Classification of events contributing to postneonatal cerebral palsy: Development, reliability, and recommendations for use.

Aim: This paper introduces the Surveillance of Cerebral Palsy in Europe (SCPE) classification of events contributing to postneonatally acquired cerebral palsy, presents its interrater reliability, and describes the cases identified in the SCPE database.

Method: The development of the classification, based on literature review and expert discussions, resulted in six main categories and 19 subcategories. The first chronological event designated as the primary event was mainly reported. Interrater reliability was assessed through online exercise providing 24 clinical vignettes representing single/complex pathways. Percent agreement and Gwet's AC1 index of reliability were estimated. Primary events were described using data of 221 children born between 2008 and 2012.

Results: Thirty-nine professionals (21 registries) participated in the reliability exercise. Substantial overall agreement was reached (0.75), with some contrast between complex (0.48, moderate agreement) and single events involved (0.89, almost perfect). The distribution of primary events showed that 32.1% were infections (category A), 23.1% head injuries (B), 15.4% related to surgery or medical interventions (C), 13.1% cerebrovascular accidents (D), 9.1% hypoxic brain damaging events of other origins (E), and 7.2% miscellaneous (F).

Interpretation: This classification allows all the events involved to be recorded while consistently reporting the primary event, and may be used in different settings.

What This Paper Adds: A standardized classification enables the description of the events contributing to postneonatal cerebral palsy (CP). The first chronological event in complex pathway leading to CP is coded. Category choice and coding of the primary event identify preventable situations. The detailed 2-level classification is easy to use in various settings. Substantial overall interrater reliability shows that main categories can be consistently differentiated.