Accuracy of the national violent death reporting system in identifying unintentional firearm deaths to children by children.

Background: In assigning manner of death (MOD) for inclusion on death certificates, medical examiners and coroners do not always apply uniform criteria. Previous research indicates surveillance statistics based on death certificates, such as the National Vital Statistics System, grossly miscount unintentional firearm deaths. The National Violent Death Reporting System (NVDRS) has taken steps to reduce variability in manner of death coding by providing uniform criteria for assigning an "abstractor manner of death" (AMD). AMD has five categories: unintentional, suicide, homicide, undetermined, and legal intervention homicide. A previous study found good accuracy of AMD coding for unintentional firearm deaths, all ages, 2003-2006, but a more recent study reported that the NVDRS undercounted self- and other-inflicted unintentional firearm deaths in which both the victim and shooter (for other-inflicted injuries) were under age 15 (2009-2018).

Findings: We replicated the recent study's sample population, identifying 924 NVDRS incidents from 2009 to 2018 in which both victim and, for other-inflicted injuries, shooter age was under 15 and AMD was homicide, suicide, unintentional or undetermined (there were no legal intervention deaths to children). We assigned a researcher-adjudicated MOD (RMD) by reviewing incident narratives. RMD was compared with AMD and with manner recorded on the death certificate. Based on RMD as the gold standard, the sensitivity, specificity, and predictive values positive and negative of the AMD for unintentional childhood firearm deaths were, respectively, 90%, 99%, 98% and 96%; 86% (24/28) of false negatives were coded by abstractors as homicides. By contrast, death certificate manner had relatively poor sensitivity (63%).

Conclusions: In our sample of 924 deaths, the abstractor manner of death generally agreed with researcher-adjudicated manner of death, though not perfectly, missing 10% of researcher-adjudicated unintentional deaths, mostly because abstractors coded these unintentional deaths as homicides. A sizable minority of false negatives were unintentional deaths where the narrative explicitly noted that adult negligence contributed to a child's unintentional shooting death. While AMD coding in NVDRS is good, it could be improved if NVDRS coding guidelines explicitly affirmed that potential prosecution for negligent manslaughter is not a contraindication to an AMD of unintentional, provided the firearm was not used to intentionally harm, threaten, or coerce.