Highlight
– A multicenter retrospective registry study (8,794 RRS activations, 35 Japanese hospitals) evaluated standardized ICU admission ratio (SIAR) and outcomes after rapid response system (RRS) activations.
– Median ICU admission rate after RRS activation was 0.33; median SIAR was 0.98. Higher SIAR was independently associated with lower odds of a composite outcome of severe neurologic disability (CPC ≥3) or death within 30 days.
– Each 0.1-unit increase in SIAR correlated with a clinically meaningful reduction in the composite endpoint (adjusted OR 0.94), suggesting that greater ICU utilization after deterioration may improve short-term neurologic and survival outcomes, though causality is unproven.
Background
Rapid response systems (RRS; also called medical emergency teams or rapid response teams) are intended to identify hospitalized patients at risk of deterioration and to provide earlier critical care evaluation and intervention, with the goals of preventing cardiac arrest, reducing unplanned ICU admissions, and improving survival and functional outcomes. ICU admission practices after RRS activations vary substantially between hospitals, reflecting differences in ICU capacity, triage thresholds, risk tolerance, and local policies. Whether higher ICU utilization after RRS activation translates into better patient-centered outcomes remains uncertain.
The standardized ICU admission ratio (SIAR) is a facility-level metric defined as the observed number of ICU admissions after RRS activation divided by the number predicted by a model that adjusts for patient- and event-level covariates. SIAR can be used to compare institutional practice intensity after deterioration while accounting for case mix. The study by Ono and colleagues (Anesthesiology, 2025) examined institutional variation in ICU admission rates and the association between SIAR and short-term outcomes following RRS activations in Japan.
Study design
This was a retrospective multicenter observational study using a Japanese in-hospital emergency registry of RRS activations between 2018 and 2022. The analysis included 8,794 RRS events from 35 participating institutions. For each hospital, investigators calculated the ICU admission rate (number of ICU admissions divided by number of RRS activations) and the standardized ICU admission ratio (SIAR), defined as observed ICU admissions divided by predicted ICU admissions.
Predicted ICU admissions were derived from a model adjusting for relevant patient- and event-level characteristics (the article provides the model specification; readers should consult the full text for covariates and model performance metrics). Associations between institutional SIAR and patient outcomes were tested using generalized estimating equation (GEE) logistic regression with clustering by hospital to account for within-hospital correlation.
Primary outcome: death within 30 days of RRS activation (events recorded during hospitalization, up to 30 days).
Secondary outcome: composite of Cerebral Performance Category (CPC) score ≥3 (indicating severe neurologic disability) or death within 30 days.
Key findings
Population and baseline measures
The cohort included 8,794 patients who experienced RRS activation across 35 Japanese hospitals over a 4–5 year period. The median institutional ICU admission rate after RRS activation was 0.33 (interquartile range [IQR], 0.21 to 0.47), indicating considerable between-hospital variation. The median SIAR was 0.98 (IQR, 0.75 to 1.17), indicating that, across sites, observed ICU admissions were close to model-predicted values on average, but with substantial institutional dispersion.
Univariable associations
In unadjusted analyses, SIAR did not show a statistically significant association with death within 30 days (β = −0.05; 95% confidence interval [CI], −0.12 to 0.01; P = 0.108). However, SIAR was significantly negatively associated with the composite outcome of CPC ≥3 or death within 30 days (β = −0.15; 95% CI, −0.27 to −0.03; P = 0.015), suggesting that institutions with higher-than-expected ICU utilization after RRS calls had fewer severe neurologic disability events or deaths.
Multivariable (adjusted) associations
After adjustment for covariates and accounting for hospital clustering, the investigators expressed the association per 0.1-unit increase in SIAR. For death within 30 days, the adjusted odds ratio (OR) per 0.1-unit SIAR increase was 0.98 (95% CI, 0.97 to 0.99; P = 0.104) — a non-significant trend toward lower mortality. For the composite endpoint (CPC ≥3 or death within 30 days), each 0.1-unit higher SIAR was associated with an adjusted OR of 0.94 (95% CI, 0.92 to 0.96; P < 0.001), a statistically robust and clinically relevant reduction in the odds of severe neurologic outcomes or death.
Interpretation of effect sizes
To aid clinical interpretation: a 0.1-unit increase in SIAR (for example, moving from 0.9 to 1.0) was associated with a ~6% reduction in odds of the composite outcome; a full 0.5-unit increase would translate into an approximately 26% reduction in odds (using multiplicative ORs) assuming linearity within the observed range. These associations were stronger for the composite endpoint than for mortality alone, suggesting an influence on neurologic preservation as well as survival.
Safety and secondary considerations
The study did not report intervention-related adverse events tied to ICU admission policy changes, and no randomized intervention was performed. Because this is observational, increased ICU utilization may carry system-level costs (bed occupancy, resource strain) not captured in the outcomes measured.
Expert commentary
What the findings suggest
Ono et al. provide important multicenter evidence that institutional propensity to admit patients to the ICU after RRS activation—measured by an adjusted and standardized metric—correlates with improved short-term neurologic and combined survival outcomes. There are plausible clinical mechanisms: earlier access to invasive monitoring, rapid organ support (vasopressors, mechanical ventilation, renal replacement therapy), protocolized sepsis care, and concentrated multidisciplinary critical care expertise may mitigate progression to catastrophic organ failure and secondary brain injury.
Key methodological strengths
– Large multicenter cohort from a national registry increases representativeness within Japan.
– Use of SIAR attempts to control for case mix and provides a facility-level metric rather than raw admission rates.
– GEE modeling with hospital clustering respects the hierarchical data structure.
– Use of a functional neurologic outcome (CPC) as part of the primary secondary endpoint emphasizes patient-centered consequences beyond survival.
Limitations and alternative explanations
– Observational design: residual confounding and unmeasured variables (e.g., staffing levels, ICU staffing model, baseline code status, limitations of care, bed availability at time of event, timing of RRS activation relative to deterioration onset) can bias associations.
– SIAR depends on the validity and calibration of the prediction model for expected ICU admissions; model misspecification could distort SIAR interpretation.
– Reverse causation: hospitals that selectively admit only sicker patients to ICU may paradoxically have worse outcomes. Conversely, hospitals with lower thresholds for ICU admission may identify and treat at-risk patients earlier, improving outcomes, but at higher resource use.
– Generalizability: Japanese hospital structure, ICU bed density, and RRS implementation models may differ internationally.
– The primary mortality outcome reached only a nonsignificant trend in adjusted analysis; the stronger signal for the composite endpoint suggests more impact on neurologic preservation than on absolute survival within 30 days.
– No cost-effectiveness or system-level impact data were reported; increased ICU utilization may have downstream effects on elective surgeries or ward throughput.
Implications for practice and policy
SIAR may be a useful performance metric for hospitals to monitor practice patterns after RRS activations and to prompt review of triage thresholds. However, adopting policies to increase ICU admissions broadly may not be the correct or sustainable approach without considering capacity, prioritization, and targeted criteria for those most likely to benefit. Institutions should couple SIAR surveillance with audits that examine case mix, outcomes by indication, time to ICU transfer, and alternatives such as high-acuity stepdown units.
Conclusion and research priorities
This multicenter retrospective study shows that higher-than-predicted ICU utilization after RRS activations (higher SIAR) is associated with a lower incidence of severe neurologic disability or death within 30 days. The findings support the hypothesis that timely access to ICU-level care after in-hospital deterioration can improve patient-centered outcomes. However, observational design limits causal inference.
Priority next steps:
– Validate the SIAR approach in other health systems and with prospective data to assess reproducibility and calibration of prediction models.
– Conduct mixed-methods studies and audits to identify which subgroups benefit most from ICU admission after RRS activation (e.g., sepsis, respiratory failure, shock, altered mental status) and to refine triage criteria.
– Evaluate system-level trade-offs including ICU capacity, cost-effectiveness, and impacts on non-RRS care pathways.
– Test care pathway interventions (stepped-wedge or cluster randomized designs) that couple earlier critical care input with structured treatment bundles to determine causal effects on neurologic outcomes and survival.
Funding and clinicaltrials.gov
The original article citation is provided below. Funding sources and clinical trial registration details should be consulted in the published paper; the present summary does not add or alter reported funding statements.
References
Ono S, Uchino S, Tokito M, Saito T, Sasabuchi Y, Sanui M. Impact of the Standardized Intensive Care Unit Admission Ratio on Outcomes in Rapid Response System Activations: A Retrospective Multicenter Study in Japan. Anesthesiology. 2025 Nov 1;143(5):1255-1265. doi: 10.1097/ALN.0000000000005689. Epub 2025 Jul 30. PMID: 40737083.
Suggested reading for context (not exhaustive)
Readers wishing to place these findings in context should consult systematic reviews and guideline documents on RRS implementation and early critical care intervention, and seek articles addressing ICU triage policies and resource allocation. Detailed appraisal of prediction model variables and performance (found in the original article) is recommended before applying SIAR locally.
