Categories
Nevin Manimala Statistics

Latent classes of substance use and delinquency in a Swedish national sample of adolescents and associated risk factors

PLoS One. 2025 May 2;20(5):e0322515. doi: 10.1371/journal.pone.0322515. eCollection 2025.

ABSTRACT

BACKGROUND: Identifying underlying subgroups might be a way to examine the development of co-occurrent substance use and delinquency. The aim of this study was to identify latent classes of substance misuse and delinquency in adolescence and which general risk factors are associated with these classes.

METHODS: Data of two waves from a national representative Swedish birth cohort was used that consisted of 4,013 randomly selected adolescents (Male = 1,798, Female = 2,201, Missing = 14) Latent class analysis was used to identify classes of substance misuse and delinquency at age 17/18 and. logistic regression analysis was used to assess risk factors at age 15/16.

RESULTS: Identified classes were: “Low/abstainers” (74.80%, n = 2858, Male = 1191 Female = 1656, Missing = 11) which acted as reference, “Alcohol only” (22.21%, n = 849, Male = 420, Female = 426, Missing = 3), “Polydrug use and crime” (2.15%, n = 82, Male = 52, Female = 30) and “High crime” (0.84%, n = 32, Male = 30, Females = 2). Factors associated with belonging to any classes engaging in substance use and delinquency were lower parental support, supervision, peer problems, and higher conduct problems, sensation-seeking behavior, distrust in society, and truancy.

CONCLUSIONS: Most people did not engage in substance use or delinquency. When accounting for less frequent behaviors such as normative adolescent drinking and one-time events of crime and drug use, about 3% of the population engaged in co-occurring substance use and delinquency. Several different factors from several domains where related to belonging to a class that used substances and/or engaged in delinquency. There were indications that the most extensive users and offender displayed a wide variety of severe level risk factors, which could have implications for targeted interventions. Though, statistical power was a problem and future research should use larger samples or alternative methods.

PMID:40315216 | DOI:10.1371/journal.pone.0322515

By Nevin Manimala

Portfolio Website for Nevin Manimala