Abstract TP353: Long-term Outcomes in Patients With Intracerebral Hemorrhage and Delayed Hospital Presentation
Background: Delays in medical care are known to be associated with worse outcomes in ischemic stroke, but outcomes in patients with intracerebral hemorrhage (ICH) and delayed presentation are unclear. We aimed to determine factors associated with prolonged delays from ICH symptom onset to hospital presentation and implications for long-term outcomes. Methods: We performed a single-center cohort study using data from consecutive ICH patients over 12 months. ICH characteristics and outcomes were prospectively collected, while time of symptom onset (or last-known-well) and emergency department arrival were retrospectively abstracted. We calculated time-to-arrival and defined prolonged delay as >24 hours. Using multivariable logistic regression, we determined factors associated with prolonged delays to presentation, then determined associations with unfavorable 3-month outcomes (modified Rankin Scale [mRS] 4-6) after adjusting for demographics and ICH severity. Results: Of 299 patients with out-of-hospital ICH, 21% (n=62) presented >24 hours from symptom onset; median time-to-arrival was 5.5 hours (IQR 1.2-17.8). There were not significant differences in age (mean 71.9±14.0 vs. 70.4±16.0, p=0.50), sex (48% vs. 50% male, p=0.80), race (89% vs. 82% white, p=0.22), or ICH size (mean 15.5±23.2 vs. 20.5±27.4cc, p=0.19) between patients presenting >24 hours and <24 hours from symptom onset, though patients with prolonged delays were less likely to have initial GCS <13 (16% vs. 34%, p=0.02) and therefore had modestly lower ICH scores (median 1 [0-2] vs. 1 [1-2], p=0.02). Patients with prolonged delays had lower 3-month mRS scores than patients who presented earlier (median 3 [1.5-4] vs. 4 [3-6], p=0.002), and lower odds of unfavorable 3-month outcome in adjusted models (OR 0.46, 95% CI 0.22-0.97). Conclusions: Outcomes in ICH patients with prolonged delays to presentation differ from those who present earlier. ICH severity in such patients may not be accurately captured by established predictors, and prognostication models should therefore account for inherent survivorship bias.