Psychosis and Criminal Offending: A Population-Based Data-Linkage Study

Abstract

This population-based case-control study examines the association between psychosis and criminal convictions in New South Wales (NSW), Australia, using data from several health and offending administrative data collections. Cases were individuals diagnosed with psychosis between 2001 and 2012 (n = 86,461). For each case, two age- and sex-matched controls with no diagnosis of psychosis were selected. Criminal convictions were identified using the NSW Reoffending Database. Cases were approximately 5 times more likely to offend compared with controls, adjusted odds ratio (aOR) = 4.68, 95% confidence interval (CI) = [4.55, 4.81], and accounted for 10% of all criminal convictions in NSW between 2001 and 2015. The prevalence of at least one criminal conviction was 30% among cases compared with 6% among controls. The results from this study confirm previous work regarding the association between psychosis and criminal convictions. More work is needed to better articulate the mechanisms for this association to enable prevention strategies to be developed.

Keywords

psychosis offending violent crime nonviolent crime data-linkage

Significant challenges exist at all stages of the criminal justice system for those with serious mental illness, including during apprehension by the police, processing by the courts, detention by custodial authorities, and reintegration back into the community on release from prison. Studies have consistently shown increased rates of offending in people diagnosed with psychosis including schizophrenia (Morgan et al., 2013; Mullen, 2006; Schanda et al., 2004). A 2009 meta-analysis of 204 studies, covering 166 individual data sets, showed psychosis to be associated with a 49% to 68% increase in the odds of a criminal conviction for violence than those without psychosis (Douglas et al., 2009). A systematic review of 27 studies reported a significant increase in the risk of reoffending in those with psychosis than those without psychosis, but no association in comparison with people with psychiatric disorders other than psychosis (Fazel & Yu, 2011). Skeem et al. (2016), using data from the MacArthur Violence Risk Assessment study, showed that 12% of violent incidents were immediately preceded by psychosis (Skeem et al., 2016).

More than one third of all convictions of homicides in England and Wales from 1996 to 1999 occurred in those with a mental disorder, with 5% having schizophrenia (Shaw et al., 2006). Almost 9% of all homicides in New Zealand from 1970 to 2000 were associated with serious mental illness (Simpson et al., 2004), and 8.8% of all homicides in New South Wales (NSW) Australia were found to have been committed during a psychotic episode between 1993 and 2002 (Nielssen et al., 2007). Another Australian study of 435 people convicted of homicide reported that schizophrenia was 13 times higher in the homicide group than the general population (Bennett et al., 2011).

Although the above evidence is compelling, the relationship between psychosis and offending is more complex and the mechanisms leading to these findings have not been well articulated by empirical studies. Serious mental illness is associated with an increased likelihood of having criminogenic factors, such as homelessness, poor social relationships, being unmarried, unemployment, substance use, poor education, and antisocial behaviors. Indeed, it has been suggested that, in many instances, mental health symptomology does not have a direct causal effect on most offending behavior (Peterson et al., 2014). Although symptoms of mental illness are generally not good predictors of criminal behavior, and are not as strongly associated with offending as some other criminogenic factors, there is however a modest association (Bonta et al., 2014). Bonta et al. (2014, p. 278) states that “From both a risk prediction and a recidivism reduction perspective, symptoms of mental illness do not appear to play a major role.” Notwithstanding this, explanations of the association between psychosis and offending include the following: (a) a direct relationship whereby psychotic symptomology is related to offending (e.g., where hallucinations tell a person to offend); (b) an indirect relationship in which symptomology has some bearing on offending, such as depressive symptoms and irritability, causing the person to act aggressively; and (c) a mediated relationship whereby symptomology is affected by external criminogenic factors, such as alcohol and other drugs, causing an individual to offend (Peterson et al., 2014). There is evidence that treatment in prison significantly delayed the time to reoffending following release among individuals with schizophrenia (18% reduction; Igoumenou et al., 2015).

Substance use is also likely to confound the association between mental illness and offending. A meta-analysis of 20 studies showed that the risk of violent offending in individuals with psychosis and comorbid substance abuse was similar to that for substance abuse alone without psychosis (Fazel et al., 2009). However, an Australian study found that individuals with schizophrenia without comorbid substance-use disorders were 2 times more likely to have a violent conviction compared with those who had never been diagnosed with schizophrenia (Short et al., 2013). Although the risk of offending among people with serious mental illness has been reported to be high, evidence suggests that comorbid substance use can trigger violence in this group. Moreover, Swanson et al. (2015) found that other risk factors such as substance abuse had a stronger association with gun violence than mental illness (Swanson et al., 2015).

No population-based study of psychosis and offending has been conducted in NSW, which the current study aimed to address. Furthermore, whereas much of the evidence has focused on the association between serious mental illness and violence, less attention has been given to the relationship with other types of offending. We examined the association between psychosis and violent and nonviolent offenses in NSW between 2001 and 2015 using data linkage of health and justice administrative data collections. We also investigated the association between criminal convictions and psychosis due to psychoactive substance use.

Method

Data Sources

We used the de-identified whole of population administrative data linked across several NSW Health and Justice systems. The NSW Ministry of Health’s Admitted Patient Data Collection (APDC) includes records for all hospital separations from all NSW public and private hospitals and day procedure centers. The APDC records include demographic information, administrative data, and coded diagnostic information. The NSW Emergency Department Data Collection (EDDC) provides information about presentations to emergency rooms in public hospitals. We extracted data on diagnosis type (see definition of psychosis), age at diagnosis, treatment episode start and end dates, gender, Aboriginality (yes, no, and missing/unknown), marital status (married including de facto, single, and missing/unknown), date of birth, and statistical local area (SLA) from the APDC and EDDC. With Aboriginal people significantly overrepresented in the justice system in Australia (28% of the total Australian population in prison are Aboriginal compared with around 3% in the general population; Australian Bureau of Statistics, 2019), we included Aboriginality in the analysis. Aboriginal status is based on any record of endorsement of Aboriginality in the data for a service contact. “Single” marital status refers to those who are single or widowed or divorced or permanently separated from their partner. The index of relative socioeconomic disadvantage (IRSD) is used to rank the socioeconomic status in each SLA by the Australian Bureau of Statistics (Australian Bureau of Statistics, 2016). IRSD is one of the four socioeconomic indexes for areas (SEIFA), which provides scores to rank the socioeconomic status in each geographic area, using data on income, education, employment, occupation, and housing. The lowest rank indicates the most disadvantaged area and the highest rank the most advantaged area. We categorized areas into disadvantaged (score 1–5) and advantaged (score 6–10).

The NSW Reoffending Database (RoD) contains information on each individual convicted of a criminal offense in NSW since 1994. The following data were extracted from the RoD: gender, Aboriginality, offense type (see primary outcome), offense date, age at the time of the offense, and principal penalty (i.e., nonsupervised order, supervised order, and custodial order) in the case of a court appearance. Minor traffic infringements were excluded from the offense data. The NSW Offender Integrated Management System (OIMS) contains demographic information on all individuals in prison in NSW and the duration of incarceration, which was used to identify whether those with or without psychosis had any prison episode between July 2001 and June 2014 (latest available at the time of linkage).

Ethics Approvals

Approvals were obtained from the NSW Population and Health Services Research Ethics Committee (HREC/15/CIPHS/17), Justice Health and Forensic Mental Health Network (G324/14), Corrective Services NSW (D15/138715), Cancer Institute NSW (2015/05/586), and the NSW Aboriginal Health and Medical Research Council (1089/15).

Study Population

Cases

Individuals who had at least one episode of hospital care (private or public) or an emergency department presentation in which the primary or additional diagnosis of psychosis was recorded were selected as cases. The cohort comprised all those diagnosed with psychosis in either the APDC, between July 2001 and December 2012, or in the EDDC, between June 2005 and December 2012. All diagnoses were made by clinicians based on the hospital system. The first psychosis admission date in either the APDC or EDDC, whichever occurred earliest, was selected as the index date for each case.

Definition of Psychosis

Diagnoses from the APDC were coded according to the International Classification of Diseases, 10th Revision (ICD-10) and diagnoses from the EDDC were coded according to the International Classification of Diseases, 9th Revision (ICD-9) or 10th Revision (ICD-10) or the Systematized Nomenclature of Medicine–Clinical Terms (SNOMED CT). Psychosis was identified according to the ICD-9 and ICD-10 codes and mapped to the relevant SNOMED codes. Mapping was done by the National Clinical Terminology and Information Service, Australian Digital Health Agency. Psychotic disorders for the purpose of this study included schizophrenia and related psychoses (F20, F22–F25, F28, F29, and 295), affective psychoses (F30.2, F31.2, F31.5, F32.3 F33.3, 296.8, and 296.9), and substance-related psychoses (F10.5, F11.5, F12.5, F13.5, F14.5, F15.5, F16.5, F17.5, F18.5, F19.5, 291, and 292). We used a hierarchical approach to psychosis with those having a diagnosis of schizophrenia and related psychoses coded as “schizophrenia and related psychoses”; any diagnosis of affective psychoses with no diagnosis of schizophrenia and related psychoses was coded as “affective psychoses” and substance-related psychoses in the absence of the other two groups was coded as “substance-related psychoses.”

Controls

Two age- and sex-matched controls without any record of a diagnosis of psychosis were randomly selected for each case from the Master Linkage Key (MLK) held by the NSW Ministry of Health. The MLK is a system of continuously updated links within and between 16 core health-related data collections in NSW that contains around 178 million records pertaining to more than 15 million individuals and was constructed by the Center for Health Record Linkage (CHeReL).

Selection of Controls

The MLK control selection sampling frame for this study comprised records from the following NSW data collections: admitted patients’ data, emergency department data, mental health outpatients’ data, Registry of Births, Deaths and Marriages, Perinatal Data Collection, Cancer Registry, Pap Test Registry, and Notifiable Conditions System. The following persons were excluded from the MLK control selection sampling frame: those identified as cases; those with a date of death prior to January 1, 2001, and those with no key demographic information (e.g., name, date of birth) on any of their MLK records, which would render them unlinkable. The SURVEYSELECT procedure in SAS (SAS Version 9.3) was used by the CHeReL to select controls.

Primary Outcome

The primary outcome of this study was a conviction, as recorded in the RoD. We analyzed RoD data for the period between July 2001 and June 2015 (latest available at the time of the linkage). Criminal convictions were coded according to the Australian and New Zealand Standard Offence Classification (ANZSOC; Australian Bureau of Statistics, 2011). We grouped offenses into violent (ANZSOC codes 111–621) and nonviolent (ANZSOC codes 711–1699). Offenses that did not result in convictions were not included in this study. We used the terms “conviction,” “criminal conviction,” and “offending” interchangeably.

Data Linkage

Data linkage was performed by the CHeReL to ensure the anonymity of the data from the researchers using probabilistic record linkage methods and the ChoiceMaker software (Borthwick, 2003). The linkage algorithm considers name reversals, name shortenings, and a limited number of keystroke errors. Once the linkages were finalized, the CHeReL created a Project Person Number (PPN) for each person identified in the linkage and assigned this PPN to every data set used for this study. The CHeReL returned the PPN and the encrypted record number from the source data sets to data custodians who independently supplied the data sets to the researchers.

Statistical Analysis

The categorical variables were described using frequencies and percentages. They were compared using the chi-square tests. Continuous variables were described using mean, standard deviation (SD), median (Mdn), and interquartile range (IQR) to present the age of cases and controls. Multivariate logistic regression models were used to assess the association between criminal convictions and psychosis as well as the other sociodemographic variables. The final multivariate model was determined using a forward stepwise approach. Risk factors were entered into the multivariate model if they had a p value of less than .100 in the univariate analysis. The final model only included statistically significant covariates (p < .05). Separate logistic regression models were fitted for both violent and nonviolent criminal convictions (at least one violent and nonviolent conviction between July 2001 and June 2015, respectively).

In a subanalysis, we used an additional variable as the presence of any substance comorbidity either within the episode of care or lifetime. The purpose of this analysis was to investigate the proportion of increase in the risk of offending in people with substance-related psychoses. All data were analyzed using SAS Version 9·4 and STATA 14.0 (College Station, TX, USA).

Results

Sociodemographic and Criminographic Characteristics

Overall, 86,461 individuals were identified as having at least one record of a diagnosis of psychosis on admission to hospital or presentation to an emergency department between July 1, 2001 and December 31, 2012 (Table 1). Of these, 48,193 (56%) were men and 38,268 (44%) were women. A total of 172,922 age- and sex-matched controls with no record of any diagnosis of psychosis were selected during the same period. Over half of the study population were 40 years or older, whereas around 5% were younger than 20 years of age at the time of the most recent diagnosis. Although the controls were matched with the cases by age and sex, age was significantly different between the groups (p < .001) due to the large sample size. Cases were more likely to be from disadvantaged areas, according to SEIFA (50% vs. 42%, p < .001). Almost one third of psychosis cases (30%) had been convicted of an offense between July 2001 and June 2015, in comparison with 6% of controls (p < .001), whereas approximately one fifth of the cases (19%) and 3% of the controls had a conviction for a violent offense (p < .001). About 11% (n = 9,744) of all cases had a prison episode between July 2001 and June 2014, whereas for controls, this was only 1% (n = 1,958; p < .001).

Table 1:

Sociodemographic (July 2001–December 2012) and Criminographic (July 2001–June 2015) Characteristics of the Study Population

Characteristic	Total(n = 259,383)		p	Men(n = 144,618; 56%)		p	Women(n = 114,765; 44%)		p
	Cases(n = 86,461) (33%)	Controls(n = 172,922) (67%)		Cases(n = 48,193) (33%)	Controls(n = 96,425) (67%)		Cases(n = 38,268) (33%)	Controls(n = 76,497) (67%)
	n (%)	n (%)		n (%)	n (%)		n (%)	n (%)
Age groups at recent diagnosis			<.001			<.001			<.001
<20 years	4,486 (5)	8,085 (4)		2,574 (6)	4,917 (5)		1,912 (5)	3,168 (4)
20−29 years	16,456 (19)	31,317 (18)		10,771 (22)	20,421 (21)		5,685 (15)	10,896 (14)
30−39 years	19,244 (22)	38,560 (23)		12,098 (25)	23,737 (24)		7,146 (18)	14,823 (19)
40+ years	46,275 (54)	94,960 (55)		22,750 (47)	47,350 (50)		23,525 (62)	47,610 (63)
Mean age at recent diagnosis (range), years	45 (16−90)	48 (15−93)		42 (16−87)	44 (15−91)		49 (16−92)	53 (16−95)
SD	70	69		66	79		74	52
Median age at recent diagnosis (IQR), years	42 (30−58)	44 (32−63)		39 (29−52)	41 (30−56)		46 (33−65)	51 (35−72)
Aboriginal			<.001			<.001			<.001
No	80,436 (93)	131,676 (76)		44,511 (92)	75,128 (78)		35,925 (94)	56,548 (74)
Yes	5,408 (6)	2,531 (2)		3,305 (7)	1,579 (2)		2,103 (5)	952 (1)
Missing/unknown	617 (1)	38,715 (22)		377 (1)	19,718 (20)		240 (1)	18,997 (25)
Marital status			<.001			<.001			<.001
Married (including de facto)	17,480 (20)	62,102 (36)		7,729 (16)	33,500 (35)		9,751 (25)	28,602 (37)
Single	55,832 (65)	44,533 (26)		32,233 (67)	23,508 (24)		23,599 (62)	21,025 (28)
Missing/unknown	13,149 (15)	66,287 (38)		8,231 (17)	39,417 (41)		4,918 (13)	26,870 (35)
SEIFA			<.001			<.001			<.001
Advantaged	37,628 (44)	74,550 (43)		20,242 (42)	40,462 (42)		17,386 (45)	34,088 (45)
Disadvantaged	43,597 (50)	72,989 (42)		24,456 (51)	39,924 (41)		19,141 (50)	33,065 (43)
Missing/unknown	5,236 (6)	25,383 (15)		3,495 (7)	16,039 (17)		1,741 (5)	9,344 (12)
Any criminal conviction			<.001			<.001			<.001
No	60,306 (70)	162,552 (94)		28,534 (59)	87,486 (91)		31,772 (83)	75,066 (98)
Yes	26,155 (30)	10,370 (6)		19,659 (41)	8,939 (9)		6,496 (17)	1,431 (2)
Any violent criminal conviction			<.001			<.001			<.001
No	70,347 (81)	167,766 (97)		35,810 (74)	91,826 (95)		34,537 (90)	75,940 (99)
Yes	16,114 (19)	5,156 (3)		12,383 (26)	4,599 (5)		3,731 (10)	557 (1)
Any prison episode			<.001			<.001			<.001
No	76,717 (89)	170,964 (99)		40,118 (83)	94,607 (98)		140 ( <1)	76,357 ( > 99)
Yes	9,744 (11)	1,958 (1)		8,075 (17)	1,818 (2)		1,669 (4)	36,599 (96)

Note. Marital status and SEIFA were determined at the episode care of most recent diagnosis. Single marital status includes those who were single or widowed or divorced or permanently separated from their partner. IQR = interquartile range; SEIFA = socioeconomic indexes for areas.

Diagnosis of Psychosis

Overall, the 86,461 cases had a total of 334,344 records of hospital contacts (i.e., hospital admissions or emergency department presentations) between 2001 and 2012. Most records of contact with health services were associated with a diagnosis of schizophrenia and related psychoses (82%), followed by affective psychoses (10%) and substance-related psychoses (8%; Table 2). Table 2 presents the first diagnosis in the data, diagnosis based on the hierarchical approach, consistent diagnosis during 2001–2012, and the most recent diagnosis. Overall, schizophrenia and related psychoses were the most common conditions in each of these groupings (69%, 76%, 62%, and 72%, respectively).

Table 2:

Characteristics of Psychosis Diagnosis by Gender, July 2001–Dec 2012

Psychosis categorization	Overall(n = 86,461)			Men(n = 48,193; 56%)			Women(n = 38,268; 44%)
Psychosis categorization	Schizophrenia and related psychoses	Affective psychoses	Substance-related psychoses	Schizophrenia and related psychoses	Affective psychoses	Substance-related psychoses	Schizophrenia and related psychoses	Affective psychoses	Substance-related psychoses
First diagnosis	59,578 (69)	13,047 (15)	13,836 (16)	33,043 (69)	5,543 (11)	9,607 (20)	26,535 (69)	7,504 (20)	4,229 (11)
Median age at first diagnosis (IQR), years	41 (29−58)	45 (32−62)	33 (24−43)	37 (27−51)	44 (31−59)	33 (24−43)	47 (32−66)	46 (33−64)	32 (24−43)
Hierarchical diagnosis	65,895 (76)	10,218 (12)	10,348 (12)	36,712 (76)	4,361 (9)	7,120 (15)	29,183 (76)	5,857 (15)	3,228 (9)
Consistent diagnosis across 2001−2012	53,649 (62)	9,871 (11)	10,348 (12)	29,609 (61)	4,152 (9)	7,120 (15)	24,040 (63)	5,719 (15)	3,228 (8)
Most recent diagnosis	62,050 (72)	12,145 (14)	12,266 (14)	34,580 (72)	5,125 (11)	8,488 (18)	27,470 (72)	7,020 (18)	3,778 (10)
Median number of diagnoses (IQR) according to the hierarchical diagnosis	2 (1−5)	1 (1−2)	1 (1−1)	2 (1−5)	1 (1−2)	1 (1−1)	2 (1−4)	1 (1−2)	1 (1−1)
Total records of psychosis diagnoses (n = 334,344)	275,255 (82)	31,485 (10)	27,604 (8)	159,048 (83)	12,685 (7)	19,577 (10)	116,207 (81)	18,800 (13)	8,027 (6)

Note. IQR = interquartile range.

Psychosis and Criminal Convictions

Out of 1,059,153 offenses committed in NSW between July 2001 and June 2015, psychosis cases accounted for 104,640 (10%) of these offenses (Table 3). The median age at first offense in both cases and controls was 30 years. The median age at the time of the first diagnosis of psychosis (across all psychosis diagnostic categories) of people convicted of a crime was 31 years (IQR = 24–40 years). The median number of convictions was 2 (IQR = 1–5) among those with psychosis and 1 (IQR = 1–2) among the controls. Those with substance-related psychoses had a higher number of criminal convictions (Mdn = 3, IQR = 1–6) than those with schizophrenia and related psychoses (Mdn = 2, IQR = 1–5) and affective psychoses (Mdn = 2, IQR = 1–3).

Table 3:

Principal Offenses (Records) by Cases (n = 26,155) and Controls (n = 10,370), July 2000–June 2015

	Cases(n = 26,155; 72%)								Controls(n = 10,370; 28%)		All offenses in NSW (cases)
Characteristic	Overall(cases)		Schizophrenia and related psychoses(n = 19,119; 73%)		Affective psychoses(n = 1,640; 6%)		Substance-related psychoses(n = 5,396; 21%)		Controls(n = 10,370; 28%)		All offenses in NSW (cases)
Median age at first offense (IQR), years	30 (22−39)		30 (23−39)		35 (26−44)		29 (22−38)		30 (22−40)
Median age at first diagnosis (IQR), years	31 (24−40)		31 (24−39)		36 (27−45)		31 (24−40)		—
Median number of offenses (IQR)	2 (1−5)		2 (1−5)		2 (1−3)		3 (1−6)		1 (1−2)
Offense type	n (%)	Rate [CI]	n (%)	Rate [CI]	n (%)	Rate [CI]	n (%)	Rate [CI]	n (%)	Rate [CI]
Violent (ANZSOC code)	30,605(29)	354.0[350.8, 357.2]	22,780(30)	345.7[342.1, 349.4]	1,509(34)	147.7[141.0, 154.7]	6,316(26)	610.4[601.0, 619.8]	7,337(31)	42.4[41.5, 43.4]	316,806 (10%)
Abduction, harassment(511−532)	1,352(1)	15.6[14.8, 16.5]	989(1)	15.0[14.1, 16.0]	62 (1)	6.1[4.7, 7.8]	301(1)	29.1[26.0, 32.5]	241(1)	1.4[1.2, 1.6]	10,107(13%)
Acts intended to cause injury (211−299)	23,850(23)	275.8[272.9, 278.8]	17,832(24)	270.6[267.2, 274.0]	1,153 (26)	112.8[106.9, 119.1]	4,865(20)	470.1[460.6, 479.9]	5,359(23)	31.0[30.2, 31.8]	229,966 (10%)
Dangerous or negligent acts(411−499)	3,005(3)	34.8[33.6, 36.0]	2,077(3)	31.5[30.2, 32.9]	203 (5)	19.9[17.3, 22.8]	725(3)	70.1[65.3, 75.2]	1,256(5)	7.3[6.9, 7.7]	54,401(6%)
Homicide(111−132)	133(<1)	1.5[1.3, 1.8]	98(<1)	1.5[1.2, 1.8]	14 (<1)	1.4[0.8, 2.3]	21(<1)	2.0[1.3, 3.1]	39(<1)	0.2[0.1, 0.3]	1,744(8%)
Robbery, extortion(611−621)	1,496(1)	17.3[16.5, 18.2]	1,156(2)	17.5[16.6, 18.6]	40 (1)	3.9[2.9, 5.3]	300(1)	29.0[25.9, 32.4]	252(1)	1.5[1.3, 1.6]	13,100(11%)
Sexual(311−329)	769(1)	8.9[8.3, 9.5]	628(1)	9.5[8.8, 10.3]	37 (1)	3.6[2.6, 5.0]	104(<1)	10.1[8.3, 12.2]	190(1)	1.1[1.0, 1.3]	7,488(10%)
Nonviolent (ANZSOC code)	74,035(71)	856.3[853.9, 858.6]	52,730(70)	800.2[797.2, 803.3]	2,990 (66)	292.6[283.9, 301.6]	18,315 (74)	1,769.9—	16,259 (69)	94.0[92.7, 95.4]	742,347 (10%)
Offenses against justice/govt. procedures(1511−1569)	17,458(17)	201.9[199.3, 204.6]	12,195(16)	185.1[182.1, 188.1]	787(17)	77.0[72.0, 82.4]	4,476(18)	432.5[423.1, 442.2]	3,593(15)	20.8[20.1, 21.5]	148,140 (12%)
Drugs(1011−1099)	11,299(11)	130.7[128.5, 132.9]	8,047(11)	122.1[119.6, 124.6]	365(8)	35.7[32.3, 39.5]	2,887(12)	279.0[270.5, 287.8]	2,635(11)	15.2[14.7, 15.8]	110,850 (10%)
Fraud, deception(911−999)	2,532(2)	29.3[28.2, 30.4]	1,717(2)	26.1[24.9, 27.3]	137(3)	13.4[11.4, 15.8]	678(3)	65.5[611.9, 70.5]	901(4)	5.2[4.9, 5.6]	40,467(6%)
Property/environmental damage(1211−1229)	7,091(7)	82.0[80.2, 83.9]	5,376(7)	81.6[79.5−83.7]	312(7)	30.5[27.4, 34.1]	1,403(6)	135.6[129.1, 142.3]	1,467(6)	8.5[8.1, 8.9]	73,600(10%)
Public order(1311−1334)	10,536(10)	121.9[119.7, 124.1]	7,862(10)	119.3[116.9, 121.8]	376(8)	36.8[33.3, 40.6]	2,298(9)	222.1[214.2, 230.2]	2,114(9)	12.2[11.7, 12.8]	97,209(2%)
Theft(811−841)	15,758(15)	182.3[179.7, 184.8]	11,059(15)	167.8[165.0, 170.7]	644(14)	63.0[58.5, 67.9]	4,055(16)	391.9[382.6, 401.4]	2,810(12)	16.3[15.7, 16.9]	152,669(2%)
Traffic(1411−1441)	2,761(3)	31.9[30.8, 33.1]	1,728(2)	26.2[25.0, 27.5]	140(3)	13.7[11.6, 16.2]	893(4)	86.3[81.1, 91.9]	1,036(4)	6.0[5.6, 6.4]	39,213(3%)
Weapons and explosives(1111−1129)	777(1)	9.0[8.4, 9.6]	528(1)	8.0[7.4, 8.7]	40(1)	3.9[2.9, 5.3]	209(1)	20.2[17.7, 23.1]	344(1)	2.0[1.8, 2.2]	13,128(3%)
Burglary(711)	4,437(4)	51.3[49.9, 52.8]	3,286(4)	49.9[48.2, 51.6]	116(3)	11.4[9.5, 13.6]	1,035(4)	100.0[94.4, 106.0]	827(4)	4.8[4.5, 5.1]	43,697(2%)
Miscellaneous(1611−1699)	1,386(1)	16.0[15.2, 16.9]	932(1)	14.1[13.3, 15.1]	73(2)	7.1[5.7, 9.0]	381(2)	36.8[33.4, 40.6]	532(2)	3.1[2.8, 3.3]	23,374(2%)
Total n (%)	104,640 (82)		75,510 (72)		4,499 (4)		24,631 (24)		23,596 (18)		1,059,153 (10%)
Sentence type
Nonsupervised order	67,171 (64)		49,612 (66)		3,149 (70)		14,410 (59)		16,230 (69)
Supervised order	18,862 (18)		12,676 (17)		835 (19)		5,171 (21)		4,285 (18)
Custodial sentence	18,787 (18)		13,222 (17)		515 (11)		5,050 (20)		3,081 (13)
Total n (%)	104,640 (82)		75,510 (72)		4,499 (4)		24,631 (24)		23,596 (18)

Note. All offenses in NSW (%) refer to the proportion of all offenses committed by people with a diagnosis of psychosis in NSW. Rates of offenses were calculated per 1,000 population. Offenses against justice/government procedures refer to breaches of custodial, community-based, and violence and nonviolence restraining orders, and other technical offenses against government/justice procedures. NSW = New South Wales; IQR = interquartile range; CI = confidence interval; ANZSOC = Australian and New Zealand Standard Offence Classification.

Conviction types in the cases and controls were similar in terms of those classified as violent (29% and 31%) and nonviolent (71% and 69%; Table 3). However, the prevalence of violent convictions was lower in those with substance-related psychoses (26%) than those with schizophrenia and related psychoses (30%) and affective psychoses (34%). The conviction rates (violent and nonviolent offenses) were significantly higher in cases compared with controls and highest in those with substance-related psychoses. Acts intended to cause injury (23% in both cases and controls), offenses against justice or government procedures (i.e., breach of custodial order offenses/community-based orders/violence and nonviolence orders, offenses against government operations/security/justice procedures; 17% and 15% in cases and controls, respectively), theft (15% and 12% in cases and controls respectively), public order offenses (disorderly conduct, regulated public order offenses, and offensive conduct; 10% and 9% in cases and controls respectively), and drug offenses (11% in both cases and controls) were the most common offense types.

The distribution of justice outcomes was similar between cases and controls, with the majority receiving nonsupervised orders (64% and 69% in cases and controls, respectively; Table 3). However, the percentage of those receiving a nonsupervised order was lowest in those with substance-related psychoses (59% vs. 66% and 70%) with a higher proportion receiving a custodial sentence (20% vs. 17% and 11%) compared with those with schizophrenia and related psychoses and affective psychoses. Among cases and controls with criminal convictions, more than 70% of those with substance-related psychoses had more than one criminal conviction, whereas 59% of those with no diagnosis of psychosis had only one conviction (Figure 1).

Figure 1:

Frequency of Offenses by Cases (by Type of Psychosis) and Controls, July 2001–June 2015

Factors Associated With Criminal Convictions

Overall, those with a diagnosis of psychosis were approximately 5 times more likely than those without a diagnosis of psychosis to have a criminal conviction (adjusted odds ratio [aOR] = 4.68, 95% CI = [4.55, 4.81], p < .001; Table 4). In the gender-stratified analysis, men and women with a diagnosis of psychosis were also more likely to have a conviction compared with men and women without a diagnosis of psychosis, respectively (aOR = 4.69, 95% CI = [4.54, 4.84], p < .001 and aOR = 6.88, 95% CI = [6.45, 7.31], p < .001 for men and women, respectively). Other factors significantly associated with an increased odds of a conviction were all types of psychosis compared with no diagnosis of psychosis, age 20 to 39 years compared with younger age (<20 years), not living in a married or de facto relationship, being Aboriginal, and living in a socioeconomically disadvantaged area. In the gender-stratified analysis, the results are similar to the overall findings except the nonsignificant result for women aged between 20 and 39 years. In both men and women, those aged 40 years and above were less likely to have a criminal conviction than younger people (<20 years of age).

Table 4:

The aOR for Criminal Convictions (at Least One Criminal Conviction Between July 2001 and June 2015 in NSW)

	Overall(n = 259,383)			Men(n = 144,618; 56%)			Women(n = 114,765) (44%)
Characteristic	n (%)	aOR[95% CI]	P	n (%)	aOR[95% CI]	p	n (%)	aOR[95% CI]	p
Psychosis
No	172,922 (67)	1		96,425 (67)	1		76,497 (67)	1
Yes	86,461 (33)	4.68 [4.55, 4.81]	<.001	48,193 (33)	4.69 [4.54, 4.84]	<.001	38,268 (33)	6.88 [6.45, 7.31]	<.001
Psychosis types
No psychosis	172,922 (67)	1		96,425 (67)	1		76,497 (67)	1
Schizophrenia and related psychoses	65,895 (25)	4.26 [4.13, 4.38]	<.001	36,712 (25)	4.26 [4.12, 4.41]	<.001	29,183 (25)	6.16 [5.77, 6.58]	<.001
Affective psychoses	10,218 (4)	2.49 [2.35, 2.65]	<.001	4,361 (3)	2.89 [2.68, 3.12]	<.001	5,857 (5)	3.86 [3.46, 4.29]	<.001
Substance-related psychoses	10,348 (4)	9.77 [9.32, 10.24]	<.001	7,120 (5)	8.24 [7.79, 8.71]	<.001	3,228 (3)	19.21 [17.42, 21.18]	<.001
Age groups at recent diagnosis
<20 years	12,571 (5)	1		7,491 (5)	1		5,080 (4)	1
20−29 years	47,773 (18)	1.21 [1.15, 1.28]	<.001	31,192 (22)	1.20 [1.12, 1.28]	<.001	16,581 (15)	1.02 [0.92, 1.13]	.677
30−39 years	57,804 (22)	1.18 [1.12, 1.25]	<.001	35,835 (25)	1.22 [1.14, 1.30]	<.001	21,969 (19)	0.93 [0.84, 1.03]	.154
40+ years	141,235 (55)	0.37 [0.35, 0.39]	<.001	70,100 (48)	0.45 [0.42, 0.48]	<.001	71,135 (62)	0.26 [0.23, 0.28]	<.001
Aboriginal
No	212,112 (82)	1		119,639 (83)	1		92,473 (80)	1
Yes	7,939 (3)	5.02 [4.76, 5.28]	<.001	4,884 (3)	4.97 [4.65, 5.33]	<.001	3,005 (3)	6.12 [5.61, 6.68]	<.001
Missing/unknown	39,332 (15)	0.07 [0.07, 0.08]	<.001	20,095 (14)	0.10 [0.09, 0.11]	<.001	19,237 (17)	0.04 [0.03, 0.06]	<.001
Marital status
Married (including de facto)	79,582 (31)	1		41,229 (28)	1		38,353 (33)	1
Single	100,365 (39)	2.05 [1.98, 2.12]	<.001	55,741 (39)	1.95 [1.87, 2.04]	<.001	44,624 (39)	1.76 [1.65, 1.88]	<.001
Missing/unknown	79,436 (30)	2.32 [2.23, 2.42]	<.001	47,648 (33)	1.93 [1.84, 2.02]	<.001	31,788 (28)	2.16 [1.99, 2.35]	<.001
SEIFA
Advantaged	112,178 (43)	1		60,704 (42)	1		51,474 (45)	1
Disadvantaged	116,586 (45)	1.55 [1.51, 1.59]	<.001	64,380 (45)	1.58 [1.53, 1.63]	<.001	52,206 (45)	1.48 [1.40, 1.56]	<.001
Missing/unknown	30,619 (12)	0.98 [0.93, 1.04]	.521	19,534 (13)	0.91 [0.86, 0.97]	.003	11,085 (10)	0.93 [0.82, 1.05]	.236

Note. The results were adjusted by age, marital status, Aboriginality, and SEIFA. Here, in the variable psychosis types, no psychosis means control group and other categories were based on the diagnostic “hierarchy” explained earlier. Marital status and SEIFA were determined at the episode care of the most recent diagnosis. aOR = adjusted odds ratio; NSW = New South Wales; CI = confidence interval; SEIFA = socioeconomic indexes for areas.

A total of 21,270 individuals in the study population had a conviction for a violent offense, of whom 16,114 (76%) had a diagnosis of psychosis (Table 1). Individuals with psychosis were more likely to have a criminal conviction for violence (aOR = 4.90, 95% CI = [4.73, 5.07], p < .001) than those who had no psychosis (Table 5). Of the cases with a conviction for violence, 74% had a diagnosis of schizophrenia and related psychoses, 20% had a diagnosis of substance-related psychoses, and 6% had a diagnosis of affective psychoses. In the gender-stratified analysis, men and women with a diagnosis of psychosis were also more likely to have a conviction for violence compared with men and women without a diagnosis of psychosis, respectively (aOR = 4.68, 95% CI = [4.50, 4.87], p < .001 and aOR = 8.80, 95% CI = [8.01, 9.66], p < .001 for men and women, respectively).

Table 5:

The aOR for Convictions for Violence (at Least One Violent Conviction Between July 2001 and June 2015 in New South Wales)

	Overall(n = 259,383)			Men(n = 144,618; 56%)			Women(n = 114,765; 44%)
Characteristic	n (%)	aOR[95% CI]	p	n (%)	aOR[95% CI]	p	n (%)	aOR[95% CI]	p
Psychosis
No	172,922 (67)	1		96,425 (67)	1		76,497 (67)	1
Yes	86,461 (33)	4.90 [4.73, 5.07]	<.001	48,193 (33)	4.68 [4.50, 4.87]	<.001	38,268 (33)	8.80 [8.01, −9.66]	<.001
Psychosis types
No Psychosis	172,922 (67)	1		96,425 (67)	1		76,497 (67)	1
Schizophrenia and related psychoses	65,895 (25)	4.60 [4.43, 4.77]	<.001	36,712 (25)	4.39 [4.21, 4.58]	<.001	29,183 (25)	8.22 [7.46, −9.06]	<.001
Affective psychoses	10,218 (4)	2.72 [2.52, 2.93]	<.001	4,361 (3)	3.05 [2.78, 3.34]	<.001	5,857 (5)	5.03 [4.34, 5.83]	<.001
Substance-related psychoses	10,348 (4)	8.15 [7.72, 8.60]	<.001	7,120 (5)	6.83 [6.42, 7.26]	<.001	3,228 (3)	18.51 [16.28, 21.04]	<.001
Age groups at recent diagnosis
<20 years	12,571 (5)	1		7,491 (5)	1		5,080 (4)	1
20−29 years	47,773 (18)	1.25 [1.17, 1.34]	<.001	31,192 (22)	1.27 [1.17, 1.37]	< .001	16,581 (15)	0.97 [0.86, 1.10]	.678
30−39 years	57,804 (22)	1.23 [1.15, 1.31]	<.001	35,835 (25)	1.29 [1.20, 1.40]	< .001	21,969 (19)	0.88 [0.78, 1.00]	.049
40+ years	141,235 (55)	0.39 [0.37, 0.42]	<.001	70,100 (48)	0.50 [0.46, 0.54]	< .001	71,135 (62)	0.24 [0.22, 0.28]	< .001
Aboriginal
No	212,112 (82)	1		119,639 (83)	1		92,473 (80)	1
Yes	7,939 (3)	4.60 [4.36, 4.84]	<.001	4,884 (3)	4.27 [4.00, 4.56]	<.001	3,005 (3)	6.24 [5.68, 6.85]	<.001
Missing/unknown	39,332 (15)	0.07 [0.06, 0.08]	<.001	20,095 (14)	0.10 [0.08, 0.12]	<.001	19,237 (17)	0.03 [0.01, 0.06]	<.001
Marital status
Married	79,582 (31)	1		41,229 (28)	1		38,353 (33)	1
Single	100,365 (39)	1.79 [1.72, 1.87]	<.001	55,741 (39)	1.61 [1.53, 1.70]	<.001	44,624 (39)	1.67 [1.53, 1.83]	<.001
Missing/unknown	79,436 (30)	1.92 [1.82, 2.02]	<.001	47,648 (33)	1.53 [1.44, 1.62]	<.001	31,788 (28)	1.97 [1.76, 2.20]	<.001
SEIFA
Advantaged	112,178 (43)	1		60,704 (42)	1		51,474 (45)	1
Disadvantaged	116,586 (45)	1.56 [1.52, 1.63]	<.001	64,380 (45)	1.59 [1.53, 1.65]	<.001	52,206 (45)	1.48 [1.38, 1.59]	<.001
Missing/unknown	30,619 (12)	0.95 [0.89, 1.02]	.132	19,534 (13)	0.88 [0.81, 0.94]	<.001	11,085 (10)	0.97 [0.83, 1.13]	.686

Note. The results were adjusted by age, marital status, Aboriginality, and SEIFA. Here, in the variable psychosis types, no psychosis means control group and other categories were based on the diagnostic “hierarchy” explained earlier. Marital status and SEIFA were determined at the episode care of the most recent diagnosis. aOR = adjusted odds ratio; CI = confidence interval; SEIFA = socioeconomic indexes for areas.

A similar analysis was conducted separately for nonviolent offenses (see Supplemental Table S1, available in the online version of this article). The results were broadly consistent with the overall analysis presented in Table 4. In addition, in the adjusted analysis, individuals who had at least one diagnosis of substance-related psychoses were also more likely to have a criminal conviction during 2001–2015 (aOR = 12.47, 95% CI = [12.00, 13.00], p < .001; data not shown), reinforcing the well-established association between drug use and criminal conviction.

Discussion

The results from this population-based case-control data-linkage study suggest that offending, as marked by a proven criminal conviction, is more common in those with a diagnosis of a psychotic disorder compared with age- and sex-matched controls with no such diagnosis. Individuals with psychosis were approximately 5 times more likely to have a conviction for a violent crime than the controls. Whereas many studies have focused on the relationship between violence and psychosis (Douglas et al., 2009; Fazel et al., 2009), we examined all types of offending, which provides a basis for further investigation into the complex relationship between psychosis and offense types. Our study has found that convictions for nonviolent offenses were also higher in those with psychosis compared with controls.

Consistent with previous research, factors significantly associated with an increased odds of a criminal conviction were the types of psychosis (substance-related psychoses having the strongest association), age (20 years or older), being Aboriginal, living in a disadvantaged area, and not being married or a de facto relationship. Overall, psychosis cases accounted for 10% of all criminal convictions (violent and nonviolent) in NSW between 2001 and 2015. These results suggest that population-level interventions and measures targeting the health and criminogenic needs of this particular group are likely to have a significant impact on crime in the state. In another study, we also showed that court diversion under the NSW Mental Health Act for those with psychosis was associated with reduced reoffending and, importantly, increased treatment was associated with reduced reoffending (Albalawi et al., 2019). Together, these findings suggest the need for better policies that manage this group to improve justice outcomes.

The median age at first criminal conviction (Mdn = 30 years, IQR = 22–39 years) for those with psychosis preceded the median age at first diagnosis of psychosis (Mdn = 31 years, IQR = 24–40 years). This is consistent with a Western Australian study where the first arrest in those with a psychiatric illness preceded the first contact with mental health services (Morgan et al., 2013). Approximately, 20% of people with psychosis (and 18% of people with schizophrenia and related psychoses) in our study had a conviction for a violent offense, which is higher than that reported in the Western Australian study of people with schizophrenia (12.9%; Morgan et al., 2013).

We found that the median age at first diagnosis for people with schizophrenia and related psychoses in our data was 41 years (IQR = 29–58 years). This might reflect the difficulties in detecting the early phases of schizophrenia due to confounding factors. For example, earlier diagnosis of substance-use disorders and antisocial behavior can be attributed to substance use alone and antisocial personality disorder, and minimization of symptoms by patients during initial contacts with mental health services. Lack of detection due to avoidance of contact with mental health services by the patients might also be another reason behind this.

An increased frequency of reoffending was observed in those with substance-related psychoses who were more likely to have been convicted for offenses. These results reinforce the well-established link between drug use and offending, most likely as a result of committing acquisitive crimes to fund problem drug use (Morgan et al., 2013). As we used the diagnostic hierarchical approach to categorize diagnoses, this suggests that substance-related psychoses alone make a significant contribution to criminal convictions. However, we found a lower prevalence of violent convictions in this group compared with those with schizophrenia and affective psychoses.

According to the Australian Bureau of Statistics, there were 216,176 Aboriginal people in NSW, equating to around 3% of the total NSW population (Australian Bureau of Statistics, 2016). However, 6% of those with psychosis in our study were Aboriginal. In the adjusted analysis, we found that Aboriginal people were more likely to be convicted of a criminal offense than non-Aboriginal people. These findings raise questions about treatment options and the lack of engagement of Aboriginal people by mental health services. This suggests that mental health services may need to be more culturally appropriate for this group to ensure they can receive equitable treatment. In another study, we identified that Aboriginal people were also less likely to be diverted into treatment by the courts than non-Aboriginal people (Albalawi et al., 2019).

Population-based studies such as ours are important to provide policy makers and planners with robust information that can be used as a basis for policy development to better manage those with serious mental illness and prevent them from coming into contact with the justice system. Similarly, it provides advocates, consumers, and the community with the necessary information to influence policy formulation in this sensitive area to improve outcomes for this group. These results can help to inform our understanding and insights about the extent of the problem, pathways leading to contact with the criminal justice system, promote a better understanding by the community of the needs of those with serious mental illness, and ultimately contribute to community safety.

Limitations of this study include having no data on certain known criminogenic risk factors such as substance use, antisocial personality disorder, criminal history preceding the year 2001, and employment status. The present analysis did not determine the temporal sequence for diagnosis and subsequent offending behavior. We did not compare individuals with psychosis with a matched sample of those with other mental health diagnoses, which could be the subject of a further linkage study. We compared those with psychosis with a random sample of age- and sex-matched individuals from a statewide data source comprising individuals who had interactions with health services for conditions other than psychosis. This study did not include those diagnosed exclusively in private clinics or treated by general practitioners. However, most mentally ill persons with psychosis are treated at some point by the public mental health services, so this number is likely to be small. Data for admitted patients were available from 2001 onward, and those presenting to the emergency departments from 2005, respectively, so diagnoses prior to this time were not available. We included records only from NSW, so those diagnosed with psychosis or convicted of a crime interstate or overseas would have been missed. We analyzed offending data for the period between July 2001 and June 2015 (latest available records), which creates the possibility that some controls may have developed psychosis between January 2013 and June 2015.

Criminogenic factors such as those listed above can be argued to increase the likelihood of more proximal and dynamic factors contributing to an offending event by creating context and contributing to the motivation to offend. Future research could include larger life domain areas covering substance use, unemployment, and other mental disorders. Future research should also focus on the assessment of mental health treatment at the community level and explore more effective treatment options with the aim of reducing future criminal behavior in people with serious mental illness.

Conclusion

The compelling results from this study confirm previous findings that individuals with psychosis are at an increased risk of criminal convictions, including those for violent crimes, and this is no different in NSW. This study suggests that, prima facie, psychosis appears to have a strong relationship with offending, including violent offending, which highlights the need for early diagnosis and treatment initiation in this group, which could potentially attenuate the risk of offending. However, this association needs to be taken in the context that serious mental illness is a factor likely to increase contact with the justice system through homelessness, unemployment, so-called public nuisance behaviors, and substance use. Efforts to reduce offending and prison numbers need to adopt interventions and strategies at all stages of the justice system, including apprehension by the police, probation and parole, and in custody, to effectively manage this population. Similarly, culturally appropriate services need to be in place for Aboriginal people. To successfully intervene and reduce offending in those with psychosis, it is likely that a multiagency approach involving health, welfare, housing, and justice is required to address both health and criminogenic needs.

Supplemental Material

supplemental_table – Supplemental material for Psychosis and Criminal Offending: A Population-Based Data-Linkage Study

Supplemental material, supplemental_table for Psychosis and Criminal Offending: A Population-Based Data-Linkage Study by Nabila Z. Chowdhury, Olayan Albalawi, Handan Wand, Stephen Allnutt, Armita Adily, Azar Kariminia, Grant Sara, Peter W. Schofield, Colman O’Driscoll, David M. Greenberg and Tony Butler in Criminal Justice and Behavior

Footnotes

Authors’ Note:

We have no conflict of interest to disclose.

ORCID iDs

Nabila Z. Chowdhury

Olayan Albalawi

Supplemental Material

Supplemental Table S1 is available in the online version of this article at

Nabila Z. Chowdhury is a PhD student at The Kirby Institute, University of New South Wales. She is working on the research project, The relationship between psychoses and offending in New South Wales—A data-linkage study: Implications for mental health policy development.

Olayan Albalawi has a PhD in biostatistics from the University of New South Wales. He is a lecturer in the statistics department, Faculty of Science, at Tabuk University in Saudi Arabia.

Handan Wand is a senior biostatistician. Besides her role in randomized trials, and large epidemiological national and international cohorts, she has led research in the development and application of new biostatistical methodologies.

Stephen Allnutt is a forensic psychiatrist and conjoint senior lecturer at the University of New South Wales. He has spent his career working in public forensic mental health services in the role of clinical director in Australia and is currently in private practice in Sydney. He has extensive experience and expertise in criminal forensic mental health.

Armita Adily has experience in population health and health services research. She is particularly interested in evidence-based policy and practice and knowledge translation. She has been involved in a number of needs assessments, workforce development planning, and continuous quality improvement programs. She is a coinvestigator of a number of studies in the areas of offenders’ health and violence at the School of Population Health, University of New South Wales.

Azar Kariminia is a senior lecturer with the Kirby Institute at the University of New South Wales. She is an epidemiologist with interest and expertise in infectious disease epidemiology and the health of marginalized populations.

Grant Sara is the director of InforMH, New South Wales (NSW) Ministry of Health, and a clinical associate professor in the School of Psychiatry, Faculty of Medicine and Health, University of Sydney.

Peter W. Schofield is a behavioral neurologist with training in epidemiology and is clinical director of the Hunter New England Local Health District Neuropsychiatry Service in New South Wales, Australia. His research interests include neuropsychiatric problems in offender populations and cognitive and dementing disorders in the general public.

Colman O’Driscoll is trained as a mental health nurse and has over 20 years’ experience across the health system, including senior leadership roles in public and private healthcare. Currently he is the chief Executive of St John of God Health Care’s NSW mental health services and leads mental health for the organisation nationally. His main area of academic interest is forensic mental health.

David M. Greenberg is currently a professor of forensic psychiatry, conjoint appointment, at the University of New South Wales. He established the NSW Statewide Court Liaison Service (Justice Health Forensic Mental Health Service), which won the NSW Premier’s Public Sector Gold Award (Social Justice Section) in 2003.

Tony Butler is program head of the Justice Health Research Program, School of Population Health, University of New South Wales. He has more than 20 years’ experience undertaking research into prisoner and offender health, covering infectious diseases, mental health, tobacco smoking, mortality, violence, traumatic brain injury, and substance use.

References

Albalawi

Chowdhury

N. Z.

Wand

Allnutt

Greenberg

Adily

Kariminia

Schofield

Sara

Hanson

O’Driscoll

Butler

(2019). Court diversion for those with psychosis and its impact on re-offending rates: Results from a longitudinal data-linkage study. British Journal of Psychiatry Open, 5(1), Article e9. https://doi.org/10.1192/bjo.2018.71

Australian Bureau of Statistics. (2011). Australian and New Zealand Standard Offence Classification (ANZSOC). https://www.abs.gov.au/ausstats/abs@.nsf/mf/1234.0

Australian Bureau of Statistics. (2016). Census of population and housing: Reflecting Australia—Stories from the census, 2016. https://www.abs.gov.au/ausstats/abs@.nsf/mf/2071.0

Australian Bureau of Statistics. (2019). Prisoners in Australia 2019. https://www.abs.gov.au/ausstats/abs@.nsf/mf/4517.0

Bennett

D. J.

Ogloff

J. R. P.

Mullen

P. E.

Thomas

S. D. M.

Wallace

Short

(2011). Schizophrenia disorders, substance abuse and prior offending in a sequential series of 435 homicides. Acta Psychiatrica Scandinavica, 124(3), 226–233. https://doi.org/10.1111/j.1600-0447.2011.01731.x

Bonta

Blais

Wilson

H. A.

(2014). A theoretically informed meta-analysis of the risk for general and violent recidivism for mentally disordered offenders. Aggression and Violent Behavior, 19(3), 278–287. https://doi.org/10.1016/j.avb.2014.04.014

Borthwick

(2003). Key concepts in the ChoiceMaker 2 record matching system. CiteSeerX.

Douglas

K. S.

Guy

L. S.

Hart

S. D.

(2009). Psychosis as a risk factor for violence to others: A meta-analysis. Psychological Bulletin, 135(5), 679–706. https://doi.org/10.1037/a0016311

Fazel

Gulati

Linsell

Geddes

J. R.

Grann

(2009). Schizophrenia and violence: Systematic review and meta-analysis. PLoS Medicine, 6(8), Article e1000120. https://doi.org/10.1371/journal.pmed.1000120

10.

Fazel

(2011). Psychotic disorders and repeat offending: Systematic review and meta-analysis. Schizophrenia Bulletin, 37(4), 800–810. https://doi.org/10.1093/schbul/sbp135

11.

Igoumenou

Kallis

Coid

(2015). Treatment of psychosis in prisons and violent recidivism. British Journal of Psychiatry Open, 1(2), 149–157. https://doi.org/10.1192/bjpo.bp.115.000257

12.

Morgan

V. A.

Morgan

Valuri

Ferrante

Castle

Jablensky

(2013). A whole-of-population study of the prevalence and patterns of criminal offending in people with schizophrenia and other mental illness. Psychological Medicine, 43(9), 1869–1880. https://doi.org/10.1017/s0033291712002887

13.

Mullen

E. P.

(2006). Schizophrenia and violence: From correlations to preventive strategies (01/02 ed., Vol. 12). Cambridge University Press.

14.

Nielssen

O. B.

Westmore

B. D.

Large

M. M.

Hayes

R. A.

(2007). Homicide during psychotic illness in New South Wales between 1993 and 2002. Medical Journal of Australia, 186(6), 301–304.

15.

Peterson

J. K.

Skeem

Kennealy

Bray

Zvonkovic

(2014). How often and how consistently do symptoms directly precede criminal behavior among offenders with mental illness? Law and Human Behavior, 38(5), 439–449. https://doi.org/10.1037/lhb0000075

16.

Schanda

Knecht

Schreinzer

Stompe

Ortwein-Swoboda

Waldhoer

(2004). Homicide and major mental disorders: A 25-year study. Acta Psychiatrica Scandinavica, 110(2), 98–107. https://doi.org/10.1111/j.1600-0047.2004.00305.x

17.

Shaw

Hunt

I. M.

Flynn

Meehan

Robinson

Bickley

Parsons

McCann

Burns

Amos

Kapur

Appleby

(2006). Rates of mental disorder in people convicted of homicide. National clinical survey. The British Journal of Psychiatry, 188, 143–147. https://doi.org/10.1192/bjp.188.2.143

18.

Short

Thomas

Mullen

Ogloff

J. R.

(2013). Comparing violence in schizophrenia patients with and without comorbid substance-use disorders to community controls. Acta Psychiatrica Scandinavica, 128(4), 306–313. https://doi.org/10.1111/acps.12066

19.

Simpson

A. I.

McKenna

Moskowitz

Skipworth

Barry-Walsh

(2004). Homicide and mental illness in New Zealand, 1970-2000. The British Journal of Psychiatry, 185, 394–398. https://doi.org/10.1192/bjp.185.5.394

20.

Skeem

Kennealy

Monahan

Peterson

Appelbaum

(2016). Psychosis uncommonly and inconsistently precedes violence among high-risk individuals. Clinical Psychological Science, 4(1), 40–49. https://doi.org/10.1177/2167702615575879

21.

Swanson

J. W.

McGinty

E. E.

Fazel

Mays

V. M.

(2015). Mental illness and reduction of gun violence and suicide: Bringing epidemiologic research to policy. Annals of Epidemiology, 25(5), 366–376. https://doi.org/10.1016/j.annepidem.2014.03.004

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.11 MB