BACKGROUND: The Pediatric Stroke Outcome Measure-Summary of Impressions (PSOM-SOI) measures neurological function across right and left sensorimotor domains (Item A), language production (Item B), language comprehension (Item C), and cognition/behaviour (Item D).OBJECTIVE: This study was a cross-cultural adaptation into French of the PSOM-SOI and an assessment of its reliability and limitations of use.MATERIAL AND METHODS: The translation and adaptation of the PSOM-SOI was followed by the assessment of its reliability in a cohort of 69 children with diagnosed acute neonatal arterial ischemic stroke. Three independent raters retrospectively scored the PSOM-SOI based on data from in-person neurological examination and results of standardized tests performed at age 7 in the cohort database. Comparison 1 (C1) involved a less experienced rater and an experienced rater and comparison 2 (C2) involved 2 experienced raters. Inter-rater reliability (IRR) was measured with Kappa coefficients.RESULTS: The cross-cultural adaptation was easily performed, and no rater had difficulties using the French PSOM-SOI. The IRR was better in C1 than C2. For Item A, the agreement in C1 (κ=0.47) and C2 (κ=0.44) was moderate. The C1 agreement was substantial for Items B (κ=0.71) and C (κ=0.70); the C2 agreement was fair for Item B (κ=0.23) and slight for Item C (κ=0.16). For Item D, the agreement was moderate in C1 (κ=0.52) and fair in C2 (κ=0.35). In all but one comparison, agreement or minor disagreement (≤0.5 points) was obtained for more than 90% of the item scores. Regarding the total score, agreement for normal function (≤0.5) versus abnormal function (>0.5) was achieved for 90% in C1 and 67% in C2.CONCLUSION: The IRR of the French PSOM-SOI gave variable results depending on the item and rater's experience, but the extent of disagreements was minor for individual items and total score. Additional prospective validation studies using the French PSOM-Short Neurological Exam to score the PSOM-SOI are needed. A dichotomised total score (cut-off≤0.5) could be used to define normal function versus poor outcome.Copyright © 2019 Elsevier Masson SAS. All rights reserved.