Perioperative medicine & outcomes
Prioritising deteriorating patients using time-to-event analysis: prediction model development and internal–external validation
Authors' conclusion
Time-varying covariate Cox models can be powerful tools for triaging patients, which may lead to more efficient and effective care in time-poor environments when the times between observations are highly variable.
Study context
- Population
- Adult inpatients, 5 Australian hospitals, 1 January 2019 to 31 December 2020
- Participants
- 150,342 admissions
- Comparison
- Cox regression vs Discrete-time logistic regression
- Outcome
- Model discrimination (cross-validated AUC) for in-hospital mortality
- Follow-up
- Within 24 h
Methodological detailsinternal–external cross validation, multicentre, time-varying covariates
- Internal–external cross validation
- Multicentre
- Time-varying covariates
Performance
As reported in the abstract
Cross-validated AUC, mortality within 24 h · 24 h
| Test or model | AUC |
|---|---|
| Cox regression | 0.96 |
| Discrete-time logistic regression | 0.93 |
AUC
Cox regression0.96
Discrete-time logistic regression0.93
Further findings
| Outcome | Cox regression | Discrete-time logistic regression |
|---|---|---|
| Cross-validated AUC, mortality within 1 week | ||
| AUC | 0.93 | 0.88 |
Calibration
- Calibration varied by hospital; mitigated by ranking patients by predicted risks
Original publication
Prioritising deteriorating patients using time-to-event analysis: prediction model development and internal–external validation. Crit Care. Published 17 July 2024. doi:10.
Composed by AI from the published abstract and checked against it. How Briefs are made →