ICU sedation, delirium & recovery
Augmenting the Simplified Acute Physiology Score II for intensive care unit mortality prediction and benchmarking: a French nationwide study
Authors' conclusion
Compared with the original SAPS II, aug-SAPS II improved calibration and discrimination and reduced between-hospital variability without requiring additional data beyond routinely collected national administrative information.
Study context
- Population
- Adult intensive care unit (ICU) stays, French national database of hospital stays, 2015–2023 (derivation), 2024 (validation), 2025 (sensitivity)ICU stays, derivation cohort
- Participants
- 2,156,332 stays
- Exposure
- Recalibrated SAPS II and augmented SAPS II vs original SAPS II
- Outcome
- In-hospital mortality prediction: calibration (Cox calibration regression, SMRs) and discrimination (AUC)
Methodological detailsnationwide, temporal external validation, sensitivity analyses
- Nationwide
- Temporal external validation
- Sensitivity analyses
Main finding
As reported in the abstract
Discrimination (AUC), derivation cohort
| Group | Value | AUC |
|---|---|---|
| SAPS II | 0.841 (95% CI 0.840–0.841) | 0.841 |
| aug-SAPS II | 0.866 (95% CI 0.865–0.866) | 0.866 |
Value
SAPS II0.841 (95% CI 0.840–0.841)
aug-SAPS II0.866 (95% CI 0.865–0.866)
AUC
SAPS II0.841
aug-SAPS II0.866
Observational data: associations, not effects.
Further findings
| Outcome | SAPS II | aug-SAPS II |
|---|---|---|
| Discrimination (AUC), temporal validation cohort (2024; 242 465 stays) | ||
| AUC | 0.835 | 0.862 |
Discrimination (AUC), temporal validation cohort (2024; 242 465 stays)
- Calibration remained close to ideal and standardized mortality ratios (SMRs) were near 1
Hospital-level random-effects variance in SMRs, validation cohort
- Decreased with aug-SAPS II
- 28.7%
- Unchanged with r-SAPS II
Sensitivity analysis, 2025 data
- Findings were consistent in 2025 data
Original publication
Augmenting the Simplified Acute Physiology Score II for intensive care unit mortality prediction and benchmarking: a French nationwide study. Crit Care. Published 10 March 2026. doi:10.
Composed by AI from the published abstract and checked against it. How Briefs are made →