Prevalence and the Need of Attention
The main finding of this study is the high prevalence of gambling disorder in violent offenders, comprehensively assessed with DSM-IV-based interviews, confirming the results from previous research (Williams et al.
2005). This prevalence is about 40 times higher than in the general Swedish population. The large epidemiological study SWELOGS (Swedish Longitudinal Gambling Study) has shown that general gambling in Sweden appears to have reached a plateau (Romild et al.
2014), but the proportion of problem gamblers in the population remains the same and gamblers with the most serious problems are even increasing.
The pronounced relationship between criminality and gambling disorder could form an incentive to pay more attention to gambling problems in the correctional system. Gambling has been reported as a usual, perhaps even normative, part of prison life (McEvoy and Spirgen
2012) which likely impairs the chances of rehabilitation, and gambling severity is a significant predictor of criminal recidivism according to a study by April and colleagues (Lahn
2005; April and Weinstock
2017). The causal relations between gambling disorder and criminality are not fully understood (Turner et al.
2013; May-Chahal et al.
2016), but there is an obvious risk for criminal relapse when a person is in gambling debt, and this two-way connection motivates increased awareness and availability of gambling disorder treatment to people with a criminal lifestyle.
Concerning types of crimes, there were no significant differences between the gambling disorder and the non-gambling disorder group. However, this study—including only violent offenders—does confirm the picture that gambling disorder might be common among offenders regardless of type of crime, and not mainly connected with economic crime or property crime, which has also been suggested (Turner et al.
2009,
2013; Cuadrado and Lieberman
2012).
The frequency of each DSM gambling criterion in the gambling disorder group in this study is interesting and shows both similarities and differences compared to previous research. About 51% of the gambling disordered group reported having committed a criminal act to finance their gambling, which is similar to reports from Abbott and colleagues, Lahn and colleagues and the Williams’ review (Abbott et al.
2005; Lahn
2005; Williams et al.
2005). However, when compared to other gambling disordered groups, the offenders with gambling disorder in this study to a markedly less extent reported having jeopardized or lost a significant relationship, job etc. (23.3%) or relied financially on others because of gambling (14.0%) (Granero et al.
2014; Christensen et al.
2015). A possible explanation could be that this group of men is limited in their ability to perceive and assess consequences of their actions. Previous research has shown impaired risk assessment in offenders, which probably is connected to the high prevalence of antisocial personality disorder (Pachur et al.
2010; May-Chahal et al.
2016). Perhaps their relatively young age and complicated life situations could affect what kind of relationships they are in, and their appreciation of them. Pachur and colleagues suggest that enhanced thinking skills programs for offenders, aimed at reducing recidivism through changing attitudes, could be more successful with an increased focus on risk taking (Pachur et al.
2010). Many criminal problem gamblers don’t see their gambling as problematic at all (Lahn
2005), which further emphasizes the need of cognitive behavioural interventions.
Early Problems and Detection
We have also showed that, even if this group of convicted men has had a substantial amount of early onset problems with e.g. schooling and demonstrates a high prevalence of conduct disorder, the participants with gambling disorder stand out with significantly worse results concerning elementary and middle school graduation. This is interesting in the perspective of the development of early deviant behavior, and previous publications on the same population show parallel results; school adjustment problems (e.g. truancy, bullying and incomplete schooling) were the most distinct significant predictors of later aggressive behavior (measured as total score on Life History of Aggression) (Wallinius et al.
2016). In line with this, an early criminal career has been associated with higher loss chasing in gambling according to a study by May-Chahal et al. (
2016).
Swedish epidemiological research has shown that gambling problems are highly overrepresented among young, marginalized men (Abbott et al.
2014) and it is probable that gambling problems and antisocial behavior develop simultaneously (Slutske et al.
2001) and perhaps catalyze each other. A part of the longitudinal study on the Dunedin birth cohort concludes that gambling disorder in young adults has much in common with addictive disorders and other externalizing behaviors, from a personality perspective (Slutske et al.
2005). Thus, young men with impulsive and delinquent behavior are clearly in the risk zone for gambling disorder and should be targets for preventive actions and treatment.
Psychiatric Comorbidity
When it comes to the extremely high prevalence of psychiatric disorders in this cohort, it speaks for the need of competent psychiatric care for young offenders (Al-Rousan et al.
2017). The criminal gamblers may be predisposed to gambling problems and form an interesting group from a biological point of view. They could represent a certain “antisocial pathway” to gambling disorder; characterized by low impulse control, high aggression levels and multiple drug use (Blaszczynski and Nower
2002; Valleur et al.
2016; Allami et al.
2017). Gupta and colleagues suggest that the “pathways model” is applicable also for adolescents, confirmed by latent class analysis which showed a distinct impulsive and antisocial subgroup among the young problem gamblers (Gupta et al.
2013). The antisocial gamblers often start at a young age (Valleur et al.
2016; Allami et al.
2017) and gambling may play a role in the evolvement of an antisocial lifestyle (Slutske et al.
2001). According to various studies looking at gambling and personality, problem gambling could be considered part of a cluster of externalizing pathology. The personality profiles of pathological gamblers and substance abusers are often dominated by negative affect and unconscientious—and disagreeable disinhibition, a combination of traits also closely connected to antisociality and impulsivity (Maclaren et al.
2011). It is believed that it is of great importance to pay more attention to the associations between both gambling disorder and substance use disorders, and criminal maintenance (McEvoy and Spirgen
2012). There is an evident connection between criminality and gambling, and even though the chronology is complex, the research on this cohort enlightens the need of efforts early in life to affect the development of both social and psychiatric problems (Nilsson et al.
2016). Further observation of this group of gamblers, through screening and the offering of treatment, would be necessary to evaluate the possible benefit of treatment interventions.
In the present study, we found significantly higher occurrence of antisocial personality disorder and three substance abuse diagnoses in the gambling disorder group; cannabis, anabolic steroids and cocaine. The variables that were independently associated with gambling disorder in the regression model were elementary and middle school graduation and cocaine abuse. Possible causal relations behind these findings are not possible to determine through this study, but our results enlighten the need of further studies.
The comorbidity between gambling disorder and substance use disorders has previously been demonstrated, and more serious alcohol problems have been shown to correlate with more severe gambling disorder and higher general dysfunction (Lorains et al.
2011; Del Pino-Gutierrez et al.
2016). Pietrzak and colleagues also demonstrated illicit drug use and severity of gambling problems to be indicators of antisocial personality disorder in treatment seeking gamblers (Pietrzak and Petry
2005). The presence of substance abuse diagnoses was very high in this study, and enlightens the need of the offering of treatment to possibly decrease relapses in gambling disorder, and other addictive disorders, and criminality. There were no significant differences between groups considering alcohol abuse, but the prevalence is considerably high in the whole cohort; 48.5%, compared to approximately 4% in the general population (Andréasson
2011). The significant overrepresentation of cocaine abuse in the gambling disorder group, still significant in the regression analysis, is interesting and might speak for a pronounced reward seeking behavior, together with problematic decision-making. Dufour and colleagues found an overrepresentation of problem gamblers among community cocaine users (Dufour et al.
2016), and alterations in neural structure in specific parts of the orbitofrontal cortex—playing an important role for drive and responsibility—have been connected to both cocaine-use disorder and gambling disorder (Adinoff et al.
2003). The field is complex; so far the studies are small and somewhat ambiguous, but there are interesting findings indicating the need of further analyses of addictive behaviors in this context (Adinoff et al.
2003; Yip et al.
2017). Our results show a connection between cocaine and gambling disorder in the studied group, even though a lot still remains unclear when it comes to the explaining pathobiology.
Limitations
The study is limited by the relatively small size of the cohort, which may have lead to undetected type II errors. The exclusion of 23 participants due to language problems may have affected the results, since men born outside Sweden seem to be at higher risk for gambling problems (Abbott et al.
2014). It cannot be precluded that the group of 109 men who declined participation in the study was different in any way considering prevalence of gambling disorder and other variables investigated. Analysis of the non-responders was limited to age and type of crime.
The cohort itself was defined by a number of factors that may limit the generalizability of the results because of selection biases. The age span was narrow (18–25 years), no females were included and all the participants were convicted of violent (including hands-on sexual) crimes. In addiction medicine research, maybe particularly concerning gambling disorder, it is important to underline that there are cross-cultural differences both within—and between countries. The current sample is a quite specific group in the Swedish population, which may of course limit the generalizability also to other similar populations (Medeiros et al.
2015; Raylu and Oei
2004).
Many parts of our data were selected retrospectively, hence a recurrent risk for recall bias. All diagnoses were based on professional assessments, and as always there are elements of subjectivity. The substance abuse diagnoses were comprehensively based on DSM-IV, but substance dependence criteria were not fully assessed. The gambling disorder diagnoses were based on a structured list of DSM-IV criteria, but no structured validated instrument was used, which may have affected the diagnostic validity of gambling disorder.
When performing multiple comparisons there is always a risk for type 1 errors. We handled this risk by using the BH method in which all
p values from the bivariate analyses were adjusted to limit the total false discovery rate to 5%, prior to the regression analysis (Hochberg and Benjamini
1990).
A main drawback with the cross-sectional design is the lack of time perspective, which also rules out the possibility to estimate risks and possible causal relationships.