essay help

International Journal of Offender Therapy and

Comparative Criminology 2017, Vol. 61(8) 919 –937

© The Author(s) 2015 Reprints and permissions:

sagepub.com/journalsPermissions.nav DOI: 10.1177/0306624X15608823

journals.sagepub.com/home/ijo

Article

Selecting a Method of Case Identification to Estimate the Involvement of People With Mental Illnesses in the Criminal Justice System: A Research Note

Melissa S. Morabito1 and Amy Blank Wilson2

Abstract Arrest and incarceration are a pervasive reality for people with mental illnesses. Wide variation, however, exists in the estimates of the percentage of people with mental illnesses who become involved in the criminal justice system. Researchers and practitioners need a variety of methods in their toolbox to maximize their ability to identify mental illness depending on available resources and needs. Yet, the benefits and costs of utilizing these different approaches have yet to be explored in the criminal justice literature. To begin exploring the utility of the different methods of case identification, we review the most commonly used approaches to identifying people with mental illnesses and end with a detailed examination of the use of behavior health records. The use of behavioral health records is a case identification method that has gained emerging support in criminal justice research in recent years.

Keywords mental illness, measurement, behavioral health

Arrest and incarceration are a pervasive reality for people with mental illnesses (Ditton, 1999; James & Glaze, 2006; Steadman, Osher, Robbins, Case, & Samuels, 2009; Teplin, 1984). However, wide variation exists in the estimates of the percentage

1University of Massachusetts Lowell, USA 2University of North Carolina at Chapel Hill, USA

Corresponding Author: Melissa S. Morabito, School of Criminology and Justice Studies, University of Massachusetts Lowell, 113 Wilder St., Lowell, MA 01854, USA. Email: melissa_morabito@uml.edu

608823 IJOXXX10.1177/0306624X15608823International Journal of Offender Therapy and Comparative CriminologyMorabito and Wilson research-article2015

920 International Journal of Offender Therapy and Comparative Criminology 61(8)

of people with mental illnesses who become involved in the criminal justice system. Some variation in estimates is understandable because different segments of the sys- tem have contact with varying subsets of this population. For example, police officers interact with a much broader subset of the population than prosecutors or correctional officers. Nevertheless, unexplained differences exist across estimates made within the same component of the criminal justice system and within the same geographic areas (see Ditton, 1999; Steadman et al., 2009). These differences make measurement and planning challenging for practitioners and policy makers.

One of the first researchers to question this inconsistency in estimates identified two methodological issues as potential causes for the variation: (a) the definition of mental illness used in research and (b) the methods used to identify cases for study (Teplin, 1983). Since Teplin’s initial work, researchers have begun to address how the lack of a standardized definition can affect estimates of those with mental illnesses who become involved in the criminal justice system (hereafter, justice-involved per- sons with mental illnesses; Draine, Wilson, & Pogorzelski, 2007; Roesch, Ogloff, & Eaves, 1995; Teplin, 1990). However, the criminal justice literature has not yet exam- ined other issues affecting estimates of this population, including the different meth- ods of case identification used to formulate estimates of the rate of mental disorders in justice-involved populations, or a cost-benefit analysis of each method.

To fill this gap, this article explores the utility of different methods of case identifi- cation, reviewing the case identification methods most commonly used to identify justice-involved persons with mental illnesses within the different criminal justice set- tings (e.g., police, courts, corrections). These case identification methods include diagnostic interviews, participant observation, and self-report surveys.

Case Identification Methods Commonly Used to Establish Estimates

Criminal justice agencies must identify people with mental illnesses under their super- vision or jurisdiction for two main reasons. First, it is imperative to have a general estimate of the proportion of individuals in their care with mental illness to make deci- sions about the allocation of resources. Second, the estimates of the number of indi- viduals with specific mental health diagnoses is also crucial, because it facilitates communication and coordination with the community-based public mental health sys- tem, which is responsible for treating people with the most serious mental illnesses. To identify justice-involved persons with mental illnesses in various criminal justice set- tings, researchers and policy makers must first establish a protocol for defining mental illness. A critical component of this decision-making process centers on the research- er’s selection of the diagnoses or behaviors included in this definition. For example, academic researchers whose interest is in understanding the scope of issues facing prison populations might include a broad range of mental health diagnoses in their definition of mental illness. In contrast, jail administrators who want to effectively target the jail’s limited resources might establish a narrow definition of mental illness that includes only those diagnoses that will identify inmates with the most serious and

Morabito and Wilson 921

persistent mental illnesses. Once a definition of mental illness is established, research- ers must then select a method for identifying people who meet the criteria established by the definition (i.e., a case identification method).

Table 1 provides a comparison of the three case identification methods most fre- quently used to establish rates of justice-involved persons with mental illnesses by some of the most cited studies in the literature.1 The first two columns organize the three methods by the criminal justice setting, and the rows document some of the most frequently cited estimates of the rates of justice involvement among people with seri- ous mental illnesses, organized by setting and case identification method. The third column compares the rates of mental illness. As illustrated by this table, substantial variation exists between both between- and within-case identification methods. As previously noted, some variation in estimates is to be expected because of differences in the volume of contact and nature of involvement in different criminal justice set- tings. However, Table 1 shows that considerable variation also exists in estimates gen- erated using the same case identification methods, within the same criminal justice setting. This point is best illustrated in the differences in estimates associated with Steadman et al. (2009), Teplin (1990), and Teplin, Abram, and McClelland (1996). These studies used diagnostic interviews to identify individuals with the most severe forms of mental illness. Yet, when compared, these estimates show a difference of 16

Table 1. Case Identification Methods.

Criminal justice setting

Case identification method

% of population with mental illness Source

Police Police contacts Observational 5.90 Teplin (1984) Observational 3.60 Engel and Silver (2001) 2.70 Jail Jail detainees Diagnostic interview 6.36 (men with

current diagnosis) Teplin (1990)

9.48 (men with lifetime disorder)

Diagnostic interview 15.0 (women with current diagnosis)

Teplin, Abram, and McClelland (1996)

18.5 (women with lifetime disorder)

Self-report survey 16.0 Ditton (1999) Self-report survey 64.20 James and Glaze (2006) Diagnostic interview 14.50 (men) Steadman, Osher, Robbins,

Case, and Samuels (2009) 31.0 (women) Prison Prison inmates Self-report survey 16.20 Ditton (1999) Self-report survey 56.20 James and Glaze (2006)

922 International Journal of Offender Therapy and Comparative Criminology 61(8)

percentage points in the number of women with a current mental health diagnosis in the two studies and a difference of 7 percentage points among men.

The literature contains a growing number of explanations for these variations in rates. Some explanations focus on the lack of precision in certain case identification methods, such as self-report surveys. Other explanations focus on variability in the definitions of what factors are considered as indicators of mental illness in the case identification process (Steadman et al., 2009; Teplin, 1983). Other explanations have pointed to the possibility that estimates deviate, given the naturally occurring temporal and geographic variations that occur within the populations being studied (Steadman et al., 2009). The accuracy of these proffered explanations cannot be established because the existing data do not provide answers as to the extent of variation in rates of involvement is attributable to the factors described above. However, the fact remains that policy makers and program developers need estimates of the number of justice- involved persons with mental illnesses that are stable and consistent (i.e., reliable). Reliable estimates of both those generally involved in criminal justice as well as esti- mates of those under the jurisdiction of localities are required to develop programs of appropriate scale and with adequate resources to meet the needs of this population. In addition, due to the focus on people with the most severe and persistent mental ill- nesses (i.e., schizophrenia spectrum and major affective disorders), in the United States’s public mental health system, these estimates will be most useful if they are also diagnostically specific. Given that none of the case identification methods in cur- rent use offers a guarantee of a reliable estimate (see Table 1), we examine the strengths and weaknesses of each method.

Diagnostic Interview

The diagnostic interview is commonly used in psychiatric research to identify people with mental illnesses for research purposes. The wide acceptance of this approach as one the most empirically defensible ways to estimate the rate of mental illnesses has led to the diagnostic interview being regarded as the gold standard in research involv- ing psychiatric diagnoses (Draine et al., 2007; Nordgaard, Revsbech, Saebye, & Parnas, 2012). The instruments used for diagnostic interviews in research settings are structured, empirically validated instruments intended to identify mental health diag- noses in study populations.

One of the most widely cited studies that used a diagnostic interview to identify justice-involved persons with mental illnesses was conducted by Teplin and her asso- ciates in Chicago during the 1980s. This study yielded several publications that exam- ined the rate of men and women with serious mental illnesses in the jail system (Teplin, 1990; Teplin et al., 1996). The researchers drew a random sample of individuals admitted to the Cook County Correctional Facility in Chicago and then used the National Institute of Mental Health Diagnostic Interview Schedule, Version III, a structured diagnostic assessment tool designed specifically to identify people with a range of psychiatric disorders (Robins, Helzer, Croughan, & Ratcliff, 1981). As Table 1 illustrates, Teplin and colleagues found that approximately 9.5% of men and 18.5% of

Morabito and Wilson 923

women entering this correctional facility met the diagnostic criteria of having a severe mental illness during their lifetime2 (Teplin, 1990; Teplin et al., 1996).

One of the more recent studies to use the structured diagnostic interview to investi- gate rates of mental disorders in the criminal justice population was conducted by Steadman et al. (2009), using samples of inmates recently admitted in five county jails in Maryland and New York. These samples of inmates were screened for mental disor- ders using the SCID—Structured Clinical Interview for the Diagnostic and Statistical Manual of Mental Disorders (4th ed.; DSM-IV; American Psychiatric Association [APA], 1994; First, Gibbon, Spitzer, & Williams, 2001), which is a revised version of the instrument used by Teplin and colleagues (1996). Steadman et al. found that 14.5% of the men and 31% of the women had a current serious mental illness.

Using structured clinical interviews as a method of case identification poses limita- tions because different diagnostic interviews are available, and each require substantial resources and have different levels of precision and diagnostic accuracy (Rogers, Sewell, Ustad, Reinhardt, & Edwards, 1995; Teplin & Swartz, 1989). The SCID (First et al., 2001) has been used in numerous studies and found to be an accurate measure of psychiatric diagnoses (Basco et al., 2000; Fennig, Craig, Lavelle, Kovasznay, & Bromet, 1994; Lobbestael, Leurgans, & Arntz, 2010). The SCID is designed to provide diagnostic information related to the full range of Axis I (i.e., schizophrenia and major depressive disorder) diagnoses. This diagnostic tool has recently been updated to ensure alignment with the diagnostic categories found in the Diagnostic and Statistical Manual of Mental Disorders (5th ed.; DSM-5; APA, 2013). There is also a separate version of this diagnostic tool, called the SCID-II that assesses Axis II (anti-social personality disorder or paranoid personality disorder) diagnoses that are present in DSM-IV (APA, 1994). The SCID-II is currently being revised to ensure alignment with the DSM-5 and is expected to be available in fall of 2015 (www.scid4.org). To use any of the SCID interviews, staff must purchase and complete training developed specifically for this diagnostic interview. The SCID was developed for use with clinicians and other mental health professionals (www.scid4.org). It can also be administered by research assis- tants, without prior clinical experience; however, these individuals are likely to require more training before they can begin using this interview (www.scid4.org).

Diagnostic interviews provide the diagnostic specificity needed to identify people in the justice system with a wide range of psychiatric disorders, while also providing the information needed to identify the subset of individuals with psychiatric disorders who will need services from the public mental health system. However, this precision comes at an expense. Diagnostic instruments, such as the SCID, represent a time- consuming, resource-intensive method of case identification. For example, the SCID is administered as a one-on-one interview that requires an average of 90 min to com- plete (Lecrubier et al., 1997). Moreover, it is recommended that interviewers adminis- tering the SCID complete SCID-specific training that takes at least 20 hr to complete (http://www.scid4.org/faq/scidfaq.html).

The Mini-International Neuropsychiatric Interview (MINI; Lecrubier et al., 1997) is a short diagnostic interview that was designed to be an alternative to longer diagnos- tic interviews such as the SCID (Sheehan et al., 1998). It is reported to be one of the

924 International Journal of Offender Therapy and Comparative Criminology 61(8)

most widely used psychiatric diagnostic assessment tools (Medical-outcomes.com). The accuracy rates of the MINI have been found comparable with those of longer diagnostic interviews such as the SCID (Lecrubier et al., 1997). However, two major differences exist between the MINI and other longer diagnostic interviews such as the SCID: administration time and diagnostic range. The MINI is completed in 15 to 20 min (Lecrubier et al., 1997). However, as compared with the SCID’s capacity to assess all DSM’s major mental health diagnoses, the MINI assesses a smaller range of diag- noses, with the capacity to identify only 17 psychiatric and substance use diagnoses (Lecrubier et al., 1997). The MINI’s diagnostic range does include major affective disorders and schizophrenia. In addition, the MINI has not been updated to align with the diagnostic categories in the DSM-5, and its diagnostic capacity is largely limited to current disorders, rather than a comprehensive assessment of current and lifetime dis- orders. Despite these limitations, the relatively shorter format of the MINI makes it a more feasible and readily administered diagnostic instrument in many situations.

Similar to the SCID and other diagnostic instruments, the MINI requires face-to- face interviews conducted by researchers who are trained in the administration of the MINI. However, despite its shortened length, conducting diagnostic interviews of any type can exceed the available time, expertise, and financial resources of many criminal justice agencies.3 Even when such resources are available, diagnostic interviews might not be a feasible method of case identification, especially in settings where even short interviews might not be possible because the person with a suspected mental illness is not interested in being interviewed and cannot be compelled to stay because he or she is not in police custody (e.g., police stations, courts). This limitation suggests that using diagnostic interviews to identify justice-involved persons with mental disorders will be restricted to a few well-funded studies in sites that can accommodate interview formats (e.g., jails, prisons).

Participant Observation

Participant observation is one of the more popular methods used to identify police interactions involving people with mental disorders (Denzin, 1989). In studies using this method, researchers with diagnostic training accompany police officers on patrol and observe and track the number and context of police–citizen interactions that involve people with mental disorders (Engel & Silver, 2001; Novak & Engel, 2005; Teplin, 1984). In Teplin’s (1984) seminal study of people with mental illnesses involved in police–citizen interactions, the observers used a structured clinical check- list designed to identify behaviors associated with diagnosable mental disorders. This checklist enabled the trained observers to make a clinically informed assessment of mental illness independent of the officers’ determination of mental illness status. Results indicated that 5.9% of police–suspect encounters involved a person with a serious mental illness (Teplin, 1984).

In a test of Teplin’s (1984) research, Engel and Silver (2001) examined police–citizen interactions involving people with mental illness. Engel and Silver’s study included a reanalysis of data collected in two earlier studies: the Project on Policing Neighborhoods

Morabito and Wilson 925

(POPN; see Parks, Mastrofski, & DeJong, 1999) and the Police Services Study (Engel & Silver, 2001). These studies examined police–citizen interactions in five metropoli- tan areas, and both studies used trained observers who accompanied officers during their shift (Parks et al., 1999). However, in the studies analyzed by Engel and Silver, the observation method differed from the method used in other studies. Rather than identifying suspects with mental illnesses, the intent of the observations in these stud- ies was to identify situations in which officers were likely to perceive of the suspects as mentally ill. Therefore, rather than a clinical-diagnostic checklist, these observers used a rating sheet to identify behaviors generally considered indicative of a mental illness (Engel & Silver, 2001). The analytic results showed that 3.6% of police con- tacts in the POPN Study and 2.7% of contacts in the Police Services Study involved a suspect with a mental illness.

These participant-observation studies have greatly advanced the field, in increasing knowledge of the rates of contact between police and people with mental illnesses. However, these studies also vary widely in their estimates of the number of people with mental illnesses involved in the criminal justice system. Even the researchers noted that the variation in estimates across studies was likely explained by the diverse methods used to identify mental illness (Engel & Silver, 2001). The small number of contacts involving persons with mental illness found in both studies (2.7%-3.6%; Engel & Silver, 2001; Teplin, 1984; 5.9%) questions whether future efforts should use these resource-intensive methods of case identification.

The use of trained observers for identification is both a strength and weakness. As compared with relying on police officers’ assessments of a suspect’s mental status, the trained observers can increase the number of contacts correctly identified as involving a person with mental illness. The observers’ training enables them to identify nuances in symptomatic behavior that are likely to be overlooked by police officers without this specialized training. However, even trained observers tend to under-identify cases because the symptoms of mental illness are cyclical, meaning those with mental ill- nesses might not be symptomatic at the time of a police encounter. Furthermore, men- tal illness may be masked by substance abuse—a difficult relationship for untrained observers to untangle. Another weakness associated with the use of trained observers is the significant expense incurred. Similar to the resource-intensive diagnostic inter- view, the costs associated with participant-observation techniques for case identifica- tion typically restrict this method to large-scale publically funded studies investigating various aspects of police behavior (Engel & Silver, 2001; Novak & Engel, 2005).

Survey

Survey methods have also been applied to case identification to generate estimates of the number of justice-involved persons with mental illnesses (Ditton, 1999; James & Glaze, 2006). For example, a study conducted by James and Glaze (2006) with two nationally representative samples used a survey to assess the rates of mental health problems among jail and prison inmates. To identify inmates with mental health prob- lems, the survey included items from a structured clinical interview organized into a

926 International Journal of Offender Therapy and Comparative Criminology 61(8)

self-administered questionnaire. The survey items asked inmates about mental health symptoms they had experienced over the past 12 months but did not collect data on symptom severity or duration. Moreover, the survey did not collect data that would allow researchers to rule out symptoms caused by issues other than mental illness, such as substance use or other medical conditions (Steadman et al., 2009). James and Glaze found that 56% of state prisoners and 64% of local jail inmates had mental health problems.

Although James and Glaze (2006) found that more than half of prison and jail inmates had mental health problems, an earlier study conducted by Ditton (1999) reported a significantly lower rate of mental illness among inmates. Similar to James and Glaze’s study design, Ditton’s study also used a self-report survey with a nation- ally representative sample of jail and prison inmates, but found that only 16% of the sample reported a mental illness. The difference in the estimates between these two studies is likely the result of the different methods used to identify mental illness. James and Glaze used self-reports of symptoms (but did not account for duration or severity or symptom-related issues, such as substance use or medical conditions), whereas Ditton’s survey assessed presence of mental illness based on inmates’ self- reports of mental or emotional problem or an overnight stay in a psychiatric facility. Ditton’s survey has been criticized by scholars who consider the items to be an overly broad indicator of serious mental illness (Draine et al., 2007).

Self-reported data collected through surveys are a relatively inexpensive way to estimate the rates of involvement of persons with mental illnesses in the criminal jus- tice system. As compared with diagnostic interviews and observational methods, the survey method requires fewer resources to collect data. For example, using self- administered surveys does not require trained interviewers to administer the survey, thus, this method substantially reduces costs associated with data collection. In turn, the lower costs of the survey method allow researchers to assess larger study samples in shorter periods.

Nevertheless, surveys have substantial limitations. Surveys must rely on less accu- rate and often overly broad indicators of mental illness because the interview format of the survey does not allow for clinically based assessments of specific mental health disorders (Teplin, 1983). This lack of clinical precision in case identification will skew the estimates of mental illness in ways that are difficult to predict. Surveys could use brief screening tools, such as the Symptom Checklist–90 (SCL-90; Derogatis & Unger, 2010), to identify people who are present with symptoms that may be associated with mental illness. However, screening tools are designed to act as triage tools that identify individuals who require further evaluation. Even when used in conjunction with other indicators such as receipt of outpatient treatment or a stay in a psychiatric hospital, these indicators lack the diagnostic specificity needed to identify individuals with the mental health disorders that are the focus of most treatment in the criminal justice and mental health system. In addition, self-report survey data are often rife with missing data, which further degrades the quality of information collected. Therefore, although surveys offer a less expensive and more expedient method of case identification, the overly broad indicators of mental illness and missing data associated with this case

Morabito and Wilson 927

identification method make the use of this method problematic in differentiating the most serious and entrenched forms of mental illness from behaviors associated with situational or personality problems.

Using Behavioral Health Records (BHR) in the Case Identification Process

In recent years, researchers have increasingly used BHR to identify justice-involved persons with mental illness (Baillargeon et al., 2010; Draine, Blank Wilson, Metraux, Hadley, & Evans, 2010; Morrissey, Cuddeback, Cuellar, & Steadman, 2007; Morrissey et al., 2006; Wilson, Draine, Hadley, Metraux, & Evans, 2011). The BHR used in these studies have typically involved insurance reimbursement claims or treatment records compiled during routine treatment. The steps involved in this method of case identifi- cation are outlined in Table 2. This table shows that this method of case identification involves a two-step process. Each step includes an outline of key decisions that need to be addressed when engaging this case identification method, which are discussed further in the text below. However, this method of visual display is not meant to imply that these decisions are sequential in nature. Rather, as the discussion below shows, many decisions are inter-related and so need to be dealt with accordingly.

Establish a Pool of Potential Cases

As outlined in Table 2, the first step of this case identification method involves developing a pool of potential cases. In this case identification method, administra- tive records are used to develop the pool by identifying individuals who have involvement with both the criminal justice and mental health systems. To develop

Table 2. The Case Identification Process Using BHR.

1. Establish a pool of potential cases

Select data files that can identify individuals who are involved in both the criminal justice system and mental health services.

Define the window of observation that will be used for each data file during the matching process.

Select a matching procedure to identify individuals who are present in both data files.

2. Identify cases where individuals have a mental illness

Select BHR that can identify individuals with qualifying mental health diagnoses from the pool of potentially eligible cases.

Identify specific mental health diagnoses that will be searched for in the behavior health records.

Define the window of observation that BHR will be searched for qualifying mental health diagnoses.

Develop procedures to address situations where individuals have multiple mental health diagnoses.

Note. BHR = behavioral health records.

928 International Journal of Offender Therapy and Comparative Criminology 61(8)

the pool, researchers must have access to two types of records: criminal justice con- tact (e.g., police, courts, or jail/prison records) and BHR. At this step, in the case identification process, it is important to consider that the matching process is most feasible when both data files are stored in electronic format and include at least one common unique identifier for each person (e.g., social security number, the person’s first and last names).

Researchers must also define the window of observation that will be used when comparing records between data files. This involves determining the specific time period that data will be extracted from each file for the matching process. For example, researchers could use police records for all arrests that occurred in 2010 in a particular jurisdiction and Medicaid eligibility files for all individuals enrolled in Medicaid dur- ing the same time period. However, if researchers are concerned that a person’s pres- ence in one data set is related in some way to their presence in the second data set, the time periods of observation for each data set could vary to account for this problem. For example, admission to jail or prison can affect a person’s eligibility for Medicaid. So, if a researcher was using records of jail admissions in the matching process, they could choose to vary the time periods used in the window of observation for each data set to account for this potential problem. In this example, the study team might decide to use all jail admissions in 2010 and Medicaid eligibility records for 2009.

The other important decision that researchers need to make during this step in the case identification process is the type of matching procedure that will be used to com- pare records across data files. There are two matching procedures that are generally used in this process: deterministic and probabilistic matching procedures. Deterministic matching procedures identify individuals across multiple data sets by comparing cases using a unique identifier, such as social security number, in each data set. In this method, the unique identifier must be exactly the same in both data sets for a match to be identified. Probabilistic matching procedures identify individuals across multiple data sets by comparing cases using multiple unique identifiers in each data set. During this matching process, an algorithm is used to assign each unique identifier with a weight that indicates how closely the unique identifier matches in the two data sets. A sum of these weights is then used to indicate the likelihood that a match exists between the two records.

Deterministic matching procedures are an expedient method for matching records; however, this method of matching cases can encounter challenges when used with criminal justice records because these records often have inaccurate or missing data. However, recent studies have overcome this problem by incorporating probabilistic matching procedures into the case identification process. For example, the Link King program (http://www.the-link-king.com/download.html; Campbell, Deck, & Krupski, 2008), which is a public domain software, increases the accuracy of the matching process by using a combination of deterministic and probabilistic matching procedures. The inclusion of probabilistic matching procedures in the case identification process allows programs such as Link King to address missing data, misspellings, or other errors in data entry, as well as the use of aliases or nicknames (Campbell et al., 2008).

Morabito and Wilson 929

Identifying Case Where Individuals Have a Mental Illness

Once the researcher has established a pool of potential cases by executing the match- ing process described above, the next step in the case identification process focuses on identifying cases where a mental illness is present. This step in the case identification process involves a number of key decision points that require consideration. Chief among these decisions is locating a record source that contains the diagnostic informa- tion needed to identify people with qualifying mental health diagnoses.

Several different types of BHR can be used during this step in the case identifica- tion process. For example, Medicaid reimbursement claims for mental health services have been used, as have prison treatment records documenting evaluations conducted as follow-up with all inmates whose intake screeners indicated that they were in need of a mental health evaluation.

Although some scholars have expressed concerns regarding the reliability of diag- noses obtained from administrative records, research has demonstrated a comparable reliability of diagnoses in records, such as Medicaid insurance claims and diagnoses obtained in routine clinical practice (Rothbard, Kuno, Hadley, & Dogin, 2004). However, much remains unknown about the consistency and accuracy of the mental health diagnoses recorded in other BHR, such as those maintained by correctional facilities.

In addition to selecting a record source to identify cases that have mental health diagnoses, it is also necessary to specify what mental health diagnoses are being searched for in these records. When selecting mental health diagnoses, it is important to work with individuals who have knowledge of the record source and how it is used in practice, so that the study team understands the range of diagnostic information contained in the records. Once this information is obtained, the number of diagnoses included in the definition has little impact on the time involved in extracting this infor- mation from the records, so the team can include as many or as few diagnoses as they need at this stage in the case identification process.

As in the previous step, it is also necessary to identify the window of observation that will be used to extract diagnostic information from the identified record sources. Establishing this window requires that the specific time period that the records will be searched for diagnostic information will be defined ahead of time. Similar to the selec- tion of mental health diagnoses, it is important for researchers to work with individuals with a working knowledge of the record source to ensure that there is equivalence of diagnostic information contained in the records during the time period that the records are being searched.

It is also important for researchers to try and identify any contextual or situational factors that need be considered when selecting the time period for observation. For example, researchers may consider selecting a window of observation that falls before the time period used in the first step of the case identification process to ensure that mental health diagnoses identified during this step were received before the person’s contact with the criminal justice used in the first step of the case identification process.

930 International Journal of Offender Therapy and Comparative Criminology 61(8)

In this method of case identification, researchers also have flexibility in how they set the beginning and end date for the window of observation. This can be useful when trying to identify populations of consumers who may have sporadic contact with the service provider completing the BHR. In cases such as this, a longer window of obser- vation may be used to adjust for gaps in service. However, in other cases when a record source such as prison classification records are being used, a shorter window of observation may be used because diagnostic information is being pulled from records related to a more systematic and reliable evaluation process.

The last important decision that has to be made during this step in case identifica- tion process is how to address situations where individuals have more than one quali- fying mental health diagnosis in their record. Researchers have not yet reached consensus on how to address this issue. Some studies address the issue of dual or multiple diagnoses by collapsing individual mental health diagnoses of interest into a broader indicator variable such as serious mental illnesses. This variable includes all individuals whose treatment records indicate at least one of the qualifying diagnoses (Draine et al., 2010; Morrissey et al., 2007; Morrissey et al., 2006; Wilson et al., 2011). Although this approach avoids supplanting clinical judgment as to which diag- nosis is “primary,” it comes at the expense of organizing people into mutually exclu- sive categories based on specific mental health diagnoses.

Other methods have been developed to address the issue of individual cases with multiple diagnoses. For example, some studies have used the most frequently occur- ring diagnosis as the diagnosis of record (Becker, Andel, Boaz, & Constantine, 2011), whereas others have relied on the diagnosis assigned during a stay in an inpatient psychiatric hospital (McCabe et al., 2012). Still, other researchers have chosen to use the diagnosis recorded during the prison intake screening and classification process (Baillargeon, Binswanger, Penn, Williams, & Murray, 2009; Baillargeon et al., 2010; Baillargeon, Williams, et al., 2009). Currently, no data are available to assess the rela- tive reliability of the different approaches to assigning diagnoses during the case iden- tification process. However, at a minimum, the different ways in which diagnostic information is dealt with during the case identification process has implications for the comparability of these estimates that must be considered in future research efforts.

Strengths and Weaknesses of the BHR Case Identification Method

The relatively low cost of using BHR to identify justice-involved persons with mental illnesses is a particular strength of this method, hereafter referred to as the BHR case identification. Unlike the case identification methods traditionally used to estimate the rates of mental illnesses in criminal justice settings, BHR case identification is the only method that does not use primary data collection. Instead, BHR case identifica- tion uses existing data, requiring only that the data be abstracted, coordinated, and formatted for analysis. The BHR case identification method substantially reduces the time involved in generating data. However, this method requires not only that project personnel have substantive knowledge of the data used in identification process but also that project personnel have the computing capabilities to abstract and organize the

Morabito and Wilson 931

data for analysis. Even so, the overall time and resources needed for BHR case identi- fication are far lower than those associated with the methods involving primary data collection.

As mentioned previously, the levels of accuracy in BHR case identification are commensurate with traditional clinical interviews conducted by service providers. The recent development of techniques such as the Link King algorithm have strengthened the accuracy and comprehensiveness of BHR case identification by lessening the potential for undercounts that can occur when deterministic matching procedures are used alone with administrative data (Campbell et al., 2008).

Equally important, the BHR case identification method is associated with other strengths that aid the case identification process. By using existing data sets (e.g., criminal justice records, insurance claims, other health records), the BHR case identi- fication method allows researchers and policy makers to capitalize on the growing volume of data stored in digital formats. For example, the Patient Protection and Affordable Care Act (ACA; 2010) requires the use of electronic health care records, and a provision in the American Recovery and Reinvestment Act (ARRA; 2009) funded the conversion of many existing health records to electronic format. These policies have created access to unprecedented levels of electronic health records, which enable researchers to include service-use analyses with case identification efforts at little to no additional costs. Furthermore, BHR case identification provides flexibility in the type and number of mental health diagnoses used to identify cases without increasing the costs. The cost-savings represents a significant advantage over primary data collection methods (e.g., diagnostic interviews), in which each diagnos- tic module increases the time and costs of obtaining the data.

Despite these appealing strengths, BHR case identification also has limitations that must be considered. First, this method provides estimates only of persons with an iden- tified mental illness. Relying on BHR means that the estimates include only those per- sons who have received treatment for mental illness. People who have not received behavioral health services or who have private insurance are not included in the esti- mates; therefore, this method can potentially produce an undercount of those with men- tal illnesses. Some researchers have sought to address this issue by using longer “look back” periods when examining BHR as a way of including cases with sporadic or low levels of receipt of mental health services (Draine et al., 2010; Wilson et al., 2011). Nevertheless, even these larger data windows cannot account for people who have not received any type of formal diagnosis or treatment or those with private insurance.

Next, the BHR case identification method has variable utility for different criminal justice agencies. This approach may be more prohibitive for some agencies such as police because it requires a great deal of resources and could not be done as often. It is more feasible for other agencies with available records. For example, jail and prison personnel could use the BHR method to get estimates and last known diagnoses for current inmates. This information could guide the use of resources and current treat- ment planning. For police, however, who deal with many people with mental illnesses in short interactions, the BHR method would not be feasible for immediate planning. This approach could provide retrospective data for policing and help paint a picture of

932 International Journal of Offender Therapy and Comparative Criminology 61(8)

the populations with which they interact which could be useful for future planning. Although police are competent at identifying mental illness when a citizen is symptom- atic (Fry, O’Riordan, & Geanellos, 2002), they may be undercounting interactions with this population because people with mental illnesses are not symptomatic all the time.

Another limitation of BHR case identification is related to the security and privacy concerns that must be addressed when matching criminal justice data with BHR. For example, legislation such as the Health Insurance Portability and Accountability Act (HIPPA; 1996) strictly controls the sharing of health records. University researchers have worked through their institutional review boards to address the privacy concerns when dealing with protected health data. However, initiatives such as the National Institutes of Health (NIH) Big Data Knowledge project (BD2K; NIH, n.d.) are developing mecha- nisms that will allow localities to share protected data both within and across systems. Initiatives such as BD2K are also creating new funding opportunities that researchers and policy makers can capitalize on to develop the data and resource infrastructure needed to use electronic records in the case identification process (NIH, n.d.). Finally, it is possible that states’ efforts to move Medicaid populations into private sector managed care cover- age, could decrease the accessibility of Medicaid data. However, at least two of the stud- ies cited in this article were conducted in states that had already transitioned the management of Medicaid to managed care entities (Becker et al., 2011; Blank Wilson, Draine, Hadley, & Metraux, 2011), thus it is reasonable to expect that even in this sce- nario, private corporations will still be required to submit these data to the state.

Discussion

Within the criminal justice field, there is near universal agreement that people with mental illnesses are overrepresented in the criminal justice system (e.g., see Ditton, 1999; James & Glaze, 2006; Steadman et al., 2009). However, the same research shows that the extent of involvement varies. Even estimates of the rates of involve- ment of people with mental illnesses in the criminal justice system formulated using the most rigorous method of case identification (i.e., diagnostic interview) vary based on geographic locations where the estimates were made (Steadman et al., 2009). This point underscores the critical importance of formulating estimates of the justice-system involvement of people with mental illnesses that are specific to the locality that will provide and develop services for this population.

Until recently, developing locally specific estimates of justice-involved persons with mental illnesses was beyond the reach of most localities because of the high expense and high level of research expertise needed to accomplish the task. However, the BHR case identification method offers practitioners and policy makers a viable method for examining rates of involvement on a local level. Although this method has limitations—particularly that it excludes people who were previously undiagnosed and those with private insurance, BHR case identification offers the benefit of provid- ing estimates of justice-system involved populations with mental illnesses that are derived from provider-reported diagnostic data, and at a fraction of the cost of con- ducting diagnostic interviews.

Morabito and Wilson 933

We hope the discussion of the methods presented in this article will guide practitioners and policy makers who are considering the use of BHR in the case identification process. For practitioners, using systematic, data-driven methods to determine local rates of involvement of people with mental illnesses maximizes the chances of services and resources being appropriately distributed. For example, police departments must decide which partnerships to pursue and trainings to use, correctional institutions must determine which medication and treatment services will be available to inmates, and reentry person- nel must link newly released offenders with the appropriate mental health services. Personnel, access, and service availability are all based on estimates of the number of people in need of these services. This approach is not the answer for every agency in every situation but may be useful for some. For some criminal justice agencies, this may be the only way to screen large populations such as arrestees and jail inmates.

Researchers have taken diverse approaches to identifying people with mental ill- ness who are involved in various criminal justice settings. Given the nature of police work, including the brevity of many police–citizen encounters, researchers have relied on trained observers to identify which of these encounters involved a person with mental illness (Engel & Silver, 2001; Teplin, 1984). Although the courts and jail set- tings have provided researchers with better access to justice-involved populations, these research efforts have lacked a standardized definition of mental illness and used multiple methods to identify people with mental illness, resulting in wide variability in estimated numbers of this population. However, the recent societal-level investments in digitizing medical records (American Recovery and Reinvestment act (ARRA: 2009) combined with the new mandate to use electronic records in health care settings (ACA, 2010) are creating volumes of digital data that can be used to identify individu- als across all levels of involvement in the criminal justice system. The availability of these new data sources has coincided with significant public investment in developing technological resources to mine these data resources; this intersection of data access availability of new methods will create opportunities for researchers to explore the involvement of people with mental illnesses at all points of the criminal justice system. Although a discussion of the full range of implications of the “big data revolution” on research with justice-involved people with mental illnesses is beyond the scope of this article, the resources generated by these efforts address limitations of the case identi- fication methods that have prohibited use of big data in the past.

The criminal justice field is entering a research era that is likely to be defined by the big data revolution, wherein large administrative data sources will become the primary data source for many studies, especially research on health and health care services use (NIH, n.d.). However, this data revolution also offers criminal justice practitioners and policy makers the infrastructure needed to use existing data to guide real-time decisions regarding mental health services that require additional development and testing. Part of such exploration should involve tests of the accuracy of the BHR case identification method as reliable information is needed, but for it to be useful for service planning in the public mental health system, it also has to be diagnostically specific. However, these tests should involve real-world considerations—such as cost-benefit analysis of increased accuracy or improved diagnostic precision—compared across case

934 International Journal of Offender Therapy and Comparative Criminology 61(8)

identification methods to enable practitioners and policy makers to use the data to make the best decisions possible based on the needs and resources present in their specific localities. Finally, these are not always mutually exclusive methods. The best way to capture the rate of mental illness including the previously undiagnosed may be some combination of these approaches. For example, it may be useful to merge jail and prison classification records with BHR or to conduct a brief survey of inmates to identify the undiagnosed. It is true that only direct screening can catch the undiagnosed and the truly unidentified, but given the resources required to conduct diagnostic interviews, this method cannot be used in a wide-scale manner. A combination of these methods could widen the net and include more people who are in need of treatment.

Conclusion

Researchers and practitioners must consider the positive and negative aspects of every method of case identification. Estimates of mental illness in criminal justice popula- tions are crucial to making public policy decisions regarding funding for programs such as Crisis Intervention Teams (CIT), mental health courts, and reentry programs. Such programs and the criminal justice knowledge base would benefit significantly if the same method was used to measure the extent of involvement of people with mental illnesses at all points in the criminal justice system. Although every police–citizen encounter does not lead to an arrest and jail time, and every jail detainee is not sen- tenced to prison, some overlap does exist among these populations.

We have demonstrated the wide variation that exists in estimates of mental illness within some criminal justice settings. Notably, even the most conservative rates dem- onstrate that a large number of people with mental illnesses become involved with the criminal justice system, and such involvement occurs at disproportionately higher rates than community samples (Teplin, 1990; Teplin et al., 1996). Thus, it is important for local jurisdictions to understand the size of the population they are trying to address, especially given that recent estimates have been shown to vary substantially across local jurisdictions (Steadman et al., 2009).

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.

Notes

1. Based on Google Scholar counts, the studies that had the most citations of rates of involve- ment were included in the table. This list is not meant to be inclusive of all studies that esti- mated rates of involvement of mental illness in criminal justice populations. These studies

Morabito and Wilson 935

are not meant to be the most recent or most rigorous in the field, rather they are indicative of the research that is most utilized in the field.

2. For serious mental illnesses, a diagnosis at any point in the lifetime of the individual will be important information for criminal justice personnel because individuals with these diagno- ses require ongoing treatment to support the management of these disorders and alleviation of symptoms during active stages of the disorders.

3. Ideally, a mental health clinician with training in interpreting these measures would conduct these interviews or would supervise non-mental health professionals.

References

American Psychiatric Association. (1994). Diagnostic and statistical manual of mental disor- ders (4th ed.). Washington, DC: Author.

American Psychiatric Association. (2013). Diagnostic and statistical manual of mental disor- ders (5th ed.). Arlington, VA: American Psychiatric Publishing.

American Recovery and Reinvestment Act (ARRA) of 2009, Pub. L. No. 111-5, 123 Stat. 115, 516 (Feb. 19, 2009).

Baillargeon, J., Binswanger, I. A., Penn, J. V., Williams, B. A., & Murray, O. J. (2009). Psychiatric disorders and repeat incarcerations: The revolving prison door. American Journal of Psychiatry, 166, 103-109. doi:10.1176/appi.ajp.2008.08030416

Baillargeon, J., Penn, J. V., Knight, K., Harzke, A. J., Baillargeon, G., & Becker, E. A. (2010). Risk of reincarceration among prisoners with co-occurring severe mental illness and sub- stance use disorders. Administration and Policy in Mental Health and Mental Health Services Research, 37, 367-374. doi:10.1007/s10488-009-0252-9

Baillargeon, J., Williams, B. A., Mellow, J., Harzke, A. J., Hoge, S. K., Baillargeon, G., & Greifinger, R. B. (2009). Parole revocation among prison inmates with psychiatric and sub- stance use disorders. Psychiatric Services, 60, 1516-1521. doi:10.1176/appi.ps.60.11.1516

Basco, M. R., Bostic, J. Q., Davies, D., Rush, A. J., Witte, B., Hendrickse, W., & Barnett, V. (2000). Methods to improve diagnostic accuracy in a community mental health setting. American Journal of Psychiatry, 157, 1599-1605.

Becker, M. A., Andel, R., Boaz, T., & Constantine, R. (2011). Gender differences and risk of arrest among offenders with serious mental illness. Journal of Behavioral Health Services & Research, 38, 16-27. doi:10.1007/s11414-010-9217-8

Blank Wilson, A., Draine, J., Hadley, T., & Metraux, S. (2011). The role of mental illness and substance abuse in explaining jail recidivism. International Journal of Law and Psychiatry, 34, 264-268. doi:10.1016/j.ijlp.2011.07.004

Campbell, K. M., Deck, D., & Krupski, A. (2008). Record linkage software in the public domain: A comparison of Link Plus, the Link King, and a “basic” deterministic algorithm. Health Informatics Journal, 14, 5-15. doi:10.1177/1460458208088855

Cuddeback, G. S., Scheyett, A., Pettus-Davis, C., & Morrissey, J. P. (2010). General medical problems of incarcerated persons with severe and persistent mental illness: A population- based study. Psychiatric Services, 61, 45-49. doi:10.1176/appi.ps.61.1.45

Denzin, N. K. (1989). The research act (3rd ed.). Englewood Cliffs, NJ: Prentice Hall. Derogatis, L. R., & Unger, R. (2010). Symptom checklist-90-revised. In I. B. Weiner & W. E.

Craighead (Eds), Corsini encyclopedia of psychology (pp. 1743-1744). Hoboken, NJ: John Wiley & Sons. doi:10.1002/9780470479216.corpsy0970

Ditton, P. M. (1999). Mental health and treatment of inmates and probationers (Special Report NCJ 174463). Bureau of Justice Statistics, U.S. Department of Justice. Retrieved from

936 International Journal of Offender Therapy and Comparative Criminology 61(8)

https://www.prisonlegalnews.org/media/publications/bojs_mental_health_and_treatment_ of_inmates_and_probationers_1999.pdf

Draine, J., Blank Wilson, A., Metraux, S., Hadley, T., & Evans, A. (2010). The impact of mental illness status on speed and means of jail release. Psychiatric Services, 61, 458-462. doi:10.1176/appi.ps.61.5.458

Draine, J., Wilson, A. B., & Pogorzelski, W. (2007). Limitations and potential in current research on services for people with mental illness in the criminal justice system. Journal of Offender Rehabilitation, 45(3-4), 159-177. doi:10.1300/J076v45n03_07

Engel, R., & Silver, E. (2001). Policing mentally disordered suspects: A re-examination of the crim- inalization hypothesis. Criminology, 39, 225-252. doi:10.1111/j.1745-9125.2001.tb00922.x

Fennig, S., Craig, T., Lavelle, J., Kovasznay, B., & Bromet, E. J. (1994). Best-estimate ver- sus structured interview-based diagnosis in first-admission psychosis. Comprehensive Psychiatry, 35, 341-348. doi:10.1016/0010-440X(94)90273-9

First, M. B., Gibbon, M., Spitzer, R. L., & Williams, J. B. (2001). User’s guide for the Structured Clinical Interview for DSM-IV-TR Axis I disorders: SCID-I, research version. New York: Biometric Research Department, New York State Psychiatric Institute.

Fry, A. J., O’Riordan, D. P., & Geanellos, R. (2002). Social control agents or front-line car- ers for people with mental health problems: Police and mental health services in Sydney, Australia. Health & Social Care in the Community, 10, 277-286.

Health Insurance Portability and Accountability Act of 1996, 42 U.S.C. § 1320d-9 (2010). James, D. J., & Glaze, L. E. (2006). Mental health problems of prison and jail inmates (NCJ

213600). Washington, DC: Bureau of Justice Statistics. Retrieved from http://www.bjs. gov/content/pub/pdf/mhppji.pdf

Lecrubier, Y., Sheehan, D., Weiller, E., Amorim, P., Bonora, I., Harnett Sheehan, K., & Dunbar, G. (1997). The Mini International Neuropsychiatric Interview (MINI). A short diagnostic structured interview: Reliability and validity according to the CIDI. European Psychiatry, 12, 224-231. doi:10.1016/S0924-9338(97)83296-8

Lobbestael, J., Leurgans, M., & Arntz, A. (2010). Inter-rater reliability of the Structured Clinical Interview for DSM-IV Axis I Disorders (SCID I) and Axis II Disorders (SCID II). Clinical Psychology & Psychotherapy, 18, 75-79. doi:10.1002/cpp.693

McCabe, P. J., Christopher, P. P., Druhn, N., Roy-Bujnowski, K., Grudzinskas, A. J., & Fisher, W. H. (2012). Arrest types and co-occurring disorders in persons with schizophrenia or related psychoses. Journal of Behavioral Health Services & Research, 39, 271-284. doi:10.1007/s11414-011-9269-4

Morrissey, J., Cuddeback, G., Cuellar, A. E., & Steadman, H. (2007). The role of Medicaid enrollment and outpatient service use in jail recidivism among persons with severe mental illness. Psychiatric Services, 58, 794-801. doi:10.1176/appi.ps.58.6.794

Morrissey, J., Steadman, H., Dalton, K., Cuellar, A., Stiles, P., & Cuddeback, G. (2006). Medicaid enrollment and mental health service use following release of jail detainees with severe mental illness. Psychiatric Services, 57, 809-815. doi:10.1176/appi.ps.57.6.809

National Institutes of Health. (n.d.). NIH big data to knowledge (BD2K). Retrieved from http:// bd2k.nih.gov/index.html#sthash.gLsy3goc.dpbs

Nordgaard, J., Revsbech, R., Saebye, D., & Parnas, J. (2012). Assessing the diagnostic validity of a structured psychiatric interview in a first-admission hospital sample. World Psychiatry, 11, 181-185. Retrieved from http://www.ncbi.nlm.nih.gov/pmc/articles/PMC3449355/

Novak, K., & Engel, R. (2005). Disentangling the influence of suspects’ demeanor and mental disorder on arrest. Policing: An International Journal of Police Strategies & Management, 28, 493-512. doi:10.1108/13639510510614573

Morabito and Wilson 937

Parks, R., Mastrofski, S., & DeJong, C. (1999). How officers spend their time with the com- munity. Justice Quarterly, 16, 483-518. doi:10.1080/07418829900094241

Patient Protection and Affordable Care Act, 42 U.S.C. § 18001 (2010). Robins, L. N., Helzer, J. E., Croughan, J., & Ratcliff, K. S. (1981). National Institute of Mental

Health Diagnostic Interview Schedule: Its history, characteristics, and validity. Archives of General Psychiatry, 38, 381-389. doi:10.1001/archpsyc.1981.01780290015001

Roesch, R., Ogloff, J. R. P., & Eaves, D. (1995). Mental health research in the criminal justice system: The need for common approaches and international perspectives. International Journal of Law and Psychiatry, 18, 1-14. doi:10.1016/0160-2527(94)00023-9

Rogers, R., Sewell, K. W., Ustad, K., Reinhardt, V., & Edwards, W. (1995). The Referral Decision Scale with mentally disordered inmates: A preliminary study of convergent and discriminant validity. Law and Human Behavior, 19, 481-492. doi:10.1007/BF01499339

Rothbard, A. B., Kuno, E., Hadley, T. R., & Dogin, J. (2004). Psychiatric service utilization and cost for persons with schizophrenia in a Medicaid managed care program. Journal of Behavioral Health Services & Research, 31, 1-12. doi:10.1007/BF02287334

Sheehan, D. V., Lecrubier, Y., Sheehan, K. H., Amorim, P., Janavs, J., Weiller, E., . . . Dunbar, G. C. (1998). The Mini-International Neuropsychiatric Interview (M.I.N.I.): The develop- ment and validation of a structured diagnostic psychiatric interview for DSM-IV and ICD- 10. Journal of Clinical Psychiatry, 59(Suppl. 20), 22-33; quiz 34-57.

Steadman, H., Osher, F., Robbins, P. C., Case, B., & Samuels, S. (2009). Prevalence of serious men- tal illness among jail inmates. Psychiatric Services, 60, 761-765. doi:10.1176/appi.ps.60.6.761

Teplin, L. A. (1983). The criminalization of the mentally ill: Speculation in search of data. Psychological Bulletin, 94, 54-67. doi:10.1037/0033-2909.94.1.54

Teplin, L. A. (1984). Criminalizing mental disorder: The comparative arrest rate of the mentally ill. American Psychologist, 29, 794-803. doi:10.1037/0003-066X.39.7.794

Teplin, L. A. (1990). The prevalence of severe mental disorder among male urban jail detainees: Comparison with the Epidemiologic Catchment Area program. American Journal of Public Health, 80, 663-669. doi:10.2105/AJPH.80.6.663

Teplin, L. A., Abram, K. M., & McClelland, G. M. (1996). Prevalence of psychiatric disorders among incarcerated women. I. Pretrial jail detainees. Archives of General Psychiatry, 53, 505-512. doi:10.1001/archpsyc.1996.01830060047007

Teplin, L. A., & Swartz, J. (1989). Screening for severe mental disorder in jails: The development of the Referral Decision Scale. Law and Human Behavior, 13, 1-18. doi:10.1007/BF01056159

Wilson, A. B., Draine, J., Hadley, T., Metraux, S., & Evans, A. (2011). Examining the impact of mental illness and substance use on recidivism in a county jail. International Journal of Law and Psychiatry, 34, 264-268. doi:10.1016/j.ijlp.2011.07.004