ICU sedation, delirium & recovery
Optimizing Agreement Between Bedside Nurse-Documented and Trained Researcher Delirium Assessments in the ICU
Authors' conclusion
A delirium model informed by bedside nurse ICDSC findings and clinical variables improves accuracy of delirium detected in the ICU
Study context
- Population
- Critically ill adults with acute respiratory failure or sepsis in ICUs of large academic hospitals, southwestern Pennsylvania1,535 matched assessments
- Participants
- 279 adults
- Comparison
- Bedside nurse Intensive Care Delirium Screening Checklist (ICDSC)-based delirium assessment vs trained researcher Confusion Assessment Method for the ICU (CAM-ICU) (reference standard)
- Outcome
- Agreement between bedside nurse and trained researcher delirium assessments (Cohen's kappa) and model prediction of positive research CAM-ICU
Methodological detailsmodel derivation with ten-fold cross-validation
- Model derivation with ten-fold cross-validation
Performance
As reported in the abstract
| Test or model | AUC | Cohen's kappa vs research CAM-ICU | F1 score (cross-validation) |
|---|---|---|---|
| ICDSC ≥ 4 | — not stated in the abstract | 0.42 | — not stated in the abstract |
| Logistic regression model | 0.87 | — not stated in the abstract | 0.72 |
AUC
ICDSC ≥ 4— not stated in the abstract
Logistic regression model0.87
Cohen's kappa vs research CAM-ICU
ICDSC ≥ 40.42
Logistic regression model— not stated in the abstract
F1 score (cross-validation)
ICDSC ≥ 4— not stated in the abstract
Logistic regression model0.72
— not stated in the abstract
Further findings
Models with more limited ICDSC information (sensitivity analyses)
- Fair to good discriminatory ability (F1)
- 0.60-0.70
- Validated cutoff model had the lowest performance
Original publication
Optimizing Agreement Between Bedside Nurse-Documented and Trained Researcher Delirium Assessments in the ICU. Crit Care Med. Published 19 September 2025. doi:10.
Composed by AI from the published abstract and checked against it. How Briefs are made →