Abstract
Twenty-one states and the District of Columbia have enacted Extreme Risk Protection Order (ERPO) statutes, which allow temporary removal of firearms from individuals who pose an imminent risk of harm to themselves or others. Connecticut was the first state to enact such a law in 1999. The law’s implementation and use between 1999 and 2013 were previously described, finding that ERPOs were pursued rarely for the first decade and that most orders were issued in response to concerns about suicide or self-harm rather than about interpersonal violence. The current study analyzes over 1,400 ERPOs in Connecticut between 2013 and 2020 in several domains: respondent demographics, circumstances leading to ERPO filing, type of threat (suicide, violence to others, or both), number and type of firearms removed, prevalence of mental illness and drug and alcohol use, and legal outcomes. Results are similar to the earlier study, indicating that ERPO respondents in Connecticut are primarily White, male, middle-aged residents of small towns and suburbs who pose a risk of harm to themselves (67.9%) more often than to others (42.8%). Significant gender differences between ERPO respondents are discussed, as are state-specific trends over time and differences between Connecticut and other states with published ERPO data.
Gun violence is a serious problem in the United States, accounting for 79 percent of homicides and 53 percent of suicides in 2020.1 Restricting individuals’ access to firearms remains politically polarizing, but in recent years, many states have found compromise in enacting laws that allow temporary removal of legally possessed guns from individuals who pose a danger to themselves or others. As of May 2024, 21 states and the District of Columbia had enacted risk-based firearm laws that temporarily prohibit gun purchase and possession.2 The gun removal orders vary slightly and have different names in their respective states, including Firearm Safety Warrant (Connecticut), Firearm Restraining Order (Illinois), Gun Violence Restraining Order (California), Risk Protection Order (Florida), and Extreme Risk Protection Order (Maryland and others). For the sake of clarity, this article refers to all risk-based firearm removal orders as Extreme Risk Protection Orders (ERPO), the most common term in academic literature on the topic.
On June 29, 1999, Connecticut became the first state to enact an ERPO law; this was in response to a highly publicized mass shooting at the Connecticut Lottery Corporation headquarters.3 Connecticut’s law allows a police officer to petition the court for a “risk warrant” (akin to a temporary ERPO in other states) when the officer has probable cause to believe that an individual, known as the ERPO respondent, poses “a risk of imminent personal injury to himself or herself or to other individuals” (Conn. Gen. Stat. § 29-38c(a)). In practice, a civilian (usually an ERPO respondent’s family member or acquaintance) contacts the police by calling 911, and the police then investigate the matter. Police may remove guns to secure the scene during their investigation, but they need additional legal authority from an ERPO to hold the guns. If two officers believe that an imminent risk exists, they file an ERPO application along with an affidavit describing the respondent’s circumstances and the reason for the ERPO request. The police must also decide whether to intervene with the respondent at the scene by arresting the individual, transporting the individual to a hospital emergency room for evaluation, both, or neither.
Once filed, the ERPO application is reviewed by a state’s attorney and a judge, and the judge determines if a temporary ERPO should be issued. In making this decision, the judge must consider recent threats or acts of violence and recent acts of cruelty to animals (C.G.S. § 29-38c(c)).3 The judge may also consider reckless gun use or display; a history of the use, attempted use, or threatened use of physical force against other persons; prior involuntary psychiatric hospitalization; and illegal use of drugs or abuse of alcohol (C.G.S. § 29-38c(c)).3 After a temporary ERPO is issued, the police serve the warrant and officially seize all firearms and ammunition from the respondent under the authority of the ERPO statute. A judicial hearing follows within 14 days. At the hearing, if a judge finds by clear and convincing evidence that the respondent poses an imminent risk to self or others, a final ERPO is issued, and the respondent’s guns are held for up to a year.
Norko and Baranoski3 and Swanson and colleagues4 previously published studies of ERPO implementation in Connecticut between 1999 and 2013. Since that time, several other authors have published studies of states’ ERPO implementation data including California,5 Washington,6 Oregon,7 Colorado,8 and Indiana.9 Some aspects of ERPO implementation are common across several states, such as ERPOs steadily increasing over time and respondents most often being White men.8,10,11 In Connecticut, ERPOs were used between 1999 and 2013 primarily in response to threats of suicide rather than interpersonal violence, which is similar to findings from Indiana.9 In contrast, data from California,5 Washington,6 and Colorado8 indicate that ERPOs are utilized most often to prevent harm to others. In Oregon, the percentage of ERPOs related to suicidality and threats of violence against others was roughly the same, with most of the sample indicating both risks.8
After adopting ERPO legislation, it often takes years for law enforcement personnel to be trained and to implement methods uniformly across a state.6,9,11 In addition, most states adopted ERPO laws relatively recently, and longitudinal analyses of trends have only been possible in Connecticut and Indiana to date.3,9,10 Thus, our study is an important addition to the literature regarding circumstances and characteristics of respondents, clinically relevant information about respondents’ mental health and substance use, potential self-harm and harm to others, and treatment usage.
Methods
Data were made available to the authors by the Connecticut Department of Mental Health and Addiction Services (DMHAS), which receives information from the courts when a temporary ERPO is granted, pursuant to C.G.S. § 29-38c. DMHAS maintains an electronic ERPO database containing information on respondent demographics, including age, gender, and race and ethnicity, as well as information about the circumstances leading to the ERPO (e.g., suicide risk, violence, substance abuse, mental illness). DMHAS also maintains an electronic repository of ERPOs filed since 2013, including police affidavits, judicial case flow records, and court orders, from which additional data could be extracted. Both of these data sources were utilized in the current study, which was approved by the Institutional Review Boards of Yale University, Duke University, and DMHAS.
The Connecticut data set is part of a larger six-state study examining ERPO implementation and jurisdictional variation.7 Because Connecticut’s data from 1999 and the first half of 2013 had been previously published, and because the original affidavits had been destroyed, the multistate study includes data from January 1, 2013, to June 30, 2020 (the end of the study period across all states). During this period, information about 1,427 temporary ERPOs issued in Connecticut was provided to DMHAS. Twenty of the DMHAS ERPO files contained incomplete or duplicate data, resulting in a total of 1,407 cases for most data analyses (exceptions are noted in the tables and text). Thirteen of the 1,407 cases (.9%) involved respondents who had previously received an ERPO during the study period, resulting in 1,394 unique respondents for the demographic analyses.
Trained research assistants coded information from the ERPO files after achieving sufficient interrater reliability (kappa score of .80) with one of the lead coders. Coded variables included the respondent’s demographic characteristics (age, gender, race and ethnicity), circumstances leading to the ERPO application (suicidality or homicidality, mental illness or cognitive impairment, and drug or alcohol use), what action was taken by the police at the scene (transport of the respondent to a hospital, arrest), and the type and number of firearms removed. For the statistical analysis, no confidence intervals for estimated population parameters were necessary, as the whole population of ERPOs was coded. Descriptive, correlational, and chi square tests for nominal variables were used to analyze data.
Data were later matched with DMHAS’s electronic medical record, which includes information from state-operated and state-funded private treatment agencies, to analyze ERPO respondents’ rates of treatment engagement. Because Connecticut does not keep a central database of gun owners, demographic data about individuals who were issued new pistol permits by the Connecticut State Police’s Special Licensing and Firearms Unit between 2013 and 2019 were compared with ERPO respondents’ demographic data.12 U.S. Census Bureau data for Connecticut13 were used to compare ERPO respondents’ demographics to state population estimates.
Results
Number of ERPOs (1999 through 2020)
Figure 1 shows the increase in Connecticut’s ERPOs since the law’s enactment in 1999. The 7.5-year study period between 2013 and 2020 had significantly more ERPOs (n = 1,407) than the preceding 13-year period, accounting for approximately two thirds of the state’s total ERPOs (n = 2,078). Although ERPOs increased over time, they remained infrequent. In 2019, based on census data and Connecticut’s firearm ownership percentage of 23.6 percent,12 one in 11,928 adults and one in 2,815 gun owners in Connecticut were subject to an ERPO.
Temporary ERPOs Issued in Connecticut from 1999 to 2020.a2020 data are from January 1 to June 30 only.
Initiation of ERPO Process
In 27.5 percent of cases during the study period (n = 387), the respondent’s spouse, dating partner, or parent of a shared child alerted law enforcement of the risk that eventually led to an ERPO. In 16.7 percent of cases (n = 235), another immediate family member called the police. Other people who called the police included licensed health care professionals (5.5%, n = 78), friends (7.1%, n = 101), and coworkers (6.0%, n = 85). In 8.4 percent of cases (n = 118), the respondent themselves called the police, and in 14.7 percent of cases (n = 207), the person who alerted law enforcement was not noted in the file. “Others” comprise the remaining 11.2 percent of people who initiated ERPOs (n = 158); they were most commonly identified as other law enforcement officers (e.g., FBI agents, out-of-state police), neighbors, crisis response line operators, and home health aides.
Respondent Demographics
Table 1 summarizes ERPO respondents’ demographic characteristics. Overall, significantly more men (92.5%, n = 1,289) than women (7.5%, n = 105) were served with ERPOs (n = 1,394; χ2(1) = 1005.64, P < .0001; phi = .849; large effect size). When compared with the gender distribution of gun permits issued in Connecticut from 2013 to 2019,12 ERPO respondents were still disproportionately male (n = 1,407; χ2(1) = 229.58, P < .001; phi = .410); medium effect size). On average, male ERPO respondents were 47.5 years old (range 16.3–98.3, SD = 16.22), and female respondents were 47.6 years old when the ERPO was issued (range 20.7–91.4, SD = 16.20). All respondents under the age of 18 were male (.3%, n = 4), as were all respondents with more than one ERPO during the study period (.9%, n = 13).
ERPO Respondents’ Demographics
Police officers reported a respondent’s race or ethnicity in 51.4 percent of cases (n = 717); some police departments used older ERPO application forms that did not prompt the officer to enter this information. Of respondents with a reported race or ethnicity, 93.9 percent (n = 673) were White, and 4.7 percent were Black (n = 34). Hispanic and Latinx, Asian, Native American and Alaskan, multiple, and “other” ethnicities made up less than one percent each of the sample. There were no significant differences in racial or ethnic distribution between male and female respondents. In comparing ERPO respondents’ race and ethnicity to Connecticut’s population data from the 2020 census,13 White individuals were significantly over-represented, and all other races were under-represented, with a medium to large effect size (n = 717; χ2(2) = 141.52, P < .001, phi = .44). When comparing ERPO respondents’ race and ethnicity with individuals who were newly issued gun permits between 2013 and 2019,12 White individuals were still significantly over-represented, but the effect size was much smaller (n = 717; χ2(2) = 6.42, P < .05, phi = .09). The odds of a White adult in Connecticut being an ERPO respondent was 3.2 times that of nonwhite adult (OR = 3.15; CI (1.38, 7.19)), but the odds of a White gun permittee being an ERPO respondent was only 1.4 times that of a nonwhite gun permittee (OR= 1.44; 95% CI (.62, 3.30)). Because the gun permit data do not identify Hispanic and Latinx as a separate category, more detailed racial and ethnic comparisons were not possible.
We next examined the urban and rural distribution of ERPOs. Connecticut is a small, densely populated state with no large metropolitan areas; its largest city has a population of 149,000. Most inhabitants, 75.2 percent, live in suburban and rural communities with less than 70,000 inhabitants, and 92.5 percent of all ERPOs were issued in those locations. Adjusting for expected cases based on population percentages in small (<70,000 inhabitants) and large towns and cities (>70,000 inhabitants) in Connecticut, significantly more ERPOs were issued in rural and suburban areas than in urban areas of the state (n = 1422; χ2(1) = 223.53; P < .001, phi = .40, medium to large effect size).
Living Arrangements
In 16.7 percent of cases (n =238), the ERPO respondent’s living arrangement was not noted in the file. Of cases with known living arrangements at the time of the ERPO, respondents most often lived with a romantic partner (41.3%, n = 491), 32.9 percent (n = 391) lived on their own, 16.0 percent (n = 190) lived with others (e.g., roommates), 9.3 percent (n = 110) lived with their parents, and .6 percent (n = 7) were unhoused. There were no significant gender differences in the living arrangements of ERPO respondents.
Reason for Firearm Removal
Table 2 describes the reason for removal of guns from ERPO respondents. Suicidality or self-harm was a concern in 67.9 percent of cases (n = 956). Risk of harm to others was a concern in 38.4 percent of cases (n = 540), and there was some overlap between the two groups: 13.5 percent of cases (n = 190) included a risk of harm to self and others. In 5.8 percent of cases (n = 82), no specific concern about harm to self or others was noted in the officer’s affidavit (n = 82). Most of these cases cited another basis for requesting the ERPO such as a history of mental illness or prior psychiatric hospitalization (n = 65), illegal drug or alcohol use (n = 32), reckless use of guns (n = 17), or a combination of these factors. In six cases, no clear reason for the ERPO could be determined based on information in the file.
Reason for Gun Removal
Suicidal Respondents
In 67.9 percent of cases (n = 956), ERPOs were issued because of the risk of self-harm. Among these respondents, 12.1 percent (n = 116) attempted suicide, surviving the attempt. The remainder exhibited suicidal ideation, threats, plans, aborted attempts, or a combination of these. In 3.9 percent of cases (n = 37), the respondent self-harmed without the intent to die, most often by cutting. There were no statistically significant differences regarding self-harm between men and women, nor between the different age groups.
Of cases where the respondent’s intended method of suicide was known through their statements or behavior (n = 375), the vast majority, 83.2 percent (n = 312), intended suicide by firearm. Almost 10 percent (n = 37) intended poisoning or overdose, 4.0 percent (n = 15) use of a sharp instrument, 1.6 percent (n = 6) asphyxiation, 1.3 percent (n = 5) jump or fall from a great height, 4.8 percent (n = 18) “suicide by cop,” and 6.4 percent (n = 24) other methods, including suicide by vehicle or lying on train tracks. The total is greater than 100% because respondents noted more than one method of suicide in some cases.
Violent Respondents
Cases were divided into use of violence (an action that inflicted or could have inflicted harm upon someone else, such as shooting at or punching someone) and threats of violence (communicating a desire or intent to harm someone else, such as saying you want to kill an individual or group), which were coded as mutually exclusive. Multiple-victim and mass shooting threats were included as a subcategory of threats of violence and were defined as threats to kill three people or more, in accordance with a previously published ERPO study by Zeoli and colleagues.7 Harm to animals was included as a subcategory of use of violence and was defined as actual harm to an animal through violent actions or neglect.
As indicated in Table 2, threats or use of violence against others were involved in 38.4 percent of ERPO cases (n= 540). Respondents used violence in 14.9 percent of ERPO cases (n = 210), with a total of 213 different targets. In two cases, the respondent harmed people who fell into more than one of the study’s coded categories, such as an intimate partner and law enforcement personnel. Intimate partners were targeted 40.8 percent of the time (n = 87), other family members 26.3 percent (n = 56), law enforcement officers 9.9 percent (n = 21), and minors 7.0 percent (n = 15). Other targets, such as friends, neighbors, coworkers, places of business, and bystanders, were identified 26.8 percent (n = 57) of the time. The target was an unclear person or group in 9.9 percent of cases (n = 21), and in .9 percent of cases (n = 2), the respondent attempted to harm a person who appeared real to him but was, in fact, a hallucination. Respondents harmed animals in 1.2 percent of cases (n = 17), and all respondents who harmed animals were male.
Respondents threatened violence against others in 27.9 percent of ERPO cases (n = 392). Of these cases, friends, neighbors, coworkers, and places of business (all coded as “others”) were most frequently threatened with violence (53.8%, n = 211). Intimate partners were threatened in 26.0 percent of cases (n = 102), other family members in 15.8 percent (n = 62), and law enforcement officers in 14.3 percent (n = 56). In 6.1 percent of cases (n = 24), the person or group threatened was unclear. Minors were threatened with violence in 4.1 percent of cases (n = 16). In another 4.1 percent of cases (n = 16), the respondent threatened to harm a person he thought was real but was, in fact, a hallucination.
Notable gender differences were observed. Male respondents used or threatened violence (39.9%, n = 519) at a significantly higher rate than female respondents (20.0%, n = 21), with a small effect size (n = 1407; χ2(1) =16.21; P < .001, phi = .10). Firearm violence was used exclusively by men, in 2.4 percent of cases (n = 34). When respondents threatened violence (n = 392), men were far more likely to threaten their intimate partners (26.8%, n = 101) than were women (6.7%, n = 1), and about half these threats involved firearms.
Respondents threatened multiple victims or a mass shooting in 3.3 percent of all ERPO cases (n = 46); all such respondents were male. Most commonly, they threatened multiple victims or mass shootings in a public area (68.9%, n = 31), such as a workplace, school, business, or health-care setting. In 15.6 percent of cases (n = 7), respondents threatened their own home, and in an equal number of cases, a general threat without a specific location was made, such as “I am going to start shooting everyone.” Anger or frustration about the respondent’s treatment at work (22.2%, n = 10) and disputes about business transactions (15.6%, n = 7) were the most common precipitants for mass shooting threats. Anger over health care was the precipitant in 11.1 percent of cases (n = 5), and 6.7 percent (n = 3) of cases involved disagreements with town or local authorities. In 15.2 percent of cases (n = 7), the respondent wanted to “stand his ground” at his home and shoot anyone trying to remove him. In 8.9 percent of cases (n = 4), frustration with school or college precipitated the threat. In 23.9 percent of these cases (n = 11), no specific precipitant for the respondent’s threat was identified.
Alcohol and Drug Use
As noted in Table 3, in 45.3 percent of cases (n = 637), ERPO respondents were reported to have used alcohol or drugs. A total of 28.5 percent of cases (n = 401) involved alcohol use, 10.5 percent (n = 148) involved drug use, and 6.3 percent of cases (n = 88) involved both. Fifty-five percent of ERPOs (n = 770) contained no indication that the respondent had used drugs or alcohol. Of the cases involving drugs (n = 236), opioids were most often used (39.4%, n = 93). Marijuana, inhalant abuse, and vague references to drugs were noted in 20.8 percent (n = 49), 22.5 percent (n = 53), and 24.1 percent (n = 57) of cases involving drugs, respectively. Stimulants and hallucinogens (e.g., methamphetamine, cocaine, psychedelic mushrooms, LSD, PCP) were involved in a total of 16.1 percent of cases (n = 38).
Mental Health Concerns and Alcohol and Drug Use
Mental Illness
In 15.1 percent of cases (n = 212), the ERPO indicated that the respondent had been diagnosed with a mental illness (excluding substance use disorders), as evidenced by a specific psychiatric diagnosis or the prescription of psychotropic medication. Depressive disorders were most common, accounting for 52.3 percent of these cases (n = 111), followed by anxiety disorders (22.6%, n = 48), bipolar disorder (22.1%, n = 47), and PTSD (19.3%, n = 41). In an additional 66.5 percent of cases (n = 936), the police noted a vague concern about mental health such as “[respondent] has been stressed out” or “seems depressed,” but no definitive evidence of a mental illness was presented. The “vague concern” category also included cases in which the police took the respondent to the emergency department for psychiatric evaluation, regardless of whether the ERPO affidavit otherwise noted concerns about the respondent’s mental health. In 6.7 percent of cases (n = 94), respondents engaged in erratic or irrational behaviors, as described in the police affidavit. Cognitive decline was noted in 4.8 percent of cases (n = 67), most of which were attributed to dementia (2.7%, n = 38).
Most ERPO respondents, 57.2 percent (n = 805), had no history of contact with DMHAS, Connecticut’s public mental health care system, either before or after the ERPO. At the time of the ERPO, 7.1 percent of respondents (n = 100) were actively engaged in treatment for a mental health or substance use disorder in the DMHAS system. In contrast to the earlier Connecticut study’s findings,3 ERPOs no longer seem to catalyze respondents’ entry into mental health treatment in the public system. In the current study, an equal number of respondents (7.1%, n = 100) received treatment in the DMHAS system during the year prior to the gun removal as received treatment in the year following the gun removal (7.1%, n = 100). Of the 100 respondents in DMHAS treatment at the time of the ERPO, 32 percent (n = 32) remained in DMHAS treatment one year later. The reasons for leaving treatment are not known and could include completion of a short-term treatment modality, premature cessation of prescribed treatment, or transfer of treatment to a private provider. There were no statistically significant differences between men and women regarding DMHAS treatment, in contrast to the earlier Connecticut study, which found that women were more likely to have a lifetime DMHAS treatment history.3
Law Enforcement Actions at the Scene
Table 4 describes police officers’ actions with ERPO respondents at the scene. In 79.1 percent of cases (n = 1,113), respondents were involuntarily transported to a hospital emergency room for evaluation on a Police Emergency Evaluation Request (PEER), pursuant to C.G.S. § 71a-503. A higher percentage of women (92.4%) were transported to the emergency room (ER) than men (78.0%); this difference was small but statistically significant (n = 1,407; χ2(1) = 12.1, P = .001, phi = .09; small effect size). In addition, 18.5 percent (n = 260) of respondents were arrested at the scene of the ERPO event. A significantly higher percentage of men (19.4%) than women (7.6%) were arrested (n = 1,407; χ2(1) = 8.88, P < .01, phi = .08; small effect size). In most cases (70.8%, n = 996), respondents were transported to the ER but not arrested. In 8.3 percent of cases (n = 117), respondents were concurrently arrested and transported to the ER. In 10.2 percent of cases (n = 143), respondents were arrested and not taken to the ER, and in 10.7 percent of cases (n = 151), respondents were neither transported to the ER nor arrested.
Action Taken by Police at the Scene
Number of Firearms Removed
The number of firearms removed from ERPO respondents varied widely, from 1 to 150 guns, with a mean of 5.85 and median of 3.0 guns, as noted in Table 5. A large minority of respondents (28.7%) had only one gun removed. Removed firearms included a mix of handguns and long guns in most cases. Men had significantly more guns on average (M = 6.06), with higher variability than women (M = 3.16); (t(1,383) =2.96, P = . 03, d = .301, small effect size; 95% CI (1.91, 3.93)). There was a small correlation between older age and an increased number of guns in both men and women (r = .10, R2= .01; P < . 01, small effect size; 95% CI (.019, .124)).
Number of Firearms Removed per Case
Court Decisions
Information about whether a final ERPO hearing was held was available in 88 percent of cases (n = 1,238). In 5.3 percent of these cases (n = 66), no hearing was held (for unclear reasons). Of the 1,172 cases where a final ERPO hearing was held, 50.9 percent (n = 596) of the orders were granted, 9.4 percent (n = 110) were denied, and 12.1 percent (n = 142) were dismissed. In 26.8 percent of cases (n = 314), an “other” outcome occurred, which included dispositions such as transferring the firearms to another owner or relinquishing them for sale or for destruction by the police. The available data do not denote reasons for denial or dismissal of cases. In analyzing which factors were associated with denial versus granting of final ERPOs (e.g., respondent demographics, suicide versus homicide risk, mental health diagnosis, alcohol or drug use, concurrent arrest), only two variables had a statistically significant association: a specific mental health diagnosis is mentioned in the ERPO application (OR = 1.69; 95% CI (1.10, 2.59)) and the respondent is threatening violence (OR = 1.90; 95% CI (1.28, 2.80)).
In most cases, the final ERPO hearing was delayed beyond the statutorily defined 14-day period because one of the parties or the court requested a continuance. In 22.7 percent of cases (n = 319), at least one court continuance was filed. In cases where the number of continuances was known (n = 224), one continuance was filed in 73.7 percent of cases (n = 165), two continuances in 16.1 percent (n = 36), three continuances in 4.9 percent (n = 11), and four or more continuances in 5.4 percent (n = 12). As a result of the continuances, final ERPO hearings occurred, on average, 70.3 days after the temporary ERPO was issued.
Discussion
Connecticut’s relatively long track record with ERPOs allows for analysis of the largest state-specific sample of orders to date. The 762 ERPOs between 1999 and mid-2013 were previously described,3,4 and the current study provides an updated analysis that includes nearly twice as many cases. In some ways, our findings are very similar to the earlier studies. The prototypical ERPO respondent in Connecticut is still a middle-aged man who lives with an intimate partner in a small town and who owns multiple guns. Suicide risk is still the most common reason for ERPOs, with interpersonal violence less common. A history of contact with the state’s mental health system is more the exception than the rule. Overall, ERPO respondents still appear to be “ordinary people in crisis” who temporarily pose a risk to themselves or others. Table 6 compares the earlier studies’ findings3,4 to the updated results, where such comparisons are possible.
Available Comparisons Reported in Three CT ERPO Study Publications
Our current study collected more granular data than the 2014 and 2017 Connecticut studies, allowing us to describe some aspects of ERPOs in more detail. For example, we can now describe the frequency of suicide attempts and actual uses of violence in ERPO respondents, not just threats, finding that respondents acted to harm themselves and others in 12.1 percent and 15.1 percent of cases, respectively. We found that, when ERPO respondents act violently, intimate partners are the most common targets. Intimate partners were not the most common targets of threatened violence, however. The group most commonly threatened are “others” such as neighbors and co-workers, but this is a large and diverse category from which it is hard to draw conclusions. We found that mass shooting threats are rare, despite many ERPO statutes’ passage in response to high-profile mass shootings. When they did occur, the threats were always made by men, most commonly in response to disputed business transactions or work-related stressors.
We also found significant gender differences between ERPO respondents. As in the earlier Connecticut studies and those from other states,6,8,9,14 the vast majority of ERPO respondents are men. Even when the male to female ratio of gun permittees in Connecticut is taken into account, men make up a disproportionately large share of ERPO respondents. We found that female respondents are taken to the emergency room for psychiatric evaluation at higher rates than men, and they are arrested at lower rates. This may be because police officers view the small number of female ERPO respondents as more “sick” than “bad,” referring them for psychiatric evaluation rather than arrest. Another possible explanation is that male ERPO respondents engage in higher rates of threats and violence, which contributes to their higher arrest rates during the ERPO incident.
The earlier Connecticut studies could not analyze data about respondents’ race/ethnicity because these data were not collected at the time. In the current study, nearly half of all ERPO cases were still missing the information, but of the cases where police officers noted a respondent’s race/ethnicity, respondents were overwhelmingly White. All other races were underrepresented when compared with the state’s population and with the state’s gun-permitted population, though the racial disparity is much less pronounced among gun permittees than the general population. The reasons for this disparity are unclear, but it raises concerns about racial equity in the ERPO process. For example, it is possible that people of color simply do not call the police when concerned about family members or others in their community out of fear of negative law enforcement interactions or for other reasons. In a California survey,15 for example, minority participants were less willing to serve as ERPO petitioners, in part out of lack of trust that the system was fair. It is also possible that police officers do not utilize the ERPO process for people of color, instead choosing the criminal-legal system to address the same behaviors that might prompt an ERPO for White individuals. Further study is necessary to understand the root causes of the apparent racial disparity among ERPO respondents.
In the current study, only a small number of ERPO cases, 15.1 percent, included a specific diagnosis or evidence of treatment for a psychiatric condition, while 66.7 percent of ERPO cases involved either a vague note by police officers, such as “seems stressed out,” or the officers took the respondent to the emergency room for evaluation. The latter response may have occurred out of an abundance of caution or out of a desire to pursue a humane alternative to arrest, but it is not possible to describe these cases further given the limitations of the available data. Thus, our data should not lead to the conclusion that ERPOs in Connecticut focus primarily on respondents with mental illness. Substance use remains a significant factor in ERPO cases, accounting for nearly half of all ERPOs. This finding is consistent with both the earlier data from Connecticut3,4 and with data from other states.6,16
Although our data about mental illness are far from conclusive, one difference from the earlier Connecticut studies is noteworthy. In the 2017 study, it appeared that ERPOs served an important “case-finding” function for the mental health system, steering respondents into DMHAS treatment services and contributing to the state’s suicide prevention efforts.4 It is no longer clear that ERPOs operate in this way. In the current study, just as many ERPO respondents dropped out of treatment in the year following the gun removal as entered treatment. This is in stark contrast to the earlier data, which indicated that 29 percent of ERPO respondents received DMHAS treatment in the year following an ERPO, compared with just 12 percent in the year prior.4 It is not clear what accounts for this change over time, but since the DMHAS database includes only data about treatment in the public health system, it is possible that respondents are now seeking treatment in the private sector. Other possible explanations include completion of a short-term treatment modality and premature cessation of prescribed treatment.
Our findings differ from other states with published data in a few areas. For unclear reasons, Connecticut ERPO respondents are significantly older, 47.6 years on average, than ERPO respondents in other states, such as Washington (42.0 years),16 California (41.8 years),5 and Indiana (42.8 years).9 Multiple-victim and mass shooting threats are cited as the reason behind ERPOs two to three times less often in Connecticut (3.3%) than in California (13.2%,), Colorado (11.5%), Florida (12.6%), and Maryland (10.3%).7 In California14 and Colorado,8 the majority of ERPOs were issued because of interpersonal violence threats, which is different from our findings, as well as those in Washington6 and Indiana.9 Rates at which the final ERPOs are granted also vary widely among states. Our study found that 50.9 percent of ERPOs are granted at final order hearings, whereas in California the rate was 32.7 percent,5 62.5 percent in Indiana,9 and 81.0 percent in Washington.6 Less than 10 percent of Connecticut hearings in which the results were known resulted in denial of the ERPO, though many other cases ended with dismissal or respondents’ agreeing to some form of legal transfer of the firearms. Further study is necessary to understand the different outcomes of final order hearings, as well as the factors associated with an ERPO being granted or denied.
Our study has several limitations. We relied on police reports and other court documents describing respondents’ behavior. By definition, police reports contain law enforcement officers’ perspectives on the events leading to the ERPO, which may be biased or may not represent the totality of respondents’ circumstances. For example, police officers’ reports about respondents’ behavior and mental illness are offered from a lay perspective. Their descriptions of race and ethnicity may be based on observations of a respondent’s physical appearance rather than self-identification, and their categorization of a respondent’s gender was always binary, perhaps preventing an opportunity for us to specify suicide risk in transgender and nonbinary individuals. In addition, the level of detail in police reports varied significantly. Thus, the data extracted from the ERPO files is limited by the information documented by law enforcement and court personnel.
Another important consideration is that our study sample includes only cases in Connecticut where a temporary ERPO was granted. It is possible that some unknown number of temporary ERPOs were requested but denied, and it is also possible that these cases differed substantively from the sample described in this study. For example, it is likely that the severity of suicidal behavior and interpersonal violence is greater in our study sample than in the sample of denied petitions. Future studies may include data about temporary ERPO applications that have been denied, but currently, the information is not collected by the courts or reported to other agencies in Connecticut.
Finally, our definition of multiple victims and mass shootings (threats to shoot three or more individuals) is different from much of the published retrospective data about mass shootings, which defines the phenomenon as killing four or more individuals. This difference cannot be avoided because, as noted by Zeoli and colleagues,7 our study sample includes cases where a mass shooting was being predicted, not those where it had already occurred. In most cases, it was simply not clear how many people a respondent intended to harm when making statements like, “I’m going to shoot this place up,” so we chose to include all cases where it was clear that three or more people could be injured by such a threat. For example, if a respondent threatened to shoot a place where at least three other people were present (according to the police report), this was coded as a multiple-victim or mass shooting threat. Because our definition of mass shooting differs from retrospective studies of mass shootings, it should not be used for direct comparison.
Conclusion
ERPO use in Connecticut has steadily increased since 1999, but it remains a rare phenomenon, affecting approximately one in 2,800 gun owners annually. ERPO respondents are most often middle-aged White men who threaten or attempt self-harm rather than interpersonal violence. Mental health concerns and substance use are common in ERPO respondents, but these data must be interpreted with caution. Future studies should seek to understand the true prevalence of mental illness in ERPO respondents, state-to-state differences in ERPO implementation, the reasons for courts’ denying or dismissing final ERPOs, and the underlying reasons for race- and gender-based differences within Connecticut.
Footnotes
Drs. Kapoor, Baranoski, and Norko acknowledge the salary support of the Connecticut Department of Mental Health and Addiction Services (DMHAS). The work described in this article was funded in part by the State of Connecticut, DMHAS, but this publication does not express the views of DMHAS or the State of Connecticut. The views and opinions expressed are those of the authors.
Dr. Norko is involved in the editorial leadership of The Journal; however, he did not participate in any aspect of this article’s review and acceptance.
Disclosures of financial or other potential conflicts of interest: None.
- © 2024 American Academy of Psychiatry and the Law
