Stavem Knut, Lossius Morten, Rønning Ole Morten
Department of Medicine, Akershus University Hospital, Nordbyhagen, Norway.
Cerebrovasc Dis. 2003;16(3):286-91. doi: 10.1159/000071129.
Stroke severity is an important determinant of outcome, however, quantitative data on the initial neurological status might be lacking in retrospective studies. We wanted to assess the reliability and validity of the retrospective use of the Canadian Neurological Scale (CNS).
In 181 patients with validated stroke, two raters scored the CNS based on medical record review. We assessed interrater reliability and construct validity of the CNS. Predictive validity was assessed by the ability of the CNS to predict 30-day and 1-year mortality.
Interrater reliability was high (kappa or weighted kappa 0.76-0.96). Correlations between similar items of prospective Scandinavian Stroke Scale scores and retrospective CNS scores ranged from 0.54 to 0.85. CNS total score was a strong predictor of death within 30 days and 1 year in multivariate models.
The retrospective algorithm for the CNS had a high to substantial interrater reliability and predictive validity. Accordingly, in retrospective stroke studies using medical record information, the CNS can be a feasible instrument to adjust for differences in stroke severity.
卒中严重程度是预后的重要决定因素，然而，回顾性研究可能缺乏关于初始神经状态的定量数据。我们旨在评估回顾性应用加拿大神经功能量表（CNS）的可靠性和有效性。
对181例已确诊卒中患者，两名评估者根据病历回顾对CNS进行评分。我们评估了CNS的评估者间信度和结构效度。通过CNS预测30天和1年死亡率的能力评估预测效度。
评估者间信度较高（kappa或加权kappa为0.76 - 0.96）。前瞻性斯堪的纳维亚卒中量表评分的相似项目与回顾性CNS评分之间的相关性在0.54至0.85之间。在多变量模型中，CNS总分是30天和1年内死亡的有力预测指标。
CNS的回顾性算法具有较高至显著的评估者间信度和预测效度。因此，在利用病历信息的回顾性卒中研究中，CNS可作为一种可行的工具来调整卒中严重程度的差异。
