A Systematic Review of Autistic People and the Criminal Justice System: An Update of King and Murphy (2014)

Studies using only quantitative methods accounted for 36 of the 47 papers (1–11, 13–26, 29–30, 32, 35–37, 39, 45–47). Additionally, six qualitative papers were reviewed (12, 34, 38, 40, 41, 42), and five used mixed methods analyses, in which authors collected both qualitative and quantitative data predominantly from surveys (27, 28, 31 43, 44) and interviews (31). Of the 47 papers, 8 were rated as 5* (17%), 25 were rated as 4* (53%), 9 were rated 3* (19%), and 5 were rated as 2* (11%).

The quality of included papers was compromised by several methodological limitations (e.g., poor generalisability, unrepresentative samples, lack of matched comparison samples, reliance on retrospective data collection, lack of standardized instruments), which impacted study outcomes and limited the extent to which findings could be relied upon. Several authors of studies that took a qualitative approach to the research reported data collection methods that were considered inadequate (e.g., 27, 28 & 40), and/or did not support their interpretation of findings with sufficient data (e.g., 34, 40, & 42), which led to an unclear link between qualitative data sources, data collection, analysis, and interpretation (e.g., 40 & 42). In addition, several authors of quantitative studies with non-randomised comparison samples did not recruit a representative sample of participants (e.g., 1, 4, 22, 29, 30–33, 36, 45, 47). Due to an overreliance on retrospective data collection methodologies and database/documentary analyses, missing data was a consistent weakness of the included studies (e.g., 4, 18 19, 24, 32, 46). Moreover, confounding variables were rarely accounted for in the design or analysis of findings (e.g., 1, 2, 4–10, 13–17, 21, 29, 30–32, 37, 46). Only one study did include a matched comparison sample controlling for confounding variables (24). Altogether, authors recruited 606 autistic adults, 1,271 adults with intellectual disabilities and autism and a population control group of 2,973 young adults (24). Those studies that utilised a mixed methods design, did not meet the criteria for both of the different components of the methods involved.

Of the 47 studies, 19 were conducted in the UK, 6 in Sweden, 6 in the USA, 4 in the Netherlands, 3 in Norway, 4 in Canada, 2 in Portugal and 3 in Australia. Of the 47 studies, authors of 9 recruited autistic people in contact with the CJS from state/nationally representative populations (14, 16, 22, 24–28 & 35). Authors of 13 studies recruited autistic people in contact with the CJS from psychiatric inpatient services (1, 4–6, 8, 11, 30, 32, 33, 36, 37, 40 & 45), 9 from prison and/or probation services (2, 9, 12, 17, 20, 23, 34, 38 & 41), 5 from the community (29, 31, 43, 44 & 47), 4 from forensic psychiatric examinations (7, 15, 42 & 46), 2 from police services (13 & 21), 2 from a residential treatment facility (18 & 19), 1 from a forensic outpatient service (3), 1 from community and inpatient forensic services (10), and 1 from a combination of prison, probation, and psychiatric inpatient services (39).

Of the 47 studies, authors of 20 recruited males only (2, 3, 9, 11, 12, 17–20, 23, 32, 33, 36–42 & 46), while authors of one study did not report the gender of participants (22), and authors of the remaining 26 studies recruited both males and females (1, 4–8, 10, 13–16, 21, 24–31, 34, 35, 43–45, 47).

Of the 47 studies, authors of 13 reported the age of autistic offenders (n = 326) ranging from 19 to 57.3 years (M = 31.50, SD = 11.03). Of the 47 studies, authors of 23 reported the gender of autistic offenders (n = 450), and most participants were male (n = 433) rather than female (n = 17). Findings are therefore biased towards autistic males.

Of the 47 studies, authors of only three studies reported the ethnicity of autistic offenders (n = 83). Ethnicities reported were White (n = 69), Black (n = 8), Asian (n = 1), and other (n = 5).

If autistic people were as likely as anyone else to be involved in the CJS we would expect the rates of autistic people in CJS populations to be about the same as the prevalence of autism in the general population (i.e., of the order of 1%).

Of the 47 studies included in the review, authors of 24 included prevalence data for autism amongst a sample of suspects or offenders (studies 1–24). A minority of studies reported using gold standard methods for diagnosing autism (2, 9, 11, 20), such as the ADI-R (Lord et al., 1994), Diagnostic Interview for Social & Communication disorders (DISCO; Wing et al., 2002) and the Autism Diagnostic Observation Schedule (ADOS; Lord et al., 2000; 2012). Several other studies used validated screening methods (17 and 23), including the Autism Spectrum Quotient (ASQ; Baron-Cohen et al., 2001), the Asperger’s Syndrome Diagnostic Scale (Myles et al., 2001). Other studies referred only to the ICD-10 or DSM-IV, perhaps due to a heavy reliance on retrospective case file data (e.g., 1, 4, 6–8).

Unbiased samples were recruited to a minority of studies (5, 7, 9, 12, 14, 15, 22). Most other studies recruited participants already hospitalised in psychiatric inpatient services who were therefore already identified as being a risk to either themselves or the public and as having a mental health condition (e.g., 1, 4, 6, 8 & 11). One study that recruited participants from the community, was also limited to people known to an outpatient psychiatric service (3). Even when unbiased samples were recruited, the methodological quality of these studies was limited as it was unclear how a diagnosis of autism was made and by whom (e.g., 5 & 7), or because samples of participants were limited by gender (e.g., 9 and 23), or type of offending behaviour (e.g., animal abuse or sexual offending; 19, 21).

Consequently, prevalence of autism amongst offending populations varied widely from 0.2% in a sample of offenders detained across seven prisons in New South Wales, Australia (12) to 62.8% in a sample of adolescent male offenders aged between 15 and 20 years who had been ordered by Juvenile Court into an adolescent sexual offenders’ program in the USA (19). Of the 16 studies to include prevalence data of autism derived from biased samples of offenders (n = 2926), 13.3% of offenders were reported to be autistic (n = 388; see Table 2). Of seven studies to include prevalence data of autism derived from mostly unbiased samples of offenders (n = 15,963), 3% of offenders were reported to be autistic (n = 486; see Table 2), though, none of the samples recruited can be said to be truly unbiased due to several factors, including the likelihood of prisons providing lower prevalence of autism due to people being diverted away from custody, as well as difficulties associated with the accuracy of diagnosis. The current prevalence of autism in offender populations is therefore slightly higher than the prevalence of autism in community samples, which range between 0.7 and 1.1% (Baxter et al., 2015; Brugha et al., 2016). Therefore, findings suggested autistic people may be somewhat over-represented in the Criminal Justice System, though biased samples and methodological limitations of the research conducted impact the reliability of these findings, for example, people recruited from the forensic psychiatric inpatient services often produce high prevalence rates as those referred are highly likely to have other mental health needs (King & Murphy, 2014). Moreover, it may be that suspects with limitations to their social behaviour (for example, poor eye contact) may be more likely to be seen as risky and therefore they may be more likely to receive custodial sentences. An added difficulty is that, due to varying samples and inconsistency in diagnosis assessment tools used when recruiting autistic suspects and offenders, it is difficult to make comparisons across studies.

Offending is known to be generally more common amongst teenage boys, rather than older men and girls (Farrington et al., 2021). If autistic people are as likely to offend as other people without autism, we would expect to see this reflected in figures for comparison samples of autistic and non-autistic people, matched for age and gender.

Of the 47 studies, authors of five reported on the prevalence of autistic people’s involvement in the CJS. Here it is essential for studies to have a comparison group of non-autistic people involved in the CJS for any conclusions to be drawn. None of the included studies reported a robust procedure for diagnosing autism, but instead relied on previous diagnosis, which on occasions was self-reported (e.g., 27 & 28). Study 29 did however, verify a self-reported diagnosis using the Social Communication Questionnaire (SCQ; Rutter et al., 2003). Studies reporting on the prevalence of involvement in the CJS for populations of autistic people were therefore of limited methodological quality due to a lack of robust assessment of autism.

Consequently, prevalence of involvement in the CJS for populations of autistic people varied widely from 0.7% in a community sample of 284 autistic adolescents and adults in Ontario, Canada (29) to 5% in a nationally representative sample of 920 autistic youth aged 15–17 years in the USA (26). Overall prevalence of involvement in the CJS across the six studies (n = 6978) was 4.3% (n = 300; see Table 2). However, what constitutes involvement in the CJS varied between studies, for example, study 26 reported on those who had been arrested, whereas study 25 reported on those convicted of at least one violent offence. Consequently, how authors define involvement in the CJS will impact on prevalence data, whereby a broader definition including any contact with the criminal justice system will increase prevalence figures as those who are arrested or questioned by the police may not go on to be convicted. A lack of research including unbiased comparison samples limits our understanding of whether those who were autistic were at more or less risk of offending in comparison to non-autistic people. Nevertheless, other studies of non-autistic adolescents (e.g., Farrington et al., 1975; 2021) suggest that these figures for autistic adolescents are much lower than for non-autistic adolescents. As reported by Farrington & Hawkins (1991), by the age of 21 years, 31% of males from working class backgrounds in England (n = 411) had a criminal conviction.

As can be seen in Table 2, types of offences committed by autistic individuals varied and included arson, property crime, sexual offending, drug offences, traffic incident/driving offences, stalking, fraud, theft, and robbery (e.g., 4, 7, 30, 31, 33, 34). Of the 47 studies included in the review, authors of 10 studies recruited a group of autistic people, as well as a comparison group of non-autistic people (1, 9, 10, 14, 24, 30, 39, 45, 47).

Authors of study 24 compared 45 autistic service-users recruited from two specialised low secure psychiatric services to 43 participants from a non-specialised low secure unit and reported autistic service-users were more likely to be younger, less likely to have a diagnosis of drug or alcohol misuse/dependence or personality disorder. Furthermore, autistic service users, when compared to non-autistic service users were less likely to have a lifetime history of physical violence or sexually inappropriate behaviour towards others or to be non-compliant with prescribed medication. They were also less likely to have a forensic history and their number of previous convictions tended to be lower (24). These findings are supported by study 10, in which findings suggested autistic people referred to forensic intellectual disability services had a lower rate of being charged with a criminal offence.

Limited research has been conducted to explore risk factors for autistic and non-autistic people for different types of offending. Authors of study 14 compared 143 autistic and 572 non-autistic juveniles with intellectual disabilities and other special educational needs involved in the CJS and reported autistic offenders were least likely to commit property offences. However, study 9 reported that at index conviction, autistic offenders were overrepresented in child sexual offending, when compared to non-autistic offenders. Though non-autistic offenders had more previous convictions (e.g., for drug related crimes), no difference in total number of prosecuted crimes was reported, although, the reliability of these findings is limited due to the sample consisting only of young male offenders detained within a prison establishment and convicted of violent offences only. Nevertheless, study 39 compared 40 autistic male offenders and 40 non-autistic offenders to a community sample of 40 autistic and 39 non-autistic people without a history of offending behaviour. Findings suggested autistic offenders scored higher for risk factors related to mental health, conduct problems, and family/childhood adversity (39).

In comparison, authors of study 30 compared autistic adults (n = 606), adults with intellectual disabilities (n = 1271), and population controls (n = 2973) and reported few group differences. Findings suggested that young autistic adults were not overrepresented in the CJS compared to their non-autistic peers and were less likely to be involved in the adult justice system than their peers. These findings are supported by the findings of study 47 in which similar rates across all forms of perpetration for autistic and non-autistic adults were reported.

Few studies compared the offending behaviour of autistic and non-autistic offenders to identify unique characteristics or similarities between these populations. Similarly, few studies compared autistic offenders to autistic people without an offending history to identify unique characteristics or risk factors amongst populations. Despite conflicting evidence regarding the similarities and differences between autistic and non-autistic offenders, authors of study 45 compared autistic males (n = 15) to a matched control group of non-autistic males (n = 15) on their ability to understand and follow court proceedings and stand trial. Findings suggested that autistic adults had more difficulty understanding and following aspects of the trial process and proceedings, despite their prior experience of the courtroom process. These findings suggest autistic adults who encounter the CJS have unique needs.

Authors of 33 of the 47 studies reported on comorbid psychiatric diagnosis of autistic offenders (see Table 2). Most authors recruited samples who were already in a mental health hospital or secure state facility (4, 11, 17 19, 30, 32, 33, 36 & 37), or who had been referred for forensic psychiatric assessment (7, 10, 15 & 34). Additionally, authors of five studies recruited from prison and probation services (study 2, 9, 21, 23, 38), while authors of one study recruited from the prison, probation, one approved premise and two hospitals (39), and authors of six studies included recruited participants from community samples (study 27, 28, 29, 31, 43, 44). However, in studies 32, 46 and 47, participants comprised autistic individuals who had interacted with the police for a variety of reasons (e.g., victimisation, mental health crisis) and therefore not all participants had a history of offending behaviour. Furthermore, two studies of poor methodological quality failed to report the type of forensic services in which autistic offenders were placed (40 & 42). Common psychiatric diagnoses reported included attention deficit hyperactivity disorder (ADHD), Tourette syndrome, intellectual disabilities, substance related disorders, psychosis, personality disorders, anxiety, mood disorders, and obsessive-compulsive disorder. Study 4 reported the prevalence of comorbid diagnosis of substance misuse and personality disorder were significantly lower in the autistic group compared to the non-autistic group. Study 33 reported the age at first contact with psychiatric services for the autistic group was lower compared to the non-autistic group, but age at first psychiatric admission was not significantly different. Further, 73.3% had psychiatric comorbidity, most commonly schizophrenia, but unlike controls, personality disorder and drug and alcohol disorders were uncommon. Additionally, a lifetime history of drug use, but not alcohol use was significantly less common for the autistic group, when compared to the non-autistic group. Study 39 reported that autistic offenders scored higher on the mental health risk factors than both typically developing offenders and typically developing non-offenders. Study 16 reported that serial offenders were more likely to have a personality disorder or autism when compared to single offenders.

Authors of five studies used a population-based cohort leading to findings that are likely to be more representative of offenders with an autism diagnosis (20, 22, 25, 26, 35). These studies showed that co-occurring intellectual disabilities and ADHD were most common, as well as psychotic disorder, drug and alcohol misuse, personality disorder, affective disorders, anxiety, tic disorder, and obsessive-compulsive disorder. Authors of study 25 compared autistic and non-autistic individuals on the Swedish national crime register and found autistic individuals who also had ADHD and/or conduct disorder had an increased risk of violent criminality. Furthermore, autistic individuals with intellectual disabilities were less likely to be convicted, and autistic individuals without intellectual disability were more likely to be convicted, compared to people without any of these conditions. However, other methodological issues affecting the reliability and validity of findings include retrospective data collection, a lack of robust assessment, and female offenders being underrepresented.

Several studies recruited participants from the autistic community, asking them about their experiences of the police in England, Wales, and Australia (31, 43, 44). Findings suggested that over 60% of respondents were dissatisfied with their experience with the police (31 & 43). The autistic community reported a lack of police awareness and knowledge of autism, a lack of information and explanation given by the police, and delays throughout the CJS (43). Furthermore, the needs of autistic individuals were often not met, and some individuals felt victimised or discriminated against by police officers (43). Interestingly, some researchers have explored whether autistic people disclose their diagnosis to the police, with findings suggesting as many as 36% never disclosed and 25% only disclosed their diagnosis on some occasions (study 43). Furthermore, their decision to not disclose was linked to a fear of discrimination or victimisation by police officers (study 43). Studies 27 and 28 recruited 35 autistic adults, 80% of whom reported to have interacted with the police at least once in their lifetime (n = 29), including being stopped by police (n = 17), being a victim of a crime (n = 17), for a mental health crisis (n = 11), suspected of a crime (n = 8), placed in handcuffs (n = 8) and convicted (n = 2). As many as 42.3% (n = 11) of respondents reported that the police had used force during the described interaction, including restraint (36.4%, n = 4), handcuffs (36.4%, n = 4), and assault (n = 3). Furthermore, none of the respondents indicated that the officer(s) were able to identify that the person with autism was vulnerable (27 & 28).

Other researchers have explored the experiences of autistic offenders in prison establishments (e.g., 12, 34 & 41). Research has highlighted the challenges surrounding the misunderstanding of autism in relation to social interactions with staff and other prisoners, which on occasions has escalated into altercations. Difficulties coping with feelings of frustration, stress, fear, upset, confusion, anxiety, and agitation that were caused by the unpredictability of the prison environment (i.e., changes in prison routine), and noise was reported as very difficult to cope with, as well as a lack of staff understanding regarding their diagnosis of autism. In contrast, some autistic offenders (n = 6) have reported coping in prison, as it provided structure and opportunities for work and study (34). It is likely that prisons vary in their ability and willingness to support autistic offenders. For example, Her Majesty’s Young Offender Institution Feltham, Her Majesty’s Prison Parc and Her Majesty’s Prison Wakefield have received autism accreditation from the National Autistic Society demonstrating their commitment to providing support to autistic offenders (Hughes, 2019; Lewis et al., 2015).

Several studies explored other risk factors associated with autistic offenders, including being male (29), single (7, 33), not having children (33), having birth defects (10), behavioural issues during childhood (14), poor educational attainment (7, 37), lack of regular employment (7, 37), poor social networks (7), having a history of verbal and physical aggression (4, 29, 32), offending/antisocial behaviour (33, 45), sexually inappropriate behaviour (33), substance misuse (31, 38), and a history of victimisation (4, 10, 19, 26, 27, 28, 31 33, 40 & 46). A range of parental and familial characteristics, such as belonging to households with lower incomes, migrant households, parental history of criminal convictions, and maternal psychiatric disorders were associated with violent criminality in autistic individuals (40). Authors of study 7 reported that 29% of the group had been in foster or institutional care as children (n = 14) and 4% had more than three changes of (non-family) carer during childhood (n = 2). Furthermore, a delayed diagnosis of autism has also been identified as a potential risk factor for offending (15, 25, 40). Several studies also reported on the motivations of autistic offenders. Common motivations included revenge, social misunderstanding, idiosyncratic rationalisations or explanations, obsessions/special interests, and social naivete (7, 15, 38).

However, autistic offenders were rarely compared to non-autistic offenders in these analyses, meaning their unique characteristics have not been identified. Nevertheless, authors of study 7 reported on a sample of 48 autistic offenders who had undergone forensic examination between 2000 and 2010 in Norway and found that 56% had no previous convictions (n = 27). Furthermore, authors of study 33 reported that a lifetime history of sexually inappropriate behaviour (n = 13) and physical violence (n = 35) were less common for the autistic offenders, when compared to non-autistic offenders. Similarly, study 9 recruited 269 violent male offenders, including 26 autistic offenders. Findings suggested there were few differences between autistic and non-autistic offenders, including prevalence of psychopathy, although, childhood placements in foster homes were overrepresented in the autistic group. Study 39 recruited arguably a more representative sample of males from an offending and non-offending population with and without autism from England and Wales from four prison establishments, two probation services, one approved premise, two secure hospitals and the community. Autistic offenders reported higher scores on the offending risk factors questionnaire (unpublished) than non-offenders and also offenders without autism. Autistic offenders scored significantly higher than the non-autistic, non-offender group on family and childhood adversity risk factors. Autistic offenders scored higher than both the autistic non-offenders and non-autistic non-offenders on the conduct problems risk factors. However, few other studies have sought to compare findings between populations, and due to biased sampling, the findings of the current review may be an overestimation of aggression and violence in autistic offenders.

Furthermore, autistic people who encounter the CJS also appeared to be vulnerable to social isolation and segregation following arrest, due to difficulties associated with interpersonal skills and communication (11, 12). For example, study 26 reported that among a nationally representative sample of 920 autistic youth, 27.5% had experienced social isolation. In addition, research also suggested as many as 97.6% of autistic offenders having a history of self-harm (n = 41; study 4 & 37). Furthermore, the prevalence of self-harm in the autistic group was significantly higher than the non-autistic group (4). In contrast, study 33 found that a history of self-harm was less prevalent for autistic service-users admitted to a secure psychiatric inpatient service, compared to non-autistic service-users (n = 17), although the difference did not reach statistical significance.

Authors of several studies identified that autistic people who encounter the CJS often have a previous history of victimisation. For example, study 4 recruited 138 service-users from a specialised forensic inpatient intellectual disabilities service and findings suggested 40.5% of autistic service-users reported being a victim of any type of abuse (n = 17), and 33.3% reported being a victim of sexual abuse (n = 14). Similarly, study 10 reported on referrals made to forensic community and secure intellectual disability services in England and Scotland and found 23.4% of autistic people had been abused as a child (n = 11). In contrast, authors of study 33 recruited a control group and compared autistic and non-autistic service users admitted to two secure psychiatric inpatient services and found a lower percentage of the autistic group had a history of childhood neglect or abuse. However, study 19 which recruited 43 male adolescents adjudicated ‘delinquent’ due to a sexual offence in a state residential facility in Pennsylvania, reported emotional abuse and neglect were more common in the autistic group compared to the non-autistic group.