A latent class analysis of changes in adolescent substance use between 1988 and 2011 in Sweden: associations with sex and psychosomatic problems.

Abstract

To characterize changes in patterns of adolescent substance use in Sweden between 1988 and 2011, and to assess whether sex and psychosomatic problems were associated with substance use and whether these associations changed over time. Secondary analysis of repeated cross-sectional survey data. Survey data were collected eight times and analyzed as four cohorts (1988-91, 1995-98, 2002-05 and 2008-11). The sample included all 15-16-year-olds in Värmland County, Sweden (n=20 057). Binary-coded substance use measures included life-time use of alcohol and tobacco, getting drunk and past school year use of inhalants. An eight-item scale was used to assess psychosomatic problems. A three-class model fitted the data best (i.e. non/low use, mainly alcohol use and polysubstance use). The patterns of substance use were different among cohorts; most notably, adolescents in the last cohort had lower odds of being included in the alcohol and polysubstance use classes rather than the non/low use class than in the earlier cohorts (all Ps<0.001). Males had higher odds than females of being in the polysubstance use class rather than the non/low use class among the first three cohorts (all Ps<0.001) but not the last. Sex was not associated with inclusion in the alcohol use class rather than the non/low use class. Adolescents who reported more psychosomatic problems had higher odds of being included in the alcohol and polysubstance use classes rather than the non/low use class (all Ps<0.001). The associations of sex and psychosomatic problems with class inclusion did not change during the study period. Between 1988 and 2011, patterns of substance use among adolescents in Sweden shifted away from polysubstance use and alcohol use to non-use or low use. Associations between patterns of substance use and sex and psychosomatic problems remained largely consistent across the study period.