Post-traumatic stress disorder and the risk of violent crime conviction in Sweden: a nationwide, register-based cohort study

Background Post-traumatic stress disorder (PTSD) has been linked to violent crime in veteran populations. However, whether there is a link between PTSD and violent crime in the general population is not known. This study aimed to investigate the hypothesised association between PTSD and violent crime in the Swedish general population and to investigate the extent to which familial factors might explain this association using unaffected sibling control individuals. Methods This nationwide, register-based cohort study assessed individuals born in Sweden in 1958-93 for eligibility for inclusion. Individuals who died or emigrated before their 15th birthday, were adopted, were twins, or whose biological parents could not be identified were excluded. Participants were identified and included from the National Patient Register (1973-2013), the Multi-Generation Register (1932-2013), the Total Population Register (1947-2013), and the National Crime Register (1973-2013). Participants with PTSD were matched ( 1:10) with randomly selected control individuals from the population without PTSD by birth year, sex, and county of residence in the year of PTSD diagnosis for the matched individual. Each participant was followed up from the date of matching (ie, the index person's first PTSD diagnosis) until violent crime conviction or until being censored at emigration, death, or Dec 31, 2013, whichever occurred first. Stratified Cox regressions were used to estimate the hazard ratio of time to violent crime conviction ascertained from national registers in individuals with PTSD compared with control individuals. To account for familial confounding, sibling analyses were conducted, comparing the risk of violent crime in a subsample of individuals with PTSD with their unaffected full biological siblings. Findings Of 3 890 765 eligible individuals, 13 119 had a PTSD diagnosis (9856 [75 center dot 1%] of whom were female and 3263 [24 center dot 9%] of whom were male), were matched with 131 190 individuals who did not, and were included in the matched cohort. 9114 individuals with PTSD and 14 613 full biological siblings without PTSD were also included in the sibling cohort. In the sibling cohort, 6956 (76 center dot 3%) of 9114 participants were female and 2158 (23 center dot 7%) were male. Cumulative incidence of violent crime convictions after 5 years was 5 center dot 0% (95% CI 4 center dot 6-5 center dot 5) in individuals diagnosed with PTSD versus 0 center dot 7% (0 center dot 6-0 center dot 7) in individuals without PTSD. At the end of follow-up (median follow-up time 4 center dot 2 years, IQR 2 center dot 0-7 center dot 6), cumulative incidence was 13 center dot 5% (11 center dot 3-16 center dot 6) versus 2 center dot 3% (1 center dot 9-2 center dot 6). Individuals with PTSD had a significantly higher risk of violent crime than the matched control population in the fully-adjusted model (hazard ratio [HR] 6 center dot 4, 95% CI 5 center dot 7-7 center dot 2). In the sibling cohort, the risk of violent crime was also significantly higher in the siblings with PTSD (3 center dot 2, 2 center dot 6-4 center dot 0). Interpretation PTSD was associated with increased risk of violent crime conviction, even after controlling for familial effects shared by siblings and in the absence of SUD or a history of violent crime. Although our results might not be generalisable to less severe or undetected PTSD, our study could inform interventions that aim to reduce violent crime in this vulnerable population.