“People with mental illness are no more likely to commit violent crimes than ordinary members of the public,” reported The Independent . “Substance abuse is the chief cause of violent crime and increases the risk equally in people with and without mental illness,” it added.
The story is based on research that looked at the risk of people with bipolar disorder committing violent crimes such as assault and robbery, compared with the general population. It found that, although bipolar disorder was associated with a higher risk of violent crime, the increase in risk was largely due to drug and alcohol abuse.
This large, well-designed study found that the increased risk of violent crime in people with bipolar disorder was largely associated with substance abuse and not with the disorder per se . There was no significantly increased risk of violent crime in individuals with bipolar disorder who had no history of substance abuse when compared with the rest of the population. These findings may help to contradict assumptions that associate bipolar disorder with violence. They should also be considered during the risk assessment and treatment of individuals with bipolar disorder who misuse alcohol and illegal drugs.
The study was carried out by researchers from Karolinska Institutet, Stockholm, and Oxford University. The study was published in the peer-reviewed medical journal Archives of General Psychiatry . It was funded by the Swedish Medical Research Council and the Swedish Council for Working Life and Social Research.
Media coverage of the study was generally fair and responsible, emphasising that people with bipolar disorder were no more likely to commit violent crimes than other members of the population, unless they also abused drugs or alcohol. The Financial Times ’s headline, “Call to improve psychiatric help” emphasised the need for improved provision of specialised drug and alcohol services for those with mental illness. The Independent ’s reporting that the study was of the “mentally ill” was misleading, since the study only looked at bipolar disorder.
This was a population-based, longitudinal cohort study that compared the risk of violent crime in people with bipolar disorder with the risk in the general population and also with siblings unaffected by the disorder. The researchers also carried out a systematic review and meta-analysis which included previous research in this area.
The researchers point out that various adverse health outcomes have been associated with bipolar disorder, including suicide, homelessness and repeat offending. But the evidence for any association between bipolar disorder and violent crime is less clear. Their aim, they say, was to quantify any possible risk of violent crime associated with bipolar disorder, and to adjust for other factors such as social class and income, early environment and genetics, and to examine the effect of substance abuse.
The researchers compared the rate of violent crime in 3,743 individuals diagnosed with bipolar disorder who were cared for in Swedish hospitals between 1973 and 2004 with that of 37,429 individuals in the general population. They also compared rates of violent crime in people with bipolar disorder with their unaffected siblings.
To identify these groups the researchers used national population-based registries in Sweden: the Hospital Discharge Registry (HDR), the National Crime Register, the national census from 1970 and 1990 and the Multi-Generation Register.
To be included in the study, patients had to have been discharged from hospital with a diagnosis of bipolar disorder according to internationally accepted definitions, on at least two separate occasions between 1973 and 2004, and had to be at least 15 years old at the start of the study. The researchers also extracted data for each of these patients about diagnoses of alcohol and drug abuse or dependence.
Researchers also identified two comparison groups of individuals who had never been hospitalised with bipolar disorder during the study period. The first was a random sample of approximately 10 individuals in the general population who were matched on birth year and sex for each individual with bipolar disorder. The second was made up of 4,059 siblings of a subgroup of 2,570 individuals with bipolar disorder. Both comparison groups included people who may have had a history of substance abuse.
Researchers also retrieved data on all convictions for violent crime from 1973 to 2004 for all individuals aged 15 (the age of criminal responsibility in Sweden) and older. Definition of violent crime included homicide, assault, robbery and rape.
They also took into account sociodemographic factors such as income, marital and immigrant status.
Using validated statistical methods the researchers used this information to identify any association between violent crime and bipolar disorder, compared with the two control groups. Only violent crime after the second diagnosis of bipolar disorder was included.
They also carried out a systematic review and meta-analysis, with searches for studies in this area between 1970 and 2009.
The researchers found that:
The researchers’ systematic review identified eight previous studies in this area. A meta-analysis that included their own study found that the odds ratios for the risk of violent crime in individuals with bipolar disorder, ranged from 2 to 9.
The researchers point out that while there is an increased risk of violent crime among individuals with bipolar disorder, most of the excess risk is associated with a history of substance abuse.
They also say that the increased risk of violent crime shown among the siblings of those with bipolar disorder weakens the relationship between a diagnosis of bipolar disorder and violent crime, and highlights the importance of genetic and early environmental factors.
Substance misuse is high in individuals with bipolar disorder, so substance abuse treatment in this group is likely to reduce the risk of violent crime.
This large well-conducted study has several strengths. Its size increases its statistical power and makes its conclusions more reliable. Its results are adjusted for possible confounders such as income. It also only included violent crime after diagnosis, which reduces the risk that hospital admission may have been triggered by a criminal conviction. The comparison population group was well matched for birth year and sex.
The authors note some limitations in its methods, which could mean the possibility that some individuals with bipolar disorder were missed and the effects of substance abuse may have been underestimated.
The study’s conclusion that bipolar disorder per se is not associated with violent crime is important, as is the strong association between bipolar disorder, substance misuse and violent crime. The findings suggest that there should be risk assessment for violent crime in patients with both bipolar and substance misuse and strengthens the case for improved treatment services for these people.