Validating narrative data on residential child injury

Problem: Devising a systematic method for analyzing and disseminating narrative descriptions of residential child injuries in Hong Kong. Method: Narrative descriptions of the injury events, sampled from a university teaching hospital, were categorized by three variables related to a residential child injury event. Four raters coded the descriptions. Results were tested on multirater reliability. Results: Satisfactory multirater kappa in coding "child's action (CA)" and "object becoming hazard (OBH)" variables confirms stability within these categories. Low agreement in coding the "other human agent (OHA)" variable revealed the conceptual and technical complexity in the definition of appropriate child supervision. Impact on industry: This study presented a systematic method for the analysis and dissemination of narrative injury data on residential child injuries, offering empirically derived content for local injury prevention programs. Results from this study address the etiology of residential childhood injuries from a process analytic perspective and bring forth intervention that acknowledges the effect of a person's environment interaction.