LAUSR

It is easy to describe the scale of gun violence in the US with large numbers and hyperbolic comparisons; more civilians have died from firearm injuries since 1968 than all the soldiers who perished in all our wars since 1775.1,2 But gun violence is also the sum of a million heartbreaking stories—of preventable tragedies, young lives cut short, loved ones left behind. In a huge and diverse nation with more guns than people and a constitutionally protected individual right to bear arms, the complex variety of circumstances and factors driving gun violence poses a daunting challenge to policies and laws designed to prevent it. A new study by Veronica Pear and colleagues3 addresses the question of whether a promising legal tool—a civil restraining order to temporarily remove guns from people behaving dangerously—has had any detectable effect on the population rate of firearmrelated injury and mortality in a single jurisdiction, San Diego County, in the years since California’s gun violence restraining order (GVRO) law went into effect in 2016.4 These researchers’ answer to the GVRO-effectiveness question is essentially, “No, but...” Using a sophisticated statistical ecological analysis to compare trends in the rates of firearm injury and death in San Diego County and a synthetic control county, they found no statistically significant association with GVRO implementation. But a telling caveat appears near the end of the article: “Our results could reflect a true absence of association or limitations of our study; further research is needed to determine which of these is the case.” One can only hope this will deter gun industry lobbyists from spinning the study’s results as definitive evidence that “gun laws do not work.” Previous studies from Connecticut5 and Indiana6 that examined individual-level outcomes have reported that similar laws, generically called extreme risk protection orders (ERPOs), indeed saved lives by preventing suicides. What might be different about San Diego County? Or, what is different about this study in particular, and what lessons might it teach for future implementation of ERPO laws and the emerging body of research to evaluate their impact? We offer 3 broad considerations. First, we must acknowledge that GVROs represent an infrequently implemented intervention to prevent a rare outcome in the population. A challenge for a risk-based firearm policy like the GVRO is that gun injuries and deaths are caused by a relatively small number of people who are not easily distinguishable from others in a large pool of putatively dangerous individuals. A California court may issue a GVRO if it finds a “substantial likelihood” that a person “poses a significant danger in the near future of causing personal injury” to self or others by having access to a firearm.7 But the indicia of likelihood that the court may consider include nonspecific risk factors, such as communicating a suicide threat.4,8 In relative terms, people who disclose that they have a plan for suicide are statistically more likely than otherwise to die from suicide, but at the same time the large majority of them—about 98%—do not, in fact, go on to die from suicide.9 For these reasons, the GVRO legal criteria for gun removal will unavoidably apply to many more people who would not actually harm themselves or others than to those who would. Accordingly, the court must carefully balance risk and rights to ensure that the GVRO policy is targeted effectively, but not too broadly. But this is why a substantial number of GVROs would be needed in order to see a meaningful effect on reduced gun violence at the population level. And herein lies a key limitation of the study by Pear and colleagues: the number of GVROs issued during the observation period was probably too small a fraction of the population at risk to expect to see a meaningful change in the + Related article