Adverse Childhood Experiences and Offending as a Function of Acquired Brain Injury Among Men in a High Secure Forensic Psychiatric Hospital

Abstract

Background

Acquired brain injury (ABI) is a serious problem that disproportionately affects individuals in correctional services, but relatively little is known about ABI risks and correlates in forensic psychiatric services.

Methods

We conducted a retrospective chart review of all admissions to a high secure forensic hospital in Ontario, Canada from January 2009 to December 2012 (n = 637) and collected data on ABI, psychiatric diagnoses, developmental disadvantage, criminal offending, and in-hospital aggression. A k-means cluster analysis was employed to assess risk factors by which men with ABI could be identified and multivariate general linear models were used to identify ABI-related differences in offending history and in-hospital aggression.

Results

One-fifth of the men had a documented ABI indicator. Based on our cluster analysis, ABI was more likely to be identified by greater adverse childhood experiences (ACEs), more health problems from pregnancy to childhood, and lower socioeconomic status, suggesting that ABI within the forensic context is associated with greater developmental disadvantage. Men with ABI had more serious pre-admission offences, but not more serious admission offences or in-hospital aggression. Men with ABI were more likely than those without to have higher scores on the Violence Risk Appraisal Guide or to be diagnosed with mood and personality disorders, and less likely to have a schizophrenia diagnosis, suggesting an association between ABI and general mental health pathologies but not with psychotic illness.

Conclusions

The disadvantage of ABI among men in forensic psychiatric hospitals is most likely evinced in antisocial behaviour rather than serious mental illness. Given that ACEs are likely to precede or co-occur with ABI, strategies that mitigate ACEs hold promise for ABI prevention.

Introduction

Acquired brain injury (ABI), defined as any injury to the brain sustained after birth, ¹ is associated with both criminal activity and major mental illness.^2,3 Forensic psychiatric patients are a high-risk group of individuals with considerably higher rates of ABI than the general population.^4,5 Despite these associations, ABI research in forensic psychiatric populations remains sparse and the relationships between ABI and offending behaviour remain incompletely understood. ⁶

ABI is similarly common in correctional settings. ⁷ Among incarcerated men, ABI has been linked to criminal conduct and recidivism,^8,9 as well as manifestations of aggression. ¹⁰ Some have debated the link between ABI and criminal conduct, ¹¹ citing instead that high reported rates of ABI in prisoner populations reflect an excess of socio-demographic risk factors for ABI. ¹² To the best of our knowledge, only one study to date has evaluated the characteristics of forensic psychiatric patients with a history of ABI, ¹³ and the putative antecedents of ABI in this setting have yet to be examined. Our aim in the present study was to expand knowledge of ABI and its risk factors in forensic patients. We examined documented reports of ABI in men in forensic care, identifying risk factors for reports of ABI and investigating their relation to aggressive and criminal behaviour.

Investigating ABI within the forensic psychiatric context may help disentangle the complex associations between mental illness and criminogenic factors. Adverse childhood experiences (ACEs), which describe abuse (psychological, physical, or sexual) or household dysfunction (substance abuse, mental illness, witnessing violence, parental separation, or criminal behaviour) within the first 18 years of life, ¹⁴ have shown to be predictive of ABI, ¹⁵ criminal offending, ¹⁶ and mental illness. ¹⁷ Empirical evidence also indicates that perinatal insults and childhood health problems, which are frequently concomitant with ACEs, are related to psychiatric illness. ¹⁸ The association between psychiatric illness and ABI is, likewise, well established, ¹⁹ but little is known about clinical implications of ABI in a forensic hospital. Lastly, and perhaps most importantly, comorbidity of ABI and psychiatric illness represents a patient burden and corresponding challenge to providing optimal psychiatric care.

The aim of this study was to investigate whether men within the forensic mental healthcare setting with a history of ABI could be identified as at risk of belonging to this group on the basis of developmental disadvantage and criminal behaviour. We hypothesized that, relative to co-patients with no history of ABI, (i) men with ABI will be collectively differentiated by ACEs, socioeconomic status (SES), and early life (pregnancy to childhood) health problems and (ii) that men with ABI will have more serious criminal offending records and behave more aggressively during the course of psychiatric hospitalization. We explored the psychiatric diagnoses and violence risk associated with ABI in order to consider the clinical implications of men with ABI in forensic psychiatric care.

Methods

This study was a secondary analysis of data collected for a study on a 4-year admission cohort of men admitted for forensic inpatient assessment. ²⁰ The original study was reviewed and approved by our hospital's research ethics board (HPRA #09.03.01 and #19.07.29) with a waiver of patient consent based on the Tri-Council Policy Guidelines for waiver of consent. ²¹ We had access to all coded data that were originally gathered through a review of each patient's medical file, which included forensic assessment reports as well as reports gathered from healthcare and criminal history records during the assessment. We did not conduct neurological exams.

Sample

We included all men admitted for inpatient forensic assessment between January 2009 and December 2012 in a high secure forensic unit in Ontario, Canada, as reported in previous studies of developmental pathways ²⁰ and general physical health. ²² The cohort comprised 638 men and 1 case was removed due to an irreparable data file loss, bringing our total sample to 637. The men were aged at least 16 years, and most were admitted following criminal charges under orders for assessment of fitness to stand trial (n = 253, 40%), eligibility for a defence of Not Criminally Responsible on account of mental disorder (n = 151, 24%), or having been found unfit (n = 161, 25%).

We identified a subgroup of 129 (20%) men as having ABI based on several coded variables regarding events consistent with the potential to cause brain injury documented in the medical file. Our definition included documented childhood brain injury with loss of consciousness (n = 13) or brain injury with dizziness (n = 10) occurring before age 15 years; 1 man had reported both for a total of 22 partcipants with childhood ABI. Dizziness was included as it is a common complaint after head trauma, second only to headache. ²³ Our definition also included adulthood brain injury since the age of 15 years (n = 115). The original data included a variable indicating whether the man experienced a serious head injury or a high fever of 103° Fahrenheit (or 39.4°C) or higher. For men with this item present, we requested information from the original medical file to verify whether a head injury, specifically, was documented. Eight men had both childhood and adulthood ABI, for a total subgroup of 129 men comprising the ABI group. In some cases, ABIs were attributed to a cause, most often stated as motor vehicle accidents, violent altercations, and stroke.

Measures

Our measures relied on variables in the existing dataset. Many variables were originally operationalized as either events recorded in the medical history or symptoms and behaviours observed and documented by clinical staff. For example, the source for information about ACEs originally included a detailed psychosocial report created by social workers who gathered information from other documentation and informants. We created a total count of ACEs from 9 variables, including but not limited to childhood experiences of abuse, including verbal abuse (nonphysical, psychological abuse), physical abuse, sexual abuse, neglect of basic emotional or physical needs, or witnessing the physical or sexual abuse of the mother. Additional ACEs included separation from the biological father before the age of 16, as well as evidence that the mother or father has a criminal record or has experienced a mental illness or problematic alcohol or drug use. All ACEs were coded as “1” if present and “0” if not, for a total possible score of 0–9 ACEs.

We examined other adverse developmental events by using variables in the dataset that recorded total counts of pregnancy, perinatal, and childhood health problems according to the patient's medical file. Pregnancy difficulties during the men's gestation included 11 items: maternal malnutrition, toxemia, exposure to toxic substances (e.g., lead), high fever, injury (from an accident or physical abuse), use of prescribed medications, alcohol or street drug use, history of addiction to alcohol or drugs, extreme psychological stress or abuse, history of mental illness, and other pregnancy difficulties. Perinatal problems during the men's birth included 12 items: low birth weight, forceps/instruments, caesarean section, prolonged labour/delivery, abnormal fetal position, umbilical cord or placental abnormality, asphyxia/anoxia, fetal distress, premature birth, Rh problems, infections, and other perinatal problems. Childhood health problems were coded based on information up to the age of 15 years and included 14 items: colic, serious childhood disease, high fever, serious infection, asthma, tuberculosis, serious allergies, hearing problems, serious headaches, blurred vision, stuttering, fainting or dizziness, major coordination problems, and other childhood illnesses or physical problems. Each health problem was coded as “1” if the medical file indicated that the problem was present, and “0” if not, for a total possible score of 0–11 pregnancy problems, 0–12 perinatal problems, and 0–14 childhood health problems.

SES during childhood was measured using a variable that indicated the highest occupation attained by men's parents up until he was aged 18 years, from 1 = none to 12 = high-level profession that requires graduate education, and 13 = other employment (n = 7). Men's primary diagnosis was that made by the attending psychiatrist and documented in the medical file. The dataset contained up to 5 psychiatric diagnoses recorded by the psychiatrist, and we included any subtype within the major diagnostic category (i.e., diagnosis of schizophrenia included all subtypes). Additional information included the number of days that men spent confined in seclusion as a result of their behaviour or presentation; for confinements of <24 h, we summed the number of hours across confinements and rounded up from 12 h to 1 day.

For the severity of the admission and pre-admission offences, we used variables that captured the Cormier-Lang Criminal History score. ²⁴ For the original data extraction, a full list of charges was available in the criminal history report contained in the medical files. The Cormier-Lang score assigns a value to each charge according to its severity, and we recorded the value for the highest-scoring charge at admission and the highest-scoring charge at any time before the admission offence. As per the Cormier-Lang method, violent charges ranged from assault (2 points) to homicide (including murder, manslaughter, and criminal negligence causing death, 28 points). Non-violent charges ranged from driving, public mischief, break and enter, and offences of similar severity (1 point) to robbery of a bank or store (unarmed; 7 points).

In-hospital aggression was measured using variables that captured assaults and threats from admission up to the date of discharge or a full year after admission, whichever came first. Specifically, in-hospital assaults comprised a count of documented times the patient made forceful physical (or sexual) contact with another patient, staff, or visitor, excluding attempted assaults and spitting. In-hospital threats comprised a count of documented spoken or written threats in which there was an unambiguous threat to cause physical harm or death to a specific patient, staff, visitor, or other person, excluding threats of emotional or financial harm and physical gestures unaccompanied by a verbal threat of physical harm. As some individuals emitted a large number of threats, this variable had been capped at 10 threats in order to reduce the burden on coding resources. To explore risk of post-discharge violence, we scored the Violence Risk Appraisal Guide (VRAG) ²⁴ by coding each item from the information available in the patient's file. The VRAG is a 12-item actuarial risk assessment tool that has yielded large predictive effects for violent recidivism in meta-analytic studies.^25,26 One VRAG item is derived from the Revised Psychopathy Checklist (PCL-R) score. ²⁷ To score this item, we used the PCL-R score on file if available. If not, we used the score on the Child and Adolescent Taxon Scale (CATS), a measure of childhood and adolescent antisociality as advised by the VRAG authors. ²⁴ We coded the CATS from file information and used the score, prorated out of 40, to score the VRAG item.

Procedure and Inter-Rater Reliability

All data were coded from individuals’ medical files using standard coding forms with detailed instructions about how variables were defined and where in the file the information could be extracted from. Variables pertaining to the assessment conducted at admission and pre-admission history were coded shortly after admission, primarily by a researcher with substantial experience reliably quantifying the information in forensic psychiatric records. About one-third of cases were coded by EH and a small number by a research student after training and evaluation of inter-rater reliability with the primary coder. In-hospital aggression was subsequently coded by EH or 1 of 8 research students. After training, each coder's reliability was evaluated on 10 cases coded independently and masked to the primary coder's coding, prior to coding further cases.

Variables tested in this study showed good agreement according to mean intra-class correlation coefficients (ICCs; mixed methods, absolute agreement, and single measures) averaged across coder pairs. A perfect agreement was obtained for admission offence severity, primary diagnosis, pregnancy problems (all reliability cases scored 0), and perinatal problems. Strong agreement was shown for childhood health problems (ICC = .75); ACEs (ICC = .87); most serious pre-admission charge (ICC = .99); number of in-hospital assaults (ICC = .92); number of in-hospital threats (ICC = .79); number of days confined (ICC = .99); and VRAG score (ICC = .87). However, agreement was poorer for SES (ICC = .57) and ABI (ICC = .64). For the present study, KB, EH, and ZH reviewed details from the medical files of each forensic patient identified in the original dataset as having an ABI and achieved consensus on inclusion of each case.

Data Analysis

Analyses were conducted using SPSS version 28, and run independently by EH and ZH to verify accuracy. Substantial skew in the variables of interest was not sufficiently reduced by log transformation, precluding discriminant analysis; therefore, we tested our primary hypothesis using k-means cluster analysis, entering pregnancy problems, perinatal problems, childhood health problems, ACEs, and SES setting the number of clusters to 2 and excluding missing cases pairwise. We tested our secondary hypothesis using a multivariate analysis of covariance in order to avoid inflating the number of tests. We ran 1 multivariate analysis for the severity of criminal history (most severe pre-index charge and index charge), with ABI as the independent variable and age at admission offence as a covariate to account for the confound of time on criminal history. We ran a second multivariate analysis for in-hospital aggression (number of assaults and threats), with ABI as an independent variable and length of stay as a covariate.

Results

Sample characteristics of men with and without any evidence of ABI are shown in Table 1. In terms of developmental disadvantage, men with ABI had lower SES, more ACEs and more childhood illnesses, but the numbers of pregnancy difficulties and perinatal problems were similar. Compared to the men with no history of ABI, men with ABI were more likely to be diagnosed with personality disorders or mood disorders, and less likely to be diagnosed with schizophrenia. Men with ABI had higher VRAG scores indicating a higher risk of violent recidivism. Regarding other clinical issues, men with ABI were more likely to have evidence of an alcohol use problem both before and since the age of 18. They also spent less time in confinement while in the hospital.

Table 1.

Sample Characteristics as a Function of Patient Group.

	Any ABI			No ABI			Total Sample
Measures	N	% or mean (SD)	95% CI	N	% or mean (SD)	95% CI	N	% or mean (SD)	95% CI
Demographic characteristics
Age at admission	129	37.36 (12.07)	35.25, 39.46	508	34.71 (12.36)	33.63, 35.79	637	35.25 (12.34)	34.29, 36.29
Education	129	10.88 (2.79)	10.40, 11.37	508	10.64 (2.99)	10.38, 10.90	637	10.69 (2.95)	10.46, 10.92
Ever married	42	33	24, 41	105	21	17, 24	147	23	20, 26
Childhood development
Socioeconomic status	67	9.01 (2.26)	8.46, 9.57	235	9.07 (2.11)	8.80, 9.34	302	9.06 (2.01)	8.81, 9.30
Number of ACEs	129	1.98 (1.76)	1.68, 2.29	508	1.38 (1.66)	1.24, 1.52	637	1.50 (1.70)	1.37, 1.63
Number of pregnancy difficulties	129	0.32 (0.57)	0.22, 0.42	508	0.27 (0.61)	0.21, 0.32	637	0.28 (0.60)	0.23, 0.32
Number of perinatal problems	129	0.23 (0.73)	0.10, 0.36	508	0.20 (0.56)	0.15, 0.25	637	0.21 (0.60)	0.16, 0.25
Number of childhood illnesses	129	0.53 (0.81)	0.39, 0.67	508	0.33 (0.71)	0.27, 0.39	637	0.37 (0.73)	0.31, 0.43
Psychiatric history
Diagnosis
Personality disorder	36	28	20, 36	73	14	11, 17	109	17	14, 20
Antisocial personality disorder	22	17	10, 24	51	10	7, 13	73	11	9, 14
Schizophrenia	44	34	26, 42	234	46	42, 50	278	44	40, 48
Other psychosis	29	22	15, 30	107	21	18, 25	136	21	18, 25
Any psychotic disorder	73	57	48, 65	340	67	63, 71	413	65	61, 69
Substance use disorder	35	27	19, 35	120	24	20, 27	155	24	21, 28
Mood disorder	27	21	14, 28	58	11	9, 14	85	13	11, 16
Prior psychiatric admission	106	85	78, 91	387	81	77, 84	493	82	79, 85
Confinement	109	7.53 (17.47)	4.22, 10.85	414	14.50 (44.60)	10.19, 18.81	523	13.05 (40.56)	9.57, 16.53
VRAG score	128	10.59 (12.17)	8.45, 12.71	502	7.91 (12.73)	6.79, 9.02	630	8.45 (12.65)	7.46, 9.44
Alcohol problem age <18	42	33	24, 41	118	23	20, 27	160	25	22, 28
Drug problem age <18	52	40	32, 49	170	33	29, 38	222	35	31, 39
Alcohol problem age ≥18	95	74	66, 81	278	55	50, 59	373	59	55, 62
Drug problem age ≥18	96	74	67, 82	307	60	56, 65	403	63	60, 67

Note: ABI = acquired brain injury; SD = standard deviation. CI = confidence intervals; ACE = adverse childhood experiences; VRAG = Violence Risk Appraisal Guide.²⁴

Differentiation of Men With ABI

The cluster analysis achieved convergence after 10 iterations and all entered variables were highly significant contributors (F = 13.21 to 574.68). Cluster 1 contained 426 cases (67%) and represented those with relatively few pregnancy, perinatal, and childhood health problems, fewer ACEs, and higher SES. Cluster 2 contained the remaining 211 cases (33%) with significantly higher developmental difficulties. Among men with ABI, 56 (43%) fell into cluster 2, compared with 155 (31%) of other men, χ² (1, N = 637) = 7.73, p = .005. This result is consistent with our primary hypothesis that men with ABI could be identified on the basis of developmental difficulties, although this was a small effect (φ = .110).

Association of Offending History and In-Hospital Aggression With ABI

Group comparisons suggested a small and significant effect of ABI on the most serious pre-admission charge and no significant effect of ABI on the most serious charge at admission, in-hospital assaults or in-hospital aggression (Table 2). The multivariate models showed the same pattern of results (Table 3). There was a significant effect of ABI on the combined criminal history while controlling for age, F(2, 607) = 4.46, p = .012, Wilks’ Λ = .99, partial η² = .014. The effect of ABI on the combined in-hospital aggression while controlling for length of stay was not significant, F(2, 463) = 0.56, p = .570, Wilks’ Λ = .998, partial η² = .002. Table 3 also shows the variable-level results for ABI, showing that ABI was significant only for the severity of the pre-admission charge, and was not significant for the remaining variables. This finding is only partly consistent with our second hypothesis.

Table 2.

Group Differences for Criminal History and In-Hospital Aggression.

	Any ABI		No ABI		Test Statistics
Measures	Mean (SD)	95% CI	Mean (SD)	95% CI	t (df)	p	Cohen's d
Most serious pre-admission offence	4.28 (4.10)	3.55, 5.00	3.37 (2.96)	3.10, 3.63	2.83 (613)	.005	0.28
Most serious admission offence	5.83 (6.77)	4.64, 7.01	5.19 (6.52)	4.62, 5.76	0.98 (630)	.327	0.097
In-hospital assaults	0.42 (1.25)	0.18, 0.66	0.39 (2.06)	0.19, 0.59	0.11 (517)	.914	0.01
In-hospital threats	1.14 (2.62)	0.64, 1.64	1.32 (3.02)	1.03, 1.61	−0.57 (518)	.568	0.06

Note. Bold font indicates a statistically significant result, using alpha = .05/4 to reduce Type I error from conducting 4 separate binary tests. Offence severity was measured by the Cormier Lang Criminal History score.²⁴

Table 3.

Multivariate Analysis of Covariance of Criminal Charges (Most Serious Pre-Admission and Admission Offences) and In-Hospital Assaults (Number of Assaults and Threats).

	Model Statistics			Pairwise Comparisons
Measures	F	df	p	Mean Diff.	Std. Error	p	95% CI
Criminal charges
Intercept	64.70	2, 607	< .001
Age at admission offence	4.47	2, 607	.012
ABI	4.46	2, 607	.012
Pre-admission charge				0.885	.324	.007	0.248, 1.522
Admission charge				0.863	.657	.190	−0.427, 2.153
In-hospital aggression
Intercept	7.36	2, 463	< .001
Length of stay	23.72	2, 463	< .001
ABI	0.56	2, 463	.570
Assaults				0.042	.071	.549	−0.097, 0.182
Threats				−0.227	.288	.431	−0.792, 0.339

Note. ABI = acquired brain injury; CI = confidence interval. The seriousness of criminal charges was measured using the Cormier Lang Criminal History Score.²⁴ Length of stay was capped at 365 days; in-hospital threats were capped at 10. Bonferroni adjustment for multiple comparisons based on estimated marginal means.

Discussion

Approximately 20% of our sample had a reported ABI, which coincides with previous findings in forensic psychiatric samples. ¹³ ABI in our sample was associated with greater developmental adversity and more serious pre-admission charges, as predicted. Men with reported ABI also had higher violence risk assessment scores. These associations, however, did not correspond to greater admission charges or in-hospital aggression for men with an ABI. Personality and mood disorders were more commonly associated with ABI, yet schizophrenia was less prevalent when compared to men with no documented ABI. In sum, ABI was associated with identifiable risk factors but not with a more challenging psychiatric presentation.

Cluster analysis based on developmental difficulties, including ACEs, early-life (i.e., pregnancy to childhood) health problems, and SES differentiated among men with and without ABI documented. The association between ACEs and physical illness is well recognized in the community literature,^15,28 but this finding is novel among forensic patients, where ACE exposure is comparatively greater. ²⁹ Furthermore, the positive relationship between ACEs and ABI may indicate a potential link to antisociality, as hostile attribution bias is thought to mediate the relationship between ACEs and aggression. ³⁰ Thus, in a forensic psychiatric population, ABI may indicate a developmental pathway involving antisociality and childhood antisocial influences. ²⁰ Early prevention strategies for ACEs and other developmental challenges may hold promise for reducing risk of ABI. Within the forensic hospital setting, patient evidence of developmental adversity may indicate a need to screen for ABI.

ABI reports were associated with more serious pre-admission offences. While this was not observed for admission offences, our measure of the most serious pre-admission offence identified the most severe charge among the individual's documented criminal history and was thus a more robust indicator of offending severity. Further, our sample included only men in a high secure facility who may have more consistently serious admission offences than other forensic patients. In a medium secure setting, ABI was associated with higher proportions of theft and failure to comply/appear charges. ¹³ Although the researchers found no significant link between ABI and the prevalence of violent crimes (i.e., assault or homicide), these comparisons were based on whether any criminal charges occurred and not on the overall offending severity. Our finding is novel with respect to identifying an association between ABI and offending severity. Therefore, documented ABI among men in forensic psychiatric care appears to be associated with offending severity and type of criminal activity.

However, reports of ABI in our sample were not associated with more in-hospital assaults or threats. While this finding was somewhat unexpected, in-hospital assaults were uncommon, averaging <1 across the entire sample, which may have limited statistical power to detect group differences. Our descriptive analyses suggested that men with reported ABI had higher VRAG scores, indicating that ABI exposure is related to increased proclivity for violent recidivism post-discharge. The discordance between violence recidivism risk and those for our hypothesis test regarding observed assaults and threats may reflect that the VRAG is not optimized for identifying the risk of in-hospital aggression.^31,32 Some scholars argue that incorporating information about neurological conditions associated with criminal and violent behaviours has the potential to improve violence risk assessment^33,34 and this may be a fruitful avenue to explore for structured assessment schemes involving in-hospital violence.

Clinically, men in the ABI group had higher rates of mood and personality disorders but lower rates of schizophrenia. Previous research has found that depression is a common sequela of ABI. ³⁵ Positive associations of ABI with personality disorders, and negative associations with schizophrenia, have previously been reported among individuals in medium secure mental healthcare. ¹³ Our study adds to this literature by providing consistent findings for men in high secure care. The identification of ABI in forensic populations is clinically valuable, because comorbid ABI may explain why some patients are more treatment resistant. ³⁶ Amantadine and memantine may show some benefit for symptoms of ABI (although effect sizes are small), ¹ which could positively affect clinical trajectories of individuals with mood and personality disorders. Clinical management strategies using the evidence-based guidelines for diagnosed mental illnesses may need adapting for individuals with ABI in the forensic psychiatric system.

The results of this study should be interpreted considering several limitations. We conducted a secondary analysis, and used existing variables rather than extracting data and code information specifically for this study of ABI. The design was retrospective and, as is common to such studies, we were unable to ascertain the temporal order of ABI, ACEs, criminal activity, and psychiatric diagnoses. This limited our ability to investigate the circumstances that may have contributed to ABI. Available data did not indicate severity of ABI. Furthermore, variables related to ABI data in the original study were extracted from patient medical files which included reports from a variety of sources, such as psychiatric assessments, self-reports, family member reports, and information from investigations in other hospitals or institutions. Self and collateral reports in particular may be biased towards moderate and severe ABI, so that men with mild or undetected ABI were not represented in our ABI group. Future research should include neurological exams and other clinical evaluations to better assess the presence and severity of ABI.