Classification of maltreatment-related mortality by Child Death Review teams: How reliable are they?

Accurate estimation of the incidence of maltreatment-related child mortality depends on reliable child fatality review. We examined the inter-rater reliability of maltreatment designation for two Alaskan Child Death Review (CDR) panels. Two different multidisciplinary CDR panels each reviewed a series of 101 infant and child deaths (ages 0-4 years) in Alaska. Both panels independently reviewed identical medical, autopsy, law enforcement, child welfare, and administrative records for each death utilizing the same maltreatment criteria. Percent agreement for maltreatment was 64.7% with a weighted Kappa of 0.61 (95% CI 0.51, 0.70). Across maltreatment subtypes, agreement was highest for abuse (69.3%) and lowest for negligence (60.4%). Discordance was higher if the mother was unmarried or a smoker, if residence was rural, or if there was a family history of child protective services report(s). Incidence estimates did not depend on which panel's data were used. There is substantial room for improvement in the reliability of CDR panel assessment of maltreatment related mortality. Standardized decision guidance for CDR panels may improve the reliability of their data.