Cardiac arrest & resuscitation
Machine Learning to Predict Individualized Treatment Effects of Sodium Bicarbonate for Patients With Out-of-Hospital Cardiac Arrest
Authors' conclusion
This study suggests the heterogeneous effects of SB on ROSC rates in patients with OHCA.
Study context
- Population
- Adults with nontraumatic out-of-hospital cardiac arrest (OHCA), 3 National Taiwan University Hospital system hospitals, 8-year databasePropensity score matching (PSM) cohort
- Participants
- 2,368 adults
- Exposure
- Sodium bicarbonate vs no sodium bicarbonate
- Outcome
- Any ROSC following resuscitation in the ED
Methodological detailsmachine learning (XGBoost-based causal forest), propensity-score matched, multicentre
- Machine learning (XGBoost-based causal forest)
- Propensity-score matched
- Multicentre
Main finding
As reported in the abstract
Predicted individualized treatment effect (ITE) of sodium bicarbonate (SB) on return of spontaneous circulation (ROSC) ranged from absolute increase to absolute reduction
24.7% increase to 28.3% reduction
Observational data: associations, not effects.
Further findings
Effect modification by tertile of predicted ITE
- Tertile of predicted ITE significantly modified effect of clinician treatment assignment on outcome
- P < 0.001
- C-for-benefit (discriminates benefit better than random allocation, also by Qini curve)
- 0.61
Factors associated with higher predicted benefit from SB
- Older age, poorer renal function, longer scene-to-hospital time, metabolic acidosis, hyperkalemia
Original publication
Machine Learning to Predict Individualized Treatment Effects of Sodium Bicarbonate for Patients With Out-of-Hospital Cardiac Arrest. Crit Care Med. Published 13 August 2025. doi:10.
Composed by AI from the published abstract and checked against it. How Briefs are made →