Association of the ICH Score With Withdrawal of Life-Sustaining Treatment Over a 10-Year Period.

Abstract

Objective: The intracerebral hemorrhage (ICH) score was developed to enhance provider communication and facilitate early severity assessment. We examined the association of the ICH score with mortality and withdrawal of life-sustaining treatment (WLST) in a large, multicenter stroke registry, and evaluated temporal trends in these associations.

Methods: We identified ICH patients from the Florida Stroke Registry from 2013 to 2022. Outcomes were WLST and in-hospital mortality. ICH scores were grouped as 0-2, 3-4, and 5-6. Importance plots identified key predictors of WLST. Model performance was assessed using AUC-ROC for logistic regression and random forest, adjusted for relevant confounders. Secondary analyses compared outcomes between 2015-2018 and 2019-2022 using stratified univariate logistic regression.

Results: In total, 12,426 patients were included (mean age 69, 55% male, 56% white). The most predictive factors associated with WLST were ICH score, age, state region, presenting level of consciousness, insurance status, and race (RF AUC = 0.94, LR AUC = 0.82). Mortality was 6.6%, 41.5%, and 66% for ICH score 0-2, 3-4, and 5-6. WLST occurred more frequently in higher ICH score groups (OR 9.35 [95% CI: 8.5-10.3] for scores 3-4; OR 18.64 [95% CI: 15.28-22.74] for scores 5-6). Early WLST (< 48 h) was more common in higher score groups (OR 2.97 [95% CI: 2.48-3.55] for 3-4; OR 9.51 [95% CI: 7.33-12.35] for 5-6).

Interpretation: Higher ICH scores were strongly associated with mortality and WLST, including early withdrawal decisions. These associations remained largely consistent over time. These observational findings underscore the need for continued attention to how prognostic scores may influence WLST decisions.