Abstract
IMPORTANCE: Depression and factors reflecting neighborhood social structure (ie, socioeconomic deprivation, ethnic heterogeneity, residential mobility, and urbanicity) have each been linked to criminal convictions. However, how the association between depression and crime varies across different neighborhood types, and the extent to which it reflects unmeasured familial confounding, remains unclear. OBJECTIVE: To examine whether the association between depression and violent and nonviolent criminal convictions varies across neighborhood types, and to assess the extent to which unmeasured familial factors contribute to the association. DESIGN, SETTING, AND PARTICIPANTS: This population-based matched cohort and sibling-comparison study used data from Swedish national registers from 1986 to 2020. Follow-up spanned from 2001 to 2020. Statistical analyses were performed from January to November 2025. The cohort included individuals with a diagnosis of depression, each matched to 5 population controls without depression by birth year, sex, and neighborhood type. EXPOSURE: Outpatient depression diagnosis (International Statistical Classification of Diseases and Related Health Problems, Tenth Revision codes F32-F33.9) recorded from 2001 to 2020. MAIN OUTCOMES AND MEASURES: The primary outcomes were violent and nonviolent criminal convictions after diagnosis, identified through the National Crime Register. Conditional logistic regression estimated odds ratios (ORs) across 4 neighborhood types (resource-limited, rural low-diversity, urban professional, and urban affluent neighborhoods), with sibling comparisons used to assess familial confounding. RESULTS: Among 571 470 matched individuals, 95 245 (36 297 male [38.1%]; median [IQR] age at first diagnosis, 20 [17-24] years) had depression. Depression was associated with increased odds of both violent and nonviolent convictions across all neighborhood types in unadjusted models. After adjustment for prior convictions, substance use disorder, and attention-deficit/hyperactivity disorder, associations remained significant in all but resource-limited neighborhoods (violent conviction OR, 1.14 [95% CI, 0.97-1.33]; nonviolent conviction OR, 1.01 [95% CI, 0.92-1.11]). A second sample included 42 585 individuals with depression and their full siblings without depression (total, 85 170 individuals). Sibling comparisons showed partial attenuation, indicating that familial confounding accounted for some, but not all, of the associations. Sibling-matched estimates were largely consistent with fully adjusted general population-matched estimates (eg, violent convictions in rural low-diversity neighborhoods: sibling-matched OR, 1.50 [95% CI, 1.33-1.69] vs general population-matched OR, 1.51 [95% CI, 1.39-1.65]). CONCLUSIONS AND RELEVANCE: In this cohort study of the Swedish general population, the association between depression and criminal convictions varied across neighborhood types and was partially explained by familial factors. These findings underscore the relevance of considering contextual and familial influences and may offer insights for prevention and intervention strategies responsive to neighborhood social environments.