Keeping children safe: a model for predicting families at risk for recurrent childhood injuries

Objective: Existing research on recurrent unintentional injury (UI) focuses on the individual child rather than family risks. This study developed a statistical model for identifying families at highest risk, for potential use in targeting public health interventions. Study design: A retrospective birth cohort study of hospital and emergency room (ER) medical records of children born in Ziv hospital between 2005 and 2012, attending ER for UI between 2005 and 2015, was conducted. Methods: Using national IDs, we assigned children to mothers and created the family entity. Data were divided into two time periods. Negative binomial regression was used to examine predictive factors in the first period for recurrent child UI in the second period. Sensitivity analyses were conducted to examine the model's robustness. Results: Eight predictive factors for child injury (P < 0.05) were found: male gender, the number of UI visits, the number of illness visits, age 36–59 months, birth weight <1500 g, maternal ER visits, siblings' UI visits, and the number of younger siblings. Some predictive factors are documented in the literature; others are novel. Five were significant in all sensitivity analyses. Conclusions: These factors can assist in predicting risk for a child's repeat UI and family's cumulative UI risk. The model may offer a valuable and novel approach to targeting interventions for families at highest risk.