Confounder selection in firearm policy research: a scoping review

Legislative firearm policies are often proposed as a way of preventing firearm-related harm. Confounding is a substantial threat to accurately estimating the causal effects of firearm policies. This scoping review characterizes the selection of potential confounders in US firearm policy evaluations in the health sciences literature. We identified empirical research articles indexed in PubMed from January 1, 2000, to January 9, 2021, that examined any of 18 prespecified firearm policies and extracted key study elements, including the exposure (firearm policy), outcomes, potential confounders adjusted for in analyses, and study approach (ie, static, uncontrolled pre-post, and controlled pre-post). There was wide variation in potential confounders within study approach/policy outcome combinations. The most common potential confounders included sociodemographic and economic variables, rurality/urbanicity, violent crime, law enforcement-related variables, alcohol use, and firearm access (mostly measured via proxies for firearm ownership). Firearm policies other than the policy being evaluated were included in the adjustment set in 23% to 44% of studies, depending on the study approach. Confounder selection was most often said to be based on prior research (n = 49, 40%) or not explicitly stated (n = 48, 39%). This scoping review provides a comprehensive resource for critically appraising the firearm policy literature and offers considerations to support more rigorous confounding control in future firearm policy research.