Arthur J. Lurigio
Departments of Criminal Justice and Psychology
Loyola University Chicago
James A. Swartz
Jane Addams College of Social Work
University of Illinois at Chicago
CRIMINAL JUSTICE AGENCIES have become de facto settings for mental health treatment and other clinical services (Lurigio & Swartz, 2000). The growing number of persons with serious mental illness (SMI) who appear at each step in the criminal justice process—from arrest to post-incarceration release—has forced professionals in corrections, who usually have no backgrounds or experiences in the mental health field, to continually face the challenges of identifying, referring, and case managing the mentally ill. With or without special training or guidelines, most criminal justice professionals (such as police officers, probation officers, prison intake workers, and parole agents) screen individuals for mental illness in their daily practice. They do so in order to make decisions about such options as diversion, segregated housing, treatment, or other specialized interventions.
Screening for SMI involves a brief initial evaluation about a client’s need for mental health services, which can be done at the point of arrest, sentencing, or imprisonment. It also can be done formally or informally, and can trigger either an immediate decision or a more comprehensive psychiatric evaluation designed to help make subsequent mental health-related decisions about a case. For example, police officers at the scene determine who should be diverted for an emergency hospitalization instead of arrested and who is at risk of attempting suicide in lockup. Probation officers conduct mental health screening as part of an overall needs assessment to determine offender classification and service brokerage. Detention and correctional officers screen incoming prison inmates and jail detainees for mental illness in order to assign them to specialized housing and programming.
This article examines the use of actuarial screening tools that have been developed to flag persons with SMI for further assessment, diagnosis, and treatment in institutional and community-based correctional facilities. In the absence of such tools, the mentally ill in the criminal justice system are likely to go unrecognized and untreated. The paper is divided into four major sections. The first discusses the prevalence of persons with SMI in jail, prison, and probation populations and the dearth of mental health services for them. The second emphasizes the use of valid and reliable screening tools as an important first step in providing services for offenders with mental illness. The third presents the results of two studies that have tested mental health screening tools for use with criminal justice populations. The fourth concludes with recommendations for the incorporation of mental health screening tools in the intake protocols of correctional departments.
Mentally Ill in Criminal Justice Settings
Over the past 20 years, several epidemiological studies have shown that substantial numbers of persons involved in the criminal justice system have SMI, such as schizophrenia, bipolar disorder, and major depression (Abram & Teplin, 1991; Abram et al., 2003; Ditton, 1999; Diamond et al., 2001; Human Rights Watch [HRW], 2003). The three largest psychiatric facilities in the United States are urban jails: the Los Angeles County Jail, the Cook County Jail (CCJ) in Chicago, and the jail at Riker’s Island in New York City (Insel, 2003). One estimate suggests that 900,000 individuals with SMI are admitted to our nation’s jails annually (Steadman, Scott, Osher, Agnese, & Robbins, 2005). Many factors explain the large numbers of the mentally ill in offender populations. These factors include transinstitutionalization, stricter civil commitment laws, homelessness, public order policing, and the fragmentation of the mental health and drug treatment service systems (HRW, 2003; Lamb & Weinberger, 1998; Lurigio, 2005; Lurigio & Swartz, 2000).
Ditton (1999) has conducted the only national study to date on the prevalence of the mentally ill in correctional populations. She reported that, at midyear 1998, an estimated 283,800 mentally ill offenders were incarcerated in our nation’s prisons and jails. A total of 16 percent of those surveyed in each population reported either a mental health condition or an overnight stay in a mental hospital. Approximately 16 percent, or an estimated 547,800 probationers, indicated that they had experienced in their lifetime a mental disorder or stayed overnight in a mental hospital.
Based on information from personal interviews, state prison inmates with a mental disorder were more likely than other inmates to be incarcerated for a violent offense (53 percent compared with 46 percent) and to be under the influence of alcohol or drugs at the time of the current offense (59 percent compared with 51 percent). They also were more than twice as likely as other inmates to have been homeless in the 12 months before their arrest (20 percent compared with 9 percent). More than three-quarters of mentally ill inmates had been sentenced to prison, jail, or probation at least once prior to their current sentence. Since admission, 61 percent of mentally ill inmates in state prison and 41 percent of mentally ill detainees in local jails reported that they had received treatment for a mental health problem, including counseling, medication, or other mental health services (Ditton, 1999).
The large number of mentally ill offenders underscores the need for effective mental health screening and treatment services. Courts have ruled consistently that jails and prisons are legally obligated to provide mental health and other medical services to detainees and prisoners (Diamond et al., 2001; Veysey & Bichler-Robertson, 2002). Left untreated, incarcerated persons with SMI have trouble adapting to life in prisons and jails, as well as following the written and unwritten rules that are inherent in the daily routines of correctional facilities (HRW, 2003; Torrey, 1995). Furthermore, prisoners and detainees with SMI are at high risk for suicide, disciplinary infractions, and victimization (Dicataldo et al., 1995; HRW, 2003). Similarly, mentally ill probationers often have trouble complying with their probation orders, such as reporting to their probation officers or finding employment and housing opportunities. These difficulties increase their risk for a technical violation or new arrest (Solomon & Draine, 1999).
The landmark case of Ruiz v. Estelle (503 P. Supp. 1265.1323 ) set forth standards for “minimally adequate mental health treatment programs” in prisons. These standards consist of the systematic mental health screening and evaluation of inmates; the capacity to ensure that treatment involves more than just inmate segregation; the provision of individualized treatment by trained mental health professionals; the maintenance of accurate and complete mental health records; the supervision and review of prescriptions; and the identification of inmates with suicidal tendencies (Jemelka et al., 1993). Other case precedents have established jail detainees’ rights to mental health treatment and aftercare services (e.g., Brad H. et al. v. City of New York et al . 185 Misc. 2d 420; 712 N.Y.S. 2d 336 [Sup. Ct. 2000]).
Notwithstanding the clear legal mandate to provide mental health services, and the prodigious numbers of offenders who need and receive such services, many mentally ill individuals remain unidentified and untreated while under the jurisdiction of the criminal justice system (Elliot, 1997; National Commission on Correctional Health Care [NCCHC], 2002; Steadman & Veysey, 1997; Teplin, 1990). For example, in a study of female jail detainees, Teplin et al. (1997) found that only 25 percent of those meeting the criteria for SMI received treatment within one week of admission. As Ditton (1999) found, nearly 40 percent of prison inmates and 60 percent of jail detainees with mental illness reported that they were receiving no mental health services during their recent incarceration.
These findings mirror those of broader studies of mental health care for prisoners (ACP et al., 1992; Jordan et al., 1992) and probationers (Lurigio et al., 2003), which also have found high rates of untreated mental illness. For example, a survey of probation departments by Boone (1995) found that only 15 percent of those responding to a national survey had programs for the mentally ill. Several factors contribute to the criminal justice system’s failure to deliver adequate treatment to individuals with SMI. These include the scarcity of mental health resources, the rapid turnover of detainees in jails, correctional staff with little clinical training, and significant increases in the criminal justice population, which exceeds the capacity of criminal justice organizations to implement mental health services for all persons who require such care (NCCHC, 2002).
Need for Mental Health Screening Tools
Clinicians use several techniques at intake to collect information for formal psychiatric evaluations. In general, they examine the type and severity of the symptoms (what the patient reports) and signs (what the clinician observes) of SMI. The two key components of the evaluation process are the construction of treatment histories, including the use of medications, and the performance of mental status examinations to evaluate current levels of cognitive, social, behavioral, and emotional functioning. Information gathering techniques include structured and unstructured interviews, symptom questionnaires, patient observations, personality inventories, and psychological tests. The clinician integrates the findings from these various sources in order to render a diagnosis and recommend a psychiatric treatment and rehabilitation plan (Nolen-Hoeksema, 2005).
The basic psychometric properties of psychiatric assessment tools are validity and reliability. The former refers to the accuracy of instrumentation and the latter to the consistency with which such instrumentation is used to collect information. The field of psychometrics has established different types of validity and reliability for different purposes of instrument construction and application (Anastasi & Urbina, 1997). Two other concepts useful in determining the accuracy of an assessment or screening tool are sensitivity and specificity. Sensitivity is the likelihood that a tool will find disease among those who have the disease, or the proportion of people with disease who have a positive test result. Specificity is the likelihood that a tool will find no disease among those who do not have the disease, or the proportion of people free of a disease who have a negative test result.
As we suggested in our preceding discussion, psychiatric assessment requires considerable time, skill, and training in the use of specialized diagnostic tools and criteria. Few correctional staff persons have such backgrounds or knowledge. Clinicians often are hired in correctional settings to perform assessments and treatment, but their time is usually limited and their caseloads are overwhelmingly high (Earley, 2006). Hence, it is infeasible and costly to conduct a full-blown psychiatric assessment on every person entering the criminal justice system.
To conserve sparse clinical time and resources, mental health screening should be employed; its goal is the identification of persons who need further evaluation. However, if mental health screening is done at all in criminal justice settings, screening tools are usually unstandardized, and screening procedures vary greatly within and among agencies. The validity and reliability of such screening results are quite low (Steadman & Veysey, 1997). For example, a survey of state prisons found most of the responding institutions had no valid data on the prevalence of various psychiatric conditions (primarily SMI) in their inmate populations—a problem attributable to antiquated information systems as well as to non-standardized screening and assessment procedures (Hornung et al., 2002).
Although no comparable surveys have been conducted of probation departments or jails, a similar degree of variation in screening and referral practices is likely to be found in those settings as well (Lurigio et al., 2003; Skeem et al., 2003). As Steadman, et al. (2005) noted about screening for mental health problems in jails: “Screening may consist of anything from one or two questions about previous treatment to a detailed, structured mental health status examination” (p. 816).
Overall, few criminal justice agencies employ brief screening tools that are constructed and validated specifically for offender populations (Swartz, 2001). Since the inception of the war on drugs, criminal justice administrators and practitioners have encouraged the use of screening and assessment tools to detect substance use disorders. However, they have paid little or no attention to the development of comparable instruments for detecting mental illness (Peters et al., 2000).
Many criminal justice agencies construct their own psychiatric screening tools, which rarely are subjected to rigorous reliability and validity studies and often are based simply on face validity, which is the lowest level of measurement accuracy. Face validity requires only that a tool appears to be measuring what it purports to measure. Still others rely on probation officers’ subjective judgments about mental health needs, which tend to grossly underestimate the number of probationers with SMI (Lurigio & Swartz, 2006). The best screening tools have high predictive validity: “most of the people who are flagged by [the tool] as being positive should, on assessment, be found to have a treatable serious mental illness” (Steadman, et al. 2005, p. 817).
The screening instrument presently used in the CCJ is typical. More than 300 detainees are processed through reception and classification every day at the jail. They are screened for psychiatric problems by mental health specialists who ask a short series of questions about previous psychiatric hospitalizations, current use of psychiatric medication, and recent thoughts of suicide. This set of questions is consistent with recommended screening practices for jails and prisons and is effective in identifying many individuals who require further assessment and treatment (American Psychiatric Association [APA], 1989; Steadman & Veysey, 1997). Nonetheless, the questions are likely to miss numerous other individuals with SMI, specifically, those who have no previous hospitalizations or current suicidal ideation, or who are taking no psychiatric medications (Teplin, 1990; Teplin et al., 1997; Teplin & Swartz, 1989). Teplin (1990) reported that these criteria missed nearly two-thirds of the mentally ill detainees who were identified with an independently administered and standardized clinical instrument.
As we noted above, screening instruments for psychiatric disorders have been validated on samples from the general population but not on samples from criminal justice populations. The accurate estimate of the underlying prevalence of a disorder is important for clinical decision-making purposes; for example, a determination of when a person’s symptoms suggest that further evaluation and treatment are warranted (Schmitz et al., 2000). Thus, a screening tool for correctional settings should be tested on criminal justice populations, where the prevalence of psychiatric disorders is higher than it is in the general population (HRW, 2003). The choice of a corrections-based screening tool also must consider the conditions that apply to screening in criminal justice settings, which place constraints on the kinds of screening instruments that are best suited for those contexts where there is a shortage of clinical services and treatment slots and a high volume of screenings, which are conducted by correctional staff persons who have no clinical or diagnostic expertise.
Hepburn (1994) recommended that screening instruments for substance use disorders in criminal justice settings have standardized and replicable scoring criteria that can be implemented and interpreted by lay interviewers. They also should be brief and easy to administer without the need for extensive training. Screening for psychiatric illnesses in a criminal justice setting is subject to the same contextual limitations and pressures as screening for substance use disorders. Therefore, the ideal psychiatric screening device would have the same properties as a substance abuse screening device.
Defining the appropriate content of a screening instrument for psychiatric disorders is much more difficult than it is for substance use disorders, for two basic reasons. First, the Diagnostic and Statistical Manual of Mental Disorders (DSM-IV-TR), Axis I (Clinical Disorders) has 15 general categories of non-substance use disorders (APA, 2002). It is possible to screen briefly for all drugs of abuse or simply for the presence of any substance use problem. However, it is impossible (at the very least, highly unwieldy) to efficiently screen for every Axis I psychiatric diagnosis. Second, even if it was possible to screen for every DSM-IV, Axis I disorder, not everyone with a disorder needs treatment. Clinical severity and treatment need lie on a continuum; even some persons with seemingly severe psychiatric disorders are able to function adequately without clinical intervention (Regier et al., 1998). The challenge is to determine which psychiatric disorders should be included in a screening tool and to define when a disorder is severe enough to warrant clinical intervention or further assessment.
Candidate Screening Instruments
A couple of approaches have been adopted to simplify standardized screening procedures for psychiatric disorders. The first determines if an individual meets the diagnostic criteria for a few diagnoses that are likely to be clinically severe and to require treatment and other interventions. This restricted subset of all DSM-IV diagnoses typically includes those that are widely regarded as SMIs: schizophrenia, bipolar disorder, and major depressive disorder. Similar to the longer assessment instruments for psychiatric diagnosis, such as the Composite International Diagnostic Interview (CIDI) (Robins et al., 1988) from which they were derived, these instruments are modular. Each module consists of a sequence of questions for diagnosing a specific disorder or class of disorders.
Among the tools that adopt a diagnostic approach to screening are the Composite International Diagnostic Interview – Short Form (CIDI-SF) (Kessler et al., 1998), the Mini-Neuropsychiatric Interview (MINI) (Sheehan et al., 1998), and the Referral Decision Scale (RDS) (Teplin & Swartz, 1989). The administration time for these relatively brief instruments can be shortened further by omitting modules that screen for disorders that are of little interest to clinicians or researchers.
Despite their administrative flexibility, problems of over- and under-identification with this class of instruments limit their usefulness, particularly in resource-constrained criminal justice settings in which accuracy is crucial. The diagnostic approach to screening equates the need for clinical intervention with diagnosis. Persons who meet the diagnostic criteria for one or more disorders are referred for further assessment and possibly treatment. Those who meet none of the full criteria for any disorders are not referred for either further assessment or treatment.
The potential drawback of such techniques is that they miss persons who have a severe disorder that is not contained in the screening tool (such as post-traumatic stress disorder, generalized anxiety disorder) but that nevertheless requires clinical intervention. These errors are called false negatives: concluding that a person with mental illness has no disorder. Such individuals might receive no treatment, while detained or under supervision, despite the clinical severity of their conditions. They are at risk for problems associated with untreated SMI (e.g., rearrest, homelessness, violence, substance abuse). Moreover, despite the administrative flexibility of being able to select the diagnoses that are included in the screening, the necessity of obtaining a DSM-IV diagnosis adds a level of complexity to the instruments that is attributable to the use of skip patterns and question probes. Given the lack of clinical and interviewing skills among corrections staff, the inclusion of even a small number of skip patterns and probes can sharply reduce the validity and reliability of the instrument.
The problems related to using a diagnostic approach to screening for psychiatric treatment have recently led to a second approach that de-emphasizes diagnosis and focuses instead on symptom severity and level of impairment (Kessler et al., 2002). Although this approach has been discussed widely in the literature (see Murphy, 2002), it has recently gained currency because large-scale epidemiological surveys, such as the Epidemiological Catchment Area (ECA) study and the National Comorbidity Survey (NCS), have found surprisingly high prevalence rates of psychiatric disorders (Kessler et al., 1994; Regier et al., 1990).
In both studies, between 20 and 30 percent of the general population met the DSM-IV criteria for at least one past-year Axis I disorder. As it is unlikely that this large a proportion of the general population required mental health treatment services, the findings of these surveys were of limited usefulness in guiding federal and state treatment resource allocations. These findings also suggested that screening for symptom severity and level of functional impairment is a more efficient way of discerning the need for psychiatric treatment (see Regier et al., 1998; Slade & Andrews, 2002).
The tools related to the symptom-severity approach are particularly applicable for use in criminal justice settings. Their advantages include the use of briefer screening instruments, without skip patterns and probes, which makes the screening tool simpler to administer. Such screening tools can be implemented by lay interviewers to identify individuals with the most severe psychiatric disorders, regardless of diagnosis. This approach conserves limited resources for only those mentally ill persons most in need of services. In other words, such tools can avoid false positives, which identify as mentally ill persons whose symptoms are not severe enough to warrant treatment. A low false positive rate is especially important in criminal justice settings, in which scarce mental health resources must be used sparingly (Steadman et al., 2005).
SMI Screening Tools for Criminal Justice
K6/K10. Among the class of instruments that take this approach to screening are the K6/K10 scales that, for the reasons we mentioned above, appear to be especially promising for use with criminal justice populations (Kessler et al., 2002). Kessler et al. (2002) began the tool-building process with a large pool of items derived from an extensive battery of psychological instruments. They then used analytic procedures, based on item-response theory, to distill a subset of 10 questions (the K10) and a completely overlapping subset of 6 questions (the K6).
The K6 asks respondents how often in the previous month they felt “nervous,” “so sad that nothing could cheer [them] up,” “restless or fidgety,” “hopeless,” “everything was an effort,” and “worthless.” These questions identified with maximum sensitivity the individuals who met the following two criteria: a past-year diagnosis of any DSMIV Axis I psychiatric disorder and a Global Assessment of Functioning (GAF) score below 60 (i.e., moderate to severe impairment in functioning) (see APA, 2002, and Endicott et al., 1976).
Further calibration of the K6/K10 scales was done to select cutoff scores that identify individuals above the 90 th percentile in symptom severity, consistent with estimates which suggest that 6-10 percent of the general population needs psychiatric treatment services at any given time (Kessler et al., 2002). In validity studies, the K6 scale has performed as well as the K10 in identifying individuals with SMI and has become the more widely used instrument (Kessler et al., 2003). The K6 is now included in national surveys, such as the National Survey on Drug Use and Health (NSDUH) and the National Health Interview Survey (NHIS).
Despite its widespread application in general population studies of psychiatric disorders, we are the only researchers to validate the K6 for use with a criminally involved sample of persons. We demonstrated the use of the K6 scale with a sample of adults who reported an arrest in the past year, and compared the classification results of the K6 screen with those obtained using a common, but unvalidated, set of screening questions (e.g., receipt of past psychiatric treatment services, use of prescribed psychiatric medications). Specifically, we compared the diagnostic accuracy of the unvalidated set of questions with the K6 classification results, and examined the characteristics of participants who are misclassified in order to understand why they were incorrectly identified by standard criteria. We also examined the characteristics of offenders who screened positive on the K-6 scale for SMI, compared with those who screened negative (Swartz & Lurigio, 2005; Swartz & Lurigio, in press).
We found that nearly 20 percent of the 1,700 participants with a past-year arrest in the 2002 NSDUH sample had a K6 score of 13 or higher, indicating that in the past year they had experienced symptoms of severe psychological distress consistent with the presence of SMI. Our findings also indicted that all the items in the K6 work equally well in detecting SMI among arrestees for both genders. In addition, we found the same pattern of item and scale consistency among different racial/ethnic and age groups. Respondents with SMI were more likely than those with no SMI to report a past-year substance use disorder that met the DSM-IV criteria for dependence or abuse; they also were more likely to have received drug abuse and mental health treatment in the past year (Swartz & Lurigio, 2005; Swartz & Lurigio, in press).
Brief Jail Mental Health Screen. The Brief Jail Mental Health Screen (BJMHS) was derived from the RDS (Teplin & Swartz, 1989) and consists of eight dichotomous questions. The first six ask respondents whether they currently believe someone is putting thoughts into, or taking them out of, their heads; whether they currently believe that other people read their minds; whether family or friends have noticed that they are more active than usual; whether they have gained or lost weight for several weeks without trying to do so; and whether they have currently felt useless or sinful. The last two questions ask whether they are currently taking any medication for emotional or mental health problems and whether they have ever been in a hospital for emotional or mental health problems.
To test the concurrent validity of the BJMHS, Steadman et al. (2005) studied samples of detainees who were and were not referred for mental health services based on the scale scores. Each of the detainees in those groups also was evaluated with the Structured Clinical Interview for DSM-IV (SCID), which provides a diagnosis-driven assessment of mental health problems. In this study, the SCID served as the gold standard, or the accepted reference or diagnostic test, for psychiatric illness. Results showed that the BJMHS correctly classified as having a diagnosable mental illness nearly 75 percent of the male detainees but only 62 percent of the female detainees. The researchers concluded that the BJMHS is, overall, a practical, simple, and efficient tool for psychiatric screening in jails, but that it has an unacceptably high false-negative rate for women.
Summary and Conclusions
Individuals with SMI are over-represented in the criminal justice system. Since Abramson’s (1972) seminal work on the criminalization of the mentally ill, such persons have become more abundant in terms of their absolute numbers and proportionate representation in correctional populations. Their prevalence in jails, prisons, and probation caseloads is likely to grow even further unless fundamental changes take place in the circumstances that contribute to their involvement in the criminal justice system. However, the historical factors that encouraged the criminalization of the mentally ill are still common today: the lack of community-based mental health services for persons with SMI, the splintering of the drug abuse and mental health treatment systems, and the legal restrictions on involuntary commitment (Lurigio, 2005).
Persons with SMI often reside in environments where crime and opportunities to commit crime are rampant, drug and crime enforcement strategies are aggressive, and mental health and drug abuse treatment programs are limited or inaccessible—all of which increase the likelihood that the mentally ill will be caught in the criminal-justice web (Draine, 1993; Fisher, Silver, & Wolff, in press). In addition, the co-occurrence of psychiatric and substance use disorders, which is more common in the criminal justice population than in the general population, elevates the risk of arrest, violent behavior, hospitalization, incarceration, recidivism, and a host of other adverse life events. The failure to treat SMI, substance use disorders, and their comorbidity jeopardizes public safety, promotes recidivism, and can result in legal liability for criminal justice departments that are unresponsive to clients’ needs for mental health services (Lurigio, 2003).
Effective screening is the first step in properly addressing the behavioral healthcare problems of the mentally ill in the criminal justice system. Mental health resources within agencies and communities are scanty; therefore, accurate identification for direct service provision or referral is paramount. As we have noted in this article, corrections professionals typically have no background, training, or experience in the assessment or treatment of mental illness. Moreover, information on treatment history and current use of psychiatric medication is not always an accurate indicator of a current mental health problem. Similarly, subjective judgments regarding mental health treatment need are fraught with errors and lead to inconsistent and biased decisions among probation officers and other professionals, including those with expertise in the mental health field (Lurigio & Swartz, 2006).
We recommend the use of the K-6 as a promising screening instrument for criminal justice populations. The BJMHS is also a viable option for such purposes but should be further refined to improve its specificity with women offenders. For both tools, more validation studies should be conducted to test their preciseness with different correctional populations. In addition, the construction of mental health screening tools for correctional clients should focus on identifying persons with comorbid psychiatric and substance use disorders. Such a screening tool is presently being developed as part of the large-scale, federally funded project known as the Criminal Justice-Drug Abuse Studies Initiative (CJ-DATS), which is funded by the National Institute on Drug Abuse (Sacks, Melnick, & Coen, 2005). Finally, feasibility, time, and resource studies must be undertaken to examine the effects of widespread mental heath screening in criminal justice settings. The bottom line is whether screening actually results in better outcomes for individuals with SMI who are involved in the criminal justice system.