Predicting 28-day all-cause mortality in patients admitted to intensive care units with pre-existing chronic heart failure using the stress hyperglycemia ratio: a machine learning-driven retrospective cohort analysis
Authors' conclusion
For patients with both acute critical illness and pre-existing CHF, SHR was an independent predictor of 28-day all-cause mortality.
Study context
- Population
- Patients admitted to intensive care units (ICUs) with acute critical illness and pre-existing chronic heart failure (CHF), Medical Information Mart for Intensive Care (MIMIC) database
- Participants
- 913 patients
- Exposure
- Higher vs lower stress hyperglycemia ratio
- Outcome
- 28-day all-cause mortality
- Follow-up
- 28 days
Methodological detailsmachine learning, MIMIC database
- Machine learning
- MIMIC database
Main finding
As reported in the abstract
Association of SHR with 28-day all-cause mortality
- Unadjusted HR per stress hyperglycemia ratio (SHR)
- 1.45 (P < 0.001)
- Adjusted HR per SHR
- 1.43 (P < 0.001)
- Risk of 28-day mortality increased with higher SHR levels
- P < 0.001
Observational data: associations, not effects.
Further findings
Subgroup analysis
- No potential risk factor (demographics, comorbidities, drugs) affected the relationship
- P for interaction > 0.05
Discrimination (ROC)
- SHR cut-off
- 0.57
- AUC for SHR larger than for admission blood glucose and HbA1c
28-day survival, SHR above vs below cut-off
- Significantly lower 28-day survival probability with SHR higher than cut-off
- P < 0.001
Machine learning prediction models
- Neural network best of four algorithms, AUC
- 0.801
- SHR a key factor for 28-day mortality by Boruta algorithm
Original publication
Predicting 28-day all-cause mortality in patients admitted to intensive care units with pre-existing chronic heart failure using the stress hyperglycemia ratio: a machine learning-driven retrospective cohort analysis. Cardiovasc Diabetol. Published 8 January 2025. doi:10.
Composed by AI from the published abstract and checked against it. How Briefs are made →