Key result
External validation of 11 mortality prediction models in critically ill patients showed that APACHE II, APACHE IV, and SAPS II had acceptable to excellent discrimination (AUROC up to 0.83) but poor calibration.
Why the study?
Having identified 43 mortality prediction models for critically ill patients in a scoping review, the authors aimed to assess their performances through external validation.
Cohort (n=3,976)
Yes
Effect estimate: AUROC 0.83 (95% CI 0.81-0.85)
External validation of ICU mortality prediction models reveals that while discrimination is often preserved, calibration is generally poor and requires recalibration for accurate use.
APACHE II/IV and SAPS II discrimination supports ICU risk stratification; poor calibration leaves clinical use uncertain without recalibration.
No takes yet. Share an insight, caveat, or question.
Cox et al. (2022) conducted a cohort in Critical illness (n=3,976). Mortality prediction models was evaluated on Hospital mortality prediction (discrimination) (AUROC 0.83, 95% CI 0.81-0.85). External validation of 11 mortality prediction models in critically ill patients showed that APACHE II, APACHE IV, and SAPS II had acceptable to excellent discrimination (AUROC up to 0.83) but poor calibration.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: