Comparing forensic and non-forensic women with schizophrenia spectrum disorders: a European study

Abstract

Studies about violence by women with severe mental disorders are rare. The aim of this paper is to analyse the sample of women diagnosed with Schizophrenia Spectrum Disorders (SSD) from the EU-VIORMED study who had offended violently and were admitted to forensic facilities (cases), and compare them to women with SSD who never exhibited violent behaviour (controls). Cases and controls matched for age and diagnosis were compared for sociodemographic, clinical, neuropsychological, and treatment-related characteristics using a standardised assessment. When compared to 36 controls, the 26 cases were significantly older, with longer duration of illness, had fewer years of education, were less likely to have children, and were more likely to have a comorbid personality disorder. Cases were less functionally impaired and scored lower on cognitive domains. There were no differences between the groups in exposure to childhood or adult violence, but a greater proportion of cases reported more frequently being witness to and victims of violence and more frequently reported being beaten, kicked, or punched. Results suggest that the emergence of violent behaviour in women with SSD might be shaped by various factors including violent victimisation, personality factors, soft cognitive impairment and perhaps as a result a more extended duration of illness.

Keywords:

• Violence
• female sex
• mental disorders
• forensic mental health services

Introduction

Generally, men are more likely to be arrested, convicted and sent to jail, and are also more likely to be treated by forensic psychiatric services than women. The size of prison populations has been increasing in the past two decades and has particularly affected the rate of incarceration of women (Butler et al., 2005; Kataoka et al., 2001; Tye & Mullen, 2006), including adolescents and adults (Anumba et al., 2012; de Vogel & Nicholls, 2016; Messina et al., 2003). In Europe the rate of women arrested for violent crimes has significantly increased in the last 40 years (Campaniello, 2019); in Canada the number of female prisoners increased by 35% in the ten years from 2000 (Derkzen et al., 2013), while the number of women in federal prisons in the USA increased by 121% compared to an 84% rise for men between 1990 and 2002 (Sacks, 2004). These increases in imprisonment rates might be related to changes in women’s criminal behaviour generally, changes in the characteristics of female offending or changes in the official responses to and the policing of female offenders (Network, 2017). Hence, until recently the female offender population received significantly less attention from policy makers than their male counterparts (Brunelle et al., 2009). Research into the female offender population has been difficult historically due to the low numbers of female offenders; they also have substantially lower reoffending rates than men (de Vogel & Nicholls, 2016) and are responsible for a significantly smaller proportion of serious and violent offences (Logan & Taylor, 2017). All these factors contribute to the sparsity of literature on female offenders. Even less is known about violent mentally disordered female offenders, referred to in this paper as forensic patients: currently around 5-18% of the forensic patients in Europe are women (Tomlin et al., 2021).

In this study we compared women with a primary diagnosis of an SSD and a history of significant interpersonal violence with a sample of matched women diagnosed with SSD who had never exhibited violent behaviour. Female offenders were recruited from multiple forensic institutions in each country; ‘controls’ were recruited from general psychiatric services. Significant interpersonal violence was defined as having committed a homicide, attempted homicide or other assault that caused serious physical injury to another person. The study aimed to shed light on socio-demographic, clinical, criminological and treatment-related characteristics of women in forensic psychiatric care and understand how they may differ from women with SSD outside forensic settings.

Methods

Participants

This study was part of a larger project that compared 221 violent cases of either sex with SSD with 177 age, sex, and diagnosis matched non-violent controls.

The study was approved by the research Ethics Committee (EC) at the coordinating centre (IRCCS Centro San Giovanni di Dio Fatebenefratelli, Brescia, Italy: n. 74–2018), and by the relevant Research Ethics Committees for each of the participating sites (see the end of the paper for EC details).

In each study centre, treating clinicians invited potential patients under their care to enter the study. After cases were identified we carefully used three matching criteria (age categories, sex and SSD diagnosis). Matched controls were recruited from local adult community and residential psychiatric services. The initial plan was to recruit 200 cases and 200 controls: however the worldwide coronavirus outbreak and the resulting restrictions from February 2020 caused recruitment to temporarily halt in every country. The degree and impact of the restrictions varied between the five countries and particularly affected control recruitment. Once recruitment restarted but as some restrictions remained it proved more feasible to over-recruit forensic cases rather than controls.

All participants provided written informed consent before entering the study after a full verbal and written description of the study’s aims and methods. Subjects also consented to allow additional information to be collected from the medical records, caregivers, family members or case-managers/clinical staff.

Socio-demographic, clinical, functional and violence assessment

All subjects were evaluated by research assistants employed by the study and centrally trained on each instrument. Socio-demographic, core clinical and criminological and violence risk data were collected using a study-specific Patient Information Form (PIF), an Index Violence Sheet (IVS) and a violence Risk Factors Questionnaire (RFQ), later cross-referenced with the medical and legal records and clinician review. DSM-5 diagnoses were based on treating clinicians’ evaluations extracted from the medical records.

Current psychotic symptoms were assessed using the Positive and Negative Syndrome Scale PANSS (Kay et al., 1987), based on a semi-structured patient interview and clinical observation. PANSS scoring used the original scoring model (Kay et al., 1987) and overall total scores ranged from 30 to 210. All research assistants underwent official centralised PANSS training in 2018 provided by the PANSS Institute and were certified PANSS raters.

The World Health Organisation Disability Assessment Schedule 2.0-WHODAS 2.0 (Ustün et al., 2010) was used to assess day-to-day functioning across six domains, including cognition, mobility, self-care, getting along, life activities and participation. Sum scores were calculated, yielding a total between 0 and 48, with higher scores indicating more severe problems.

Cognitive assessment

Cognition was tested using the Brief Assessment of Cognition in Schizophrenia-BACS (Keefe et al., 2004). It assesses six cognitive domains: verbal memory and learning, working memory, motor function, verbal fluency, processing speed and executive function (Keefe et al., 2004).

All assessment instruments were available in official and validated translations.

Statistical analyses

Frequencies and percentages for categorical variables and means and standard deviations for continuous variables were evaluated. Chi-squared or Fisher’s exact test was used according to the nature of the data to compare the categorical variables between the experimental groups. The continuous variables’ distribution was established by histogram plots and normality tests. T-tests or the non-parametric Mann-Whitney tests were used for these variables, employing tests for independent samples to account for the fact that our groups consisted of distinct individuals, not perfectly matched pairs or the same individuals measured under different conditions. Additionally, the chi-squared test was deemed not applicable for the Social Support variable, which was assessed through a multiple-choice format. This format allowes participants to select more than one option, making the data incompatible with the assumptions of the chi-squared test, which requires mutually exclusive categories.

Finally, logistic regression models were constructed to quantify the association between the two groups (dichotomous dependent variable) and the variables that significantly differed between cases and controls (independent variables). Each model was adjusted for potential confounders (e.g., sex and country). Then, different multiple logistic regression models were tested to identify the best (in terms of predictive performance) factors with the highest predictive risk or protective value for violence.

All analyses were carried out in SPSS software (IBM Corp. Released 2019. IBM SPSS Statistics for Windows, Version 26.0. Armonk, NY: IBM Corp); the significance level was set at 0.05.

Results

Of 87 female patients who were potentially eligible, 25 declined to participate (10 cases, 27.8% and 15 controls, 29.4%). ‘Cases’ and ‘controls’ refusals differed significantly between the five countries (p = 0.002 and p < 0.001, respectively). In particular, cases’ refusal rates in Poland were lower than in the other countries, while controls’ refusal rate was higher in Germany and in Poland than in the other countries. Unfortunately, we were unable to collect data on refusers given the prohibition of ECs to acquire any information on these subjects.

Within the wider study group of patients who gave consent, there were 26 female cases with SSD and a history of at least one act of serious interpersonal violence and 36 female controls with SSD who had never been violent. Recruitment country did not significantly differ between the two groups (p=.431). The majority of patients were white (92.3% of cases and 94.4% of controls, p=.538). Cases were significantly younger than controls (p=.012).

Table 1 presents the sociodemographic characteristics of the female sample. Marital status did not significantly differ between cases and controls (p=.265). Most women in both groups were single (69.2% of cases, 83.4% of controls), even though among cases a higher percentage (23.1%) were divorced/widowed. By way of contrast, controls were more likely than cases to have children (88.9% vs 61.5%, p=.011). There was also a significant difference in the level of education between the two groups (p=.027), with controls reporting more years of education (14, SD= 3.4) compared to cases (12.3, SD= 2.5). No significant differences were found in occupational status.

Table 1. Socio-demographic characteristics of female forensic patients with ssd and controls.

	Forensic group N = 26 N (%)	Control group N = 36 N (%)	p-value
Country of recruitment
Austria	4 (15.4%)	4 (11.1%)	.431
Germany	6 (23.1%)	8 (22.2%)
Italy	2 (7.7%)	3 (8.3%)
Poland	8 (30.0%)	18 (50.0%)
United Kingdom	6 (23.1%)	3 (8.3%)
Ethnicity
White	24 (92.3%)	34 (94.4%)	.538
Middle Eastern or Arab (incl. all North African nations)	0	1 (2.8%)
Asian	1 (3.8%)	0
Black/African/ Central or South American (Hispanic)	0	0
Don’t know/won’t say	1 (3.8%)	1 (2.8%)
Age (Years), Mean (SD)	42.4 (12.4)	35.4 (9.0)	.012
Marital status
Married or cohabiting	2 (7.7%)	3 (8.3%)	.265
Single	18 (69.2%)	30 (83.4%)
Divorced or widowed	6 (23.1%)	3 (8.3%)
Children
Yes	16 (61.5%)	32 (88.9%)	.011
No	10 (38.5%)	4 (11.1%)
Education years, Mean (SD)	12.3 (2.5)	14.0 (3.4)	.027
Highest occupational status
Never worked/ Student/housewife	2 (7.7%)	6 (16.7%)	.753
Unskilled worker	11 (42.3%)	13 (36.1%)
Skilled worker	11 (42.3%)	15 (41.7%)
Professional	2 (7.7%)	2 (5.6%)
Social support #
None	4 (15.4%)	4 (11.1%)	–
Family	20 (76.9%)	29 (80.6%)
Friends	8 (30.8%)	10 (27.8%)
Other patients	4 (15.4%)	1 (2.8%)
All	0	0

#Multiple-choice question: chi-square test was not possible.

Criminological history of the forensic sample

Table 2 presents key characteristics of the cases, including the nature of their index violence, their mental state around the time of that violence, the presence of substance use disorders and contact with mental health services (MHS) and diagnoses before the index violence, whether they were prescribed and compliant with medication at the time if their index violence, and finally their arrest and conviction history in relation to the index violence. Notably, the majority of the index violence incidents were considered to be reactive/impulsive (62.5%), and a significant portion of cases were already in contact with MHS (80.0%) at the time they committed their index violence.

Table 2. Criminological and clinical characteristics of the forensic sample at the time of the index violence.

	Forensic cases N = 26 N (%)
Type of crime (index violence)
Homicide	12 (46.2)
Attempted homicide	3 (11.5)
Assault with injury	7 (26.9)
Used a weapon/force	1 (3.8)
Other	3 (11.5)
Type of index violence
Reactive/Impulsive	15 (62.5)
Premeditated	9 (37.5)
Abnormal mental state (during the index violence)
No	0 (0)
Yes	26 (100)
Type of abnormal mental state (during the index violence)
Psychotic disorder	26 (100)
Mood disorder	0 (0)
Other	0 (0)
Substance use disorders (at the time of the index violence)
No	13 (56.5)
Yes	10 (43.5)
Already in contact with MHS (at the time of the index violence)
No	5 (20.0)
Yes	20 (80.0)
Diagnosis before the index violence if already in contact with MHS
Psychotic disorder	18 (90.0)
Personality disorder	1 (5.0)
Mood disorder	1 (5.0)
Prescribed medications (at the time of the index violence)
Antipsychotics	14 (82.4)
Antidepressants	1 (5.9)
Mood stabilizer	2 (11.8)
Pharmacological compliance (at the time of the index violence)
No	17 (94.4)
Yes	1 (5.6)
History of conviction or arrest for the index violence: FIRST admission at the time of index violence
No	5 (19.2)
Yes (current forensic unit)	7 (26.9)
Yes (another forensic unit)	5 (19.2)
Yes (prison)	9 (34.6)

Psychopathology, psychosocial functioning and cognition

There was a trend for case to have been older at first contact with mental health services (p =. 133). Cases had a significantly longer illness duration and were significantly older than controls Table 3. Cases and controls were similar with respect to the nature of their SSD diagnosis, but were more likely to have a comorbid diagnosis of personality disorders (36%) than controls (5.7%) (p=.003). Lifetime substance use problems and suicide attempts were common in both cases and controls.

There were no significant differences in mean PANSS scores suggesting similar levels of positive symptoms, negative symptoms, general psychopathology, and overall symptom severity among cases and controls at the time of testing (Table 4
). Cases functioned better socially on the WHODAS, reflecting lower disability and better functioning. Cases had non-significantly lower scores on non-memory tasks, but had significantly worse executive function as measured by the Tower of London test.

Table 3. Clinical characteristics of female forensic patients with ssd and controls.

	Forensic group N = 26 N (%)	Control group N = 36 N (%)	p-value
Illness duration (Years), Mean (SD)	15.4 (10.6)	9.3 (7.9)	.016
Age of first contact with DMHs (Years), Mean (SD)	27.0 (11.2)	23.1 (8.3)	.133
Type of SSD diagnosis
Schizophrenia	22 (84.6%)	28 (77.8%)	.184
Schizoaffective disorders	2 (7.7%)	8 (22.2%)
Delusional disorder	1 (3.8%)	0
Brief psychotic disorder	0	0
Schizophreniform disorder	1 (3.8%)	0
Drug-induced psychosis	0	0
Comorbidity with Personality disorders
No	16 (64.0%)	33 (94.3%)	.003
Yes	9 (36.0%)	2 (5.7%)
Type of Comorbid personality disorders
Borderline personality disorder	6 (66.7%)	1 (50.0%)	.391
Antisocial personality disorder	2 (22.2%)	0
Other	1 (11.1%)	1 (11.1%)
Lifetime substance use
Never	12 (46.2%)	13 (37.1%)	.479
Yes	14 (53.8%)	22 (62.9%)
Lifetime attempted suicide/self-harm behaviours
Yes	12 (46.2%)	16 (47.1%)	.944
No	14 (53.8%)	18 (52.9%)
Collaboration skills in the last year
Actively seeks treatment, willing to collaborate	13 (50.0%)	25 (69.4%)	.275
Wants to be helped, but lacks motivation	3 (11.5%)	4 (11.1%)
Passively accepts the treatment/intervention	7 (26.9%)	5 (13.9%)
Does not show attention or compreh. for treatment efforts	1 (3.8%)	2 (5.6%)
Actively refuses the treatment/intervention	2 (7.7%)	0
Medications
No	0	1 (2.8%)	.392
Yes	26	35 (97.2%)
Antipsychotics
No	0	0	–
Yes	26	35

History of victimisation and violence exposure

For most areas explored, there were no significant differences between female cases and female controls (Table 5
). We found no significant differences in whether the forensic cases and controls were ever exposed to violence. However, there was a notable difference in the frequency burden with which they were exposed to violence in the past (p=.034). Cases were more likely to report often being exposed to violence (62.5%), whereas controls predominantly reported that it was a rare experience (66.7%). More cases were also often themselves the victims of violence compared to their non-forensic counterparts (p=.023), and were more likely to have been beaten, kicked or punched by someone (p=.020).

Table 4. Clinician-administered assessment tools in female forensic patients with ssd and controls.

	Forensic group N = 26 Mean, SD	Control group N = 36 Mean, SD	p-value
PANSS
Positive symptoms	16.4 (7.3)	15.7 (6.2)	.700
Negative Symptoms	18.7 (5.9)	18.1 (7.5)	.766
General psychopathology	35.8 (12.1)	35.4 (10.5)	.877
Total score	72.1 (22.9)	69.0 (22.8)	.624
WHODAS 2.0
Total score	7.0 (6.8)	16.7 (7.6)	<.001
BACS
Verbal memory #	37.6 (12.3)	37.3 (13.7)	.937
Working memory#	15.3 (3.3)	16.2 (4.7)	.412
Token motor task #	58.0 (14.4)	65.1 (18.0)	.121
Verbal fluency #	40.9 (12.1)	45.2 (14.4)	.244
Symbol coding task #	40.5 (9.7)	44.2 (11.1)	.206
Tower of London #	12.7 (5.4)	15.3 (4.2)	.050

PANSS: positive and negative syndrome scale; WHODAS 2.0: World Health Organisation Disability Assessment Schedule 2.0; BACS: brief assessment of cognition in schizophrenia.

#Raw scores.

Multivariable logistic regression model

We then conducted a multivariable logistic regression using all the factors that differed between cases and controls as shown in Tables 1, 3 and 4 (Table 5). Backward selection with elimination was employed to determine the most impactful factors, which are presented in Table 6. The results indicate that age in years, number of years in education, comorbidity with personality disorders, WHODAS 2.0 total score, and having experienced physical violence, that was being beaten, kicked, or punched by someone, significantly predicted the likelihood of being a case as compared to a control. Specifically, as age in years increased by one, the odds of being a forensic case increase by a factor of 1.02 (with a 95% confidence interval [CI] from 1.00 to 1.05). Conversely, for each additional year in education, the odds decrease by a factor of 0.88 (95% CI: 0.82-0.95). Individuals with a comorbid personality disorder were substantially more likely to be cases, with an odds ratio of 4.02 (95% CI: 1.91-8.46). Greater WHODAS 2.0 total scores were associated with a reduced likelihood of being a case, with an odds ratio of 0.92 (95% CI: 0.89-0.95). Finally, individuals who have been assaulted in the past by someone had increased odds of being forensic cases, with an odds ratio of 2.46 (95% CI: 1.42-4.24).

Table 5. History of victimisation and violence in female forensic patients with ssd and controls.

	Forensic group N = 26 N (%)	Control group N = 36 N (%)	p-value
Witness of physical and/or sexual violence in the family
Yes	9 (34.6%)	12 (36.4%)	.889
No	17 (65.4%)	21 (63.6%)
When witness of violence
Early childhood – Childhood - Adolescence (0-18 years)	7 (77.8%)	11 (91.7%)	.368
Adulthood	2 (22.2%)	1 (8.3%)
Frequency of witness of violence?
Rarely	2 (25.0%)	8 (66.7%)	.034
Occasionally	1 (12.5%)	3 (25.0%)
Often	5 (62.5%)	1 (8.3%)
Medical attention needed after witnessed violence
Yes	2 (22.2%)	1 (8.3%)	.368
No	7 (77.8%)	11 (91.7%)
Victim of physical and/or sexual violence in the family
Yes	9 (36.0%)	10 (30.3%)	.647
No	16 (64.0%)	23 (69.7%)
When victim of violence
Early childhood – Childhood - Adolescence (0-18 years)	6 (75.0%)	8 (80.0%)	.800
Adulthood	2 (25.0%)	2 (20.0%)
Frequency of victim of violence
Rarely	2 (28.6%)	6 (60.0%)	.023
Occasionally	1 (14.3%)	4 (40.0%)
Often	4 (57.1%)	0
Medical attention needed after violence
Yes	2 (25.0%)	1 (10.0%)	.396
No	6 (75.0%)	9 (90.0%)
Beaten, kicked or punched by someone?
Yes	16 (64.0%)	11 (33.3%)	.020
No	9 (36.0%)	22 (66.7%)
When beaten, kicked or punched
Early childhood – Childhood - Adolescence (0-18 years)	7 (43.8%)	6 (54.5%)	.581
Adulthood	9 (56.3%)	5 (45.5%)
Frequency of beaten, kicked or punched
Rarely	7 (53.8%)	8 (72.7%)	.575
Occasionally	3 (23.1%)	2 (18.2%)
Often	3 (23.1%)	1 (9.1%)
Medical attention needed after having beaten, kicked or punched
Yes	3 (20.0%)	1 (9.1%)	.446
No	12 (80.0%)	10 (90.9%)

Table 6. Results of logistic models. association between socio-demographic and clinical variables (independent variables) and the two groups (cases vs controls).

Independent variable (reference category)	Odds Ratio [95% CI]	p-value
Age (Years)	1.02 [1.00–1.05]	.050
Education years	0.88 [0.82–0.95]	.002
Comorbidity with Personality disorders (Yes)	4.02 [1.91–8.46]	<.001
WHODAS 2.0 Total score	0.92 [0.89–0.95]	<.001
Beaten, kicked or punched by someone (Yes)	2.46 [1.42–4.24]	.001

Discussion

To our knowledge, EU-VIORMED is the first international study of women with SSD who have committed a violent offence and were admitted to forensic psychiatry units. Studies on women with mental disorders and who have offended are few (Carabellese et al., 2018, 2019; de Vogel et al., 2014, 2019; Giacco et al., 2023): most found that schizophrenia with emotional instability, often in the context of a comorbid borderline personality disorder and past histories of wider violent behaviour and long histories of contact with psychiatric services, was common. Our findings provide an initial understanding of the sociodemographic and clinical characteristics of forensic women with SSD compared to women with SSD who have never exhibited violent behaviour. Overall, the emerging picture is that women with SSD who have been violent and are managed by forensic services had evidence of complex and enduring mental health needs, social disadvantage, poor histories of educational achievement and limited evidence of successful employment experiences.

Sociodemographic characteristics and clinical profiles

We found significant differences between forensic cases and controls in terms of education, and clinical characteristics: cases had lower levels of education, and had a longer duration of illness than controls. Educational attainment and employment have long been known to represent areas of vulnerability for those in contact with the criminal justice system. Samples of individuals with mental disorders are also often noted for their low rates of school completion and stable employment. The results of this study support the past literature showing first the poor academic achievement and then the poor employment records of female individuals with SSD who are in custodial settings (Hurley & Dunne, 1991; Landgraf et al., 2013; Ribeiro et al., 2015). There are important implications from these findings for tailoring psychological and other interventions to the needs of forensic patients, as well as the clear need for effective and well-resourced interventions to specifically target improving the educational and employment prospects of this vulnerable cohort.

The current study found that 74% of women had more than one psychiatric diagnosis and cases were more likely to have a comorbid personality disorder, specifically borderline personality disorder. These high rates of psychiatric comorbidity are not surprising, but highlight the mental health complexities that these women face, and emphasise how forensic psychiatric services for women must offer the treatment programmes and staff expertise to treat and care for such individuals with multiple and varied psychiatric needs.

The results of this study are comparable to Wolf et al. (2023), who compared violent vs non-violent female forensic patients (Wolf et al., 2023). Like the present study, Wolf et al. (2023) found that violent cases were older than non-violent controls and had lower educational achievement. Differing from the Wolf et al. (2023) paper, this study found no differences in symptom profiles or measures of substance use. These different results could well be due to the difference in the control groups, with Wolf et al. (2023) using non-violent forensic patients and the current study non-violent civil patients. In an earlier study, Dean et al. (2006) compared violent and non-violent women with psychosis in the community: like our study Dean et al. (2006) found that violent offending was associated with comorbid personality disorder and types of victimisations.

Risk factors for violent behaviour

There is an emerging consensus about what the risk factors for violence are in men with SSD based on work such as the McArthur study (Monahan et al., 2001) and later meta-analyses (Whiting et al, 2022; Witt et al., 2013). These risk factors include a past history of violence, active substance use disorders, non-adherence to medications, younger age, co-occurring personality disorders, poor impulse control, unstable living situations, perceived threats or command hallucinations and cognitive impairments. By way of contrast, the factors in women are less well established, and may differ. It may be that amongst women in forensic care, there are those who have done badly at school and in work, had children but were unable to look after them, were victimised and expressed aggressive tendencies from childhood but were not prosecuted until they then seriously harmed someone else (Hodgins, 2022). As a result there is a growing literature that suggests that violence risk assessments needs to be sex informed (Van Voorhis et al., 2010), and our results provide some support for these developments.

Functioning and cognitive assessment

Functional impairment was significantly greater among controls compared to cases, suggesting that those in forensic settings, despite being older and with a longer history of psychiatric illness, retained better functional skills or might have received more targeted support to maximise functional difficulties. This is a finding common to earlier studies of men, where forensic samples have also retained higher levels of functioning. The often-advanced explanation was that the presence of negative symptoms was protective against violence. There were no significant cognitive differences between the two female groups, as has also typically been found in men (Iozzino et al., 2021), although cases generally scored lower on cognitive domains, implying potential cognitive challenges that could influence their risk for violence.

Victimisation history and violence exposure

Trauma is a risk factor for the development of mental ill health, substance use and offending (Dolan & Whitworth, 2013). Our results show that both forensic cases and controls reported a history of victimisation, but that the cases were more often themselves the victims of violence.

Since a significant proportion of the data collected about victimisation came from a review of case notes, we lacked the richness of a dedicated, focused victimisation instrument. Whilst it was frequently documented that women with SSD had suffered abuse or developmental trauma, often the nature, including type, timing, frequency, and perpetrator of these traumatic experiences was not clearly documented. The women in this study stood out for their rates of being victims of violence as well as being violent offenders in adulthood. Historically, as seen above, emphasis has been placed on the experience of trauma and abuse in childhood and adolescence; however, the presence of abuse and victimisation throughout the lifespan is becoming increasingly well recognised. In the current study, 27 of the 58 women (58.5%) had been assaulted during their life, and this proportion was significantly greater among cases. This highlights the significant difficulties and social disadvantages that female forensic patients have and continue to experience. It emphasises that, beyond treatment for their mental disorder and steps to manage their offending, other interventions need to be broader and more holistic in their focus, looking at education, employment and of course relationships and that all interventions need to adopt a trauma-informed approach.

Limitations

The modest size of this multinational sample cannot be considered generalisable to the wider forensic psychiatric population. Many potentially eligible patients, close to 30% in both samples, refused to participate, and in line with good ethical practice, we were unable to collect any data on them. It was impossible to determine whether the refusers differed clinically or in risk-relevant characteristics from those who were recruited into the study. The relatively small sample size in our study may impact the statistical power and limit the external validity of our findings. Consequently, caution is warranted in extending the implications of our findings beyond the specific characteristics of our study sample.

Moreover, we recognise the importance of addressing the issue of multiple comparisons, which poses a risk of both Type I and Type II errors. Given the multitude of variables examined in our study, there is an increased likelihood of chance findings, potentially resulting in false positives (Type I errors) or overlooking true effects (Type II errors). The risk of spurious associations should be considered. Future research with larger sample sizes and appropriate statistical corrections for multiple comparisons is warranted to enhance the reliability and validity of the observed relationships.

Assessment of the patients’ symptoms was based on the current interview, rather than an assessment of the symptoms at the time of the violent offence. The personality disorder assessment relied on clinicians’ judgement and the clinical notes, and was not based on a structured instrument. Cognitive performance could have been affected by the patients’ psychotic symptoms at the time of testing and their current treatment, which could affect performances. The histories of victimisation and trauma were based on self-reports without third party verification.

Conclusions

The study’s findings have implications for clinical practice, intervention strategies, and policy development in addressing the unique needs of forensic women with SSD. Based on the results of the present study, it can be concluded that more gender-responsive service design and treatment strategies are needed in forensic psychiatry: for instance, treatment programmes should address issues such as trauma. Considering the vulnerability of these traumatised women, there is a substantial risk of being victimised again (Finkelhor, 2011), and preventing revictimization should be a high priority in treatment (Covington & Bloom, 2006).

However, the evidence from this cross-sectional study does not allow a complete understanding of the directionality of the associations we found, and we cannot rule out reverse causality for some of these. Despite possible bias and underestimation, our study confirms that the manifestation of violent behaviour among women with SSD is multifactorial. Factors such as long duration of illness, stories of victims of violence and borderline personality disorder may play a key role in aggressive behaviour: hence, all these areas need to be carefully handled in order to maximise the chances of a successful rehabilitation of women with SSD who have offended.

Ethical approval

The project was approved by relevant local or national ethical committees of each country the first approval was obtained by the St. John of God Ethical Committee (coordinating centre) on July 20th, 2018 (permission n. 74–2018); subsequent permissions have been obtained in each of the other recruiting countries according to national and local policies. Here with the details:

• Austria (Medical University of Vienna): approval EK Nr: 1603/2018, date 24.8.2018.

• Germany (Central Institute of Mental Health, Mannheim): approval 2018-582 N-MA, date 10.7.2018.

• Poland (Institute of Psychiatry and Neurology, Warsaw): approval 35/2017, date 7.6.2018.

• UK (King’s College, London): approval 18/SW/0264, date 10.12.2018.

Consent for publication

Participants have been informed about the aims of the project and signed informed consent to the participation and publication of results

Authors’ contributions

G.D.G., J.H., J.W., L.I., and M.P. designed the study. G.D.G., L.I., R.N., J.H., J.W., M.K., M.P., M.Z., A.C., S.G., and M.L. interpreted the data, wrote, read, and edited the paper.

Acknowledgments

Collaborators: R. Oberndorfer, A. Reisegger, T. Stompe, (Clinical Division of Social Psychiatry, Department of Psychiatry and Psychotherapy, Medical University of Vienna, Austria); B. Horten, A. Giersiefen, J. Schmidt (Central Institute of Mental Health, Mannheim, Germany); R. Ruiz (Institute of Psychiatry, Psychology and Neuroscience, King’s College London); I. Markiewicz, M. Ozimkowicz, M. Pacholski (Institute of Psychiatry and Neurology, Warsaw).

Acknowledgments are also due to: Austria: M. Koch, S. Stadtmann, A. Unger, H. Winkler (Clinical Division of Social Psychiatry, Department of Psychiatry and Psychotherapy, Medical University of Vienna, Austria), A. Dvorak (Justinzanstalt Goellersdorf, Goellersdorf, Austria), A. Kastner (Klinik für Psychiatrie mit forensischem Schwerpunkt, Linz, Austria). Germany: H. Dressing, E. Biebinger (Klinik für Forensische Psychiatrie Klingenmünster), C. Oberbauer (Klinik für Forensische Psychiatrie und Psychotherapie Wiesloch), M. Michel (Klinik für Forensische Psychiatrie und Psychotherapie Weinsberg). Italy: G. Tura, A. Adorni, S. Andreose, S. Bignotti, L. Rillosi, G. Rossi (IRCCS Fatebenefratelli, Brescia), L. Castelletti, G. Rivellini (REMS ULSS9 Scaligera, Verona, Italy), F. Lazzerini, A. Veltri (REMS AUSL Toscana Nord-Ovest), G. Nicolò, C. Villella (REMS ASL Roma 5), A. Vita, P. Cacciani, G. Conte, A. Galluzzo (Department of Mental Health, ASST Spedali Civili, Brescia). Poland: I. Markiewicz, M. Ozimkowicz, A. Pilszyk, M. Pacholski (Institute of Psychiatry and Neurology, Warsaw), A. Welento-Nowacka (Forensic Department, Mental Health Hospital in Starogard Gdański). United Kingdom: N. Blackwood (Institute of Psychiatry, Psychology and Neuroscience, King’s College London).

Acknowledgments are also due to Professor Gianluca Castelnuovo (Faculty of Psychology, Catholic University of Sacred Heart, Milan) for his support to the realisation of this paper.

Disclosure statement

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

Data availability statement

The project will fully embrace the open access data policy of H2020 to make data FAIR (Findable, Accessible, Interoperable, and Re-usable), and all data gathered in the framework of the project are stored in a public repository (https://doi.org/10.5281/zenodo.4442372) accessible to all scientists willing to carry out additional analyses.

Additional information

Funding

The EUropean Study on VIOlence Risk and MEntal Disorders (EU-VIORMED) project has received a grant from European Commision (Grant Number PP-2- 3-2016, November 2017–October 2020) and is registered on the Research Registry - https://www.researchregistry.com/ - Unique Identifying Number 4604. The funding source had no role in the design and in the conduct of the study, and had no role in data analyses, in the interpretation of results and in the writing of the study report.