Perioperative medicine & outcomes
Development and validation of delirium prediction models for noncardiac surgery patients
Authors' conclusion
Using preoperative and intraoperative variables, simple static models performed as well as a dynamic delirium prediction model that also included postoperative variables.
Study context
- Population
- Noncardiac surgical patients, major quaternary medical center; training Jan 2016–Jun 2020, validation Jul 2020–Jan 2022training 51,677
- Exposure
- Static random survival forest and Cox models (preoperative ± intraoperative variables) and a dynamic landmark model (pre-, intra- and postoperative variables)
- Outcome
- Validation performance
Methodological detailsdevelopment and validation, single-centre
- Development and validation
- Single-centre
Main finding
As reported in the abstract
| Group | C-statistic | Brier score | Overall c-index |
|---|---|---|---|
| Static random forest, preoperative | 0.81 (95% CI: 0.79, 0.82) | 0.04 | |
| Static random forest, preoperative + intraoperative | 0.86 (95% CI: 0.84, 0.87) | 0.04 | |
| Dynamic model | 0.84 (95% CI: 0.83, 0.85) |
C-statistic
Static random forest, preoperative0.81 (95% CI: 0.79, 0.82)
Static random forest, preoperative + intraoperative0.86 (95% CI: 0.84, 0.87)
Dynamic model— not stated in the abstract
Brier score
Static random forest, preoperative0.04
Static random forest, preoperative + intraoperative0.04
Dynamic model— not stated in the abstract
Overall c-index
Static random forest, preoperative— not stated in the abstract
Static random forest, preoperative + intraoperative— not stated in the abstract
Dynamic model0.84 (95% CI: 0.83, 0.85)
Observational data: associations, not effects.
Further findings
Events
| Outcome | Training dataset | Validation dataset |
|---|---|---|
| Delirium incidence | 2795 | 912 |
Parts of this Brief are withheld: the automatic checks could not confirm them against the abstract.
Original publication
Development and validation of delirium prediction models for noncardiac surgery patients. J Clin Anesth. Published 18 November 2023. doi:10.
Composed by AI from the published abstract and checked against it. How Briefs are made →