A Systematic Examination of the Influence of Indicators of Mental Illness on Deaths in the Texas Criminal Justice System

Abstract

The publicized deaths of individuals such as Sandra Bland raise questions about whether, and how, the American criminal justice system should handle individuals who display symptoms of mental illnesses. The current study seeks to establish whether there is a stage in the criminal justice process at which the presence of an indicator of a mental illness represents a particularly high risk of death in custody. Using multilevel logistic regression and data on 10,667 civilian deaths that occurred in the custody of criminal justice agencies across the state of Texas, from January 1st, 2005, through October 30th, 2020, results suggest that indicators of mental illness represent the highest risk of death when an individual is in the custody of police. At later stages in the criminal justice system process, particularly jail and prison, indicators of a mental illness reduce the chances that someone will die in custody. Policy implications are discussed.

Keywords

Race and crime/justice gender and crime/justice law enforcement corrections mental illness

Introduction

The publicized deaths of individuals such as Sandra Bland, Patrick Warren Sr., Janice Dotson-Stephens, and Angelo Quinto raise questions about whether, and how, the American criminal justice system should handle individuals who display symptoms of mental illnesses (Clarke, 2019; Burke, 2021a; Burke, 2021b; Montgomery, 2019). While these cases illustrate the worst possible outcome, the death of an individual, lawmakers and citizens are left to wonder if they represent a larger pattern of deleterious interactions between those experiencing a mental health crisis and employees of the American criminal justice (CJ) system.

There is supporting research that suggests that individuals with mental illnesses are often labeled as “dangerous, unpredictable, incompetent, and unable to function in society” (Yanos, 2018, p. 4). Movies, TV shows, and news media commonly misrepresent both mental illness and the people who suffer from such illnesses (Stuart, 2006; Scarf, et al., 2020; “Stigma, prejudice, and discrimination…”, 2020). According to a U.S. study from 2006, half of all characters who were portrayed as having a mental illness in prime-time programming, were also portrayed as being violent (Stuart, 2006). A more recent study in the Journal of the American Medical Association (JAMA) specifically notes that the 2019 film Joker “continues the tradition of movies depicting individuals with mental illness as violent” (Scarf, et al., 2020, p. 1). In addition to movies and television programming, celebrities or political figures have given misleading public statements about many aspects of mental illness (LaPierre, 2012; Neill, 2005). For example, in 2012, the head of the NRA, Wayne LaPierre, gave a speech in Washington D.C. in the wake of a mass shooting, in which he classified people with mental illnesses as “genuine monsters – people so deranged, so evil, so possessed by voices and driven by demons that no sane person can possibly ever comprehend them” (LaPierre, 2012, para. 9).

Given the prevalent and consistent application of negative labels to people who suffer from mental illnesses, we are again left to wonder if criminal justice system personnel are affected by these stigmas to the extent that showing symptoms of a mental illness becomes a risk factor for death in the custody of these actors. The current study seeks to examine this question in more detail by analyzing 10,667 civilian deaths that occurred in the custody of criminal justice agencies across the state of Texas, from January 1st, 2005, through October 30th, 2020. Texas is the second largest state in the United States when ranked by population, at approximately 11 million people behind California, yet there are approximately 16,000 more people in Texas jails and prisons, and 170,460 more people on probation and parole in Texas, compared to California. (“US states – ranked…”, 2021; “The facts…”, 2020). In addition, for decades, a Texas statute has required law enforcement and correctional agencies to file a report within 30 days of the death of an individual in their custody (Texas Code of Criminal Procedure, 2021). According to the Texas Code of Criminal Procedure, “The director shall make a good faith effort to obtain all facts relevant to the death and include those facts in the report” (Texas Code of Criminal Procedure, 2021). The custodial death reports were made publicly available on the Texas Attorney General's (AGs) office webpage in 2016 (“Custodial Death Report”, 2020), making Texas an ideal place for an exploratory look at the effects of mental illness on deaths in custody.

Literature Review

Mental Illness in the Criminal Justice System

According to the National Institute of Mental Health (NIMH), almost one in five U.S. adults live with mental illness (“Mental illness”, 2022). This, however, encompasses a wide range of illnesses that include a broad array of symptoms and symptom severity. Accordingly, the NIMH further divides the concept of “mental illness” into two broad categories known as “any mental illness or AMI” and “serious mental illness or SMI”. AMI is defined as a mental, behavioral, or emotional disorder of any level of severity, while an SMI is a mental, behavioral, or emotional disorder “resulting in serious functional impairment, which subsequently interferes with or limits one or more major life activities” (“Mental illness”, 2022, para. 4). In 2020, approximately 21% of all U.S. adults had a disorder classified as AMI, while 5.6% of U.S. adults (14.2 million people) had a SMI. In addition, almost 50% of juveniles between the ages of 13 and 18 had, at some point in their lives, a disorder classified as AMI and approximately 22% of those had serious impairment (“Mental illness”, 2022).

While the prevalence of mental illness is high in the United States, so too is the lack of treatment for such disorders. In 2017, the Treatment Advocacy Center, a non-profit agency focused on the timely and effective treatment of severe mental illness, applied the NIMH prevalence statistics for schizophrenia and severe bipolar disorder to census data and created estimates of these conditions by state (Fuller, 2017). Their finding suggests that 8.2 million U.S. adults had one of these two conditions, and that 3.9 million people with these diagnoses were untreated. In Texas, approximately 3.3% of the total adult population had at least one of these two disorders in 2017, which equates to almost 691,000 people. Of those, approximately 47.5% were untreated (Fuller, 2017).

The high prevalence of SMIs coupled with the lack of treatment available or obtained has created a notable issue for the American criminal justice system. In fact, it is estimated that “individuals with severe mental illness generate no less than 1 in 10 calls for police service and occupy at least 1 in 5 of America's prison and jail beds” (Fuller et al., 2015). Research shows that 20 − 25% of individuals who have been arrested in the United States have some form of behavioral health condition (Abramson, 2021; Livingston, 2016). The situation has become so severe for American jails that the Los Angeles, Chicago, and New York jails each hold more individuals with mental illnesses than any psychiatric hospital in the U.S. (“Serious mental illness…”, 2016; Vogel, et al., 2014) and the Harris county jail (servicing the Houston, TX area) is the largest mental health service provider in the state of Texas (Natarajan et al., 2016). In addition, the Bureau of Justice Statistics (BJS) reports that 41% of all state and federal prison inmates have a history of mental health problems, with 14% of state prisoners and 8% of federal prisoners reporting having experienced serious psychological distress less than 30 days prior to being interviewed for the BJS study (Maruschak, et al., 2021). According to the same study, only 40% of state prisoners and 26% of federal prisoners who were deemed to have experienced recent serious psychological distress reported receiving treatment (including prescription medications, therapy from a trained professional, or both) for a mental health problem.

Mental Illness and Death in Interactions with Police

As previously noted, a significant percentage of individuals with mental illness have contact with law enforcement each year in the United States (Fuller et al., 2015; Manderscheid et al., 2009, Abramson, 2021). Some research shows that police deal with the mental health calls for service at even higher rates. For example, in the Netherlands, Van den Brink, et al. (2012) found that about 50% of the calls for service involved individuals with a mental illness. Their study showed that those individuals who were connected successfully with services were more likely to continue receiving treatment. Other research shows that the norms of the community and the police department determine how officers deal with people who have one or more mental illnesses (Morabito, 2007). For example, in rural communities, the police may be more familiar with specific individuals in need of care; hence, officers are more likely to deal with these individuals in a more informal way. Conversely, although larger departments may have more resources, such as specialized crisis units to respond to mental health calls for service, the response to these calls can often be more time-consuming and complex than in urban areas (Compton et al., 2010).

Irrespective of the organizational and community differences, interacting with individuals with mental illnesses can potentially be more volatile as some individuals have the propensity to become violent (Jun, et al., 2020). This perceived response from officers sometimes leads to the elevated use of force during such interactions (Jun, et al. 2020; Rossler & Terrill, 2017). In fact, Jun et al. (2020) found that in New York and Baltimore, regardless of their involvement in criminal behavior, individuals with a SMI were more likely to have physical force used against them by police. In turn, individuals with mental illnesses may develop a response that is conditioned on fear toward officers, which perpetuates a cycle of negative interactions (Boscarato et al., 2014; Meade, et al., 2015). While an increase in the volatility of interactions can reasonably be assumed to increase the chances of a death occurring during a police/citizen encounter, estimating how often a mental health condition contributes to a death during these interactions is difficult. This is because mental health variables are not commonly known or consistently recorded surrounding these events in the U.S. (Fuller et al., 2015).

One program run by the BJS does attempt some measurement of the impact of mental health on deaths during police/public interactions. The Arrest Related Death (ARD) program, began in 2003 but was terminated in 2014 “due to concerns over the program's coverage and reliability” (Banks, et al., 2019, p. 1). Subsequently, a pilot study of a redesigned version of the program began in 2015. The revised ARD program consisted of a search of open-source records (news sources, available publicly posted records from government agencies, etc.) for deaths that occurred in the United States during an arrest, in a given time frame. Following this, a survey was sent to police agencies across the country to inquire about any deaths that had occurred during an arrest in the given year. The reports from the agencies were then used to cross check the open-records search and a report was compiled regarding the deaths. During the 2015 pilot study, 303 qualifying deaths were identified. Of those, 211 were cases in which a person was killed by the police. The remaining deaths included suicides and manners of death classified as “other”. Of the 211 deaths in which a person was killed by police, 18.5% of those individuals made suicidal statements and 19.4% exhibited mental-health problems (Banks, et al., 2019). Fuller et al. (2015) echo these estimates and add that “the risk of being killed during a police incident is 16 times greater for individuals with untreated mental illness than for other civilians approached or stopped by officers” (p. 1).

Mental Illness and Death in Correctional Facilities

Sources estimate that 20% of the people in jails around the United States, and 13–15% of adult state prison inmates have a serious mental illness or SMI (Maruschak, et al., 2021; “Serious mental illness…”, 2016). In Texas, 30% of jail inmates have one or more SMIs (Natarajan et al., 2016). Maruschak et al. (2021) conducted face-to-face interviews with a national sample of state and federal prisoners across the U.S., who were aged 18 or older. They found that the most diagnosed condition in both state and federal prisons was major depressive disorder, followed in state prisons by bipolar disorder and in federal prisons by anxiety disorder. Vogel et al. (2014) note that the American criminal justice system was not designed to treat mental health issues, and as a result, the majority of inmates with mental health problems are not treated, or properly treated, while incarcerated. Specifically, they note that only one-third of state prisoners, a quarter of federal prisoners, and one in six jail inmates with a mental health condition receive treatment (Vogel et al., 2014). Maruschak et al. (2021) report that only about a third of state inmates and one-fifth of federal inmates who reported recent severe psychological distress also reported receiving medications for a mental health condition. In addition, only 30% of state inmates and 15% of federal inmates who reported recent severe distress reported receiving counseling or therapy from a trained professional.

Deaths of individuals with mental health conditions occur in correctional facilities for a number of reasons. Correctional environments, which would be stressful to most people, are especially problematic for those with mental disorders (Natarajan et al., 2016). Everything from the lights and noises of the correctional unit to the real or perceived threats of violence, are more challenging for those with mental illnesses to navigate. Conditions of the unit can enhance the symptoms of a SMI which then increases the likelihood of the inmate displaying behavioral problems (“Serious mental illness…”, 2016). This can either result in inmates being involved in a higher prevalence of violent interactions with other inmates or guards (Natarajan et al., 2016), or in the inmate spending a disproportionate amount of time in solitary confinement (“Serious mental illness…”, 2016). Both outcomes can lead to a higher chance of death for inmates with a mental illness as an increase in violent interactions increases their vulnerability to a fatal assault (Natarajan et al., 2016), while more time in solitary confinement can contribute to higher rates of suicide (“Serious mental illness…”, 2016). Some estimates suggest that half of all suicides in correctional facilities are committed by the 13 to 20% of inmates with SMIs (“Serious mental illness…”, 2016). In jails, state prisons, and federal prisons, the leading non-natural cause of death is suicide (Carson, 2021a; Carson, 2021b).

Other Causes of Death in the Custody of the Criminal Justice System

In addition to the possible presence of a mental illness, other factors are known to contribute to a person's chances of dying in the custody of the U.S. criminal justice system. Using data from the National Vital Statistics System (NVSS) and Fatal Encounters data (FE), Edwards et al. (2019) found that the risk of being killed by the police is determined by one's race, gender, and age. The authors found that the risk of death in police custody was highest for Black men, who are about 2.5 times as likely to be killed by police during their lifetime compared to White males. The study showed that Black women were 1.4 times more likely to be killed by the police than White women. American Indian men were between 1.2 and 1.7 times more likely to be killed compared to White (non-Hispanic) men, and American Indian women were more likely to be killed at a rate between 1.1 and 2.1 times that of White women. Interestingly, while Hispanic men were found to be between 1.3 and 1.4 times as likely as White men to be killed during conflicts with the police, Hispanic women were much less likely ─between 12 and 23% ─to be killed by police than White women (Edwards et al., 2019).

According to the BJS, White prisoners accounted for just over half of all deaths in state prisons, while making up less than one-third of the total state prison population (Carson, 2021b). The same study indicated that 95.6% of all inmate deaths in state prisons were male and that the mortality rates of inmates in state and federal prisons increased steadily over time from 2010 to 2018. Another study published by the BJS reports that White individuals comprised 59% of jail inmate deaths in 2018 and that “females held in local jails had a higher rate of mortality (162 deaths per 100,000 female inmates) than males had (152 deaths per 100,000 male inmates)” (Carson, 2021a, p. 1). In both jails and prisons, the leading causes of death are natural (including all manner of illnesses and pre-existing conditions), followed by suicide, drug and alcohol intoxication, and then homicide (Carson, 2021a; Carson, 2021b).

In addition, the interaction of these variables complicates the discussion. Monique Brown, an epidemiologist from the Arnold School of Public Health at the University of South Carolina, notes that while there are no known differences in genetic makeup between White and Black people, which might cause conditions such as high blood pressure, diabetes, mental health disorders, or heart disease, “there are many social factors at play. Historically, African Americans have had less access to health care and lower incomes” (Horn, 2020, para. 7). Because the leading causes of death in jails and prisons are natural causes, including pre-existing medical conditions (Carson, 2021a), factors such as the inability to access health care prior to entering the CJ system potentially increase the odds that racial minorities, and Black people specifically, will experience higher rates of death in the criminal justice system.

Finally, while demographic factors pertaining to death in custody may be somewhat sparsely examined, variables relating to the agencies in which the deaths occurred are not examined at all. In 2019, BJS released a report on the first phase of their revised ARD program, in which open-sourced records were used to better collect information on arrest related deaths across the U.S. (Banks, et al., 2019). This report noted that California, followed by Texas, reported the highest number of arrest related deaths across the U.S. In fact, nearly half (42%) of all 424 deaths identified in the study period [June 1, 2015- August 31, 2015] occurred in five states – California, Texas, Florida, Georgia, and Ohio. These findings are supported and further examined in reports created by private companies or organizations. For example, mappingpoliceviolence.org reports that five of the top 24 cities across the U.S., that reported multiple deaths in police custody between 2013 and 2020, were Texas cities. Further, Houston, Texas had the fifth highest reported number of deaths in police custody in the U.S. during that time period (“Mapping police violence”, 2021). In addition, Texas comprised 10.6% of the prison deaths that occurred across the country in 2019 (Carson, 2021c)

While this knowledge is interesting, it is not, in and of itself, particularly useful for future policy implications, other than potentially for those policies relating to the reporting of deaths in custody. It does, however, suggest that there is an area of research which has not been fully examined. More specifically, how do variables pertaining to the place of death impact the likelihood of death? Questions such as these have yet to be answered in the literature.

Current Study

The current study seeks to examine the following research questions:

At what stage, if any, in the Texas criminal justice process (arrest, jail, prison, or medical/drug treatment) does the presence of an indicator of a mental illness represent a higher risk for death in custody?

Is death in custody, at any stage in the Texas criminal justice process, more likely, due to demographic variables such as race, sex, and age?

Has the likelihood of death in custody at any stage of the Texas criminal justice process changed over time?

Where possible to examine, do agency variables, such as the size of the agency and the ratio of officers to the number of people served, affect the likelihood of death in custody at any particular stage in the Texas criminal justice process?

Data

The Texas Code of Criminal Procedure (Art. 49.18) specifies that if a person dies in the custody of, or during an interaction with, a peace officer, or while incarcerated in a jail, correctional facility, or state juvenile facility, the law enforcement agency or correctional facility “shall” investigate and file a written report with the state Attorney General (AG) within 30 days of the death. It also states that the Texas AG's office is charged with receiving those reports and with making them available to all interested parties, with the exception of any information deemed to be “privileged”¹ (Texas Code of Criminal Procedure, 2021). In accordance with the Texas Code of Criminal Procedures, the Texas AG's office posts custodial death reports on their website, in their raw form (“Custodial death report”, 2020). Please note that while it is mandatory for Texas agencies to submit a custodial death report when someone dies in their custody, what is required in that report is somewhat vague. Texas does have a standard form used for this purpose but the statute mandating the report does not require that all possible fields be completed. As a result, some variables including sex, race, age, the year of death, and the existence of a pre-existing medical condition are reported across the examined time frame with low rates of missing data (0%, 0.3%, 0.4%, 0%, and 1.4% respectively), while information regarding the presence of indicators of a mental illness is missing in over 87% of cases. This missing data issue also varies by agency. For example, hospitals or infirmaries that treated individuals under the custody of the Texas criminal justice system were the most consistent in reporting whether an individual was displaying a symptom of a mental illness at the time of death. The Texas prison system (TDCJ) was the least consistent in doing so. Given the large sample size, researchers can still examine the impact of a display of indicators of a mental illness, but results should be considered exploratory².

The Texas Justice Initiative (TJI) is a non-profit organization that collects and publishes Texas data for the purpose of government agency oversight (“About us”, 2021). In that role, TJI continuously pulls death-in-custody reports from the Texas AG's webpage and codes and combines them into an Excel spreadsheet for research use. All death information used in the present research, including characteristics of deceased individuals and information about the location and agency where the death occurred, was obtained from the TJI webpage (“Explore the data”, 2021). This data includes 10,667 reported deaths that occurred in agencies across the state of Texas from January 1^st, 2005 to October 30^th, 2020.

Information about correctional or policing agencies (number of licensed peace or correctional officers, populations served, etc.) was obtained from a number of sources. More specifically, population data for states, towns, cities, or counties was pulled from the U.S. Census Bureau, based on the 2010 census (“Quick facts: Texas”, 2021)³. The number of licensed peace officers working in university, city, or county law enforcement agencies was pulled from the FBI's Crime Data Explorer (Crime data explorer [CDE], 2021). Because of the length of time required for the FBI to compile information into the CDE, the most recent year available is 2019. That is the data that was pulled for the present analysis. While 2019 is near the end of the time frame being examined, it was the year that data was most consistently reported across all agencies. Further, upon examination of agencies that reported multiple years of data to the FBI, there were no large increases or decreases in the number of officers from 2005 to 2019. Therefore, data from this year was chosen to examine differences between agencies, as opposed to within agencies.

Data was submitted to the AG's office from any policing or correctional agency that had state licensed officers and a death that occurred in the examined time period. These include agencies classified as school district, medical center, airport, and transit police. Accordingly, information about school district populations and school district police departments came from each respective independent school district's webpage (for example, see: “Police Department”, 2021). The same is true for medical center, airport, transit, and correctional agency information (For examples see: “About DART Police”, 2021; “DFW Airport Department…”, 2021; “Texas Department of Criminal Justice [TDCJ]”, 2021).

Analytical Strategy

Because the current data covers a time period of 16 years, and spans across agencies that serve a wide range of populations (under 100 people, to several million), many agencies reported more than one death in custody. In fact, there were 10,667 deaths reported by 508 agencies. Because many agencies reported more than one death, characteristics of the agency where the incident occurred (things like population served and the number of officers in that agency) were not unique to each event. The result is a regression violation that, if not addressed, could cause an over-inflation of estimates of statistical significance (Menard, 2010).

In the current study, multilevel logistic regression, run with the HLM version 8.1 software package, was used to address this lack of independence of events. Multilevel modeling allows for the nesting of deaths within agencies (Raudenbush et al., 2004). Accordingly, level-1 of the reported model consists of characteristics that are unique to each death event (the age, sex, and race/ethnicity of the individual who died, the year of the death, whether the individual illustrated any indication of a mental health problem, and whether the individual died of a pre-existing medical condition). Level-2 consists of agency characteristics (the number of licensed police or correctional officers in the agency, and the ratio of officers per 1000 people in the population served). In addition, version 8.1 of the HLM software package allows three options for the handling of missing data. In the present study, a level-1 deletion list was used to ensure the same number of units across analyses. In this case, when the MDM file (the data file format that HLM uses) is created, the program is told to “Delete missing data when: running analyses”. Once this is done, you can create a list of level-1 variables that will be used in the analyses, from which to base deletion on. This means that the number of observations used in the analyses will remain consistent (“Missing data in…”, 2020). Finally, logistic regression is being used here because the dependent variables are dichotomous (1 = death occurred in police [jail, prison, etc.] custody, 0 = death occurred in the custody of some other CJ system actor), and this analytical strategy is appropriate for such outcomes (Menard, 2010).

There is one exception to the above noted analytical strategy. Deaths that occurred in Texas prisons were examined using a single-level logistic regression model. This procedural change is due to the fact that only two prison agencies report deaths in custody in Texas. Those agencies are the Texas Department of Criminal Justice, or TDCJ, and the Texas Youth Commission (TYC), which became the Texas Juvenile Justice Department (TJJD) in 2011 (Sunset Advisory Commission, 2011). The TYC and TJJD together reported only five deaths in custody over the 16-year period being examined, which is not enough to examine separately. This effectively means that all of the deaths from Texas prisons (99.9%) came from a single agency; the TDCJ. It is reasonable to assume that different units under the umbrella of the TDCJ have different death in custody rates. Unfortunately, while the form allows agencies to record the address where each death occurred, in the case of TDCJ deaths, this information is commonly left blank or a P.O. Box or address for a central agency such as the TX Department of Public Safety, the State Counsel for Offenders, or the Office of the Inspector General is entered. Of the 6,971 deaths that occurred in a TDCJ facility, none of them had an address for an actual unit entered. As a result, agency characteristics such as the number of licensed correctional officers or the number of officers per 1000 inmates can only be calculated for the agency as a whole. With no variability in the level-2 variables for this agency, the prison model does not need multilevel modeling. Agency variables were removed, and all other variables were examined using a single level logistic regression model.

For the current analysis, model statistical significance will be tested using the model chi-square (G_M) statistic. Model substantive significance will be established using the likelihood ratio, or McFadden, R² $(R_{L}^{2})$ . Statistical significance of predictors will be determined with the Wald test statistic, and fully standardized regression coefficients (See Menard, 2010 for a complete description of how these are calculated for logistic regression models) will be used to determine the substantive significance of predictors.

Variables

The current data contains information about individuals who died in custody in some part of the Texas criminal justice system (CJS). There is no information in the available data on those who were taken into custody and did not die. Therefore, it is not possible, without seeking and receiving additional data from individual policing and correctional agencies across the state of Texas (over 2,700 agencies), to truly predict death in custody (“Current Statistics”, 2021). It is currently possible, however, to examine variables that might predict where in the process someone will die. The form that Texas agencies must submit to the state AG's office to report a death in custody, contains a selection indicating if the individual died in police custody (with separate entries for during, or after, an arrest), in a jail setting, in a hospital or medical facility, in TDCJ (prison), or in some other setting (these include things like detox facilities). For the present analysis 14.9% of the sample died in police custody, while 65.3% died in a prison setting, 11.9% died in a jail, 4.3% died in a hospital or infirmary, and 0.5% died in some other setting. The remaining 3% did not have a code for the custody type when the death occurred and are thus coded as missing data. Because so few deaths occurred in locations classified as “other”, there were not enough cases to examine them using multivariate analysis. As a result, 10,290 deaths (96.5% of the total sample)⁴ were examined. Each of these custody types was dummy coded (1 = the death did occur in this form of custody; 0 = the death did NOT occur in this form of custody) and used as the dependent variables in the respective models. In this way, it is possible to examine differences in statistically significant predictors of death in each area of the Texas criminal justice system.

Independent variables in the present analysis include whether the person exhibited any signs of a mental health problem, whether the individual died of a pre-existing condition that was present before they entered the system, the age, sex, and race/ethnicity of the deceased individual, the year of the death, the agency size, and the ratio of licensed officers per 1000 people in the population served. Where indicators of mental health problems are concerned, agencies were simply asked to specify whether the individual who died had exhibited any indicators of a mental health problem (1 = yes, 0 = no). No further information on the indicator, or on the mental health problem, was required, though some agencies did record more detailed information in an optional narrative section of the report. This, however, was a rare event. Accordingly, we simply examined whether the deceased individual gave an indicator that they had a mental health problem or condition.

The death in custody form that each agency must submit to the state AG's office prompts the agencies to specify whether the individual's death was caused by a condition that was present before they came into contact with police or otherwise came into the custody of a criminal justice agency in the state of Texas. In the present study, this variable was coded as 1 = they did die of a condition that was present prior to entering the system, 0 = they did not die of a condition that was present prior to entering the system. It is important to note here that 69% of the sample had received a medical examiner or coroner's evaluation to determine the cause of death. While more cases may have been examined by a medical expert, there was no indication of this in the report, and no amended report filed to note that this had occurred at a later time. Age was recorded as the decedent's age in years at the time of death. Sex was recorded as 0 = female and 1 = male. Race/ethnicity was recorded as White, Hispanic, Black, or Other. For the present analysis, each racial/ethnic category was dummy or dichotomously coded into a separate variable where 1 = membership in that racial/ethnic group, and 0 = no membership in that racial/ethnic group (or membership in some other racial or ethnic category). The death year was simply the year that the individual died in the custody of the Texas CJS.

Agency size was simply measured as the number of licensed peace or correctional officers in the agency where the death occurred. The ratio of officers per 1000 people in the population served was calculated by dividing the population size by 1,000 and then dividing the number of officers by that result. Please note that if an agency served a city, then the population of the city was used, but if the agency served a school district, for instance, then the student population of the school district was used. Every effort was made to determine and record the actual population that each agency served. While previous studies do not include these variables, they are incorporated here to explore their potential impact on the current outcome variables. Though simply having more officers, or a greater ratio of officers per individuals served, will not automatically make them more ethical or better trained, it seems reasonable to assume that some deaths may be avoided if officers are not working in agencies suffering from personnel shortages. Having a higher number of officers in an agency might, for example, allow for reduced back up times for police officers or more frequent cell checks for correctional officers. Reduced back-up response times, in turn, could decrease police officer anxiety and make them less likely to react with violence when other options are available. More frequent cell checks could mean that symptoms of drug withdrawal, medical conditions, or mental health conditions are caught before they escalate to a deadly level (Barajas & Novack, 2021). While these scenarios are only possibilities, they illustrate that these variables have some potential to influence in-custody deaths at every stage of the Texas criminal justice system.

Findings

Table 1 illustrates some descriptive statistics for the variables used in the present study.

Table 1.

Descriptive Statistics.

	Death in Police Custody (n = 1,593)			Death in a Jail Setting (n = 1,269)			Death in a Prison Setting (n = 6,971)			Death in a Hospital or Infirmary (n = 457)			Total Sample (n = 10,667)
	Mean/%	Min	Max	Mean/%	Min	Max	Mean/%	Min	Max	Mean/%	Min	Max	Mean/%	Min	Max
Number of Officers	1,735.6	1	5264	540.4	3	5,264	25,000			1,310.5	6	5,264	16,848.3	1	25,000
Ratio of Officers	1.5	0	13.8	.9	0	144.1	164.5			.9	.06	4.3	108.8	0	164.5
Age	34.9	11	80	42.5	14	89	55.3	18	92	46.2	18	85	49.8	11	93
Sex
Male	95.4%			85.5%			95.9%			91%			94.3%
Race
White	38%			46.7%			42.6%			38.7%			42.1%
Black	26.4%			27.1%			29.9%			34.8%			28.8%
Hispanic	33.1%			24.9%			27%			24.5%			27.8%
Other	2.4%			1.3%			.5%			2%			1%
Death Year		2005	2020		2005	2020		2005	2020		2005	2020		2005	2020
Pre-Existing condition	2.3%			25.9%			38%			36.5%			30%
Mental Health problems	5.1%			5.4%			.5%			7.9%			2.5%

From the breakdown of the table, it is clear that individuals who die in police custody in Texas are generally younger (35 years of age on average) than those who die in other parts of the system, with the oldest average age of death occurring in prison settings (55 years on average). In addition, prison and jail settings typically have more correctional officers, and more officers per 1,000 people served, than do police or other law enforcement agencies (hospital or transit police, for example). The vast majority of citizens at each stage of the Texas criminal justice system are male and a smaller majority at each step are White. Between 26% and 38% of people who died in Texas jails, prisons, or hospitals (while in custody), died of a pre-existing medical condition that they had before they entered the CJ system, while only 2.3% of those who interacted with police did so.

Perhaps most germane to the present study, the descriptive statistics suggest that the largest percentage of people who died after indicating that they may have had some form of mental illness, did so while in custody at a Texas hospital or infirmary (7.9%). The lowest percentage of deaths involving an indicator of mental illness occurred in Texas prisons (0.5% of deaths), with police custody deaths falling in between at 5.1%. If we were to remove missing data and look at valid percentages, or percentages of only the 1,331 cases where mental health indicators were reported, then 34% of those people died in police custody, 25.6% died in Texas jails, 6.1% died in prisons, and 33% died in a hospital or infirmary while in state custody.

Table 2 illustrates the results of the multi-level logistic regression model for death in police custody, versus any other form of custody in the Texas criminal justice system. From the table, we can see that the model is statistically significant, but substantively fairly weak in that it explains only 3.6% of the variance in the dependent variable. In order of strength, the statistically significant predictors of death in police custody were NOT having a pre-existing condition, being younger, and showing symptoms of a mental illness.

Table 2.

Predictors of Death in Police vs Other Custody Types.

Dependent Variable	Model Statistics	Independent Variables	Unstandardized Coefficients (b)	Standardized Coefficients (b)*	Significance (Wald Statistic)
Death in Police Custody (0 = No; 1 = Yes)	G_m = 105.165df = 10(p = .000) $R_{L}^{2}$ = .036	Number of Officers in Agency	.0003	.405	.365
		Officers per 1000 People	−.095	−.874	.139
		Age at Death	−.042	−.076	.001
		Sex (Male)	.574	.016	.187
		Death Year	−.155	−.085	.240
		Race (Hispanic)	.428	.023	.239
		Race (Black)	.474	.026	.264
		Race (Other)	.425	.005	.804
		Mental Illness Symptoms	.574	.027	.015
		Pre-Existing Condition	−2.357	−0.129	.000

The results of the multi-level logistic regression model for deaths in Texas jail facilities are demonstrated in table 3. Please note that, in this model, there is a multicollinearity issue between the number of officers in a facility and the ratio of officers per 1,000 people served (VIF = 184.869 and 185.564 respectively). This is likely due to mandatory caps on the populations that can be housed in jails, as well as relatively stable annual jail confinement rates in some areas⁵(Kang-Brown, et al., 2021). In fact, the Texas Commission on Jail Standards (TCJS) publishes annual reports that contain information on both inmate populations and licensed jailer turnover rates (“TCJS annual jail…”, 2020). While older reports have inconsistent data collected, in 2010, the TCJS reported that the Texas jail system was running at 70.7% capacity with 66,064 inmates, while in 2019 the commission reported the jail system to be running at 68.6% capacity with 65,719 inmates (“TCJS annual jail…”, 2020). The licensed jailer turnover rates were more variable, however, with the 2019 report noting that “individual county monthly rates ranged from 0% to 50%” (Wood, 2020, p. 17). In addition, the Texas Administrative Code (2014, title 37, part 9, chapter 275, section 275.4) states that the officer to inmate ratio must be at least one jailer to every 48 inmates Texas code of criminal procedure article 49. (Texas Code of Criminal Procedure, 2021). The code does allow for the mandate of staffing in excess of this minimum standard, as deemed necessary, but this ratio is the set minimum in Texas jails. Given that the inmate population has remained stable over time, and that the ratio of inmates to jailers is held to a minimum standard, the ratio of officers per 1,000 jailed individuals was removed from the model, eliminating the collinearity concern. The number of officers in the unit remained in the model due to the variability reported in the 2019 TCJS annual report (Wood, 2020).

Table 3.

Predictors of Death in Jail vs Other Custody Types.

Dependent Variable	Model Statistics	Independent Variables	Unstandardized Coefficients (b)	Standardized Coefficients (b*)	Significance (Wald Statistic)
Death in Jail Custody (0 = No; 1 = Yes)	G_m = 59.115df = 9(p = .000) $R_{L}^{2}$ = .020	Number of Officers in Agency	−0.001	−.653	.003
		Age at Death	.009	.008	.370
		Sex (Male)	−.654	−.009	.075
		Death Year	−.028	−.007	.732
		Race (Hispanic)	−.497	−.013	.110
		Race (Black)	−.416	−.011	.184
		Race (Other)	−.357	−.002	.670
		Mental Illness Symptoms	−.436	−.010	.029
		Pre-Existing Condition	.576	.015	.061

The tabled results indicate that the model is statistically significant (p = .000), but as with police custody deaths, this model is substantively weak ( $R_{L}^{2}$ = .020 or 2% explained variance in the dependent variable). Of the nine predictors in the model, only two had a statistically significant impact on whether someone died in jail custody, as opposed to the custody of some other part of the Texas CJ system. In order of substantive significance, people were more likely to die in a Texas jail versus some other point in the Texas criminal justice system, when the jails had a low number of officers employed there, and when individuals did NOT illustrate any signs of a mental illness or condition.

Table 4 illustrates the results of the single-level logistic regression model for deaths in Texas prisons.

Table 4.

Predictors of Death in Prison vs Other Custody Types.

Dependent Variable	Model Statistics	Independent Variables	Unstandardized Coefficients (b)	Standardized Coefficients (b*)	Significance (Wald Statistic)
Death in Prison Custody (0 = No;1 = Yes)	G_m = 498.424df = 8(p = .000) $R_{L}^{2}$ = .294	Age at Death	.076	.414	.000
		Sex (Male)	.711	.059	.012
		Death Year	.379	.623	.000
		Race (Hispanic)	.477	.076	.008
		Race (Black)	.644	.104	.000
		Race (Other)	−1.516	−.053	.056
		Mental Illness Symptoms	−1.663	−.237	.000
		Pre−Existing Condition	−.954	−.156	.000

Recall that this model is reviewing deaths from only one agency, the TDCJ, over the relevant time frame. As a result, agency variables are not necessary, nor is multi-level modeling. The subsequent model is statistically (p = .000) and substantively ( $R_{L}^{2}$ = .294) significant. The predictors, together, explain 29% of the variance in the dependent variable (death in prison vs death in some other part of the Texas CJ system). Seven of the eight predictors had a statistically significant impact on the outcome variable. In order of strength, those variables are death year (more deaths occurred in TDCJ later in the examined timeline), age (older individuals were more likely to die in prison compared to other parts of the system), symptoms of a mental illness (people who exhibit symptoms of a mental illness are LESS likely to die in the custody of TDCJ), pre-existing conditions (people who had a pre-existing condition coming into the system were also LESS likely to die in prison custody), race (Black and Hispanic individuals are more likely to die in TDCJ), and gender (males are more likely to die in this part of the Texas CJ system).

Finally, table 5 indicates the results of the multi-level logistic regression model that examines deaths while in custody and receiving treatment in a hospital or infirmary. This model is statistically significant (p = .000), but substantively weak ( $R_{L}^{2}$ = .025). Two variables make it statistically significantly more likely that people will die at this stage in the Texas criminal justice system, compared to all other stages. In order of strength, those predictors are age at death (being older), and the presence of a pre-existing medical condition.

Table 5.

Predictors of Death Hospitals or Infirmaries vs Other Custody Types.

Dependent Variable	Model Statistics	Independent Variables	Unstandardized Coefficients (b)	Standardized Coefficients (b*)	Significance (Wald Statistic)
Death in Custody at a Hospital or Infirmary (0 = No; 1 = Yes)	G_m = 69.310df = 10(p = .000) $R_{L}^{2}$ = .025	Number of Officers in Agency	.0002	0.184	.497
		Officers per 1000 People	−.080	−0.502	.458
		Age at Death	.032	0.040	.003
		Sex (Male)	.139	0.003	.741
		Death Year	.131	0.049	.271
		Race (Hispanic)	.172	0.006	.645
		Race (Black)	.336	0.012	.432
		Race (Other)	−.002	0.000	.998
		Mental Illness Symptoms	.011	0.000	.975
		Pre−Existing Condition	.751	0.028	.032

Discussion and Implications

The first research question asks at what stage in the Texas CJ system does the presence of an indicator of a mental illness represent a higher risk of death for those in custody? Multivariate results suggest that such indicators represent the highest risk when an individual is in the custody of police. At later stages in the CJ system process, particularly jail and prison, indicators of a mental illness reduce the chances that someone will die in custody. Empirical studies have reported a mental illness rate of between 25% −50% among individuals who had interactions with the criminal justice system (Abramson, 2021; Livingston, 2016; Van den Brink et al., 2012). Fuller et al. (2015) noted that at least one in four fatal police encounters involve a mentally ill person. This is also echoed in this current study as about 34% of valid cases of those who died in police custody illustrated a symptom of a mental illness.

Several reasons may explain this result. First, some citizen interactions with police officers involve arrests, where the police may employ a certain degree of force. For example, the high-profile cases of Eric Garner in 2014 and George Floyd in 2020, both involved some police force. Second, in these situations, mental illness may increase the chances that an individual could behave in an unpredictable manner when given instructions by a police officer (Shemin, 2019). This, in turn, might lead to an officer perceiving a threat to their life or safety, particularly given the common negative stereotypes involving the violence or “dangerousness” of those who display indicators of mental illness (Yanos, 2018). This perception of a threat to one's life or safety may then increase the chances that the police officer will initiate or elevate the use of force in an interaction. For example, Officer Reynaldo Contreras from the Killeen Texas Police department shot and killed 52-year-old Patrick Warren, Sr. after his family called and requested a mental health check (Burke, 2021a). Bell county has mental health deputies that are commonly dispensed for such calls, and one had been sent the previous day to help Mr. Warren. On the day of the incident, however, the mental health deputy was not available, and police officers were sent instead. Upon encountering an unarmed but emotionally distressed Mr. Warren, an officer first attempted to tase him. When this did not end the confrontation, the officer shot and killed the man (Burke, 2021a). While a grand jury failed to indict the officer for his actions (“Grand jury clears…”, 2021), this case clearly highlights the importance of addressing mental illness issues in the criminal justice system.

Prior research has reported that individuals with mental health issues were 16 times more likely than other civilians to be killed during interactions with law enforcement officers (Fuller et al., 2015). In fact, many individuals with mental illness lack access to appropriate mental health treatment or do not receive proper medical care, leaving law enforcement to deal with these cases (Shemin, 2019). In many instances, the city or county, and the police department, lack the resources and skills needed to deal with mental health crises that arise.

The second research question asks whether demographic variables have an impact on the area of custody in which a person dies in the Texas CJ system. Multivariate results suggest that the answer to this question is affirmative. Specifically, individuals who are classified as Black or Hispanic, are more likely to die in prisons, versus other parts of the Texas CJ system. Younger individuals are more likely to die in police custody, while older individuals are more likely to die in the custody of prisons or while being treated in hospitals or infirmaries. Finally, males are statistically significantly more likely to die in prison environments as opposed to other custody types.

At least part of the race/ethnicity related results may be explained by the overrepresentation of Black individuals at this stage of the CJ system. In 2020, Black individuals represented 12.9% of the population for the state of Texas (“Quick facts…”, 2021), yet in 2019, they represented 32.6% of the population of people incarcerated by the TDCJ (Texas Board of Criminal Justice, 2021). Hispanic people, however, do not show prima facie evidence of being overrepresented in the Texas prison system. For these individuals, the increased likelihood of death specifically in the custody of Texas prisons, may be due to factors such as differential homicide or suicide rates, disparities in deaths from conditions such as COVID-19, which can be contracted while in prison, or any number of other influences that are beyond the scope of the present study. The impact of age is likely related to systemic differences in the length of custody for individuals in prisons versus law enforcement. Many individuals remain in the custody of the correctional system for long periods of time, growing older, and hence are more likely to perish from health-related or other conditions either in the prison, or while being treated in a hospital or infirmary. Finally, as with race, the significance of gender may be a byproduct of overrepresentation. As of 2019, males represented 91.6% of the TDCJ population (Texas Board of Criminal Justice, 2021).

Indirect effects may also play a role. For example, the current data suggests that there is a statistically significant positive bivariate correlation between being Black and having a pre-existing medical condition (r = .129, p = .000). This is not surprising as racial minorities, and especially Black people, are more likely to be economically disadvantaged and to have limited access to health care (Horn, 2020). Multivariate models in the present analysis indicate that those with pre-existing conditions are more likely to die in custody when being treated in a hospital or infirmary. This may suggest that Black individuals are also more likely to die in this part of the Texas CJ system, because of their higher likelihood of having a pre-existing condition. Black individuals consisted of 12.9% of the Texas population according to 2020 Census data (“Quick facts…”, 2021), but they accounted for 34.8% of the custodial deaths in hospitals or infirmaries in the current sample.

Research question three asked whether the likelihood of death in custody at any stage of the Texas criminal justice system has changed over time. Multivariate results indicate that the year of death was only relevant at one point in the Texas system. Individuals were more likely to die in the state prison system, the TDCJ, later in the examined timeline, as opposed to earlier. Figure 1 visually illustrates an inconsistent but overall increasing trend in deaths over time in this agency. It should be noted that increases that occurred in 2020 may be due to the COVID-19 pandemic (Griswold, 2021).

Figure 1.

Deaths in the custody of the TDCJ across time.

These results are consistent with Carson (2021b), who found that state and federal prison mortality rates rose steadily from 2010 to 2018.

Finally, research question four inquired as to whether agency variables played a role in predicting any of the current outcome variables. Univariate results suggested that correctional agencies have more officers per 1,000 people served than do policing agencies. This observation makes some sense, given that police are working with the general public and correctional officers are working with individuals who are both captive and at least suspected of having committed a criminal offense. This point is made even more clear when examining the multivariate models, as they suggest that deaths in jail custody increase when there are fewer officers employed in the agency. Recall that jails, in part, act as the first place of detainment when an individual is brought into the system. As a result, these individuals may have medication or other medical or mental health needs that have not yet been identified or addressed, including potential difficulties associated with the withdrawal from an illegal substance (American Academy of Family Physicians, 2017). These threats to the safety or life of detained individuals must be closely monitored and addressed by jail staff. If there are not enough individuals employed to effectively perform these tasks, it could mean that detained individuals do not pass through this part of the Texas criminal justice system alive (Barajas & Novack, 2021).

The findings in regard to indicators of mental health have significant policy implications regarding factors contributing to in custody deaths. On the one hand, it is important to reduce encounters between law enforcement officers and agitated individuals with mental health issues in order to minimize the chance of using lethal force. There should be a larger prevalence of mental health deputies, as described by the Texas Health and Human Services department, who are trained and available to respond to emergency calls that involve individuals who are displaying symptoms of a mental health condition (“Mental health deputy…”, 2021). Through these mental health deputies, individuals should be directed to crisis intervention centers, where available, for psychiatric treatment instead of dispatching law enforcement officers to deal with the situation. In addition, the American Psychological Association (Abramson, 2021) notes a growing trend of police departments teaming with mental health clinicians either out in the field or behind the scenes. These include programs such as the CAHOOTS public safety system in Eugene, Oregon, where if a call comes through the 911 system involving a non-violent mental health issue, a medic and a trained mental health crisis worker are sent to the scene. In 2019, these teams were sent over 24,000 times and they only requested police backup 150 times. Options that directly involve mental health clinicians, may save taxpayers money, and improve safety for citizens and police officers alike (Abramson, 2021).

Because officers in many departments lack the expertise to adequately work with individuals with mental health issues, law enforcement agencies should also mandate additional trainings to enhance officers” knowledge on mental illness conditions, symptoms, treatments, and tactics to de-escalate the situation when encountering mentally ill individuals. If possible, training should also include some effort to educate officers regarding a more empirically supported estimate of the dangerousness or prevalence of violence among those diagnosed with different mental illnesses. There is some precedent for such training, as an anti-stigma intervention program was given to Swedish officers during their basic police officer training (Hansson & Markstrom, 2014). Even six months after the completion of the program, officers who went through the training had better mental health literacy, better attitudes towards people with mental illnesses, and reported greater willingness to interact with and live around people with mental illnesses, compared to officers who did not go through the training. With programs such as these, perhaps, the negative labels seen on television could be put into better perspective for officers, which in turn, would hopefully make these stigmas less of an unconscious factor when making split second decisions to escalate the level of force used.

This study represents the first effort by researchers to systemically analyze death in custody data in Texas and addressed several important questions regarding custodial death. However, this study is not without limitations and results should be interpreted with caution. First, the data contains a fairly large percentage of missing cases where indicators of mental health are concerned, which may potentially bias the current findings. This limitation suggests that the death in custody reporting program should be improved. Improved standardized reporting requirements should be implemented to encourage or mandate more in-depth and extensive accounts of all custodial death incidents, and of all variables involved in each incident. Second, the analysis only included information on individuals who died in custody. Without information on those who entered police or other custody and lived, we were not able to fully analyze predictors of death in the custody of the Texas CJ system.

Future studies should employ data on all individuals who entered the criminal justice system to explore the previously mentioned questions. In addition, the measure of mental illness in the present study is simply whether the deceased individual displayed a symptom of a mental illness at the time of their death. In future studies, it is imperative to examine the specific mental health indicators that most commonly lead to death at every stage of the criminal justice system. Other factors should also be studied, including whether an individual has been diagnosed with a mental illness, as well as indicators of whether a person has received treatment for a mental health disorder, and the nature of that treatment (medication, psychotherapy, or both). More knowledge about these issues will allow researchers to determine if they reduce the likelihood that an individual will die in the custody of the criminal justice system, or in any particular part of the criminal justice system (police, jail, prison, etc.). Finally, if a detailed description of the death in custody event is collected on a more consistent basis, it would allow future researchers and mental health clinicians to ascertain whether an individual may have been displaying indicators of a mental illness that the CJ system practitioner could not identify. This, in turn, could potentially lead to further training for employees of the American criminal justice system.

Research suggests that only 2–4% of those diagnosed with a serious mental illness commit acts of violence (DeAngelis, 2022), yet movies portray individuals with mental illnesses as consistent and calculating in their violence (see Silence of the Lambs (1991), American Psycho (2000) and Joker (2019) for examples of this depiction of individuals with mental illness). The current data suggests that in the Texas criminal justice system, people who indicate that they are suffering from a mental illness are most likely to die during their interaction with police. It is possible that lacking somewhat extensive training on the handling of individuals with mental illnesses, it may be very difficult for police officers to counteract potentially subconscious depictions of people who suffer from mental illness, when they are called to make split second decisions about force in the uncontrolled environments where they are required to work.

Despite knowing that he was responding to a mental health call, and despite having another officer present with him, Officer Reynaldo Contreras felt it necessary to shoot and kill Patrick Warren Sr. in the front yard of his home when the unarmed man did not respond appropriately to his commands. In the moments after the shooting, he is heard on camera telling another officer that he had no choice but to shoot (“Grand jury clears…”, 2021). Currently, the data suggests that it is police officers who are the most likely to kill those who display symptoms of mental illness, at least in the state of Texas. A future question to focus on is whether officer assessments of danger in situations such as these are based entirely on the circumstances at the scene, or whether perceptions of persons with mental illnesses, which can potentially be countered by training, play a role.

Footnotes

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article

ORCID iD

Wendi Pollock

Notes

Author Biographies

Wendi Pollock is an associate professor of Criminal Justice at Texas A&M University Corpus Christi. Her research interests include police-public interactions, the long and short-term impact of arrest, race/ethnicity and gender issues in the American criminal justice system, and the impact of crime media consumptions on the actions of criminal justice practitioners. Her recent publications have appeared in journals such as The Police Journal: Theory, Practice and Principles, Race and Justice, and the International Journal of Police Science & Management.

Deborah Sibila is an assistant professor of Criminal Justice at Texas A&M University Corpus Christi. Her research interests include policing, intimate relationship violence, immigration, and gender offending/victimization related issues. Her recent publications have appeared in journals such as the International Journal of Police Science and Management, the Journal of Strategic Security, and The American Review of Public Administration.

Durant Frantzen is professor and Chair of the Department of Social Sciences at Texas A&M-San Antonio. His research interests are in the areas of community corrections and domestic violence interventions, with a focus on high-risk offender populations. He has published in numerous journals in his field and is coauthor of The Death Penalty: Constitutional Issues, Commentaries and Case Briefs published by Taylor and Francis, and Everyday Ethics for the Criminal Justice Professional published by Carolina Academic Press.

Fei Luo is an associate professor of Criminal Justice at Texas A&M International University. Her research interests include policing, race/ethnicity, gun policy, immigration and crime, intimate partner violence, and application of quantitative methodology. Her recent publications have appeared in journals such as Race and Justice, Journal of Crime and Justice, American Journal of Criminal Justice, and Policing: An International Journal.

Alex del Carmen is a Professor and Associate Dean of the School of Criminology, Criminal Justice and Public Administration. His research specialty is on racial profiling in policing. He has written numerous articles and books on the topic and others related to policing. He is a Fulbright Specialist and has made academic presentations throughout the world.

References

About DART police. (2021). Retrieved on January 24, 2021 from: https://www.dart.org/about/dartpolice/dartpolice.asp

About us. (2021). Retrieved on January 24, 2021 from: https://texasjusticeinitiative.org/about

Abramson

. (2021, July 1). Building mental health into emergency responses: More cities are pairing mental health professionals with police to better help people in crisis . Retrieved from: https://www.apa.org/monitor/2021/07/emergency-responses

American Academy of Family Physicians. (2017). Incarceration and health: A family medicine perspective (position paper). Retrieved from: https://www.aafp.org/about/policies/all/incarceration.html

Banks

Plenty

Couzens

Lee

Brooks

Scott

K.M.

Whyde

. (2019, July). Arrest-related deaths program: Pilot study of redesigned survey methodology . Bureau of Justice Statistics. Retrieved from: https://bjs.ojp.gov/library/publications/arrest-related-deaths-program-pilot-study-redesigned-survey-methodology

Barajas

Novack

. (2021, November 1). Locked up and left to die. Texas Observer. Retrieved from: https://www.texasobserver.org/locked-up-and-left-to-die/

Boscarato

Lee

Kroschel

Hollander

Brennan

Warren

. (2014). Consumer experience of formal crisis-response services and preferred methods of crisis intervention. International Journal of Mental Health Nursing, 23(4), 287–295. https://doi.org/10.1111/inm.12059

Burke

. (2021, May 16). Policing mental health: Recent deaths highlight concerns over officer response . NBC News. Retrieved from: https://www.nbcnews.com/news/us-news/policing-mental-health-recent-deaths-highlight-concerns-over-officer-response-n1266935

Burke

. (2021, January 14). Texas man fatally shot by police during mental health check, family calls for officer’s arrest. NBC News. Retrieved from: https://www.nbcnews.com/news/us-news/texas-family-calls-officer-s-arrest-after-man-fatally-shot-n1254297

10.

Carson

E. A

. (2021, April). Mortality in local jails, 2000–2018–statistical tables. Bureau of Justice Statistics. Retrieved from: https://bjs.ojp.gov/content/pub/pdf/mlj0018st.pdf

11.

Carson

E. A

. (2021, April). Mortality in state and federal prisons, 2001–2018 – statistical tables. Bureau of Justice Statistics. Retrieved from: https://bjs.ojp.gov/content/pub/pdf/msfp0118st.pdf

12.

Carson

E. A

. (2021, December). Mortality in state and federal prisons, 2001 − 2019–statistical tables. Bureau of Justice Statistics. Retrieved from: https://bjs.ojp.gov/content/pub/pdf/msfp0119st.pdf

13.

Clark

. (2019, July 2). Mentally ill Texas woman “ignored to death” after five months in jail . Prison Legal News. Retrieved from: https://www.prisonlegalnews.org/news/2019/jul/2/mentally-ill-texas-woman-ignored-death-after-five-months-jail/

14.

Compton

M. T.

Broussard

Hankerson-Dyson

Krishan

Stewart

Oliva

J. R.

Watson

A. C

. (2010). System- and policy-level challenges to full implementation of the crisis intervention team (CIT) model. Journal of Police Crisis Negotiations, 10(1–2), 72–85. https://doi.org/10.1080/15332581003757347

15.

Crime data explorer. (2021). Retrieved January 14, 2021 from: https://crime-data-explorer.app.cloud.gov/officers/agency/TX0040000/pe

16.

Current statistics. (2021). Retrieved on December 30, 2021 from: https://www.tcole.texas.gov/content/current-statistics

17.

Custodial death report. (2020). Retrieved on November 10, 2020 from: https://oagtx.force.com/cdr/cdrreportdeaths

18.

DeAngelis

. (2022, July 11). Mental illness and violence: Debunking myths, addressing realities. American Psychological Association. Retrieved from: https://www.apa.org/monitor/2021/04/ce-mental-illness

19.

DFW airport department of public safety. (2021). Retrieved on January 24, 2021 from: https://www.dfwairport.com/dps/

20.

Edwards

., Lee

., & Esposito

. (2019, August 20). Risk of being killed by police use of force in the United States by age, race-ethnicity, and sex. Proceedings of the National Academy of Sciences of the United States of America. Retrieved on February 9, 2021, from: https://www.pnas.org/content/pnas/116/34/16793.full.pdf

21.

Explore the data. (2021). Retrieved on December 1, 2020 from https://texasjusticeinitiative.org/datasets/custodial-deaths

22.

Fuller

D.A.

Lamb

H.R.

Biasotti

Snook

(2015, December). Overlooked in the undercounted: The role of mental illness in fatal law enforcement encounters. Treatment Advocacy Center. Retrieved from: https://www.treatmentadvocacycenter.org/overlooked-in-the-undercounted

23.

Fuller

D. A

. (2017, May 3). Research weekly: 2016 prevalence of treated and untreated severe mental illness by state . Treatment Advocacy Center. Retrieved from: https://www.treatmentadvocacycenter.org/fixing-the-system/features-and-news/3828-research-weekly-2016-prevalence-of-treated-and-untreated-severe-mental-illness-by-state

24.

Grand jury clears officer who shot unarmed central Texas man with mental health issues. (2021, May 21). KBTX3 News. Retrieved on December 9, 2021 from: https://www.kbtx.com/2021/05/21/grand-jury-clears-officer-who-shot-unarmed-central-texas-man-with-mental-health-issues/

25.

Griswold

. (2021, Nov. 5). Texas leads nation in COVID-19 deaths in prisons . Spectrum News. Retrieved from: https://spectrumlocalnews.com/tx/south-texas-el-paso/news/2021/11/05/texas-leads-nation-in-covid-19-deaths-in-prisons

26.

Hansson

Markstrom

. (2014). The effectiveness of an anti-stigma intervention in a basic police officer training programme: A controlled study. BMC Psychiatry, 14(55), 1–8.

27.

Horn

. (2020, June 8). COVID-19 highlights underlying racial health disparities: African Americans are more likely to have underlying health conditions that undermine COVID-19 survival . Retrieved on November 10, 2020 from: https://www.sc.edu/uofsc/posts/2020/06/covid_health_disparities.php#.X6m-UFBMFPY

28.

Jun

H. J.

DeVylder

J. E.

Fedina

. (2020). Police violence among adults diagnosed with mental disorders. Health & Social Work, 45(2), 81–89. https://doi.org/10.1093/hsw/hlaa003

29.

Kang-Brown

Montagnet

Heiss

. (2021, January). People in jail and prison in 2020 . Retrieved from: https://www.vera.org/downloads/publications/people-in-jail-and-prison-in-2020.pdf

30.

LaPierre

. (2012, December 21). NRA: Full statement by Wayne LaPierre in response to Newtown Shootings . The Guardian. Retrieved from: https://www.theguardian.com/world/2012/dec/21/nra-full-statement-lapierre-newtown

31.

Livingston

J. D

. (2016). Contact between police and people with mental disorders: A review of rates. Psychiatric Services, 67(8), 850–857. https://doi.org/10.1176/appi.ps.201500312

32.

Manderscheid

R. W.

Atay

J. E.

Crider

R. A

. (2009). Changing trends in state psychiatric hospital use from 2002 to 2005. Psychiatric Services, 60(1), 29–34. Retrieved from http://ps.psychiatryonline.org/doi/pdf/10.1176/ps.2009.60.1.29

33.

Mapping police violence. (2021). Tretrieved on April 7, 2021, from: https://mappingpoliceviolence.org/cities

34.

Maruschak

L. M.

Bronson

Alper

. (2021, June). Survey of prison inmates, 2016: Indicators of mental health problems reported by prisoners . Bureau of Justice Statistics. Retrieved from: https://bjs.ojp.gov/library/publications/indicators-mental-health-problems-reported-prisoners-survey-prison-inmates

35.

Meade

Steiner

Klahm

C. F. I. V

. (2015). The effect of police use of force on mental health problems of prisoners. Policing and Society, 27(2), 229–244. https://doi.org/10.1080/10439463.2015.1049602

36.

Menard

(2010). Logistic regression: From introductory to advanced concepts and applications. Sage Publications, Inc.

37.

Mental health deputy. (2021). Retrieved on December 10, 2021, from: https://www.hhs.texas.gov/providers/behavioral-health-services-providers/crisis-service-providers/mental-health-deputy

38.

Mental illness. (2022, January). National Institute of Mental Health . Retrieved on March 6, 2022 from: https://www.nimh.nih.gov/health/statistics/mental-illness

39.

Missing data in HLM 8.1. (2020). Retrieved on December 4, 2021, from: https://ssicentral.com/wp-content/uploads/2020/12/Missing_data_81-1.pdf

40.

Montgomery

. (2019, May 7). Sandra Bland, it turns out, filmed traffic stop confrontation herself . The New York Times. Retrieved from: https://www.nytimes.com/2019/05/07/us/sandra-bland-video-brian-encinia.html

41.

Morabito

M. S.

(2007). Horizons of context: Understanding the police decision to arrest people with mental illness. Psychiatric Services, 58(12), 1582–1587. https://doi.org/10.1176/ps.2007.58.12.1582

42.

Natarajan

Trigilio

Stamm

Gnaedinger

Manautou

Nguyen

Cavnar-Thompson

Wittenmeyer

Zybay

. (2016, November). Preventable tragedies: How to reduce mental health related deaths in Texas Jails. University of Texas School of Law Civil Rights Clinic. Retrieved from: https://law.utexas.edu/wp-content/uploads/sites/11/2016/11/2016-11-CVRC-Preventable-Tragedies.pdf

43.

Neill

U.S

. (2005, August 1). Tom cruise is dangerous and irresponsible. The Journal of Clinical Investigation, 115(8), 1964–1965. https://doi.org/10.1172/JCI26200

44.

Police department. (2021). Retrieved on January 15, 2021 from: https://www.aldineisd.org/about/departments/police-department/

45.

Quick facts: Texas. (2021) Retrieved on January 14, 2021 from: https://www.census.gov/quickfacts/TX

46.

Raudenbush

Byrk

Cheong

Congdon

Toit

M. D.

(2004). HLM6: Hierarchical linear & nonlinear modeling. Scientific Software International, Inc.

47.

Rossler

M. T.

Terrill

. (2017). Mental illness, police use of force, and citizen injury. PoliceQuarterly, 20(2), 189–212.

48.

Scarf

Zimmerman

Winter

Boden

Graham

Riordan

B.C.

Hunter

J.A.

(2020, April, 24). Association of viewing the films Joker or Terminator: Dark Fate with prejudice towards individuals with mental illness . JAMA Network Open, available from: https://jamanetwork.com/journals/jamanetworkopen/fullarticle/2764842

49.

Serious mental illness (SMI) prevalence in jails and prisons. (2016, September). Treatment Advocacy Center. Retrieved from: https://www.treatmentadvocacycenter.org/evidence-and-research/learn-more-about/3695

50.

Shemin

. (2019). Mental health on trial: An in-depth look at the criminalization of mental illness in the United States criminal justice system. (Master’s thesis). Available from ProQuest. (ProQuest No. 13865636)

51.

Stigma, prejudice and discrimination against people with mental illness. (2020, August). Retrieved on November 22, 2021, from: https://www.psychiatry.org/patients-families/stigma-and-discrimination

52.

Stuart

. (2006, February). Media portrayal of mental illness and its treatments: What effect does it have on people with mental illness? CNS Drugs, 20(2), 99–106. https://doi.org/10.2165/00023210-200620020-00002

53.

Sunset Advisory Commission. (2011, July). Final Report. Retrieved from: https://www.sunset.texas.gov/public/uploads/files/reports/Youth%20Commission_Juvenile%20Probation_Ombudsman%20Final%20Report%202011%2082%20Leg.pdf

54.

TCJS Annual Jail Reports. (2020). Retrieved on July 11, 2022 from: https://www.tcjs.state.tx.us/reports/#1570702351860-8bc65270-53ae

55.

Texas Administrative Code. (2014). Retrieved on July 11, 2022 from: https://texreg.sos.state.tx.us/public/readtac$ext.TacPage?sl=R&app=9&p_dir=&p_rloc=&p_tloc=&p_ploc=&pg=1&p_tac=&ti=37&pt=9&ch=275&rl=4

56.

Texas Code of Criminal Procedure. (2021). Texas code of criminal procedure article 49. Retrieved on January 24, 2021 from: https://statutes.capitol.texas.gov/Docs/CR/htm/CR.49.htm

57.

Texas department of criminal justice [TDCJ]. (2021). Retrieved on January 15, 2021 from: https://www.tdcj.texas.gov/divisions/hr/coinfo/emp-co.html

58.

The facts: State-by-state data. (2020). Retrieved on December 3, 2021, from: https://www.sentencingproject.org/the-facts/#detail?state1Option=Texas=state2Option=South%20Carolina

59.

US states – ranked by population 2021. (2021). Retrieved on December 3, 2021, from: https://worldpopulationreview.com/states

60.

Van Den Brink

R. H.

Broer

Tholen

A. J.

Winthorst

W. H.

Visser

& Wiersma

. (2012). Role of the police in linking individuals experiencing mental health crises with mental health services. BMC Psychiatry, 12(1), 171–178. https://doi.org/10.1186/1471-244X-12-171

61.

Vogel

Stephens

K. D.

Siebels

. (2014). Mental illness and the criminal justice system. Sociology Compass, 8(6), 627–638. https://doi.org/10.1111/soc4.12174

62.

Wood

. (2020, February 1). Texas Commission on Jail Standards 2019 Annual Report. Retrieved from: https://www.tcjs.state.tx.us/wp-content/uploads/2020/07/2019-TCJS-Annual-Jail-Report.pdf

63.

Yanos

P. T.

(2018). Written off: Mental health stigma and the loss of human potential. New York, NY: Cambridge University Press.