Skip Navigation
American College of Physicians Logo
  • Subscribe
  • Submit a Manuscript
  • Sign In
    Sign in below to access your subscription for full content
    INDIVIDUAL SIGN IN
    Sign In|Set Up Account
    You will be directed to acponline.org to register and create your Annals account
    INSTITUTIONAL SIGN IN
    Open Athens|Shibboleth|Log In
    Annals of Internal Medicine
    SUBSCRIBE
    Subscribe to Annals of Internal Medicine.
    You will be directed to acponline.org to complete your purchase.
Annals of Internal Medicine Logo Menu
  • Latest
  • Issues
  • Channels
  • CME/MOC
  • In the Clinic
  • Journal Club
  • Web Exclusives
  • Author Info
Advanced Search
  • ‹ PREV ARTICLE
  • This Issue
  • NEXT ARTICLE ›
Original Research |19 December 2017

In-State and Interstate Associations Between Gun Shows and Firearm Deaths and Injuries: A Quasi-experimental Study Free

Ellicott C. Matthay, MPH; Jessica Galin, MPH; Kara E. Rudolph, PhD, MPH, MHS; Kriszta Farkas, MPH; Garen J. Wintemute, MD, MPH; Jennifer Ahern, PhD, MPH

Ellicott C. Matthay, MPH
From University of California, Berkeley, School of Public Health, Berkeley, and University of California, Davis, Sacramento, California.

Jessica Galin, MPH
From University of California, Berkeley, School of Public Health, Berkeley, and University of California, Davis, Sacramento, California.

Kara E. Rudolph, PhD, MPH, MHS
From University of California, Berkeley, School of Public Health, Berkeley, and University of California, Davis, Sacramento, California.

Kriszta Farkas, MPH
From University of California, Berkeley, School of Public Health, Berkeley, and University of California, Davis, Sacramento, California.

Garen J. Wintemute, MD, MPH
From University of California, Berkeley, School of Public Health, Berkeley, and University of California, Davis, Sacramento, California.

Jennifer Ahern, PhD, MPH
From University of California, Berkeley, School of Public Health, Berkeley, and University of California, Davis, Sacramento, California.

Article, Author, and Disclosure Information
Author, Article, and Disclosure Information
This article was published at Annals.org on 24 October 2017.
  • From University of California, Berkeley, School of Public Health, Berkeley, and University of California, Davis, Sacramento, California.

    Disclaimer: The analyses, interpretations, and conclusions of this report are attributable to the authors and not to the California Department of Public Health or National Institutes of Health.

    Acknowledgment: The authors thank the following funding sources: the Eunice Kennedy Shriver National Institute of Child Health and Human Development; the National Institutes of Health Office of the Director; the University of California, Berkeley Committee on Research; and the Heising-Simons Foundation.

    Grant Support: By grant DP2HD080350 from the Eunice Kennedy Shriver National Institute of Child Health and Human Development and the National Institutes of Health Office of the Director (Ms. Matthay, Ms. Galin, Ms. Farkas, and Dr. Ahern); the University of California, Berkeley Committee on Research (Ms. Matthay, Ms. Galin, Ms. Farkas, and Dr. Ahern); and grant 2016-219 from the Heising-Simons Foundation (Dr. Wintemute).

    Disclosures: Ms. Matthay, Ms. Galin, Ms. Farkas, and Dr. Ahern report grants from the National Institutes of Health and University of California, Berkeley, during the conduct of the study. Dr. Wintemute reports grants from Heising-Simons Foundation during the conduct of the study. Disclosures can also be viewed at www.acponline.org/authors/icmje/ConflictOfInterestForms.do?msNum=M17-1792.

    Editors' Disclosures: Christine Laine, MD, MPH, Editor in Chief, reports that she has no financial relationships or interests to disclose. Darren B. Taichman, MD, PhD, Executive Deputy Editor, reports that he has no financial relationships or interests to disclose. Cynthia D. Mulrow, MD, MSc, Senior Deputy Editor, reports that she has no relationships or interests to disclose. Deborah Cotton, MD, MPH, Deputy Editor, reports that she has no financial relationships or interest to disclose. Jaya K. Rao, MD, MHS, Deputy Editor, reports that she has stock holdings/options in Eli Lilly and Pfizer. Sankey V. Williams, MD, Deputy Editor, reports that he has no financial relationships or interests to disclose. Catharine B. Stack, PhD, MS, Deputy Editor for Statistics, reports that she has stock holdings in Pfizer and Johnson & Johnson.

    Reproducible Research Statement:Study protocol: Not applicable. Statistical code: Available from Ellicott Matthay (e-mail, ematthay@berkeley.edu). Data set: Death and hospital visit data are available from the California Department of Public Health Vital Records and the Office of Statewide Health Planning and Development; data on dates and locations of gun shows are available from Ellicott Matthay (e-mail, ematthay@berkeley.edu).

    Requests for Single Reprints: Ellicott C. Matthay, MPH, 50 University Hall, University of California, Berkeley, Berkeley, CA 94704; e-mail, ematthay@berkeley.edu.

    Current Author Addresses: Ms. Matthay, Ms. Galin, Drs. Rudolph and Ahern, and Ms. Farkas: 50 University Hall, University of California, Berkeley, Berkeley, CA 94704.

    Dr. Wintemute: Violence Prevention Research Program, UC Davis Medical Center, 2315 Stockton Boulevard, Sacramento, CA 95817.

    Author Contributions: Conception and design: E.C. Matthay, J. Galin, K.E. Rudolph, G. Wintemute, J. Ahern.

    Analysis and interpretation of the data: E.C. Matthay, J. Galin, K.E. Rudolph, G. Wintemute, J. Ahern.

    Drafting of the article: E.C. Matthay, J. Galin, J. Ahern.

    Critical revision of the article for important intellectual content: E.C. Matthay, K.E. Rudolph, K. Farkas, G. Wintemute, J. Ahern.

    Final approval of the article: E.C. Matthay, J. Galin, K.E. Rudolph, K. Farkas, G. Wintemute, J. Ahern.

    Provision of study materials or patients: G. Wintemute.

    Statistical expertise: E.C. Matthay, K.E. Rudolph.

    Obtaining of funding: G. Wintemute, J. Ahern.

    Administrative, technical, or logistic support: E.C. Matthay, J. Galin.

    Collection and assembly of data: E.C. Matthay, J. Galin, K. Farkas, G. Wintemute, J. Ahern.

×
  • ‹ PREV ARTICLE
  • This Issue
  • NEXT ARTICLE ›
Jump To
  • Full Article
  • FULL ARTICLE
  • FULL ARTICLE
    • Abstract
    • Methods
    • Results
    • Discussion
    • Appendix: In-State and Interstate Associations Between Gun Shows and Firearm Deaths and Injuries: Supplemental Information, Methods, and Results
      1. References
  • Figures
  • Tables
  • Supplements
  • Audio/Video
  • Summary for Patients
  • Clinical Slide Sets
  • CME / MOC
  • Comments
  • Twitter Link
  • Facebook Link
  • Email Link
More
  • LinkedIn Link

Abstract

Background:

Gun shows are an important source of firearms, but no adequately powered studies have examined whether they are associated with increases in firearm injuries.

Objective:

To determine whether gun shows are associated with short-term increases in local firearm injuries and whether this association differs by the state in which the gun show is held.

Design:

Quasi-experimental.

Setting:

California.

Participants:

Persons in California within driving distance of gun shows.

Measurements:

Gun shows in California and Nevada between 2005 and 2013 (n = 915 shows) and rates of firearm-related deaths, emergency department visits, and inpatient hospitalizations in California.

Results:

Compared with the 2 weeks before, postshow firearm injury rates remained stable in regions near California gun shows but increased from 0.67 injuries (95% CI, 0.55 to 0.80 injuries) to 1.14 injuries (CI, 0.97 to 1.30 injuries) per 100 000 persons in regions near Nevada shows. After adjustment for seasonality and clustering, California shows were not associated with increases in local firearm injuries (rate ratio [RR], 0.99 [CI, 0.97 to 1.02]) but Nevada shows were associated with increased injuries in California (RR, 1.69 [CI, 1.16 to 2.45]). The pre–post difference was significantly higher for Nevada shows than California shows (ratio of RRs, 1.70 [CI, 1.17 to 2.47]). The Nevada association was driven by significant increases in firearm injuries from interpersonal violence (RR, 2.23 [CI, 1.01 to 4.89]) but corresponded to a small increase in absolute numbers. Nonfirearm injuries served as a negative control and were not associated with California or Nevada gun shows. Results were robust to sensitivity analyses.

Limitation:

Firearm injuries were examined only in California, and gun show occurrence was not randomized.

Conclusion:

Gun shows in Nevada, but not California, were associated with local, short-term increases in firearm injuries in California. Differing associations for California versus Nevada gun shows may be due to California's stricter firearm regulations.

Primary Funding Source:

National Institutes of Health; University of California, Berkeley; and Heising-Simons Foundation.

Firearms are a leading cause of morbidity and mortality in the United States and accounted for more than 36 000 deaths and nearly 85 000 injuries in 2015 (1). Ownership increases the risk for suicide, homicide, and unintentional firearm death and injury in the home (2–8). Greater availability and ownership of firearms also contributes to the higher rate of firearm deaths and injuries (hereafter called “firearm injuries”) in the United States than in other high-income countries (9–12). Gun shows account for 4% to 9% of annual firearm sales (13–15) and 3% of gun owners' most recent gun acquisitions (16). However, many of these transfers do not involve a background check (16) and firearms from gun shows are disproportionately implicated in crimes (17, 18). Little is known about how gun shows contribute to firearm injuries in the United States.
More than 4000 gun shows are held annually in the United States (19). These shows, which can attract thousands of attendees and hundreds of sellers, generate a temporary and diverse source of new and used firearms, ammunition, and related equipment in a competitive market where sales may be subject to less oversight (15, 20). Consequently, gun shows may increase local firearm ownership and use and affect subsequent rates of firearm injury. State regulations also differ markedly, which may modify the association between gun shows and firearm injuries. In particular, interstate activity and the flow of firearms from less to more restrictive states have been documented previously (21), and this pattern, which may limit the effectiveness of regulations in states that have them, may also extend to gun shows.
Using a quasi-experimental, difference-in-differences design, we exploited the natural variation in the timing and location of gun shows to investigate whether they are associated with increased injury rates and whether this association varies by the state in which the gun show is held. California has some of the most restrictive firearm laws in the country, including a comprehensive set of statutes regulating gun shows (22, 23). In contrast, Nevada has some of the least restrictive laws in the country and no explicit regulations on gun shows (22). Thus, comparing pre–post differences in California and cross-border differences between California and Nevada gun shows may provide useful information on these different policy environments.
No formal evaluations have assessed the effects of policies regulating gun shows. Observational evidence from 5 states suggests that such activities as anonymous and undocumented sales are less frequent in California, where they are prohibited, than in states where they are legal (24). Previous evidence has also linked firearms purchased at gun shows to crimes (17, 18), but to our knowledge, only 1 study has examined the association between gun shows and subsequent firearm injuries. Duggan and colleagues (25) examined weekly violent firearm deaths in ZIP codes in the immediate vicinity of gun shows in California and Texas. They found no association and suggested that California's gun show regulations have no effect on violent firearm deaths. However, the study was criticized as having low statistical power, incomplete identification of gun shows, and an analytic approach ignoring California's requirement that buyers wait 10 days between purchasing and obtaining a firearm (26, 27).
We addressed these gaps while assessing whether firearm injuries increase in nearby California areas immediately after gun shows in California and Nevada. We hypothesized that gun shows lead to increased rates of firearm injury.

Methods

Overall Approach

We used a quasi-experimental, difference-in-differences design (28, 29). First, we compared firearm injury rates for the 2 weeks immediately before and after each gun show in California regions within convenient traveling distance of the show. Then, we compared this difference for the California populations exposed to California versus Nevada gun shows. This approach is advantageous because each region's characteristics, other than the occurrence of a show, are unlikely to change appreciably over so short a time. Thus, each region serves as its own control, allowing us to adjust for community-level characteristics that may be associated with firearm injuries.

Firearm Injuries

We identified fatal and nonfatal firearm injuries in California between 2005 and 2013 using death records from the California Department of Public Health Vital Records and emergency department and inpatient hospitalization records from the Office of Statewide Health Planning and Development. External cause-of-injury coding in California's hospital discharge records is mandatory, subject to ongoing quality assurance measures, and considered 100% complete (30). Emergency department records from before 2005 are not available.

Gun Show Data

We compiled dates and locations of gun shows in California and Nevada between 2005 and 2013 using published lists in the Big Show Journal. This source was the most comprehensive; other magazines (Gun and Knife Show Calendar and Gun List Magazine) and online sources (we considered 11 major Web sites) did not cover the entire study period or included fewer listings (95% vs. 65% coverage). We used ABBYY FineReader 12 character recognition software to convert scanned images of show listings to electronic alphanumeric data (31).

Database Construction

Regions considered local to gun shows were determined using the Google Maps Distance Matrix API (32) by measuring the typical driving time between each ZIP code centroid in California and each geocoded gun show location. Little evidence exists on how far or how long the effects of gun shows extend (27). Thus, we selected reasonable time frames and travel distances to balance capturing short-term effects with estimating stable rates and to include regions likely to be affected by gun shows while excluding regions so distant that unrelated firearm injuries might obscure potential relationships. We tested the sensitivity of our results to chosen time frames and travel times in several sensitivity analyses. “Before” periods were the 14 days before each show; “after” periods were the 14 days after the 10-day waiting period from the start of the show for California or after the start of the show for Nevada, which has no waiting period. ZIP code centroids within a 60- or 120-minute drive were considered to be within traveling distance of California and Nevada shows, respectively. In California, most persons can access a gun show within a 60-minute drive every few weeks, but we hypothesized that some persons in California would be willing to travel farther to Nevada's comparatively unregulated environment.
ZIP codes were occasionally local to several gun shows at the same time. This was problematic when the “before” period of a later gun show (show B) overlapped with the “after” period of an earlier show (show A). Without consideration of this overlap, the ZIP code would be misclassified as “unexposed” for examination of show B when it was “exposed” to show A. In these cases, we excluded the overlapping ZIP code from analyses of show B (hereafter “overlap exclusions”). Restricting to ZIP codes far enough from the border to eliminate the need for overlap exclusions did not alter the findings (results available on request). Throughout, rates are reported per 100 000 persons in regions within traveling distance of shows.

Statistical Analysis

We did a difference-in-differences analysis (28, 29) at the gun show–period level using multivariable Poisson mixed-effects regression. The main outcome measure was the rate of firearm injury. The full model specification was as follows:
graphic
View largeDownload slide
graphic
View largeDownload slide
where ytks was the count of firearm injuries at time t in the region surrounding gun show s in state k; β0, the intercept; Xt, the period (after vs. before); Xk, the state of the show (Nevada vs. California); Xm, month indicators to account for seasonality; ρcs and ρc, random-effects intercepts to account for clustering by gun shows nested within cities; log(dtsk), an offset for the number of at-risk persons; and ϵtsk, the error term. Statistical testing of the dispersion parameter indicated that a Poisson model was more appropriate than a negative binomial model. Under this specification, exp(β1) estimates the rate ratio (RR) associated with gun shows in California, exp(β1 + β3) estimates the RR associated with gun shows in Nevada, and exp(β3) estimates the difference-in-differences estimate—the ratio of RRs—capturing the increase in firearm injury rates after Nevada shows compared with that after California shows.
P values less than 0.05 were considered statistically significant. Data were processed by using SAS, version 9.3 (SAS Institute), and R, version 3.2.1 (R Foundation), and regression analysis was done using the lme4 package (33) in R, version 3.2.1. This study was approved by the State of California and University of California, Berkeley, Committees for the Protection of Human Subjects.

Subgroup and Secondary Analyses

To examine variation by firearm injury type, we did subgroup analyses for intentional interpersonal violence, intentional self-harm, unintentional injuries, and injuries of undetermined intent (Appendix Table 1). Because the exposure periods and geographic regions defined for California and Nevada shows were not identical (with vs. without a waiting period; a 60- vs. a 120-minute drive), we also stratified the analysis by state. In addition, we did analyses restricted to specific gun shows and affected regions to examine potential associations along known firearm trafficking routes between Reno and San Francisco and between Las Vegas and Los Angeles. We also tested the association between California gun shows and California firearm injuries ignoring the 10-day waiting period, because activities other than legal firearm purchases (such as ammunition or parts purchases, illegal purchases, and repairs) may affect firearm injuries and do not have a waiting period. We tested the association between gun shows and nonfirearm injuries as a negative control to assess whether common causes of firearm and nonfirearm injuries confounded our findings (34).

Appendix Table 1. ICD-9 and ICD-10 External Cause-of-Injury Codes Used to Identify and Classify Firearm Deaths and Injuries

Appendix Table 1. ICD-9 and ICD-10 External Cause-of-Injury Codes Used to Identify and Classify Firearm Deaths and Injuries
Finally, differences in associations between California and Nevada gun shows may be due to differing characteristics of regions exposed to California shows versus those exposed to Nevada shows. To address this potential source of variation, we restricted the entire analysis to regions similar to those exposed to Nevada gun shows. We tightly matched on ZIP code characteristics that differed between regions exposed to California versus Nevada shows and may modify the association between gun shows and firearm injuries (35, 36). These were population density, percentage of veterans, median income, median age, percentage of white non-Hispanic persons, hunting licenses per capita, and the overall rate of firearm injury between 2005 and 2013. We tested a range of matching approaches, all of which produced similar matches. Further details on the matching approach and characteristics of this restricted analysis are in the  Appendix.

Power Calculations

To confirm that our study had sufficient statistical power, we did a power analysis using simulated data that were generated to be similar to the observed data (Appendix Figure 1) (37). We applied the main analysis regression approach to each simulated data set and recorded the proportion of simulations with a significant association. This analysis indicated that our study had 87.8% power to detect increases in firearm injuries as large as or larger than those seen for Nevada shows and 84.2% power to detect the difference between California and Nevada gun shows.
Appendix Figure 1.

Comparison of observed and simulated distribution of firearm deaths and injuries during the 2 weeks after gun shows in nearby regions.

The unit of analysis is the gun show. The figure presents the distribution of the observed and simulated number of firearm deaths and injuries, per gun show, ≤2 wk after each gun show, in regions within driving distance of each show, by state. For example, in both the observed and simulated data, just fewer than 300 gun show regions had no firearm deaths or injuries ≤2 wk after the show.

Bias Analysis

To assess the potential role of residual confounding due to unmeasured factors, we did a quantitative bias analysis. We estimated the characteristics of an unmeasured confounder that would yield the observed associations between gun shows and firearm injuries, if the true effect were not statistically significant.

Role of the Funding Source

This research was funded by the Eunice Kennedy Shriver National Institute of Child Health and Human Development; the National Institutes of Health Office of the Director; the University of California, Berkeley, Committee on Research; and the Heising-Simons Foundation. The funding sources had no role in study design, conduct, data collection, data analysis, preparation of the manuscript, or the decision to submit the manuscript for publication.

Results

We identified 640 gun shows in California and 275 in Nevada between 1 January 2005 and 31 December 2013 (Table 1). Shows were held on weekends, usually at convention centers or county fairgrounds, and lasted 2 to 3 days. Some shows returned to the same locations at regular intervals, whereas others were held at irregular times and locations. Overlap exclusions were more common for Nevada shows than for California shows because those in Nevada were more frequent and held in fewer locations. Appendix Figure 2 is a map of gun show locations. Table 1 provides characteristics of California and Nevada gun shows and total person-weeks of exposure to gun shows.

Table 1. Characteristics of California and Nevada Gun Shows and Population Exposure to Gun Shows

Table 1. Characteristics of California and Nevada Gun Shows and Population Exposure to Gun Shows
Appendix Figure 2.

Locations of gun shows in California and Nevada.

Table 2 presents the number and rate of local firearm injuries before and after California and Nevada gun shows. In the 2 weeks preceding California shows, 15 000 firearm injuries occurred in at-risk regions, but before–after rates remained stable for California shows. For Nevada shows, only 44 firearm injuries occurred during the preshow period (Table 2). However, firearm injury rates increased from 0.67 injuries (95% CI, 0.55 to 0.80 injuries) per 100 000 persons to 1.14 injuries (CI, 0.97 to 1.30 injuries) per 100 000 persons in California regions exposed to Nevada shows.

Table 2. Unadjusted Analyses of the Association Between Firearm Deaths and Injuries in California and Gun Shows in California and Nevada

Table 2. Unadjusted Analyses of the Association Between Firearm Deaths and Injuries in California and Gun Shows in California and Nevada
After adjustment, California shows were not associated with increases in firearm injuries (RR, 0.99 [CI, 0.97 to 1.02]), but Nevada shows were associated with significant cross-border increases in firearm injuries in California (RR, 1.69 [CI, 1.16 to 2.45]) (Table 3). The difference between states was significant: Gun shows in Nevada were associated with a 70% greater increase in firearm injuries than those in California (ratio of RRs, 1.70 [CI, 1.17 to 2.47]). This association corresponds to a rate difference of 0.46 (CI, 0.36 to 0.57) per 100 000 persons, or a 0.3-SD increase relative to the biweekly variability in rates across locations. In terms of cases, the association corresponds to 30 additional injuries in regions exposed to the 161 Nevada shows.

Table 3. Adjusted Analyses of the Association Between Firearm Deaths and Injuries in California and Gun Shows in California Versus Nevada

Table 3. Adjusted Analyses of the Association Between Firearm Deaths and Injuries in California and Gun Shows in California Versus Nevada
The association with Nevada shows was driven by significant increases in firearm injuries from interpersonal violence (RR, 2.23 [CI, 1.01 to 4.89]). Results for analyses stratified by gun show state (Table 4) or restricted to regions similar to those exposed to Nevada shows (Table 5) were consistent with those from the main analysis. No significant relationships existed between gun shows and firearm injuries along known trafficking routes or when excluding California's 10-day waiting period (Appendix Table 2).

Table 4. Adjusted Analyses of the Association Between Firearm Deaths and Injuries in California and Gun Shows in California and Nevada, by State

Table 4. Adjusted Analyses of the Association Between Firearm Deaths and Injuries in California and Gun Shows in California and Nevada, by State

Table 5. Adjusted Analyses of the Association Between Firearm Deaths and Injuries in California and Gun Shows in California Versus Nevada, Restricted to Regions Similar to Those Exposed to Nevada Gun Shows

Table 5. Adjusted Analyses of the Association Between Firearm Deaths and Injuries in California and Gun Shows in California Versus Nevada, Restricted to Regions Similar to Those Exposed to Nevada Gun Shows

Appendix Table 2. Secondary Analyses for the Adjusted Association Between Firearm Deaths and Injuries Along Firearm Trafficking Routes and Excluding California's 10-Day Waiting Period

Appendix Table 2. Secondary Analyses for the Adjusted Association Between Firearm Deaths and Injuries Along Firearm Trafficking Routes and Excluding California's 10-Day Waiting Period
In sensitivity analyses of geographic range and duration of exposure (Appendix Table 3), associations between California shows and firearm injuries were consistently null. For Nevada gun shows, changes in firearm injuries remained statistically significant for shorter (1-week) and longer (3-week) periods but were not statistically significant for smaller geographic ranges (60-minute drive), which yielded very few cases, or larger geographic ranges (120- and 180-minute drives for California and Nevada guns shows, respectively), which covered large portions of California. Nevada shows were significantly associated with increases in self-directed intentional firearm injuries when longer periods were examined.

Appendix Table 3. Sensitivity Analyses for the Adjusted Association Between Firearm Deaths and Injuries in California and Gun Shows in California Versus Nevada

Appendix Table 3. Sensitivity Analyses for the Adjusted Association Between Firearm Deaths and Injuries in California and Gun Shows in California Versus Nevada
The  Appendix provides bias and negative control analyses. In brief, for the association between Nevada gun shows and California firearm injuries to be spurious, another factor would have to match the precise geographic and temporal pattern of the 275 Nevada gun shows and also be strongly associated with firearm injuries in California, corresponding to RRs of at least 1.5 or 2. This factor would also have to be up to 80% more prevalent in the 2 weeks after Nevada gun shows than in the 2 weeks before. Bias analysis results were similar for the difference-in-differences estimate comparing Nevada with California. Associations between both California and Nevada gun shows and nonfirearm injuries were null, or were statistically significant because of the large number of nonfirearm unintentional injuries (n = 6 065 633) but not meaningfully different from the null.

Discussion

We examined the association between California and Nevada gun shows and short-term changes in local firearm injuries in California. Using a quasi-experimental, difference-in-differences design, we took advantage of natural variation in the timing and location of gun shows and differences between California and Nevada firearm regulations to compare this association by state. Firearm injuries in California remained stable after California gun shows but increased by a small but significant amount after Nevada shows.
Several factors could explain our findings. First, although we did not formally assess the effect or enforcement of firearm policies in either state, the absence of an increase in firearm injuries after California gun shows may be evidence that California's strict regulatory environment, both gun show–related and otherwise, mitigates potential risk from gun shows through deterrence. Among other restrictions, California requires that all private transfers be documented by a licensed dealer and include a background check (22). It also enforces restrictions on gun shows (23), including a range of security- and enforcement-related planning and reporting practices, that may deter the illegal firearm activity historically seen at gun shows (15, 17, 18, 20). Specialized firearm enforcement agents from the California Department of Justice also do surveillance at California gun shows. In contrast, Nevada does not require background checks or documentation for private transfers and places no regulations on gun shows. Thus, California's measures may prevent illegal activities that could lead to increases in interpersonal firearm injuries.
A second possibility is that California's regulations and 10-day waiting period motivate buyers to cross into Nevada when seeking a faster, less-regulated source of firearms. This mechanism, which suggests displacement rather than deterrence, would imply that even if California's regulations are mitigating risk from gun shows within its borders, travel to less-restrictive states may threaten the effectiveness of California's laws. Indeed, interstate gun trafficking, including that between Nevada and California, is well-documented and fueled in part by gun shows (18–21, 38, 39).
A third possibility is that gun shows affect Californians near Nevada differently from Californians in the rest of the state. However, analyses restricted to regions similar to those along the California–Nevada border produced results consistent with the main analysis, suggesting that the characteristics of border communities are not major drivers of the observed differences.
Gun show occurrence was not randomized. Thus, a fourth possibility is that the observed association is due to uncontrolled confounders. However, gun shows were not associated with nonfirearm injuries, providing evidence that the results are not due to confounding by common causes of firearm and nonfirearm injuries (34). Furthermore, the quantitative bias analysis indicated that for the observed associations to be spurious, at least 1 factor would have to match the geographic and temporal pattern of the gun shows, be strongly associated with firearm injuries, be unevenly distributed between California and Nevada, and change markedly in prevalence in the 2 weeks after gun shows compared with the 2 weeks before. Identifying a factor that fits these criteria is challenging, which strengthens confidence in our results. Similar bias analyses have been used to bolster evidence of the association between firearm ownership and suicide (40).
Our null findings for California gun shows are consistent with those of Duggan and colleagues (25). However, our study was the first to our knowledge to assess interstate associations and suggests that travel across state lines may be important. Our study avoided several limitations highlighted in previous critiques of Duggan and colleagues' study (26, 27) by being well-powered statistically, analyzing data from the show-period level rather than the ZIP code–week level, and accounting for California's 10-day waiting period. Our approach was also strengthened by inclusion of nonfatal injuries and unintentional and intent-undetermined firearm injuries, rather than only firearm suicides and homicides. In addition, we examined geographic areas defined by driving distances and incorporated overlap exclusions for regions simultaneously exposed to 1 show and unexposed to another.
This study had limitations. First, all nonexperimental studies are subject to residual confounding. We minimized the effect of potential confounders by comparing identical regions over brief periods, during which factors other than gun shows are unlikely to vary; we also did negative control and quantitative bias analyses to assess the sensitivity of our results to an unobserved confounder. Second, few firearm injuries occurred in regions exposed to Nevada gun shows; however, rates for this region were derived from 13 037 052 person-weeks of exposure (Table 1). Third, cause-of-death and cause-of-injury classification on death and discharge records is imperfect, although studies suggest that the degree of misclassification is not substantial enough to alter major trends and patterns (41, 42). Fourth, we did not examine associations with firearm injuries in Nevada populations. Future research on the effects of gun shows in Nevada and other states would be valuable. Fifth, data on nonfatal injuries include most hospital visits for firearm injuries but do not include military hospitals. Lastly, evidence suggests that firearms purchased at gun shows and recovered from crime scenes are rarely found in the immediate region or period after shows (27). However, these patterns do not preclude the possibility of a proximate effect, particularly because first use of a gun may predate its recovery from a crime scene.
In conclusion, gun shows are an important source of firearms and may offer an opportunity for regulatory intervention. Results from this study suggest that California gun shows are not associated with short-term increases in firearm injuries but that Nevada shows are associated with cross-border increases in firearm injuries in California. Differences in regulations may explain this pattern, but alternative explanations exist, and the short-term increase in firearm injuries attributable to gun shows is small relative to the number of firearm injuries in places exposed to gun shows. Better understanding the long-term effects of gun shows over larger geographic regions, the effects of gun show policies, and the patterns of acquisition and use of firearms would provide important evidence to inform future efforts to prevent firearm injuries.

Appendix: In-State and Interstate Associations Between Gun Shows and Firearm Deaths and Injuries: Supplemental Information, Methods, and Results

Firearm Injury Classification Codes

We identified fatal and nonfatal firearm injuries in California between 2005 and 2013 using International Classification of Diseases, 9th and 10th Revisions, external cause-of-injury codes contained in death records from the California Department of Public Health Vital Records and in emergency department and inpatient hospitalization records from the Office of Statewide Health Planning and Development. To examine variation by firearm injury type, we did subgroup analyses for intentional interpersonal violence, intentional self-harm, unintentional injuries, and injuries of undetermined intent. International Classification of Diseases codes used in this analysis are presented by subgroup in Appendix Table 1.

Power Calculations

To confirm that our study had sufficient statistical power, we did a power analysis using simulated data that were generated to be similar to the observed data. We used the observed number of firearm deaths and injuries for the 2 weeks before each gun show (in regions within convenient traveling distance of each show) and simulated the number of firearm deaths and injuries in the 2 weeks after each show. Appendix Figure 1 presents the observed and simulated distributions of firearm deaths and injuries for the 2 weeks after gun shows. The distributions are very similar, suggesting that the power analysis is based on simulated data that accurately reflect the observed data.
We applied the main analysis regression approach to each simulated data set and recorded the proportion of simulations with a significant association. This analysis indicated that our study had 87.8% power to detect increases in firearm deaths and injuries as large as or larger than those seen for Nevada shows and 84.2% power to detect the difference between California and Nevada gun shows.

Gun Show Locations

Appendix Figure 2 presents the locations of all gun shows identified in California and Nevada between 2005 and 2013. Locations included in each analysis depend on the analysis specification (geographic range [driving distance] and duration of preexposure and postexposure periods). In particular, some shows in northeastern Nevada were not included in analyses restricted to Nevada shows within a 120-minute drive of California but were included in sensitivity analyses extending to longer driving distances.

Secondary and Sensitivity Analyses

Little evidence exists on how far or how long the effects of gun shows extend (27). Thus, we selected reasonable time frames and travel times to balance capturing short-term effects with estimating stable rates and to include regions likely to be affected by gun shows while excluding regions so distant that unrelated firearm injuries might obscure potential relationships. We then tested the sensitivity of our results to chosen time frames and travel times.
Appendix Table 3 presents the results of these sensitivity analyses. Associations between California shows and firearm deaths and injuries were consistently null. For Nevada gun shows, changes in firearm injuries remained statistically significant for shorter (1-week) and longer (3-week) periods but were not statistically significant for smaller geographic ranges (60-minute drive), which yielded very few cases, and larger geographic ranges (120- and 180-minute drives for California and Nevada guns shows, respectively), which covered large portions of California. Nevada shows were significantly associated with increases in self-directed intentional firearm injuries when examining longer periods.
We restricted additional secondary analyses to specific gun shows and affected regions to examine potential associations along known firearm trafficking routes between Reno and San Francisco and between Las Vegas and Los Angeles. We also tested the association between California gun shows and California firearm injuries ignoring the 10-day waiting period, because activities other than legal firearm purchases, such as ammunition or parts purchases, illegal purchases, and repairs, may affect firearm injuries and do not have a waiting period.
Appendix Table 2 presents the results of these secondary analyses. No significant relationships existed between gun shows and firearm injuries along known trafficking routes or when California's 10-day waiting period was excluded.

Sensitivity Analysis Restricted to Regions Similar to Those Exposed to Nevada Gun Shows

One important consideration in interpreting the results of the main analysis is that characteristics of the regions exposed to gun shows may modify the association between shows and firearm deaths and injuries. The observed differences in associations between California and Nevada gun shows and firearm deaths and injuries may be due to differences in the characteristics of the regions exposed to California gun shows versus those exposed to Nevada gun shows. For example, regions exposed to Nevada gun shows tend to be more rural and have lower rates of firearm death and injury (Table 2 and Appendix Table 4).

Appendix Table 4. Distribution of Characteristics of Regions Exposed to Gun Shows Before and After Restriction to Regions Similar to Those Exposed to Nevada Gun Shows

Appendix Table 4. Distribution of Characteristics of Regions Exposed to Gun Shows Before and After Restriction to Regions Similar to Those Exposed to Nevada Gun Shows
To address this potential source of variation, we restricted the entire analysis to regions similar to those exposed to Nevada gun shows. We identified these regions by tightly matching on ZIP code characteristics that differed between regions exposed to California versus Nevada shows and may modify the association between gun shows and firearm deaths and injuries. These were population density, percentage of veterans, median income, median age, percentage of white non-Hispanic persons, hunting licenses per capita, and the overall rate of firearm injury between 2005 and 2013. We used 1-to-many greedy Mahalanobis distance matching (a generalization of nearest-neighbor matching based on Euclidean distance) with replacement and a caliper of 0.01 SDs of the distance measure (43). This means that several ZIP codes exposed to California shows could be matched to each ZIP code exposed to a Nevada show. We discarded ZIP codes with characteristics with values outside the range of those observed for ZIP codes exposed to Nevada gun shows. Other matching approaches, such as optimal or nearest-neighbor matching based on the propensity score, produced nearly identical matches. Although restricting to the California region along the California–Nevada border exposed to both California and Nevada gun shows was not possible because the populations were too sparse to estimate stable rates of firearm deaths and injuries, this approach provides a close approximation by restricting locations to those very similar to this border region.
Of the 1769 ZIP codes in California, 490 remained after restriction and 192 of those were matched more than once because of replacement. Appendix Table 4 presents the distribution of the potentially modifying characteristics before and after restriction and compared with ZIP codes exposed to Nevada gun shows. After restriction, the remaining ZIP codes were very similar to those exposed to Nevada gun shows. Compared with all California ZIP codes, the restricted set is less densely populated; includes more veterans and non-Hispanic whites; and has higher median income, median age, and hunting licenses per capita.
Table 5 presents the results of the restricted analysis. Results are nearly identical to those of the main analysis, suggesting that modification by these factors is not a driver of the observed differences in associations between California and Nevada gun shows and firearm deaths and injuries.

Negative Control Analysis

We tested the association between gun shows and nonfirearm injuries as a negative control to assess whether common causes of firearm and nonfirearm injuries confounded our findings (34). Using the same data sources and analytic approach as in the main analysis, we found that neither California nor Nevada gun shows were meaningfully associated with short-term increases in nonfirearm injuries (Appendix Table 5). Although several of the tested associations were statistically significant, this finding was driven by the large number of nonfirearm unintentional injuries (n = 6 065 633), and the RRs were effectively null. These results provide further evidence that the results are not due to confounding by common causes of firearm and nonfirearm injuries.

Appendix Table 5. Negative Control Analysis for the Adjusted Association Between Nonfirearm Injury Deaths and Hospital Visits in California and Gun Shows in California Versus Nevada

Appendix Table 5. Negative Control Analysis for the Adjusted Association Between Nonfirearm Injury Deaths and Hospital Visits in California and Gun Shows in California Versus Nevada

Quantitative Bias Analysis for an Unobserved Confounder

To assess the potential role of residual confounding due to unmeasured factors, we did a quantitative bias analysis for 2 of the measured associations: that between gun shows in Nevada and firearm deaths and injuries in California and that between state of gun show and increases in firearm deaths and injuries after shows.
Association Between Gun Shows in Nevada and Firearm Deaths and Injuries in California
We estimated the characteristics of an unmeasured confounder that would yield the observed association between gun shows in Nevada and firearm deaths and injuries in California, if the true effect were not statistically significant. To do this, we used the bias equation presented by VanderWeele and Arah (44) for the RR and applied it to the estimated RR of the association between Nevada gun shows and firearm deaths and injuries in California (exp[β1 + β3] in the main regression analysis).
We defined the following random variables: Let A be a binary indicator representing exposure to Nevada gun shows (that is, the period is the 2 weeks after Nevada gun shows versus the 2 weeks before), let Y be the rate of firearm deaths and injuries per 100 000 population in California, let X be the measured covariates controlled in the main analysis, and let U be an unmeasured confounder. Following VanderWeele and Arah's analysis, we made 3 assumptions: first, that the association between U and Y does not vary between strata of A; second, that the association between U and A does not vary between strata of X; and third, that U is binary. Under these conditions, the bias in the conditional causal RR is defined as the ratio between the observed RR and the true conditional causal RR and is computed as:
graphic
View largeDownload slide
graphic
View largeDownload slide
where γ is the association between U and Y, defined as γ = E(Y|a,x,u = 1)/E(Y|a,x,u = 0). The association between U and A is defined as δ = P(U = 1|a = 1,x)/P(U = 1|a = 0,x).
We estimated the corrected lower confidence bound of the RR for the association between Nevada gun shows and the rate of firearm deaths and injuries in California (observed RR, 1.69 [CI, 1.16 to 2.45]) across a range of bias scenarios. We tested values of γ (the relative association of U with Y) ranging from 1 to 3, values of δ (the relative association of U with A) ranging from 1 to 3, and prevalence of U among the exposed (P(U = 1|a = 1,x)) ranging between 0.1 and 0.8. This analysis tells us how prevalent U must be and how strong the U–A and U–Y relationships would have to be for an uncontrolled confounder to explain the association observed in our study.
Appendix Figure 3 presents the results of this analysis. Across all of the scenarios we considered, an unmeasured confounder would need to be associated with both gun shows and firearm deaths and injuries with RRs of at least 1.5 or 2 to yield the observed association, if the true effect were not statistically significant.
Appendix Figure 3.

Bias analysis results for the association between Nevada gun shows and California firearm deaths and injuries.

Each graph represents a scenario for the prevalence of U among the exposed (P(U = 1|a = 1,x), which ranges from 0.1 to 0.8). In each plot, the x-axis measures the association between the unmeasured confounder and firearm deaths and injuries in California, the color of each line indicates the association between the unmeasured confounder and exposure to Nevada gun shows, and the y-axis displays the corrected lower confidence bound for the given bias scenario. For example, when the prevalence of U is 0.1, the RR for the U–gun shows association is 3, and the RR for the U–firearm deaths and injuries association is 3, then the association between Nevada gun shows and California firearm deaths and injuries would still be statistically significant, with a corrected lower confidence bound above 1. RR = rate ratio.

This analysis informs our interpretation of the results. For the association between Nevada gun shows and firearm deaths and injuries in California to be spurious, another factor would have to match the geographic and temporal pattern of the 275 Nevada gun shows and be strongly associated with firearm deaths and injuries in California. This factor would have to be notably more prevalent after Nevada gun shows than before, corresponding to RRs of at least 1.5 or 2 for a confounder that is up to 80% more prevalent in the 2 weeks after Nevada gun shows than in the 2 weeks before. Identifying a factor that fits these criteria is challenging. Similar bias analyses have been used to strengthen evidence of the association between firearm ownership and suicide (40). One possibility is that this factor is a marker or artifact of Nevada gun shows; for example, if persons at higher risk for firearm deaths and injuries come to California areas near Nevada shows when Nevada shows are occurring, or happenings at Nevada gun shows prompt persons in nearby California to use their firearms in ways they otherwise might not, then we might see the observed association. There may be other explanations as well.
Association Between State of Gun Show and Increases in Firearm Deaths and Injuries After Gun Shows
We also estimated the characteristics of an unmeasured confounder that would yield the observed association between the state in which the gun show was held and increases in firearm deaths and injuries after gun shows, if the true effect were not statistically significant. Again, we used the bias equation presented by VanderWeele and Arah (44), but in this case, we applied it to the ratio of RRs for the association between the state in which the gun show was held (Nevada vs. California) and increases in firearm deaths and injuries after gun shows (exp[β3] in the main regression analysis).
For this application, we defined the following random variables: Let A be a binary indicator representing the state in which the gun show was held (Nevada vs. California), Y be the change in rate of firearm deaths and injuries per 100 000 population in the 2 weeks before gun shows compared with the 2 weeks after, X be the measured covariates controlled in the main analysis, and U be an unmeasured confounder. Following VanderWeele and Arah's analysis, we made the same 3 assumptions as above, and the bias in the conditional causal RR is defined as above.
We estimated the corrected lower confidence bound for the ratio of RRs estimate of the association between the state of the gun show and increases in firearm deaths and injuries after shows (observed, 1.70 [CI, 1.17 to 2.47]) across a range of bias scenarios. Again, we tested values of γ (the relative association of U with Y) ranging from 1 to 3, values of δ (the relative association of U with A) ranging from 1 to 3, and prevalence of U for Nevada gun shows (P(U = 1|a = 1,x)) ranging between 0.1 and 0.8. This analysis tells us how prevalent U must be for Nevada gun shows and how strong the U–A and U–Y relationships would have to be for an uncontrolled confounder to explain the association observed in our study.
Appendix Figure 4 presents the results of this analysis. Across all of the scenarios we considered, an unmeasured confounder would need to be associated with both the state of the gun shows and increases in firearm deaths and injuries after shows with RRs of at least 1.5 or 2 to yield the observed association, if the true effect were not statistically significant.
Appendix Figure 4.

Bias analysis results for the association between state of gun show and increases in firearm deaths and injuries after gun shows.

Each graph represents a scenario for the prevalence of U for Nevada gun shows (P(U = 1|a = 1,x), which ranges from 0.1 to 0.8). In each plot, the x-axis measures the association between the unmeasured confounder and increases in firearm deaths and injuries after gun shows, the color of each line indicates the association between the unmeasured confounder and the state of the gun show, and the y-axis displays the corrected lower confidence bound for the given bias scenario. For example, when the prevalence of U is 0.1 for Nevada gun shows, the RR for the U–A association is 3, and the RR for the U–Y association is 3, then the association between the state of the gun show and increases in firearm death and injuries after gun shows would still be statistically significant, with a corrected lower confidence bound above 1. RR = rate ratio.

This analysis informs our interpretation of the results. For the association between state and increases in firearm deaths and injuries after shows to be spurious, another factor would have to match the geographic and temporal pattern of the 915 gun shows in both states. This factor would also have to be strongly associated with both the state of the gun show and changes in firearm deaths and injuries immediately before and after the shows, corresponding to RRs of at least 1.5 to 2 for a confounder that is up to 80% more prevalent for Nevada shows than California shows. Identifying a factor that fits these criteria is challenging. Similar bias analyses have been used to strengthen evidence of the association between firearm ownership and suicide (40).

References

  1. Centers for Disease Control and Prevention; National Center for Injury Prevention and Control. Web-based Injury Statistics Query and Reporting System (WISQARS). 2005. Accessed at www.cdc.gov/injury/wisqars on 22 January 2016.
  2. Kellermann
    AL
    ,  
    Rivara
    FP
    ,  
    Rushforth
    NB
    ,  
    Banton
    JG
    ,  
    Reay
    DT
    ,  
    Francisco
    JT
    ,  
    et al
    Gun ownership as a risk factor for homicide in the home.
    N Engl J Med
    1993
    329
    1084
    91
     PubMed
    CrossRef
     PubMed
  3. Kellermann
    AL
    ,  
    Rivara
    FP
    ,  
    Somes
    G
    ,  
    Reay
    DT
    ,  
    Francisco
    J
    ,  
    Banton
    JG
    ,  
    et al
    Suicide in the home in relation to gun ownership.
    N Engl J Med
    1992
    327
    467
    72
     PubMed
    CrossRef
     PubMed
  4. Kellermann
    AL
    ,  
    Somes
    G
    ,  
    Rivara
    FP
    ,  
    Lee
    RK
    ,  
    Banton
    JG
    .  
    Injuries and deaths due to firearms in the home.
    J Trauma
    1998
    45
    263
    7
     PubMed
    CrossRef
     PubMed
  5. Dahlberg
    LL
    ,  
    Ikeda
    RM
    ,  
    Kresnow
    MJ
    .  
    Guns in the home and risk of a violent death in the home: findings from a national study.
    Am J Epidemiol
    2004
    160
    929
    36
     PubMed
    CrossRef
     PubMed
  6. Wiebe
    DJ
    .  
    Homicide and suicide risks associated with firearms in the home: a national case-control study.
    Ann Emerg Med
    2003
    41
    771
    82
     PubMed
    CrossRef
     PubMed
  7. Miller
    M
    ,  
    Azrael
    D
    ,  
    Hemenway
    D
    .  
    Firearm availability and unintentional firearm deaths, suicide, and homicide among 5-14 year olds.
    J Trauma
    2002
    52
    267
    74
     PubMed
     PubMed
  8. Azrael
    D
    ,  
    Hemenway
    D
    .  
    ‘In the safety of your own home': results from a national survey on gun use at home.
    Soc Sci Med
    2000
    50
    285
    91
     PubMed
    CrossRef
     PubMed
  9. Grinshteyn
    E
    ,  
    Hemenway
    D
    .  
    Violent death rates: the US compared with other high-income OECD countries, 2010.
    Am J Med
    2016
    129
    266
    73
     PubMed
    CrossRef
     PubMed
  10. Hemenway
    D
    ,  
    Miller
    M
    .  
    Firearm availability and homicide rates across 26 high-income countries.
    J Trauma
    2000
    49
    985
    8
     PubMed
    CrossRef
     PubMed
  11. Hemenway
    D
    ,  
    Shinoda-Tagawa
    T
    ,  
    Miller
    M
    .  
    Firearm availability and female homicide victimization rates among 25 populous high-income countries.
    J Am Med Womens Assoc (1972)
    2002
    57
    100
    4
     PubMed
     PubMed
  12. Killias
    M
    .  
    International correlations between gun ownership and rates of homicide and suicide.
    CMAJ
    1993
    148
    1721
    5
     PubMed
     PubMed
  13. Cook
    PJ
    ,  
    Ludwig
    J
    .  
    Guns in America: Results of a Comprehensive National Survey on Firearms Ownership and Use
    Washington, DC
    Police Foundation
    1996
  14. Hepburn
    L
    ,  
    Miller
    M
    ,  
    Azrael
    D
    ,  
    Hemenway
    D
    .  
    The US gun stock: results from the 2004 national firearms survey.
    Inj Prev
    2007
    13
    15
    9
     PubMed
    CrossRef
     PubMed
  15. Wintemute
    GJ
    .  
    Inside Gun Shows: What Goes On When Everybody Thinks Nobody's Watching
    Sacramento, CA
    Violence Prevention Research Program
    2009
  16. Miller
    M
    ,  
    Hepburn
    L
    ,  
    Azrael
    D
    .  
    Firearm acquisition without background checks: results of a national survey.
    Ann Intern Med
    2017
    166
    233
    9
    CrossRef
     PubMed
  17. Braga
    AA
    ,  
    Kennedy
    DM
    .  
    Gun control in America: gun shows and the illegal diversion of firearms.
    Georgetown Public Policy Review
    2000
    6
    7
    24
  18. U.S. Department of the Treasury; Bureau of Alcohol, Tobacco and Firearms
    Following the Gun: Enforcing Federal Laws Against Firearms Traffickers
    Washington, DC
    US Gov Pr Office
    2000
  19. Gun Shows: Brady Checks and Crime Gun Traces
    Washington, DC
    Bureau of Alcohol, Tobacco and Firearms
    1999
  20. Cook
    PJ
    ,  
    Braga
    AA
    .  
    Comprehensive firearms tracing: strategic and investigative uses of new data on firearms markets.
    Ariz Law Rev
    2001
    43
    277
    309
  21. Knight
    B
    .  
    State gun policy and cross-state externalities: evidence from crime gun tracing.
    Am Econ J Econ Policy
    2013
    5
    200
    29
    CrossRef
  22. Fleegler
    EW
    ,  
    Lee
    LK
    ,  
    Monuteaux
    MC
    ,  
    Hemenway
    D
    ,  
    Mannix
    R
    .  
    Firearm legislation and firearm-related fatalities in the United States.
    JAMA Intern Med
    2013
    173
    732
    40
     PubMed
    CrossRef
     PubMed
  23. A.B. 295, 1999 State Assemb., Reg. Sess. (Cal. 1999). Accessed at http://leginfo.legislature.ca.gov/faces/billTextClient.xhtml?bill_id=199920000AB295 on 13 December 2016.
  24. Wintemute
    GJ
    .  
    Gun shows across a multistate American gun market: observational evidence of the effects of regulatory policies.
    Inj Prev
    2007
    13
    150
    5
     PubMed
    CrossRef
     PubMed
  25. Duggan
    M
    ,  
    Hjalmarsson
    R
    ,  
    Jacob
    BA
    .  
    The short-term and localized effect of gun shows: evidence from California and Texas.
    Rev Econ Stat
    2011
    93
    786
    99
    CrossRef
  26. Hemenway
    D
    .  
    How to find nothing.
    J Public Health Policy
    2009
    30
    260
    8
     PubMed
    CrossRef
     PubMed
  27. Wintemute
    GJ
    ,  
    Hemenway
    D
    ,  
    Webster
    D
    ,  
    Pierce
    G
    ,  
    Braga
    AA
    .  
    Gun shows and gun violence: fatally flawed study yields misleading results.
    Am J Public Health
    2010
    100
    1856
    60
     PubMed
    CrossRef
     PubMed
  28. Angrist
    JD
    ,  
    Pischke
    JS
    .  
    Mostly Harmless Econometrics: An Empiricist's Companion
    Princeton, NJ
    Princeton Univ Pr
    2009
  29. Dimick
    JB
    ,  
    Ryan
    AM
    .  
    Methods for evaluating changes in health care policy: the difference-in-differences approach.
    JAMA
    2014
    312
    2401
    2
     PubMed
    CrossRef
     PubMed
  30. Centers for Disease Control and Prevention
    Strategies to improve external cause-of-injury coding in state-based hospital discharge and emergency department data systems: recommendations of the CDC Workgroup for Improvement of External Cause-of-Injury Coding.
    MMWR Morb Mortal Wkly Rep
    2008
    57
    1
    16
     PubMed
  31. ABBYY FineReader. 2016. Accessed at www.abbyy.com/en-us/finereader on 9 November 2016.
  32. Developer's guide: Google Maps Distance Matrix API. 2016. Accessed at https://developers.google.com/maps/documentation/distance-matrix/intro on 3 September 2016.
  33. Bates D, Maechler M, Bolker B, Walker S, Christensen RHB, Singmann H, et al. lme4: Linear Mixed-Effects Models using ‘Eigen' and S4. 2016. Accessed at https://cran.r-project.org/web/packages/lme4/index.html on 14 December 2016.
  34. Lipsitch
    M
    ,  
    Tchetgen Tchetgen
    E
    ,  
    Cohen
    T
    .  
    Negative controls: a tool for detecting confounding and bias in observational studies.
    Epidemiology
    2010
    21
    383
    8
     PubMed
    CrossRef
     PubMed
  35. Morin R. The demographics and politics of gun-owning households. Washington, DC: Pew Research Center; 2014. Accessed at www.pewresearch.org/fact-tank/2014/07/15/the-demographics-and-politics-of-gun-owning-households on 10 July 2017.
  36. Parker
    K
    ,  
    Horowitz
    JM
    ,  
    Igielnik
    R
    ,  
    Oliphant
    B
    ,  
    Brown
    A
    .  
    America's complex relationship with guns
    Washington, DC
    Pew Research Center
    2017
  37. Arnold
    BF
    ,  
    Hogan
    DR
    ,  
    Colford
    JM
    Jr
    ,  
    Hubbard
    AE
    .  
    Simulation methods to estimate design power: an overview for applied research.
    BMC Med Res Methodol
    2011
    11
    94
     PubMed
    CrossRef
     PubMed
  38. Wintemute
    GJ
    ,  
    Romero
    MP
    ,  
    Wright
    MA
    ,  
    Grassel
    KM
    .  
    The life cycle of crime guns: a description based on guns recovered from young people in California.
    Ann Emerg Med
    2004
    43
    733
    42
     PubMed
    CrossRef
     PubMed
  39. Braga
    AA
    ,  
    Wintemute
    GJ
    ,  
    Pierce
    GL
    ,  
    Cook
    PJ
    ,  
    Ridgeway
    G
    .  
    Interpreting the empirical evidence on illegal gun market dynamics.
    J Urban Health
    2012
    89
    779
    93
     PubMed
    CrossRef
     PubMed
  40. Miller
    M
    ,  
    Swanson
    SA
    ,  
    Azrael
    D
    .  
    Are we missing something pertinent? A bias analysis of unmeasured confounding in the firearm-suicide literature.
    Epidemiol Rev
    2016
    38
    62
    9
     PubMed
     PubMed
  41. Caveney
    AF
    ,  
    Smith
    MA
    ,  
    Morgenstern
    LB
    ,  
    Lisabeth
    LD
    .  
    Use of death certificates to study ethnic-specific mortality.
    Public Health Rep
    2006
    121
    275
    81
     PubMed
    CrossRef
     PubMed
  42. Mohler
    B
    ,  
    Earls
    F
    .  
    Trends in adolescent suicide: misclassification bias?
    Am J Public Health
    2001
    91
    150
    3
     PubMed
    CrossRef
     PubMed
  43. Ho
    D
    ,  
    Imai
    K
    ,  
    King
    G
    ,  
    Stuart
    EA
    .  
    MatchIt: nonparametric preprocessing for parametric causal inference.
    J Stat Softw
    2011
    42
    1
    28
    CrossRef
  44. VanderWeele
    TJ
    ,  
    Arah
    OA
    .  
    Bias formulas for sensitivity analysis of unmeasured confounding for general outcomes, treatments, and confounders.
    Epidemiology
    2011
    22
    42
    52
     PubMed
    CrossRef
     PubMed
Appendix Figure 1.

Comparison of observed and simulated distribution of firearm deaths and injuries during the 2 weeks after gun shows in nearby regions.

The unit of analysis is the gun show. The figure presents the distribution of the observed and simulated number of firearm deaths and injuries, per gun show, ≤2 wk after each gun show, in regions within driving distance of each show, by state. For example, in both the observed and simulated data, just fewer than 300 gun show regions had no firearm deaths or injuries ≤2 wk after the show.

Appendix Figure 2.

Locations of gun shows in California and Nevada.

Appendix Figure 3.

Bias analysis results for the association between Nevada gun shows and California firearm deaths and injuries.

Each graph represents a scenario for the prevalence of U among the exposed (P(U = 1|a = 1,x), which ranges from 0.1 to 0.8). In each plot, the x-axis measures the association between the unmeasured confounder and firearm deaths and injuries in California, the color of each line indicates the association between the unmeasured confounder and exposure to Nevada gun shows, and the y-axis displays the corrected lower confidence bound for the given bias scenario. For example, when the prevalence of U is 0.1, the RR for the U–gun shows association is 3, and the RR for the U–firearm deaths and injuries association is 3, then the association between Nevada gun shows and California firearm deaths and injuries would still be statistically significant, with a corrected lower confidence bound above 1. RR = rate ratio.

Appendix Figure 4.

Bias analysis results for the association between state of gun show and increases in firearm deaths and injuries after gun shows.

Each graph represents a scenario for the prevalence of U for Nevada gun shows (P(U = 1|a = 1,x), which ranges from 0.1 to 0.8). In each plot, the x-axis measures the association between the unmeasured confounder and increases in firearm deaths and injuries after gun shows, the color of each line indicates the association between the unmeasured confounder and the state of the gun show, and the y-axis displays the corrected lower confidence bound for the given bias scenario. For example, when the prevalence of U is 0.1 for Nevada gun shows, the RR for the U–A association is 3, and the RR for the U–Y association is 3, then the association between the state of the gun show and increases in firearm death and injuries after gun shows would still be statistically significant, with a corrected lower confidence bound above 1. RR = rate ratio.

Appendix Table 1. ICD-9 and ICD-10 External Cause-of-Injury Codes Used to Identify and Classify Firearm Deaths and Injuries

Appendix Table 1. ICD-9 and ICD-10 External Cause-of-Injury Codes Used to Identify and Classify Firearm Deaths and Injuries

Table 1. Characteristics of California and Nevada Gun Shows and Population Exposure to Gun Shows

Table 1. Characteristics of California and Nevada Gun Shows and Population Exposure to Gun Shows

Table 2. Unadjusted Analyses of the Association Between Firearm Deaths and Injuries in California and Gun Shows in California and Nevada

Table 2. Unadjusted Analyses of the Association Between Firearm Deaths and Injuries in California and Gun Shows in California and Nevada

Table 3. Adjusted Analyses of the Association Between Firearm Deaths and Injuries in California and Gun Shows in California Versus Nevada

Table 3. Adjusted Analyses of the Association Between Firearm Deaths and Injuries in California and Gun Shows in California Versus Nevada

Table 4. Adjusted Analyses of the Association Between Firearm Deaths and Injuries in California and Gun Shows in California and Nevada, by State

Table 4. Adjusted Analyses of the Association Between Firearm Deaths and Injuries in California and Gun Shows in California and Nevada, by State

Table 5. Adjusted Analyses of the Association Between Firearm Deaths and Injuries in California and Gun Shows in California Versus Nevada, Restricted to Regions Similar to Those Exposed to Nevada Gun Shows

Table 5. Adjusted Analyses of the Association Between Firearm Deaths and Injuries in California and Gun Shows in California Versus Nevada, Restricted to Regions Similar to Those Exposed to Nevada Gun Shows

Appendix Table 2. Secondary Analyses for the Adjusted Association Between Firearm Deaths and Injuries Along Firearm Trafficking Routes and Excluding California's 10-Day Waiting Period

Appendix Table 2. Secondary Analyses for the Adjusted Association Between Firearm Deaths and Injuries Along Firearm Trafficking Routes and Excluding California's 10-Day Waiting Period

Appendix Table 3. Sensitivity Analyses for the Adjusted Association Between Firearm Deaths and Injuries in California and Gun Shows in California Versus Nevada

Appendix Table 3. Sensitivity Analyses for the Adjusted Association Between Firearm Deaths and Injuries in California and Gun Shows in California Versus Nevada

Appendix Table 4. Distribution of Characteristics of Regions Exposed to Gun Shows Before and After Restriction to Regions Similar to Those Exposed to Nevada Gun Shows

Appendix Table 4. Distribution of Characteristics of Regions Exposed to Gun Shows Before and After Restriction to Regions Similar to Those Exposed to Nevada Gun Shows

Appendix Table 5. Negative Control Analysis for the Adjusted Association Between Nonfirearm Injury Deaths and Hospital Visits in California and Gun Shows in California Versus Nevada

Appendix Table 5. Negative Control Analysis for the Adjusted Association Between Nonfirearm Injury Deaths and Hospital Visits in California and Gun Shows in California Versus Nevada

Clinical Slide Sets

Terms of Use

The In the Clinic® slide sets are owned and copyrighted by the American College of Physicians (ACP). All text, graphics, trademarks, and other intellectual property incorporated into the slide sets remain the sole and exclusive property of the ACP. The slide sets may be used only by the person who downloads or purchases them and only for the purpose of presenting them during not-for-profit educational activities. Users may incorporate the entire slide set or selected individual slides into their own teaching presentations but may not alter the content of the slides in any way or remove the ACP copyright notice. Users may make print copies for use as hand-outs for the audience the user is personally addressing but may not otherwise reproduce or distribute the slides by any means or media, including but not limited to sending them as e-mail attachments, posting them on Internet or Intranet sites, publishing them in meeting proceedings, or making them available for sale or distribution in any unauthorized form, without the express written permission of the ACP. Unauthorized use of the In the Clinic slide sets will constitute copyright infringement.

This feature is available only to Registered Users

Subscribe/Learn More
Submit a Comment

3 Comments

Michael Xu

UTHSC

October 24, 2017

IIlegal

It is already federal crime in all states for both private individuals and businesses to sell a firearm across state borders without background check. A California resident cannot drive to NV, buy a firearm, and then travel back to California. He or she must buy the firearm, have the private individual/business transfer the firearm to a California firearm dealer, at which time a back ground check will be conducted.

Seth Breitbart

EHMC

January 3, 2018

Politically Motivated Pseudoscience

As the previous commenter, Michael Xu, correctly stated, it is already a federal crime in all 50 states for both private individuals and businesses to sell a firearm across state borders without a background check. It is also illegal for a person to buy a gun in a state where he/she does not maintain a residence. Again, as Michael Xu correctly stated, a California resident cannot drive to Nevada, buy a firearm, and then travel back to California: Such transaction is already illegal.

If, for example, a California resident wants to buy a gun from a seller from Nevada. The California resident and the Nevada gun seller must go through several steps. First, the California resident contacts his local FFL (e.g. a local gun store where a background check can be done) and tells them that he wants to buy an out of state gun. Next, he will pay the Nevada gun seller. The Nevada gun seller will then transfer the gun to the gun store designated by the buyer, where a background check will be completed. For this, the California resident must pay another fee. Thus, the entire argument of this article, which infers that California residents are travelling over the state border to legally buy guns at Nevada guns shows, is verifiably untrue.

It is a shame that the Annals of Internal Medicine would publish such a clearly politically motivated article that and can be debunked at a glance.

Alan R. Ertle, MD, MPH, MBA

Mercy Medical Group, Inc.

January 3, 2018

Highly Questionable Association

It is likely that everyone in a health profession in the United States agrees that there is too much gun violence and too many firearms-related injuries. Matthay and colleagues’ study attempts to make an association between the relatively under-regulated gun shows of Nevada and the increase in firearm death and injuries in the immediate post-gun-show period in California zip codes where those California populations were deemed close enough to drive to both certain California and Nevada gun shows. What stands out first is the difference in total firearm death and injuries for California-based gun shows versus Nevada-based gun shows. In the two weeks after gun shows there were 14,893 total firearm deaths and injuries after California-based gun shows and 74 total firearm deaths and injuries for Nevada-based gun shows, or about 0.5% of the California-based gun show total. These numbers are totals for the eight-year period of the study. This vast difference in raw number of events during the eight-year study period must point out an equally vast difference in measured at-risk populations in California between California-based and Nevada-based gun shows, despite a much closer number of total gun shows analyzed in the final regression for California (585) and Nevada (161). One can think about the raw data in several different ways. For the Nevada-based gun shows, there are about 5.5 gun-related events per year in the two weeks running up to the gun shows (44 events over 8 years) and 9.25 gun-related events per year in the two weeks after (74 events over 8 years). Alternatively, there was an average of 0.27 gun-related events per Nevada-based gun show in the two weeks prior to the guns shows (44 events and 161 gun shows) and there was an average of 0.45 gun-related events per Nevada-based gun show in the two weeks after (74 events and 161 gun shows). On the surface, the total firearm deaths and injuries in the study population before and after the Nevada-based gun shows seems to be quite small, particular given the eight-year study period and 161 analyzed Nevada-based gun shows. It calls into question whether the power calculation to actually determine true differences rather than by-chance differences was accurate.
Additionally, it is not clear if anyone from California actually went to the Nevada-based gun shows, whether they actually purchased weapons or ammunition at the Nevada-based gun shows, or whether any weapon or ammunition from a Nevada-based gun show was actually used to conduct firearms-related violence in any of the California incidents measured in the two-week post-gun-show period. It would seem that it would be necessary to have this data in order to make any assertion that Nevada-based gun shows had anything to do with the events they measured. If you put one variable into a model, it is always likely that you will find an association if you look hard enough. I find the study totally underwhelming and highly questionable in its conclusion that, “Firearm injuries in California remained stable after California gun shows but increased by a small but significant amount after Nevada shows.”

PDF
Not Available
Citations
Citation

Matthay EC, Galin J, Rudolph KE, et al. In-State and Interstate Associations Between Gun Shows and Firearm Deaths and Injuries: A Quasi-experimental Study. Ann Intern Med. 2017;167:837–844. [Epub ahead of print 7 November 2017]. doi: https://doi.org/10.7326/M17-1792

Download citation file:

  • Ris (Zotero)
  • EndNote
  • BibTex
  • Medlars
  • ProCite
  • RefWorks
  • Reference Manager

© 2019

×
Permissions

Published: Ann Intern Med. 2017;167(12):837-844.

DOI: 10.7326/M17-1792

Published at www.annals.org on 7 November 2017

©
2017 American College of Physicians
4 Citations

See Also

Firearm Injury After Gun Shows: Evidence to Gauge the Potential Impact of Regulatory Interventions
Associations Between Gun Shows and Firearm Deaths and Injuries
Associations Between Gun Shows and Firearm Deaths and Injuries
Associations Between Gun Shows and Firearm Deaths and Injuries
Associations Between Gun Shows and Firearm Deaths and Injuries
View MoreView Less

Related Articles

Reducing Firearm Injuries and Deaths in the United States: A Position Paper From the American College of Physicians
Annals of Internal Medicine; 169 (10): 704-707
Public Perceptions of Firearm- and Non–Firearm-Related Violent Death in the United States: A National Study
Annals of Internal Medicine; 169 (10): 734-737
Firearm Involvement in Violent Victimization and Mental Health: An Observational Study
Annals of Internal Medicine; 169 (8): 584-585
Fatal Firearm Incidents Before and After Australia's 1996 National Firearms Agreement Banning Semiautomatic Rifles
Annals of Internal Medicine; 169 (1): 62-64
View MoreView Less

Related Point of Care

Preventing Firearm-Related Death and Injury
Annals of Internal Medicine; 170 (11): ITC81-ITC96
View MoreView Less

Related Topics

Hospital Medicine

Hospital Medicine.

CME/MOC Activity Requires Users to be Registered and Logged In.
Sign in below to access your subscription for full content
INDIVIDUAL SIGN IN
Sign In|Set Up Account
You will be directed to acponline.org to register and create your Annals account
Annals of Internal Medicine
CREATE YOUR FREE ACCOUNT
Create Your Free Account|Why?
To receive access to the full text of freely available articles, alerts, and more. You will be directed to acponline.org to complete your registration.
×
The Comments Feature Requires Users to be Registered and Logged In.
Sign in below to access your subscription for full content
INDIVIDUAL SIGN IN
Sign In|Set Up Account
You will be directed to acponline.org to register and create your Annals account
Annals of Internal Medicine
CREATE YOUR FREE ACCOUNT
Create Your Free Account|Why?
To receive access to the full text of freely available articles, alerts, and more. You will be directed to acponline.org to complete your registration.
×
link to top

Content

  • Home
  • Latest
  • Issues
  • Channels
  • CME/MOC
  • In the Clinic
  • Journal Club
  • Web Exclusives

Information For

  • Author Info
  • Reviewers
  • Press
  • Readers
  • Institutions / Libraries / Agencies
  • Advertisers

Services

  • Subscribe
  • Renew
  • Alerts
  • Current Issue RSS
  • Latest RSS
  • In the Clinic RSS
  • Reprints & Permissions
  • Contact Us
  • Help
  • About Annals
  • About Mobile
  • Patient Information
  • Teaching Tools
  • Annals in the News
  • Share Your Feedback

Awards and Cover

  • Personae (Cover Photo)
  • Junior Investigator Awards
  • Poetry Prize

Other Resources

  • ACP Online
  • Career Connection
  • ACP Advocate Blog
  • ACP Journal Wise

Follow Annals On

  • Twitter Link
  • Facebook Link
acp link acp
silverchair link silverchair

Copyright © 2019 American College of Physicians. All Rights Reserved.

Print ISSN: 0003-4819 | Online ISSN: 1539-3704

Privacy Policy

|

Conditions of Use

This site uses cookies. By continuing to use our website, you are agreeing to our privacy policy. | Accept
×

You need a subscription to this content to use this feature.

×
PDF Downloads Require Access to the Full Article.
Sign in below to access your subscription for full content
INDIVIDUAL SIGN IN
Sign In|Set Up Account
You will be directed to acponline.org to register and create your Annals account
INSTITUTIONAL SIGN IN
Open Athens|Shibboleth|Log In
Annals of Internal Medicine
PURCHASE OPTIONS
Buy This Article|Subscribe
You will be redirected to acponline.org to sign-in to Annals to complete your purchase.
CREATE YOUR FREE ACCOUNT
Create Your Free Account|Why?
To receive access to the full text of freely available articles, alerts, and more. You will be directed to acponline.org to complete your registration.
×
Access to this Free Content Requires Users to be Registered and Logged In. Please Choose One of the Following Options
Sign in below to access your subscription for full content
INDIVIDUAL SIGN IN
Sign In|Set Up Account
You will be directed to acponline.org to register and create your Annals account
Annals of Internal Medicine
CREATE YOUR FREE ACCOUNT
Create Your Free Account|Why?
To receive access to the full text of freely available articles, alerts, and more. You will be directed to acponline.org to complete your registration.
×