This nationwide retrospective cohort study evaluated whether outcomes differ for critically ill children treated in pediatric intensive care units (PICU) versus adult intensive care units (AICU) in Germany. The investigators focused on children receiving complex intensive care treatment (CICT) and sought both quantitative estimates of in-hospital case fatality by treating department and qualitative insights into reasons why children are admitted to AICUs.
The analysis used the German nationwide hospital discharge dataset for the years 2016 to 2023. Included cases were patients aged ≥ 28 days and < 18 years who received CICT. Using department codes in the discharge data, cases were categorized as treated in PICU, AICU, or interdisciplinary departments.
Total cases included in the cohort numbered 143,034. Distribution by treating department was 67.8% in PICUs, 14.0% in AICUs, and 18.2% classified as interdisciplinary.
The primary outcome was in-hospital case fatality. Odds ratios (OR) for death were estimated using generalized linear mixed models that included the hospital as a random effect to account for clustering. Models were adjusted for age, presence of acute organ dysfunction, and chronic conditions. The authors also estimated excess deaths attributable to differences in treating department and conducted a survey of pediatric and adult intensivists; the survey results were analyzed qualitatively.
In crude (unadjusted) comparisons, the OR for in-hospital death for cases treated in PICUs versus AICUs was 1.14 (95% CI 1.03 to 1.26), indicating a modestly higher crude case fatality in PICUs compared with AICUs.
After adjustment for age, acute organ dysfunction, chronic conditions, and accounting for hospital-level clustering, the association reversed. The adjusted OR for death in PICU versus AICU was 0.73 (95% CI 0.63 to 0.84), indicating a lower risk of in-hospital death for children treated in PICU settings when controlling for measured confounders.
When PICU and interdisciplinary cases were combined and compared with AICU cases, the fully adjusted OR was 0.61 (95% CI 0.54 to 0.70), further supporting lower risk when care involved pediatric-intensivist environments or interdisciplinary pediatric-capable settings.
Using the observed adjusted differences, the authors estimated excess deaths across the study period associated with AICU care. The estimate of excess deaths was 100 for the primary classification. When interdisciplinary cases were reclassified as pediatric, the estimated excess deaths rose to 191. These figures represent the study's calculated excess mortality burden linked to treatment outside routine pediatric intensive care settings.
A survey of pediatric and adult intensivists accompanied the quantitative work. The qualitative analysis identified three primary domains that underlie admissions of children to AICUs: capacity constraints (limited PICU bed availability), organizational factors (referral pathways, hospital structure), and issues related to clinical expertise (availability of routine pediatric intensive care experience). The abstract reports these thematic domains but does not provide further methodological details of the survey in the abstract itself.
Crude mortality figures suggested slightly higher case fatality in PICUs versus AICUs, but after risk adjustment the opposite was true: children treated outside of routine pediatric intensive care settings had higher risk-adjusted case fatality. The authors conclude that their findings support treating critically ill children in settings with routine pediatric intensive care experience, highlighting potential outcome consequences when children receive complex intensive care outside PICUs.
The study used secondary, fully anonymized data from the Federal Statistical Office; the authors report that this did not require ethics approval. The original dataset remains at the Federal Statistical Office and can be accessed by qualified researchers at designated research data centers after request and confidentiality agreement. The authors declared no competing interests and followed relevant reporting guidelines.
The abstract does not report detailed survey methodology, response rates, or the content and coding approach used in the qualitative analysis. Granular patient-level information beyond the listed adjustment variables (for example, detailed severity scores, timing of interventions, or specific diagnoses) is not provided in the abstract. Other potential limitations, such as residual confounding, misclassification of department codes, or temporal trends over the 2016–2023 study period, are not elaborated in the abstract and would require examination of the full manuscript and supplementary materials for confirmation.