Article Text

PDF
G459(P) Inter-rater reliability of two outcome scoring tools in paediatric head injury
  1. E Casselden1,
  2. FJ Kirkham2,
  3. AJ Durnford3
  1. 1Faculty of Medicine, University of Southampton, Southampton, UK
  2. 2Paediatric Neurology, University Hospital Southampton, Southampton, UK
  3. 3Wessex Neurological Centre, University Hospital Southampton, Southampton, UK

Abstract

Aims Head injury is the commonest cause of childhood-acquired disability. Evaluation of clinical management and research strategies are dependent on reliable and accurate paediatric-specific outcome tools. The Glasgow Outcome Scale-Extended (GOS-E) is a standard tool used in adults; two modified versions have been proposed: GOS-E Peds and the King’s Outcome Scale for Childhood Head Injury (KOSCHI). Inter-rater reliability and retrospective use of GOS-E Peds has not been examined. The KOSCHI developers recommend pilot comparisons between observers prior to use. Both would ideally be used by various healthcare professionals, promoting widespread use. We aimed to examine the inter-rater reliability of both tools.

Methods Two observers, a consultant paediatric neurologist and senior medical student, assessed outcome at discharge and last follow-up, for a sample of 30 patients taken from an existing, retrospectively-acquired database of children admitted to our centre with brain injury requiring intracranial pressure monitoring, between 1999 and 2013. Information gathered from both hospital computer databases and clinical notes was scored according to the published scales. Observers were blind to the other’s outcome, which was undertaken independently. A Cohen’s Kappa test was utilised to assess inter-rater reliability at both timepoints for each tool. For KOSCHI, Kappa was also calculated for both major categories and subcategories.

Results Of the 30 patients assessed, 5 were excluded due to death or missing reports. Time from discharge to last follow-up ranged from 5–360 weeks. At discharge, k = 0.199 for GOS-E Peds, demonstrating a ‘poor’ level of agreement. KOSHI scores at discharge showed greater agreement, k = 0.264 (‘fair’). Both tools showed greater reliability at follow-up, with ‘fair’ agreement. (GOS-E Peds k = 0.477 and KOSCHI k = 0.307. KOSCHI reliability improved when combining subcategories into major outcome groups, with k = 0.324 (‘fair’) at discharge, and at follow-up, 0.590 (high ‘moderate’), in line with previously reported inter-rater reliability.

Conclusion k = 0.41 is thought to be adequate; GOS-E Peds has satisfactory inter-rater reliability at follow-up, KOSCHI only has when subcategories are considered together. Both appear worse at discharge. Further assessment of the reliability of paediatric head injury outcome scores is required. Future research should involve increased sample size with multiple raters of different specialities and grades.

Statistics from Altmetric.com

Request permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.