Discrimination and Calibration of the ICH Score for 30-Day Mortality After Surgical Evacuation of Spontaneous Intracerebral Hemorrhage at an Indonesian Tertiary Referral Center

Abstract

Background: The ICH Score is the most widely used prognostic grading scale for spontaneous intracerebral hemorrhage (ICH), but its accuracy after surgical hematoma evacuation has not been evaluated in Indonesia.

Objective: To evaluate the discrimination and calibration of the ICH Score for 30-day mortality in surgically treated patients.

Methods: Retrospective cohort of 81 consecutive adults with spontaneous ICH who underwent hematoma evacuation at Dr. Mohammad Hoesin General Hospital, Palembang (July 2024–June 2025). Discrimination was summarized by the area under the receiver operating characteristic curve (AUC, DeLong 95% confidence interval [CI]) and calibration against the original stratum risks; component associations were tested exactly, and adjusted component effects were bounded across all datasets consistent with the observed tables.

Results: Twenty-nine (35.8%) died within 30 days: none of 20 with scores 0–1, 4 of 36 (11.1%) with a score of 2, and all 25 with scores 3–4. The AUC was 0.958 (95% CI 0.917–0.998). A cutoff of ≥3 gave 86.2% sensitivity and 100% specificity. Total deaths were close to expectation (observed/expected 0.94), yet stratum-level calibration was poor, with an exact Hosmer–Lemeshow-type p of 0.011 and a recalibration slope of 3.00 (95% CI 1.88–6.21). Age ≥80 years (odds ratio [OR] 26.84) and intraventricular hemorrhage (OR 15.24) were associated with death; only the latter remained significant in every admissible dataset.

Conclusion: After surgical evacuation the ICH Score ranked 30-day mortality well, but its original risk estimates were miscalibrated. Locally recalibrated risks may suit counseling better than the original percentages but need external validation; a high score should not define futility.