Related Experiment Videos
The inter- and intraobserver reliabilities of a new classification system for ischaemic stroke: the South London
1Guy's, King's and St. Thomas' School of Medicine, King's College London, 5th Floor, Capital House, 42 Westmont Street, SE13QD, London, UK.
Objectives:
This study aims to determine the inter- and intraobserver reliabilities of an aetiological classification devised as part of a community-based stroke register.
Methods:
Patients with first-ever acute ischaemic stroke were clinically assessed and received investigation according to a set protocol. Results of the clinical assessment and investigations were used to determine an aetiological stroke subtype for each patient by observer 1. Aetiological subtypes consisted of extracranial large artery atherosclerosis (LAAec), intracranial large artery atherosclerosis (LAAic), high-risk cardioembolism (CH), medium-risk cardioembolism (CM), small vessel occlusion (SVO), other aetiology (OTH), no aetiology identified (NA) and multiple probable or multiple possible aetiology (MPA). The same data were distributed to a further four observers along with the criteria for the classification system. Two of the observers were retested on the same patients after of period of 8 weeks. Inter- and intraobserver agreement was determined using the kappa statistic, which gives the chance-adjusted percentage agreement.
Results:
Forty-five consecutive patients were included. The overall kappa statistic for ischaemic stroke was 0.91 indicating excellent agreement. Kappa statistics were highest for the more frequent subtypes of SVO (0.97) and CH (0.97). Substantially high kappa statistics were also obtained for the less-frequent categories of LAAec (0.91), CM (0.84), NA (0.89) and MPA (0.87). A low kappa statistic was obtained for the category OTH (0.16), which had a low frequency of reporting, indicating poor agreement. The kappa statistic for probable categories was higher than the kappa statistic for all stroke subtypes at 0.96. Intraobserver agreement between first and second assignments of subtype diagnoses for both observers reached excellent agreement with kappa statistics of 0.83 and 0.85.
Conclusion:
The aetiological classification system, designed for use in the investigation of the epidemiology, stroke subtype and their relation to the natural history of stroke in a multiethnic inner city population, allows high inter- and intrarater agreements of subtype diagnosis.