Characterizing Arrests and Charges Among Individuals With Serious Mental Illnesses in Public-Sector Treatment Settings

Approximately one million jail bookings in the United States every year involve persons with serious mental illnesses (1), and such individuals are overrepresented at all stages of the criminal legal system. Those with serious mental illnesses are more likely to get arrested (1), are more likely to receive a jail sentence for misdemeanors (2), spend more time in jail, and experience greater recidivism after release (1, 3). The most recent national data estimate that 26% of people incarcerated in jail and 14% of people incarcerated in prison meet criteria for serious psychological distress (4). Further, 44% and 37% of people in jail and prison, respectively, have been told in the past by a mental health professional that they have a mental disorder (4). The overrepresentation of individuals with serious mental illnesses in the criminal legal system has been linked to myriad factors, including deinstitutionalization, inadequate funding of community-based mental health services, police officers’ limited options for resolving situations, and a broad array of social disadvantages that both people with serious mental illnesses and people committing crimes are more likely to experience (e.g., poverty, unemployment, and housing instability) (3, 5).

The nature of criminal legal system entanglement is complex. One study of legal involvement among more than 600 individuals with schizophrenia found that being a victim of crime was the most common type of involvement (67%), followed by being on parole or probation (26%), being arrested for assault (13%), having other miscellaneous encounters not resulting in arrest (13%), being cited for a major driving violation but without arrest (11%), being arrested for parole or probation violation (10%), and being charged with disorderly conduct (9%) (6). A study of 13,851 persons with serious mental illnesses found that 27% were arrested in a 10-year period, almost exclusively for minor crimes, such as property offenses, crimes against public order (e.g., disturbing the peace), public indecency, and motor and drug offenses (7). Property offenses, including trespass, have been reported as the most common charges among individuals with serious mental illnesses, followed closely by alcohol or drug possession and disorderly conduct (7–10). Among 99 mental health court participants in North Carolina, 90% of offenses in the 2-year period before court entry were misdemeanors; the remaining 10% were felonies, 83% of which pertained to theft. Among 133 individuals with serious mental illnesses and a history of incarceration who were being treated in a community mental health center in Atlanta, common self-reported misdemeanor offenses included property crimes and crimes against public order (11).

Taken together, data from available research suggest that most arrests in this population are for minor charges that may be appropriate for prearrest diversion programs that focus on connecting people to behavioral health and other social services. At the same time, the handful of available studies have relied on data that are at least 15 years old, and several lack detailed charging information. In order for upstream interventions to be best informed and most effective, we need a clearer, current understanding of what individuals with serious mental illnesses are being charged with and which sociodemographic or clinical risk factors increase individuals’ risk of arrest.

This analysis used Record of Arrest and Prosecution (RAP) sheets from a sample of patients with serious mental illnesses in southeast Georgia to characterize the nature of lifetime arrests and charges. First, for descriptive purposes, we determined the proportion of patients who had ever been arrested, and among those who had, we determined the number of lifetime arrests and number of lifetime charges. Second, we examined associations of number of arrests with sex, race, educational attainment, current homelessness, psychiatric diagnostic category, and presence of a co-occurring substance use disorder. We hypothesized, on the basis of prior literature in both the general population and among persons with serious mental illnesses, that the number of arrests would be predicted by male sex and African American race (12, 13), lower educational attainment (11, 14), homelessness (15, 16), having a psychotic (as opposed to a mood) disorder (17), and having a co-occurring substance use disorder (15, 16, 18). Third, we examined the most common specific charges for the earliest two arrests and the most recent two arrests to determine whether the types of charges differed. Such exploratory work is necessary to better understand how “accumulation of criminal justice involvement” (19) affects criminal legal, social, health, and mental health outcomes among individuals with serious mental illnesses.

Methods

Setting and Sample

Data for this secondary analysis came from 240 participants recruited as part of a parent project evaluating a new model of recovery-oriented community navigation. Recruitment for the parent study took place between December 2014 and June 2018 at three inpatient psychiatric facilities in southeast Georgia—a state psychiatric hospital and two crisis stabilization units. Most participants (N=198, 82%) were recruited from two settings in Savannah, and 42 (18%) were recruited from a unit in Brunswick. The Savannah metropolitan statistical area has a population of 389,494 (55% White, 32% Black or African American, and 7% Hispanic or Latino), with a median age of 36.1, a median household income of $58,178, and a poverty rate of 16% (20). The Brunswick metropolitan statistical area has a population of 115,939 (68% White, 23% Black or African American, and 5% Hispanic or Latino), with a median age of 41.9, a median household income of $44,887, and a poverty rate of 20% (20).

Eligibility criteria for this analysis were the same as for the parent study and included the following: 18–65 years of age, diagnosis of a psychotic or mood disorder, having had two inpatient psychiatric admissions within the past 12 months, inability to complete activities of daily living or social role functioning in at least two areas (e.g., navigating services, caring for personal business affairs, and obtaining or maintaining employment), absence of known or suspected intellectual disability or dementia, and having capacity to give informed consent.

Procedures and Materials

Participants were referred to the parent study from clinicians at the three recruitment sites. After written informed consent was obtained by following processes approved by the New York State Psychiatric Institute’s and other institutional review boards, an assessment lasting 2–4 hours was conducted for the parent project in the inpatient setting in the several days prior to discharge. Clinical research diagnoses for psychotic and mood disorders, and substance use disorders, were made with the Structured Clinical Interview for DSM-5 Disorders (21).

The Georgia Bureau of Investigation’s Georgia Crime Information Center (GCIC) provided RAP sheets, with participants’ consent. GCIC receives monthly crime and arrest reports from >600 state and local law enforcement agencies. Information on arrests and charges is stored in the GCIC’s crime database and is summarized in the RAP sheet. We extracted the following information from each participant’s RAP sheet: number of lifetime arrests, number of lifetime charges, and whether any charge within an arrest was a felony (otherwise, the arrest was classified as a misdemeanor, unless it was an unclassified charge type). Additional data were extracted for each individual’s earliest two arrests and most recent two arrests, including all charges within the arrest event.

Data Analysis

Descriptive statistics were examined. Poisson regression—a generalized linear model form of regression analysis used to model count data—was conducted to determine associations between number of arrests and sex, race, educational attainment, current homelessness, psychiatric diagnostic category, and presence of a co-occurring substance use disorder, while controlling for age (given that older patients would have had more time across the lifespan to be arrested). All analyses were conducted by using IBM SPSS, version 25.

Results

Sociodemographic and Clinical Characteristics of the Study Sample

Sample characteristics are shown in Table 1. The mean age was 35.9, and 65% of participants were male. The vast majority were non-Hispanic (95%), with roughly half being White (48%) and half being Black or African American (48%). The mean years of education was 11.0, most were single and never married (62%), and most were unemployed (87%). Roughly two-thirds were diagnosed as having a psychotic disorder (65%), and the remainder had a mood disorder (35%). The proportion with a co-occurring substance use disorder was 61%.

Charges and Arrests in the Study Sample

Among the 240 participants, 171 (71%) had been arrested in Georgia at least once, and among those, the mean±SD number of lifetime arrests was 8.6±10.1 (median=6) and the mean number of lifetime charges was 12.6±14.6 (median=7). Ninety-eight participants (41% of the overall sample, and 57% of those who had ever been arrested) had a felony charge in their history. Arrest dates ranged from 1976 to 2018.

Associations Between Number of Arrests and Six Variables

In the Poisson regression on number of arrests, age was included as a covariate in the model. As shown in Table 2, of the six variables of interest included in the model (in addition to age), only currently experiencing homelessness was not an independently significant predictor when the analysis controlled for the effects of the other variables. Educational attainment was the strongest predictor, with a risk ratio (RR) of 2.96. Other meaningful predictors were African American race (RR=1.54), the presence of a substance use disorder (RR=1.20), and the presence of a mood disorder (RR=1.36). Contrary to our expectations, female sex was associated with more arrests (RR=1.40), when the analysis controlled for the other variables.

Most Common Charges

To examine specific charges, among participants with four or more lifetime arrests (N=99), we extracted specific charge codes for the first two arrests and the most recent two arrests. A ranking of the 15 most common charges from this extraction is provided in Table 3. Charges with a frequency of >5% included criminal trespass (the most common charge, at 7% of all charges), willful obstruction of law enforcement officers (7% of all charges), disorderly conduct (6%), and shoplifting (6%).

Table 4 gives rankings of the charges with a prevalence greater than 2% within the first and second arrests (which occurred at mean ages of 19.6±4.2 and 21.3±4.8, respectively) and within the most recent two arrests (which occurred at mean ages of 33.4±10.3 and 34.9±10.6, respectively). The earliest arrests included charges for marijuana possession and driving under the influence (15% of all charges), followed by shoplifting and burglary (12% of all charges), whereas the most recent charges included probation violations (11% of all charges), failure to appear for a court appointment (5%), and officer obstruction and giving false information (14%). Driving under the influence became much less frequent between the earliest and most recent arrests (7% of charges, compared with 1%), and the same was true of marijuana-related charges (13% of charges, compared with 2%). On the other hand, disorderly conduct became much more common in the most recent arrests (9%, compared with 3%), as did charges related to officer obstruction and giving false information (14%, compared with 6%).

Discussion

At least five findings are noteworthy. First, 71% of a sample of patients recruited from three inpatient psychiatric facilities in Georgia had been previously arrested in Georgia, which means that the actual percentage would likely have been higher if the analysis had included data from other states. This high rate is consistent with prior literature involving individuals with serious mental illnesses (1, 15). Second, among those who had been arrested, the average number of lifetime arrests in Georgia was 8.6, and the average number of lifetime charges was 12.6. Third, among those ever arrested in Georgia, almost half (43%) had never been charged with a felony, and most of the charges were for misdemeanor or unclassified offenses. As policy change pertaining to criminal legal system reform (such as bail reform for misdemeanors) proceeds, the special needs of those with serious mental illnesses—who are at high risk of persistent inequities—must be intentionally and strategically addressed so that they can share equitably in the benefits of reform. Furthermore, racial inequities must be at the forefront of reform, not only with regard to the general population but also among those with serious mental illnesses.

Fourth, having lower educational attainment (<12 years), being female, being Black or African American, the presence of a co-occurring substance use disorder, and the presence of a mood disorder were associated with a greater number of arrests. The finding of more arrests among Black or African American participants was likely a reflection of the impact of structural racism across multiple systems (e.g., housing, employment, criminal justice [22], and health care), which contributes to the disproportionate arrests and incarceration of people of color, perhaps especially among those with serious mental illnesses. The lack of association between arrests and homelessness could be due to the fact that the sample was economically disadvantaged, and many participants who did not identify as homeless were nevertheless unstably housed (i.e., living with various friends or family members).

Fifth, the ranking of the most common charges varied substantially over time, from the earliest arrests that occurred at an average age of 20–21 years to the most recent arrests that occurred at an average age of 33–35 years. The earliest arrests commonly involved marijuana possession, driving under the influence, and shoplifting or burglary charges, whereas recent arrests more often included probation violations, officer obstruction–related charges, disorderly conduct, and failure to appear in court. (According to Georgia code, misdemeanor obstruction of an officer is defined as occurring when a person “knowingly and willfully obstructs or hinders any law enforcement officer in the lawful discharge of his official duties.” This, along with the misdemeanor “giving false name, address, or birthdate to an officer,” is, in essence, being uncooperative and getting in the way of the officer doing his or her job.) Some of this variation over 42 years (the arrest dates ranged from 1976 to 2018) is undoubtedly related to changes in policing and processing practices (e.g., less enforcement of marijuana possession). At the same time, future research should explore the extent to which the evolution of an individual’s criminal legal system history is related to the changes in the severity of his or her illness and to the psychosocial consequences of having a long-term serious mental illness. Just as Lorvick and colleagues (19) described how the “accumulation of criminal justice involvement” leads to a higher prevalence of unmet mental and physical health care needs, so too might the accumulation of adverse psychosocial consequences related to long-term serious mental illnesses lead to a change in the nature of arrest charges. That is, as more arrests and more charges accrue, the likelihood of failure to appear and probation violation increases, and such violations are also likely driven partly by the manifestations of long-term serious mental illnesses, including neurocognitive impairments and other symptoms, as well as lack of transportation and other social adversities. Having a long-term serious mental illness also appears to be associated with a higher likelihood of charges that may be illness related (e.g., disorderly conduct). Much more research is needed to understand the dynamic relationships between behavioral illness severity; socioeconomic disadvantage, including housing insecurity; and events leading to arrests (23).

Diverse policies, practices, and programs have been and continue to be developed to reduce criminal legal system entanglement among individuals with serious mental illnesses (24–27). Interventions benefit from targeting different points of the sequential intercept model (28): intercepts 1 (emergency services and law enforcement), 2 (booking, arraignment, and detention), 3 (jails and courts), 4 (reentry from jails or prisons to community), and 5 (probation and parole). Our findings may be informative. First, community-based mental health services (“intercept 0”) have a key role to play in reducing arrests and providing police with alternatives to arrest. New interventions and service models are needed that ensure appropriate interventions are available at the right time to mitigate the conditions or risk factors associated with police contacts and arrests (29). The findings reported here suggest that a potentially high-impact approach toward this goal would be for mental health professionals to ask regularly about clients’ legal involvement and work with clients on adherence to court mandates so that they do not experience the negative collateral consequences of failure to appear (30) and probation violations that can so easily result in additional arrests.

Second, we hypothesize that charges such as willful obstruction of officers and giving false name, address, or birth date to an officer (i.e., charges related to suboptimal interactions with the officer) may stem from prior negative experiences with the criminal legal system, including low perceived procedural justice (which concerns perceptions of neutrality, impartiality, or fairness; respect and dignity; having a voice; and trustworthy and transparent processes). Several North American studies examining experiences in police encounters (both crisis- and non–crisis-related) have indicated that individuals with serious mental illnesses feel very vulnerable when interacting with police officers and that how the officer treats them influences their feelings toward the encounter and their level of cooperation with the police (31–33). This is consistent with research originating in social psychology on policing and procedural justice: how people are treated by an authority in terms of fairness, dignity, and voice has implications for cooperation, perceptions of police legitimacy (34), and motivation to comply with the law (35). As such, new approaches need to be developed (for criminal legal system personnel) to improve empathy, fairness, dignity, and voice, which would improve perceived procedural justice and thus respect toward officers (which may ultimately reduce the likelihood of charges such as willful obstruction).

Third, because some charges (e.g., shoplifting and theft by taking) could stem in part from socioeconomic disadvantage and poverty—as opposed to the serious mental illness itself—progress in this area will remain limited as long as those with serious mental illnesses are disenfranchised. Many jurisdictions have been experimenting with public health–informed responses to people who engage in relatively minor criminal activities that may be a result of extremely adverse social conditions, recognizing that police may not be at all necessary in many of these situations and that other types of professionals may be better equipped to interact with vulnerable members of the community (36). Fourth, because some charges (e.g., disorderly conduct and “terroristic threats,” the latter including threatening “to commit any crime of violence”)—especially later in the course of serious mental illness—likely stem from illness manifestations, officers need more tools at their disposal to directly link symptomatic or behaviorally disturbed individuals to a receptive and accessible mental health system.

At least two methodological limitations must be acknowledged. First, although the study had the unique advantage of having highly objective data on lifetime arrests and charges, the RAP sheets captured only arrests occurring in the state of Georgia. As such, the figures presented here are undoubtedly underestimates of actual arrests. Furthermore, the RAP sheets we obtained had highly reliable information only on arrest history; conviction and sentencing history could not be relied on to be complete. Other research has documented inequities at later points in the criminal justice system. For example, Hall et al. (2) reported that a major mental illness diagnosis was associated with more than a 50% increase in the odds of a jail sentence for misdemeanor arrestees. Second, although internal validity was high, given the relatively homogeneous nature of the sample, generalizability might be limited because of the particular sociodemographic and clinical characteristics of the study sample (e.g., all participants were enrolled from public-sector inpatient settings, indicating a high level of socioeconomic disadvantage and clinical severity). Multisite studies, with larger and more representative samples, are warranted.

Conclusions

Given the substantial reforms that are under way, such as those in many police departments and in behavioral health crisis response systems across the United States, more fine-grained research on the exact charges experienced by people with serious mental illnesses over the life course is merited, as is more insight into the nature of the situations in which these charges are issued. Such findings, including those reported here, might point to both policy and programmatic solutions—within the criminal legal and mental health sectors—to reduce arrests and incarcerations in this population.