Interrater agreement of the diagnosis and classification of a first seizure in childhood. The Dutch Study of Epilepsy in Childhood
- H Stroink1,
- C A van Donselaar2,
- A T Geerts3,
- A C B Peters4,
- O F Brouwer5,
- O van Nieuwenhuizen4,
- R F M de Coo3,
- H Geesink6,
- W F M Arts3
- 1Department of Neurology, St Elisabeth Hospital and TweeSteden Hospital, the Netherlands
- 2Rudolf Magnus Institute of Neuroscience, Epilepsy Section, University Medical Centre, Utrecht, the Netherlands
- 3Department of Paediatric Neurology, Erasmus Medical Centre/Sophia Children’s Hospital, Rotterdam, the Netherlands
- 4Department of Paediatric Neurology, University Medical Centre, Utrecht, the Netherlands
- 5Department of Paediatric Neurology, University Hospital, Groningen, the Netherlands
- 6Stichting Epilepsie Instellingen, Heemstede, the Netherlands
- Correspondence to: H Stroink Department of Neurology, St Elisabeth Hospital, PO Box 90151, 5000 LC Tilburg, The Netherlands; H.StroinkElisabeth.nl
- Received 31 March 2003
- Accepted 28 June 2003
- Revised 22 June 2003
Abstract
Objective: To assess the interrater agreement of the diagnosis and the classification of a first paroxysmal event in childhood.
Methods: The descriptions of 100 first paroxysmal events were submitted to two panels each consisting of three experienced paediatric neurologists. Each observer independently made a diagnosis based on clinical judgment and thereafter a diagnosis based on predefined descriptive criteria. Then, the observers discussed all patients within their panel. The agreement between the six individual observers was assessed before discussion within each panel and after that, between the two panels.
Results: Using their clinical judgement, the individual observers reached only fair to moderate agreement on the diagnosis of a first seizure (mean (SE) kappa 0.41 (0.03)). With use of defined descriptive criteria the mean (SE) kappa was 0.45 (0.03). The kappa for agreement between both panels after intra-panel discussion increased to 0.60 (0.06). The mean (SE) kappa for the seizure classification by individual observers was 0.46 (0.02) for clinical judgment and 0.57 (0.03) with use of criteria. After discussion within each panel the kappa between the panels was 0.69 (0.06). In 24 out of 51 children considered to have had a seizure, agreement was reached between the panels on a syndrome diagnosis. However, the epileptic syndromes were in most cases only broadly defined.
Conclusions: The interrater agreement on the diagnosis of a first seizure in childhood is just moderate. This phenomenon hampers the interpretation of studies on first seizures in which the diagnosis is only made by one observer. The use of a panel increased the interrater agreement considerably. This approach is recommended at least for research purposes. Classification into clinically relevant syndromes is possible only in a very small minority of children with a single seizure.
