eTable 1. Mental Disorder Diagnostic Categories (ICD-10 and ICD-8 Codes)
eTable 2. Incidence of First Subjection to Crime(Violent Offense) by Covariates, Among Those With and Without Mental Disorder
eTable 3. Incidence Rate Ratio of First Subjection to Crime (Any Crime and Violent Offense) With Adjustment by Covariate Combinations, Among Those With Mental Disorder Compared to Those Without Mental Disorder
eTable 4. Incidence Rate Ratios (IRR) for First Subjection to Crime, by Type of Crime, for Those With and Without Any Mental Disorder (Men)
eTable 5. Incidence Rate Ratios (IRR) for First Subjection to Crime, by Type of Crime, for Those With and Without Any Mental Disorder (Women)
Customize your JAMA Network experience by selecting one or more topics from the list below.
Dean K, Laursen TM, Pedersen CB, Webb RT, Mortensen PB, Agerbo E. Risk of Being Subjected to Crime, Including Violent Crime, After Onset of Mental Illness: A Danish National Registry Study Using Police Data. JAMA Psychiatry. 2018;75(7):689–696. doi:10.1001/jamapsychiatry.2018.0534
What is the incidence of police-reported experiences of being subjected to crime, including violent crime, after onset of mental illness, and do persons with specific mental disorders have increased risk of being subjected to crime compared with those without mental illness?
In this national cohort study of more than 2 million individuals, the incidence of experience of any crime and violent crime was increased among those with mental illness. The association was seen across the diagnostic spectrum in both men and women, with the strongest associations found for those with substance use disorders and personality disorders.
Mental illness across the diagnostic spectrum is associated with increased risk of police-reported experiences of being subjected to violent and nonviolent crime.
People with mental illness are more likely to have contact with the criminal justice system, but research to date has focused on risk of offense perpetration, while less is known about risk of being subjected to crime and violence.
To establish the incidence of being subjected to all types of criminal offenses, and by violent crimes separately, after onset of mental illness across the full diagnostic spectrum compared with those in the population without mental illness.
Design, Setting, and Participants
This investigation was a longitudinal national cohort study using register data in Denmark. Participants were a cohort of more than 2 million persons born between 1965 and 1998 and followed up from 2001 or from their 15th birthday until December 31, 2013. Analysis was undertaken from November 2016 until February 2018.
Cohort members were followed up for onset of mental illness, recorded as first contact with outpatient or inpatient mental health services. Diagnoses across the full spectrum of psychiatric diagnoses were considered separately for men and women.
Main Outcomes and Measures
Incidence rate ratios (IRRs) were estimated for first subjection to crime event (any crime and violent crime) reported to police after onset of mental illness. The IRRs were adjusted for cohort member’s own criminal offending, in addition to several sociodemographic factors.
In a total cohort of 2 058 063 (48.7% male; 51.3% female), the adjusted IRRs for being subjected to crime associated with any mental disorder were 1.49 (95% CI, 1.46-1.51) for men and 1.64 (95% CI, 1.61-1.66) for women. The IRRs were higher for being subjected to violent crime at 1.76 (95% CI, 1.72-1.80) for men and 2.72 (95% CI, 2.65-2.79) for women. The strongest associations were for persons diagnosed as having substance use disorders and personality disorders, but significant risk elevations were found across almost all diagnostic groups examined.
Conclusions and Relevance
Onset of mental illness is associated with increased risk of exposure to crime, and violent crime in particular. Elevated risk is not confined to specific diagnostic groups. Women with mental illness are especially vulnerable to being subjected to crime. Individual’s own offending accounts for some but not all of the increased vulnerability to being subjected to crime.
People who experience mental illness are more likely to come into contact with the criminal justice system. Research to date has focused more on the association between mental illness and an elevated risk of crime perpetration1,2 than on the heightened vulnerability to being subjected to nonviolent or violent crime. However, evidence has emerged to indicate that risk of being subjected to crime may be at least as great, if not greater, than the risk of crime perpetration3 among persons with mental illness. Studies4-6 of individuals with severe mental illnesses, such as schizophrenia, have reported strikingly high rates of self-reported experiences of being subjected to crime. A survey7 of individuals in contact with secondary mental health services in London, United Kingdom, found rates as high as 40% for self-reported past-year experience of crime with any offense and 19% for past-year violent crime specifically compared with 14% and 3%, respectively, for a control sample obtained from a contemporaneous national crime survey. While increased risk of being subjected to crime among those with mental illness is in itself worthy of a preventive focus, its importance is further justified by evidence that being subjected to crime may be associated with a range of other adverse outcomes, including poor symptomatic and functional mental health recovery.8 Being subjected to crime has also long been acknowledged to have an important etiological role in the development of mental illness, and particularly so early-life trauma and abuse,9 but less attention has been paid to the potential for mental illness to increase vulnerability to such violence.
To date, the association between mental illness and risk of being subjected to crime has been examined mainly in cross-sectional surveys of selected samples of individuals with severe mental illnesses.4 Less is known about how risk might vary across the psychiatric diagnostic spectrum, specifically after onset of mental illness, or at a population level. Such an approach was previously undertaken to examine the association between mental illness and crime perpetration and found that risk after mental illness onset extended across the psychiatric diagnostic spectrum.2 While being exposed to crime, and violent crime in particular, is increasingly regarded as an important but largely neglected public health problem,10 the incidence of crime experiences by those with mental illness compared with the general population without such illness has not yet been established. Therefore, robust evidence is lacking to inform the development of preventive strategies, including initiatives aimed at improving the experience of persons with mental illness who report being subjected to crime and subsequently seek justice.11
In this national cohort study, we examined for the first time to date in a population registry the incidence of being subjected to crime by all types of criminal offenses, and by violent crimes separately, after onset of mental illness across the full diagnostic spectrum. We further explored the potential influence of offense perpetration given the known overlap between those who are exposed to crime and those who perpetrate crime and violence,12 as well as the putative role of mental illness in determining the extent of that overlap.13
This national cohort study (N = 2 058 063) consisted of all persons born in Denmark between 1965 and 1998 who were alive at their 15th birthday (having excluded 650 individuals who could not be linked to their mother in the Civil Registration System). The Civil Registration System14 contains the personal identification number, sex, date and place of birth, continuously updated vital status, and parents’ personal identification numbers. Each resident is assigned a unique personal identification number at birth or at point of first address in Denmark, through which it is possible to link information within and between registers.
The study was approved by the Danish Data Protection Agency. According to Danish law, informed consent is not required for register-based studies.
Data on individuals exposed to reported criminal offenses in Denmark were extracted from the Administrative System of the National Police, available from 2001 onward and including data on all offenses reported to police, including those not pursued after report. The events coded by the Danish police force represented herein as “being subjected to crime” are reported by individuals as having been a “victim of crime” in the terminology of the actual reports. The dependent variables considered were first subjection to crime event (including all registered criminal offenses) and first subjection to violent crime (ie, all violent crimes, including physical assault, aggravated acquisitive crimes, violent threats, and sexual offenses). The 5 most common types of being subjected to crime—thefts, simple violence, threats, robberies, and severe violence—were examined separately in relation to the presence or absence of any mental disorder.
Information on mental illness was obtained from the Psychiatric Central Research Register,15 which contains data relating to all admissions to psychiatric hospitals since 1969 and all outpatient contacts and emergency department visits since 1995. Diagnoses were assigned based on International Classification of Diseases (ICD) coding according to the eighth revision (ICD-8)16 until 1993 and the 10th revision (ICD-10)17 from 1994 onward. We classified diagnoses into 8 groups based on the broad ICD-10 categories (eTable 1 in the Supplement).18 For each mental disorder, date of onset was defined as the first day of the first psychiatric contact (inpatient, outpatient, or psychiatric emergency care unit) for the diagnosis of interest. Two approaches were taken to considering the association between diagnostic category and being subjected to crime. First, each of the 8 categories was included alone in separate analyses (the single-diagnosis model). Second, mutually exclusive categories were created according to the hierarchical logic of the ICD-10 classification system. Individuals with psychiatric contacts belonging to more than one diagnostic category were allowed to move upward in the hierarchy as they accumulated new diagnoses over time but not downward.
From the National Crime Register, we extracted information on penal code violations.19 Only guilty verdicts resulting in custodial sentences, suspended sentences, conditional withdrawal of charges, fines, and sentences to psychiatric treatment were included.
To assess the potential confounding influence of parental socioeconomic status, we used information on paternal income (in quartiles plus an “unknown” category) and highest level of maternal education (coded as primary, secondary, or tertiary) at each cohort member’s 15th birthday. These covariate data were extracted from the Integrated Database for Labour Market Research, containing information from the 1970 Population and Housing Census20 and annually updated information from 1980 onward.21 We also included the following covariates in all analyses: age in 1-year bands, calendar year in 1-year bands, and unknown paternal identity.
All cohort members were followed up from 2001 or from their 15th birthday, whichever came last, until their first crime event, death, emigration, or the end of follow-up on December 31, 2013. Analysis was undertaken from November 2016 until February 2018. This longitudinal national cohort study was analyzed using Poisson regression with the GENMOD procedure (SAS, version 9.1.4; SAS Institute Inc). Poisson regression was used to conduct a time-to-event survival analysis, with the number of person-years at risk used as an offset variable to enable incidence rate ratio (IRR) estimation.22 We calculated the incidence rate as the number of first crime events per 1000 person-years at risk. From the Poisson regression models, IRRs were calculated for each diagnostic category group vs a reference category of no recorded mental disorder. The 95% CIs for these IRRs were estimated using maximum likelihood. We retained information on an individual’s first diagnosis within each group occurring before the date of first registered subjection to crime for any criminal offense. Psychiatric diagnostic category and cohort member’s own criminal offending were handled in the analyses as being time varying,23,24 while all other covariates were fixed at the start of follow-up. The unit of analysis for the study remained at the level of the individual for all analyses. Sex-specific analyses were conducted for both outcomes (any subjection to crime and violent crime), and models were then fitted for the ICD-10 hierarchical classification controlling for cohort member’s own criminal offending, along with other potential confounders. The incidence rates for any first subjection to crime and first subjection to violent crime were examined for each of the covariates among those with and without any mental disorder.
Variation in incidence rates across levels of each examined covariate was identified (Table 1 for any offense and eTable 2 in the Supplement for violent offenses). Among individuals in the cohort with any recorded mental disorder, rates of being subjected to crime were considerably higher, even after adjustment for sociodemographic factors, than those among individuals without mental disorder (IRR, 1.68; 95% CI, 1.65-1.71 for men [Table 2] and IRR, 1.71; 95% CI, 1.68-1.73 for women [Table 3]).
Considering each diagnostic category separately (the single-diagnosis model in Table 2 and Table 3), positive associations between specific categories of mental disorders and the incidence of subjection to crime were found across the psychiatric diagnostic spectrum for both men and women, with the exception of developmental disorders (where a negative association was found) and intellectual disability for men (for which no statistically significant association was found). For both sexes, the strongest associations were observed for substance use disorders (IRR, 2.61; 95% CI, 2.53-2.69 for men and IRR, 3.18; 95% CI, 3.06-3.32 for women) and personality disorders (IRR, 2.23; 95% CI, 2.15-2.32 for men and IRR, 2.00; 95% CI, 1.95-2.06 for women). When the ICD-10 hierarchical classification was fitted, the pattern was essentially unaltered for both sexes.
Although a similar pattern of results was observed when subjection to violent crime was considered specifically, the magnitude of the associations was considerably greater, particularly among women (Table 4 and Table 5). The adjusted IRRs for being subjected to violent crime among those with any mental disorder were 2.10 (95% CI, 2.05-2.15) for men and 2.99 (95% CI, 2.92-3.06) for women. Significant positive associations between having a specific category of mental disorder and the incidence of subjection to violent crime were found across the psychiatric diagnostic spectrum (the single-diagnosis model in Tables 4 and 5), with the only exception being developmental disorders for men. The strongest associations were again found to be for those with substance use disorders (IRR, 3.45; 95% CI, 3.33-3.58 for men and IRR, 7.04; 95% CI, 6.69-7.41 for women). When the ICD-10 hierarchical classification was fitted, the pattern was again essentially unchanged for both men and women.
Incidence rates for cohort member’s own criminal offending are listed in Table 1 and eTable 2 in the Supplement, along with rates for each of the covariates previously considered. The strength of the associations observed between mental disorders and being subjected to crime was attenuated by adjusting for cohort member’s own criminal offending (Tables 2, 3, 4, and 5), although the apparent confounding influence varied by diagnostic category. The reduction in strength of association for those with substance use disorders was greatest (20%-30% IRR reduction for any offense and approximately 35% IRR reduction for violent crime offense). The 2 sociodemographic covariates of paternal income and maternal education made little difference in attenuating the IRRs over and above the far stronger potential confounding influence of cohort member’s own criminal offending, as summarized in eTable 3 in the Supplement.
The 5 most common types of being subjected to crime obtained from police data were the following police data codes: thefts (code 1336xxx), simple violence (code 1252xxx), threats (code 1292xxx), robberies (code 1280xxx), and severe violence (code 1255xxx). Incidence rates were raised for each of these crime types among those with any mental disorder compared with those with no mental disorder, with stronger associations seen for women (eTable 4 and eTable 5 in the Supplement).
To our knowledge, this is the first study to systematically establish on a national basis the incidence of being subjected to crime with any offense, and being subjected to violent crime separately, after onset of mental illness across the full spectrum of psychiatric diagnoses. In this study of more than 2 million persons, we identified an elevated incidence of crime across multiple diagnostic categories compared with individuals without mental illness in the population. Incidence rate ratios adjusted for sociodemographic confounders were particularly high for those with substance use disorders and personality disorders, but they were also raised for individuals with other diagnoses, including severe mental illnesses. When the incidence of being subjected to violent crime was examined specifically, similar patterns of elevated risk were found, but the magnitude of the associations observed was consistently greater. The strongest associations overall were found for women with substance use disorders: incidence rates of being subjected to violent crime were 7 times higher than for women without mental illness.
While previous studies have typically reported prevalence rather than incidence estimates and have focused on severe mental illnesses, the pattern of our findings is broadly in line with the literature,3-7 including one of the few previous studies using population linkage to police data.25 Our findings augment the existing literature by demonstrating that the heightened vulnerability to being subjected to crime with any offense, and violent crime in particular, extends to those with a wide range of mental illnesses, is true of those reporting to police, occurs after mental illness onset, and is not confined to individuals with mental disorders in the population who are treated as inpatients. Our results regarding the association between mental illness onset and subsequent subjection to crime are likely to be conservative given that individuals who are subjected to crime before mental illness onset will have been censored from the analysis.
In the present study, relative risks were consistently higher for women with mental illnesses than for men with mental illnesses, particularly for being subjected to violent crime, a finding that has been noted in previous studies,26 at least in the context of severe mental illnesses. In the general population, with the exception of specific offenses (eg, sexual crimes and domestic abuse or violence), women are less likely to be exposed to crime than men.27 Whatever ordinarily protects some women from being subjected to crime appears to be eroded by the presence of mental illness. The reasons for this are likely to be multifactorial,26 but it is reasonable to hypothesize that causal mechanisms arising directly from the onset and influence of mental illness may be particularly relevant for women.
We also found evidence that cohort member’s own criminal offending explained some of the elevated risk of being subjected to crime observed after mental illness onset, even after further adjustment was made for a range of covariates likely to be associated with both being subjected to crime and offense perpetration. The degree of reduction in strength of association appeared to differ by diagnosis (ie, it was greatest for those with substance use disorder) and was slightly greater for being subjected to violent crime, as well as among men. Criminal offending is unlikely to be a straightforward confounder of the association between onset of mental illness and being subjected to crime but may instead lie on the causal pathway between the 2 phenomena (in either direction) for some individuals. Few previous studies have considered risk of offending and being subjected to crime among those with mental illness in the same study cohort,25,28 despite the overlap between perpetrators of and subjection to criminal activity being well established in the general population.12 The overlap between perpetrators of and individuals subjected to criminal activity likely arises in part from the presence of shared risk factors for the 2 outcomes among individuals with and without mental illness (eg, comorbid substance misuse problems,29,30 psychiatric symptoms,13 comorbid personality problems,31 and conflicted social relationships32).
The present study had several key strengths, including the large size of the national cohort, the long duration of follow-up, and the minimization of selection, attrition, and information biases commonly encountered in other studies, which are advantages arising from the use of interlinked national registers. The full spectrum of psychiatric diagnoses was examined, incidence of being subjected to crime after mental illness onset rather than prevalence alone was estimated (thereby reducing potential reverse causality bias), mental health data came from outpatient and emergency department and inpatient registers, and both violent and nonviolent crime experiences reported to police were examined. The investigation was also strengthened by its consideration of the dual risks of crime receipt and perpetration in the same study cohort.
However, this study had some important limitations. While systematic studies validating all of the diagnoses presented in this study are not available, many of the key diagnoses (eg, schizophrenia, dementia, affective disorders, depressive disorder, and childhood disorders) have been validated, with reassuring results.33,34 It should be noted that all diagnoses were made by a treating clinician and often based on a period of clinical observation rather than a single clinical or research interview. Reliance on such clinically determined diagnoses enables results to be more readily generalized to clinical settings, where structured diagnostic interviews are used infrequently. However, we were unable to identify individuals with mental disorders in the population either not receiving any treatment or receiving treatment only in primary or private care. For those disorder categories characterized by lower rates of outpatient, emergency, or inpatient contact (eg, anxiety disorders), our findings likely reflect risk of being subjected to crime for a subgroup who may have more severe disorder, comorbid problems, or other adversity that has contributed to their need for mental health care beyond primary care. Also, although the cohort included individuals aged up to 45 years, we could not examine the entire period of risk for onset of mental disorder, particularly for disorders with later onset, such as those in the organic disorders category. While we could assess the potential confounding or mediating influences of several sociodemographic factors and investigate the overlap between crime receipt and perpetration events in the same individuals, we were unable to elucidate the underlying mechanisms. A complex and multifactorial causal pathway is likely to explain the associations observed between mental illness and being subjected to crime operating at the following 3 levels: (1) individual (eg, specific symptoms, cognitive impairments, and comorbidities), (2) familial (eg, interpersonal conflict and isolation), and (3) neighborhood (eg, urban or rural and local crime density). A longitudinal model of causation has been previously proposed that considers the likely role of early-life and more recent factors.31
With regard to the generalizability of our findings, overall crime rates in Denmark, as documented by international subjection to crime surveys,35 are generally comparable to those of other industrialized countries, although rates of being subjected to violent crime may be lower than in some settings. The most recently reported annual prevalence of being subjected to violent crime in Denmark (3.3%) is close to the average rate (3.0%) but lower than in either the United States (4.3%) or the United Kingdom (5.4%), for example.35 Finally, it is important to note that reliance on official police records of crime receipt and perpetration ignores experiences and behaviors that are not reported, and there is some evidence that reporting rates may vary between countries, being higher in more affluent settings.35 Also, persons with mental illness may be more reluctant or less able to report crime experiences to police than those without mental illness.36
After onset of mental illness, individuals may experience a heightened vulnerability to being subjected to crime and violence. The focus to date in clinical practice and research on offending may have been at the expense of neglecting the risk of being subjected to crime, including violent crime. For example, risk assessment in clinical settings is dominated by consideration of risks of suicide and violence, while risk of crime receipt is largely ignored. Similarly, for those in contact with the criminal justice system, identification of mental health need and provision of support and treatment are offered almost solely to offenders. At a policy level, our findings have the potential to contribute to efforts to remedy public misconceptions about mental illness, often fueled by selective and pejorative media reporting, with the ultimate aim of reducing stigma. Our results highlight the need for further research to determine more fully why some people with mental illnesses are especially vulnerable to being subjected to crime (eg, those with substance use and personality disorders) and to develop effective interventions to reduce the elevated risk.
Accepted for Publication: February 19, 2018.
Corresponding Author: Kimberlie Dean, PhD, School of Psychiatry, University of New South Wales, Roundhouse, Long Bay Complex, PO Box 150, Matraville, New South Wales, Australia 2036 (email@example.com).
Published Online: May 23, 2018. doi:10.1001/jamapsychiatry.2018.0534
Author Contributions: Drs Laursen and Agerbo had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.
Study concept and design: Dean, Laursen, Pedersen, Agerbo.
Acquisition, analysis, or interpretation of data: All authors.
Drafting of the manuscript: Dean.
Critical revision of the manuscript for important intellectual content: All authors.
Statistical analysis: Laursen, Agerbo.
Obtained funding: Dean, Mortsensen.
Administrative, technical, or material support: Pedersen, Mortsensen, Agerbo.
Study supervision: Laursen, Pedersen.
Conflict of Interest Disclosures: None reported.
Funding/Support: This study was supported by a grant from Justice Health and Forensic Mental Health Network. Dr Dean is supported financially by Justice Health and Forensic Mental Health Network. Drs Laursen, Pedersen, Mortensen, and Agerbo are supported financially by The Stanley Medical Research Institute and The Lundbeck Foundation Initiative for Integrative Psychiatric Research, iPSYCH. Dr Webb is supported financially by the European Research Council.
Role of the Funder/Sponsor: The funding sources had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; or decision to submit the manuscript for publication.