Validation and Refinement of PreICH scale to Identify Intracerebral Hemorrhage Versus Large-Vessel Occlusion

Background and Purpose- Prehospital large-vessel occlusion (LVO) scales identify severe stroke syndromes but may not distinguish LVO from intracerebral hemorrhage (ICH). We aimed to prospectively validate the PreICH scale, with the primary diagnostic objective of differentiating ICH from confirmed LVO, and to explore whether additional hemorrhage-oriented variables could refine its performance. Methods- We conducted a prospective observational study of consecutive stroke-code activations evaluated before neuroimaging by a vascular neurologist. PreICH was calculated prospectively. Patients with calculable PreICH and valid final diagnosis were included. The primary diagnostic cohort comprised confirmed LVO and ICH. Secondary cohorts included ischemic stroke versus ICH and the overall stroke-code cohort, including stroke mimics. Multivariable NIHSS-adjusted models identified variables associated with ICH. A modified PreICH score (mPreICH) was derived post hoc and evaluated as exploratory apparent performance. Results- Among 1012 screened activations, 982 patients were analyzed: 597 ischemic strokes, 91 ICH, and 294 stroke mimics. The LVO-versus-ICH cohort included 144 LVO and 91 ICH. NIHSS and RACE were higher in ICH than in ischemic stroke, but did not differ between LVO and ICH (NIHSS, 13 [IQR, 7-20] versus 15 [5-23], P=0.300; RACE, 5 [2-8] versus 6 [2-8], P=0.435). In the LVO-versus-ICH cohort, PreICH showed an AUC of 0.758 (95% CI, 0.696-0.820), whereas RACE did not discriminate LVO from ICH (AUC, 0.530 [95% CI, 0.453-0.607]). The exploratory mPreICH showed apparent AUCs of 0.835 (95% CI, 0.785-0.884) for ischemic stroke versus ICH and 0.798 (95% CI, 0.740-0.856) for LVO versus ICH. Conclusions- In this prospective stroke-code cohort, severity-based scales distinguished ICH from the overall ischemic stroke population but showed limited ability to differentiate LVO from ICH. An exploratory modified PreICH scale incorporating additional hemorrhage-oriented variables improved apparent discriminative performance, including in the LVO-versus-ICH setting. External validation is required before potential implementation in prehospital decision-making.