You are here

Article Text

Article menu

Download PDFPDF

Study protocol
Evaluation of California's Armed and Prohibited Persons System: study protocol for a cluster-randomised trial
  1. Garen J Wintemute1,
  2. Laurel Beckett2,
  3. Philip H Kass3,
  4. Daniel Tancredi4,
  5. David Studdert5,
  6. Glenn Pierce6,
  7. Anthony A Braga6,
  8. Mona A Wright1,
  9. Magdalena Cerdá1
  1. 1Violence Prevention Research Program, Department of Emergency Medicine, School of Medicine, University of California, Davis, Sacramento, California, USA
  2. 2Department of Public Health Sciences, School of Medicine, University of California, Davis, Davis, California, USA
  3. 3Department of Population Health and Reproduction, School of Veterinary Medicine, University of California, Davis, Davis, California, USA
  4. 4Department of Pediatrics, School of Medicine, University of California, Davis, Sacramento, California, USA
  5. 5Schools of Medicine and Law, Stanford University, Palo Alto, California, USA
  6. 6School of Criminology and Criminal Justice, Northeastern University, Boston, Massachusetts, USA
  1. Correspondence to Dr Garen J Wintemute, Violence Prevention Research Program, Department of Emergency Medicine, School of Medicine, University of California, Davis, Sacramento, CA 95817, USA; gjwintemute{at}ucdavis.edu

Abstract

Background and objective Too little is known about the effectiveness of efforts to prevent firearm violence. Our objective is to evaluate California's Armed and Prohibited Persons System (APPS), a law enforcement intervention that seeks to recover firearms from individuals who purchased them legally but subsequently became prohibited from having access to firearms. Prohibitions usually arise from events suggesting an increased risk for future violence.

Design and study population This group-randomised trial involves approximately 20 000 APPS-eligible individuals in 1041 communities. Randomisation was performed at the community level, to early or later intervention (Group 1 and Group 2, respectively) with stratification by region, population and violent crime rate.

Methods APPS is being implemented by the California Department of Justice. The principal outcome measure is the incidence of arrest for a firearm-related or violent crime. Primary analysis will be on an intention-to-treat basis, comparing individuals in Group 1 and Group 2 communities. Analyses will focus on time to event, using proportional hazards regression with adjustment for the clustered nature of the data and incorporating individual- and community-level characteristics. Secondary analyses will examine the effect of the intervention on an as treated basis, effects on subgroups, and effects on community-wide measures such as crime rates.

Discussion APPS may have a significant impact on risk for future violence among members of its target population. The findings of this study will likely be generalisable and have clear implications for violence prevention policy and practice.

Trial registration number NCT02318732.

https://doi.org/10.1136/injuryprev-2016-042194

Statistics from Altmetric.com

    Request Permissions

    If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

    Introduction

    Firearm violence is a significant public health and safety problem, ranking among the leading causes of death in the USA.1 One overarching prevention strategy is to prohibit access to firearms by individuals believed to be at high risk for committing violent acts. Background checks and denials of firearm purchases by prohibited persons, important components of this strategy, reduce the incidence of arrest for violent and firearm-related crime by at least 25% among individuals who are directly affected.2–4 Preventing purchases by prohibited persons is a successful application of the principle of means reduction that may have deterrent effects as well, and it has gained wide acceptance.5–7

    Almost no attention has been given, however, to interventions focused on individuals who pass background checks and acquire firearms legally, then later become prohibited persons. Transitions to prohibited-person status are important because they usually follow events associated with increased risk for future violence. Examples include convictions for violent crimes,8–10 hospitalisations for acute mental illness associated with dangerousness to self or others,11–13 and domestic violence restraining orders (DVROs).14 ,15

    It is highly likely that a large number of firearm owners in the USA have purchased firearms legally and have since become prohibited persons. Millions of individuals legally purchase firearms every year,16 and perhaps 10% of purchasers have one or more pre-existing but non-prohibiting criminal convictions.17 Within 5 years of purchasing firearms, 5–10% of legal firearm purchasers who have any prior criminal history become prohibited persons following convictions for crimes committed after the firearms were purchased.18 Other firearm purchasers become prohibited persons following DVROs or mental health emergencies. Given the size of this population, failure to address continued firearm possession by prohibited persons may have significant negative effects on public health and safety.

    For the first time, a systematic, large-scale effort is being implemented to enforce existing restrictions on possession of firearms by prohibited persons. The Armed and Prohibited Persons System (APPS), an initiative of the California Department of Justice (CA DOJ), uses that agency's archive of firearm transfers to identify firearm owners among individuals who become prohibited persons under California or federal law. APPS then seeks to recover those firearms and thereby prevent both interpersonal and self-directed violence. After several years as a pilot programme, during which APPS intervened with more than 10 000 prohibited persons and recovered more than 10 000 firearms,19 the programme is being scaled up for statewide implementation with funding by a special legislative appropriation.20

    We report here the protocol for a controlled evaluation of APPS: a group-randomised trial involving more than 1000 communities that are home to nearly 80% of California residents. Our primary strategic goal is to determine whether APPS works—specifically, whether it reduces risk of future violence, both among those directly affected and at the population level. Incident arrests will be our primary outcome measure, and our principal specific hypothesis is that exposure to the APPS intervention will be associated with a reduced incidence of subsequent arrest for firearm-related and violent crime.

    Background

    Federal statutes prohibit the purchase and possession of firearms by any person convicted of a felony or a domestic violence misdemeanour or served with a final DVRO; anyone who is ‘an unlawful user of or addicted to any controlled substance’, ‘adjudicated as a mental defective’ or ‘committed to any mental institution’ and others.21 Criminal convictions account for most denials of purchase.16

    Many states have enacted additional prohibitions. California, for example, has a 10-year prohibition for persons convicted of essentially all violent misdemeanours. It also imposes a 5-year prohibition for persons admitted to licensed psychiatric facilities for evaluation following a determination that, due to acute mental illness, they pose an immediate danger to themselves or others. Respondents to temporary DVROs are prohibited for the duration of the orders.

    California courts are required to report to CA DOJ all dispositions of criminal prosecutions and all DVROs. Licensed psychiatric facilities must report emergency admissions resulting from dangerousness to self or others. CA DOJ continually screens these reports electronically to identify individuals who have experienced a prohibiting event. It then conducts automated searches for those individuals in its Automated Firearms System (AFS)—a compilation of records of handgun transfers beginning in 1996, rifle and shotgun transfers beginning in 2014, assault weapon registrations, voluntary weapon registrations, and applications for permits to carry concealed firearms. A ‘hit’ suggests that a prohibited individual is a firearm owner. A tentative match is scrutinised by APPS personnel; if the match is verified, that individual is added to a registry known as the APPS database.

    The APPS intervention is implemented at the community level. When a community is selected, Criminal Intelligence Specialists at CA DOJ review records for all potentially eligible individuals to determine current addresses and reverify that each individual is still a prohibited person and a firearm owner. They provide a hardcopy dossier on each individual to a field enforcement team.

    Each field team consists of four to seven CA DOJ special agents who specialise in APPS enforcement. Field operations are individualised; teams review each subject's dossier prior to contact and develop a specific intervention plan. No advance notice is given to subjects who are to be contacted, as such notice would provide an opportunity to thwart firearm recovery efforts. Teams make multiple efforts if needed to contact the individual. This can involve speaking with family members or associates at several locations. Once contact is made, agents explain the law and its application to that individual and recover all firearms to which he or she has access. This commonly includes firearms beyond those recorded in AFS and may include firearms owned by other members of the prohibited person's household.

    Individuals who acknowledge having firearms surrender them voluntarily in about 80% of cases. Search warrants are obtained if acknowledged firearms are not surrendered. When individuals deny still having firearms, they are asked to provide supporting evidence, such as a receipt from a licensed retailer or a local law enforcement agency. Efforts to recover firearms do not proceed further if no evidence is produced, but the case remains open.

    Recovered firearms are held by CA DOJ or local agencies. Owners have 180 days to arrange sale of their weapons to a licensed retailer. Owners may retrieve their weapons if their prohibited status expires during that time (eg, when a temporary DVRO expires) and a background check does not find them to be prohibited for other reasons. Weapons held beyond 180 days may be destroyed.

    Methods

    Overview

    The study design takes advantage of the fact that, as CA DOJ expands APPS from a pilot project to a statewide enforcement effort, it cannot implement the intervention across all of California simultaneously. We worked with CA DOJ to design a stratified group-randomised trial with community-level allocation to early or later APPS intervention. The trial start date was February 1, 2015. Our planned period of observation is up to 24 months following intervention.

    We are comparing communities that have been assigned to early APPS intervention (Group 1) with communities assigned to later APPS intervention (Group 2). There is no untreated group. Our principal outcome measure is the incidence of arrest for firearm-related and violent crimes, measured at the individual level and aggregated at the community level. A beneficial effect of the intervention would be expected to register as a reduced incidence of subsequent arrest among APPS-eligible individuals in Group 1 communities as compared with the concurrent experience of those in Group 2 communities (figure 1), particularly in the time period before APPS intervention begins in Group 2. The primary analysis will include data only for individuals residing in study communities as of the trial start date.

    Figure 1
    Figure 1

    Hypothetical effects of the Armed and Prohibited Persons System (APPS) intervention on eligible individuals, measured at the community level.

    Formation of the study population

    Randomisation was performed at the community level, with stratification prior to randomisation by region, population, and violent crime rate. We used the 2010 US Census to identify all California cities and towns (hereafter, cities) and census-designated places. Populations were available from the Census, and 2009–2011 average violent crime rates for cities were calculated using data from the California Justice Statistics Center.22

    Two small groups of communities were excluded prior to randomisation. One comprised those where APPS had recently completed enforcement actions or where local enforcement agencies had established parallel interventions. The second included small and isolated communities, typically in the far northern part of the state or east of the Sierra Nevada and excluded by CA DOJ for logistical reasons, where a very small number of APPS-eligible individuals resided. Exclusions reduced the population of study communities by 20.7%, to 27.9 million residents.

    We grouped remaining communities into five regions defined by the California Department of Social Services on the basis of their geographic and labour market continuity.23 Within each region, we classified cities as having an above- or below-median population and an above- or below-median violent crime rate. This established a 2×2 stratification of cities within each region, producing four strata. Within each region and stratum, cities were randomly assigned in a 1:1 ratio to Group 1 or Group 2.

    Crime rates were not available for CDPs; crime reporting is agency-based, and CDPs do not have police agencies. We stratified CDPs by region and population and assigned them to groups as described for cities.

    Data sources

    Principal data for this study are provided by CA DOJ.

    Armed and Prohibited Persons System

    Compiled since 2005, the APPS database contains identifiers for armed and prohibited persons, pointers to records in other CA DOJ files that provide grounds for the determinations that each individual is (a) a firearm owner and (b) a prohibited person, and key information on the nature and duration of prohibitions.

    APPS encounter database

    These data include detailed reports, filed electronically, of all field team encounters and attempted encounters with eligible individuals. Records comprise detailed circumstantial data for all contacts and attempted contacts, a free-text narrative describing encounters and their outcomes, and information on any firearms recovered.

    APPS operations reports

    These are redacted copies of APPS enforcement daily operations reports. We are using them to monitor progress of the intervention and fidelity of implementation according to the randomisation scheme.

    Criminal history system

    California's criminal records are now almost entirely electronic; new records have been electronic since 1980. More than 10 million individuals are in the system. Records contain individual identifiers and the details of all reported criminal justice transactions (arrests, dispositions, etc).

    Automated firearms system

    AFS data were described previously.

    Other sources

    We will obtain data on community demographic and socioeconomic characteristics from the Census and the American Community Survey.

    Monitoring

    We are monitoring for fidelity of implementation through regular meetings with CA DOJ senior staff, records monitoring, and field observations (ride-alongs). By means of Internet searches and our meetings with CA DOJ, we are monitoring for evidence of contamination—unplanned exposure to APPS—that occurs through media publicity, parallel interventions and requests from communities for intervention outside the allocation scheme. We are also monitoring for adverse events related to APPS enforcement.

    Analysis plan

    The primary analysis will be on an allocation (intention-to-treat) basis. Individuals whose data will be the subject of the analysis will be APPS-eligible persons believed to reside in study communities as of the trial start date. Each community's allocation status (Group 1 or Group 2) will be assigned to all eligible persons believed to reside in that community at that time, along with a time-varying indicator for the initiation of APPS in their assigned community.

    We will conduct a secondary as-treated analysis to take account of the fact that communities may not receive the APPS intervention according to the initial allocation scheme. Our allocation scheme is not binding. Communities remain free to request APPS intervention at any time postrandomisation, and CA DOJ remains free to schedule any community for APPS intervention when circumstances justify doing so in their judgment.

    APPS encounter data will allow us to estimate a dose–response relationship for the ‘intensity’ of the exposure actually received (eg, whether or not contact was made with the subject, whether or not firearms were recovered) and the outcomes of interest. Demographics, criminal history and firearm ownership status for individuals and a wide array of community-level characteristics will be used to adjust for any imbalances between the observable characteristics of the early and later treatment groups, account for contamination or intervention outside assignment status, and support subanalyses. Time-varying covariates will be employed where appropriate.

    The primary endpoint will be an incident arrest for a firearm-related or violent crime. The short follow-up makes final dispositions, such as convictions, an insensitive measure of incident criminal activity; the probability of type I error based on the use of arrest is considered to be substantially less than the probability of type II error based on the use of conviction.24 Relying on convictions would also introduce additional confounders.25–27

    Secondary outcome measures will include incident arrests for crimes of other types, DVROs, and emergency psychiatric hospitalisations for dangerousness to self or others. We will assess community-wide crime rates for violent and firearm-related crimes—as a group and for the specific offences most likely to be affected by APPS—and other crimes.

    We will assess APPS's effects on subgroups defined by the nature of the prohibiting event (ie, criminal conviction, controlled substance abuse, DVRO, psychiatric hospitalisation). Other subgroup analyses will evaluate specific ancillary research questions. For example, we will categorise individuals who are prohibited because of criminal convictions according to their prior criminal histories, to determine whether there are criminal history subgroups for which APPS is counterproductive.

    Our analysis will focus on time to event, using proportional hazards regression, adjusted for the clustered (hierarchical) nature of the data (individuals within communities and communities within regions). The primary predictor will be whether the individual resides in a community assigned at the beginning of the trial to early intervention (Group 1) or later intervention (Group 2). For the primary event-history analysis, APPS-eligible individuals will be the units of analysis and the observation period will be restricted to the calendar period between the initiation of the APPS intervention in Group 1 and the initiation of APPS intervention in Group 2 within each region.

    To assess the validity of the primary analysis, we will undertake sensitivity analyses employing multiple ancillary methods and assessing multiple outcomes. These will include as-treated event-history analyses, where the primary predictor is the actual delivery of the intervention, treated as a time-varying covariate. This approach will permit incorporation of the pretreatment period for individuals who serve as their own historical controls.28 Additional analyses will incorporate data for individuals in study communities who became eligible for the APPS intervention after the trial start date but before the intervention was implemented in their communities.

    Finally, we will compare communities in Groups 1 and 2 before and after exposure on rates of specified crimes, chosen a priori and jointly with CA DOJ, as exemplars for each of the following: crimes most likely to be influenced by APPS; other firearm-related or violent crimes, less likely to be influenced by APPS and crimes involving neither firearms nor violence.

    Power analysis

    Power becomes a problem in group-randomised trials when a small number of groups is included,29 ,30 but that is not the case here. We have randomised 529 communities to Group 1 and 512 to Group 2. We assume that we will have at least 10 000 individuals in each group, observed during at least a 6-month period falling after the intervention for Group 1 and before the intervention for Group 2. Our studies and others suggest that it is reasonable to assume a 6-month incidence of arrest for a firearm-related or violent offence (the primary outcome measure) of between 5% and 10% among persons with recent criminal convictions,2 ,17 DVROs31 or emergency psychiatric hospitalisation.11 ,12 ,32

    Power calculations also depend on the intra-class correlation coefficient (ICC).33–35 ICC is not known for short-term incident criminal outcomes in California. We anticipate ICCs will be lower than 0.02.34 ,36–39 Even under conservative assumptions, we are likely to have excellent power to detect an impact of a 20–30% reduction in the short-term incidence of arrest for violent offences.

    Limitations

    There are exclusions—communities recently subject to the intervention or in remote areas—but their residents account for less than 20% of the state's population, and the study population includes rural communities. Contamination may occur or communities may receive the intervention outside the allocation scheme. Individuals may migrate between communities. We will monitor for these and can account for them in our analyses. In our primary intention-to-treat analysis, such departures will bias effect estimates toward the null. The 24-month period of follow-up does not allow for a full assessment of the intervention's effects, but the natural decrease in risk of recidivism following an index event suggests that benefits attributable to the intervention will be greatest in the near term and decline over time.8–10

    Other limitations concern the data available to APPS. The Automated Firearm System is still largely limited to handgun owners, who must be 21 or older, and does not include transactions before 1996. Reporting of criminal convictions is incomplete. These missing data will most likely reduce the population of affected individuals by causing misclassification of persons eligible for APPS either as non-owners of firearms or as non-prohibited persons. And while arrests are an imperfect measure of criminal activity, we can estimate APPS's effect on offending from its effect on arrests.

    Discussion

    APPS is a unique and now large-scale intervention that is seen as a model for similar efforts in other states. Many states have laws authorising or requiring judges to order firearm recovery when issuing DVROs, or requiring law enforcement officers to recover firearms when serving DVROs.40 Enforcement of these laws is inconsistent and has not been evaluated, however. Results from the only published outcome evaluation of such an intervention were inconclusive.41

    While it is plausible that APPS and other interventions that recover firearms from prohibited persons reduce their risk for violence, their effects are not known. Such interventions could even be counterproductive for important subsets of the population, as appears to be true of mandatory arrest in cases of domestic violence.42 ,43

    APPS's potential benefit to both communities and individuals lies in the fact that, if it reduces risk for firearm-related violence, fewer instances of harm to self and others will occur. Individuals who would otherwise have been victims of crimes (or self-harm) that APPS prevents will be spared injuries, economic losses and perhaps disability or death. Individuals subject to APPS who would have committed those crimes will not face prosecution and incarceration. Healthcare, criminal justice, productivity, and other costs associated with violence that has been prevented will not be incurred. These potential effects are substantial.

    Conclusion

    APPS is a unique intervention that seeks to prevent violence by recovering firearms from prohibited persons. This large, cluster-randomised trial should provide a robust assessment of its effects on important outcome measures under controlled conditions. The results are likely to be generalisable and may have important implications for law enforcement efforts to prevent firearm violence. Further research may be needed to determine APPS's effectiveness as implemented without the constraints imposed by randomisation.

    Acknowledgments

    The authors are grateful to their colleagues at the California Department of Justice, who exemplify public service at its best and without whose continuing collaboration this study would not be possible, and to Pam Keach and Vanessa McHenry of the Violence Prevention Research Program for outstanding management and technical support.

    References

    1. ↵
      Centers for Disease Control and Prevention. Web-based Injury Statistics Query and Reporting System (WISQARS). National Center for Injury Prevention and Control, Centers for Disease Control and Prevention. http://www.cdc.gov/injury/wisqars/index.html (accessed 29 Jul 2016).
    2. ↵
      1. Wintemute GJ,
      2. Wright MA,
      3. Drake CM, et al.
      Subsequent criminal activity among violent misdemeanants who seek to purchase handguns: risk factors and effectiveness of denying handgun purchase. JAMA 2001;285:1019–26. doi:10.1001/jama.285.8.1019
      OpenUrlCrossRefPubMedWeb of Science
    3. ↵
      1. Wright MA,
      2. Wintemute GJ,
      3. Rivara FA
      . Effectiveness of denial of handgun purchase to persons believed to be at high risk for firearm violence. Am J Public Health 1999;89:88–90. doi:10.2105/AJPH.89.1.88
      OpenUrlCrossRefPubMedWeb of Science
    4. ↵
      1. Swanson JW,
      2. Robertson AG,
      3. Frisman LK, et al.
      Preventing gun violence involving people with serious mental illness. In: Webster DW, Vernick JS, eds. Reducing gun violence in America: informing policy with evidence and analysis. Baltimore, MD: The Johns Hopkins University Press, 2013:33–51.
    5. ↵
      1. Barry CL,
      2. McGinty EE,
      3. Vernick JS, et al.
      Two years after Newtown-public opinion on gun policy revisited. Prev Med 2015;79:55–8. doi:10.1016/j.ypmed.2015.05.007
      OpenUrlCrossRefPubMed
    6. ↵
      1. Barry CL,
      2. McGinty EE,
      3. Vernick JS, et al.
      After Newtown—public opinion on gun policy and mental illness. N Engl J Med 2013;368:1077–81. doi:10.1056/NEJMp1300512
      OpenUrlCrossRefPubMedWeb of Science
    7. ↵
      1. Wintemute GJ
      . Support for a comprehensive background check requirement and expanded denial criteria for firearm transfers: findings from the Firearms Licensee Survey. J Urban Health 2014;91:303–19. doi:10.1007/s11524-013-9842-7
      OpenUrlCrossRefPubMed
    8. ↵
      1. Blumstein A,
      2. Nakamura K
      . Redemption in the presence of widespread criminal background checks. Criminology 2009;47:327–59. doi:10.1111/j.1745-9125.2009.00155.x
      OpenUrlCrossRefWeb of Science
    9. ↵
      1. Kurlychek MC,
      2. Brame R,
      3. Bushway SD
      . Scarlet letters and recidivism: does an old criminal record predict future offending? Criminol Public Pol 2006;5:483–504. doi:10.1111/j.1745-9133.2006.00397.x
      OpenUrlCrossRef
    10. ↵
      1. Kurlychek MC,
      2. Brame R,
      3. Bushway SD
      . Enduring risk? Old criminal records and predictions of future criminal involvement. Crime Delinq 2007;53:64–83. doi:10.1177/0011128706294439
      OpenUrlCrossRefWeb of Science
    11. ↵
      1. Lidz CW,
      2. Mulvey EP,
      3. Gardner W
      . The accuracy of predictions of violence to others. JAMA 1993;269:1007–11. doi:10.1001/jama.1993.03500080055032
      OpenUrlCrossRefPubMedWeb of Science
    12. ↵
      1. Newhill CE,
      2. Mulvey EP,
      3. Lidz CW
      . Characteristics of violence in the community by female patients seen in a psychiatric emergency service. Psychiatr Serv 1995;46:785–9. doi:10.1176/ps.46.8.785
      OpenUrlCrossRefPubMedWeb of Science
    13. ↵
      1. Ilgen MA,
      2. Burnette ML,
      3. Conner KR, et al.
      The association between violence and lifetime suicidal thoughts and behaviors in individuals treated for substance use disorders. Addict Behav 2010;35:111–15. doi:10.1016/j.addbeh.2009.09.010
      OpenUrlCrossRefPubMed
    14. ↵
      1. Holt VL,
      2. Kernic MA,
      3. Lumley T, et al.
      Civil protection orders and risk of subsequent police-reported violence. JAMA 2002;288:589–94. doi:10.1001/jama.288.5.589
      OpenUrlCrossRefPubMedWeb of Science
    15. ↵
      1. Jordan CE,
      2. Pritchard AJ,
      3. Duckett D, et al.
      Criminal offending among respondents to protective orders: crime types and patterns that predict victim risk. Violence Against Women 2010;16:1396–411. doi:10.1177/1077801210389680
      OpenUrlCrossRefPubMedWeb of Science
    16. ↵
      1. Karberg JC,
      2. Frandsen RJ,
      3. Durso JM, et al
      . Background Checks for Firearm Transfers, 2013–14—Statistical Tables. Report No: NCJ 249849. Washington, DC: Bureau of Justice Statistics, 2016.
    17. ↵
      1. Wintemute GJ,
      2. Drake CM,
      3. Beaumont JJ, et al.
      Prior misdemeanor convictions as a risk factor for later violent and firearm-related criminal activity among authorized purchasers of handguns. JAMA 1998;280:2083–7. doi:10.1001/jama.280.24.2083
      OpenUrlCrossRefPubMedWeb of Science
    18. ↵
      1. Wright MA,
      2. Wintemute GJ
      . Felonious or violent criminal activity that prohibits gun ownership among prior purchasers of handguns: incidence and risk factors. J Trauma 2010;69:948–55. doi:10.1097/TA.0b013e3181cb441b
      OpenUrlCrossRefPubMed
    19. ↵
      1. Nicas J
      . New push tries to confiscate firearms from felons, mentally ill: gun-confiscation programs start to launch across the country but have their critics. Wall Street J. http://www.wsj.com/articles/SB10001424127887323971204578625622172850706 (accessed 26 Jul 2013).
    20. ↵
      California Legislature. SB 140, Chapter 2, Statutes of 2013, enacted 1 May, 2013.
    21. ↵
      18 United States Code § 922(d).
    22. ↵
      California Justice Statistics Center. Crimes Data File 2003–2012. Office of the Attorney General, California Department of Justice. http://oag.ca.gov/crime/cjsc/stats/crimes-clearances (accessed 31 Jul 2013).
    23. ↵
      California Department of Social Services. The regions of California: recommended grouping of the counties for regional studies. Sacramento, CA: California Department of Social Services, 2002.
    24. ↵
      Bureau of Justice Statistics. Survey of state criminal history information systems, 2010. Washington DC: Bureau of Justice Statistics, 2011.
    25. ↵
      1. Tonry M
      . Punishing race: a continuing American dilemma. New York, NY: Oxford University Press, 2011.
    26. ↵
      1. Unnever JD,
      2. Gabbidon SL
      . A theory of African American offending: race, racism, and crime. New York, NY: Routledge, 2011.
    27. ↵
      1. Stolzenberg L,
      2. D'Alessio SJ,
      3. Eitle D
      . Race and cumulative discrimination in the prosecution of criminal defendants. Race Justice 2013;3:275–99. doi:10.1177/2153368713500317
      OpenUrlCrossRef
    28. ↵
      1. Crowley J,
      2. Hu M
      . Covariance analysis of heart transplant survival data. J Am Stat Assoc 1977;72:27–36. doi:10.1080/01621459.1977.10479903
      OpenUrlCrossRefWeb of Science
    29. ↵
      1. Murray DM
      . Design and analysis of group-randomized trials. New York, NY: Oxford University Press, Inc, 1998.
    30. ↵
      1. Donner A,
      2. Klar N
      . Design and analysis of cluster randomization trials in health research. Hoboken, NJ: John Wiley & Sons, Inc, 2000.
    31. ↵
      1. Wintemute GJ,
      2. Frattaroli S,
      3. Vittes KA, et al
      . Firearms and Domestic Violence Education and Intervention Project. Final Report to the California Department of Justice of Process and Outcome Evaluations. Sacramento, CA: Violence Prevention Research Program, 2012.
    32. ↵
      1. McNiel DE,
      2. Binder RL,
      3. Fulton FM
      . Management of threats of violence under California's duty-to-protect statute. Am J Psychiatry 1998;155:1097–101. doi:10.1176/ajp.155.8.1097
      OpenUrlCrossRefPubMedWeb of Science
    33. ↵
      1. Killip S,
      2. Mahfoud Z,
      3. Pearce K
      . What is an intracluster correlation coefficient? Crucial concepts for primary care researchers. Ann Fam Med 2004;2:204–8. doi:10.1370/afm.141
      OpenUrlAbstract/FREE Full Text
    34. ↵
      1. Murray DM,
      2. Blitstein JL
      . Methods to reduce the impact of intraclass correlation in group-randomized trials. Eval Rev 2003;27:79–103. doi:10.1177/0193841X02239019
      OpenUrlCrossRefPubMedWeb of Science
    35. ↵
      1. Campbell MK,
      2. Thomson S,
      3. Ramsay CR, et al.
      Sample size calculator for cluster randomized trials. Comput Biol Med 2004;34:113–25. doi:10.1016/S0010-4825(03)00039-8
      OpenUrlCrossRefPubMedWeb of Science
    36. ↵
      1. Campbell MK,
      2. Fayers PM,
      3. Grimshaw JM
      . Determinants of the intracluster correlation coefficient in cluster randomized trials: the case of implementation research. Clin Trials 2005;2:99–107. doi:10.1191/1740774505cn071oa
      OpenUrlCrossRefPubMedWeb of Science
    37. ↵
      1. Campbell MK,
      2. Mollison J,
      3. Grimshaw JM
      . Cluster trials in implementation research: estimation of intracluster correlation coefficients and sample size. Stat Med 2001;20:391–9. doi:10.1002/1097-0258(20010215)20:3<391::AID-SIM800>3.0.CO;2-Z
      OpenUrlCrossRefPubMedWeb of Science
    38. ↵
      1. Cox DR,
      2. Wermuth N
      . A comment on the coefficient of determination for binary responses. Am Stat 1992;46:1–4.
      OpenUrl
    39. ↵
      1. Sampson RJ,
      2. Raudenbush SW,
      3. Earls F
      . Neighborhoods and violent crime: a multilevel study of collective efficacy. Science 1997;277:918–24. doi:10.1126/science.277.5328.918
      OpenUrlAbstract/FREE Full Text
    40. ↵
      Law Center to Prevent Gun Violence. Domestic Violence & Firearms Policy Summary. http://smartgunlaws.org/domestic-violence-firearms-policy-summary/ (accessed 30 Apr 2016).
    41. ↵
      1. Wintemute GJ,
      2. Frattaroli S,
      3. Wright MA, et al.
      Firearms and the incidence of arrest among respondents to domestic violence restraining orders. Inj Epidemiol 2015;2:1–11. doi:10.1186/s40621-014-0032-1
      OpenUrlCrossRefPubMed
    42. ↵
      1. Maxwell CD,
      2. Garner JH,
      3. Fagan JA
      . The preventive effects of arrest on intimate partner violence: research, policy and theory. Criminol Public Pol 2002;2:51–80. doi:10.1111/j.1745-9133.2002.tb00107.x
      OpenUrlCrossRef
    43. ↵
      1. Sherman LW,
      2. Harris HM
      . Increased death rates of domestic violence victims from arresting vs. warning suspects in the Milwaukee Domestic Violence Experiment (MilDVE). J Exp Criminol 2015;11:1–20. doi:10.1007/s11292-014-9203-x
      OpenUrlCrossRef

    Footnotes

    • Contributors All authors contributed to the design of the study. GJW drafted the manuscript. All authors provided critical editorial input.

    • Funding This study is funded by grant number 2014-R2-CX-0012, awarded by the National Institute of Justice, Office of Justice Programs, U.S. Department of Justice; and award number 14-6100 from the California Department of Justice. The opinions, findings, and conclusions or recommendations expressed in this publication/programme/exhibition are those of the author(s) and do not necessarily reflect those of the Department of Justice.

    • Competing interests None declared.

    • Ethics approval This study has been approved by the Institutional Review Board of the University of California, Davis (IRB ID 553213-2) and classified as posing minimal risk to study subjects.

    • Provenance and peer review Not commissioned; internally peer reviewed.