Risk of Subjection to Violence and Perpetration of Violence in Persons With Psychiatric Disorders in Sweden

Key Points Question What is the incidence of subjection to violence or perpetration of violence in persons with psychiatric disorders? Findings In this nationwide cohort study of 250 419 individuals with psychiatric disorders in Sweden, in the decade after the onset of their conditions, fewer than 7% of patients had either been subjected to violence severe enough to require specialist medical treatment or had perpetrated violence. Meaning Persons with psychiatric disorders were approximately 3 to 4 times more likely than their siblings without psychiatric disorders to be either subjected to violence or to perpetrate violence.


Population Sample
Our original sample consisted of all children born in Sweden 1973-1993 who could be linked to both of their biological parents (n=2,176,150). We chose these years as they captured information on all psychiatric disorders and covariates (as the National Patient and Crime registers were available from 1973) and also on outcomes (as the legal age of responsibility in Sweden is 15 years and we wanted the youngest cohorts to have sufficient time to have outcomes). We excluded those who had emigrated (n=73,717) or died (n=17,420) before age 15 as well as those who lacked data on parental socioeconomic measures (n=5149), thus resulting in a final population size of 2,079,864 individuals (95.6% of the targeted population). All of the patients and controls were selected from this sample.

Definitions of Measured Confounders
Immigrant status was defined as having at least one biological parent who was born in a non-Nordic country. Parental history of psychiatric morbidity and violent criminality indicated whether either biological parent had been diagnosed for any psychiatric disorder or had a violent criminal conviction.
Low family income indicated whether the individual's disposable family income (e.g., standardized net sum of earnings and benefits averaged across both biological parents) measured at the end of the year that they turned 15 years of age ranked in the bottom decile of the population. If this information was missing, we used data from the previous year or until it became available. We adopted this nonlinear definition because earlier studies have shown that rates of violent convictions are heavily concentrated in the most deprived groups in Sweden. 1,2 Similar findings have also been observed, albeit less pronounced, for violent victimization in Finland. 3,4 Low parental educational attainment indicated that neither biological parent had achieved secondary school qualifications.

Validation of Diagnoses
The National Patient Register (NPR) offers a near full coverage (>99%) of all somatic (including surgery) and psychiatric inpatient care discharges since 1973 and specialist outpatient visits since 2001. 6 A 2011 review of the validity of all inpatient care diagnoses reported in the NPR found that the positive predictive values typically ranged between 85-95%. 6 We are unaware of any validation studies that have examined victimization diagnoses specifically but there is some evidence suggesting that the severity levels of the victimization injuries are valid. 7, 8 We also note that the violent victimization diagnoses have been used in large-scale epidemiological studies conducted not only in Sweden 9 but also in comparable registers in Finland 4 and Denmark. 10

Complementary Sensitivity Analyses
We carried out sensitivity analyses using alternative measurement definitions and model specifications to test for the stability of the fully adjusted co-sibling estimates (Model IV). First, we used a stricter definition of psychiatric disorders by requiring at least two separately occurring episodes and also for violent victimization severity by only considering inpatient care cases of victimization and homicidal deaths. We further tested for the associations by only considering the "core" psychiatric disorders (e.g., anxiety, depression, bipolar disorder and schizophrenia). Second, we tested for alternative matching criteria, either by only including unaffected same-sexed siblings or unaffected siblings of both sexes born within four years of the patients. Third, to test for potential misclassification bias, we excluded individuals who had been diagnosed with having been exposed to an unarmed brawl or fight (ICD-10 code: Y04). Lastly, we tested for violent crime arrests (derived from the National Criminal Suspects Register) instead of violent crime convictions as outcome.

Moderation Effects
We tested for moderation effects by specifying a number of additional statistical models that included interaction terms for the following research questions. First, we asked whether the associations © 2020 Sariaslan A et al. JAMA Psychiatry between being diagnosed with any psychiatric disorder and the violence outcomes were moderated by the individual's history of violent victimization and perpetration. Second, we asked whether the association between being diagnosed with any of the "core" psychiatric disorders (e.g., anxiety, depression, bipolar disorder and schizophrenia) and the violence outcomes were moderated by comorbid personality disorder, alcohol use disorder and/or drug use disorder. There was little evidence of any of the examined moderation effects as none of the interaction terms were statistically significant (all p>0.05 following Bonferroni corrections for multiple testing