LAUSR

Background Paediatric head injury is common, yet 80% are mild and require no investigation or treatment. Decision-making around neuroimaging is guided by clinical decision rules. Whether parents and clinicians interpret questions within these algorithms in the same way is not known. Understanding this is fundamental for public facing algorithms and parental guidance. We aimed to determine the inter-observer reliability of Pediatric Emergency Care Applied Research Network (PECARN) head injury algorithm completion between clinicians and parents of head injured children. We conducted a prospective, observational, feasibility study. Children over 2 years old, presenting to the Royal Manchester Children’s Hospital within 24 hours of a head injury, were recruited. Children were excluded if they required immediate resuscitation or non-accidental injury was suspected. Parents were given a questionnaire to complete whilst awaiting assessment, with the PECARN criteria in lay-man’s language. Following assessment, the treating clinician completed a similar questionnaire, blinded to the parent’s answers. 57 children were included. The mean age was 5.8 years (SD 4.07). The most common mechanism of injury was a fall from standing (46%), with only 6 parents reporting a dangerous mechanism. Arrival Glasgow Coma Scale (GCS) was 15/15 in 98% of children. The inter-rater reliability for individual PECARN criteria was highly variable, with kappa values ranging from −0.10 to 1.00. There was poor agreement on whether the child was alert and acting normally (K −0.10, SE 0.49) but perfect agreement on other features of altered mental state including agitation, repeated questioning and taking longer than normal to respond (all K 1.0). The high variability in agreement between clinicians and parents when assessing PECARN criteria highlights differences in how these groups evaluate head injured children. This has significant implications for public facing algorithms, such as those used by NHS 111.