Clinimetric properties of scales for neurological assessment and monitoring after stroke: a series of reviews-Reliability of items within scales

Abstract

Abstract title (25 word limit)
Clinimetric properties of scales for neurological assessment and monitoring after stroke: a series of reviews- Reliability of items within scales.
Introduction
Neurological assessment and monitoring is crucial for identifying deterioration, and initiating changes in management after stroke. Measurement scales must be adequate for purpose and perform well across a range of properties. This series of reviews aimed to allow comparison of the reliability of assessment items within scales.

Methods
A scoping review first identified 24 stroke neurological assessment scales. Searches for literature for each scale were then run across 5 databases using COSMIN search strategy. Multiple methods were reported for calculating inter-rater reliability; however this presentation focuses on the papers that kappa (k) statistics. Landis and Koch classifications were applied to aid comparison within and between items and scales.

Results
Data was available across 13 of the 24 scales (38 papers). Results have been tabulated to allow comparison. Reported reliability across scale items ranged widely and no items consistently classify as very good. Ataxia, facial palsy, visual fields, gaze, and extinction/neglect seem to be less reliable than items such as motor power and LOC- commands.

Conclusions
The reliability across all scales, even those frequently used in practice, was inconsistent. Furthermore, there is no agreed standard for the level of reliability of scales for clinical practice. These findings indicate that more work is needed to improve standards of reliability in neurological assessment and monitoring. Mechanisms to achieve this could be changes in the way specific items are assessed and attention to consistent application of scales in practice.