Suicide Prevention Effects of Extreme Risk Protection Order Laws in Four States

Firearm Suicide and the Rationale for ERPOs

Features of the epidemiology of suicide in the United States provide a compelling rationale for ERPOs as an important suicide-prevention tool. Suicides in the United States were on the wane in the 1980s through the 1990s, but the 21st century marked a disturbing reversal of that trend.¹ Although mortality from most other causes declined, the suicide rate climbed by more than 30 percent for the next two decades, reaching a record annual high of nearly 50,000 deaths in 2022. Suicide increased fastest in nonurban areas and among younger people.¹ Suicide now ranks as the second leading cause of death among those aged 10 to 34.¹ Easy access to firearms has played a significant role in the trajectory of the nation’s suicide epidemic. Between 2019 and 2021, when the nation saw a large increase in firearm sales and the number of first-time gun owners,⁹ the nonfirearm suicide rate declined whereas the firearm suicide rate increased by eight percent in the population in the United States overall and increased by 18 percent among adolescent to young adult males living in the nonurban South.¹

Many factors may contribute to suicidal acts (neurobiological vulnerabilities, psychological distress, experiences of loss, social pressures and isolation, peer and social media influences, economic strain, and cultural norms), but the enabling factor of access to lethal means often makes the difference between suicidal behavior and suicide mortality.¹⁰ On average, approximately nine out of ten people who attempt suicide survive nonfirearm methods of self-injury,¹¹ and those who survive are unlikely to die from suicide in a subsequent attempt.¹² In contrast, firearm-involved suicide attempts are rarely survivable; only about one out of ten people survive,¹¹ and those few who survive are often left with severe disability.¹³ These statistics highlight a remarkable public health opportunity to reduce the suicide rate through policies such as ERPO laws that limit access to firearms specifically for persons known to be at risk of suicide.

In 2021, an estimated 12.3 million adults in the United States had serious thoughts of suicide, 1.7 million made a suicide attempt,¹⁴ and just under 50,000 died.¹ Although firearms are involved in only a small proportion of suicide attempts (about 5%),¹⁵ they account for a large proportion of suicide deaths (about 52%).¹ From a public health perspective, then, although it is a difficult long-term challenge to mitigate the deleterious causal factors, both distal and proximal, that incline people to self-injurious behavior, a more immediate and effective way to stop so many people from dying is to limit access to firearms for people at manifest risk of harm to self, whatever their motivation. Research suggests that only a small minority of individuals who are prevented from suicide through denial of access to a firearm will substitute an alternative method and eventually die from suicide.¹² Also, as already suggested, the large majority of survivors of suicide attempts die of other causes later in life. Risk does not inevitably lead to harm.

ERPOs are designed as a public health-oriented law that does not restrict the rights of law-abiding gun owners unless they are deemed to be a danger to themselves or others.¹⁶ ERPOs fall within a small class of constitutionally permissible, politically tenable, narrowly focused legal or regulatory schemes that can disqualify individuals from purchasing or possessing firearms (for example, having a record of involuntary civil commitment or a felony criminal conviction).¹⁷ A key difference between ERPOs and these other laws is that ERPOs impose a short-term disqualification whereas the others typically confer lifetime disqualifications. These other gun-prohibiting criteria have long been recognized as legitimate constraints on the right to bear arms as set forth in Justice Antonin Scalia’s opinion in District of Columbia v. Heller (2008), to the effect that “the right secured by the Second Amendment is not unlimited” (Ref. 18, p 626).

As of May 2024, 21 states and the District of Columbia have enacted ERPO laws. All states’ ERPO statutes authorize law enforcement to initiate an ERPO.¹⁹ Some states also authorize family members, clinicians, educators, or other categories of individuals to initiate an ERPO by petitioning a court directly when they observe someone engaging in behaviors demonstrating risk of firearm violence. Most ERPO statutes provide a two-step legal process beginning with an ex parte order for firearm removal in exigent or emergency circumstances, followed by a timely opportunity for a court hearing between seven days and one month following the temporary ERPO, depending on the state. The respondent may be represented by a lawyer in ERPO proceedings, although most states do not pay for a respondent’s lawyer.

At an ERPO hearing, the state has the burden of proving that the individual continues to pose a substantial risk of harming self or others. If a judge finds such evidence sufficient to grant the order according to statutory criteria, a local law enforcement agency or licensed firearm dealer retains the ERPO respondent’s firearms for no longer than a year, in most states. An ERPO also prohibits the respondent from purchasing firearms for the duration of the order. After an ERPO has been issued, a respondent may request a hearing to reconsider the order. The terms of such requests vary among the states.

ERPOs are civil orders that carry no criminal implications unless they are violated. ERPOs were initially designed as a risk-based public safety intervention that could be applied most appropriately to people without criminal legal involvement. In practice, however, ERPOs are being used in different kinds of cases in conjunction with criminal law enforcement and prosecution. Specifically, ERPOs have been used in response to criminal incidents involving firearms, cases where police make an arrest or have probable cause to make an arrest, and where prosecutors exercise discretion in proceeding with a criminal case or withhold or withdraw charges. In this regard, ERPOs can be seen in different contexts as either a form of criminal diversion or enhanced law enforcement. There are legal and public health policy arguments for and against the use of ERPOs in these ways. More research is needed on both their effectiveness and fairness.

ERPO laws were designed to provide robust due process protections, including the requirement for a timely court hearing following an ex parte order, the opportunity for the respondent to be represented by a lawyer, and the state having the burden of proof, typically by the standard of clear and convincing evidence. Such due process protections are considered especially important for ERPOs to the extent that these laws regulate and restrict individuals’ rights based on judgements of future risk and are not a sanction for prior injurious behavior. Legal scholars Joseph Blocher and Jacob Charles examined the features of ERPO laws in relation to constitutional requirements for prehearing deprivation of rights, Supreme Court guidelines for seizures prior to a full hearing, standard of proof principles, and burden of proof in the context of extreme risk; they concluded that ERPOs fit well within the established constitutional due process framework.¹⁶

Subsequently, in the aftermath of the Supreme Court’s 2022 decision in New York State Rifle and Pistol Association v. Bruen,²⁰ new legal challenges to ERPOs arose, especially regarding due process concerns with gun seizure under an ex parte order prior to a hearing. Following the Bruen opinion, lower courts must now utilize a new method of scrutiny, viewing such challenges solely through the lens of the “text, history, and tradition” of the constitution and thus by appealing to historical analogs or precedents in past laws. This new test and diverging interpretations of its applicability to various firearm restrictions brought a season of uncertainty in gun rights jurisprudence, both with respect to the Second Amendment and due process protections^19,21.

Most recently, in its first post-Bruen gun rights case, United States v. Rahimi,²² the Supreme Court held that when an individual has been found by a court to pose a credible threat to the physical safety of another, that individual may be temporarily disarmed consistent with history and without running afoul of the Second Amendment. The Rahimi decision reaffirmed the constitutionality of federal laws restricting access to firearms for respondents to domestic violence restraining orders issued after notice and a hearing. While the opinion did not explicitly address ERPOs or ex parte proceedings, its reasoning suggests that risk-based restrictions adjudicated in a civil court proceeding can be constitutionally permissible, especially when based on a court’s individualized determination of a credible threat. Although the Rahimi decision also did not address the potential due process concerns that some critics of ERPOs have raised, academic legal analysis has concluded that ERPOs fit comfortably within the constitution’s protections of an individual’s right to a hearing prior to a lasting deprivation of a protected interest.¹⁶

Table 1 summarizes key statutory features of ERPO laws in the four states that were included in the study.

By legal design, the qualifying criteria for an ERPO do not require a mental illness diagnosis and with good reason; ERPOs focus on emergent indicators of risk of imminent harm to self or others, irrespective of the cause or source of the risk. Despite stigmatizing public opinion to the contrary,^23,24 the large majority of people with diagnosable mental illnesses are not at risk of violent behavior directed at others as a result of their illness.²⁵ That being said, acute mental illness is a significant risk factor for suicide, and a substantial proportion of ERPO respondents are described in ways that suggest, implicitly or explicitly, that they are experiencing a mental health crisis. A Connecticut study found that law enforcement officers detained and transported 55 percent of ERPO respondents to a hospital emergency department, ostensibly for evaluation and treatment of an acute behavioral health condition.²⁶ Moreover, 46 percent of respondents in the study had a diagnosable behavioral health disorder recorded in the state’s public behavioral health system; three out of four of those with a matched record had a mental health diagnosis, either alone or co-occurring with a substance use disorder, and the remainder had a substance use disorder only.²⁶

Critics of ERPOs sometimes lament that an ERPO does nothing to address the underlying causes of dangerous behavior but merely mitigates risk associated with exposure to firearms. For their part, proponents of ERPO laws describe this focus as appropriate; by reducing firearm exposure specifically for people at high risk, the chance of a lethal outcome is lessened and an opportunity opens for other interventions that can address a person’s motivation for self-harm. Evidence from ERPO implementation studies shows that, when law enforcement serves an ERPO and finds the respondent to be in a mental health crisis, they often proceed to deliver the individual to a hospital emergency department for evaluation and continuing treatment, as needed.²⁶ A longstanding body of state laws, separate from ERPO laws, provides police officers with the authority to detain and transport an individual who shows signs of mental disorder and dangerousness.²⁷ In many cases, then, ERPOs not only can provide a second chance at life but a window of time and opportunity for respondents in crisis to access therapeutic or other interventions that could address their underlying motivations for suicide.

An emerging body of research has described the implementation of ERPO laws in a handful of states and local jurisdictions, but effectiveness studies have been limited.^28,–,32 Early studies examined suicide outcomes in small samples of individuals exposed to risk-based firearm-removal laws in Connecticut (n = 762) and Indiana (n = 395).^26,33 These studies linked ERPO respondent databases to state death records and identified individuals who had later died of suicide, whether by firearm injury or other methods. Using a counterfactual analysis based on case fatality rates (CFRs), these studies estimated that for every 10 to 20 firearm removal actions, one life was saved by averting a suicide.

Kivisto and Phalen later compared trends in population-level suicide rates in Connecticut and Indiana with trends in synthetic control states without ERPO laws. Using a quasiexperimental analysis, these researchers estimated that Indiana’s gun removal law was associated with a 7.5 percent reduction in gun suicides over 10 years and that Connecticut’s law was associated with a 13.7 reduction in the “post-Virginia Tech period.”³⁴ A commentary by Swanson discussed the uncertainty of this type of study’s conclusions with respect to population-level outcomes, given the relatively small number of individuals actually exposed to the law’s restrictions.³⁵

Another synthetic control study in San Diego County did not find a significant effect of ERPOs in reducing population rates of violent crime and suicide.³⁶ Still, given that only a small number of ERPOs had been issued (355 orders over a period of four years in a population of 2.6 million), it is hardly surprising that statistically significant population-level effects could not be detected.³⁷

For this article, we used the same analytic method that Swanson and collaborators described in their studies conducted in Connecticut and Indiana.^26,33 The advantage of the current study is that the estimates of effectiveness of ERPOs are based on a much larger and more diverse sample of ERPO respondents combined across four states.

Method

This study is part of a larger project³² analyzing ERPO use and outcomes associated with ERPO laws in California (Cal. Penal Code § 18,100–18,205), Connecticut (Conn. Gen. Stat. § 29-38c), Maryland (MD Code, Public Safety § 5–6), and Washington (RCW 7.94). These states were selected to reflect at least some of the geographic and statutory diversity that characterizes ERPO states and to take advantage of access to ERPO data in certain states where academic institutions and prominent researchers in gun violence prevention and policy had existing working relationships with each other and with state entities empowered to provide such data to researchers. Colorado and Florida were included in the larger study but excluded from the current analysis of suicide outcomes owing to limitations of available data. In Colorado, there were not enough ERPO cases to support a valid study of suicide outcomes. In Florida, we were unable to obtain death record data. In California and Washington, public records requests allowed us to obtain ERPO casefiles. The Connecticut data were made available to us through a collaborative arrangement between the site investigators and relevant state agencies. In Maryland, we were granted access to nonpublic ERPO court records. All requests were made for ERPO cases through June 30, 2020, but the start date varied by state, depending on when the law went into effect.³²

We obtained death records from each state’s vital records repository (typically housed in the Department of Health) to identify ERPO respondents who had died. In California, we used Lexis Nexis to identify decedents and then obtained their death certificates to determine cause of death. For decedents who were matched in the search, we merged variables containing the date and cause of death with the larger study database. For suicide cases, we also obtained the specific mechanism of injury.

Trained research associates coded ERPO petitions and the attached supporting affidavits to classify descriptive information about the respondents, including the reasons that an ERPO had been requested from the court. Each research associate double-coded petitions with a lead coder until they reached an interrater reliability score of .80. This study protocol was approved by the institutional review boards at Michigan State University, Johns Hopkins Bloomberg School of Public Health, University of Washington, University of California Davis, Yale University, Duke University, and the Connecticut Department of Mental Health and Addiction Services.

We followed the method of counterfactual analysis that was used in Swanson and collaborators’ previous single-state studies of ERPOs’ effectiveness in preventing suicide.^26,33 Specifically, we used published case fatality rates (CFRs) for each method of suicide to extrapolate the number of nonfatal suicide attempts that had likely occurred to produce the corresponding number of recorded deaths among ERPO respondents. Published national CFRs were as follows: ingesting poison or cutting or piercing with a sharp object, one percent fatality; intentional drug overdose, two percent; jumping from a high place or in front of a moving vehicle or object, 28 percent; gas inhalation, 31 percent; hanging, 53 percent; drowning, 56 percent; and self-inflicted firearm injury, 90 percent fatality.¹¹

We then used the CFR-extrapolated estimates of suicide attempts to create an alternative data array under the counterfactual assumption that these respondents had not been subject to an ERPO and had access to firearms. The analysis also relies on an evidence-based assumption regarding the probability that, in the absence of ERPO, the average ERPO respondent who attempted suicide would have used a firearm rather than some other method of self-injury. For this probability, we applied our previously published estimate of p = .39.²⁶ That probability estimate was initially calculated from a state-level regression analysis of its linear association with states’ household firearm possession rate.

We applied that result to the counterfactual array to estimate the number of excess fatalities that could have been expected in the absence of gun removal and ultimately the number of granted ERPO petitions needed to prevent one suicide. We conducted the analysis first for the entire pooled study population and then for the subset of 2,850 ERPO respondents with a documented suicide concern as part of their initial ERPO petition.

We also calculated the number of ERPOs needed to prevent one suicide using an alternative estimate of the probability that a male gun owner who attempts suicide will use a gun, an estimate that does not rely on a state-level ecological correlation. In a recent article, Matthew Miller and his colleagues recalculated the probability in question by applying CFRs to a database of suicide death records pertaining to a large number of known handgun owners in California.³⁸ They calculated the probability to be approximately 28 percent. In what follows, we present the estimate of the number of ERPOs needed to prevent one suicide using both of these probability assumptions (our original 2017 estimate and Miller and colleagues’ updated version) set within a range of possible estimates on a curve.

In addition, we calculated the ratio of firearm-involved suicides to all suicides (FS/S) among ERPO respondents and compared that with an evidence-based counterfactual that could have been expected in the absence of an ERPO. In general, higher proportions of gun suicides are associated with higher absolute suicide rates, because of the high lethality of firearm suicide attempts relative to attempts with other methods.¹¹ Because the large majority of ERPO respondents are male gun owners,^{26,30,33,36,39} an informative comparison for the counterfactual can be made to suicide decedents in largely male gun-owning populations. We used data on suicide methods reported by the Centers for Disease Control and Prevention (CDC) for males in five states in the South and Mountain West with the highest rates of firearm ownership: Alabama, Louisiana, Mississippi, Montana, and Wyoming. In those populations, approximately 70 percent of male suicides involve firearms.¹

Results

The study population of ERPO respondents was drawn from California (n = 1,386), Connecticut (n = 1,407), Maryland (n = 1,347) and Washington (n = 442), for a total of 4,583 ERPO respondents in the pooled database. The petitioners in these cases were as follows: Connecticut, 100 percent law enforcement; California, 97.6 percent law enforcement; Maryland, 56.8 percent law enforcement, 31.7 percent intimate partner, 7.4 percent other family, 3.6 percent “other”; and Washington, 88.8 percent law enforcement, 5.3 percent intimate partner, 4.4 percent other family.

In the pooled study population, 45 individuals went on to die by suicide, 20 by firearm injury and 25 by other methods. Ninety-one percent of all the suicide decedents were male (44 of 45), and 97 percent of the firearm suicide decedents (24 of 25) were males. The average age of respondents in the study was 43.7 years. By extrapolating from published national case fatality rates (CFRs) for each method of suicide recorded,¹¹ we estimated that there were approximately 22 firearm suicide attempts (.44% of the study population) at 90 percent lethality resulting in the 20 firearm suicide deaths observed. Using the same method, we estimated that there were 794 suicide attempts by other methods (15.1% of study population), ranging from one percent to 42 percent lethality across methods used, resulting in 25 other suicide deaths (see Table 2).

Following the method described by Swanson and collaborators,²⁶ we initially assumed a probability of p = .39 that a male gun owner who attempts suicide will use a gun in his attempt, and we further assumed that, but for the ERPO, the respondent would have had access to firearms. Accordingly, we reassigned 39 percent of counterfactual suicide attempts for each method of suicide to firearms with a CFR of 90 percent and the remaining 61 percent to the less lethal method observed for each case as reported. Recalculating from the counterfactual CFR distribution the number of deaths that hypothetically would have occurred without the ERPO, we estimated 315 hypothetical expected deaths in the counterfactual. Subtracting the 45 observed deaths, we estimated that there were 269 averted suicide deaths attributable to the ERPO intervention. Dividing the total number of ERPOs in the pooled four-state study population by the estimated number of lives saved, we calculated that, for every 17 ERPOs, one potential suicide death was prevented, as shown in Table 2.

We conducted the same analysis on the smaller group of ERPO respondents in which there had been an indication of suicide ideation or attempt coded in the ERPO petition. Specifically, there were 2,850 ERPO respondents with a baseline suicide concern (with or without violent threats directed at others as well) in the pooled database. Within that subgroup, our death record match identified 19 firearm-involved suicides and 19 suicides by other methods, totaling 38 suicide deaths. Using the aforementioned method, we estimated that there were 21 firearm suicide attempts at 90 percent lethality resulting in the 19 firearm-involved deaths. Likewise, we estimated that there were 630 suicide attempts by other methods combined, with an average of three percent lethality resulting in 19 observed nonfirearm suicide deaths.

Recalculating from the counterfactual CFR distribution the number of deaths that hypothetically would have occurred without an ERPO in the subgroup with a suicide concern, we estimated 252 hypothetical expected deaths in the counterfactual compared with 38 observed deaths in the data, for an estimated 214 averted suicide deaths. As before, we divided the total number of ERPOs in the subgroup by the estimated number of lives saved in this group to calculate our estimate that, for every 13 ERPOs issued in cases involving a known suicide risk, one life was saved.

Using Miller and colleagues’³⁸ alternative estimate of the probability that a male gun owner who attempts suicide will use a gun, the estimated number of ERPOs needed to avert one suicide in these data are 24 for the whole sample and 18 for the subset with a suicide concern mentioned in the petition. Figure 1 displays the curvilinear relationship between the assumed probability that a male gun owner who attempts suicide will use a gun and the number of ERPOs needed to prevent one suicide. The higher curve represents all ERPO respondents in the study, and the lower curve represents the subset of ERPO respondents for whom a suicide concern was documented in the ERPO petition.

Figure 1. Association between the probability that a male gun owner who attempts suicide will use a firearm in the attempt and the estimated number of ERPOs needed to prevent one suicide death: evidence from 4,583 ERPOs in four states. ERPO = extreme risk protection order.

As further evidence consistent with ERPO effectiveness in suicide prevention, Figure 2 displays the close correspondence between the state-level firearm possession rate and the proportion of male suicides that involved firearms (FS/S) in five states with the highest rates of gun ownership,³⁹ compared with ERPO respondents, who are depicted as having 100 percent firearm ownership prior to their ERPO. Even though not every ERPO case resulted in a gun removal, most of them did, and the assumption justifying the ERPO is that the person could have obtained access to a gun without the ERPO. The gap between the assumed pre-ERPO (or without ERPO) gun ownership rate and the post-ERPO FS/S ratio is consistent with the ERPO having an effect by removing firearms from individuals at risk of suicide. That the FS/S ratio was still 44 percent suggests that ERPO respondents, as a group, were not entirely deterred from accessing firearms in making a suicide attempt. There may also have been problems in effectively removing firearms in some cases.

Figure 2. Firearm ownership rate and proportion of male suicides that involve guns in selected high gun ownership states, U.S. adult males, and in ERPO respondents. ERPO = extreme risk protection order; FS/S = firearm suicides to all suicides.