Citation

Kalesan B, Galea S, Fagan JA. Am. J. Epidemiol. 2017; 186(7): 897-898.

Copyright

(Copyright © 2017, Oxford University Press)

DOI

10.1093/aje/kwx280

PMID

unavailable

Abstract

We thank Cook et al. for their interest in our paper, "The Hidden Epidemic of Firearm Injury: Increasing Firearm Injury Rates During 2001-2013" (1). They suggested that we have overstated the increase in nonfatal firearm assault injuries (2) based on limitations in the publicly available data from the Nonfatal Injury Reports in the Centers for Disease Control and Prevention's Web-Based Injury Statistics Query and Reporting System (WISQARS). In Web Appendix 1 of our article (1), we provided details regarding the data from the National Electronic Injury Surveillance System-All Injury Program (NEISS-AIP) that we used. These data were discussed by Cook et al. in detail in their recent AJPH paper "Constant Lethality of Gunshot Injuries From Firearm Assault: United States, 2003-2012" (3), in which they presented their core concern.

We agree with Cook et al. that the NEISS-AIP data should be viewed with caution (summarized in our paper's limitation section) while questioning whether the flaws in those data undermine our results. Cook et al. suggested that the sample of hospitals included in NEISS-AIP, the foundation for WISQARS, is not representative of the country and that inferences from these data are flawed (3). A full reckoning with their points rests on the logic behind their challenge to the NEISS-AIP data set. Two points are worth highlighting.

First, Cook et al. assumed that the distribution of unknowns follows the distribution of knowns (3). We wonder whether this assumption introduces more questions than it answers. Missingness at a high rate (40% in these data) suggests that the evidence in the "unknowns" is dependent on the parameters of the unobserved values. This is a testable hypothesis, and there are empirical solutions, such as...

Language: en