A systematic review of offender mental health stigma: commonality, psychometric measures and differential diagnosis

ABSTRACT

Stigmatic attitudes appear to vary across different mental health diagnoses, and offenders with mental illness have been shown to elicit more negative stigmatic attitudes than offenders without mental illness. Stigma and discrimination can have detrimental effects on an individual's recovery, treatment and even employment opportunities. This systematic review aimed to report the commonality of research into stigma towards offenders with mental health conditions, to explore if different mental health diagnoses were associated with differential rates of stigma in offenders, and to ascertain which psychometric measures have been used to capture such stigmatic attitudes. Twelve studies were included in the review with varied populations and study locations. The vast majority reported negative stigmatic attitudes towards offenders with mental illness when compared to control groups, with neither a criminal history nor a mental illness. Results also indicated that the diagnoses with particularly high levels of stigma were psychopathy and schizophrenia. Psychometric measures used to capture stigma varied considerably and rarely was the same measure used across studies which limited comparisons. This review highlights a number of key points for advancing research in the area which are discussed along with strengths and limitations.

KEYWORDS:

• Stigma
• offenders
• criminal history
• mental health stigma
• negative attitudes

Stigma in mental health

Link and Phelan (2001) define stigma as ‘the convergence of interrelated components [existing] when elements of labelling, stereotyping, separation, status loss and discrimination occur together in a power situation that allows them’ (p. 377). Definitions of stigma tend to include a description of pejorative attitudes and beliefs which can lead to discrimination towards others. People with mental health conditions may be a particular target of such stigma e.g. ‘public stigma’ or, indeed, may experience ‘self stigma’, a differentiation made explicitly within the ‘Mental Illness Stigma Framework’ by Fox et al. (2018). Public stigma encompasses related terms including stereotypes, prejudice and discrimination, with each of these terms being considered important in a consecutive process by which stigma develops (Corrigan et al., 2004). Self-stigma reflects the internalisation of such attitudes by stigmatised person. The overall literature on stigma is relevant due to the damaging effects stigma has on those with mental illness, which include, but are not limited to, an increased prevalence of suicide (Schomerus et al., 2015), reduced employment and housing opportunities, and barriers to accessing healthcare and treatment (Overton & Medina, 2008). The latter points have been hypothesised to be at least partly because health professionals may share similar stigmatic attitudes as the general public (Jorm et al., 1999; Lauber et al., 2000).

Given the inherent link between self-stigma and public stigma (West et al., 2014; Wood et al., 2014), it is conceivable that efforts to reduce public stigma may also reduce self-stigma. For these reasons, various anti-stigma campaigns have aimed to reduce public stigma. For example, ‘Time to Change 2009’ (Time to Change, 2021) in England has shown some improvements in aspects of stigma such as intended behaviour and a small improvement in attitudes through education and raising awareness about mental health (Corrigan et al., 2012; Evans-Lacko et al., 2014). This provides hope that efforts to increase the public understanding of stigma may be productive.

Differential mental health diagnoses

Stigmatic attitudes appear to vary across different mental health diagnoses (Crisp et al., 2000; Parle, 2012). The most stigmatised diagnoses have frequently been found to be schizophrenia (Read et al., 2006; Wood et al., 2014) and borderline personality disorder (BPD) (Catthoor et al., 2015). There has been less research into public stigmatic attitudes towards BPD than that of schizophrenia but perceptions of frustration and fear amongst the public toward personality disorders have been found (Adebowale, 2010). Research has shown that negative public attitudes towards those with a diagnosis of schizophrenia often involve beliefs around dangerousness and unpredictability (Angermeyer & Matschinger, 2003; Crisp et al., 2000). It is thought that schizophrenia might be particularly stigmatised due to a small minority of people with this diagnosis behaving dangerously and the media exaggerating the link between schizophrenia and violence (Crisp et al., 2000). As a result, these perceptions are generalised to all individuals with the condition (Crisp et al., 2000). However, some diagnoses appear to be associated with less stigma; for instance, Wood et al. (2014) highlighted that anxiety disorders may be relatively less stigmatised. However, the literature considering different levels of stigma between diagnoses is relatively under-developed.

Stigma towards offenders

Stigma towards offenders, or those who have previously committed a crime, has also been associated with the development of wider stereotypes of dishonesty and danger (Hirschfield & Piquero, 2010). Research shows that violent behaviour may be a particular source of stigma (Hardcastle et al., 2011) and that sex offenders are amongst the most highly stigmatised subgroup of offenders (Tewksbury & Lees, 2006). Public attitudes towards offenders are more negative towards those convicted of sexual offences than other non-sex offences (Craig, 2005). Therefore, stigma derived from sex offending behaviour may occur in a somewhat different way than that derived from violent offending more broadly (Hogue, 1993; Weekes et al., 1995), and furthermore, may interact differently with other sources of stigma (e.g. mental health stigma).

Of course, one difference between offenders and people with mental health diagnoses is that offenders generally have demonstrated behaviour that may reasonably lead others in society to experience fear of harm. Arguably, however public stigmatic attitudes frequently extend beyond the actual risk of danger likely caused and may serve to paradoxically prevent an offender from exiting the circumstances or factors that maintain the offending.

Joint stigma

Given the aforementioned research, it seems highly likely that people with both mental health problems and a criminal history will experience a ‘dual stigma’. Indeed, offenders with mental illness have been shown to elicit more negative attitudes than offenders without mental illness (Rade et al., 2016). Similarly, once arrested, offenders with mental illness have been found to be held in custody for longer periods than those without (Solomon & Draine, 1995). However, the relationship between stigma that arises from a mental health condition and a criminal history is not well researched; it is unclear whether one source of stigma is more important than the other, whether the relationship between the two sources of stigma is additive or interactive, or whether both sources of stigma might be related to broader, more general attitudes towards disadvantaged people in society.

The possibility of an interaction between offending and mental health problems as sources of stigma is important to consider in the context of the relationship between offending and mental health. Whilst there is a higher likelihood of offenders experiencing mental health problems than the general community, only a minority of people with mental health conditions are violent or have a history of offending and they are more likely to be a victim than a perpetrator of violent crime (Brekke et al., 2002). Despite this, some specific mental health disorders are more strongly associated with crime, at least on a group level. For instance, having a psychotic disorder increases the prevalence for being convicted of a crime (Morgan et al., 2013), although the relationship between violence and certain types of symptoms (e.g. persecutory delusions) may be overall more important (Coid et al., 2013). Such an interaction could take on many forms, including the possibility that one source of stigma may mitigate against the other. This could be possible if, for instance, mental health problems were seen as a less ‘personal’ explanation for offending, and perhaps more amenable to change through treatment (Morgan et al., 2013).

Yet it is hard to answer questions about the relationship between these two sources of stigma without answers to more basic research questions, for instance, how common is the experience of dual stigma, and how might it be best assessed. West et al. (2014) and Rade et al. (2016) have commented on the sheer lack of research into the stigmatisation of forensic psychiatric groups and the focus of stigma research being on single sources of stigma. A comprehensive and systematic review of the literature that has examined such dual stigma therefore appears an important step towards improving the state of the current research. Building such a body of research may be of particular relevance to the patients of forensic psychiatric services (West et al., 2014), and to those in prison, where the rates of mental health problems are high (Diamond et al., 2001).

A particular issue which requires consideration through such a review is the best way to assess dual stigma. There is no shortage of mental health stigma measures. In a critical review of mental illness stigma measures, over 400 were identified, a situation that has been described as ‘overwhelming’ (Fox et al., 2018). However, many of these measures as highlighted by Fox et al. (2018) did not have adequate psychometric evaluation and all are specific to mental health stigma. Furthermore, specific measures of stigma towards offenders appear to be less common. Previous papers and reviews have not focused specifically on measures of offender mental health stigma, so research which has considered this area may have utilised measures adapted from mental health stigma. Yet, significant adaptation may be necessary (e.g. to content of vignettes as well as questionnaire items) to develop suitably valid and meaningful measures, and it is unclear, overall, which approaches to measurement of stigmatic attitudes in this population have the most empirical basis. Understanding the current literature would provide valuable information about whether these measures already exist or if there is a need for further development of specific measures.

Aims

To the author's knowledge, there has not been a systematic review of the literature surrounding offenders with mental health problems from the perspective of the stigmatiser. Therefore, the current systematic review aimed to understand how common stigma towards offenders with mental health conditions was (research question one); ascertain which measures have been used to capture such stigma (research question two) and determine if different mental health diagnoses were associated with differential rates of stigma in offenders (research question three).

Method

This systematic review was registered on the International Register of Prospective Systematic Reviews (PROSPERO) (registration number: CRD42020191145, 17/09/20).

Eligibility

Inclusion criteria

The current paper sought empirical research which met identified criteria that aligned with the aims of the systematic review. All criteria had to be met to be included. The inclusion criteria were:

(a) empirical research studies which developed or applied a measure of stigma adopting a quantitative stigma score (studies which adopted tools measuring stigma without a quantitative aspect were excluded). The measure had to be stigma ‘of another’, i.e. studies that considered ‘self-stigma’ were excluded.

Studies were required to have considered stigma in relation to offenders with mental health difficulties:

(b) the term ‘offender’ did not need to be specifically mentioned but could be implied through phrases such as ‘history of a criminal conviction’, ‘residing in a forensic psychiatric hospital or prison’, ‘contact with the criminal justice system’; (c) a phrase such as ‘mental health condition’ or ‘mental health difficulty’ could be explicitly used or a specific mental health diagnosis such as schizophrenia or depression was also considered sufficient; (d) studies were required to measure stigma of offenders with mental health conditions and therefore the inclusion of both offender and mental health dimensions was required within the study. This could have been evidenced through a vignette including information about criminal history and the use of a mental health stigma questionnaire or a specific questionnaire investigating offender mental health stigma, for example; (e) stigma as a concept was considered to include broader negative attitudes and stereotypes but was required to be multifaceted (with more than one facet of stigma) to be included. This was due to the vast range of constructs described in the literature as defining stigma and the commonality amongst them was a multifaceted approach. The division between them was that they did not agree on the same facets to define stigma. For example, studies that focused on one facet of stigma, such as sympathy or dangerousness were not included. However, studies which included a broader measure of stigma and a measure of dangerousness were included where data were only extracted from the broader measure. Data derived from the additional single faceted measures were excluded from the current review; (f) articles must have been published in peer-reviewed journals only; (g) articles must have been written in the English language; (h) articles must have been published after January 2009 and (i) participants included in the research studies must have been aged 18 years or over.

Exclusion criteria

In addition to the inclusion criteria, studies were excluded under additional specific circumstances: (a) addiction in relation to drug or alcohol use was not considered a primary or secondary mental health condition; (b) current or historical sexual offence or offence related to sexual abuse (this was to avoid including stigma that was specific to this form of offending, which may operate differently than stigma derived from violent offending more generally [Hogue, 1993; Weekes et al., 1995]); (c) qualitative measures of stigma or negative attitudes including individual experiences of stigma such as self-stigma or anticipated stigma; (d) any description of a learning disability, brain injury, dementia, cognitive impairment or neurodevelopmental condition in any given vignettes; (e) research involving ‘exonerees’ defined as individuals who have previously been wrongfully convicted. These criteria were to ensure that stigma solely in relation to offenders with psychiatric diagnoses was considered as the focus of the current review.

Search strategy

The following databases were searched; MEDLINE, PsycINFO, EMBASE, PsyArticles, ProQuest criminal justice and the National Criminal Justice Reference Service (NCJRS) as they were considered significant in relation to mental health, stigma and offender research. The search terms used were (Stigma* or stereotype* or prejud* or ‘negative attitude*’ or discrim* or ‘public attitude*)’ AND (Schizo* or Psycho* or ‘personality disorder*’ or depress* or bipolar or ‘mood disorder’ or ‘mental health’ or ‘mental illness)’ AND (Offend* or forensic or prison* or probation or ‘secure unit’ or crim* or justice). The NCJRS did not have capacity for searching articles using ‘OR’ terms and was therefore searched using the least restrictive option using broad terms and the results were manually searched by the primary author. For the remaining journals, the abstract and title searches were carried out with a date limitation of the start of 2009–July 2020. Reference lists were also checked for key research articles; however, this did not yield any further studies that had not already been identified within the main searches. Searches were conducted on the 3rd August 2020.

Identification and selection of studies

To identify and select studies relevant to the systematic review questions, the search strategy outlined above was employed. The titles and abstracts of the search results were screened by the primary author against the inclusion and exclusion criteria previously described. Duplicate articles across journals were also removed, see Figure 1 for further detail. All data extraction was completed by the primary author. The final studies were checked against eligibility criteria by a fellow named author in order to reduce bias. Both authors agreed that all of the selected studies met the eligibility criteria.

Figure 1. PRISMA study selection flowchart.

Studies included in the review

Figure 1 shows a PRISMA flow chart describing the overall process of study selection. This began with initial screening where 3196 studies were identified, and a further 99 from the NCJRS. Following the removal of duplicates, this resulted in 1584 studies to be screened along with an additional 98 from the NCJRS as due to its setup it was not possible to remove duplicates digitally. Following the screening of titles of abstracts, 157 full-text articles were screened against eligibility criteria resulting in 12 eligible studies.

Data extraction

Data was extracted in three parts, broadly following each research question. The first detailed the demographics of the study including the sample, research aims, findings in relation to stigma and study location. The second detailed the measures used in each study, relevant psychometrics and mean stigma scores (total and subscales). In order to understand if there was a presence of stigma towards offenders with mental health conditions, stigma scores were compared with that of control groups (where neither an offending history nor mental health condition was present). Where studies were applying a previously validated measure (as opposed to developing a novel measure), comparisons were made with control groups (when provided or where possible) using t tests to understand if differences between the means were statistically significant. In order to ensure a consistent approach to the identification of an appropriate comparison sample, and to use a sample that was most comparable to the identified sample, a brief protocol was employed (see Figure 2).

Figure 2. Flowchart describing the selection of comparative research studies.

Methodological quality assessment

Study quality assessment was completed using the Appraisal Tool for Cross-sectional studies (AXIS tool) for quantitative research (Downes et al., 2016). The AXIS tool consists of 20 questions to critically appraise observational research studies; examples include sample size justification, a clearly defined target population and statistical methods to allow for replication. The AXIS tool does not have a numeric scale or a final score. Instead, it asks for the presence or absence of each quality area. However, previous research employing this tool has reported how many of the 20 criteria were met, giving a score out of 20 (e.g. Wong et al., 2018) and therefore this was replicated in the current review (see Table 1).

Table 1. Overview of the final studies selected for the current review.

Study ID	Authors of study	Sample size gender split and mean age (SD)	Participant type/job role	Study location	Research aims/questions	Summarise main findings in regard to stigma	Quality of study AXIS criteria /20
1.	Nee and Witt (2013)	243 (total) 70% female 30% male 35 (13.18) yrs	General Public	UK	This study predicted: ‘(i) those with mental health problems would be seen as more likely to commit crime; (ii) participants’ own familiarity with mental health problems and/or criminal behaviour would result in a less negative, stereotypical response towards individuals with mental health problems. (iii) increased participant age will result in a less negative, stereotypical response to individuals with mental health problems.’	‘The sample were significantly more likely to think that a character would ‘possibly’ commit future crime if he had mental illness in comparison to the control, but crimes were expected to be minor’	20
2.	Garcia et al. (2020)	290 (total) 53% female 47% male <1% prefer not to say 37.31(11.52)yrs	General public	United states	‘To understand public perceptions of the relationship between mental illness, perceived criminality and race. Increase understanding of stigmatization of mentally ill. Hypothesis: Vignettes depicting mental illness would be associated with higher levels of criminality; familiarity with mental illness or criminality would be associated with lower levels of stereotypical beliefs.’	‘The presence of a mental health diagnostic label elicited higher levels of risk of criminality. The public appear to view individuals with mental illness as being more dangerous, it was not specific to mental health diagnosis.’	19
3.	Rao et al. (2009)	108 (total) 86% female 14% male 43.2 (1.2) yrs	Health professionals 58% qualified nurses 13% healthcare assistants 9% did not state profession 20% doctors	South East, England, UK	‘Aimed to assess stigmatized attitudes among health professionals. Research Questions: 1. Do health professionals have more stigmatizing attitudes towards schizophrenia than brief psychotic episodes? 2. Do health professionals have more stigmatizing attitudes towards patients admitted to a secure hospital than somebody who has been diagnosed with schizophrenia alone?’	‘Participants had highly stigmatized attitudes towards patients from a forensic hospital. This suggested that health professionals have stigmatized attitudes towards an illness such as schizophrenia and this is worse towards patients from a secure hospital.’	19
4.	Sowislo et al. (2017)	2207 (total) 61.5% female 38.5% male 43.4 (13.4) yrs	General Public	Basel-Stadt, Switzerland	Compared stigma in relation to psychiatric symptoms, to that related to the type of psychiatric service use. ‘Compared stigma around BPD with schizophrenia and alcohol dependence. Understood differences in stigma between different psychiatric inpatient services such as forensic settings.’	‘Desired social distance was significantly lower in relation to psychiatric service use than to psychiatric symptoms. Overall, symptoms of alcohol dependency, behaviour endangering others, and the fictitious character's being male tend to increase stigmatization. The character being hospitalized in a psychiatric unit at a general hospital and also respondent and familiarity with psychiatric services tend to decrease stigmatization’	18
5.	Adjorlolo et al. (2018)	113 65% female 35% male 75% <30 yrs 35% >30 yrs	Qualified Mental Health Nurses	Ghana	‘This study investigated mental health nurses’ attitudes toward mental illness, as well as punishment-oriented attitudes (i.e. conviction proneness and punitiveness) as predictors of their attitudes toward offenders with mental illness. Second, the study examines whether mental health nurses’ demographic backgrounds, namely, gender, age, and years of practice, have significant influence on their attitudes toward offenders with mental illness.’	‘The nurses’ scores in conviction proneness and criminal blameworthiness significantly predict negative attitudes toward the offenders even after controlling for their attitudes toward mental illness.’	18
6.	Lammie et al. (2010)	58 50% 29 female 41.4% 24 male 8.6% 5 prefer not to say <21 yrs = 0 21–30 yrs = 19 31–40 yrs = 16 41–50 yrs = 15 51+yrs = 4 Unknown = 4	Nursing disciplines in forensic wards	UK	‘To examine practitioner attitudes towards patients within forensic mental health care; to identify whether qualitative and quantitative approaches provide different insights. that participants who work within the medium secure unit would rate the fictitious patient less favourably than those who work within the low secure unit, due to their contact with patients who are considered to require a higher level of security. Stigma hypotheses: That male nurses, across both sites, would rate the fictitious patient higher on all factors, with the exception of fear and danger, than female nurses. That older participants would be less fearful of the fictitious patient and rate him lower on the factor of dangerousness than younger participants.’	‘Significant minority of negative attitudes in relation to desire for social distance. Quantitative results showed high stigma scores for avoidance and segregation. There were no significant differences in attitudes between medium and low secure settings. However overall, males reported more negative attitudes in relation to blame and avoidance and younger participants demonstrated more negative attitudes than older participants in relation to fear and danger. While fear and blame were low overall, males reported more blame and younger practitioners reported more fear. This may indicate that experience (contact) reduces stigma in forensic settings although we cannot assume this from the study.’	20
7.	Frailing and Slate (2016)	196 (total) 55% female 45% male 119 Florida 60% female 40% male 82% 18–25 yrs	Students	Southern Texas & Florida	‘This research reported on the measurement of criminal justice and criminology students’ attitudes towards people with mental illness, before and after a class on criminalisation of mental health in offenders.’	The ‘results indicate that criminal justice and criminology students’ attitudes toward people with mental illness, offenders with mental illness, and community-based mental health services were significantly more positive at the conclusion of our classes than they were at the beginning. These outcomes were unaffected by choice of instrument or research location.’	18
8.	Perkins et al. (2009)	404 (total) 67% female 33% male 52.18 (16.08) yrs	General public	Indiana	‘1.An adult male with schizophrenia who is actively engaged in competitive, wage-based community employment will elicit less social distance than one who is unemployed. 2. An adult male with schizophrenia who has a past history of misdemeanor criminal conduct will elicit less social distance than one with a past history of felony criminal conduct.’	‘The individual who was gainfully employed (vs. unemployed), or who had a prior misdemeanour (versus felony) criminal offense, elicited significantly less stigma. Employment may destigmatize a person coping with both psychiatric disability and a criminal record.’	19
9.	Batastini et al. (2014)	465 (total) 66.7% female 33.3% male Not reported 18–24 yrs – 84% 25–34 yrs – 12.1% 35–44 yrs – 2.2% 45–54 yrs – 1.8%	University psychology students	West Texas	Hypothesised that regardless of education or experience, the job applicant with a history of both mental illness and criminal involvement will be rated the least desirable candidate for the job. They hypothesized that ‘prior experience with a mentally ill or criminal justice involved person would be associated with less stigmatized attitudes toward the respective job applicant.’	‘Applicants with a history of both mental illness and criminal behaviour were perceived as the least acceptable candidates for employment. However, this finding did not hold true when participants (i.e. the hypothetical employers) were exposed to a brief explanation about the benefits of employment.’	18
10.	Durand et al. (2017)	116 50.9% female 49.1% male 26.8 (10.77) yrs	General public	International but most common Europe, North America & Asia	The study ‘hypothesized a negative relationship between high expression of psychopathic traits and stigmatization towards psychopaths, and also hypothesized that this negative relationship would be strongest within interpersonal-affective features due to their association with fearlessness.’	‘The presence of psychopathic traits, particularly those related to boldness, was negatively correlated with the degree of stigmatizing behaviours towards psychopaths.’	18
11.	Weaver et al. (2019)	358 77% female 23% male 28.49 (9.02) yrs	College students majoring in social work (35%) or criminal justice (65%)	University of Southern Mississippi, US	‘This study investigates attitudes toward offenders living with mental illness among a cross-section of college students.’	‘Results indicated that Social Work students were less likely to have negative stereotypes (than criminal justice students) toward offenders with mental illness and tended to be more supportive of their potential for rehabilitation. The two groups of students appeared to share ambivalence regarding the dangerousness and culpability of offenders living with mental illness.’	19
12	Batastini et al. (2017)	138 29.7% female 70.3% male 49.43 (12.84) yrs	23.1% Judges, 24.3% Prosecutors, 52.7% Public Defendants	Mississippi	The ‘primary purpose of this study was to identify the prevalence of stigmatizing beliefs among judges, prosecutors, and public defenders. It was hypothesized that defence attorneys would self-report significantly less biased and stigmatizing attitudes about mental illness in general and seriously mentally ill defendants than both judges and prosecutors.’	‘Public defenders, relative to both judges and prosecutors, endorsed more compassionate attitudes about defendants with mental illnesses. While judges and prosecutors endorsed more negative stereotypes about mental illness and perceived mentally ill defendants as a greater risk to the community, mean scores across groups suggested moderately positive attitudes overall.’	19

Analysis

The data were analysed using a narrative synthesis model to describe the literature at present regarding offender mental health stigma, to understand the measures used to capture this information and to suggest future research ideas. Where possible the impact of differential mental health diagnoses on levels of stigma in offenders was also considered.

Results

Twelve studies were identified as eligible and therefore included in the final dataset for the current systematic review.

Study characteristics

Population samples in the studies varied and included the general public (n = 5), university students (n = 4), healthcare professionals (n = 3) and one study included legal professionals (judges, prosecutors and public defendants). Study locations included the United States of America (USA) (n = 5), the United Kingdom (n = 3), India (n = 1), Ghana (n = 1), Switzerland (n = 1) and an international study including participants from across Europe, Asia and the USA. The majority of studies selected mental health stigma questionnaires and the use of a vignette to specify a criminal offence or background. Two studies employed a specific offender mental health stigma questionnaire called Attitudes Towards Mentally Ill Offenders (ATIMO) (Church et al., 2009).

Participant characteristics

The study sample sizes ranged from 58 to 2207 (N = 4696). Females were over-represented in the review (see Table 1), 11 out of 12 studies had more than 50% female participants with the exception of a single study conducted with legal professionals (Batastini et al., 2017) in which the majority (70.30%) were male. Where reported (n = 9), the mean age ranged from 21.65 years (SD = 2.60) to 52.18 years (SD = 16.08).

Quality assessment

The selected studies scored highly against the AXIS criteria (range = 18–20) with two studies scoring the full 20 out of 20. The majority of studies lacked justification of sample size, such as the use of a power analysis or lacked a statement around the size chosen for the study. Another criterion often unmet was a description of measures taken to categorise non-responders from study samples. Table 1 gives an overview of all of the studies included in the review.

Research question 1. How common is stigma towards offenders with mental health conditions?

To understand the specific stigma deriving from offending and mental health conditions, ideally stigma scores from vignettes describing offenders with mental health conditions would be compared with vignettes describing non-offenders with or without a mental health condition. Unfortunately, rarely were many of the studies set up in this way. However, two studies compared offenders with and without mental health difficulties, producing similar results. In both Garcia et al. (2020) and Nee and Witt (2013), those without mental health difficulties had significantly lower scores on sympathy subscales and higher stigma in comparison to those with a mental health difficulty. In Nee and Witt’s (2013) study it appeared that the offending history with and without mental health diagnosis was associated with higher levels of stigma in comparison to a control group, therefore, showing the impact of offending history on stigma levels. Alternatively, Garcia et al. (2020) found that participants judged the likelihood of a future crime as greater when a mental health diagnosis was added to a vignette containing otherwise the same offending history, suggesting that the mental health condition was associated with an increase in stigma.

In considering the question of how common stigma towards offenders with mental health conditions is from another perspective, the protocol previously outlined (Figure 2) was followed and five studies were selected. Table 2 shows comparisons between the samples obtained and control samples. The comparisons revealed that in a study amongst hospital staff, stigmatised attitudes were higher towards those admitted to a forensic hospital than those admitted to hospital with schizophrenia or a brief psychotic episode (Rao et al., 2009). Another study had similar findings where higher levels of social distance were desired when vignettes described a forensic unit in comparison to a general hospital with a psychiatric unit (Sowislo et al., 2017). A further study found significantly higher stigmatic levels on a fear/dangerousness subscale for a vignette with an offender with bipolar in comparison to a control group who had neither an offending nor a psychiatric history (Batastini et al., 2014). However, the responsibility subscale between these groups scored in the opposite direction indicating significantly more responsibility was given to the control group (Sowislo et al., 2017). The results taken together indicate a somewhat mixed message. There was no difference on a scale of willingness to help or social distance between the groups.

Table 2. Mean stigma scores of identified studies and comparison groups.

Authors of study	Name and reference of measure	Direction of score	Brief description of what measure operationalises	Mean score	Comparison/control group mean score	T test	Is offender mental health stigma significantly higher than control?
Nee and Witt (2013)	No name (5 questions)	10-point Likert Higher scores = less stigma for trust and sympathy and rehab potential. Higher scores = more stigma for likelihood and severity of future crime	Vignettes included an offender with a mental health condition of either depression/schizophrenia	Mean response % categorised by depression or schizophrenia	Control group (no mental illness no criminal background)
			Subscales:
			Trustworthiness	Depression: 65.00 (SD = 17.85, n = 44) Schizophrenia: 69.46(SD = 15.47, n = 37)	72.79 (SD = 16.52, n = 43)	Depression: t (85) = 2.11, p = 0.04* Schizophrenia: t (78) = 0.93, p = 0.36	Yes for depression
			Sympathy	Depression: 72.27 (SD = 16.82, n = 44) Schizophrenia: 74.59(SD = 23.99, n = 37)	53.02 (SD = 22.20, n = 43)	Depression: t (85) = 4.58, p < .001* Schizophrenia: t (78) = 5.17, p < .001*	Yes for depression & schizophrenia
			Likelihood future crime (scores reported are possibility to definitely commit crime)	Depression: 5.02 (SD = 2.61, n = 44) Schizophrenia: 4.81(SD = 2.95, n = 37)	2.65 (SD = 2.70, n = 43)	Depression: t (85) = 4.16, p < .001* Schizophrenia: t (78) = 3.42, p < .001*	Yes for depression & schizophrenia
			Severity of future crime (minor crime category)	Depression: 4.21 (SD = 1.86, n = 32) Schizophrenia: 4.12(SD = 1.91, n = 24)	3.38 (SD = 1.54, n = 16)	Depression: t (46) = 1.54, p = 0.13 Schizophrenia: t (38) = 1.29, p = 0.20	No
			Rehab potential	Depression: 67.8 (SD = 19.13, n = 32) Schizophrenia: 75.00(SD = 16.55, n = 26)	75.29 (SD = 20.65, n = 17)	Depression: t (47) = 1.27, p = 0.21 Schizophrenia: t (41) = 0.05, p = 0.96	No
Garcia et al. (2020)	No name (5 questions as used in Nee & Witt, 2013)	10-point Likert (1 = lower stigma/positive, 10 = higher/negative)	Perceptions: sympathy, trustworthy, future crime likelihood and severity, rehab potential. Means reported for only one subscale:	Schizophrenia condition only one reported. Other categories are grouped as ‘mental illness’
		Higher stigma for both mental illness and schizophrenia for future crime	Likelihood of future crime	Mental Illness grouped: 5.11 (SD = 2.09, estimated n = 217) Schizophrenia: 5.23 (SD = 1.97, estimated n = 72)	Control Group from Nee and Witt (2013) 2.65 (SD = 2.70, n = 43)	Mental Illness grouped: t (258) = 6.69, p < .001* Schizophrenia: t (113) = 5.90, p < .001*	Yes for mental illness grouped and for schizophrenia specifically
Rao et al. (2009)	Attitude to Mental Illness Questionnaire (AMIQ) Luty et al. (2006)	5-point Likert (max +2 min −2, neutral/don't know 0) Total score between −10 and +10 Lower scores indicate negative attitudes, higher = positive	Stigmatised attitudes	Admitted to forensic hospital – Broadmoor No subscales reported Forensic hospital: −1.2 (SD = 3.12, n = 108)	Control group- general public from Luty et al. (2006) validation study 5.86 (SD = 2.40, n = 879)	Forensic hospital: t (985) = 27.83, p < .001*	Yes
Batastini et al. (2014)	Attribution Questionnaire (AQ-27) Corrigan (2003)/ Brown (2008) subscales	9-point Likert Higher score = higher stigma (Some subscales reverse scored)	Stigma/stereotypes using Brown’s (2008) subscales (Fear/dangerousness, help/interact, responsibility, forcing treatment, empathy).	Non psychoeducation group Bipolar I disorder and theft jail sentence	Control group
			Fear/Danger	Bipolar I disorder: 22.23 (SD = 11.59, n = 56)	16.16 (SD = 9.32, n = 55)	Bipolar I disorder: t (109) = 3.04, p = .003*	Yes
			Responsibility	Bipolar I disorder: 13.25(SD = 4.80, n = 56)	16.60 (SD = 4.11, n = 55)	Bipolar I disorder: t (109) = 3.95, p < .001*	Yes
			Help/interact	Bipolar I disorder: 22.55 (SD = 9.14, n = 56)	21.16 (SD = 8.18, n = 55)	Bipolar I disorder: t (109) = 0.84, p = .40	No
	Social Distance Scale Link et al. (1987), Martin et al. (2000)	4-point Likert Higher scores = more desired social distance	Social distance	Bipolar I disorder: 13.92 (SD = 4.29, n = 56)	12.71 (SD = 3.42)	Bipolar I disorder: t (109) = 1.64, p = .10	No
Durand et al. (2017)	Attribution Questionnaire (AQ-20) Corrigan et al. (2003) / Brown (2008) Subscales	9-point Likert Higher scores = higher stigma	Stigma, originally developed to measure schizophrenia but was replaced psychopathy Split into Brown’s (2008) subscales:	Psychopathy with conviction of theft	Control Batastini et al. (2014)
			Fear/dangerousness	Psychopathy: 22.48 (SD = 12.00, n = 116)	16.16 (SD = 9.32, n = 55)	Psychopathy: t (169) = 3.44, p < .001*	Yes
			Help/interact	Psychopathy: 37.21 (SD = 8.70, n = 116)	21.16 (SD = 8.18, n = 55)	Psychopathy: t (109) = 10.01, p < .001*	Yes

Note: Asterisk denotes a significant difference (p < .05) between groups.

Often studies found significantly more stigma for a forensic group than a control group with neither (mental health or offender) labels (Batastini et al., 2014; Durand et al., 2017; Rao et al., 2009). The specific subscales found to have a greater stigma towards offenders with mental health difficulties in comparison to control groups were ‘fear/danger’, ‘responsibility’ (Batastini et al., 2014; Durand et al., 2017) likelihood of future crime (Nee & Witt, 2013; Garcia et al., 2020) and ‘Trust’ (Nee & Witt, 2013). Subscales showing little difference between the two groups were ‘social distance and perceived dangerousness’, ‘willingness to help’ (Batastini et al., 2014) and ‘rehabilitation potential’ (Nee & Witt, 2013). An exception was a study where offenders were diagnosed with psychopathy; this induced significantly higher levels of fear/dangerousness when compared to a control group (Durand et al., 2017).

Research question 2. Which measures are used to capture stigma towards offenders with mental health conditions?

This review also intended to understand which measures have been used to capture offender mental health stigma in the literature. An overview of all the stigma measures used in the included studies can be found in Table 3. Out of a total of 12 studies, only a maximum of three used the same measure which was the Attribution Questionnaire (AQ, Brown, 2008; Corrigan et al., 2003) and the Community Attitudes to Mental Illness (CAMI, Högberg et al., 2008; Taylor & Dear, 1981) and both included different versions. Most measures were only adopted by a single study. Table 4 shows each measure selected in the current review and associated psychometrics. Many of the measures shown in Table 4 were found to be self-report and validated. The Cronbach's alpha for each measure is reported in Table 4 and ranged from 0.53 to 0.96. Only three measures had subscales with Cronbach's alpha in the 0.50 range, the remaining measures were above acceptable levels of reliability (>0.60).

Table 3. Quick reference list of stigma measures.

Measure	ADJORLOLO 2018	BATASTINI 2014	BATASTINI 2017	DURAND 2017	FRAILING 2016	GARCIA 2020	LAMMIE 2010	NEE 2013	PERKINS 2009	RAO 2009	SOWISLO 2017	WEAVER 2019
Attitudes toward mentally ill offenders (ATIMO, Brannen et al., 2004) Adapted versions of: ATMIO (Church et al., 2009)
Attitude to mental illness questionnaire (AMIQ) Luty et al. (2006)
Attitudes and beliefs about psychopathy (ABP) Smith et al. (2014)
Attribution questionnaire (AQ-27; Corrigan et al., 2003) Attribution questionnaire (Brown, 2008)
Community attitudes toward mental illness (CAMI) (Taylor & Dear, 1981; Swedish version) CAMI adapted 20 item version (Högberg et al., 2008)
Bogardus social distance scale (Bogardus, 1925)
Modification of Self-stigma of mental illness scale (Corrigan et al., 2006) Stereotype subscale only
Social distance scale (Link et al., 1987)
Survey of attitudes (Steadman & Cocozza, 1977)
5 questions by Nee and Witt (2013)

Table 4. Psychometrics for each measure included in the review.

Authors of study selected for review	Name and reference of measure	Psychometric of the measure α = Cronbach's alpha (Validation study)	Mental health condition referred to	Vignette
Nee and Witt (2013)	No name (5 questions)	Not reported	Depression schizophrenia	Yes
Garcia et al. (2020)	No Name (5 questions, as used in Nee & Witt, 2013)	Not reported	Grouped as ‘mental illness’ (referring to depression & schizophrenia)	Yes
Rao et al. (2009)	Attitude to Mental Illness Questionnaire (AMIQ) Luty et al. (2006)	α = 0.933 (Luty et al., 2006)	Admitted to forensic hospital (Broadmoor)	Yes
Sowislo et al. (2017)	Modification of the Bogardus social distance scale (Bogardus, 1925)	α = 0.92 (von dem Knesebeck et al., 2013)	Psychiatric hospital with forensic unit (borderline personality disorder and acute psychosis)	Yes
Adjorlolo et al. (2018)	Community attitude toward mental illness (CAMI) Högberg et al. (2008)	Open-mindedness α = 0.77 Fear/avoidance α = 0.81 Community mental health α = 0.67 Total α = 0.79	Schizophrenia	Yes
Lammie et al. (2010)	Attribution Questionnaire- (AQ-27) Corrigan et al. (2003)	α = .70–96 Corrigan et al. (2003)	Schizophrenia	Yes
Weaver et al. (2019)	ATMIO Brannen et al. (2004)	Negative Stereotypes α = .86 Rehabilitation/Compassion α = .70 Community Risk α = .61 Diminished Responsibility α = .56 (Church et al., 2009).	Mentally ill	No
Frailing and Slate (2016)	Survey of attitudes Steadman and Cocozza (1977)	α = .63 to .82 (Steadman & Cocozza, 1977)	Mental illness	No
	Community Attitudes towards Mentally Ill (CAMI) adapted version Taylor and Dear (1981)	α = .86 (Thompson et al., 2014).	Mental illness	No
Perkins et al. (2009)	Social Distance questions (Link et al., 1987)	α  = 0.87 (Perkins et al., 2009)	Schizophrenia	Yes
Batastini et al. (2014)	Attribution Questionnaire (AQ-27) (Brown, 2008) subscales	Fear/dangerousness α = 0.93 Help/interact α = 0.82 Responsibility α = 0.60 (Brown, 2008)	Bipolar I disorder	No
	Social Distance Scale Link et al. (1987); Martin et al. (2000)	α = 0.87 (for 6 item version, Martin et al., 2000)	Bipolar I disorder	No
Durand et al. (2017)	Attribution Questionnaire, (AQ-20) Corrigan et al. (2003) / Brown (2008) Subscales	α = 0.53–0.93 (Durand et al., 2017)	Psychopathy	Yes
	Attitudes and Beliefs about Psychopathy (ABP) Smith et al. (2014)	α = 0.50–0.86 (9 subscales, Durand et al., 2017)	Psychopathy	No
Batastini et al. (2017)	Adapted version of: Attitudes Toward Mentally Ill Offenders (ATMIO) Church et al. (2009); Brannen et al. (2004)	α = .73 to .88 (Church et al., 2009).	Mentally ill	No
	Community Attitudes towards Mentally Ill (CAMI) adapted version Taylor and Dear (1981)	α = .86 (Total score, Thompson et al., 2014).	Mentally ill	No

It appeared that general mental health stigma measures were frequently used in conjunction with a vignette which depicted someone with a mental health problem and a criminal conviction in order to understand forensic stigma (see Table 4). This was the case for all but two studies (Batastini et al., 2017; Weaver et al., 2019), where a measure specifically designed to measure stigma in offenders called ATIMO was developed by Brannen et al. (2004) was used. This speaks to a debate by Fox et al. (2018) about the frequent use of different measures in the stigma literature and outlines the significantly high number of stigma measures.

Research question 3. Are different mental health diagnoses associated with differential rates of stigma in offenders?

The most commonly specified mental health diagnosis used across all of the studies was schizophrenia (n = 5), followed by the generic descriptor ‘mental illness’ (n = 4). Other examples less often used included, depression (n = 2), ‘forensic hospital patient’ (n = 2), bipolar disorder (n = 1) and psychopathy (n = 1). It should be noted that on some occasions multiple diagnostic labels were included in one research paper.

Due to the lack of consistent use of stigma measures, comparing results across studies with different diagnostic labels was not possible for most of the selected studies. However, two of the studies did investigate differential diagnoses as part of their research question and therefore will be considered in more detail here. The first was Nee and Witt (2013) who compared the impact of changing the mental health condition from schizophrenia to depression. The results found that stigma scores were significantly higher on a scale of ‘likelihood to commit a future crime’ for vignettes that included mental health diagnoses in comparison to a control group (Nee & Witt, 2013). Sympathy levels were high for both schizophrenia and depression, and significantly higher than the control group (with no mental illness or criminal background). Neither of the two diagnostic categories induced significantly different scores from one another on most questions indicating that the diagnoses type did not, in isolation, induce stigmatised views (Nee & Witt, 2013). The exception was ‘rehabilitation potential’ where participants felt offenders with schizophrenia had higher potential for rehabilitation than those with depression, however, no difference was found in comparison to a control group. When comparisons were made against someone with a past criminal conviction and no mental health condition, the only significant difference in stigma scores was on the sympathy subscale, where the presence of a mental health label appeared to receive higher levels of sympathy than someone without a diagnostic label (Nee & Witt, 2013). A similar finding of higher sympathy for those with schizophrenia was found using the same stigma questions as Nee and Witt (2013) by Garcia et al. (2020). Schizophrenia in addition to an offending history, was found to have higher levels of future crime in comparison to a control group and to someone with the same offending history, showing the impact of this particular diagnostic label (Garcia et al., 2020).

Discussion

The current systematic review aimed to summarise the research within the literature in offender mental health stigma and consider whether different mental health diagnoses were associated with differential rates of stigma towards offenders. The review also set out to understand which measures had been used in the literature to capture such stigmatic attitudes towards this population. As hypothesised the review highlighted that the combination of mental health and offending increased negative stigmatic attitudes. It also highlighted that there are a number of measures specific to offender mental health stigma that are available and we argue that these tools should receive further development, focus and revision as opposed to novel tools being developed. Finally, some tentative conclusions can be drawn about comparative rates of stigma between mental health disorders, but overall this question was limited by lack of research. Overall, it can be observed that the literature in stigma in relation to offenders with mental health problems is relatively under-developed; given this, much of the following narrative considers particularly how future research might develop the field most effectively.

This is the first systematic review to approach the stigma of offenders with mental health difficulties, from the perspective of the stigmatiser. The studies selected were from a wide variety of countries across the world such as Ghana, the United States and the United Kingdom. The eligible studies included in this review suggested there was evidence of stigma towards offenders with mental health conditions. Moreover, the amount of stigma towards offenders with mental health difficulties appeared to be notably higher than that towards people without mental health difficulties or a history of offending. In regards to the question of the impact of different diagnostic terms, it is noted that most studies adopted a general term such as ‘mental illness’ rather than specific diagnostic labels, which is an important finding given the evidence of stigma attached to specific diagnostic terms (Pescosolido et al., 1999). Those which did specify a diagnosis suggested that schizophrenia and psychopathy were more stigmatised when compared to other mental health conditions, such as depression or neutral control groups. Finally, the measures used to capture stigmatic attitudes were unfortunately inconsistent between studies. Infrequently was the same measure used in more than one study (Table 3). Due to this variance, only limited comparisons across research studies were possible.

The findings speak to the presence of a possible ‘dual stigma’ towards the combined effect of an offending history and mental health difficulty. The findings from the current review echo those from reviews which have considered stigma in relation to mental health conditions (Parcesepe & Cabassa, 2013; Sheehan et al., 2016) and offending (Feingold, 2021). It begs the question of whether the combination of offending and a diagnostic label induces higher stigmatic attitudes, or if the presence of one of the two factors has a dominating influence on stigma. Unfortunately, there was not enough data to explore this fully. Future research should delve deeper into a better understanding of the combination of offender and mental health stigma, and how it affects public attitudes. To do this, the same stigma measure could be applied to different contexts and settings, with results offering some agreement about psychometric factors that make up the key elements of stigmatic judgement in the studied populations. In addition, research should include specific mental health diagnoses rather than general terms, as well as different types of offending. These research topics would necessitate large-scale sampling and a range of experimental studies. Research in these areas would support measurable attitudinal change as targeted by anti-stigma intervention research. Once there is a basis of research in these areas, it would be important to understand how stigmatic attitudes could go on to affect an individual's behaviour.

Frequently, the tools used to measure offender mental health stigma were primarily mental health stigma questionnaires (as shown in Table 4), but with the addition of a vignette to specify a particular mental health condition or an offending history. A problem for the literature, highlighted by this review, and congruent with previous research, is that the field is at saturation point with around 400 different stigma measures available (Fox et al., 2018). There is very little consensus about which measures are most suitable for which types of research question, and little evidence of replication across different samples. Some of the selected papers used specific offender mental health stigma measures, which did not rely on the use of a vignette or adaptation. The most frequently used measures were the Attribution Questionnaire by Corrigan et al. (2003) with an adapted factor structure by Brown (2008) and the Community Attitudes to Mental Illness (CAMI) by Taylor and Dear (1981) and adapted by Högberg et al. (2008). Link et al. (2004) highlight the importance of selecting measures based on the concept that is of interest and also the availability of validated measures. In the first instance, they advocate for adapting previously validated measures before considering the development of a new measure (Fox et al., 2018; Link et al., 2004). It seems necessary to highlight this viewpoint given its downstream impact on the current study and other researchers seeking to meta analyse or systematically review multiple studies.

There are also competing views around whether measures for mental health stigma should be adapted with specific diagnoses in mind (Pescosolido et al., 1999). Certainly, the current review suggested some evidence of a difference in public stigma between different diagnoses. Therefore, it would be important to research the use of diagnosis-specific measures of stigma in relation to offenders to better understand these differences. This could be done through group comparisons with a variety of symptoms associated with different mental health difficulties, where it might be possible to see the impact of particular elements of a diagnostic presentation eliciting a particular response, such as fear. It would then be possible to compare if these emotional or stereotypical responses are aligned with the known risks of those particular symptoms or associated mental health difficulties. Further research into the combining effect of the offender and mental health diagnoses would inform the necessity for specific or generalised terms when measuring mental health stigma.

Despite a vast majority of negative stigmatic attitudes, there was some positive evidence. Reassuringly, three studies found social work and criminology students, as well as public defenders, were less likely to have negative stigmatic attitudes and demonstrated compassionate views (Batastini et al., 2017; Frailing et al., 2016; Weaver et al., 2019). However, this was not held constant amongst students from other courses or amongst judges and prosecutors (Batastini et al., 2017; Weaver et al., 2019). Both of these specific populations appeared to have higher levels of education and training in relation to offending and even mental health, therefore education may have the potential to mitigate levels of stigma (Batastini et al., 2017; Frailing et al., 2016; Weaver et al., 2019). Understanding positive evidence is supportive in developing anti-stigma programs that act to reduce levels of stigma in the wider community.

Strengths and limitations

As highlighted, the current review included studies completed across the globe, including Ghana, Texas and the UK showing the diversity of the sample but unfortunately also the possible spread of negative stigmatic views across continents. In addition, the selected studies had diverse populations, from the general public to mental health professionals, and whilst the amount of stigma reported differed, the vast majority had negative stigmatic attitudes. Due to the wide variety of stigma measures, it was not possible to fully compare measures across studies and the use of highly specific offender mental health stigma measures meant that neutral control groups for comparisons were not available. This demonstrates an advantage of using adaptable vignettes in stigma research which would allow for previously validated measures to be easily compared to one another even with differential diagnostic and offending labels. A limitation to the review was that it focused on studies printed in the English language which inevitably has excluded some international research. An additional limitation was that searching was limited to articles published in the last ten years. The justification for limiting the publication date was to provide an up-to-date account of the current literature and to answer the research question around the commonality of offender mental health stigma research. Original authors of measures have been referenced as well as validation studies for the measures which are listed and in some cases, they pre-dated 2009.

Conclusions and recommendations

Given this review is the first to combine research in understanding the literature around stigma towards mental health and offenders, it highlights a number of key points for advancing research in the area. Firstly, it suggests a high level of stigma towards individuals with a psychiatric and an offending history. Further research is needed to better understand this complex relationship. Research could include studies where multiple conditions are compared, similar to the methodology used by Nee and Witt (2013). Secondly, the current review also re-emphasises the importance of selecting available validated measures, either specific to offender mental health stigma or with an adaption such as a vignette to allow for comparisons between studies and also within groups in large-scale studies. Finally, findings of this review contribute to measuring and understanding stigma towards those in vulnerable positions. It encourages further intervention-based research to bring about change and reductions in stigma. This is not only important for public stigma and the way individuals are treated in the community, but also for reductions in self-stigma which all together have ramifications for an individual's recovery and rehabilitation.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Data availability statement

The data is all in published articles available through commonly used subscription services.