Validity and reliability of estimating the Scandinavian Stroke Scale score from medical records

Background: Reliable estimation of severity of neurological impairments early after stroke is essential for research and audit of acute stroke care. Obtaining this information prospectively requires significant resources. We wished to assess the reliability of estimating the Scandinavian Stroke Scale (SSS) score retrospectively from routine hospital admission records. Methods: Acute stroke admissions to a large urban hospital were examined and had their SSS scored by an experienced physician within 4 h of the examination performed by the medical admissions team. Two examiners ( a trained research nurse and a second physician), blinded to the patients' clinical condition, later independently estimated retrospective SSS scores using information documented in the medical admission notes. Results: Fifty patients were recruited [ median age 73 years ( interquartile range 61, 79)]. Weighted kappa statistics for agreement between domains of the face-to-face and retrospective SSS were as follows: consciousness 0.73, eye movements 0.60, arm motor power 0.83, hand motor power 0.71, leg motor power 0.81, orientation 0.81, speech 0.80, and facial palsy 0.53. The intraclass correlation coefficient for face-to-face and retrospective SSS composite scores was 0.97 (95% CI 0.96 - 0.98), p < 0.0001. Interobserver reliability for the different components of the retrospective SSS was excellent ( kappa values greater than 0.75) apart from consciousness (0.71) and eye movements (0.58). Conclusions: The composite SSS score and most of its individual components can be reliably estimated retrospectively from routine hospital admission records. This method is potentially useful both in observational studies and in case-mix adjustment for audit purposes. Copyright (C) 2004 S. Karger AG, Basel.