Serotonergic function in aggressive and impulsive behaviour: Research findings and treatment implications
disorders (Brown et al., 1979, 1989; Cocarro, 1992; Siever prediction that ATD would cause an increase in aggressive/
et al., 1993; Cocarro et al., 1995; Cocarro et al., 1996a; impulsive responding, but results have been inconsistent.
Cocarro et al., 1997a; Soloff et al., 2003; New et al., 2004).
Only two of the studies (Soloff et al., 2003; New et al., Smith et al. (1986) used the Buss paradigm to evalu-
2004) included both men and women; reduced 5-HT ate the effect of ATD created through dietary depletion
response was observed in men only. O’Keane et al. (1992) in normal men. In this paradigm participants have to
confirmed these observations among incarcerated men deliver electric shocks in response to a stimulus tone to a
with antisocial personality disorder. Reduced serotoner- partner who, unknown to them, does not exist. Although
gic responsiveness to a fenfluramine challenge has also reduced central 5-HT metabolism was achieved, this did
been described in criminals with a history of conduct not alter levels of aggressiveness. In an adapted design
disorder (Cherek and Lane, 1999) and impulsive offend- using higher levels of provocation, however, an inverse
ers with personality disorder in a high security hospital relationship between tryptophan levels and aggressive
setting (Dolan et al., 2001). Two studies (Manuck et al., responding was found (Pihl et al., 1995). Cleare and Bond
1998), in non-patient samples have shown negative corre- (1995) also demonstrated increased subjective feelings of
lations between prolactin response and life-time history aggression during a competitive task in normal people.
of aggression and self-reported impulsivity. This effect was only observed, however, in those with
higher self-reported aggression ratings. Although these
Meta-chlorophenylpiperazine (mCPP) acts as releasing findings suggest that lowering 5-HT levels may not pre-
agent and reuptake inhibitor at most serotoninergic recep- dispose normal individuals to aggressive overreactions
tors. Studies using it as a pharmacological challenge have to environmental stimuli, it is possible that 5-HT may
largely confirmed observations from fenfluramine challenge modulate aggression in people with significant psycho-
studies with people who have BPD (Hollander et al., 1994; pathology or high base rates of aggression. In line with
Rinne et al., 2000). Moss et al. (1990) found blunted 5-HT this suggestion, Bjork et al. (2000) found that ATD pro-
sensitivity in a sample with ASPD and comorbid substance duced a marked increase in ratings of aggression during
abuse disorders. A general pattern of inverse correlation provocation among people with high but not low trait
between prolactin responses and Buss–Durkee hostility aggression. This differential effect of ATD has also been
inventory assaultiveness subscale scores was found in all found in healthy individuals using other paradigms (e.g.
participants, including the healthy controls. A negative rela- Dougherty et al., 1999). The few other published reports
tionship between prolactin response and the irritability sub- on use of ATD in clinical samples, however, have found
scale was also found in two mCPP studies in patients with no such effects in them, for example Salomon et al. (1994)
mixed personality disorders (Cocarro et al., 1991, 1997b). who studied people with intermittent explosive disorder
(IED) and McCloskey et al. (2009).
In order to characterize further the relationship between
serotonergic response and psychopathology, Cocarro et al. As serotonin appears to be related primarily to the
(1996b) divided their sample of 20 patients with personality impulsive type of aggression, through inhibition of impul-
disorder into ‘blunters’ and ‘non-blunters’ to a fenfluramine sive responding, it might be more fruitful to study the
challenge. Negative correlations with 5-HT function were effects of serontonergic manipulation on impulsivity tasks
observed in both groups, but differed in detail. Prolactin rather than aggression. LeMarquand et al. (1999a,b) thus
values correlated best with assault scores in blunters but studied sons of alcoholic fathers and found increased
with indirect and verbal aggression in non-blunters, sug- error rates on a behavioural inhibition impulsivity task
gesting that reduced 5-HT may be associated specifically but no difference in aggressive responding after ATD.
with assaultive behaviour. This fits with previous obser- Walderhaug et al. (2002) described increased impulsive-
vations by Cocarro et al. (1989) of a strong relationship ness on a continuous performance task after ATD among
between 5-HT function and physical/motor but not idea- healthy controls. One research group, however, found
tional aspects of assaultiveness and impulsivity, tending to improved performance (Crean et al., 2002). Most studies
support a behavioural inhibition model of 5-HT function in did not detect any effect of ATD on impulsivity tasks, such
aggressive behaviour. This is broadly in line with findings as continuous performance tasks, stop tasks (Cools et al.,
by Dolan and Anderson (2003) showing that 5-HT function, 2005; Clark et al., 2005) or Go/No-Go tasks (LeMarquand
assessed by fenfluramine challenge, correlated negatively et al., 1999; Rubia et al., 2005).
with the impulsive–antisocial component of the PCL-SV
but not its arrogant/deceitful factor. Other healthy volunteer studies have described how
5-HT affects learning. It has been shown that lowering 5-HT
Acute tryptophan depletion (ATD) studies impairs the ability to respond appropriately to rewarding or
punishing stimuli, thereby altering the individual’s capacity
5-HT manipulation through ATD provides a further way to adapt behaviour according to feedback (Rogers et al.,
of assessing links between 5-HT activity and aggression 1999; Cools et al., 2005). This is an exciting new line of work,
or impulsivity. The literature just reviewed leads to a but it has not yet been done with samples selected for their
aggression or impulsiveness.
© 2014 by Taylor & Francis Group, LLC 309
Disorders of brain structure and function and crime
Brain imaging studies and 5-HT in the ASPD group. Improving serotonergic function might,
thus, be useful for people with ASPD.
While there is consistent evidence for a role of 5-HT in impul-
sive aggression, how it exerts this is not yet understood. More Treatment implications of
specifically, it is not known how 5-HT modulates brain activ- serotonergic research
ity and which 5-HT receptor types may be involved.
The inverse relationship between measures of 5-HT func-
Neuroimaging studies allow the localization of such tion and indices of impulsivity and aggression raises the
effects in vivo. PET studies have been used to identify the possibility that drugs which enhance serontonergic activity
distribution of 5-HT receptors in humans. Conflicting may be effective in reducing impulsive aggressive behav-
findings have been reported on the relationship between iour. Experimental studies in non-clinical samples have
aggression and receptor density. Parsey et al. (2002) found provided some support for this suggestion. Knutson et al.
a negative correlation between 5-HT1A binding sites and (1998) reported reduced hostility and increased social co-
life-time aggression in healthy men. Rabiner et al. (2002), operation in healthy volunteers after a 4-week trial of fluox-
however, again only with men, found no associations, while etine. Among men with a history of criminal behaviour,
Witte et al. (2009) reported a positive relationship between Cherek and Lane (2001) showed that a single fenfluramine
prefrontal 5-HT1A binding sites and aggression scores in a challenge significantly reduced impulsive responses during
mixed sex sample. Differences in sample composition and a reward choice paradigm. More recently, Kamarck et al.
aggression measures are likely to have contributed to these (2009), in a randomized controlled trial of citalopram in
differences in findings. Frankle et al. (2005), investigat- a large sample of healthy men and women, found that it
ing serotonin transporter distribution, found significantly reduced their anger, hostility and aggression. A meta-analy-
reduced serotonin transporter availability in the anterior sis of animal studies investigating the effect of serotonergic
cingulate in individuals with impulsive aggression com- agents on aggression (Carrillo et al., 2009) concluded that
pared with healthy controls. 5-HT1A receptor binding in increasing 5-HT levels had an inhibitory effect on aggres-
prefrontal structures has been shown to be lower too in sion. To date, though, there have been few well-conducted
violent people than in healthy controls (Meyer et al., 2008). trials of serotonergic drugs with patients with impulsive
aggressive behaviour. The National Institute for Clinical
PET has also been used to investigate regional meta- Excellence (NICE) in England offers no support for such
bolic activity in response to serontonergic stimulation. New interventions for BPD or ASPD (National Collaborating
et al. (2002) compared the effect of an mCPP challenge in 13 Centre for Mental Health 2009a,b).
impulsive aggressive participants and 13 healthy controls,
and found decreased brain activations in the orbitofrontal Table 12.7 provides an overview of such treatment trials
cortex and anterior cingulate, regions known to be involved with patients with personality disorder for whom ratings of
in inhibition of impulsive aggression. In a separate study, the impulsivity or aggression are included; sample sizes of the
New group found that a course of fluoxetine had a normal- first few shown are too small for statistical analysis.
izing effect on prefrontal cortex metabolism in people with
impulsive aggressive behaviour, and that these changes cor- A consistent picture is beginning to emerge which holds
related with clinical improvement (New et al., 2004). some promise for treatment, despite the limitations of the
studies listed, which include small sample sizes, high drop
Brain activations associated with neuropsychological out rates and short trial duration of trials. SSRIs appear
tasks after serontonergic manipulation have been stud- to be effective in the treatment of anger, impulsivity and
ied using fMRI. Anderson et al. (2002) showed enhanced aggression, an effect which seems to be independent of
activations in the lateral orbitofrontal cortex in healthy reduced depression or other psychopathological improve-
controls in a behavioural inhibition task after an mCPP ments. This emerging picture is corroborated by studies
challenge. Other studies with normal volunteers also sug- with people with a wide range of Axis I disorders, and no
gested enhanced activations in prefrontal areas during a personality disorder, who tend to show improvement on
Go/No-Go task after citalopram (Del Ben et al., 2005) and anger and aggression ratings after SSRI treatment. Primary
mirtazapine (Völlm et al., 2006) challenge. This is consist- disorders have included depression (Fava et al., 2000);
ent with the fMRI demonstration by Rubia et al. (2005) of repeated suicidal behaviour (Verkes et al., 1998); dysthy-
reduced right orbital prefrontal activation during a Go/No mia (Hellerstein et al., 1993); panic disorder (Neuger et al.,
Go task after tryptophan depletion. These findings suggest 2002); pervasive developmental disorder (Couturier and
a possible interaction between serontonergic function and Nicholson, 2002); head injury (Kant et al., 1998); bipolar
task specific brain activations, but the application of these depression (Mammen et al., 2004); PTSD (Davidson et al.,
methods to patient groups is still in its infancy. One recent 2004); intellectual disability (Janowsky et al., 2005); and
study by Völlm et al. (2010) used fMRI to study men with post-stroke depression (Choi-Kwon et al., 2006). Rubey et al.
ASPD, using behavioural inhibition and reward paradigms (1996) reported improvements in anger ratings, regard-
following mCPP challenge. Serotonergic modulation of less of primary diagnosis, in a small adult patient sample.
reward pathways was impaired during the Go/No Go task
310 © 2014 by Taylor & Francis Group, LLC
Serotonergic function in aggressive and impulsive behaviour: Research findings and treatment implications
Table 12.7 SSRI treatment studies in personality disordered and impulsive–aggressive patients
Author, year Drug, dose Trial type Duration of Participants Measures aggression Outcome
trial or impulsivity
Cocarro et al., Fluoxetine, titrated Open trial 6 weeks 2 BPD, 1 ASPD Overt Aggression Scale Reduction in all OAS
1990 to 60 mg (OAS) Self-rated global and self-ratings from
8 weeks 5 inpatients aggression (from Hopkins week 1.
Cornelius et al., Fluoxetine Open trial Symptom Checklist-90)
Not stated 10 outpatients with Impulsivity sub-scale Large decrease in
1990 20–40 mg BPD Buss–Durkee Hostility impulsivity. No overall
8 weeks Inventory (BDHI) change in hostility
Heller, 1994 Sertraline, up to Open trial 11 with any Hostility Subscale of on BDHI; change on
200 mg (six patients) Open trial 12 weeks personality disorder the Symptom Checklist SCL-90.
Kavoussi et al., or fluoxetine, SCL-90
1994 20–40 mg (four 7 weeks 27 outpatients with Questionnaire ratings of Improvement with
patients) BPD; mood swings, chronic both drugs on both
12 weeks 22 completed – 13 anger, emptiness/ measures; fluoxetine
Sertraline, flexible treatment, boredom, emotional pain superior due to better
dosing Cross-over 9 controls tolerability.
trial: 6 weeks 46 patients with Overt Aggression Scale Significant decrease
Salzmann et al., Fluoxetine, up to RCT1 + 6 weeks BPD (OAS) of irritability and
placebo, or overt aggression from
1995 60 mg the opposite, 40 outpatients with Profile of Mood States week 4 onwards,
then 12 weeks personality disorder (POMS) maintained at week 8.
Silva et al., Fluoxetine, Open trial open follow up and history of McLean Hospital Overt Significant
1997 20–60 mg 8 weeks impulsive aggression Aggression Symptom improvement in anger
(27 treatment, 13 checklist-revised (OAS-R) in treated group.
Cocarro and Fluoxetine, RCT 12 weeks placebo) Clinical Impulsivity Scale
38 women with Significant
Kavoussi, 1997 20–60 mg BPD, all outpatients; Anger, Irritability, and improvements in
20 treatment Aggression Questionnaire clinical measures
Rinne et al., Fluvoxamine, mean RCT 18 controls (AIAQ) and Impulsivity Scale
2002 166 mg Overt Aggression Scale- scores from week 1 of
20 individuals with Modified (OAS-M) treatment, continued
Reist et al., Citalopram, mean Open trial cluster B personality until week 7.
2003 dose 45.5 mg disorders or Subscales mood swings, Sustained reduction in
intermittent explosive anger and impulsivity of irritability
Simpson et al., Fluoxetine, 40 mg/ RCT disorder from the Borderline Personality and overt aggression
2004 placebo (all general population Disorder Severity on OAS-M and CGI
receiving DBT2) 20 women with Index but no improvement
BPD: 9 fluoxetine in depression.
11 placebo Barratt Impulsivity Scale Improvement on
(BIS) rapid mood swings
Buss Durkee but not impulsivity
Hostility Inventory and anger.
Overt Aggression Scale
(OAS-M) Improvement on
State Trait Anger all scales except BIS
Expression Inventory motor impulsivity,
(STAXI) BDHI assault,
Overt Aggression Scale negativism and
(OAS) suspicion.
No advantages for
fluoxetine
(Continued)
© 2014 by Taylor & Francis Group, LLC 311
Disorders of brain structure and function and crime
Table 12.7 (Continued) Trial type Duration of Participants Measures aggression Outcome
RCT trial or impulsivity
Author, year Drug, dose 45 women with Improvement in all
8 weeks BPD; Overt Aggression Scale three drug groups for
Zanarini, 2004 3 groups: 14 fluoxetine (OAS) chronic dysphoria and
olanzapine 16 olanzapine impulsive aggression;
fluoxetine 15 both olanzapine and
olanzapine + combination superior
fluoxetine to fluoxetine alone.
1Randomized controlled trial.
2Dialectical Behavioural Therapy.
Reductions in overt aggression have also been observed is to respond to similar degrees of threat, frustration,
in child and adolescent outpatients in a trial of citalopram or aversive circumstances with an aggressive outburst.
(Armenteros and Lewis, 2006). It is of note, however, that This proposal is consistent with findings that serotonin
two trials involving women found SSRIs had no such effects plays a particular role in the impulsive type of aggression,
(Rinne et al., 2002; Simpson et al., 2004). This is consistent and probably affects actions more than hostile or aggres-
with our earlier observations that the relationship between sive thoughts or feelings.
serontonergic dysfunction and impulsive or aggressive
behaviour is less well established in women than in men. Nevertheless, more research is needed, especially with
women and with people whose aggression is instrumental
Several authors have suggested SSRIs as first line and planned. The little work done so far with these groups
intervention for symptom-specific treatment of impul- suggests that they may have different serotonergic system
sive-aggressive symptoms, notwithstanding the modest relationships with their aggression. Future research should
evidence (e.g. Soloff, 1998). This strategy was adopted in the also further investigate possible interactions between
practice guidelines of the American Psychiatric Association neurotransmitter systems. Brain imaging studies using
(2001) on BPD, but more recent Cochrane reviews (Lieb et carefully designed behavioural paradigms may illuminate
al., 2010; Bateman, 2012) have concluded that these recom- this, and also possibly more focal deficits. Genotyping, to
mendations can no longer be supported; there is better identify genetic markers of serontonergic function, would
evidence for other pharmacological agents, such as mood provide another promising approach. Finally, the evidence
stabilizers or atypical antipsychotics, at least for BPD (see for treatment of patients with impulsive-aggressive behav-
also chapter 16). More well-conducted randomized con- iour using SSRIs is merely promising; more large scale
trolled trials are needed. In particular, treatment studies RCTs are needed, in well defined samples, to translate
of hospitalized offenders are lacking. It remains to be seen findings about serotoninergic dysfunction in impulsive
whether SSRI treatment will be effective in those with aggression into clinically relevant outputs.
higher baseline levels of impulsive aggression.
Serotonergic function: conclusions Implications of current
knowledge of brain
The literature shows robust evidence for an inverse rela- structure and function
tionship between serotonin levels and both aggression and for forensic mental health
impulsivity. These relationships have been observed in a practice and research
wide range of samples, providing support for the suggestion
that the role of 5-HT/serotonin is not restricted to people Studies of brain structure and function continue to provide
with pathologies, but mechanisms are not fully understood. evidence that abnormal behaviour in any of its cognitive,
One suggestion has been that serotonergic dysfunction emotional or direct action elements has some physical
leads to a decreased ability to inhibit behaviour as well as foundation in the brain. In a growing field, we have focused
impairing capacity to suppress behaviours through learn- on the literature which has the most direct relevance to
ing of their adverse effects, including punishment. Cocarro offenders and offender-patients, but much of the more
and Kavoussi (1996) proposed that: general literature is important in its focus on symptoms of
potential relevance to offending. When asked in court, for
the threshold to act aggressively, given the proper envi- example, it maybe useful to say that such an individual with,
ronmental circumstances, is modulated by overall say, psychotic symptoms, has evidence of organic brain
5-HT system function. The lower the functional sta-
tus of the 5-HT system, the more likely the individual
312 © 2014 by Taylor & Francis Group, LLC
Implications of current knowledge of brain structure and function for forensic mental health practice and research
structural damage or dysfunction that is more consistent while we have not ‘proven’ that this subject is inno-
with that found among others with such symptoms than cent, we demonstrated that her behavioural and func-
among healthy people. Shergill’s group have shown struc- tional anatomical parameters behave as if she were.
tural brain changes among people with hallucinations (e.g. Spence (2008) further advocated caution in reliance on
O’Daly et al., 2007) and the Bentall–Blackwood group have neuro-imaging for detecting lies, having reviewed 16 imag-
shown functional abnormalities among people with delu- ing studies in this field. Greater activation of the prefrontal
sions (e.g. Blackwood et al., 2004). Lui and colleagues (2009) lobes, as with most deviance, is associated with lying com-
have found early associations between psychotic symptoms pared with telling the truth, but sample sizes are generally
and brain structure, particularly reduction of grey matter small, there is much variation between the findings and
in the temporal lobe, before antipsychotic medication is the investigators have never replicated their own findings.
established, with evidence, in addition, of changed func- To all this, one might add, the samples are all apparently of
tional connectivity. The order of association is not clear, healthy people, many of them students and most, although
although others (Wexler et al., 2009) have found that brain not all, men. Then, too, would the findings necessarily gen-
connective tissue (white matter) loss appears to be associ- eralize to offender samples, so often struggling with a range
ated with cognitive impairment while the grey matter loss of psycho- and neuropathology?
was common to groups with psychotic symptoms without
cognitive impairment. Rasetti et al. (2009), after examining So, we return to our theme of relying mainly on studies
healthy siblings and healthy unrelated controls as well as of brain structure and function to inform clinical work and
people with schizophrenia, found that the neuropathol- further research to improve understanding of the mecha-
ogy, here amygdala dysfunction, was characteristic only of nisms that link mental disorder and crime. They may
the patient group, so concluded that this dysfunction at inform treatment on the one hand and perhaps enhance
least was a feature of illness rather than inherent difficul- techniques for determining recovery and safety on the
ties. With increasingly thoughtful and inventive protocols, other. Where a treatment is truly effective, being able to
however, a sense persists that there is still a long way to go show that it has fundamentally changed structure or func-
before we can be much more definite than invoking pre- tion in the brain would be a substantial advance. An excit-
frontal, temporal and limbic system damage or dysfunction ing study in this regard, given the link between passivity
to account for most behavioural anomalies – perhaps a bit delusions and serious violence (chapter 14), was one from
more or less of one or the other in some conditions than Spence and colleagues (1997), which used a joystick pro-
others, but still quite a general picture of what is going on. tocol with men with schizophrenia during PET scanning.
Hyperactivation of parietal and cingulate cortices was con-
Spence has investigated truth telling of a more direct fined to those men who had passivity delusions, and not
kind (Spence et al., 2008b) using an fMRI protocol with the seen in the others; it was no longer evident 4–6 weeks later
hypothesis that, focusing on the ventro-lateral prefrontal in those men whose passivity delusions had recovered. The
cortex and the anterior cingulate gyrus, response time study is however, tantalizing. Where are the replications?
would be longer and activation greater when endorsing How much can really be inferred from just seven men with
false rather than true statements. Applying this to a woman passivity delusions, six schizophrenic peers without such
with alleged Munchausen syndrome by proxy, when, as delusions and six healthy men? There has been little lon-
requested, she endorsed the accusations of others and not gitudinal study in this field to date, but surely that is the
when she gave her own story, they argue: way forward.
© 2014 by Taylor & Francis Group, LLC 313
13
Offenders with intellectual disabilities
Edited by Written by
Pamela J Taylor William R Lindsay
Gregory O’Brien
John L Taylor
Clinical and legislative Adaptive functioning
definitions Adaptive functioning is a broad concept referring to an
individual’s ability to cope with the day-to-day demands
The term ‘intellectual disability’ (ID) is gaining international of his/her environment. Thus, an assessment of adaptive
currency. Its use corresponds with the terms ‘learning dis- function must take into account a person’s age, environ-
ability’, commonly used in health and social services in the ment and cultural expectations.
UK, and ‘mental retardation’ in North America and in the
international (ICD-10, World Health Organization (WHO) Age of onset criterion
1992) and US (DSM-IV, American Psychiatric Association, The ‘age of onset’ criterion, according to general inter-
1994) diagnostic and classification manuals. Although peo- national consensus, means below the age of 18 years
ple with intellectual disability do not constitute a homoge- (American Psychiatric Association, 1994; British
neous population, ICD-10 and DSM-IV include three core Psychological Society, 2000), although ICD-10 does not
criteria for what they call mental retardation, and we will specify a criterion age.
refer to as intellectual disability:
●● significant impairment of intellectual functioning; Legislative terms
●● significant associated impairment of adaptive or social Mental health legislation in England and Wales,
Scotland, and Northern Ireland may apply to people with
functioning; and intellectual disability as well as to people with mental ill-
●● age of onset within the developmental period before ness or personality disorder, and this is generally true in
most countries around the world. The extent to which
adulthood. mental impairment or intellectual disability is explicit in
All three criteria must be present for a diagnosis of intel- the legislation varies between countries, and indeed also
lectual disability. over time. In the Mental Health Act (MHA) 1983, for exam-
ple, which covers only England and Wales, the term mental
Impairment of intellectual function disorder included the explicit and defined categories of
Impairment of intellectual function, particularly in a ‘mental impairment’ and ‘severe mental impairment’. The
forensic context, should be assessed using an individually Mental Health Act 2007, however, removes such sub-cate-
administered, reliable and valid, standardized test, such as gorization and replaces ‘mental impairment’ with ‘learning
the third edition of the Wechsler Adult Intelligence Scale disability’, while retaining the intent that such disorder
(WAIS-III; Wechsler, 1999). Using such a test, based on may render a person liable to detention, as, for example,
normal distribution of general intelligence, a ‘significant’ set out in the Code of Practice for England (Department of
impairment of intellectual functioning is conventionally Health, 2008). As is often the case, however, legal concepts
understood to be a score of more than two standard are not co-terminous with clinical concepts. The Mental
deviations below the population mean. According to this Health Act 2007 legal concept of learning disability requires
approach, an IQ score of less than 70 is indicative of intel-
lectual disability. Intelligence scores on such a measure abnormally aggressive or seriously irresponsible conduct
are used as the basis for distinguishing between levels of on the part of the person concerned…
severity.
314 © 2014 by Taylor & Francis Group, LLC
Crime and people with intellectual disabilities
to be coupled with Crime and people with
…a state of arrested or incomplete development of mind intellectual disabilities
… which includes significant impairment of intelligence
and social functioning In the late nineteenth and early twentieth centuries, criminal
behaviour and intellectual disability were firmly linked in the
and this distinction is retained in the guidance in the ideology of the menace of the feeble-minded (Trent, 1994).
English Code of Practice. The Code recommends specialist In 1889, Kerlin (reviewed by Trent, 1994, p.87) suggested that
assessment. (See chapter 3 of this book and chapter 34 of vice was not the work of the devil, but the result of physical
the English Code of Practice for further details.) infirmity and an inability to perceive moral sense both of
which were inherited and non-remediable. Kerlin’s views
The Adults with Incapacity (Scotland) Act 2000 and directly challenged the optimism of earlier authorities that
the UK Mental Capacity Act (2005) have clarified the role viewed people with intellectual disability as full of potential
of mental capacity in the responsibility of an individual for and remediable by suitable education. These more pessimis-
his/her own actions. Having intellectual disability and/or tic views became dominant for the next 50 years. Terman
another form of mental or cognitive impairment does not (1911), an author of one of the earliest IQ tests, wrote:
automatically absolve an individual from responsibility for
his/her actions. The assumption must be made that adults there is no investigator who denies the fearful role of
have capacity unless there is evidence to the contrary. The mental deficiency in the production of vice, crime and
approach to be taken in assessing an individual’s capacity delinquency… not all criminals are feeble-minded but
for decision-making is described in detail in the legisla- all feeble-minded are at least potential criminals (p.11).
tion and the accompanying secondary legislation, codes of Goddard (1922), author of The Criminal Imbecile, concluded:
practice and policy guidance. the results of the most careful studies indicate that some-
where in the neighborhood of 50 per cent of all criminals
People with intellectual are feeble minded (p.106).
disability detained in secure Despite the long alleged association between delinq uency
health service facilities in the UK and impaired intellectual functions, it is clear neither
whether people with intellectual disability commit more
In March 2006, in England and Wales, a total of 14,625 crime than those without (Lindsay et al., 2004) nor whether
patients were resident in NHS or independent hospitals those who do offend have any distinctive patterns in terms
under a section of the MHA 1983 (The Information Centre, of the nature or frequency of their offending (Holland, 2004).
2009). Of this total, 1098 (7.5%) were detained under the This lack of clarity is due in large part to methodological
categories of mental impairment/severe mental impair- problems in prevalence studies in this area (Sturmey et al.,
ment. Of the 12,132 patients detained in NHS facilities, 5.6% 2004). Inclusion criteria used in prevalence studies vary
(684) were categorized as having mental or severe mental and are often unclear. This can affect prevalence rates,
impairment, whereas the corresponding percentage in particularly if individuals with IQ scores in the ‘borderline
independent hospitals was 16.6% (414) of the total 2,493 intelligence’ range (70–85) are included. Differences in
patients detained there. During 2005–2006, a total of 25,740 measurement also mean that studies are not strictly compa-
people were admitted to NHS facilities (including high rable, for example whether intellectual disability is assessed
security hospitals) under the MHA 1983 (The Information through formal IQ tests, educational history or both. Further,
Centre, 2009). Most of these detentions (8,435) were under sampling bias and filtering effects result from the nature of
civil sections of the Act, less than 1% were in the mental the base population – whether this was a true community
impairment/severe mental impairment categories. Of the sample, a prison sample or a hospital sample (Holland et al.,
1304 detentions under court disposal or prison transfer 2002). Even within highly selected samples, estimates vary.
orders during this period, over 4% were categorized as A Scottish prison survey of the early 1990s, for example,
mental impairment/severe mental impairment. suggested that less than 0.5% of prisoners had intellectual
disability, while poor literacy and low educational attain-
The proportion of people in the general population ment were also problems (Davidson et al., 1995). Using the
with IQ scores under 70 is approximately 2.5% (assuming Quick test (Ammons and Ammons, 1962), the 1997 Office
a normal distribution), so almost double the expected of National Statistics survey of prisoners in England and
number of people with impaired intellectual function are Wales found that 5% of male and 9% of female sentenced
compulsorily detained in NHS facilities under the Mental prisoners together with 11% of remand prisoners, regardless
Health Act 1983; in the independent hospital sector the of gender, scored in the lowest band (25 or below), prob-
figure is as high as one in six detained patients. More recent ably indicating intellectual impairment (Singleton et al.,
NHS admissions under civil sections of the 1983 Act are 1998). Loucks (2007), in the most recent review of the lit-
in the anticipated range, but admissions under criminal erature, also acknowledged methodological problems, but
sections remain somewhat higher than expected (The
Information Centre, 2009).
© 2014 by Taylor & Francis Group, LLC 315
Offenders with intellectual disabilities
estimated that 20–30% of prisoners have intellectual dif- are now fewer options for their diversion from criminal
ficulties sufficient to impair their ability to benefit from ser- justice proceedings (Sturmey et al., 2004). This issue is
vices in prison, including offender treatment programmes. reflected in reports from the UK Prison Reform Trust, as
part of its No One Knows initiative to highlight the predica-
There is a dearth of good quality studies of other kinds ment of prisoners with intellectual disability and of those
too, for example comparing the prevalence of offending with less severe learning difficulties (Jones and Talbot,
in whole populations of people with intellectual disability 2010; O’Brien, 2008b). Loucks (2007) reports on studies that
with that in populations without (Sturmey et al., 2004) or of have indicated that people with intellectual disability may
comparative recidivism rates for offenders with and without account for up to 7% of prisoners while those with lesser
intellectual disability. Recidivism rates for offenders with or specific learning difficulties may account for up to 32%.
intellectual disability have generally been reported as high. In another report by the Prison Reform Trust (Jacobson,
In their study of 250 detained male intellectual disability 2008), recommendations were made for diversion of those
patients in the UK, Gibbens and Robertson (1983) reported individuals with intellectual disability away from the crimi-
a re-conviction rate of 68%, while Lund (1990) found a re- nal justice system to alternative provision. The Bradley
offending rate of 72% in a follow-up of Danish offenders with Report (Bradley, 2009) emphasizes that the needs of peo-
intellectual disability who had been detained on statutory ple with intellectual and developmental disabilities in the
orders. He also found a doubling of sex offending incidents criminal justice system require collaborative planning and
when he compared sentencing in 1983 to that of 1973. He person-centred care by a range of agencies.
suggested that this rise may have been a result of a policy of
de-institutionalization whereby people with intellectual dis- Although it would appear that service development for
ability are no longer indefinitely detained in hospitals, but offenders with intellectual disability has been consistent
he did not set this finding in the context of any change in with the development of all intellectual disability services,
overall sex offending rates over the same period. He thought entry into secure services often involves people being sent
that, given de-institutionalization, those with higher pro- by service commissioners to out-of-area facilities, whether
pensities towards offending would be preferentially dis- NHS or independent sector hospitals, with a resultant
charged, as perhaps also appearing more able; in fact, they drain on the resources of local services (Crossland et al.,
would also be more likely to be arrested at the scene of any 2005), and potential further exclusion of the clientele from
incident and possibly less able to defend themselves. their local community. In turn, this can lead to a lack of
alternative pathway development, local staff failing to
Linhorst et al. (2003) reported that, among 252 offend- develop knowledge and skills for helping these people, and
ers with developmental disabilities who completed a case geographical variability in the type and quality of provi-
management community programme, just 25% were re- sion available (Sturmey et al., 2004). It has been suggested
arrested within 6 months of case closure, while 43% of that the resources currently invested in out-of-area secure
those who dropped out of the programme were re-arrested services would be better directed towards developing local
during the same period. These figures are similar to those community-based support services (National Development
for large samples of general offenders studied over the same Team, 2007). Another concern about specialist secure
period, albeit in the USA; Langan and Levin (2002) found hospital development is that it means that institutions for
that among 300,000 prisoners of all kinds, 30% were re- people with intellectual disability are being re-established,
arrested within 6 months of release, while the re-arrest rate perhaps on a smaller scale, and specifically for people with
for 79,000 general offenders on probation was 43% (Langan offending or challenging behaviour, but nevertheless re-
and Cunniff, 1992). institutionalization is taking place.
The historical scaremongering concerning the associa Theories of offending
tion between low IQ and offending was influential in setting applied to people with
up large institutions for people with intellectual disability, intellectual disabilities
although there were also compassionate arguments for
this institutional policy, such as the notion that people with It is possible that low IQ itself may be a vulnerability factor
intellectual disability required separate, supportive cultures for offending, but the evidence is complex. There is a body
in order to live fulfilling lives. It was not until after 1970 of literature supporting a relationship between low IQ and
that the institutionalization philosophy changed in the UK, higher rates of offending (e.g. Hirschi and Hindelang, 1977;
and people with intellectual disability were (re)integrated Goodman et al., 1995) and there have been a number of
in community settings. In Europe and Australasia, imple- studies linking the combination of socioeconomic depriva-
mentation of deinstitutionalization policies has resulted in tion, antisocial influences and lower IQ to a higher rate of
substantial changes in all aspects of service organization delinquency. West and Farrington (1973) reported that only
and delivery in this field. These changes have been particu- 9% of multiple offenders had measured IQ scores of 100 or
larly evident in both the design and location of services for
offenders with intellectual disability, which work for those
with uncomplicated intellectual disability, although there
316 © 2014 by Taylor & Francis Group, LLC
Offenders with intellectual disabilities and additional diagnoses
greater while 28% of recidivist delinquents scored below 90. positive reinforcement in association with criminal sub-
Farrington (1995), reviewing a number of large scale studies cultures and family influences as well as the development
on the development of criminal careers, found that measures of self-control through appropriate social learning from
at 8–10 years old which are significant predictors of adult positive role models. Hirschi (1969) wrote that the suc-
criminality are troublesome behaviour, an unco-operative cess of social learning was dependent on four factors:
family, poor housing, poor parental behaviour and low IQ. attachment, commitment, involvement, and belief, using
the terms in his own specific ways. ‘Attachment’ referred
There is much less evidence, however, concerning the to the extent to which the individual identified with the
relationship between delinquency and IQ scores around or expectations and values of others within society, such as
greater than two standard deviations below the mean (i.e. IQ parents and teachers. ‘Commitment’ invokes a rational
≤70). McCord and McCord (1959) evaluated an early inter- element in criminality; individuals make subjective evalu-
vention with 650 underprivileged boys in Massachusetts. ations about the loss they will experience following arrest
The Cambridge–Somerville Youth Study was set up ‘to and conviction. ‘Involvement’ deals with the balance
prevent delinquency and to develop stable elements in the between engagement in ordinary activities, such as work,
characters of children’ (ibid, p.2). The boys were divided into education and other occupational activities, and opportu-
325 matched pairs and assigned to treatment and control nity to consider delinquency; the less that individuals are
conditions. There was a relationship between IQ and rates involved with the day-to-day activities of society the more
of conviction in that, for the treatment group, 44% of those likely they are to engage in criminal activity. He consid-
in the IQ band 81–90 had a conviction while 26% of those ered ‘belief ’ as the extent to which individuals accept the
with an IQ above 110 had a conviction. The 10% of individu- laws of society as being reasonable mores to which they
als in the lowest IQ group (less than 80), however, had a convic- would conform. There is a wealth of evidence supporting
tion rate of 35%, that is, lower than that recorded in the IQ this hypothesis, leading to recommendations that com-
band 81–90, but there was a twist. The highest percentage munity engagement and quality of life should be central
going to penal institutions (19%) was in the lowest IQ band, treatment components in programmes for offenders with
and none from the highest was sentenced to imprisonment. intellectual disability (Lindsay and Taylor, 2005). Measures
The results were similar in the control group. Maughan to promote pro-social influences and community integra-
et al. (1996) also found that the rate of adult crime among tion must be coupled with more specific interventions for
boys who had significant reading difficulties (an indication offending behaviours such as fire-setting, assault or sexual
of developmental and intellectual disability) was lower than offending. The evidence base thus informs clinical work
the rate of adult crime in the general population compari- with offenders with intellectual disability.
son group. This finding held true independently of psycho-
pathology and/or measured social abilities. This suggests Offenders with
that the relationship between intelligence and delinquency intellectual disabilities and
is not simple and linear when considering individuals 1.5 or additional diagnoses
more standard deviations below the mean.
People with intellectual disability have a high prevalence of
The work of Farrington (1995, 2000) and others – psychiatric disorders, including mental illness (Reid, 1972),
including the work of Patterson and his colleagues on other pervasive developmental disorders and/or neuro-
the relationships between family interactions and social psychiatric disorders related to the cause of their brain
learning (e.g. Snyder and Patterson, 1995) – established the damage (Gillberg et al., 1986).
strong links between social and environmental factors and
the development of criminality. Based on the results of their Intellectual disability and mental illness
longitudinal research, Huesmann et al. (1987) suggested
that, in early childhood, those with lower intellectual abili- The prevalence of mental illness, including the psychoses
ties are prone to developing aggressive behaviour because and the neuroses, among adults with intellectual dis-
of difficulties in learning more complex non-aggressive, ability is estimated to be in the region of 20% (Cooper
pro-social interpersonal skills. Aggressive behaviour, in et al., 2007; Taylor et al., 2004a). If challenging/problem
turn, may result in failure to develop intellectually, due to behaviour is included as a diagnostic category, the
its isolating and alienating effects that minimize oppor- reported prevalence jumps to around 40% (e.g. Cooper
tunities for effective education. Novaco and Taylor (2004) et al., 2007). Evidence that psychiatric disorders are com-
found evidence that this dual-process social learning model mon among offenders with intellectual disability comes
has applicability for people with intellectual disability and from both inpatient (e.g. Day, 1997; Novaco and Taylor,
significant offending histories. 2004) and community studies (e.g. Lindsay et al., 2002).
Findings may, however, be a function of these studies
The ability to identify with the values of society – or having been carried out in clinical settings. What is clear
lack of it – has long been a core concept in sociological
theories of criminality. Control theory (Hirschi, 1969) is
focused on the learning of criminal behaviours through
© 2014 by Taylor & Francis Group, LLC 317
Offenders with intellectual disabilities
is that psychiatric disorder is common in this popula- Intellectual disability and pervasive
tion and carries major implications for treatment and developmental disorders
management.
Pervasive developmental disorders, especially autism, are
Treating people with mental illness in the context common among adults with intellectual disability. The
of intellectual disability best evidence at present is that autism is no more common
Drug treatment for mental illness in offenders with intel- among offenders with intellectual disability than people
lectual disability follows the same principles as in all with intellectual disability who do not offend (O’Brien,
psychiatric practice, but framed according to their needs 2002b). The management of those adults with intellec-
arising from their developmental disabilities. First, and tual disability and autism in offender services, however,
most importantly, there are three key principles: can be particularly daunting, as the interaction of autism
1. start low; with offending is complex. Autism may result in offending
2. go slow; and through obsessional fixations but, more often, through
3. avoid polypharmacy. mutual misunderstandings between the person with
The notion of ‘start low’ is intended to reflect the fact that autism and others, and resultant mismatch in their behav-
doses of most psychotropic medications in adults with iours; sometimes the two mechanisms act together. Also
intellectual disability should be started at around half the common is the propensity of people with autism to panic,
usual adult dose (O’Brien, 2002a). This is necessary partly and to have sudden aggressive outbursts in this context.
to avoid common major side-effects such as movement Treating offenders with pervasive developmental
disorders, sedation, and other dose-related problems, all disorders in the context of intellectual disability
of which are more likely to appear in this group. Secondly, These matters are relevant to the management of people
the suggestion to ‘go slow’ refers to incremental tapering on the autism spectrum who have offended. First, drug
of dosage. It is recommended that dose is altered over a treatment can be helpful. This is particularly so for those
period of about twice the time usual in mainstream psy- individuals who experience anxiety and are likely to panic
chiatric practice. This is both to avoid side-effects and to or become aggressive. They often respond well to low
monitor very carefully changes in clinical response, which dose antipsychotic therapy, such as risperidone. Selective
can take longer to appear in this population than among serotonin reuptake inhibitors (SSRIs) may also be helpful
people with illness alone. Finally, avoidance of polyphar- in this context. Either medication may be helpful for those
macy is a reflection of the fact that, while several drugs are individuals on the autism spectrum whose offending is
often required in this population, numbers should be kept driven by obsessional fixations.
to a minimum. Adverse drug interactions are more likely
among people for whom metabolic disorders are common. Individuals with autism are likely to find the experience of
inpatient treatment or offence-related interventions anxiety
It is important to consider the timing of drug use. Very provoking, and symptom worsening is common at times dur-
often, when first admitted into services, people are in a ing such treatment. Consequently, in addition to optimizing
state of distress and intra-psychic disorganization, partly inter-personal functioning and psychological organization
due to mental illness and partly to a range of other factors. at the start of a course of treatment, through drug therapy, it
Prompt treatment can then be very helpful in preparing the is important to monitor the progress closely throughout, in
way for structured offence-related interventions, described order to detect early any need for modifying medication or
later in this chapter. An essential step in engaging the adding support over the course of the programme.
patient in a treatment milieu which enables other interven-
tions to proceed lies in helping him/her to become more Intellectual disability in people
psychologically organized. In addition, in the course of with epilepsy and other
offence-related treatments, it is common for problems of neuropsychiatric disorders
anxiety and mood problems to worsen. Indeed, in the face
of the emotions engendered in such therapy psychotic As intellectual disability from childhood is generally
symptoms and may re-emerge and will also need careful explained by very early brain damage, developmen-
prescription and management. tal disorders or a combination of the two, a range of
neuropsychiatric disorders is common among adults with
For further reading about management of mental intellectual disability, epilepsy among them (Tyrer et al.,
illness, and in particular the drug treatment of men- 2006). As with mental illness in people with intellectual
tal illness in the context of intellectual disability, see disability, the key is to provide optimal treatment as early
O’Brien (2002a), and for further notes on the manage- as possible. This helps the individual to access their latent
ment of pervasive developmental disorders in this con- cognitive resources, facilitates engagement in other treat-
text of younger offenders with intellectual disability see ments, including psychological interventions, and prevents
O’Brien and Bell (2004).
318 © 2014 by Taylor & Francis Group, LLC
Genetic disorders, intellectual disability and offending: Genotypes and behavioural phenotypes
further damage. Perhaps contrary to popular belief, it is classifications (WHO, 1992) presented in a manner consist-
as uncommon in clinical forensic practice to encounter ent with that reported by Lindsay et al. (2006a), with par-
offending during epileptic fits or post-ictal confusion or ticipants from high security settings having a higher rate
fugues among people with intellectual disability as it is of these diagnoses than participants from the other two
more generally (see also chapter 12). It is far more common settings. In addition, having an ICD-10 dissocial personal-
find individuals who may be behaviourally disturbed and/ ity diagnosis was a significant predictor of level of security.
or aggressive as part of irritability in the context of poor
epilepsy control, highlighting the need for prompt and Genetic disorders,
effective anti-convulsant therapy. intellectual disability and
offending: Genotypes and
Intellectual Disability and behavioural phenotypes
Personality Disorders
The links between specific genetic syndromes and
Reported prevalence of personality disorders among offending behaviour are as complex as they are some-
offenders with intellectual disability has varied quite times controversial. Although the process of drawing
widely. In reviewing this literature, Lindsay (2007) made inferences from individual research studies as well as
several recommendations for considering their nature in from their combination is dealt with at length in chap-
this group, including: ter 8, we deal with some of the issues here too. This is
a. greater use of behavioural observation and informant because there are some matters of particular importance
to people with intellectual disability generally, and also
information to make diagnostic classifications; and there are some conditions in this field which are more
b. greater awareness of cultural factors affecting the specific than in the field of mental illness in the inher-
ited component of the resultant difficulties, conditions
diagnosis for this group. People with intellectual such as fragile X or Prader–Willi syndrome (see below).
disability have often lived more restricted lives than Nevertheless, any proposition that inheritance of a spe-
those in the mainstream population and consequently cific genotype might result in an individual’s carrying
have less opportunity to experience a range of social out a particular offence is fraught with difficulty. The
and sexual relationships which, in turn, may have history of eugenics renders it unpalatable, but, at a more
hindered personality development. This must be taken practical level, the possible intervening variables are
into account when making a diagnosis of personality extensive. Notwithstanding the need for extreme caution
disorder, as must the lower levels of occupational to be observed in any proposal that a particular genotype
activity, and higher levels of necessary dependency. might result in offending behaviour, recent research
Other contextual factors too, such as higher levels of on behavioural phenotypes has gone to great lengths to
suggestibility, may affect responses to questionnaire clarify mechanisms of expression of the genotype. This
assessments. may be of value to inform and pave the way for care and
In a large multi-centre study, Lindsay et al. (2006a) com- intervention. In this context, the useful concept of a
pared rates of personality disorder in offenders with intel- behavioural phenotype may be defined as ‘a characteristic
lectual disability in community, low/medium secure and pattern of social, linguistic, cognitive and motor obser-
high secure settings, using DSM-IV criteria (American vations which is consistently associated with a biologi-
Psychiatric Association, 1994). The reliability between cal/genetic disorder’ (O’Brien and Yule, 1995).
independent raters was generally over 80%, and the assess-
ment and diagnosis of personality disorder among people The observed phenotype of any one genetic condition
with intellectual disability appeared valid. The prevalence varies from time to time and from person to person: there
of personality disorder across the study population was is nothing in this definition that suggests that the behav-
39%. Perhaps unsurprisingly, antisocial personality disor- ioural features of a genetic syndrome are inevitable, fixed,
der (ASPD) was the most frequently diagnosed, and signifi- or irremediable. When considering the genetic basis of the
cantly more individuals receiving this diagnosis were in the behavioural phenotypes observed, and how these might
high secure setting than in the other two settings. There contribute to offending behaviour, certain themes are
were no diagnoses of dependent personality disorder, apparent. These are:
which might have been anticipated given the extent of the 1. the role of generalized intellectual disability;
study group’s developmental delay. When data for the three 2. the relationship between genotype severity and pheno-
groups were combined, personality disorder was positively
associated with risk of violence according to the Violence type severity;
Risk Appraisal Guide (VRAG; Quinsey et al., 1998). 3. the different considerations which apply in respect of
Evaluating data from the same study groups, Hogue
et al. (2006) reported that ICD-10 personality disorder progressively deteriorating central nervous system
syndromes and non-progressive conditions;
© 2014 by Taylor & Francis Group, LLC 319
Offenders with intellectual disabilities
4. the mechanism of action of discrete genes, in certain the phenotype, especially in terms of degree of intellectual
common and/or familiar syndromes; disability. This has major implications for their behaviour.
When it comes to recording and interpreting genetic find-
5. behavioural phenotype expression through gene– ings with respect to offending behaviour, one important
environment interactions. issue is to be clear whether the genetic finding displays
such quantitative elements.
The mediating role of generalized
intellectual disability The Different Considerations
in Respect of Progressively
In considering the pathway from genotype to behavioural Deteriorating Cns Syndromes and
phenotype to offending behaviour, the first issue is whether Non-Progressive Conditions
the observed behaviour is mainly a reflection of the breadth
and/or severity of intellectual disability which is typical of Special consideration must be given to those conditions
the genetic syndrome in question. Overall, there is a strong which feature progressively deteriorating CNS function,
association between occurrence of disturbed behaviour because the manifest behavioural phenotype in such syn-
and degree of intellectual disability (Gillberg and O’Brien, dromes changes over time, with implications for the occur-
2000). In the assessment of offending behaviour, the com- rence of disturbed and/or offending behaviour. With loss of
mon occurrence of restless and over-active behaviour, often skills, and disorganized and often (early in course) disinhib-
compounded by excitability and/or aggression, tends to be ited behaviour, the risk of aggressive offending behaviour
more problematic at lower than higher levels of tested IQ, may be increased. In non-deteriorating conditions, by
the exception to this being at the very lowest end of the IQ contrast, improvement and development are the norm,
spectrum. Here, many are so disabled that disturbed behav- through education and other influences aimed at optimiz-
iour is not an issue, but rather apathetic and listless behav- ing self-organization and related skills.
iour dominates the presentation. This tends not to apply to
other behaviours figuring prominently in some of the more The mechanism of Action of
specific syndromes, such as the insatiable over-eating in Discrete Genes in Certain Common
Prader–Willi syndrome and the compulsive self-injury in and/or Familiar Syndromes
Lesch Nyhan syndrome (see also O’Brien and Yule, 1995).
In some genetic syndromes, certain behaviours present When the action of a single gene has a strong effect, then
over the whole IQ range, as, for example, in Prader–Willi these simple principles guiding the pathway from genotype
syndrome. Thus, while many of the behavioural anomalies to phenotype begin to unravel. In this case, the essential
which figure among the behavioural phenotypes of genetic steps are as follows (O’Brien and Pearson, 2005):
syndromes mainly reflect low IQ, some have a more specific ●● genes code for proteins, that is they provide a template
organic basis.
for the production of a specific protein;
The implication of these observations is that in the ●● proteins design, build and develop all body systems;
assessment of any offending behaviour in the context of a
genetic syndrome which has a putative behavioural pheno- they comprise the matrix against which other tissue
type, clinical assessment should include expert psychologi- constituents are taken up and laid down in the body,
cal testing of intellectual function. and proteins regulate this, in all body tissues;
●● including the brain;
The relationship between genotype ●● which controls behaviour;
severity and phenotype severity ●● so, any variation in a gene involved in any aspect of
brain development or maturation may result in a
Since the beginnings of genetic research, one fundamental behavioural phenotype.
tenet has been that more extreme variations in genotype Some insight into the nature of the mechanisms of expres-
are associated with more extreme variations in phenotype. sion of gene–behaviour associations can be gleaned from
Many of the whole chromosome replication syndromes, consideration of three of the most widely studied con-
for example, feature very severe levels of intellectual dis- ditions, namely: fragile X syndrome, in which a micro-
ability, and, in some cases, also other phenotypic features anatomical effect on neuronal dendritic arborization over
such as life-threatening congenital cardiac abnormalities the course of brain maturation is seen; Lesch–Nyhan syn-
and other health problems. Mosaicism, where there is drome, where the gross impact of an aberrant metabolite
admixture of normal and abnormal cell lines, also impacts on the whole organism is manifest; Prader–Willi syndrome,
directly on phenotypic expression; there is a direct cor- in which a gender-specific imprinting effect on a psychosis
relation between the degree of mosaicism and phenotypic gene is postulated (see also below).
expression. Thus, individuals who have a greater propor-
tion of normal cell lines typically show milder variants of
320 © 2014 by Taylor & Francis Group, LLC
Genetic disorders, intellectual disability and offending: Genotypes and behavioural phenotypes
Behavioural Phenotype Expression affected individual should include assessment of the cogni-
Through Gene–Environment Interactions tive and behavioural traits described above.
The behavioural expressions of these (primarily) CNS Fragile X syndrome
genes are not simple direct results of gene on behaviour. The gene underlying fragile X syndrome (FMR-1) is located
In the definition of behavioural phenotype, the pattern of on the distal arm of the X chromosome, Xq27.3, and is
behaviour, given a particular genotype, is characteristic, not associated with a large expansion of a sequence of CGG
universal, or non-mutable. All behaviour – especially offend- (cytosine–guanine–guanine) trinucleotide repeats. This
ing behaviour – is to an extent contingent on the personal gene exerts its impact on brain development by regulat-
environment and the reactions of others, but environment ing neuron dendritic arborization. In affected individuals,
is thought to have a more powerful role in some conditions there is failure of inhibition of dendritic arborization, which
(see also chapter 8). results in too many inter-neuronal connections: effectively,
a reduction of the pruning effect on cerebral structure,
There is now growing evidence, however, that demon- which is part of normal development over adolescence.
strates how reactions of parents and carers to behaviour in Consequently, the brain of affected individuals is larger –
the developing individual has a shaping effect in even some around 10% heavier – than the normal young adult brain,
of the most florid features of behavioural phenotypes. This but many of the connections detract from functional adap-
has been shown elegantly in the context of research on an tation, rather then add to it.
important behaviour which had previously been thought
to be independent of personal or social environment – There is a direct correlation between the length of
that is, the socially inappropriate laughter of children with the repeat CGG sequence, and the severity of phenotypic
Angelman syndrome (see below). Oliver et al. (2002) found expression, in physical, intellectual and behavioural terms.
evidence that, on the contrary, the laughter of these chil- Phenotypic expression of fragile X syndrome depends on
dren is heavily dependent on reactions by others, especially the sex of the affected individual. Intellectual disability in
the social reactions of their parents to their laughter in males is typically mild to moderate, while females – having
their early years. Such findings may indicate new directions one normal X chromosome – generally show IQ in the low
for management and facilitating optimal development normal to borderline range. Affected boys show a combina-
among affected individuals. tion of an atypical form of autism spectrum disorder, and
an attention deficit hyperactivity disorder (ADHD)-type
For each of these important conditions, we describe pattern of overactivity. Repetitive behaviour and social
common considerations with regard to offending behav- anxiety are prominent, but theory of mind test results are
iour, in particular the implications of the behavioural phe- less impaired than in typical autism. As they develop, boys
notype for assessment of offending behaviour. become less overactive, indeed, many are quite underactive
and listless by adulthood, but the autistic-type features and
Angelman syndrome social anxiety are more persistent among men (Turk, 1992;
This syndrome shares the same deletion on chromosome Hagerman, 2005). Girls and women generally have few
15q (11–13) as Prader–Willi syndrome, but in Angelman autistic features, but some social anxiety.
syndrome this is maternally derived. The intellectual dis-
ability of affected individuals is usually in the severe to Clinical note on assessment of offending
profound range. Lack of speech is characteristic. Facial behaviour in a person with fragile X
features are usually characteristic, with a prominent jaw, Most such people are not aggressive, nor are they likely to
wide mouth, with widely spaced teeth and thin upper lip, engage in offending behaviour. Nevertheless, as such syn-
flat occiput, mid-facial hypoplasia and deep-set eyes. The dromes go, fragile X, is common, and the condition’s behav-
behavioural features include general motor restlessness ioural phenotype of social cognitive deficits, which feature
and overactivity, short attention span, ataxia, and, notably, ADHD, and autistic traits can result in an affected individual
a prominent pattern of episodic excessive and socially being involved in behaviour which may lead to allegations of
inappropriate laughter. This resulted in the now discred- offending. In assessment of any alleged offending behaviour
ited eponym – ‘happy puppet syndrome’, which families here, the first issue will usually be to clarify whether either
and carers find unhelpful, and insulting. of the two common developmental disorders – ADHD or
Clinical note on assessment of offending autism – is also present. If so, the task then is to explore the
behaviour in a person with Angelman syndrome role of the disorder in the offending behaviour.
Behaviour may be misinterpreted as deliberately socially
awkward, and may be regarded as offensive and threat- The behaviours of people with fragile X syndrome, in
ening. Any allegation of offending behaviour – especially common with others who are on the autism spectrum,
of interpersonal offending behaviour – on the part of an are also liable to provoke unfortunate or even bullying
© 2014 by Taylor & Francis Group, LLC 321
Offenders with intellectual disabilities
reactions on the part of others. This common problem onwards. If unchecked, the resultant obesity can be crippling
is also relevant to the assessment of any allegations of and life-threatening. This is preceded in early life by difficul-
offending, whether it is the alleged offender or the victim ties establishing eating and, often, failure to thrive. With
who is affected by fragile X syndrome. Careful analysis of careful attention and strict supervision, weight management
the interpersonal actions and reactions is important to an from childhood into late adulthood can be attained.
understanding of what happened and to developing strate-
gies for preventing repetition. The behavioural and psychiatric features of the con-
dition have been subject to close study. There is a well-
Lesch–Nyhan syndrome documented distinctive pattern of self-injury, which
Lesch–Nyhan syndrome occurs almost exclusively in takes the form of skin-picking (Boer and Clarke, 1999).
males, caused by gene mutations on the X chromosome, Affected individuals often have mood problems, with
resulting in an almost complete lack of activity of the anxiety and depression. Paranoid psychosis is common
enzyme hypoxanthine guanine phosphoribosyl transferase in Prader–Willi syndrome, compared with other indi-
(HPRT). This enzyme normally plays a key role in the recy- viduals with a similar degree of intellectual disability.
cling of the purine bases, hypoxanthine and guanine. When Genetic family pedigree studies have revealed that,
it is inactive purines are not salvaged, but rather degraded, where this psychosis presents in paternal deletion cases,
while, also, as a compensatory mechanism, increased it is merely a reflection of familial heredity. In those
synthesis of purines takes place (Deutsch et al., 2005). in whom there is maternal disomy of chromosome 15
Uric acid is overproduced, leading to clinical features of only, however, not all individuals develop the psychosis,
gout, and, given the stage of development of the affected although most do, and this is independent of familial
person, intellectual disability, spastic cerebral palsy, invol- heredity (Boer et al., 2002).
untary, choreoathetoid movements and aggressive behav-
iour, including self-mutilation, all exacerbated under Clinical note on assessment of offending
stressful circumstances (Nyhan, 1976; Palmour, 1983). Self- behaviour in a person with Prader–Willi
mutilation is severe and compulsive, with unrestrained syndrome
patients biting off digits and parts of their lips, even In a condition in which insatiable over-eating occurs,
though they experience pain and scream while doing so. allegations of offending behaviour related to obtaining
Aggression is also compulsive and sufferers have been food might be expected, or, in children, sweets or candy.
heard apologizing while hurting others. The mechanisms In clinical experience, however, this is not common out-
behind the neurology of the condition are incompletely side the home. The mood problems and psychosis which
understood, but may include abnormalities of dopamin- commonly present in the condition are, in some cases,
ergic, GABAergic and glutamatergic neurotransmitter sys- important considerations in the assessment of offending
tems (Deutsch et al., 2005). behaviour of affected individuals.
Clinical note on assessment of offending
behaviour in a person with Lesch–Nyhan There have been court cases after people with Prader–
syndrome Willi syndrome have died in states of extreme obesity, and
The aggressive behaviour is profoundly distressing and their carers have been called to account for how they have
potentially very damaging to both the affected individu- dealt with the known trait of over-eating. Such cases have
als and anyone with whom they come into close contact. not resulted in prosecution: it is accepted that limiting food
The behaviour is so obviously part of the syndrome, that intake in these individuals may be very challenging. It is
it would be almost unheard of to prosecute a sufferer. The likely, however, that such cases will continue to be tested in
condition has, however, been of particular interest to those court – particularly as successful techniques of managing
studying genetic mechanisms relevant to aggression. the over-eating of the condition become more widely avail-
able, and it might be suggested that extreme obesity in the
Prader–Willi syndrome syndrome is avoidable.
In most cases, Prader–Willi syndrome is caused by a dele-
tion on the paternal chromosome 15 (q11–13) (a deletion Tuberous sclerosis
on the same chromosome of maternal origin results in Tuberous sclerosis is a complex autosomal dominant
Angelman syndrome, see above). intellectual disability is neurocutaneous multisystem condition, most often involv-
variable, from severe through to normal IQ range, but most ing either chromosome 9q34.3 or 16p13.3. The typical
affected individuals are in the mild intellectual disability presentation of the full-blown syndrome is of hamartias (a
range. Irrespective of IQ, all affected people show a pattern of focal malformation of disorganized tissue types), hamarto-
insatiable over-eating of carbohydrates from mid-childhood mas (benign tumours), true neoplasms, skin lesions, intel-
lectual disability, behavioural abnormalities – especially
autism, ADHD and disturbed sleep – and seizures. The
clinical presentation of the condition is, however, extremely
322 © 2014 by Taylor & Francis Group, LLC
Genetic disorders, intellectual disability and offending: Genotypes and behavioural phenotypes
variable, from individuals who only have mild cutaneous Early research findings in XYY syndrome concluded
lesions, and are often undiagnosed, through to those who that affected individuals had an exceptionally aggressive
have the most severe forms of the condition, depending and violent behavioural profile and were predisposed
upon the location and extent of the lesions, particularly in to criminal activities (Sandberg et al., 1961; Jacobs
the brain and kidney. About half have intellectual disability. et al., 1965; Hook, 1973). These conclusions were based
In most individuals the disorder is non-progressive, but for on findings in psychiatric and penal institutions, with
a minority, with brain and kidney involvement, progressive inherent sample biases. Assumption of a direct relation-
degeneration occurs. ship between the extra male chromosome and criminal-
ity oversimplifies the genotype–phenotype relationship.
Clinical note on assessment of offending Theilgaard (1984), in a true community sample, found
behaviour in a person with tuberous sclerosis little difference in violence or anti-social behaviour
Some people with tuberous sclerosis, especially young between XYY, XXY and healthy control men; although
adults, can be very aggressive and disturbed in behaviour. the XXY men had more often been arrested, they had
The behaviour is more often restless and disorganized than not more often been convicted. The XYY men reported
deliberately violent. Autism and ADHD in the condition more difficulties in childhood, including teasing or chal-
may be the correlates of alleged offending in some cases. lenge by other children on account of their size. Ratcliffe
(1999) concluded that there is a moderate, but impor-
Williams syndrome tant, increased incidence of antisocial behaviour in XYY,
This syndrome is associated with a microdeletion on chro- but this is by no means invariable, and is not a simple
mosome 7. The microdeletion accounts for disruption of or inevitable effect of having an extra Y chromosome.
the elastin gene, which contributes to the vascular and Society’s response to the large physical stature, intel-
connective tissue pathology associated with the syndrome. lectual disability, and tendency to impulsivity of affected
Intellectual disability is usually in the moderate range. individuals all operate as intermediary risk factors.
There is a characteristic ‘elf-like’ facies, with prominent
cheeks, a wide and long philtrum (infra-nasal depres- In fragile X syndrome, the early research indicated
sion), flat nasal ridge, and heavy orbital ridges. There is a that the phenotypic expression resembled that of autism
distinctive cognitive profile, with impaired visuo-spatial (reviewed in Gillberg and O’Brien, 2000). This resulted in
processing abilities, but relatively superior verbal abilities. the application of the term AFRAX syndrome (autism–
The social behaviour of affected individuals features a fragile X), with the proposition that the fragile X (now
superficial pattern of affable conversation, which has been known as FMR-1) gene might be an ‘autism gene’.
referred to as ‘cocktail party’ syndrome, which tends to Subsequent research has revealed important differences
mean that the general abilities of affected individuals are between the fragile X syndrome behavioural pheno-
over-estimated by others, especially on first meeting. type and classical autism, or Kanner syndrome. While
individuals with fragile X syndrome do have social and
Clinical note on assessment of offending language difficulties which are on the broader autism
behaviour spectrum, in other respects the phenotype is unique –
The unusual social behaviour of the condition, coupled with social anxiety being particularly prominent.
with its characteristic appearance, are so prominent that
affected individuals are often referred to psychiatric ser- These syndromes highlight some of the important les-
vices, usually in childhood (Howlin and Udwin, 2006), sons of direct relevance to the consideration of offending
where one of the key concerns is sexual vulnerability on behaviour among people with specific genetic syndromes.
the part of young women. Their affable and misleadingly These include the need to consider:
effective social behaviour may predispose to being victims ●● whether any claims for the genetic basis of behaviour
of crime more generally. In the assessment of any alleged
offence, a full cognitive and social assessment of the are derived from skewed or biased samples;
affected individual is crucial, in addition to the standard ●● whether appropriate clinical assessment and
investigations of any such event.
measurement approaches have been employed;
Historical Lessons From Research ●● whether sufficient attention has been paid to the
Into Two Genetic Syndromes
complex intervening social and interpersonal variables
The history of research into behavioural phenotypes between any psychological traits which are linked to the
provides lessons in the need for a cautious and careful syndrome in question, on the one hand, and manifest
approach, particularly in clinical diagnosis and assessment. offending behaviour, on the other.
By these means, the clinician is well-placed to bring genetic
syndromes of intellectual disability and their behavioural
phenotypes into the assessment and understanding of
offending behaviour, where relevant.
© 2014 by Taylor & Francis Group, LLC 323
Offenders with intellectual disabilities
Alcohol and substance misuse disability were either alcohol abusers or were intoxicated
at the time of the offence. In a further investigation, Hayes
In general, studies have found that people with intellectual (1996) reported that of individuals with intellectual dis-
disability are less likely to misuse alcohol or other drugs ability appearing before two rural courts in New South
than the mainstream population (Stavrakaki, 2002). In a Wales, 90% had consumed some alcohol on the day of the
study of 329 people with intellectual disability, Rimmer alleged offence. A further study in Australia (Klimecki et al.,
et al. (1995) found that less than 5% of individuals used 1994) reported that 45% of first offenders, 71% of second
alcohol at all. Consistent with these findings, Sturmey offenders, 67% of third offenders and 88% of fourth offend-
et al. (2003) estimated that fewer than 5% of people with ers had a history of substance abuse. Cockram et al. (1998)
intellectual disability have a co-existing substance related reported that two-thirds of a sample of 20 offenders with
disorder and that those individuals who do have such a intellectual disability were identified by family members as
disorder are likely to be in the mild or borderline ranges misusing substances. In a larger scale study of 247 offend-
of intellectual disability, since these people are well able ers with intellectual disability in Scotland, non-sexual
to access sources. Taggart et al. (2006) found that 67 offenders had a far higher rate of alcohol related crime
individuals with intellectual disability in a UK region had than sexual offenders (Lindsay et al., 2006a). Overall, these
significant substance related problems – this translates to findings, mainly from Australia, suggest that while alcohol
an estimated 0.8% of the adult population with intellec- abuse may be fairly unusual among people with intellectual
tual disability. This is likely to be an underestimate, how- disability overall, it is far more common among offenders
ever, as the study involved only people known to services. with intellectual disability.
An important effect of de-institutionalization for people
with intellectual disability is that they have thus become Care pathways for offenders
able to access alcohol and drugs like everyone else and, as with intellectual disabilities
availability of substances has increased, so for people with
intellectual disability living in the community prevalence The care pathways of offenders with intellectual disability
of substance misuse has increased somewhat in recent share, necessarily, many of the characteristics of the care
years (Annand and Stavrakaki, 2007). pathways of all other offenders. Treatment and manage-
ment strategies occur within a given legal and national
Despite this relatively low prevalence, clinicians (e.g. policy context, and there is a limited range of options.
Krishef and DiNitto, 1981) have suggested that, within this There are, however, certain differences. These are, to some
minority of substance misusers, patterns of misuse may dif- extent, a reflection of the nature of people with intellec-
fer from those in the general population. People with intel- tual disability and their offending behaviour, but the care
lectual disability tend to be older at first use and have less pathways mainly reflect societal attitudes towards them,
physiological dependency but, when problems do arise, they as evidenced in the development of services and their
are likely to be similar to those experienced by the general underlying policies. The complex interplay of these factors
population including social, biomedical, occupational and may be drawn together in consideration of the typical steps
family difficulties. Westermeyer et al. (1996) found that, to be considered in building a care pathway for an offender
although individuals with intellectual disability generally with intellectual disability.
used less alcohol, the lower quantities nevertheless precipi-
tated the typical problems associated with misuse amongst Step One – The Offence, Including
their intellectually average peers. A particular problem seems Context and Contributory Factors
to be that when people with intellectual disability use alco-
hol, a far higher proportion – almost 50% – than in the gen- Elsewhere in this chapter, we have described the character-
eral population also misuse other substances (McGillicuddy istics of the offending behaviour of people with intellectual
and Blane, 1999). McGillivary and Moore (2001) compared disability, and how this relates to their intellectual and cog-
the rate of self-reported alcohol and substance use in 30 nitive function, educational attainment and social context.
offenders with intellectual disability to a control group of 30 Their offending profile does, to some extent, set the scene
non-offenders with intellectual disability. The offender group for a care pathway that shares some of the characteristics
reported greater use of both legal and illegal substances than of that for offenders with more average abilities, while hav-
the controls and many reported that they had been under ing differences in emphasis. A real case for differences in
the influence of alcohol or drugs at the time of committing care pathway is clearer for those with more severe intel-
their offences. Significant knowledge about substances was lectual disability, whose offending behaviour is also more
found in both groups, although the offenders had greater divergent from that of the general population, and whose
overall awareness about alcohol and drugs. developmental characteristics are such that mainstream
services are inappropriate to their needs (Holland, 1997).
In their report on prisoners with intellectual disabil-
ity in New South Wales, Australia, Hayes and McIlwain
(1988) found that around 66% of offenders with intellectual
324 © 2014 by Taylor & Francis Group, LLC
Care pathways for offenders with intellectual disabilities
Step two – Recognition and Reporting Step Four – Diversion From Custody
Although offending in the wider community by more Diversion from custody for offenders with intellectual disabil-
able offenders with intellectual disability may leave them ity entails referral to an appropriate specialist team, offering
more susceptible to arrest, there is some evidence of expertise in assessment, formulation and treatment. Such
under-reporting of their offending behaviour in residen- assessment/formulation is best carried out as early as possible
tial settings. The reluctance of carers and/or statutory in the care pathway – and is often carried out prior to involve-
authorities to report to the police is most marked for ment of police and other statutory agencies, particularly in
those with more severe disabilities, where ability to form respect of offenders already known to services. Assessment
intent is genuinely more severely impaired. In one study, does not necessarily result in automatic acceptance into spe-
Lyall et al. (1995) investigated such reporting by staff in cialist treatment programmes. In fact, careful assessment and
community residential services. They found that staff formulation will typically result in a range of options, including:
declared themselves to be unlikely to record and report
even quite serious offending behaviour by their residents, No further action
including assaults resulting in serious injury. Such atti- This is where there is no indication of need for a clinical
tudes, which may be described as misplaced tolerance disposal – for example, in respect of an individual who has
on their part, do not, in fact, serve the interests of their previously been through a programme of treatment, and
residents well, although they tend to be reinforced by the that further intervention is not warranted.
police and Crown Prosecution Service (CPS) alike. Among
people who may already be struggling to understand Referral to mainstream (non-intellectual
social boundaries and mores, lack of action in this respect disability) forensic mental health services
may be regarded by transgressors as endorsement of their which is often indicated for intellectually higher func-
behaviour, and has been identified as one important fac- tioning individuals, who are able to access and use such
tor in the maintenance of ongoing offending behaviour services.
among adult offenders with intellectual disability (O’Brien
and Bell, 2004). Consideration of community/outpatient
treatment and support by specialist intellectual
Step Three – Criminal Prosecution disability services
To be successful, such services typically require the invo-
A variety of prevalence studies, using various method- cation of some statutory power, whether under crimi-
ologies, indicate that people with intellectual disability nal justice (e.g. probation) or mental health legislation
may be over-represented among those seen at police (e.g. Lindsay et al., 2006). Many clinicians involved in such
stations, which findings are in line with the observations work prefer to operate such services within a clinical jus-
of the extent not only of offending behaviour among tice framework, because of the particular statutory agency
this population but also the likelihood that this will be involvement which that option entails (Holland et al.,
immediately identified, as just outlined. The numbers of 2002), and the positive results from such community-based
adults with intellectual disability who are known to ser- work (e.g. Lindsay et al., 2006b; Linhorst et al., 2003).
vices who are actually charged and convicted of offences,
however, would appear to be low (see Holland et al., 2002 Consideration of admission to inpatient (secure)
and Murphy and Mason, 1999 for reviews). Overall, the provision
reluctance by care workers to record and report offending Most developed countries have some kind of secure hospi-
behaviour by people with intellectual disability is paral- tal/health service facility available for offenders with intel-
leled by similar reluctance on behalf of police and CPS to lectual disability when their needs cannot be met within
process such offences. Indeed, these reviewers report that, prison, generic (non-intellectual disability) forensic mental
in a number of studies, staff stated that their disinclina- health services or specialist community intellectual dis-
tion to report any such offending was partly based on their ability services. In a minority of cases there is a need for a
expectation that no prosecution would ensue without secure setting in order to contain a significant level of risk
a huge struggle on their part. It is important to engage of harm to self or others. There is current concern that the
the police and prosecution services, at the highest levels, pathways into and between services for this group of people
in dialogue about such circumstances so that each can are neither dependent on the nature of the offending behav-
understand the difficulties and limitations of the other in iour nor on the clinical needs of the individuals concerned,
ensuring that community safety extends to families, other but rather on the availability and accessibility of local health
care workers and clinical staff while continuing effort and social care services (O’Brien, 2004). Admission of a per-
to enhance the management and safeguard the rights of son into secure provision frequently entails a long period of
those people with intellectual disability who do offend. time a long way from his/her usual home, cutting any useful
ties and making discharge back to the community less likely.
© 2014 by Taylor & Francis Group, LLC 325
Offenders with intellectual disabilities
At present, therefore, too many individuals remain offenders with intellectual disability, Novaco and Taylor
in long-term secure care. As suggested earlier, the poli- (2004) found that 47% had been physically violent on at
cies of de-institutionalization and community care which least one occasion following admission to the specialist
produced positive change in service provision for people forensic service. This finding was replicated by McMillan
with intellectual disability took little or no account of the et al. (2004) in a study also involving offenders with intel-
subgroup who offend. In the UK at least, developments of lectual disability detained in secure settings. Anger is rarely
hospital-based services, similar to those for people with assessed routinely in intellectual disability services, but
mental illness, have been provided on an ad hoc basis Law et al. (2000) found that more than 60% of people with
rather than strategically planned for holistic care along intellectual disability referred to a community-based ser-
a continuum of need. The institutional model has crept vice for challenging or offending behaviours were found to
back, with inadequate complementary community services have clinically significant anger.
(Murphy et al., 1996). Although that is beginning to change,
improvements remain patchy. Aggression carries high costs for services generally
and staff specifically as well as potentially very serious
Looking Ahead consequences for individuals (e.g. Attwood and Joachim,
1994; Bromley and Emerson, 1995). In a study of aggres-
What should be the way forward? In the management of sion against staff working in an NHS intellectual disability
offenders with intellectual disability, it is important not service, Kiely and Pankhurst (1998) found that there were
to adopt a ‘one-size-fits-all’ approach. Services and sys- almost five times more incidents of patient violence than
tems working with these people must be responsive to were recorded in the Trust’s sister mental health service.
their learning needs, to their styles and preferences and Following aggressive incidents, staff reported feeling wary
to those of the professionals operating the services if they of the perpetrator, and less confident in their own abilities.
are to achieve good outcomes. Also, while the delineation
of discrete steps in the care pathway is helpful to frame Assessment of anger among people
thinking, in actual case work the separate steps are often with intellectual disability
not readily discernible. Nevertheless, it is important that Despite the importance of anger and aggression problems
clinicians and case managers retain a clear picture of the for people with intellectual disability, literature on reliable
structure that lies behind this pathway, and the principles and valid measures of these phenomena in this popula-
that should underpin it including: inclusiveness, equity, the tion is sparse (Taylor, 2002). Studies by Benson and Ivins
evidence-base, rationality, accessibility, (cost) efficiency, (1992) and Rose and West (1999) have indicated that a
accountability, and social responsibility. To support this modified self-assessment measure of anger reactivity (the
approach, a recent UK Department of Health multi-centre Anger Inventory) has some limited reliability and valid-
systematic study has been undertaken to investigate the ity with people with intellectual disability. Oliver et al.
demographic, individual, offence and service character- (2007) reported that the Modified Overt Aggression Scale
istics of adults with intellectual disability referred to a (MOAS; Sorgi et al., 1991), an informant-rated measure
range of service settings in order to examine and define of the frequency and severity of aggression (verbal and
the service pathways and determinants of their offending physical against self, others and property), had high levels
and antisocial behaviour (Carson et al., 2010; Lindsay et al., of inter-rater reliability when used with a small number of
2010; O’Brien et al., 2010; Wheeler et al., 2009). people with intellectual disability as part of a treatment
outcome study.
Assessment and treatment
of anger and aggression Novaco and Taylor (2004) evaluated the reliability
and validity of several specially modified anger assess-
Prevalence and impact ment measures with detained male offenders with intel-
Studies of the prevalence of aggression, which includes lectual disability. The Novaco Anger Scale (NAS; Novaco,
more than infliction of actual physical violence, have been 2003), Spielberger State-Trait Anger Expression Inventory
conducted in large samples of people with intellectual dis- (STAXI; Spielberger, 1996), both self-report measures
ability across three continents. They indicate aggression of anger disposition, and the Provocation Inventory (PI;
rates in community populations of 11–16% in centralized Novaco, 2003), a self-report anger reactivity scale, along with
service provider surveys (Harris, 1993; Sigafoos et al., 1994) the Ward Anger Rating Scale, an informant-rated anger
and direct carer interview studies (Hill and Bruininks, attribution measure (WARS: Novaco, 1994) were evaluated.
1984; Smith et al., 1996). Aggression rates were consist- The modified anger self-report measures were found to
ently higher in institutional than in community settings have high internal consistency and less robust, but reason-
– at between 35% and 40%. In a study involving detained able, test–retest stability. The STAXI and NAS showed sub-
stantial inter-correlation, evidence of concurrent validity
for these instruments. staff WARS ratings of patient anger
326 © 2014 by Taylor & Francis Group, LLC
Assessment and treatment of anger and aggression
were found to have high internal consistency and to cor- the treatment of aggression in people with intellectual
relate significantly with patient anger self-reports. Anger, disability are impossible.
self-reported by the patients, was significantly related to
their record of assault behaviour in hospital. The NAS was The most extensive literature concerning treatment
found to be significantly predictive of whether the patient of their aggression is in the applied behavioural analy-
physically assaulted others following admission to hospital sis (ABA) field. Taylor and Novaco (2005) summarized
and of the total number of physical assaults. this, describing several extensive reviews, and concluded
that ABA-type behavioural interventions that are generally
Anger scales such as the NAS and PI are ‘nomothetic’, applied to low functioning individuals in institutional set-
which means that they are best suited to detecting mean tings may not be as effective for the anger and aggression
differences for groups of people with reference to norma- encountered among higher functioning but intellectually
tive data, rather than highlighting clinically significant disabled offenders who display low frequency, but very seri-
changes for individuals. Taylor et al. (2004b) further devel- ous aggression and violence, and live in relatively uncon-
oped the Imaginal Provocation Test (IPT; originally devel- trolled environments.
oped by Novaco, 1975) as an alternative ‘idiographic’ anger
assessment procedure for people with intellectual disabil- Treatment of anger and aggression using cognitive
ity. It is easy to administer, and taps key elements of the behavioural therapy (CBT) based interventions has now
experience and expression of anger (emotional reaction, been extensively evaluated with children, adolescents,
behavioural reactions, and anger control). The IPT indices adults with mental illness and offenders (see Novaco and
(anger reaction, behavioural reaction, anger composite and Taylor, 2006 for a review). Taylor (2002) and Taylor and
anger regulation) had respectable internal reliabilities and Novaco (2005) have also reviewed numerous single case
concurrent validity. In a small controlled trial, the IPT anger and case-series studies as well as uncontrolled group anger
reaction, behavioural reactions and anger composite indi- treatment studies of individual and group therapy for
ces were shown to be sensitive to clinical change following people with intellectual disability; these have generally indi-
treatment for anger (Taylor et al., 2004b). cated positive outcomes. Novaco’s (1975) anger treatment
approach that incorporates Meichenbaum’s (1985) stress
Willner et al. (2005) developed the Profile of Anger inoculation paradigm frequently provides the platform on
Coping Skills (PACS) to measure specific skills in managing which these modified interventions are based. More recently,
angry situations among people with intellectual disabil- a number of small treatment trials have shown the effective-
ity. Informants are asked to rate the person’s use of eight ness of group CBT for anger over waiting-list/no-treatment
anger management strategies in specific anger-provoking control conditions with people with intellectual disability
situations salient to that individual. The strategies assessed living in community settings (Lindsay et al., 2004; Rose et al.,
include use of relaxation skills, counting to 10, walking 2000; Willner et al., 2005; Willner et al., 2002). In the follow-
away calmly, requesting help, use of distraction activities, up study by Lindsay et al. (2004), the numbers of physical
cognitive re-framing and being assertive. The PACS was assaults by treatment group participants and those in the
found to have acceptable test–retest and inter-rater reli- waiting-list control condition were monitored for equiva-
ability coefficients and to be sensitive to change associated lent time periods following completion of treatment. It was
with an intervention to reduce anger. Following involve- found that significantly fewer participants in the treatment
ment in a community-based anger management group, condition physically assaulted others compared to those in
informant ratings showed that participants’ PACS scores the control condition (14% versus 45% respectively).
were significantly improved compared with scores for those
in a no-treatment control group. In the treatment group, There have also been a small number of studies of CBT
coping skills had improved significantly in terms of cogni- for anger with offenders with intellectual disability that
tive re-framing, assertiveness, walking away calmly, and have yielded positive outcomes. Allan et al. (2001) and
asking for help. These latter two areas of improvement were Lindsay et al. (2003) reported on group CBT for anger for a
maintained at 6-months follow-up. series of five women and six men with intellectual disability.
The participants were living in community settings, how-
Treatment for anger disorders among ever they had all been referred following violent assaults
people with intellectual disability which had resulted in CJS involvement. In both studies,
Taylor (2002) reviewed psychopharmacological treat- improvements which were reported for all participants at
ment of aggression for people with intellectual disability, the end of treatment were also maintained at 15-months
but found little reported empirical research. Reviews by follow-up. Burns et al. (2003) evaluated a CBT-framed group
Baumeister et al. (1998), Brylewski and Duggan (1999) anger management intervention for three offenders with
and Matson et al. (2000) suggested that, due to meth- intellectual disability residing in a specialist NHS medium
odological problems with this research, firm conclusions secure unit. Using multiple assessment points to carry out
about the effectiveness of psychotropic medications in time series analysis, the results were mixed in terms of self-
reported anger and informant rated aggression measures.
The authors suggest that the relatively short length of the
© 2014 by Taylor & Francis Group, LLC 327
Offenders with intellectual disabilities
unmodified intervention and unstable baseline measures of the scales, sex offenders scored significantly higher than
contributed to the limited treatment effects observed. other types of offenders, non-offenders or a group of non-
offending normal men. The QACSO has also been shown to
Taylor et al. (2002, 2004c, 2005) evaluated individual CBT be sensitive to change over time (Lindsay et al., 1998; Rose
for anger among detained male patients with mild–border- et al., 2002). There is also some preliminary evidence that
line intellectual disability and violent, sexual and fire-setting the QACSO offences against children scale may differentiate
histories in a series of concatenated waiting-list controlled between offenders against children and offenders against
studies. The 18-session treatment package includes a adults, with the latter scoring significantly lower on this
6-session broadly psycho-educational and motivational pre- scale (Lindsay et al., 2007).
paratory phase, followed by a 12-session treatment phase
based on individual formulation of each participant’s anger Treatment of intellectually disabled sex offenders
problems and needs. The treatment phase follows the The evidence base for the treatment of inappropriate sexual
classical cognitive behavioural stages of cognitive prepara- behaviour in men with intellectual disability is reason-
tion, skills acquisition, skills rehearsal and then practice in ably extensive (Lindsay, 2002; Courtney and Rose, 2004).
vivo. These studies showed significant improvements on Until the late 1990s, behavioural management approaches
self-reported measures of anger disposition, anger reactiv- remained the most common basis for psychological treat-
ity and behavioural reaction indices following interven- ment of their sexual offending (Plaud et al., 2000). These
tion in the treatment groups compared with scores for authors noted that the purpose of a behavioural treatment
the control groups. These differences were maintained for programme is to improve behavioural competency in daily
up to 4 months following treatment. Staff ratings of study living skills, general interpersonal and educational skills
participants’ anger disposition did not reach statistical sig- as well as the more specialized skills related to pro-social
nificance, but tended to converge with patient self-reports. sexual behaviour. Griffiths et al. (1989) developed a compre-
hensive behavioural management régime for sex offenders
In summary, there are some indications that reliable with intellectual disability. Their programme included mod-
and valid assessment measures are being developed for ifying deviant sexual behaviour through education, training
offenders with intellectual disability who are angry and/ in social competencies, improving relationship skills, and
or aggressive. Limited but growing research evidence sug- relapse prevention through alerting support staff and train-
gests that CBT based interventions may be effective for ing on issues related to responsibility. In a review of 30 cases,
this population. they reported no re-offending and described a number of
successful case studies to illustrate their methods.
Assessment and treatment
of sexually aggressive In their review, Plaud et al. (2000) also describe aver-
behaviour among people sion therapy techniques and masturbatory retraining tech-
with intellectual disability niques in some detail. Although there are few reports on
the use of these methods with offenders with intellectual
Assessment disability, Lindsay (2004) successfully employed imagined
The Socio-Sexual Knowledge and Attitudes Test (SSKAT; aversive events to control deviant sexual arousal and
Wish et al., 1980) has been used widely to consider the role routines.
and relevance of sexual knowledge in sexual offending by
a person with intellectual disability. Griffiths and Lunsky A major recent development in the use of psychological
(2003) have comprehensively revised the SSKAT and shown treatment for sex offenders with intellectual disability has
that sexual offenders have superior knowledge to non- been the employment of cognitive and problem-solving
sexual offenders in a number of key areas. techniques within therapy. Among mainstream offenders,
Hanson et al. (2002) reported, in a meta-analytic study, that
Developments in sex offender treatment work have those treatments that used cognitive techniques showed
emphasized the centrality of cognitive processes in the greater reductions in recidivism rates than treatments
planning, commission and post hoc justification of sexually using other techniques, including behavioural treatments.
offensive incidents by people with intellectual disability A central assumption in cognitive approaches is that sex
(Hudson et al., 1999). Broxholme and Lindsay (2003) and offenders are likely to hold a number of cognitive distor-
Lindsay et al. (2007) have developed the Questionnaire tions about sexual behaviour and its expression that, for
on Attitudes Consistent with Sex Offences (QACSO), an them, provide justification for their sexual aggression.
assessment of attitudes consistent with, or permissive of, Cognitive distortions fall into a number of categories, but
sexual offending. This measure assesses attitudes towards generally include: complete denial that an offence occurred,
rape, voyeurism, exhibitionism, dating abuse, stalking, mitigation or even denial of responsibility, denial of harm to
homosexual assault and offences against children. The the victim, denial or mitigation of intent to offend, thoughts
QACSO scales were shown to have good internal reliability of entitlement, and mitigation through claims of altered
(Cronbach alphas of around 0.8 in each case) and on each mental states such as depression or intoxication.
328 © 2014 by Taylor & Francis Group, LLC
Fire-setting behaviour among people with intellectual disability
Lindsay et al. (1998) reported a series of case studies of offenders, all with intellectual disability. Re-offending rates
treatment of paedophiles, exhibitionists and stalkers with were reported for up to 12 years after the index offence.
intellectual disability using cognitive behavioural interven- There were no differences between the groups in IQ; the
tions in which various forms of denial and mitigation of the sex offender cohort tended to be older than the other two
offence were challenged over treatment periods of up to 3 cohorts. The female offenders had higher rates of mental
years. Strategies for relapse prevention and the promotion illness although about one-third of the men also had men-
of self-regulation were also components of the treatment. tal illness. The differences in re-offending rates between
Across these studies, participants consistently reported posi- the three groups was highly significant with rates of 24%
tive changes in cognitions during treatment. Each of these for male sex offenders, 19% for female offenders but 59%
reports provides examples of how cognitive distortions are for other types of male offenders. Number of offences by
elicited and challenged during treatment. This component recidivists following treatment was also reduced – to about
of the intervention was evaluated directly using the QACSO a quarter to a third of those recorded before treatment, indi-
(Lindsay et al., 2007). Reductions in the number of endorse- cating a considerable amount of harm reduction as a result
ments given to cognitive distortions were found following of interventions.
extended treatment periods. These treatment gains were
maintained for at least 1 year following cessation of treat- Based on the limited evidence available it is possible
ment. More importantly, more lengthy follow-up of some to conclude tentatively that psychologically informed and
cases (4–7 years) showed that none had re-offended. structured interventions appear likely to improve out-
comes for sex offenders with intellectual disability, CBT
Rose et al. (2002) reported on a 16-week group treat- to have a positive effect on offence-related attitudes and
ment for five men with intellectual disability who had cognitions, and that longer periods of treatment will yield
perpetrated sexual abuse. The group treatment included better outcomes that are maintained for longer.
self-control procedures, consideration of the effect of the
offences on their victims, emotional recognition and strate- Fire-setting behaviour
gies for avoiding risky situations. Individuals were assessed among people with
using the QACSO, a measure of locus of control, an assess- intellectual disability
ment of knowledge of sexually inappropriate behaviour,
and a victim empathy scale. Significant differences from Prevalence of fire-setting with
pre- to post-treatment were found only on the locus of intellectual disability
control scale. It has to be acknowledged that the length of It has been suggested that fire-setting is over-represented
treatment was short in comparison to most sex offender amongst offenders with intellectual disability (e.g. Day,
treatment programmes (usually 12–18 months), neverthe- 1993; Raesaenen et al., 1994) but, as with so many other
less, participants had not re-offended at 1-year follow-up. areas with respect to intellectual disability, this research
too is sparse and limited by methodological problems, and
Unfortunately, treatment comparison studies fall well the prevalence may, in fact, be unremarkable. Irrespective
short of optimum experimental standards, and it is impor- of the scale of the problem, there is general agreement
tant to consider the results in the light of this. Lindsay and that that there are people with intellectual disability
Smith (1998) compared seven individuals who had been in who have histories of fire-setting who need specialist
treatment for 2 or more years with a group of seven who assessment, treatment and management. In a hospital-
had been in treatment for less than 1 year. The comparisons wide study of male forensic inpatients with intellectual
were serendipitous in that time in treatment reflected the disability, Taylor et al. (2002b) found that 20% of them
length of their probation orders made by the court. There (26 of 129) had convictions for arson prior to admission.
were no significant differences between the two groups in
terms of severity or type of index offence. The 1-year treat- Assessment of intellectually disabled fire-setters
ment group showed significantly poorer progress and were Literature on clinical practice with fire-setters with intel-
more likely to re-offend than those treated for at least 2 lectual disability is even more limited than prevalence
years. This suggests that shorter treatment periods may be research. Murphy and Clare (1996) interviewed 10 fire-
of limited value for such men. setters with intellectual disability about their cognitions
and feelings prior to and after setting fires, using a newly
Keeling et al. (2007) compared treatment outcomes for developed Fire-Setting Assessment Schedule (FSAS).
11 ‘special needs’ sexual offenders and matched mainstream Participants were also asked to rate their feelings in rela-
controls. These were convenience samples, and the authors tion to a series of fire-related situations described in a
note the difficulties in ensuring the equivalence of treatment new 14-item Fire Interest Rating Scale (FIRS) (Murphy
and assessments between the two groups. There were no and Clare, 1996). The construction of the FSAS was guided
assessed differences between the groups following treat-
ment and no re-offending reported at 16 months follow-up
for either. Lindsay et al. (2006b) compared 121 sex offend-
ers with 105 other types of male offenders and 21 female
© 2014 by Taylor & Francis Group, LLC 329
Offenders with intellectual disabilities
by the functional analytical approach to fire-setting pro- Two group follow-up sessions were held, the first 6 weeks
posed by Jackson et al. (1987), in which it is proposed that after the original group and the second after 6 months.
fire-setting is associated with a number of psychological Unfortunately, although well designed, this work rests
functions, including the need for peer approval, need for wholly on description and no outcome measures are given –
excitement, need to alleviate or express sadness, mental just an indication that the men ‘responded positively’.
illness, or a wish for retribution. Murphy and Clare (1996)
found that their study participants identified antecedents Taylor et al. (2002b) reported a group study involving
to fire-setting with more reliability than they could the 14 men and women with intellectual disability and arson
consequences. The most frequently endorsed antecedents convictions who were assessed pre- and post-treatment
were anger, followed by being ignored and then feelings on a number of fire-specific, anger, self-esteem and depres-
of depression. The FSAS has proved to be clinically useful sion measures. The intervention comprised 40 CBT based
since its inception, but was little further investigated until group sessions over 4 months that involved work on offence
Taylor et al. (2002) used it to assess the effectiveness of a cycles, education about the costs associated with set-
treatment programme for 14 fire-setters with intellectual ting fires, training of skills to enhance future coping with
disability. Consistent with the results of the Murphy and emotional problems associated with previous fire-setting
Clare (1996) study, Taylor et al. (2002b) found that anger, behaviour, and work on personalized plans to prevent
being ignored and depression (in rank order) were the most relapse. Following treatment, significant improvements
frequently endorsed antecedents to and consequences of were found in all areas assessed, except depression.
fire-setting in this group according to ratings on the FSAS.
A case series of four detained men with intellectual
Treatment of intellectually disabled fire-setters disability and convictions for arson offences were evalu-
Rice and Chaplin (1979) conducted a study of social skills ated before and after completion of the same intervention
training for two groups of five fire-setters in a high security (Taylor et al., 2004d). The patients engaged well, and all
psychiatric facility in North America. People in one of the showed high levels of motivation and commitment that
groups were reported to have mild to borderline intel- were reflected in generally improved attitudes with regard
lectual disability. After treatment, both groups improved to personal responsibility, victim issues and awareness of
significantly on a reliable observational rating scale of risk factors associated with their fire-setting behaviour.
role-played assertive behaviour. At the time of reporting,
eight of the 10 patients in this study had been discharged The same methods were used in a further case series of
for around 12 months and none had been convicted or six women with a similar range of problems (Taylor et al.,
suspected of setting fires. 2006), again, the participants engaged well and all com-
pleted the programme. Their scores on measures related to
Clare et al. (1992) reported a case study of a man with fire-specific treatment targets generally improved following
mild intellectual disability who had been admitted to a the intervention. All but one of the participants had been
secure hospital after two convictions for arson. He had a discharged to community placements at 2-year follow-up,
prior history of arson and making hoax calls to the fire ser- and there had been no reports of participants setting any
vice. On transfer to a specialist inpatient unit he received a fires or engaging in fire risk-related behaviour.
comprehensive treatment package, which included social
skills and assertiveness training, development of coping There is, then, little more than pilot work to support
strategies, covert sensitization, and surgery for a signifi- therapeutic interventions for people with intellectual dis-
cant facial disfigurement; significant clinical improvements ability who set fires, but that work does seem to offer prom-
were observed in targeted areas. Discharged to a commu- ise, based on CBT methods delivered in a group setting.
nity setting, this man had not engaged in any fire-related
offending behaviour at 30 months follow-up. Assessment and management of
risk of offending and/or harm
Hall et al. (2005) described application of cognitive to others among offenders
analytical therapy (CAT), an integrative model of short- with intellectual disabilities
term psychotherapy (Ryle, 1993), to arsonists with intel-
lectual disability, with the aim of reformulating the origins Advances have been made in the general forensic research
of the distress and maladaptive coping strategies that had field in the development of measures designed to predict
resulted in fire-setting behaviour. They also described the violence and sexual aggression amongst those with a his-
delivery of a 16-session group CBT approach for six male tory of offending (e.g. Banks et al., 2004; see also chapter
fire-setters with intellectual disability, detained in an NHS 22). This work has now been extended to include offenders
medium secure setting, to help them identify personal risk with intellectual disability. Quinsey et al. (2004) demon-
factors associated with their fire-setting and develop alter- strated that the Violence Risk Appraisal Guide (VRAG;
native coping strategies to reduce their risk of re-offending. Quinsey et al., 1998), an established actuarial risk measure
in the general offender literature, has good predictive accu-
racy when used with intellectual disability offenders. Gray
330 © 2014 by Taylor & Francis Group, LLC
Legal and ethical considerations in working with offenders with intellectual disabilities
et al. (2007) conducted a more extensive investigation of instruments, some developed specifically for this group,
the VRAG with 145 patients with intellectual disability and have good reliability, discriminative validity and predictive
996 mainstream patients, all discharged from hospital hav- validity with offenders with intellectual disability.
ing been admitted with serious mental illness, intellectual
disability or personality disorder – and having been con- Legal and ethical
victed of a criminal offence or having exhibited behaviour considerations in working
that might have led to a conviction in different circum- with offenders with
stances. They found that the VRAG predicted re-conviction intellectual disabilities
rates in the intellectual disability sample with an effect size
as large as that for the non-intellectual disability sample. Intellectual Disability and the
Criminal Justice System
This important research on the assessment and man-
agement of risk in offenders with intellectual disability People with intellectual disability are vulnerable at each
has continued in a study involving 212 people in a range of stage of the criminal justice process – pretrial, during any
security settings: high, medium and low hospital security, court hearing and on sentencing. Their disability may
and community forensic intellectual disability services (the affect their capacity to understand their rights on arrest,
212 Multi-Centre Risk Study: Hogue et al., 2006). The most to deal with police questioning/interrogation, and/or to
complex presentations, in particular those with comorbid provide valid statements or confessions. Then, if the case
personality disorder, were found in the more secure set- comes to court, their capacity to enter a plea, to under-
tings; more participants in the high security group had stand court proceedings, and/or to instruct their counsel
sustained both index and previous convictions for violence. may be limited. On sentencing, particularly if this is to
prison, their capacity to access and make use of the various
Lindsay et al. (2008) combined the total cohort of educational, treatment and rehabilitative packages offered
offenders with intellectual disability from the 212 Multi- in the mainstream CJS may impede their progress, and
Centre Risk Study to evaluate the predictive validity of a even compromise their rights to fair treatment.
range of static and dynamic risk assessments. They found
that the VRAG, the Short Dynamic Risk Scale (Quinsey, ‘Responsibility’ and ‘competency’ or ‘capacity’ are key
2004), and the Emotional Problems Scale (Prout and concepts in criminal justice systems around the world and
Strohmer, 1991) showed significant areas under the curve are particularly pertinent to offenders with intellectual
(AUC), using receiver operator characteristics (ROC) disability. In the pretrial phase, defendants with intellec-
analyses (for an explanation of these measures see chap- tual disability are exceptionally vulnerable to giving self-
ter 22) in relation to the prediction of violent incidents. incriminatory statements or confessions. In fact, in England
The Static-99 (Hanson and Thornton, 1999) also showed and Wales, the Confait case, which ultimately led to the
a significant AUC in relation to the prediction of sexual Police and Criminal Evidence Act (PACE) 1984 through a
incidents. With the same study sample, Taylor et al. (2010) route including a judicial enquiry (Fisher, 1977) and the
reviewed the psychometric properties and predictive Royal Commission on Criminal Procedure 1981, was of two
validity of the HCR-20 (Webster et al., 1997 and chapter 17 year olds and a man with intellectual disability, who
22). Exploratory factor analysis found that the H Scale had been ‘prompted’ into a false confession. PACE and its
(historical items) constituted three factors (delinquency, accompanying Codes of Practice have particular provisions
interpersonal function and personality disorder) while the for people with intellectual disability with regard to police
C (clinical items) and R (risk items) scales made up dis- questioning and confessional evidence (Sanders and Young,
tinctly separate factors. Using ROC analyses, they found 2000). Before beginning an interview with a person with
that all three scales were predictive of incidents recorded intellectual disability, or who appears to have intellectual
prospectively over a 12-month period, concluding that the disability, the presence of an ‘appropriate adult’ is required.
HCR-20 is a robust instrument for guiding clinicians to An appropriate adult is distinct from a legal advisor, and is
reach clinically consistent and defensible decisions. more likely to be a relative or guardian of the interviewee,
or someone with experience of working with people with
In an extension of the 212 Multi-Centre Risk Study, intellectual disability (e.g. social worker or community
Morrissey and colleagues (Morrissey et al., 2005, 2007a,b) nurse) who is not employed by the police service.
investigated the utility, discriminant and predictive valid-
ity of the Psychopathy Checklist–Revised (PCL-R: Hare, The issue of suggestibility of accused persons with
2003). The results show that the PCL-R predicts both good intellectual disability during police interviews has been
response to treatment and positive moves from high to well researched. Gudjonsson (1992) argued that people
medium secure conditions, both within 2 years of assess- with intellectual disability were more susceptible to yield-
ment. The PCL-R did not, however, predict institutional ing to leading questions and shifting their answers under
violence at a better than chance level. interrogation by the police and, as such, more liable
Overall, risk assessment research has demonstrated
that several well established actuarial, dynamic and clinical
© 2014 by Taylor & Francis Group, LLC 331
Offenders with intellectual disabilities
to give false confessions. He developed the Gudjonsson Retardation (CAST-MR: Everington and Luckasson, 1992)
Suggestibility Scales (GSS; Gudjonsson, 1997), which are assesses competence in three areas: basic legal concepts,
used widely. Clare and Gudjonsson (1993) found that par- skills to assist the defence counsel and understanding of
ticipants with intellectual disability confabulated more court procedures. The CAST-MR was used by 45% of psy-
and were more acquiescent during interrogative interview chologists surveyed about practices used when evaluating
while Everington and Fulero (1999) found that people with competence to stand trial among juveniles (Ryba et al.,
intellectual disability were more likely to alter their answers 2003), but this and similar assessments have a number of
in response to negative feedback. Beail (2002), however, in a limitations. These include the lack of an underlying con-
review of a number of studies involving the GSS, questioned ceptual structure, no standardized administration proce-
whether artificial test situations were similar enough to dures, no criterion based scoring, and limited normative
real-life interrogation situations: data (Otto et al., 1998).
…because the results are based on an examination of In law, a criminal conviction generally requires proof
semantic memory whereas police interviews are more beyond reasonable doubt of two elements – the act (actus
concerned with episodic or autobiographical event mem- reus) and the state of mind, or intent, necessary for this
ory. Also, experienced events usually involve multi-modal to have been a crime (mens rea). Historically, people with
sensory input, resulting in a more elaborate trace in asso- severe intellectual disability were considered incapable
ciative memory (p.135). of forming such intent and thus not responsible for their
Beail concluded that the GSS may be limited in its applica- actions (Fitch, 1992). Traditionally, in common law systems,
bility to criminal justice proceedings. judgment about responsibility for acts on the part of a per-
son with intellectual disability was made in terms of his/
These concerns about competence to be interviewed her ability to distinguish right from wrong, but, in the US
are important because confessional evidence should at least, courts are moving away from this dichotomous
be considered to be valid only if it is voluntary, know- approach to ‘moral understanding’ in favour of case-by-case
ing and intelligent (Baroff et al., 2004). People with consideration (Baroff et al., 2004) (see also chapters 2 and 5).
intellectual disability can be particularly vulnerable to
coercion, threats and promises of leniency, thus raising Ethical Issues and Offenders
concerns about the ‘voluntariness’ of their confessions. with Intellectual Disability
Understanding of the concepts of the right to silence and
other rights to protect oneself have rarely been tested As Sturmey and Gaubatz (2002) have indicated, issues
with suspects who have intellectual disability, and yet concerning offenders in secure settings pose classic prob-
such interviewees are probably more likely to answer lems for professional ethics. People with intellectual dis-
questions in the manner and direction they believe they ability may be particularly vulnerable to various forms of
are expected to, the so-called social desirability bias abuse involving the application of dangerous treatment
(Baroff, 1996), or simply to want to please. A valid confes- and research interventions without due process. While
sion rests on the suspect’s understanding that in waiving some people with intellectual disability will be able to
his or her rights s/he may be placing him- or herself in understand the elements necessary for consent to treat-
jeopardy. In the stressful and confusing context of arrest ment, many cannot and even fewer are likely to be able to
and interrogation it can be exceptionally difficult for comprehend all the elements necessary for participation in
suspects with intellectual disability to make a reasoned clinical research (Arscott et al., 1998, 1999). The tension lies
choice about information they will volunteer or with- in the fact that it is crucially important also to avoid any
hold, or to grasp the implications of their responses to situation where discriminatory decisions might exclude
police questions (Baroff et al., 2004). people with intellectual disability from potentially benefi-
cial or benign treatment or from research because of erro-
In assessing competency or fitness to stand trial and neous assumptions about their capacity to give consent. A
enter a plea, a defendant’s abilities in the following areas balance is needed (Sturmey et al., 2004).
should be considered: (a) understanding of the crime of
which they are accused; (b) knowledge of the purpose Valid consent requires that people are provided with
of the trial and the roles of the principal players; and (c) accurate information about the treatment or research
ability to instruct one’s counsel (Baroff et al., 2004). In the intervention, that they have capacity to make a decision
UK, assessment and management of a claim to be unfit about it and that they understand the consequences
to plead or stand trial follows the same path, regardless of the decision and that their decision to participate
of the nature of the disorder (see chapter 2). There are a is voluntary (Lord Chancellor’s Department, 1999); It
number of assessments which may be of particular value is possible to make information about treatment and
in assessing a defendant’s understanding of court pro- research understandable and accessible to people with
ceedings when s/he has intellectual disability, all from the intellectual disability (Arscott et al., 1998, 1999); however,
USA. The Competence Assessment to Stand Trial–Mental research is needed concerning functional assessments
332 © 2014 by Taylor & Francis Group, LLC
Conclusions
of the elements (comprehension, assimilation, recall and that was not just unhelpful and ineffective (Mostert,
decision-making) required for capacity to give valid con- 2001), but resulted in the wrongful conviction of parents
sent (Iacono and Murray, 2003). of sexual abuse of their children and separated their chil-
dren from their families. It is important, therefore, that
These important considerations aside, the extent to treatment and research with this vulnerable group are
which decision-making by offenders with intellectual dis- not simply well intended, but are supported by evidence
ability can ever be totally voluntary and free of a degree of their effectiveness and offered to offenders with intel-
of coercion is a matter that needs to be aired openly by lectual disability within a robust and transparent ethical
practitioners in this field. This would encourage clarity consent framework.
for clinicians and researchers, ethics committees, and
offenders themselves, in this difficult area. In most juris- Conclusions
dictions, treatment for offenders may be mandated by the
courts. The need to punish offenders, to protect the pub- The policy of de-institutionalization of services for people
lic, and to rehabilitate creates a tension. In Europe, clini- with intellectual disability has had an impact on offend-
cians are fortunate that they do not have to struggle at all ers with intellectual disability, who are now more visible
with giving evidence in capital cases, but even in the USA, than before (Taylor and Lindsay, 2007). Larger numbers of
in 2002, the Supreme Court outlawed the death penalty people with intellectual disability who transgress society’s
for people with mental retardation (Atkins). The decision mores are being dealt with through regular CJS channels,
acknowledged that such defendants face a greater risk and the courts are sending more people with intellec-
of wrongful conviction and warrant special protection tual disability to forensic mental health programmes for
accordingly, although, as French (2005) describes, the rul- offence-related interventions (Lindsay and Taylor, 2005).
ing still allows the state to determine how such offenders Services, research and practice for this offender group have
are to be assessed and measured, and provision of appro- grown apace since the early 1990s (Lindsay et al., 2004),
priate services remains patchy. As a number of people but, as illustrated, even in the field of epidemiology the
have shown, including French (2005) and Sturmey et al. research has limitations. Most evidence for treatments and
(2004), psychological assessment, treatment and research intervention comes from naturalistic studies or single case
in the intellectual disability field have been subject to work. Beyond the need to develop the evidence base for
fashions and trends which have at times been far from treating offenders with intellectual disability, we anticipate
benign, including enactment of sterilization laws (Sofair that, in common with other health, criminal and human
and Kaldjian, 2000). Now, in Europe, the USA and many service areas, the most pressing future issue is understand-
other countries, there are legal protections against abuse ing how to translate knowledge into practice and make a
in this respect. real difference to the lives and prospects of this subgroup
of people with intellectual disability, improving their safety
Unproven treatments or interventions have the at home and in the wider community.
potential to do harm even when the intention is benign.
Facilitated communication is an example of one such
© 2014 by Taylor & Francis Group, LLC 333
14
Psychosis, violence and crime
Edited by Written by
Pamela J Taylor Pamela J Taylor
Sue Estroff
1st edition authors: Paul Mullen, Pamela J Taylor and Simon Wessely
The personality of any normally constituted person times the rate in the comparison group. Teplin’s group then
must be capable of at least a certain flexibility, reviewed all 31 US empirical studies since 1990 of violence
otherwise the machinery for doubt would be absent, among people with psychosis. Choe et al. (2008) found that,
and what is more irrefutable proof of madness than an overall, 12–18% of all patients – resident in hospital or in the
inability to have doubt. No, no, to ensure sanity there community – had been perpetrators of violence, but 35%
must at least be the elements of internal disagreement had been victims of violence in the 6–18 months prior to
ever present in a personality (Ustinov, 1977). the study. Just three of the studies had evaluated perpetra-
tion and victimization in the same sample. One found that,
Vulnerable to violence and in an outpatient commitment sample over a short period
vulnerable to being violent (4 months), a higher proportion of the patients had been vic-
timizers than victimized, but in the other two studies having
The psychoses are intrusive and often chronic debilitating ill- become a victim of violence was far more prevalent than
nesses that profoundly alter the lives of those who have them. inflicting it (Brekke et al., 2001; Brunette and Drake, 1997).
Among their most common symptoms are delusions, which
allow little room for doubt. Among their many unfortunate Violence to self is another indicator of harmful distress
consequences they leave sufferers very vulnerable to violence which sometimes affects others as well as the sick person.
in all its directions. The risk of becoming a victim of violence Among people with schizophrenia and similar disorders,
is higher among people with schizophrenia and similar ill- non-fatal self-directed violence occasionally takes on a
nesses than among the general population, but risk engaging specifically destructive form if body parts are involved in
in destructive acts to self or harming others is also elevated. the delusional symptoms (e.g. Chand et al., 2010). More
commonly than this, self-directed violence may be fatal;
Victimization is mainly dealt with in chapter 28. Here the risk of suicide by people with psychosis is higher than
we provide only an illustrative summary of the consistent their risk of homicide. Data from the National Confidential
findings in this area. Among about 700 people attending Inquiry into Suicide and Homicide (NCISH) (see also chap-
psychiatric services in the UK (Walsh et al., 2003) and ter 28) are illustrative. For the period April 2000–December
962 in Australia (Chapple et al., 2004), it has been shown 2004 (NCISH, 2006), 6367 Inquiry cases of suicide, about
that 16–18% of people with psychosis become a victim of 27% of the total suicides for the period, were defined by
violence during the course of a year, with social exclusion having had some contact with mental health services in
factors, such as homelessness, substance misuse and more the 12 months before death; 1,145 (19%) had schizophrenia
severe psychopathology increasing the risk of this happen- or other delusional disorders and 2,821 (46%) depression
ing. In the United States of America (USA), Teplin et al. or bipolar disorder. There were 2,670 homicide convictions
(2005) drew from the 32,449 people participating in the and 14 cases of murder charges where the defendant was
nationwide USA Crime Victimization Survey to compare unfit to plead or not guilty by reason of insanity over a
crime victimization among 483 men and 453 women with similar period (April 1999–December 2003); 141 (5%) of the
severe mental illness (SMI) and their peers without. The homicides were by people with schizophrenia, 806 (30%) by
former were randomly selected, but stratified for ethnicity, people with any lifetime psychiatric diagnosis. In the public
from 16 mental health treatment agencies in Chicago. Over mind, the risk of a person with psychosis being violent to
25% of the people with severe mental illness had become others is generally the source of greater concern than the
the victim of a violent crime in the preceding 12 months, 11 other types of vulnerability to violence. Constant efforts
are needed to inform the public better and, above all, to
334 © 2014 by Taylor & Francis Group, LLC
Vulnerable to violence and vulnerable to being violent
diminish fear about violence to strangers. This is, in fact, discrimination invoke the contributions to violence of
less common on the part of psychosis sufferers than within common comorbid conditions. One person’s altruistic
other diagnostic groups (Johnson and Taylor, 2003) or the interpretation, however, can feed another’s search for
wider public (e.g. NCISH, 2006). We are, however, writing reasons for re-labelling and rejecting those seeking help
for a text on forensic psychiatry, so the main focus of this from their services. Since the 1983 edition of this text
chapter will be on violence perpetrated by people with there have been many publications on the numerical
schizophrenia or similar psychoses. relationship between psychosis and violence, but it is less
clear that there has been a commensurate increase in
A great deal is known about a few individuals with psy- understanding how psychosis may predispose a person to
chosis who have killed another person, often driven to the being the recipient or perpetrator of violence or of how to
deed by their illness. Worldwide, at least since 1800, the public decrease the number and seriousness of violent incidents
and political imagination has been captured by spectacular through therapeutic intervention with the individual and
cases. Not infrequently these tragic figures then inadvertently his/her social environment.
become the impetus for new mental health law, or mental
health aspects of criminal law, often bearing the names of Psychosis and Perpetration of
those who were killed, and for specialist secure forensic Violence: Some Illustrative Cases
mental health services. Post hoc, efforts are made to insert and Their Consequences
the more pragmatic mix for science and humanity which is
essential both for the specialist services to be successful and In England, widely reported links between psychosis and
for established generic services to improve management of infamous crimes which had implications for law and men-
complex cases, and so prevent harm. Evidence for the effec- tal health or social services first became prominent in the
tiveness of most current strategies and services, however, nineteenth century. In 1800, Hadfield shot at King George
remains slight, as does an evidence-based understanding of III (Walker, 1968). Hadfield was put on trial for high trea-
the nature of links between psychosis and violence. son, but his post-traumatic psychosis was recognized.
Powerful arguments for compassion on the one hand and
Some Strengths and Limitations pragmatics on the other resulted in a finding of:
of the Literature
Not Guilty: he being under the influence of Insanity at the
Much of the research has focused on the epidemiologi- time the act was committed.
cal question of whether violence is associated with psy-
chosis more often than would be expected by chance. Since he was manifestly dangerous, legislation was hastily
In the 1980s, evidence in this area was incomplete and enacted and implemented retrospectively, thus allowing
inconclusive, and each of the three main claims – that his detention in hospital. McNaughton’s trial for murder in
violence was more, less, or as common among people 1843 brought elaboration of the insanity rules, still referred
with psychosis than without it – had some support to by his name. It also brought some of the best descrip-
(Taylor, 1982). A main limitation was that research fol- tions of links between psychosis and serious violence,
lowed a pattern of inquiring about violence in groups of as not only was his concurrent mental state described,
people with psychosis who were in treatment or about but also witnesses who had known him long before the
psychosis in established criminal populations; there were culminating homicide were cross-examined about the
no true community surveys (Monahan and Steadman, development of his delusions (West and Walk, 1977). This
1983). Since the 1990s, several good epidemiological stud- is not the place for a long list of notorious and/or influen-
ies have emerged. After a period of consistent findings tial English cases, but a leap forward to the latter part of
and growing certainty about the elevated risk of violence the twentieth century will show their continuing power.
among people with psychosis, however, ever larger more On 17 December 1992, Christopher Clunis killed Jonathan
recent studies seem to be beginning to call this into ques- Zito. The men did not know each other and the attack, in
tion again. A major problem in this field is that political full public view, was unprovoked and, at least to onlook-
overtones seem inescapable, both in the commissioning ers, at random. Clunis had sought help. In the 4 years prior
of research and in its interpretation and application. to the fatal attack he had seen 43 different psychiatrists
From a clinical perspective, the size of the relation- (Ritchie et al., 1994), and it is hard to imagine that any one
ship between psychosis and violence matters primarily of them could have developed an in-depth understanding
in terms of planning effective services with adequate or therapeutic relationship with him. He was a man with a
capacity. From a civil and legal rights perspective, there near classic social decline into schizophrenia, but he had
has been much concern that an epidemiology which difficulty in engaging in treatment, had given warnings of
establishes a link between psychosis and violence may violence and sometimes used illicit drugs, so his diagnosis
merely serve to stigmatize an already disparaged group, varied as he moved between services. On one occasion
and put their civil liberties in peril. Where an association his condition was even described as ‘manipulation for a
is found between the two, those who fear an increase in bed’. Sadly, the landmark change was not that services
© 2014 by Taylor & Francis Group, LLC 335
Psychosis, violence and crime
became uniformly more receptive to such people. The services came to a disastrous end. He ran amok in a school,
Ritchie report, together with the charitable trust set up by and killed several schoolgirls. On arrest, he said:
Jonathan Zito’s widow (see also chapter 28) was, however,
instrumental in establishing the principles that every Anything and everything has become unbearable. Time
homicide by a person who had been in contact with psy- and time again I tried to kill myself but I could not do it. I
chiatric services before the homicide would be subject to wanted to be arrested and get the death sentence.
an individual independent inquiry (Department of Health, He was later executed, but the Prime Minister’s immediate
1994) and that there should be a National Confidential response was:
Inquiry into these homicides (see also this chapter below The imperfection of the law has become clear and certain
and chapter 28). It also influenced many of the changes in changes to Japan’s psychiatric system should be consid-
mental health legislation for England and Wales, finally ered for mentally ill offenders.
enacted in 2007. In 2003, the Japanese parliament, the Diet, enacted the
Law Concerning Medical Treatment and Observation
Other parts of the UK have had experience with similar for People Who Commit Serious Harm to Others Under
individuals for whom psychosis, for a variety of reasons, was the Condition of Lost Mind and the Like 2003, which,
not much helped by services, and led to serious violence. among other provisions, for the first time enabled the
The extent, however, to which understanding that mental development of specialist forensic psychiatric services
health services could and should have a more effective (Yoshikawa and Taylor, 2003; see also chapter 5).
role in preventing such tragedies is global, and transcends
cultures and less porous national boundaries. Another man The cases that drive service change tend to involve
who had been trying to get help for his psychotic illness was offences by men, and there are common claims that foren-
Andrew Goldstein in New York. He had had a good deal of sic mental health services have been particularly oriented
attention, and had accumulated over 3,500 pages of hospital towards men to the detriment of treating women. Among
psychiatric records by the time he committed a homicide, people with psychosis, as those without, recorded violence
but he too was chaotic and often skipped appointments. Just is more common among men than women, so more men
days before the killing, a case worker sent him a note saying present to services. As we shall see, however, proportion-
that if he had not called by a specified date, his case would ately more violence among women may be accounted for
be closed (Winerip, 1999). Three days before that deadline, by psychosis than among men, but then, most research has
he pushed a young woman, Kendra Webdale, to her death necessarily been done with men. Our examination of the
under a subway train. Later that year, the New York State work on violence in relation to psychosis will, therefore,
Assembly and Senate passed The New York Mental Hygiene mainly reflect research with men only, or with men form-
Law 1999 (MH Law 9.60), also known as ‘Kendra’s Law’, and ing the majority of any sample. Where issues more specific
injected money into developing an ‘Assisted Outpatient to women can be examined, these are mainly covered in
Treatment Program’. For adults still capable of living in the chapter 20.
community but likely to be unsafe without supervision,
this required attendance at a designated clinic, included a Psychosis and crime:
mandate for medication and, as necessary, periodic blood The epidemiology
tests/urinalysis, individual or group therapy, day treatment,
educational or vocational training, and supervision of living Relative Proportions of People with and
arrangements. Preliminary evaluation of the programme without Psychosis Who are Violent
suggested that it had been effective in reducing violent and
suicidal behaviour as well as rates of hospital admissions, Even if single cases are powerful, it is scientifically inap-
so the law was renewed in 2005, but with a requirement for propriate to draw from them the inference that there is
further evaluation. Subsequent research replicated these more than chance or ‘special circumstances’ that link
positive findings (Swartz et al., 2009, 2010), with the quali- psychosis and violence. Swanson et al. (1990) may be
fication that this New York programme is probably more regarded as the pioneers of an acceptable epidemiol-
comprehensive than other such programmes in the USA. ogy of psychosis and violence, using US Epidemiologic
Improvements in many aspects of the lives of the people in Catchment Area (ECA) interview data from over 10, 000
the programme were reported, including reduction in use of people who participated in a household survey. This
inpatient facilities and arrests, without those people feeling study was, thus, of a general population based sample,
unduly stigmatized and without adverse effect on their use albeit excluding some groups of particular interest in
of other services. this context, such as homeless people. Just over 12% of
the people with schizophrenia had reported being vio-
In Japan, in June 2001, the by now familiar story of a lent within the year prior to interview, compared with
young man with schizophrenia, well known to psychiatric about 2% of the people without mental disorder. When
services – ‘institutionalized more often than he had been schizophrenia alone was considered, there was just a
arrested’ – and yet never quite connecting with those
336 © 2014 by Taylor & Francis Group, LLC
Psychosis and crime: The epidemiology
four-fold elevation in rate of violence. Depression alone analyses were applied. In this cohort, only three conditions
was associated with about twice the rate of violence of independently affected violence rates at age 21: schizophre-
the general population. Having more than one diagnosis, nia spectrum disorder, alcohol dependency and cannabis
however, had a substantial effect, with over 30% of the dependency, with a five-fold (criminal convictions) to seven-
people with schizophrenia and substance misuse disor- fold elevation in (self-reported) violence rates among those
ders being violent. with schizophrenia spectrum disorder (Arseneault et al.,
2000). No study is without its m ethodological problems,
Interpretation of other studies requires the observation however. Here, although bias by subgroup exclusion was
that while some are diagnostically precise, and some refer eliminated, difficulties lie in the sample characteristics.
only to schizophrenia, others take a broader definition of There were just six people in the subgroup with criminal
psychosis, albeit noting that schizophrenia is the most convictions for violence at age 21, and only 13 in the vio-
common of the psychoses to be associated with violence. lence self-report group. At age 26, the relationship between
Still other studies take the concept of ‘major mental dis- schizophrenia spectrum disorders and violence remained
order’, which seems to more or less equate with psycho- similar, but numbers in the groups of interest were still
sis. The picture for schizophrenia and schizophrenia-like small (Arseneault et al., 2003).
psychoses is generally clearer than that for bipolar and/
or depressive psychosis. It may even be that mania per se Proportions of People Who
is protective against violence, indeed some have argued Kill Who Have Psychosis
that, providing that a person has a fixed belief in his or her
omnipotence, then violence is unnecessary. Häfner and Another popular way of estimating the frequency of asso-
Böker (1973) found just one case of mania among their ciation between psychosis, generally schizophrenia, and
national study series of homicides in Germany 1955–64 and violence has been to study homicide figures. Homicide
Schipkowensky (1968) found just four over the 40 years up is chosen for this purpose because it is a crime with a
to 1965 in Sofia, Bulgaria. Craig (1982) failed to find a single high clear-up rate, and so it is generally considered that
episode of violence among 20 consecutive admissions to studying national or regional criminal statistics in this
hospital with mania, despite the high levels of anger and regard will yield something very close to the real figures.
agitation among these people. The stories for bipolar disor- It is, though, also an unusual and generally non-recidivist
der and depressive psychoses in these studies, however, dif- crime (see also chapter 19). Furthermore, like other
fer from the much more recent study using Swedish health violence, there has been consistent evidence over time
and criminal justice registers in Sweden (Fazel et al., 2010). that the contribution of people with mental disorder to
Like other recent studies, there is no separate considera- homicide statistics will depend in part on the base rate
tion of mania in this study, but rather 3,743 individuals with for homicide (Schipkowensky, 1973; Coid, 1983; Reiss and
two or more discharge diagnoses of bipolar disorder were Roth, 1993), which varies substantially internationally (e.g.
identified in the 32-year period 1973–2004 and compared Reiss and Roth, 1993). Large et al. (2009) have, however,
with over 37, 000 general population controls. There was posed a persuasive challenge to this assumption, based
a small elevation of risk among those with pure bipolar on a systematic review and meta-analysis of all published
disorder, but people with bipolar disorder in combination studies, which included diagnostic assessments, from
with substance abuse disorders were about 6.5 times more developed countries, between 1960 and 2008. This does
likely to be violent. not, of course, demolish the argument that, untreated,
schizophrenia or similar psychoses may impose some risk
Focus on schizophreniform psychoses suggests that the for violence. The figures which follow underscore that,
order of elevation in rate of violence by people suffering but also suggest the importance of risk factors likely to
with one is something in excess of four times the general be common to all homicides, and the obvious point that
population rate. This is fairly consistent from Swanson tackling these more general risk factors, including avail-
through other well constructed association studies emerg- ability of weapons, alcohol and illicit drugs and social
ing through the 1990s, despite the fact that details of method adversities may exert an influence on people with psycho-
differed. Five studies linked health and criminal records in sis as much as people without.
Nordic birth cohorts (Hodgins, 1992; Hodgins et al., 1996;
Tiihonen et al., 1997; Brennan et al., 2000; Ortman, 1981, The proportion of homicides attributed to people with
available only in Danish). An Australian study linked health psychosis across a wide range of countries is remarkably
and criminal records through the electoral role in Australia consistent, with the tiny population of Iceland being the
(Wallace et al., 2004). The Dunedin birth cohort provides only real outlier: Australia (New South Wales) 6.7% (Nielssen
for a true community survey, relying on repeated interview et al., 2007); Australia (Victoria) 7.1% (Wallace et al., 1998);
and records searches in a longitudinal design with remark- Austria 5.4% (Schanda et al., 2004); Finland 5.7% (Laajasalo
able follow-up rates of over 90% (see chapter 2 of Moffitt and Hakkanan, 2005, 2006); Germany 8.9% (Erb et al.,
et al., 2001). For estimates of the quantitative relationship 2001; Hafner and Boker, 1973); Iceland 28% (Petursson and
between psychosis and violence, however, cross-sectional
© 2014 by Taylor & Francis Group, LLC 337
Psychosis, violence and crime
Gudjonsson, 1981); Singapore 8.2% (Koh et al., 2006); Sweden representative of the US general population. People were
8% (Lindqvist, 1989), 8.9% (Fazel and Grann, 2006); New sampled from every state and from homeless or temporar-
Zealand 3.8% (Simpson et al., 2004); the UK 5–8% (NCISH, ily housed populations as well as people in stable homes,
2010); USA (Contra Costa County, California) 10% (Wilcox, and longitudinal data were collected in two waves about
1985). There is also some consensus on trends over time, three years apart (2001–2003 and 2004–2005). Only the
with indications of a reduction in the proportion of homi- institutionalized were excluded. The research question
cides committed by people with a mental disorder at least was about whether mental illness at the first wave of data
since the 1950s (Taylor and Gunn, 1999; Simpson et al., collection predicted violence between the first and second
2004). Some recent increase and restablilization is evident interviews. It did, but only when associated with a history
in England and Wales (NCISH, 2010); in Nordic countries, of violence and/or other antisocial behaviour, substance
as in other Western countries, there has been an increase abuse, and/or certain demographics (sex, age, income) and
in overall homicide rates during the 1990s and 2000s, but contextual factors, which were stressors of various kinds.
with alcohol and other drug misuse rather than psychosis
accounting for this rise (Gudjonsson and Petursson, 2007). The authors concluded that, because mental illness
did not independently predict future violent behaviour,
The Impact of People with Psychosis these findings challenge perceptions that mental illness is
on Violence in the Population a leading cause of violence in the general population. This
is persuasive research, as far as it goes with the longitudi-
Swanson and colleagues (1990) emphasized that their work nal model, but the conclusions seem influenced as much
was suggestive only of a higher relative risk of violence by policy considerations and concerns about stigmatiz-
by people with psychosis. Among the more than 10 000 ing people with psychiatric disorders as with science. It
respondents in three of the four US cities participating is hard to think of anyone in this research field who has
in the ECA survey (Baltimore, Raleigh-Durham and Los ever suggested that mental disorder is a leading cause of
Angeles) just 3% of the violence was accounted for by peo- violence. The common claim since 1990 is that psychosis
ple with schizophrenia. Measuring the impact of psychosis is a small but significant contributor – and, more to the
in this way not only yields figures which are more helpful point, a contributor that may be amenable to prevention
for public health initiatives, but makes allowance for pos- or measurable decrease through appropriate treatment.
sible differences in violent episodes per person and some Furthermore, this study neither considers directional rela-
other confounding factors. Wallace et al. (2004) showed tionships between substance misuse and psychosis nor
the potential relevance of this in finding that people with impact of treatment. The UK based National Comorbidity
schizophrenia were convicted of about twice as many Study of about 1.4 million people attending general prac-
offences over a lifetime as people without. Accordingly, tices 1993–98 (Frischer et al., 2004) suggested a greater
there have been a number of studies of ‘population attrib- vulnerability of people with psychosis going on to develop
utable risk’. In the Wallace et al. (2004) study this was very substance misuse disorders than the other way around.
similar (3.2%) to the Swanson figure. Fazel and Grann When the MacArthur study of mental disorder and vio-
(2006) studied linked crime and health registers for people lence group examined the relationship between psychotic
over the age of criminal responsibility (15) in Sweden for symptoms and violence in terms of symptoms at time-1
the 13-year period 1988–2000. The average population for predicting violence at time-2, without taking account of
the period was just over 7 million, and the number of peo- treatment of or change in those symptoms, they found no
ple discharged from hospital with a diagnosis of psychosis relationship between symptoms and violence (Appelbaum
just over 98, 000. The population attributable risk for psy- et al., 2000). Subsequently, however, they demonstrated
chosis generally was 5.2% and for schizophrenia specifically symptom change over those periods (Appelbaum et al.,
6.3%. In the UK, however, in a computer assisted interview 2004) and an overall relationship between specific psy-
household survey in the year 2000, Coid et al. (2006) found chotic symptoms and violence (Monahan et al., 2001).
this risk to be just 1%, little different from the likely repre- Interpretation of longitudinal data on people with psy-
sentation of psychosis in the community. chosis, especially when they have been identified by their
patient status, must take account of treatment and illness
Longitudinal Risk Based Studies changes if it is to be useful.
Elbogen and Johnson (2009) have taken a welcome longi- Fazel et al. (2009a) examined hospital and criminal
tudinal perspective on the relationship between psychosis records data on just over 8000 patients with schizophrenia
and other mental disorders. They used the US National subsequent to their first recorded admission in the period
Epidemiologic Survey on Alcohol and Related Conditions and 80, 000 general population controls in Sweden (1973–
(NESARC) database to study this. The advantages of this 2006). After adjusting for demographics, these patients
database are that it is large – 34,653 participants – and were about twice as likely as the controls to have had
at least one violent conviction (13.2% : 5.3%). Substance
abuse comorbidity, however, was a strong mediator; such
338 © 2014 by Taylor & Francis Group, LLC
Psychosis and crime: The epidemiology
patients were over four times more likely to have a violent abuse, and use it to exclude people from services, sometimes
conviction than their general population peers while those with disastrous consequences (e.g. Health Inspectorate
with schizophrenia alone had only a slightly increased odds Wales, 2008). Diagnosis of any kind is an approximate art,
ratio (1.2; CI 1.1–1.4). In an interesting additional analysis, and the kind of research diagnoses used in these studies
it was found that when the people with schizophrenia were tend to be reliable operationalized descriptions of presenta-
compared with their siblings who did not have the disor- tion rather than diagnoses in a conventional medical sense,
der, the substance misuse comorbidity effect was much with its implications of understanding causation, context
less, suggesting that genetic or early environmental factors and course (Taylor and Gunn, 2008). Could reliance on
also had their part to play. This finding further underscores measurement of psychotic symptoms rather than focusing
something with which practising forensic clinicians are on diagnosis be more helpful?
already familiar, that substance misuse among people
with psychosis must be assessed, and, where the misuse Psychotic symptoms as a more useful or valid
amounts to more than a simple palliation of symptoms, variable to test for relationship with violence?
must be treated in its own right. More importantly, it is The prevalence of psychotic symptoms is substantially
not, in most cases, an alternative explanation for violence higher than the prevalence of diagnosed or treated psy-
by people with psychosis. Rather the inter-relationships chotic illness. This was suggested in Mackay’s (1857) his-
between illness and other relevant factors must be seen as torical description of ‘popular delusions and the madness of
many and complicated. crowds’, but there is now also systematic scientific evidence
in support of this observation. In the Netherlands Mewal
The Epidemiology of Psychosis and Health Survey and Incident (NEMESIS) study at least 17.5%
Violence: Need for More Knowledge of participants reported at least one psychotic symptom,
and Less Interpretation? in a population in which the prevalence of a DSM-III-R
non-affective psychosis was 2.1% (van Os et al., 2000). This
So, a small but significant relationship between psychosis led to the idea of a symptomatic continuum between the
and perpetration of violence has been consistently reported general population and people diagnosed with psychosis,
in a variety of studies since 1990. Fazel et al. (2009b) and that a more instructive route to treatment may lie
brought most of the studies just described together with a in investigating factors which influence the modulation
few additional ones in a systematic review; the additional of psychotic symptoms – and their effects – rather than
studies were of relatively small geographical areas and/or being centred on diagnosis (Verdoux and van Os, 2002).
patient samples defined by an episode of treatment so long Other researchers, in other countries, have also found that
as there was a general population comparison group. They the prevalence of psychotic symptoms exceeds expected
conducted a meta-analysis despite the diverse range of rates of disorder. Mojtabai (2006), for example, in a US
methodologies, with different sampling techniques, differ- household survey, found that 5% of over 38,000 respondents
ent measures of psychosis and/or violence, different times described what he referred to as ‘psychotic-like’ symptoms,
scales for those measurements and procedural differences. so named because the data had been collected by trained
Although they did some separate analyses according to lay researchers rather than clinicians. In an Australian
gender, and separate homicide from other violence, they national household survey, Scott et al. (2006) found that,
mixed cross-sectional and longitudinal work. They con- among over 10,500 respondents, 4.5% endorsed at least
cluded nonetheless that there is about twice the risk of one delusional experience. In the UK National Comorbidity
violence among people with ‘pure’ psychosis compared Survey 2,406 respondents completed two interviews, on
with the general population, but about eight times the risk average 18 months apart; 414 (10.9%) had at least one
if substance misuse is involved. This may be vital informa- psychotic symptom at first interview and 128 (3.3%) had
tion, but the preferred emphasis on the substance misuse symptoms persistent between the two (Wiles et al., 2006).
could mislead the unwary: A cross-national study by Nuevo et al. (2012) included
256,445 people from the general population of 52 countries
Most of the excess risk appears to be mediated by sub- and found a 6% prevalence for hallucinations and 5–8% for
stance abuse comorbidity. The risk in these patients with delusions, with 12.5% having at least one psychotic symp-
comorbidity is similar to that for substance abuse without tom, albeit with much variation from country to country.
psychosis.
Few of these studies were in a position to go one step Link and Stueve (1994) were probably the first to rec-
further and cite comorbidity with personality disorder, in ognize the role of symptoms independently of diagnosis in
some cases since they often relied on measures of personal- the generation of violence. They identified 367 patients in
ity disorder for identifying the violence (e.g. Swanson et al., various stages of treatment and 286 never treated com-
1990). In practice, however, psychiatrists often do take this munity controls, all from New York State. Everyone was
step and infer personality disorder from comorbid substance interviewed using the Psychiatric Epidemiology Research
© 2014 by Taylor & Francis Group, LLC 339
Psychosis, violence and crime
Interview (PERI; Dohrenwend et al., 1986) which, among a those who did not mainly by having retained an unimpaired
full range of symptoms, allows recording of three referred interest in sexual relationships while struggling with the
to as ‘threat/control-override symptoms’ (TCO symptoms). emotional aspects of interpersonal relationships. Official
There has been some debate as to whether these constitute statistics for men detained in hospitals under mental health
true psychotic symptoms, as the questions are: ‘how often do legislation with restrictions on discharge at about the same
you feel that … your mind was dominated by forces beyond your period indicated that 10% had been convicted of a sexual
control? … thoughts were put into your head that were not your offence (Home Office, 1997). Smith and Taylor (1999) exam-
own? … that there were people who wished to do you harm? ’ ined a complete national (England and Wales) sample for
They do seem, however, to reflect the kind of experience the month of May 1997 and identified 84 male sex offenders
as many patients with psychosis would describe them. The with schizophrenia at that time. For most, antisocial sexual
violence measures were hitting someone in the last year and/ acts were only part of their offending repertoire. All but four
or fighting and/or weapon use in the previous 2 years. There of them had been psychotic at the time of the index offence,
was a difference between patients and never treated con- and about half had hallucinations or delusions directly
trols on these violence measures, but the significance of this relevant to the offending. Fazel et al. (2007a,b) conducted a
difference was reduced by entering the range of psychotic much more substantial study, using Swedish national regis-
symptoms in a preliminary regression analysis. Several alter- ters. They identified all 8,495 male sexual offenders on record
native statistical models were tested, but when TCO symp- for the 3 years 1988–2000 and compared them with a ran-
toms were treated separately from the other symptoms, dom sample of nearly 20, 000 men in the general population.
and potentially relevant demographic variables were also Twenty per cent of such crimes had been committed by men
allowed for, the TCO symptoms accounted for the violence, who were or had been hospitalized for psychiatric disorder.
rendering the patient–non-patient status non-significant. Men with psychosis were just over five times more likely to
have a psychotic illness than the general population (OR: 5.2,
Swanson et al. (1996) applied measures of these symp- 95% CI: 3.9–6.8; schizophrenia specifically OR: 4.8, 95% CI:
toms to the ECA data, referred to above. They provided fur- 3.4–6.7). The temporal and causal pathways between sexual
ther endorsement of the potential importance of psychotic offense and psychiatric disorder deserve further scrutiny.
symptoms in the absence of illness in relation to occur-
rence of violence as participants with TCO symptoms were Psychosis and arson
about five times as likely as those without to have been It is now generally accepted that, in England and Wales at
violent in the previous year, but here the independence of least, the largest group of women detained in secure hospi-
the symptom relationship from diagnosis or service use is tals according to offence – about 40% – is of women who
less clear. Mojtabai’s (2006) household survey, also already have set fires or been convicted of arson (e.g. Coid et al.,
referred to, took up the psychotic symptom–violence ques- 2000; see also chapter 20). From Swedish national registers,
tion too. Attacks intending to hurt, violence to an intimate Anwar et al. (2011) extracted all 1,340 men and 349 women
partner and arrest for aggravated assault were each about who had been convicted of at least one offence of arson in
five times more likely among people who had reported the 13-year period 1988–2000, and compared them with a
psychotic symptoms, here not necessarily TCO symptoms. sample of over 40,000 of the general population over the
age of criminal responsibility (age 15), randomly selected
On logical and empirical grounds, then, there is reason for sex and year. People with non-schizophrenic psychosis
to explore further the role of symptoms, with or without had a substantially elevated rate of arson, but those with
diagnosis, in understanding the link between psychosis schizophrenia were most at risk, with men over 20 times
and violence. more likely and women nearly 40 times more likely to have
had this illness if they had been convicted of arson (men:
Psychosis and Offending Other OR: 22.6, CI: 14.8–34.4; women 38.7, CI: 20.4–73.5).
than Interpersonal Violence
Psychosis and other offending
Psychosis and sex offending People with schizophrenia or other chronic psychoses may
Many would regard sex offending as a form of violence but, be more likely to be convicted of all sorts of less serious
as it may have somewhat different implications for treat- crime than their healthy peers, although in this context,
ment, separate mention seems useful. Sexual violence and less serious crime such as property damage or vagrancy is
offending has been neglected relative to research on violence much less studied. A factor in any such excess rates may be
in schizophrenia. Philips et al. (1999), in a small series of the greater likelihood of arrest without detection of people
high security hospital patients, found that about 15% had with psychosis (Robertson, 1988), but also there may be
been convicted of a sexual offence or had behaved in an vulnerabilities through social marginalization of various
antisocial sexual way. In almost all of these cases the sexually kinds, including homelessness and poverty, and, for the
offensive behaviour had post-dated the onset of the illness
and occurred in the context of psychotic symptoms. The
men who showed such behaviours were distinguished from
340 © 2014 by Taylor & Francis Group, LLC
Pathways into violence through psychosis: Distinctive or common to most violent offenders?
increasing proportions who complicate their illness with neurodevelopmental nature of the disorder, and this is the
alcohol or drug use, a need to finance this ‘medication’. area where ever more evidence is called for (e.g. Yung and
There is some evidence that, when committed by people McGorry, 2007). Although some of the maternal factors
with psychosis, this sort of crime is more likely than violent cited may be indirect evidence of social disadvantage, in
crime to follow from rational motives (Taylor, 1985b). many cases this is an area of research where all too little
is known. Early work, to an extent, captured the public
Pathways into violence imagination, but seemed to lay blame rather than provide
through psychosis: explanations of potential therapeutic value, for example
Distinctive or common to on parenting styles (e.g. Fromm-Reichmann, 1948; Bateson
most violent offenders? et al., 1956) or the family more generally (e.g. Laing and
Esterson, 1964; Wynne et al., 1958). The more scientific
Discovery of patterns in the individual and/or his/her 1960s work on expressed emotion was about emotional
environment prior to the onset of the illness or behaviour climate in families once the disorder was established (e.g.
of concern is useful for establishing cause, course and iden- Brown et al., 1962, 1972; Leff and Vaughn, 1985). More
tifying potential intervention points. This certainly applies recently, evidence has been accumulating that early expe-
to schizophrenia and similar disorders, and early treatment rience of abuse and/or neglect is not only associated with
is increasingly seen as important such that, at best, resolu- psychosis, but also in the causal pathway (Read et al.,
tion of prodromal states might prevent the development 2005), but this group highlighted the relative poverty of
of the full blown illness (e.g. Mrazek and Haggarty, 1994) work in the field. They noted that their systematic search
or, at worst, rigorous treatment of the illness per se could of the literature identified just 23 articles about child abuse
prevent cumulative damage. Insofar as offending behaviour and schizophrenia – 0.05% of all the articles captured on
among people with psychosis is consequent upon the ill- schizophrenia overall, compared with 4.1% on the genetics
ness, then this too might be prevented. The difficulty is that of schizophrenia and 8% on brain damage or dysfunction
the prodromal phase of illness is often not distinctive, with and schizophrenia.
depression, anxiety, stress and substance misuse among the
features, and treatment with medication carries its own set The environmental influences cited above may be
of risks. taken as part of the ongoing debate about the social causes
and contexts of mental illness. Kirkbride and Jones (2011)
Although schizophrenia and similar psychoses are gen- examined data in support of taking a primary prevention
erally regarded as illnesses because, at least in full blown approach to schizophrenia. They argued that by taking an
form, they emerge after a long period of relative normality, ‘eco-epidemiological view’ it is apparent that:
in many cases there is an argument to be made that the
schizophrenias, like conduct and personality disorders, are variation exists along a number of other fascinating
also developmental disorders. Chapter 8 provides evidence domains, including migration and minority status,
of genetic influences in the emergence of schizophrenia as place of birth and upbringing, life events and social
well as antisocial behaviour, and also substance abuse, the disadvantage, pre- and perinatal stressors (i.e. famine
latter so often complicating either or both of the former or viruses).
conditions. Two important points emerge. The first is that
we are only just beginning to learn about the particular Similarly, in a recent re-analysis of the leading social
genetic organization that may contribute to the psycho- cause hypotheses, Hudson (2005) concluded that:
ses, and even underpin overlap between it and these other
conditions. Secondly, even in the presence of a genetic role the idea that the impact of SES (socioeconomic status)
in the condition or its associated disorders, a substantial on mental illness is mediated by economic stress received
amount of the variation in behaviour in these conditions the strongest support, with this model substantially fitting
must still be accounted for through environmental dif- the data.
ferences. Environmental candidates for which there is at
least moderate evidence include epigenetic factors such As these two perspectives coalesce, we find further
as obstetric complications (e.g. Geddes et al., 1999), mater- impetus to take a developmental, socially situated view not
nal use of alcohol and other drugs during pregnancy (e.g. just of causal pathways to psychosis, but of the environ-
Frank et al., 2001) and maternal infections during preg- ment in which symptoms are experienced and expressed.
nancy (e.g. Adams et al., 1993). The literature has tended to
emphasize externally observable physical differences (e.g. This is not the place, however, for an extended review
Lane et al., 1997) or delays in motor or cognitive develop- of pathways into the psychoses per se but rather to explore
ment (e.g. Marenco and Weinberger, 2000) between people the pathway through psychosis to violence. Longitudinal,
with schizophrenia and those without as evidence of the often birth cohort, studies have provided invaluable infor-
mation on pathways into crime and violence (see chapter
7). Again, the underlying hope is that a map of critical
developmental stages and of points of deviance from them
could assist with early intervention and prevent accumula-
tion of harm. One of the most often cited findings is of two
© 2014 by Taylor & Francis Group, LLC 341
Psychosis, violence and crime
principal pathways, albeit clearly evidenced only for men, an episode of schizophrenia, several studies made similar
of early-onset life-course persistent offending/violence and observations with respect to people who had schizo-
late-onset adolescent-limited offending/violence (Moffitt, phrenia, suggesting that if people with such illnesses did
1993a; Moffitt et al., 2001). The first is likely to reflect an offend, it might be a later complication of their illness,
interaction between individual genetic and physical factors rather than the illness being more-or-less coincidental.
on the one hand and environmental factors on the other, Walker and McCabe (1973), for example, compared the
while the second is much more environmentally dependent. records of three 1963/4 cohorts of people – those who
were detained in hospital as offenders with mental dis-
Given the prevalence of schizophrenia in the general order, everyone found guilty of an indictable offence and
population, it is hardly surprising that the birth and child- everyone admitted to a mental hospital – and found the
hood cohort studies which include repeated, multi-source mentally disordered offenders were significantly older
based evaluations over time include too few people with than the other two groups. Häfner and Böker (1973), with
schizophrenia, schizophrenia spectrum disorders or other a 1955–1964 German national cohort of homicides, found
psychoses to offer adequate information on pathways into that, although a person might kill during a first episode of
violence for such groups. The Cambridge study, for exam- schizophrenia or depression, this was very unusual and
ple (see chapter 7), included no one with schizophrenia or that people with schizophrenia were generally 10–15 years
similar psychoses, while in the Dunedin birth cohort of over older than people without a mental disorder, and people
1000 individuals, there were just six at age 21 with a schizo- with an affective psychosis even older still. Mowat (1966),
phrenia spectrum disorder and a conviction for a violent in his study of morbid jealousy among high security hospi-
offence, and just 13 with the disorder and a self-reported tal patients, added the observation that some subgroups
history of violence (Arseneault et al., 2000). Nevertheless, of homicidal offenders with psychosis might be older still –
the Dunedin study both at age 21 and at age 26 (Arseneault here suggesting that the older age of the morbid jealousy
et al., 2003) hints at more than one route into violence for reflected the length of time needed for the delusions to
this group. A regression analysis showed that about one- develop to the point of driving violence. There is some
third of the risk of the adult schizophreniform disorder suggestion from later research, however, that the age dif-
violence could be accounted for by childhood psychotic ferential may have decreased since inpatient hospitaliza-
symptoms, but childhood violence independently and tion has been less routine; in an English pretrial prisoner
together with these symptoms also accounted for a small study the age difference between alleged offenders with
but significant effect. This finding, however slight, pro- and without psychosis was 3–4 years (Taylor and Gunn,
vides some prospective evidence for the idea that there 1984; Taylor, 1987).
may indeed be different pathways into psychosis related
violence – one related chiefly to the illness and likely to be Two different research approaches from Sweden
symptom driven and one embedded in comorbidity and hinted at two developmental paths. Historical data for
historically more complex social relationships. a Stockholm county patient discharge cohort revealed
two peaks of offending onset, more typical of men than
Larger, mainly Nordic country birth cohort studies, women, and different from the offending pattern of the
without the richness of data of these interview studies, healthy comparison group (Lindqvist and Allebeck, 1990).
can and do answer simpler questions about relative age of A Swedish birth cohort study, for which hospital and
onset of psychosis and offending, while criminal justice and criminal records were obtained, found a peak of onset
health service case register studies also contribute. Studies, of offending in the mid-teenage years and a second peak
albeit not always prospective studies, with a starting point after age 21 for the men but a single later peak for the
of childhood abuse indicate that this also is worth investi- women (Hodgins, 1992). Wessely et al. (1994) found similar
gation as an antecedent of both psychosis and antisocial variation in male : female offending patterns when com-
behaviour and, probably, the combination. Studies which paring case register people with and without schizophre-
track psychotic symptoms over time, and their temporal nia who had had any contact with psychiatric services in
relationship to violence are also developmental studies, but South London. The possible clinical relevance of the two
of a rather different kind, and will be dealt with separately. developmental pathways was suggested by an English
high security hospital resident cohort in which two dis-
Early Versus Late Onset of Offending tinct groups of people with psychosis were apparent:
Distinctions Among People with Psychosis those who had been unremarkable until the onset of their
illness, with an index offence almost invariably reported
East (1936) was among the earlier researchers to draw as having been driven by psychotic symptoms and those
attention to the older average age of people with major who had established conduct and/or emotional disorders
mental disorder who had killed compared either to con- in childhood, continuous with adult personality disorders,
victed killers without such disorder or people with psy- who had also developed an illness indistinguishable from
chosis who had not offended. During the 1960s and 1970s, schizophrenia. The index offence in the comorbid group
when inpatient treatment was still more routinely used in
342 © 2014 by Taylor & Francis Group, LLC
Pathways into violence through psychosis: Distinctive or common to most violent offenders?
was significantly less likely to have been driven by psy- than their non-offending peers (Munkner et al., 2003b).
chotic symptoms (Taylor et al., 1998). It is unclear whether there were distinct groups of
early- and later-onset offenders among these people
For people with schizophrenia, therefore, concepts of with schizophrenia, with the early-onset group follow-
early- and late-onset offending are promising. Although not ing a more typically criminogenic pattern, or whether
formally tested, it appears that both the earlier and the later age at presentation and/or diagnosis was an artefact
onset groups tend to be older at onset than their respective of antisocial behaviour distancing them from health
non-psychotic counterparts and desistance from offend- services.
ing tends to occur earlier than in any of the non-psychotic
samples; this may be related to treatment. The other major Gosden et al. (2005) also used Danish criminal justice
area of pathway research here lies in attempts to estab- (NCR) and health (DPCR and Danish National Cause
lish whether the illness or the offending start first, or at of Death Register) registers to conduct a prospective
least whether mental health service involvement precedes study of the careers of 794 15–19-year-old males and
criminal justice service use as, even in prospective studies, 54 15–19-year-old females in a complete 1992 national
it is much harder to date precisely the onset of particular cohort of people in contact with the Danish Probation
behaviours, or the point at which they may be regarded as and Prisons Service, followed up in 2001. From birth
pathological. to index day (the start of the study), just over 3% of the
cohort of 732 male and 48 female survivors with ade-
In Finland, Laajasalo and Häkkänen (2005) studied a quate records had been diagnosed with schizophrenia
consecutive series of people with schizophrenia who had and 4.5% with any psychosis; 12 of the young men, but
been charged after a homicide between 1983 and 2002. none of the young women, had been convicted of serious
They observed little difference in index offence characteris- violent crimes, including homicide, and 382 of the young
tics, but that, for some, the homicide was the only offence men and 15 of the young women of lesser violence; just
while others had sustained at least one criminal convic- 59 young men and seven young women had ever been
tion before the onset of their illness; the latter group were admitted as psychiatric inpatients. Two variables were
significantly more likely to have had established behaviour independently related to a diagnosis of schizophrenia/
problems in childhood. While these findings tended to sup- psychosis – previous psychiatric admission and previ-
port the two-pathway model, Fresán et al. (2004) studied ous violence – offering yet further evidence for two
a consecutive series of 75 outpatients with schizophrenia subgroups of people whose psychosis is complicated by
in Mexico, and found that worse premorbid adjustment violence, and also that they tend to be treated differently
through childhood and adolescence was characteristic from an early stage.
of the violent group – so here the suggestion was of two
developmental pathways into schizophrenia, but one main The extant research, therefore, despite limitations in
route into schizophrenia with violence. The latter sample, method and varying findings, makes a strong case for
however, was small, and the violence confined to levels continued pursuit of pathways to the co-occurrence of
which were apparently manageable in the community. psychosis, psychotic symptoms, and violence. It is through
these exploratory and descriptive studies that we will
Munkner et al. (2003a) used the Danish Psychiatric develop the next generation of more targeted and precise
Central Register (DPCR), the National Crime Register inquiry which, of necessity, will precede launching effective
(NCR) and the Civil Registration System (CRS) to iden- intervention trials.
tify all 4,619 live individuals in Denmark born on or
after 1 November 1963 with a diagnosis of schizophrenia First Episode of Illness Offending
and who had attained the age of criminal responsibil-
ity; 41% had been convicted of any criminal offence, Humphreys et al. (1992) were among the first to study
17% of at least one violent offence. Men were more violence, ascertained from multiple sources, including self-
likely to have sustained convictions than women. Again, report, family or friend reports and records, in a consecutive
men and women differed in the temporal relationship series of 148 men and 105 women referred while in a first
between onset of offending and first presentation to episode of schizophrenia; the study period was 1979–81
health services with their psychotic illness; 71% of the and the area was within a 35-mile radius of a hospital in the
men with a criminal record but only 37% of such women south of England. Most had not been violent at this stage,
had established their criminal career before the onset but a fifth of them had behaved in a way that was threaten-
of their illness. Of all violent offenders, 58% of the 700 ing to the lives of others. This did not, however, imply that
men and 21% of the 81 women had sustained their first the violence had necessarily preceded the illness. About
violent conviction before their first psychiatric contact. half of the incidents occurred after the person had been
In a second study with the same cohort, however, it was ill for at least a year, and violence was more common the
apparent that those with schizophrenia who had been longer the person had been ill. Volavka et al. (1997) found
convicted of a criminal offence were older at first con- a similar overall figure for violence during a first episode of
tact with services and first diagnosis of schizophrenia
© 2014 by Taylor & Francis Group, LLC 343
Psychosis, violence and crime
schizophrenia, at a similar time (1978), when drawing on group’s conclusions might be misleading because of the
a sample of 570 men and 447 women from three develop- diagnostic heterogeneity of the samples included in their
ing and seven developed countries (including the UK); this review. Morgan and Fisher tried to focus on more diag-
overall figure of 20% violent in their first episode, however, nostically homogeneous studies. Although they found the
obscured a three-fold difference between the developing evidence to be weaker, they considered the likelihood of
(31.3%) and developed (10.5%) countries. a link between trauma and psychosis to be enhanced by
some plausible explanatory links through the impact of
In a study of 280 psychotic twins and 210 healthy stress on the dopaminergic system. We have subsequently
co-twins Coid et al. (1993) established that, not only did found volumetric brain differences which may differen-
a higher proportion of the psychotic twins (21%) have a tiate traumatized men with psychosis who have been
criminal record than their co-twins (11%), but also that violent from those who have not, the former suffering
there was a significant correlation between offending and thalamic loss and the latter hippocampal and frontal lobe
onset of illness. Specific examination of the violent offend- loss (Kumari et al., in press).
ers showed that the violence clearly post-dated the onset
of the illness in 12 of the 14 men and both women with Subsequent to these reviews, Scott et al. (2007), with
schizophrenia who had been violent. One of the remaining an Australian sample of 10,641 in the National Survey of
men had been in prodromal phase of his illness and for one Health and Wellbeing, effectively endorsed this view, albeit
only did the offending clearly precede psychotic symptoms referring to people traumatized at any age. There was not
(Taylor and Hodgins, 1994). only a relationship between number and type of events and
endorsement of delusional experiences, but also, where
Large and Nielssen (2011) conducted a systematic a diagnosis of post-traumatic stress disorder (PTSD) had
review of studies of violence during a first episode of schizo- been made, there was almost four times the rate of report-
phrenia. They identified nine studies (Dean et al., 2007; ing delusions among the traumatized than when they did
Foley et al., 2007; Harris et al., 2010; Humphreys et al., 1992; not reach criteria for PTSD. Bebbington et al.’s (2004) study
Milton et al., 2001; Spidel et al., 2010; Steinert et al., 1999; of 8580 adults in a household survey – the British National
Verma et al., 2005; Volavka et al., 1997), with a total of 2,545 Survey of Psychiatric Morbidity – also suggested that type
cases. From various meta-analyses, the pooled estimate of of abuse was relevant; people reporting having been sexu-
serious violence during the first episode from all the studies ally abused before the age of 16 were nearly four times more
except Spidel was 16.6% (CI 12.9–21.3%). This occurrence likely to have psychosis while people who had experienced
of serious violence was associated with ‘forensic history’, violence were twice as likely to be psychotic.
duration of untreated illness and total symptom scores.
As with all material in this field, however, it is perhaps
Childhood Trauma in the Pathway better to consider the results as suggestive rather than
to Violence Through Psychosis conclusive. One substantial prospective study in Australia
had diametrically opposite results, although later studies
The way in which seriously adverse childhood experiences in the Netherlands, Germany and Britain have endorsed the
fit into the pathways through psychosis to violence is, as association between childhood trauma and later psychosis.
yet, unclear. Studies generally do not explicitly map them The Australian study was of people who had experienced
into such developmental models, but rather into the causes verified abuse (Spataro et al., 2004). No relationship was
of psychotic symptoms or psychosis (e.g. Read et al., 2005) found between childhood sexual abuse and psychosis,
or the emergence of violence (e.g. Widom, 1989; Widom although there were some important biases in this study;
and Maxfield, 2001) separately. Work with people with the average age of the people who had suffered abuse,
schizophrenia in just one of the English high security hos- in the early 20s, was significantly lower than that of the
pitals suggested that early-childhood victim experiences comparison group, so ‘time at risk’ for schizophrenia was
clustered in the conduct disorder subgroup, but that was a significantly different. This may not have been very impor-
small sample (n = 101), selected for perceived high threat to tant as Kelleher et al. (2008), in a small British study of 12
others (Heads et al., 1997). to 15 year olds, found an association between reports of
abuse or bullying and psychotic symptoms even within
A systematic review by Read et al. (2005) covered this age group. Read and colleagues (2005), however, raised
published articles between 1872 and 2004, locating 46 a further interesting point with respect to the Spataro
after expanding the diagnosis to the wider concept of psy- study; although prospective studies of the effects of verified
chosis. They found, overall, a strong association between abuse are notionally methodologically among the most
reports of childhood abuse and schizophrenia, psycho- robust, many children never report their abuse, and so a
sis and psychotic symptoms, particularly delusions and ‘verified group’ may be unusual. Further, there may, after
hallucinations, with a suggestion of a ‘dose–response’ all, be little difference between prospectively documented
relationship – the more severe the abuse, the higher the and retrospectively reported maltreatment in this context
chance of psychosis. In their 2007 ‘critical review’, how- (Scott et al., 2012). Janssen et al.’s (2004) style of prospective
ever, Morgan and Fisher (2007) argued that the Read
344 © 2014 by Taylor & Francis Group, LLC
Psychosis, comorbid mental disorders and violence
study in the Netherlands differed in that it relied on collec- We have already made reference to the small-scale study
tion of self-reported childhood abuse in an adult general of Heads et al. (1997) in the UK. Swanson et al. (2002), in
population sample of 4,045, but at a time when the partici- the USA, found that, while early victimization in men with
pants were psychosis free. They were reassessed two years severe mental illness was directly associated with violence,
later. Depending on whether the psychosis measure was of for women factors both in mental state and environment
diagnosis or symptoms, the rate of psychosis in the abused current to the research evaluation mediated its effects.
group was between 2.5 and 9 times that of those who had This is important, as it may indicate differing inter-
reported no abuse. Spauwen et al. (2006), in a longitudinal vention opportunities, here between women and men.
German study, reported on 2,524 people who were 14 to 24 Participants were 280 women and 522 men with complete
years old at the time of first interview, and found an asso- data on victimization, violent behaviour and clinical and
ciation between childhood trauma, including abuse, and environmental variables from a larger study of sexually
psychotic symptoms, particularly when the trauma had transmitted diseases among people with severe mental
been associated with intense fear. Fisher et al. (2009), using disorder attending public health services. Victimization
data from the Aetiology and Ethnicity in Schizophrenia was determined by self-report. Two-thirds of the partici-
and other Psychosis (ÆSOP) case (390) control (391) study pants had a diagnosis of schizophrenia or schizo-affective
found a relationship between childhood abuse and adult illness, a further 17% bipolar disorder, and the rest ‘major
psychosis only in the women. depression’ or ‘other serious mental disorders’. The work is
particularly interesting, however, because, taking psycho-
Childhood trauma is probably one of the many risk sis/severe mental disorder as the underlying problem, it
factors for adult psychosis, and probably one for violent identified a hierarchy of impact of victim experiences on
behaviour also. Could there be an interaction between violent outcomes:
all three? The reality is that almost any unfortunate ●● People who had been a victim of physical or sexual
outcome in terms of illness or behaviour has been found
to relate to childhood trauma or adversity (Green et al., abuse before the age of 16, but not victimized as adults,
2010; McLaughlin et al., 2010; Kessler et al., 2010). A great were no more likely than those who did not report
deal more work needs to be done to clarify inter-relation- childhood victim experiences to be violent as adults.
ships, and how they work. To what extent are violence ●● Those who reported no such childhood experience, but
and psychosis related in such circumstances only by did report victimization as an adult were more likely to
the common antecedent of childhood trauma? To what report violence as an adult.
extent might they be related through the mediation of ●● Those who had both childhood and adult victim
substance misuse – the individual needing to ‘block out’ experiences were the most likely to have been violent
the memory of the experience, resorting to illicit drugs themselves.
to do so, and thus inducing psychosis, or at least trigger- The association of repeated victimization with later vio-
ing it in the context of a propensity for it? Bebbington lence was significant for both women and men, although
et al. (2011), on further work with the British National conferring greater risk on the men. Mediating factors for
Survey of Psychiatric Morbidity, found that the associa- the women were more serious illness (more hospitaliza-
tion between sexual abuse before age 16 and later psy- tions and more symptoms at evaluation) and recent home-
chosis was mediated by anxiety and depression, but not lessness. In the statistical models created here, substance
by heavy cannabis use or revictimization in adulthood. misuse was not an independent influence. This may be
The Janssen group’s explanation was that distress caused because of its complex relationships with abuse experience
by traumatic experiences may consolidate non-clinical and disorder. On the one hand it may be part of the risk
psychotic experiences, leading to psychotic illness (Bak taking behaviour of conduct/personality disorder, but on
et al., 2005). Neither of these studies, however, went on the other it may be a means of blocking out unpleasant
to explore how violence perpetration might come into experiences – victimization, psychosis or both.
this. Finally, there is the possibility too that there may
be a more or less direct relationship between childhood Psychosis, comorbid mental
trauma and psychotic symptom formation. There are disorders and violence
three major psychological theories about delusion for-
mation (Bentall and Taylor, 2006) and here the more apt When some researchers use the terminology of comorbid-
would rest in the idea that psychosis may arise out of an ity, they go beyond clinical conditions to include violence
unconscious attempt to construct a model of construing and antisocial behaviour per se (e.g. McCord and Enslinger,
the world that will improve self-esteem (Colby, 1977) or 1997). Others take a narrower operational view that two or
to attribute negative experiences to agents other than more conditions as defined in one of the disease classifica-
self (Zilger and Glick, 1988). tion systems must be present at the same time (Clarkin and
Kendall, 1992). Here we take Feinstein’s (1970) intermediate
The ultimate question here, however, is about interac-
tions between childhood trauma, psychosis and violence.
© 2014 by Taylor & Francis Group, LLC 345
Psychosis, violence and crime
position: the presence of ‘an additional clinical entity that 95% CI: 1.18–2.91). At first sight, the personality disorder
has existed or that may occur during the clinical course of rate may seem surprising in a general psychiatry cohort,
a patient who has the index disease under study’. Jaspers when only 20% of English high security hospital residents
(1923) took the view that the concept of comorbidity arises had such comorbidity (Taylor et al., 1998). There are two
out of the categorical approach to diagnosis. The American possible explanations. First, the Moran group included
Diagnostic and Statistical Manual (DSM-IV) partly deals schizoid personality disorder in their calculations, and
with this through the structural framework of multiaxial secondly a lower proportion of their study patients had
classification (American Psychiatric Association, 1994). schizophrenia. Both the general and the high security
Then, the term ‘dual diagnosis’ became popular through hospital studies suggested that schizophrenia was less
the 1990s, typically referring to the combination of a likely than other psychoses to be comorbid with person-
psychotic illness with a substance-misuse diagnosis. For ality disorder, although the Moran group found that this
forensic mental health practitioners, however, this rarely difference was not significant. Blackburn, by contrast,
captures the multiplicity of clinical problems suffered by favouring dimensional measures of personality, found
people who come to their services. Compton et al. (2005), that personality disorder comorbidity was the rule rather
in a study of nearly 2000 people in contact with general than the exception among high security hospital residents
psychiatric services in the USA, amply illustrated the range with a primary psychosis diagnosis, and that they were
of concurrent social disadvantage and dysfunction, includ- similar to patients with primary personality disorder on a
ing offending, presented by people with the combination of range of personality measures (e.g. Blackburn, 1974).
schizophrenia-spectrum and substance-misuse disorders,
but even this does not capture the evidence of developmen- In Finland, Putkonen et al. (2004) interviewed peo-
tal disorder and antecedent disadvantage that is apparent ple whom they described as constituting a nationally
for a substantial subgroup. Perhaps terminology such as representative group of homicide offenders with a major
‘triply troubled’ (Lindqvist, 2007) or even ‘multiply troubled’ mental illness, using the Structured Clinical Interview for
would better capture the complexity of presentations. DSM-IV Axis I and II disorders (SCID). They found that
more than half of their sample of 90 had a personality dis-
Psychosis, Personality order, and nearly three-quarters had a substance misuse
Disorder and Violence diagnosis. In this series all of those with major mental
illness and personality disorder also had a substance mis-
There is perhaps little distinction to be made between con- use disorder – triply troubled. Nestor et al. (2002), work-
cepts of anomalies in developmental pathways and comor- ing in the USA with a smaller group of 26 hospitalized
bidity of personality disorder with psychosis; indeed it is psychotic men who had killed another person, relied on
arguable that personality disorder should not be diagnosed the Psychopathy Checklist – Revised (PCL-R; Hare, 1991)
without substantial verified evidence of conduct and emo- as the measure of personality. They, like the UK research-
tional disorders through childhood, but there is now also a ers, found two distinct groups, but in this case one with
literature on personality measures made among adults with low psychosis and high PCL-R ratings and the other with
schizophrenia. At one time, it would have been anathema the converse. They also found evidence of distinct neu-
to make diagnoses of personality disorder and psychosis in ropsychological characteristics between the groups. In a
the same person, as operational definitions required exclu- review, Nestor (2002) went on to suggest four key dimen-
sion of psychosis before making a diagnosis of personal- sions of personality dysfunction in the context of people
ity disorder (Hare and Cox, 1978). Then, too, it has been with psychosis who had been seriously violent: impulse
shown in longitudinal studies that schizotypal personality control, affect regulation, narcissism, and paranoid cog-
disorder so commonly progresses to schizophrenia, but not nitive personality.
other illnesses, that those two conditions might even be
regarded as varying presentations of essentially the same Psychosis and the PCL-R
condition, whereas other personality disorders, such as
borderline personality disorder, do not progress in this way The PCL-R (Hare, 1991) has increasingly been used as a
(e.g. McGlashan, 1983). measure with offenders with schizophrenia, although the
meaning of doing so is not entirely clear. Taking the two
Moran and colleagues (2003) interviewed 670 of the factor model, which Hare has favoured, there is no reason
UK 700 psychosis cohort using the rapid version of the to suppose that the repeated antisocial behaviour factor
Personality Assessment Schedule (PAS-R) (Tyrer and would be confounded by diagnosis; it is difficult, however,
Cicchetti, 2000): 186 of them (28%) had a comorbid per- to know how the affective impairment factor relates to
sonality disorder, and nearly twice as many in this group the affective blunting and/or incongruity commonly seen
were violent (OR: 1.71, CI: 1.05–2.79). There was an inde- as a part of schizophrenia. Be that as it may, Tengström
pendent association with substance misuse (OR: 1.85, et al. (2004) set out to clarify inter-relationships between
346 © 2014 by Taylor & Francis Group, LLC
Psychosis, comorbid mental disorders and violence
such illness, PCL-R scares, substance misuse and offending ●● psychotic and other symptoms lead to self-medication;
among 202 male offenders with schizophrenia, with and ●● medication side-effects lead to self-medication with
without substance misuse disorders, and 78 men with a
primary personality problem. Among the men with schizo- illicit substances;
phrenia, those with high PCL-R scores (26 +) committed ●● medication non-compliance leads to substance use;
more offences that those with scores below the cut-off; ●● boredom and lack of structure lead to substance use;
men with high PCL-R scores but no schizophrenia commit- ●● exposure to adverse social environments leads to
ted more offences than those with high scores and schizo-
phrenia. A high PCL-R score was associated with similar substance use and aggressive behaviour;
rates of offending within disorder groups, regardless of the ●● substance abuse exacerbates psychotic and other
presence or absence of a substance misuse disorder. They
concluded that traits yielding high PCL-R scores accounted symptoms;
for offending better than any associated substance misuse ●● substance abuse reduces impulse control;
disorder, among men with psychosis and primary personal- ●● substance abuse is a proxy measure for personality
ity disorder alike.
disorder.
Psychosis and Substance Misuse Mueser et al. (1998) reviewed evidence for various of these
possibilities, suggesting that only the evidence for the self-
Substance misuse, particularly of alcohol and/or a range medication model was weak.
of more stimulant illicit drugs, is undoubtedly a major
contributor to violence by people without major mental More recent studies of the psychosis–substance misuse–
illness, so it would be very surprising indeed if it were violence links, with wide geographical spread but smaller
not a risk factor for violence among people with psycho- and more selected patient groups, present a mixed picture.
sis if they have access to such substances. Early stud- In the Czech Republic, for example, Vevera et al. (2005)
ies, however, were suggestive of an inverse relationship did not find an elevated rate of violence among those with
between substance misuse and violence among people psychosis and substance misuse, while in Switzerland
with schizophrenia, whether examining homicide (Häfner Modestin and Wuermle (2005) and in Turkey Erkiran et al.
and Böker, 1973) other crime (Virkkunen, 1974), alleged (2006) emphasized the increased probability of criminal
crime (Taylor, 1993a,b) or simply general psychiatric behaviour among men with schizophrenia, with or without
patients admitted to hospital (Tardiff and Sweillam, 1980). substance misuse. Larger community-based studies have
Almost certainly two factors were operating here, the shown that relationships between illness and substance
lower availability of substances, including alcohol, dur- misuse disorder may not be associated with all types of
ing these years and, during the earlier ones at least, the offending (Wallace et al., 2004), or may be affected by
protective effect of long periods of institutionalization. genetic or early environmental factors (Fazel et al., 2009a).
From the 1990s, there appeared to be increasing accept-
ance that if people with psychosis had substance misuse Why such discrepancies? There may be special meth-
problems, sometimes identified formally as a substance odological issues in identifying substance abuse, with
misuse disorder and sometimes not, then, as a group, offenders perhaps preferring to attribute their offending
they were at increased risk of violent behaviour. This to alcohol or drugs than to mental illness. Preliminary
seemed apparent, for example, in the USA-based ECA examination of attribution of offending among 113 jail
Survey, in which rates of violence among people with detainees in the USA, for example, suggested that neither
schizophrenia and substance misuse disorder were many substance misuse nor mental illness had a substantial
times higher than rates among those with schizophrenia effect on offending, but offences were more likely to have
alone (Swanson et al., 1990). It was also true, however, that been attributed to substance abuse than mental illness
people with schizophrenia appeared to be more vulner- ( Junginger et al., 2006). This hints at an even greater
able to using substances. Regier et al. (1990), for example, problem, that ‘substance abuse’ or ‘substance misuse’
showed that people with schizophrenia were about four may refer to categorization based on diagnostic manual
times as likely to misuse substances and people with criteria, to scores on dimensional screening tools, occa-
mania about six times as likely to do so as people without sionally to measures of substances or their metabolites
the disorders. Tiihonen and Swartz (2000) provide a good in blood or urine, or to unqualified reports of use. Most
framework for understanding both the likely higher rate of the more substantial studies use diagnostic criteria,
of substance misuse and its role in increasing the risk of so it is reasonable to think of comorbidity here, and this
violence among people with psychosis, suggesting that choice probably creates an underestimate of the risks
their hypotheses are not mutually exclusive: posed by substances – among people with psychosis
as well as those without. The other major issue is that
use of substances by people with and without psychosis
undoubtedly changes over time, with changing access,
mores and fashions in use. Over 25 years of admissions
(1975–1999) to the English high security hospitals, for
example, problem drinking in the year prior to the index
© 2014 by Taylor & Francis Group, LLC 347
Psychosis, violence and crime
offence or act was strongly associated with illicit drug need multi-faceted management and treatment plans in
use over the same period, and the proportion of patients order to approach a recovery trajectory.
who had been consuming to problem levels showed a
linear increase over time (McMahon et al., 2003). Problem Clinical characteristics
drinking was defined as taking in excess of 21 units of psychosis associated
of alcohol per week during that period. The increase with violence
affected the women more than the men and cut across all
diagnostic groups, however those with psychosis as the A number of psychotic symptoms have been implicated in
sole diagnosis were least affected. acts of violence, especially delusions. Deficits in capacity
for empathy and affective states, in particular anger, which
Implications for Practice of may accompany or even be induced by the psychosis, have
Findings About Comorbidity and/ also been implicated. Difficulties arise in making sense
or Distinctive Pathways into Violence of these findings, because, in a field where multiple and
by People with Psychosis complicated inter-relationships between a plethora of fac-
tors might be expected to influence interpersonal violence,
There is, then, growing evidence, from a variety of research research tends to focus on simple, single links. So, ques-
approaches, of at least two developmental pathways into tions are more likely to be about whether or not symptoms
offending by people with schizophrenia – one in which are associated with violence rather than the factors which
conduct disorder and antisocial behaviours are estab- might render a particular delusion or hallucination risky.
lished early, probably before the onset of the psychosis, Much of what follows in this chapter refers to psychotic
and the other in which the onset of the psychosis clearly symptoms in the context of a long-standing psychotic ill-
precedes violent offending, sometimes by many years, ness, but, as noted, there may well be a continuum between
and with the individual having appeared unremarkable symptom presentation and full blown illness. Furthermore,
until the onset of the illness. There may be a possible the border between normality and pathology is sometimes
further variant, with antisocial behaviour and psychosis hard to distinguish, for example between ‘over-valued ideas’
emerging more or less simultaneously. There is much and full delusions, and such ideas/delusions may persist
more to be understood about how experience of major in apparent isolation from a diagnosis of schizophrenia or
trauma and/or substance misuse fit into these pathways. similar illness. Studies in the field often do not make abso-
The implications of the likely range from having psy- lute distinctions and, indeed, it may not matter in practice
chosis alone to having a number of inter-related condi- except insofar as monodelusional states (e.g. see Behavioral
tions means that no single strategy for prevention and Sciences and the Law, issue 3, 2006) or disorders of passion
management of violence among people with psychosis is (see chapter 12) lack the bizarre qualities of some schizo-
likely to be effective (see also Volavka and Citrome, 2011). phrenic states and may be more easily missed.
Refinement of definition of subgroups to allow for pres-
ence or absence of significant developmental problems, Psychotic Symptoms and Violence
presence or absence of certain personality types or traits,
and presence or absence of substance misuse is likely Delusions and threat/control-override symptoms
to be necessary for effective clinical practice and useful Beliefs, by definition, require some sort of acceptance
new research alike. In addition, it now seems that it is that a proposition is true in the absence of evidence for it.
as important to be vigilant for the possibility of early or Most people have beliefs of some kind, whether religious,
current experiences of sexual or physical abuse. This may political or something else. This in itself is not patho-
need emphasizing, because Young et al. (2001) noted that logical, but some beliefs have characteristics which are.
asking about victim experience is not yet routine in clini- A pathological belief is referred to as a delusion. Kräupl
cal practice, and also that a diagnosis of schizophrenia is Taylor (1979) defined a ‘psychotic delusion’ as a belief based
one of the factors that renders such inquiry particularly on an absolute conviction of the truth of a proposition
unlikely. Here the emphasis has been on pathways into that is idiosyncratic, incorrigible, ego-involved and often
violent behaviour, and the likely treatment needs that preoccupying. Reference to ‘incorrigibility’ is consistent
these may raise. In going on to explore how specific clini- throughout the literature, from Jaspers (1923) to Oltmanns
cal characteristics may trigger violence, it is worth saying (1988), and reiterated in the DSM-IV (American Psychiatric
that even those people who, like McNaughton, seem to Association, 1994). In these terms, delusions are beliefs
have been unremarkable before psychosis devastated which are not just held without evidence, but rather in the
their lives and its symptoms drove them to some terrible face of contrary evidence, and are not amenable to change.
act, are likely to have complex needs. They, their families This conceptualization is helpful in many circumstances,
and their communities will be so changed by the combi- and perhaps particularly so when the beliefs are bizarre. It
nation of their illness and act that they too are likely to
348 © 2014 by Taylor & Francis Group, LLC
Clinical characteristics of psychosis associated with violence
may, however, provide insufficient guidance for clinicians equivalent of, respectively, passivity delusions, thought
when beliefs centre on more ordinary social situations. insertion and persecutory delusions, but researchers gener-
Mullen and Maack (1985), for example, noted that in cases ally treat them as if they are.
of morbid jealousy it is not uncommon for the partner who
was believed to be unfaithful to end up with the postulated The most commonly researched question about delu-
lover; people who have paranoid delusions about their sions or TCO symptoms and violence has been about
neighbours may indeed have neighbours who have become overall frequency of association. Although such studies
antagonistic to them or even reported them to the police consistently demonstrate an association, as symptoms
if they have been acting strangely or dangerously in this and violent acts are measured during long periods of time
context, but some truth in the neighbours’ hostility does which are not necessarily co-terminous, they can do little
not necessarily invalidate classification of the paranoid more than indicate to clinicians that this is an area of con-
belief as a delusion. In such cases, indicators of the way cern. A common choice is to test the relationship between
in which the belief was formulated, or the way in which any delusion and any violence reported as having occurred
it is maintained may be more helpful than evidence of its at any time over the 12 months prior to interview (e.g.
truth or falsity per se. To complicate matters further, it has Mojtabai, 2006; Swanson et al., 1996); some have adopted
long been recognized that the apparently consistent view a generally similar strategy, but with some more serious
of a delusion as ‘incorrigible’ cannot be defended either. violence included in analysis if it occurred up to 5 years
Beck’s (1952) description of change in a delusion suffered previously (e.g. Link and Steuve, 1994). This is where the
by a man with schizophrenia in response to psychotherapy point about incorrigibility or changeability becomes impor-
is generally taken as the first clinical step in recognizing tant. This sort of difficulty is perhaps best illustrated by
this, followed by others adding evidence to support such comparing publications on the matter from the MacArthur
a position (e.g. Sacks et al., 1974; Rudden et al., 1982) and risk assessment study. The oft quoted evidence against a
the development of scales for measuring delusions along relationship between delusions and violence comes from
one or more dimensions (e.g. Hole et al., 1979; Kendler the Appelbaum et al. (2000) report, which was, in fact,
et al., 1983; Shapiro, 1961; Strauss 1969; Brett-Jones et al., about the predictive value of delusions. In this early report,
1987; Taylor et al., 1994; Peters et al., 1999). Delusions may which included evaluations of mental state at baseline
fluctuate in another sense too – they are not necessarily (time-1) and re-examination at 10-weekly intervals for up
experienced consistently throughout the whole period to a year, the MacArthur group had not yet made the all
for which they are generally a problem. Myin-Germeys important intermediate test of whether delusions at time-1
et al. (2001) developed an ‘experience sampling method’ predicted delusions at time-2 (or subsequently). In a later
to find out how much of the time people with delusions study Appelbaum et al. (2004) confirmed that one-third of
were aware of them and/or preoccupied with them. On the subgroup of patients with delusions at any measured
average, people reported awareness of their delusions for interview no longer had them 10 weeks later, so there was
about a third of their time, this period being characterized indeed symptomatic change. Violent action was associated
by more negative affect. Withdrawal from social and other with persistence of delusions. Over the whole study period
‘distracting’ activities increased the likelihood of delusional there was a small but significant relationship between both
awareness. Bell et al. (2006) provide a useful review of the delusions and violence and hallucinations and violence
growing number of standardized rating schedules to facili- (Monahan et al., 2001).
tate reliable determination of a delusion.
Individual case reports, such as those referred to above
A concept which is closely related to that of the delu- have long implicated delusions as drivers to violence, and
sion, but possibly not fully co-terminous with it, is of some historical samples, unselected for violence, have also
threat/control-override (TCO) symptoms. Link and Stueve done so (Wilkins, 1993). In both criminological (Häfner
(1994) highlighted this cluster as a strong correlate of and Böker, 1973) and clinical samples (Rofman et al., 1980)
violence. It is constituted by three symptoms, as rated on there were further indications of this likelihood, but sys-
the Psychiatric Epidemiologic Research Interview (PERI; tematic documentation of a quantitative association prob-
Dohrenwend et al., 1980): ably started with our UK prisoner studies, in which men
●● a feeling that one’s mind is dominated by forces who had been under pretrial remand in prison, mostly for
4 weeks or less, were asked about their mental state dur-
beyond one’s control; ing interview and around the time of the alleged offence;
●● a feeling that thoughts are being put into one’s mind they were also asked, specifically, how they explained their
actions in relation to the index offence (Taylor, 1985b). It
that are not one’s own; was found that almost all of those with psychosis had been
●● a feeling that there are people who wish to do one symptomatic at the time of the offence, regardless of the
type of offence (Taylor, 1985), but specific inquiry about
harm. motive for the offence revealed that only about 40% of the
Use of the word ‘feeling’ has raised doubts for some men actually attributed the offending behaviour to any
clinicians about whether these symptoms are truly the
© 2014 by Taylor & Francis Group, LLC 349
Psychosis, violence and crime
psychotic symptom. Where a symptom was linked with relationship between delusions and violence. Skeem et al.
offending, it was almost invariably a delusion, although not (2006), for example, reported no evidence of association
necessarily recognized by the man as such. Further exami- between TCO symptoms and violence in an emergency
nation of the data revealed that there was a significant room cohort (n = 132) followed for 6 months, but the people
relationship between acting on delusions and the more seri- were selected on a criterion of ‘high risk’ and the sample
ous violence. Findings from a later, records based study of was diagnostically heterogeneous. Given that psychosis
homicide offenders in Finland fitted well with these British emerged as a protective factor against violence, it is likely
data in that over 90% of these Finnish men and women had that the case mix prevented adequate assessment of the
been symptomatic at the time of the offence, and, given question about TCO symptoms and violence.
the seriousness of the crime, a rather higher proportion
of offences had been directly attributed to the psychotic So, there is consistent evidence of a general association
symptoms (two-thirds) (Laajasalo and Häkkänen, 2006). between delusions and violence, and growing evidence
That said, Laajasalo and Häkkänen did not find that ‘exces- that when such acts are attributed to abnormal beliefs,
sive violence’, meaning more than was necessary to kill, was they are likely to be particularly serious. Delusions, how-
more likely among the delusionally driven. Swanson et al. ever, are common in psychosis and many who suffer them
(2006a), drawing on a sample of over 1,400 people from are never violent, so the next question must be about
across the USA, also reported that presence of psychotic whether it is possible to determine which are the more
symptoms was linked with more serious violence. Teasdale risky delusions and/or in which contexts they may be most
et al. (2006), using McArthur risk study data on men and likely to result in violence. Personality may be a mediating
women who were discharged general psychiatric patients in factor. When personality disorder is treated in terms only
the USA, found a sex difference in that there was an associa- of current trials, it appears to be almost invariably present
tion between delusions and violent acts for the men, but not (e.g. Blackburn et al., 2003), and thus not especially useful
the women, when experiencing threat delusions specifically. in discriminating likely relevance of delusions. Confining
attribution of personality disorder only to those with
With regard to TCO symptoms, a comparison of conduct or emotional disorders established in childhood
patients and never treated community controls in New does, however, discriminate. Those who are unremarkable
York showed that TCO symptoms best differentiated before onset of their illness are more likely to have been
people who had been violent and people who had not, driven to their offence by their delusions than those with
regardless of patient status or demographic character- comorbid personality disorder (Taylor et al., 1998). A com-
istics (Link and Stueve, 1994). So, with TCO symptoms, parison of such patients between Scotland on the one hand
as with delusions, presence of the symptoms rather than and England and Wales on the other confirmed that, while
having a diagnosis of psychosis and/or being in treatment the prevalence of premorbid personality disorders may
seems to be the critical factor. Although, in this study, vary between communities, this inter-relationship with the
symptoms and various forms of violence were measured impact of delusions is consistent (Taylor et al., 2008). Other
over different time periods, Swanson et al. (1996) subse- possibilities lie in the characteristics of the delusions and
quently tested the relationship between TCO symptoms in social interactions about them, the latter perhaps being
and violence using US ECA data with reference to the of particular importance as violence in the context of psy-
year prior to interview and to whole lifetime, confirming a chotic illness seems to be exceptionally likely to be directed
relationship over both periods. Link et al. (1998) replicated towards people in the family, or who are close socially
the finding of a relationship in an Israeli sample. Two (Johnston and Taylor, 2003; Nordström and Kullgren, 2003;
smaller studies in European countries have added further Nordström et al., 2006).
support for the role of this cluster. In Norway, Bjørkly and
Havik (2003) found that TCO symptoms were present Most of the standard mental state assessment sched-
close in time to a serious violent act; in Austria, Stompe ules which inquire about psychotic symptoms, including
et al. (2004) showed that TCO symptoms did not distin- those which are useful in measuring change, such as the
guish between offenders with psychosis and non-offender Comprehensive or Brief Psychiatric Rating Scales (CPRS:
patients with psychosis, but did distinguish the seriously Åsberg et al., 1978; BPRS: Overall and Gorham, 1962), allow
violent from other offenders. Hodgins et al. (2003) interro- rating of delusions by main content, such as persecutory,
gated longitudinal data on 128 men discharged from psy- passivity or religious. In these terms in the pretrial prisoner
chiatric hospitals in four countries. Comorbid personality study, the main distinguishing quality for the men driven to
disorder was more likely in the subsequently violent group act violently on their delusions was passivity (Taylor, 1999).
but, after controlling for personality disorder, ‘severe posi- Paranoid delusions were more common, but occurred at
tive symptoms’ were more likely among the men who were a similar rate in both groups. The TCO cluster may be so
violent on follow-up. powerful, when it occurs, because the passivity element
endorses violent actions on the belief in being threatened.
Rare, apparently dissenting studies may not be measur-
ing true delusions and/or may not be primarily testing the The Maudsley Assessment of Delusions Schedule
(MADS) (Taylor et al., 1994) was developed to describe
350 © 2014 by Taylor & Francis Group, LLC
Clinical characteristics of psychosis associated with violence
other characteristics of delusions, particularly factors attendant distress; thirdly, it may be that the social interac-
maintaining them, or their consequences. Thus, with the tion modifies the belief over time, along a pathway which
MADS, the patient-rated most important delusion is meas- finally makes action imperative.
ured along nine dimensions: belief maintenance factors
(e.g. seeking evidence for the belief, finding it, dealing with Perhaps the mechanism by which the belief has formed
hypothetical contradiction), level of conviction, of preoc- in the first place is critical here. There are parallels in
cupation, of systematization, of its idiosyncracy, affective the three main routes to belief formation that have been
impact of the belief, actions on it, including talking about hypothesized. One explanation lies in ‘logical’ explana-
it, withdrawal because of it and level of insight. In a series tions for perceptual abnormalities (Maher 1974, 1988).
of 83 general hospital patients, some sort of action on the Such abnormalities not only refer to hallucinations, but
belief designated by the patient as the most important to also impairments such as deafness (Thewisson et al., 2005)
him or her was common, and violent action in just one or hyperacute perceptions, such as enhanced or selective
28-day period had occurred in about one-quarter of the ability to recognize negative emotions in others (Davis
cases (Wessely et al., 1993). and Gibson, 2000). With respect to impairments, there are
even some data to show that healthy people may develop a
Qualities associated with violent action on a delusion paranoid state in the context of hypnotically induced deaf-
included first having acted on it in another, less threaten- ness (Zimbardo et al., 1981). The Bentall group has consist-
ing way – by seeking evidence for it, and, for some, having ently demonstrated selective attention to threat related
found evidence for it; secondly, being affectively distressed input (e.g. Bentall and Kaney, 1989; Bentall et al., 1995). A
by the belief – especially frightened by it; and, thirdly, the second possibility is that delusions constitute a form of
nature of response to hypothetical challenge to the belief personal defence. Psychoanalysts have long held a view that
(Buchanan et al., 1993). Hypothetical challenge means that paranoia emerges as a form of resolution of difficulties with
a proposition related to the belief is created by the inter- self-esteem (e.g. Colby, 1977), while others have emphasized
viewer who then asks the person with the belief whether, a so-called ‘self-serving bias’, which allows attribution of
if the proposition were true, it would affect the belief. any good outcomes to self and all negative outcomes to
An example would be if a person believed that she had a others (Campbell and Sedikides, 1999). There is even some
transmitter implanted in her head by the aliens who were evidence from CBT-based studies that improvement in self-
persecuting her, the interviewer might say that it might be esteem may be associated with improvement of psychotic
possible to do a very sensitive X-ray of the area, and ask a symptoms, albeit here delusions of grandeur and negative
skilled radiologist to read it. The interviewer would then symptoms (Jones et al., 2010). The third is a cognitive route
ask: ‘If this specialist could find no evidence of the device, (Magaro, 1980) by which people with certain cognitive
would that change the position? Would you feel reassured?’ styles process information differently from the ‘average
In the series of general psychiatric patients under study, person’, in particular through ‘jumping to conclusions’ (e.g.
Buchanan et al. found that no one abandoned his/her belief Huq et al., 1988; Abroms et al., 1996). The possible routes to
in these circumstances but, while some ignored such chal- delusion formation are not necessarily mutually exclusive.
lenge, others became even more convinced of the validity
of their belief, or developed it further in some way. People So, how might these various theories apply when vio-
in this latter group were more likely to have acted violently. lence is an issue? Carlin et al. (2005) found that offender
patients with psychosis were more likely to make exter-
The MADS incorporates the gentlest way of posing nal attributions for negative events than those without
potentially contradictory evidence to the responder’s self- psychotic symptoms, regardless of delusion type within
designated most important belief, but how, and under what the psychotic group. Hurn et al. (2002) showed how
range of circumstances do people routinely talk about their response to hypothetical challenge might differ accord-
beliefs and/or their delusions? What sort of response do ing to mechanism in belief formation. Those psychotic
they get? To what extent is the affective response to the patients who rejected hypothetical contradiction were
belief influenced by social context or interaction – perhaps more likely to have beliefs founded in hallucinatory expe-
a more direct or affectively laden challenge? To what extent riences, beliefs which they rated more strongly as ‘truth-
is the propensity to modify a delusion rather than ignore ful’; patients who accepted the hypothetical challenge
challenge or potentially contradictory evidence intrinsic? were more likely to report that their belief had affected
their behaviour and interfered with their thoughts and
The MADS findings, then, suggested three possible their lives. Freeman et al. (2007) explored ‘safety behav-
pathways to violence in the context of delusions: first, a iours’ in relation to paranoid delusions. Safety behaviours
primary delusional effect – the person acts on a delusion were first described in relation to anxiety (Salkovskis,
because the belief is sufficient in itself; secondly, action may 1991). They are most commonly avoidant, and have
follow as a result of the affective distress, apparently caused the risk of maintaining the pathology, because the per-
by the delusion, but perhaps also because the social climate son attributes their continued safety to their safety
between the person with the delusion and his/her signifi- behaviours rather than, perhaps, a false perception or
cant others is profoundly altered by the delusion and/or its
© 2014 by Taylor & Francis Group, LLC 351
Psychosis, violence and crime
interpretation of the dangers. It was this that prompted Interpersonal communications
Freeman and colleagues to ask about their relationship about delusions and violence
to delusions. All but four of 100 patients with paranoid
delusions reported carrying out at least one safety behav- The fact that McNaughton talked to other people about his
iour in the month prior to interview. Here too the most most important delusion, of being the object of systematic
common was avoidance, but 24 of the patients had been persecution, and that other people responded in various
aggressive. Greater use of safety behaviours was associ- ways, may well be typical of ordinary exchanges about such
ated with a higher level of distress about the delusions, beliefs, but there are few empirical data. The first study
and, in turn, with the greater likelihood of a history of using the MADS established that over 90% of people with at
violent acts against others and suicide attempts. least one delusion, interviewed within a week of admission
to a psychiatric unit, reported having spoken to someone in
One of Junginger’s (2006) main interests was in their social circle about that delusion in the previous 28 days
the constancy or variation of delusions over time, and (Taylor et al., 1994). If one enters a phrase like ‘talking about
another in whether, insofar as they do vary, their charac- delusions’ into Google, then pages of entries appear with
teristics are consistent in different phases of illness. His either relatives of people suffering from delusions asking
brief review of previous studies showed how little work how they should respond, or others – some relatives, some
had been done in this field, with somewhat contradictory psychologists or other professional clinicians – offering
results. In his small series (n = 54) of delusional patients, advice, but, in fact, this advice is not based on systematic
40 reported at least one episode of violence which had study. Langlands et al. (2008) consulted 45 ‘consumers’,
coincided with a delusion, 16 of them motivated by it. 60 carers and 52 clinicians from Australia, Canada, New
Eight patients had at least two episodes of delusionally Zealand, the UK and the USA about essential preliminaries –
motivated violence, separated in time, and on each occa- they called it ‘first aid’ – for helping people with psychosis.
sion the violence had been motivated by a delusion with They sought consensus through the Delphi method. ‘How
the same characteristics. Junginger proposed a simple to deal with delusions and hallucinations’ and ‘how to deal
method to aid prediction of delusion development, and with communication difficulties’ were among the nine cat-
thus, where this has been observed to underpin violence egories of needs which were endorsed by more than 80% of
in a particular case, to aid assessment or risk of vio- the participants.
lence. He suggests characterizing delusions by gaining
a detailed narrative of the experience and then applying A systematic review of talking about delusions (Fadhli
questions to each element of it: Who? What? Where? and Taylor, under submission), however, revealed that
When? How? Why? The following is how it might work if only the MADS study and one other (McCabe et al.,
applied to Daniel McNaughton, from descriptions by his 2002) have collected data on such communication in
father and others: a structured way, and neither have data on outcome of
routine exchange. The McCabe study recruited about half
Father: [Daniel] said he wished me … to put a stop to a of patients attending two outpatient clinics in London
persecution raised against him – it was some of the gen- and most of the psychiatrists treating them (39 in total)
tlemen connected with the conservative parties [who] in to participate in an observational study. They noticed
Glasgow [where]… I assured him there was no such perse- that patients actively tried to talk about their psychotic
cution … he shook his head, and said there certainly was a symptoms, but each such interaction lasted for about 67
system of persecution [what] existing against him … seconds, and occurred, on average, 1.4 times per inter-
view (range 0–4), with the psychiatrists commonly ‘re-
the interviews on the same subject were frequent for a year aligning’ the focus of the session away from the symptom,
and a half – he told me … he had left the city of Glasgow by responding with another question, or laughing, or, if
… that he went to England to avoid them, and even to a relative or carer were present, speaking to that person
France, to escape from the persecution; that he had no instead. The MADS study, as noted above, showed that a
sooner landed in France than he saw the spies following ‘set piece’ research exchange about the patient-designated
him there [the ‘where’ expanding and developing]. most important belief – the hypothetical challenge –
could result in a change in the belief, in the direction of its
Alexander Johnstone, MP: [McNaughton said] … that he intensification or elaboration.
could get no rest, night nor day [when], on account of
being watched … In theory, there is a more substantial literature on
Edward Thoms Monroe, MD: [McNaughton said] that the impact of communication about delusions in the
they wanted to murder him [why they were following more actively therapeutic context of cognitive behaviour
him] …he had on one or two occasions found something therapy for people with psychosis. It is arguable, however,
pernicious in his food [how] …the system destroying his that such communication differs considerably from the
health [how] (Old Bailey Proceedings, 2011, with analytic range of everyday talking about beliefs which may occur
comments inserted). as relatives or carers seek variously to reassure, or perhaps
352 © 2014 by Taylor & Francis Group, LLC
Clinical characteristics of psychosis associated with violence
argue with the sufferer, because the impact of arguments mask associations between hallucinations and violence in
from a loved one, whether warm and supportive or har- treated samples.
assed, distressed and irritable, would be likely to have a
different impact from interactions with someone who Rudrick (1999) and Barrowcliff and Haddock (2006)
would be regarded as an essentially neutral figure by the have reviewed the literature of various periods on com-
patient. Even here, however, there is not much informa- mand hallucinations. Barrowcliff and Haddock considered
tion. Jones et al. (2004; and see also below) conducted studies in three groups: those examining an association
a systematic review of CBT for schizophrenia, with the between command content and compliance in general
express criterion that any treatment included must have psychiatric and community samples; those doing the same
amongst its aims the ‘correction of misperceptions, irra- in offender patient samples; and the role of psychological
tional beliefs and reasoning biases.’ Just two of the 19 factors in understanding the relationship between compli-
trials identified met the analytical standards set by Jones ance and command. They identified 17 published studies of
and colleagues according to generally accepted criteria compliance with command hallucinations, of which three
and measured delusions over time. One of them showed a referred to harm to others (Rogers et al., 2002; Fox et al.,
slight advantage for CBT over ‘standard treatment’, which 2004; Lee et al., 2004). Factors associated with a greater
disappeared when allowance was made for missing rat- likelihood of action on hallucinations included beliefs about
ings on the assumption that these reflected serious con- the hallucinations, such as the likely consequences of not
tinuing illness (Drury et al., 2000) and the other described complying, ‘knowing’ the identity of the voice, attitude
no effect at all (Durham et al., 2003). These studies do to the voice(s), seriousness of the command, concurrent
not, however, make reference to aggression or violence, mood and more social factors like placement at the time
except in some cases as an exclusion from trial criterion, – being in the community rather than in hospital may
and, indeed, none makes reference to side effects of any increase likelihood of compliance with violent commands –
kind. An extended discussion of the extent to which these and self-perception of social rank (e.g. Fox et al., 2004).
studies leave us in an uncertain position on how talking Monahan et al. (2001) found a weak relationship between
about delusions affects them or affects the behaviour of hallucinations and violence over 1 year of follow-up of a
the sufferer is provided elsewhere (Taylor, 2006a). The mixed diagnosis sample of people discharged after a brief
few subsequent specifically therapeutic developments general psychiatric hospital admission.
are discussed below, with one now having considered
aggression. Other psychotic symptoms and violence
None of the other ‘positive symptoms’ of a psychotic illness
Hallucinations and violence has been shown to be of any great relevance to actions
Hallucinations may occur in any sensory modality. There is (Hafner and Boker, 1973; Taylor, 1985b). There have been
a small literature with data on gustatory and/or olfactory suggestions that negative symptoms are either neutral or
hallucinations in the context of delusions of being poisoned protective, but much may depend on context here, as such
(Mawson, 1985; Mowat, 1966), but, in the context of the difficulties often lead to ‘encouragement’ towards activities
functional psychoses, most concern has been expended on and in this context, negative symptoms have been associ-
auditory hallucinations, particularly command hallucina- ated with irritable violence (Nilssen et al., 1988). The only
tions. There is an expectation that if the commands are to other phenomenon worth considering is ‘motivelessness’,
do harm, then there is a risk that this will indeed follow. It which is not strictly a symptom but was at one time taken
may be partly for this reason that, taken together, research as almost pathognomonic of a psychotic killing. Generally, a
findings suggest that auditory hallucinations per se are not motiveless crime is taken to be one in which there has been
particularly associated with violence (e.g. Taylor, 1985b; very definite action on the part of the offender, but it has not
Rogers et al., 1988). They may, however, add to risk of action been possible to formulate any reason at all for it, at least up
on congruent delusions (Taylor et al., 1998). Further, the to the time of trial. It is a commonly expressed lay view that
usual absence of violence may be an artifact of staff con- a ‘senseless’ crime must imply mental abnormality, although
cern about the possible risk from command hallucinations psychiatrists should always try to demonstrate disorder
and their implementation of effective management strate- from evidence independent of the crime. Wilmanns (1940)
gies. One study of the period (Hellerstein et al., 1987) iden- described 18 killers who showed no clear evidence of psy-
tified 151 hallucinating patients among 789 consecutive chosis at the time of their crime, but went on to develop
hospital admissions and found no excess of violence in the schizophreniform psychoses in prison. Gillies (1965) noted,
hallucinating group. The people with hallucinations were, in a different series, that, although some went on to develop
however, significantly more likely to have been secluded or florid symptoms of psychosis, others merely remained
under 1:1 staff assignments. Thus, staff interventions may affectless and withdrawn, much as at the time of the
offence. It must remain in some doubt as to whether those
in either series who went on to develop a florid psychosis
© 2014 by Taylor & Francis Group, LLC 353
Psychosis, violence and crime
had really been free from illness at the time of the offence. It thus made logical sense for the MacArthur risk assessment
Lanzkron (1963) presented the view that no less than 27% of group to include anger measurement among their research
his series of homicide offenders had become psychotic after tools when studying a cohort of discharged psychiatric
their offence. It would be compatible with current views on patients, and it was not surprising that they found an asso-
the impact of major trauma that psychiatric illness, includ- ciation (Monahan et al., 2001). Their group was, however,
ing the psychoses, could be precipitated by such major life heterogeneous by diagnosis, so the next question is whether
events as a killing, trial and imprisonment. anger may be the critical link to violence among people with
psychosis. In a systematic review of the literature on this
Häfner and Boker (1973) found that motive was unre- more specific area of anger work, 13 studies were identified
corded for less than 20% of the German mentally abnormal and, in all but one of them, mean anger scores were elevated
homicide series, and that this proportion was not peculiar among those who had been violent compared with those
to those with psychosis. It varied little between diagnostic who had not (Reagu et al., 2013, in press). In order to make
groups. In our early pretrial prisoner study (Taylor, 1985b), use of such findings clinically, however, as Novaco (e.g. 1994)
there was a much lower proportion who could give no has repeatedly emphasized, it is important to understand
account of motives – 8% among the psychotic men and 7% the context of the anger – is it driven by the internal state, as
in the non-psychotic group, the difference probably being might be expected of paranoid states (e.g. Beck, 1999; Hareli
accounted for by the fact that the latter study was inter- and Weiner, 2002), or is it driven by tense exchanges and/or
view based and allowed for a good deal of probing. A small the potentially negative emotional climate that may surround
number of the men were motiveless in a different sense. No a person who is chronically ill with a psychotic illness? There
positive motive could be recorded as their ‘antisocial behav- is much more to be learned, but it is not too soon to be asking
iour’ almost amounted to a negative symptom of their ill- clinical questions.
ness, aimlessness possibly being a better word to describe
the reason for what had happened. In this case, the offence Environmental factors which
was not violent. may be relevant to violent
outcomes among people
Empathy, Psychosis and Violence with functional psychosis
Derntl et al. (2009) have suggested a general deficit of all Environment with respect to people who have psychosis
empathic abilities in people with schizophrenia compared is generally taken to mean social environment, and this
with the general population. In a systematic review of empa- is where we will focus, but it is worth noting that factors
thy and offending, in which they identified 35 relevant stud- beyond the personal may also be important. Schory et al.
ies, Joliffe and Farrington (2004) found a strong relationship (2003) compared records of humidity, wind speed and low
between low cognitive empathy and offending, particularly barometric pressure in one city in the USA (Louisville,
violent offending, and a weak relationship between affec- Kentucky) with documentation of psychiatric emergency
tive empathy and offending. The question then arises as to room visits, violence data from the city police department and
whether impaired empathy may be an important mediator suicide data from the county medical examiner. Diagnoses
of violence among people with schizophrenia or similar psy- are not recorded, but acts of violence and e mergency visits
chosis. In a systematic review of this (Bragado-Jimenez and rose during periods of low barometric pressure.
Taylor, 2012), six eligible studies were identified, but sample
selection, and empathy, illness and violence measures dif- The Family and Close Social Circle
fered between the studies, as did procedures of study. Data
were thus too heterogeneous for meta-analysis. It was more When violence is perpetrated by people with psychosis, it
likely than not for a relationship to be found in the better con- is so often within the family that it is arguable that research
trolled studies between perceptual or cognitive empathy and to investigate family interactions and needs in this context
violence, but the literature was not consistent. Responsive should be the highest priority. The National Confidential
empathy has not been studied in this context. Inquiry into Homicide for England and Wales confirms
that people with a history of mental illness, of contact with
Anger, Psychosis and Violence psychiatric services or with symptoms at the time of the
offence are less likely to kill strangers than those without
Anger is an emotion which affects people with and without (Shaw et al., 2004). In Sweden, Nordström et al. (2006)
psychosis alike, but there is evidence that it is more prevalent reported on all 48 homicide offenders between 1992 and
among people with psychosis than their healthy peers (e.g. 2000 who had schizophrenia and showed that those who
Freeman et al., 2001; Green et al., 2006). Anger is neither nec- had killed a family member were more likely to have been
essary nor sufficient to explain all violence, nor is it inherently delusional and less likely to have been intoxicated or had a
dysfunctional, but there is evidence that it has a significant
association with violent acts (Anderson and Bushman, 2002).
354 © 2014 by Taylor & Francis Group, LLC
Environmental factors which may be relevant to violent outcomes among people with functional psychosis
criminal career. Most were known to psychiatric services, and Vaughn, 1985). The term ‘expressed emotion’ (EE)
although few of them were still in contact, and only two was used to indicate qualities in emotional climate, with
were taking medication. In an earlier study (Nordström high EE reflecting tension, emotional over-involvement,
and Kullgren, 2003), they had investigated all 207 men of 18 critical remarks and even outright hostility on the part
years of age and over who had committed at least one vio- of the family member(s). Initially the main thrust of the
lent crime and presented for a first specialist forensic psy- research was in showing that not only was a person who
chiatric evaluation between 1992 and 2000, with the main had schizophrenia more likely to relapse when family
research question: do violent offenders with schizophrenia members had high EE in relation to him/her but also
who attack family members differ from those with other that the high EE was the provocation (Brown et al., 1972).
victims? As in our English high security hospital study Subsequent studies have endorsed the position that high
(Johnston and Taylor, 2003), injuries to family victims were EE in this context carries at least twice the risk of relapse
more likely to have been serious or fatal than injuries to for the person with schizophrenia, and that about two-
people outside the family. Beyond that, the Swedish results thirds of people returning to a high context after treat-
suggested that there had been a substantially greater fam- ment would be expected to relapse (Butzlaff and Hooley,
ily burden from the illness of those who had attacked fam- 1998; Stanhope and Solomon, 2007). A more multidirec-
ily. This was echoed in a study in Austria (Stompe et al., tional model of understanding the emergence of high
2006) which found that family burden with schizophrenia expressed emotion and its consequences is now better
was twice as high among those with schizophrenia who recognized (e.g. Kavanagh, 1992), based on the stress and
had also offended as among those who had not. This was, coping model of Lazarus and Folkman (1984). The finding
however, complicated by the parents of some of the men that not only may high EE be apparent in certain fam-
also having had schizophrenia. In general, family burden ily relationships but also in particular professional care
from psychotic illness has received a little attention, and staff–patient relationships (Moore and Kuipers, 1999)
much of that focused solely on family members other provides further evidence, if needed, that it is more useful
than the person with psychosis. Cousins et al. (2002) have to explore the interpersonal dynamic than to attribute
described a way of assessing this in people living with attitudinal difficulties. The burden of care on the families
and/or caring for people with Parkinson’s disease, which of people with schizophrenia may be substantial, and
could be applied to help families of people with psychosis, probably the more so since the ever growing shift towards
but in addition, Greenberg (Greenberg et al., 1994; Chen treating in the community any but the most acutely ill
and Greenberg, 2004) points out, we must also take into (Awad and Voruganti, 2008). Factors which increase the
account positive contributions to families by people with burden include the perception on the part of the relatives
psychosis in order to have an accurate view of household of the severity of symptoms and their ability to cope with
and family dynamics. them (e.g. Barrowclough and Parle, 1997), the mood of the
person with psychosis (Boye et al., 2001) and perceived
Earlier studies are also consistent on the vulnerability hostility (Estroff et al., 1994, 1998; Swanson et al., 1997).
of relatives and household members, perhaps especially Relatives who are less well informed about the illness, and
mothers (Estroff and Zimmer, 1994; Estroff et al., 1994, tend to attribute blame for its disruptive qualities to the
1998; Steadman et al., 1998; Tardiff et al., 1997; Vaddadi person with the illness rather than the illness itself appear
et al., 1997). Violence is not contingent on proximity per to be at higher risk for developing high expressed emotion
se, so what possible explanations might there be? The (Chan, 2010; Kavanagh, 1992). It is less clear, however,
sequencing of symptoms and any conflict is unclear, but how schizophrenia, family burden, high EE and violence
three, possibly inter-related aspects of close relationships relate – there has been too little enquiry. In a preliminary
may be relevant: the roles in key relationships, emotional study, part of an ongoing, longitudinal study of people
climate in them and the way in which they may influence with schizophrenia and their relatives and professional
symptom development. Roles within the family tend to be carers, we encountered both patient and relative centred
altered by chronic illness. Parents, particularly mothers, difficulties in recruiting relative participants. Among
often find themselves taking over various aspects of life those relatives who did participate, half had high EE.
management, tasks which in other circumstances would Qualitative analysis of 5-minute speech samples, of unin-
not be regarded as appropriate for the age of their ‘child’. terrupted talking about the relationship with the patient
This has been found to have an association with violent according to five written prompts (Magana et al., 1986),
behaviour between the parties concerned (Estroff et al., revealed that the core concern was about the identity of
1994), fuelled in particular by the financial dependence the person with schizophrenia (Rowntree et al., 2011).
of the person with psychosis, and a mutually hostile and This polarized relatives with high EE who regarded the
threatening relationship between parent and adult child. person as a patient and the delusions as distressing and
salient, and those with low EE who viewed the individual
Emotional climate in relationships between people as ‘the person s/he had always been’. Where violence had
with schizophrenia and their families has been an area
of interest since the 1960s (e.g. Brown et al., 1962; Leff
© 2014 by Taylor & Francis Group, LLC 355
Psychosis, violence and crime
occurred, it was, perhaps contrary to expectations, in the disagree, such as: ‘You think that going for help probably
latter group. wouldn’t do any good.’ Adherence was a binary variable,
derived in response to: ‘In the past six months, were there
Residential proximity creates opportunities for rela- times when you thought you should go to a doctor or
tives to comment on delusions to the person who has clinic… but did not go?’ Answers to four questions defined
them, or discuss them, but this has not been systematically perception of treatment effectiveness: ‘As a direct result of
studied (Fadhli and Taylor, under submission). We have the services I received: (a) I deal more effectively with my
already mentioned research showing that responsiveness daily problems; (b) I am better able to control my life; (c) I
to hypothetical challenge to a delusion may be associated am getting along better with my family; (d) my symptoms
with violence, but we can find no systematically collected are not bothering me very much.’ Perceived treatment need,
information on how families deal with the symptoms with positive perceptions of treatment and treatment adherence
which their psychotic member suffers. There is, however, were all associated with reduced odds of violence over the
plenty of evidence in publicly accessible search engines 6 months of outpatient care. Diagnostic differences did not
that family members are sufficiently concerned about such affect perceived treatment need, but people with psychosis
matters to seek advice on how to respond to reported delu- were more likely to perceive their treatment as effective and
sions and that there are people offering them apparently report treatment adherence than people in other diagnos-
untested advice. Locating the pathways to and expression tic groups.
of violence within the families of people with psychosis
complicates research and challenges simpler causal formu- Day et al. (2005) completed a similar study in Wales
lae. As we argued earlier with regard to ‘unpacking’ diagno- and north-west England of 228 people who had been hos-
sis into actual symptoms, it is necessary, in the same way, pitalized with schizophrenia or schizo-affective disorder.
to increase the granularity with which we consider family Attitudes towards treatment and treatment adherence
processes, traditions, affective and cognitive traditions to were predicted by insight, attitudes towards staff and
illuminate the links between violence and psychosis. the treatment experience. In particular, people with poor
insight, with a poor relationship with the prescribing psy-
The Clinical Environment chiatrist and who had experienced any coercion in their
treatment were more likely to have a negative attitude to
Some difficulties in the nature of dialogue between patients that treatment. Research conducted by and in collabora-
and their clinicians have already been noted (McCabe et al., tion with investigators with first hand experience of hospi-
2002; Estroff 2004). Such difficulties in communication talization provides a range of findings about the experience
may contribute directly to violence. Omerov et al. (2004) and outcomes of psychiatric inpatient treatment (Russo
compared staff and patient experiences of the same violent and Wallcraft, 2011). The first few hospitalizations may
incidents in a hospital in Sweden, where staff were able represent an exquisitely vulnerable time, during which
to identify fewer than 50% of the provocations that the enduring experiences of trauma or of authentic therapeutic
patients reported experiencing. One of the more consist- alliance are formed. It is particularly important that people
ent discrepancies in perception arose over medication. who might benefit from mental healthcare do not avoid it
Secker et al. (2004) picked up this theme in a qualitative later due to the lingering effects of prior humiliation, vio-
study undertaken in a south London hospital, in England. lence, or coercion in treatment settings.
They considered the most striking theme to be lack of staff
engagement with the patients, and a particular inability Zygmunt et al. (2002) reported a systematic review of
to look at the world through the eyes of the patients (see the literature on strategies to improve treatment adher-
also Gilburt et al., 2008). In a partner study, Benson et al. ence. Studies of psycho-educational interventions with
(2003) found patients reluctant to talk with researchers, patients or their families were popular, but typically ineffec-
but striking congruence between one patient participator tive. Problem-solving or motivational techniques directed
who had been violent and the staff in the discourse about specifically at treatment adherence appear most likely to
two incidents, in particular in attributing blame. Bowers be successful in improving this. We would also suggest that
et al. (2006) tested the relationship between specific nurse more attention may be needed in the support and supervi-
training in prevention and management of violence in the sion of staff. Moore and Kuipers (1992) and Moore et al.
inpatient setting, with rather discouraging results. (1992) did much of the early work on expressed emotion
in staff relationships to patients, finding that, generally,
Elbogen et al. (2006), using the MacArthur study multi- where critical style was found, it was not typical of the staff
site cohort of over 1,000 patients discharged into the concerned, but rather seemed responsive. Since then, there
community after a brief hospitalization, studied patient have been about 27 studies in this field, yielding mixed
perceptions of their treatment need, adherence and effec- results both on prevalence of the difficulty and on its conse-
tiveness, and compared these self-ratings with community quences (Berry et al., 2011). Some studies had small samples
violence. Perceived treatment need was couched in nega- and did not attempt prevalence reporting, for example one
tive statements, to which patients were asked to agree or study of 20 staff and 20 patients did not find high expressed
356 © 2014 by Taylor & Francis Group, LLC
Management and treatment
emotion, but did note critical comments – mainly about mood disorders attending public health services in four
enduring features which were seen as being ‘within the states of the USA and, indeed, found that apart from a past
patient’s control’. One important factor in the differences history of violent victimization, homelessness and violence
may lie in length and intensity of relationships – there has in the surrounding community were the factors associated
to be close contact over at least 3–6 months before it would with higher rates of patient violence. The lack of accessible
be expected that any such difficulties might emerge. Moore or affordable residential options in safe settings for people
and colleagues have specifically studied staff–patient pairs with psychosis may contribute to perpetuating their use
in secure hospital settings (Moore et al., 2002), where 55/61 and need of violence that may be context appropriate, but
staff who participated had high expressed emotion in the nonetheless detrimental clinically and socially. If absent
specified relationship, which correlated exclusively with from research consideration, these contextual influences
irritability, argumentativeness and violence on the part of on violence may be inaccurately attributed to symptoms
the patients. The rate was much higher than for staff work- or substances.
ing with long-term patients in general services, where a
quarter of staff showed high expressed emotion, but here Management and treatment
the patient problems were ‘difficult’ and socially embarrass-
ing behaviours’, a finding endorsed in other studies. More Practically useful, high-quality research information about
information is needed about the impact of such difficul- treatments specifically for people with psychosis who have
ties, but it seems likely that when clinical staff recognize been violent is difficult to obtain. The gold standard for
their emergence, they may use the recognition as a cue to treatment trials is always considered to be the randomized
distance themselves from the patient, and this may even controlled trial (RCT), notwithstanding its undoubted
be a factor in rejection from services; where not, then the limitations in tending to focus on average, consenting
main effect may be on quality and extent of patient benefit and co-operating men. People with psychosis who need
from treatment. It is arguable that enhanced recognition forensic mental health services are rarely ‘average’, often
of developing expressed emotion could be used to clinical having atypical presentations, combinations of disorders
advantage. or medication resistant psychosis, they often go through
periods when they are reluctant to take treatment, some-
The Wider Community times frankly refusing, and, although the majority are men,
10–20% are women, for whom the evidence base on medi-
The main factors in the wider community which are likely cation is much less good (see also chapter 20). Some who
to affect violence by people with psychosis are almost have engaged in the most serious or repetitive violence may
certainly similar to those which affect people without even be regarded as unique, and, perhaps for these few,
psychosis, or indeed mental disorder of any kind, and these treatment trials might best be created as individualized
are availability of alcohol and other drugs. We have already trials, with on–off–on design around treatment of their
noted that there seems to have been an increase in the use specific problems.
of substances by people with psychosis over time in England
and Wales at least, as licensing laws have relaxed, street Cure et al. (2005), nevertheless, provided a tantalizing
drugs have become more readily available, and ‘care in the report of the range of randomized trials that they consid-
community’ has left many people with psychosis struggling ered could be relevant to the provision of forensic mental
to maintain symptom control and/or with little to distract health services. They searched 29 electronic databases of
or occupy them. Experiences which are more problematic literature for the period 1955–2000, none of them, they
for people with psychosis than their healthy peers are observed, standing out as a definitive source of ‘foren-
dealing with stigma and the higher rates of community sic studies’. Of approximately 22,000 studies identified,
victimization they experience (see opening paragraphs). 409 were described as relevant to highly aggressive or
They are also probably more vulnerable to a range of other aggressive people with psychosis, and were made available
significantly stressful life events which may have an impact through the Cochrane Central Register of Controlled Trials.
on their propensity for violence (e.g. Silver and Teasdale, The authors offered further encouragement in describing
2005). Becoming homeless is among these. Such problems the range of treatments evaluated, noting that a higher pro-
are in part related to the tendency for people with psychosis portion of the studies than in other mental health fields had
and other major mental disorders to find themselves reset- been pursued for at least 6 months and suggested that addi-
tled in problem communities (Silver, 2000; Silver et al., 2002) tional studies were still likely to be discovered or underway
and on social support for their finances, sometimes explic- at the time of their searches. Nevertheless, they hardly
itly contingent on complying consistently with treatment; stimulated much further interest in the work, by describing
some are without supports of any kind. Swanson et al. the quality of most of the studies as ‘poor’ according to the
(2002) studied factors associated with the 1-year preva- Jadad scale (Jadad et al., 1996).
lence of violence among 802 adults with psychosis or major
The principles of treatment are considered in chap-
ter 23, emphasizing the therapeutic context and the
© 2014 by Taylor & Francis Group, LLC 357
Psychosis, violence and crime
importance of establishing sound treatment engagement for schizophrenia, the Patient Outcomes Research Team
in order to sustain specific treatments. As far as possi- (PORT) also provides regular updates for evidence based
ble, it is important first, to be working within an explicit, treatment recommendations (e.g. Dixon et al., 2009).
evidence-based multidisciplinary and interagency frame-
work; secondly, to be ensuring that people are stabilized Medication
on medication which has maximum impact on all the
features of psychosis – not just the positive symptoms – Medication for people who are acutely psychotic with a risk
while inflicting the minimum unwanted effects; thirdly, to of being violent is dealt with in chapter 23. There is insuffi-
be making full use of relevant psychological interventions; cient space for comprehensive coverage of the evidence on
and, fourthly, to be employing all relevant educational and the range of new (second generation antipsychotic/SGA)
occupational techniques to enhance the prospect of gain- and older ( first generation antipsychotic/FGA) medica-
ing healthy independence or, in some terminology, moving tions, singly and in combination, for treating schizophrenia
along the recovery pathway. or similar psychoses per se, but, as medication has long
been recognized as such a fundamental component of
For people with psychosis, it will be important to make the treatment of schizophrenia and similar psychoses, at
some specific efforts to improve or maintain physical least since the study of May (1968), it may be useful here
health in the face of the combination of an illness which to refer to one or two of the more recent reviews of this.
often imposes anergia on the sufferer and the medications First, it is useful to have a working knowledge of the range
which themselves have a tendency to promote weight gain of antipsychotic medications available, and here again, to
and, in some cases, metabolic diseases such as diabetes. avoid repetition, we refer the reader to chapter 23 for a brief
The specialist clinical models and medico-legal frame- description of the subgroups within these main classes and
works for enhancing continued treatment adherence with of the main individual drugs.
a health service lead, such as provisions within mental
health legislation, are mainly covered in chapter 24, and Kane et al. (2003) conducted a consensus survey of
those with a criminal justice system lead, such as com- expert opinion on medication for psychosis, which covered
munity or suspended prison sentences with a mental not only medication preferences but also aspects of pre-
health treatment requirement, are dealt with in chapter scribing practice. There was a high return rate (over 90%) on
25. Only some of the issues more specific to psychosis will a very detailed questionnaire derived from literature review,
be considered here, with the same caveat, as in the more which brought consensus on nearly 90% of the options
general treatment chapter (23), that this is a field that is offered for rating. Atypical antipsychotics were preferred.
under more-or-less continuous update. We recommend Three to six weeks on a particular drug was considered
here, and there, that, for both physical and psycho-social an adequate trial, but most would continue for four to ten
treatments, the reader keep checking sources such as the weeks if there was at least a partial response. When switch-
Cochrane Library of systematic literature reviews (www. ing between oral antipsychotics, cross-titration was con-
thecochranelibrary.com), which deals exclusively with ran- sidered to be the best strategy, while for a switch to depot
domized controlled trials (RCTs), and the Database of medication, stress was placed on maintaining oral antipsy-
Abstracts of Reviews of Effects (DARE) (www.crd.york. chotic medication until therapeutic levels of the depot were
ac.uk/cms2web/), which goes beyond randomized con- attained. Emphasis was placed on the importance of moni-
trolled trials in its scope. The reader should also be aware toring physical health, particularly metabolic status and
that, from time to time, previously well-regarded and useful sexual health. Antipsychotic medications tend to impair
reviews, particularly Cochrane reviews, are removed from sexual function, most noticeably among men, and may be a
the site before new reviews are available, so it is worth reason for discontinuing such medication, with or without
saving reviews in areas of interest. In addition, in England, the knowledge of the doctor (Baldwin and Mayers, 2003).
standard guidance in treatment, as far as possible based
on such reviews, is available on the National Institute of Notwithstanding the apparent confidence expressed by
Health and Clinical Excellence website: www.nice.org. clinicians in the Kane study, evidence for an advantage of
uk. In fact, the increments of change are generally quite any particular antipsychotic medication in any particular
small and the principles of working remain solid. While circumstance is actually quite slight. Such misplaced con-
specific treatment guidance ought to be very similar across fidence may reflect over-reliance on open studies which,
national boundaries, there may be some differences with Leucht and colleagues (2009) suggest, systematically favour
respect to medication, as there may be more or less cau- second generation drugs. Lepping and colleagues (2011)
tion about licensing certain products in different countries. showed, in a systematic review of 300 data sets, that find-
Thus, where national guidelines are available these will ings of change in drug trials are generally measured in
be most useful. In the USA, for example, the American terms of increments on scales which are of limited clinical
Psychiatric Association provides Practice Guidelines and relevance. They are at pains to point out that this tends to
occasional Task Force Reports (www.psych.org). Specifically be about differences between drugs, and that they are not
arguing that, overall, drugs have negligible effects in clinical
358 © 2014 by Taylor & Francis Group, LLC
The words you are searching are inside this book. To get more targeted content, please make full-text search by clicking here.
Forensic Psychiatry_ Clinical, Legal and Ethical Issues Book 1
Discover the best professional documents and content resources in AnyFlip Document Base.
Search
Forensic Psychiatry_ Clinical, Legal and Ethical Issues Book 1
- 1 - 50
- 51 - 100
- 101 - 150
- 151 - 200
- 201 - 250
- 251 - 300
- 301 - 350
- 351 - 400
- 401 - 450
- 451 - 500
- 501 - 514
Pages: