Assessment of Extreme Risk Protection Order Use in California From 2016 to 2019

Key Points Question What were the characteristics of extreme risk protection order (ERPO) petitioners and respondents, and what were the spatial and temporal patterns in the policy’s use in California during the first 4 years of its implementation? Findings This cross-sectional study of 1076 ERPO respondents found that most respondents were men and white individuals and that most petitioners were law enforcement officers. Use of ERPOs increased substantially from 2016 to 2019, and there was large variation in uptake between counties. Meaning The results of this study could inform stakeholders involved with ERPO policy adoption, implementation, education, and outreach as well as researchers seeking to further evaluate these policies.


Introduction
Firearm injury and death are major public health problems in the US.In 2018, there were 39 740 deaths from firearms, 24 432 (61.5%) from suicide and 13 958 (35.1%) from homicide. 1 That year, 51% of all suicides and nearly 75% of all homicides involved firearms. 1In California, where firearm deaths result nearly equally from homicide and suicide, there were 3040 such deaths in 2018. 1 Homicide and suicide rates in California vary significantly at the county level. 2

Key Points
Question What were the characteristics of extreme risk protection order (ERPO) Meaning The results of this study could inform stakeholders involved with ERPO policy adoption, implementation, education, and outreach as well as researchers seeking to further evaluate these policies.

+ Invited Commentary + Supplemental content
Author affiliations and article information are listed at the end of this article.Many firearm injury prevention strategies target members of groups considered to be at increased risk of harming themselves or others.For example, federal law prohibits firearm purchase and possession for those convicted of a felony or a domestic violence misdemeanor, subject to a domestic violence restraining order, designated an "unlawful user of or addicted to any controlled substance," "adjudicated as a mental defective or … committed to any mental institution," and those who have renounced their US citizenship as well as fugitives and persons "unlawfully in the United States." 3 Some state laws place prohibitions on additional groups.However, many who harm themselves or others with firearms are not prohibited persons. 3,4e mass shooting in California demonstrated how the existing law failed to prohibit certain dangerous individuals from possessing firearms and provided legislative momentum for California's extreme risk protection order (ERPO) policy.In April 2014, Santa Barbara County sheriff's deputies conducted a welfare check in response to a call from the concerned parents of Elliot Rodger, aged 22 years and living in Isla Vista, California, who was not prohibited from owning firearms.The deputies lacked probable cause to arrest Rodger, to put him on an involuntary mental health hold, or to search his apartment, and they did not check state records of firearm purchases or view the concerning online videos mentioned by Rodger's mother. 5Three weeks later, Rodger killed 6 and injured 14 using a combination of firearms, stabbing, and vehicle ramming in a self-proclaimed act of retribution.

Open
Shortly thereafter, California adopted an ERPO law, similar to existing risk warrant laws in Connecticut and Indiana. 6 California, ERPOs are called gun violence restraining orders (GVROs).Since January 1, 2016, the GVRO policy has allowed for temporary removal of firearms and ammunition from individuals who pose a risk of violence to themselves or others but who are not otherwise prohibited from possessing firearms and ammunition and for whom other risk reduction interventions (eg, arrest, other protective orders, emergency psychiatric hospitalizations) are not available or have failed. 6dividuals subject to GVROs are also prohibited from purchasing firearms and ammunition for the duration of the order.10] Our 2019 analysis of a subset of California GVRO cases 11 found that such laws may also be useful for preventing mass shooting events.That analysis reported 21 cases in which orders were sought with the intent to prevent mass violence. 11No mass violence occurred in those cases, but such data provide only limited support for a claim that ERPOs directly prevent mass violence.

JAMA Network Open | Public Health
Assessment of Extreme Risk Protection Order Use in California, 2016-2019 Publicized cases have demonstrated the breadth of GVRO use, including prevention of interpersonal violence, suicides, and threatened mass shootings.For example, in a suburb of Sacramento, police petitioned for a GVRO in response to a call from a concerned relative and recovered firearms from a man threatening to commit suicide. 12In San Diego, a GVRO was granted for a man with dementia who made threats to shoot his wife and neighbor, believing they were having an affair. 12In another San Diego case, a GVRO was used to recover a semiautomatic rifle with "significant killing capability" from a man who praised a recent mass shooter and made threats to bring his gun to work. 13 of May 2020, 19 states and the District of Columbia have ERPO or similar risk warrant statutes.A better understanding of California's experience may inform policy and practice in California and elsewhere.This study describes California's GVRO respondents and petitioners as well as the temporal and geographic variation in the policy's use during its first 4 years.

Data Sources
Data on GVROs were obtained from the California Department of Justice (CA DOJ) California Restraining and Protective Order System (CARPOS) for the period January

Statistical Analysis
We determined counts and percentages of GVROs and characteristics of respondents and petitioners from 2016 to 2019.Because we presented data for the entire population of respondents, rather than a sample, measures of error, such as confidence intervals, were not calculated.We mapped counts and rates of respondents per county from 2016 to 2019 and analyzed geographic clustering with spatial data analysis. 14To examine spatial dependence, we calculated univariate Moran I using a firstorder Queen contiguity matrix to define neighboring counties (ie, counties are considered neighbors if they share a boundary).We used a permutation test, using 999 random permutations, to determine statistically significant spatial autocorrelation (at 2-tailed α = .05).Significant positive spatial autocorrelation for GVRO counts (or rates) would mean that neighboring counties have issued a similarly high or low number of GVROs (or GVROs per population), indicating that either the use of GVROs in 1 county increases the use of GVROs in neighboring counties or that some exogenous factor that is also spatially clustered is contributing to GVRO use.

Results
There were 1094 respondents to GVROs during 2016 to 2019.Of these, 999 (91.3%) were men and 654 (59.8%) were white individuals (Table (Figure 3).The count of respondents per county roughly matches population size.There was no significant spatial clustering in county rates of GVRO respondents per population (eTable 2 in the Supplement).

Discussion
ERPO and risk warrant policies provide flexible, proactive, and individualized legal tools to temporarily prohibit possession and purchase of firearms and ammunition by individuals who pose a risk of harm to themselves or others but who do not meet existing prohibiting criteria.While both law enforcement and family or household members can petition for GVROs in California, petitioners in the first 4 years were overwhelmingly law enforcement officers.
GVROs were issued more often over time.However, rates of service decreased, and orders do not take effect until they are served.Prior research on restraining has reported high nonservice rates. 17,18Anecdotally, we understand reasons for nonservice include that the situation was resolved prior to service or that another prohibiting civil or criminal mechanism was used, rendering the GVRO unnecessary.We were unable to distinguish between unreported service and nonservice, both of which may vary over time, geographically, and by agency.Further exploration of court records and temporal and geographic variation could help clarify drivers of service rates and reporting within and between counties.
After 4 years of implementation, most counties had issued GVROs.San Diego County drove the increase, with more than 50 times the number of respondents in 2019 (268) than in 2016 (5).San Diego City Attorney Mara Elliott has repeatedly publicly endorsed GVROs, encouraged their use as a firearm violence prevention measure, and funded a team devoted to this effort. 13Her endorsement may help explain the disproportionate increase in use in San Diego, suggesting that local leadership may play a role in local policy use.San Diego has developed and begun implementing law enforcement training in other counties, which may to some extent account for the disproportionate use of GVROs in southern California compared with other areas of the state and explain the geographic clustering of counts of GVROs we observed.
Clustering of county counts of respondents was significant, but clustering of rates of respondents (per 100 000 residents) was not.We believe rates are unstable at this early stage because use is changing rapidly.Furthermore, these rates are potentially misleading because rates could vary, both absolutely and relatively, depending on choice of any of several plausible denominators (eg, a county's total population, population of men, population of firearm owners, or population of firearm owners who are a danger to themselves or others).Despite the increase in orders over time, GVRO uptake in California appears slow relative to other states.Compared with California's 71 respondents in the first year, Oregon, with a population one-tenth of that of California, issued 42 orders in the first 6 months of policy implementation. 19ryland, where more types of individuals are permitted to petition for ERPOs, issued approximately 445 orders in the first 11 months. 20Such discrepancies may owe in part to differences in policies, firearm culture, or more stringent firearm purchasing laws in California, which may result in fewer nonprohibited and at-risk firearm owners relative to other states.Comparative work and further investigation may help with understanding such differences; however, heterogeneity in states' policies should be carefully considered in any multistate analysis.
Evaluations of risk warrant legislation (which predates ERPO laws) indicate that such policies may be primarily effective for suicide prevention. 8,9Estimating a number needed to treat for GVROs in California may be helpful for measuring the policy's effectiveness and for making comparisons among states.Further research will allow understanding of the frequency with which GVROs in California are being used for prevention of self-directed vs other-directed harm.
Although uptake in California may have been slower than in other states, more than 1000 Californians were subject to GVROs.This suggests that the policy filled a gap in existing firearm violence prevention strategies, although these data do not allow us to measure the policy's effects on violence prevention.Additionally, the rapid increase in respondents seen from 2018 to 2019 likely did not reflect a rapid increase in the underlying need for GVROs (ie, a sudden increase in the number of people who were a danger to themselves or others).Instead, the increase may reflect a growing awareness of GVROs, which could be an indication of an emergent upward trend in use as awareness, training, and media coverage continue.
These results may be useful to policy makers, law enforcement officers, county officials, and other stakeholders in California as well as those in states considering or enacting ERPO legislation.
For example, our finding that few GVROs were issued in California before 2019 casts doubt on the recent suggestion that ERPOs in California and other states had major effects on suicide. 21Given that the findings in California may not be generalizable across the country, similar studies of early ERPO use in other states could be useful for comparison with these findings.These findings raise important questions for our continuing process and outcome research, which involves review of case court records, respondent criminal arrests and mortality, and interviews with key stakeholders.For example, why does uptake in California appear slower than in other states?Why are most petitioners law enforcement officers?What explains the variation in county-level use over time?What is the effectiveness of this intervention in preventing interpersonal and self-directed violence?Answering these questions may improve future application of ERPO laws.

Limitations
This study has limitations.Law enforcement or court personnel must report GVROs to CA DOJ for archive in CARPOS, but manual data entry leaves room for error.The data files from CA DOJ contain information only on the last order issued per respondent.Therefore, we were unable to describe multiple orders per person or total number of orders issued.This limitation necessarily resulted in an undercount of emergency and ex parte orders, given that we expect that respondents whose last order was an order after hearing had a prior emergency or ex parte order.Furthermore, some of the increase over time may be attributable to classifying respondents according to their last order, but this is likely minimal considering that the 2019 data, which we received from CA DOJ in a separate and subsequent installment, indicated that only 36 respondents had an order in 2016 to 2018 and another in 2019.
Anecdotal evidence suggests that in some counties, civil or criminal domestic violence restraining orders or other prosecutions that may eventually result in a firearm prohibition are sought concurrently with GVRO petitions.We are unsure of how common this practice of seeking GVROs as a so-called backup to other means of prohibition is and whether it has become more common.
Consequently, we cannot discern whether and to what extent this affected the increase in respondent counts over time.CARPOS does not contain information on other orders or prohibitions, but ongoing work will explore GVRO respondents' co-occurring orders and other prohibitions.
Clustering of GVRO counts may reflect the clustering of counties by population in California rather than the clustering of GVRO use.Nonetheless, we believe that factors aside from county population, such as local champions and available resources, are driving varying county-level GVRO use.Additionally, because these data do not include denied petitions, our analysis may not cover the full scope of California's GVRO use.Because these data do not contain information on risk, respondents' health or criminal histories, targets of harm, and other case details, this analysis can only describe the early use of GVROs and cannot provide insight into the reasons for the policy's use (eg, self-directed or other-directed harm) or the policy's consequences.

Conclusions
This study describes the early utilization of an ERPO law.Petitioners were overwhelmingly law enforcement officers; uptake has increased slowly from 2016 to 2018, with rapid increase in 2019; and local public champions may be driving particularly rapid uptake in some areas.The geographic and temporal variation in use suggests that ERPO uptake, implementation, or need may vary as a function of local, time-varying characteristics, such as awareness of the policy, protocols for implementation, willingness to petition for and issue ERPOs, and the use of other mechanisms for firearm relinquishment (eg, criminal protective orders).The results of this study may inform policy makers and other stakeholders in California and elsewhere involved with ERPO policy implementation, education, and outreach.

Findings
petitioners and respondents, and what were the spatial and temporal patterns in the policy's use in California during the first 4 years of its implementation?This cross-sectional study of 1094 ERPO respondents found that most respondents were men and white individuals and that most petitioners were law enforcement officers.Use of ERPOs increased substantially from 2016 to 2019, and there was large variation in uptake between counties.

Figure 3 .Respondents
Figure 3. Number of Gun Violence Restraining Order Respondents in California Counties, 2016 to 2019 16ughly matches that of firearm owners in California (ie, 64% white, 20% Latinx, 4% black, and 9% other race/ethnicity)15; however, compared with the state as a whole (ie, 38% white, 39% Latinx, 6% black, and 14% Asian),16a larger percentage of GVRO respondents were white individuals and smaller percentages were Latinx or Asian individuals.The mean age of respondents was 41.8 years (range, 14 to 98 years) (Table).Law enforcement officers petitioned for 1053 of 1094 orders (96.3%).The annual number of respondents increased by 910%, from 71 in 2016 to 717 in 2019, with a dramatic ).The racial/ethnic distribution of GVRO respondents JAMA Network Open | Public Health Assessment of Extreme Risk Protection Order Use in California, 2016-2019 JAMA Network Open.2020;3(6):e207735.doi:10.1001/jamanetworkopen.2020.7735(Reprinted) June 18, 2020 3/10 Downloaded From: https://jamanetwork.com/ on 09/16/2023 (24.1%) issued at least 1 in each year since the policy took effect (Figure 2).County-level spatial clustering in GVRO respondent counts was significant (observed Moran I, 0.18, mean [SD] Moran I from reference distribution, −0.02 (0.05); z value, 3.90; P = .005).The observed Moran I statistic (0.18) indicated a weak positive association between the number of respondents in any county and that of its neighbors.Counties in southern California as well as Santa Clara County most frequently Table.Characteristics of Gun Violence Restraining Order Respondents, From the California Restraining and Protective Order System, 2016 to 2019 a Two orders had missing petitioner types in 2016; 1 respondent had missing birthdate.All data are displayed for the last order per person during the entire study period (2016-2019).