassification of location or timing of care transitions between wards and ICUs could bias estimates.
- Residual confounding by unmeasured variables (socioeconomic status, prehospital care, do-not-resuscitate orders) cannot be excluded.
- Heterogeneity within the ‘ICU’ and ‘general ward’ categories (e.g., presence of high-dependency beds, pediatric-trained nurses) was not captured and may influence outcomes.
- Observational design precludes definitive causal inference.
Implications for practice and policy
The study provides evidence supportive of policies that prioritize ICU-level care for children requiring invasive ventilation. Policymakers and hospital leaders should consider:
- Assessing pediatric ICU capacity and geographic distribution to identify gaps in access to specialized critical care.
- Implementing triage guidelines that favor ICU admission for children anticipated to need mechanical ventilation, where capacity permits.
- Investing in workforce development (pediatric intensivists, specialized nursing, respiratory therapists) and pediatric-specific training for ward staff in hospitals that must manage ventilated children.
- Exploring telemedicine, regional transfer agreements, and temporary high-dependency units as scalable interventions to improve outcomes while capacity is expanded.
Research agenda
Priorities for future research include:
- Prospective studies or registries capturing physiologic severity scores, ventilator parameters, and detailed timing of interventions to better adjust for baseline risk.
- Cluster or stepped-wedge implementation studies testing interventions such as tele-ICU support, ward ventilator bundles, and enhanced monitoring to evaluate effect on outcomes.
- Cost-effectiveness analyses comparing centralization versus capacity-building strategies, accounting for transfer logistics and family-centered outcomes.
- Qualitative work exploring barriers to ICU admission and clinician decision-making in different hospital contexts.
Conclusion
Sakurai et al. report that most mechanically ventilated children in Japan are treated on general wards and that ward-based care is associated with higher in-hospital mortality than ICU care after propensity adjustment. The findings are consistent with the notion that specialized ICU structures and processes contribute to better outcomes for ventilated children. Given study limitations, the results should be interpreted as strong signals for health systems to assess pediatric critical care capacity and consider measures ranging from regionalization to targeted process improvements to ensure ventilated children receive the level of care that optimizes survival and recovery.
Funding and clinicaltrials.gov
Refer to the original publication for study-specific funding statements and trial registration. (Sakurai Y et al., Crit Care Med. 2025.)
References
1. Sakurai Y, Michihata N, Osada K, Kobayashi S, Sakamoto W, Uchida Y, Ishii K, Yokohari H, Kurosawa H, Isogai T, Matsui H, Fushimi K, Yasunaga H. Patient Backgrounds and Outcomes of Mechanically Ventilated Children Treated in ICUs Versus General Wards in Japan: A Retrospective Cohort Study Using a National Inpatient Database. Crit Care Med. 2025 Dec 1;53(12):e2497-e2505. doi: 10.1097/CCM.0000000000006901. Epub 2025 Oct 17. PMID: 41104918.