Early Cumulative Risk in a Nordic Welfare State Context: Longitudinal Effects on Adolescent Problem Behaviors

The most vulnerable youth are those who experience multiple risks during early development. The cumulative risk (CR) hypothesis suggests that the likelihood of experiencing problem outcomes increases with the sheer number of risk factors. This hypothesis has received considerable support, but gaps in knowledge remain. The majority of research is based on samples from the United States or the United Kingdom. It is not evident that established CR indicators have the same impact on individuals in other country contexts, such as Nordic welfare states. Also, there is a need for longitudinal studies to understand the general and specific patterns of prediction from CR in early development to multiple outcomes in adolescence, a period of risk for problem behaviors, and to test potential gender moderation and non-linear effects. We address these gaps by examining CR in relation to adolescent substance misuse, externalizing behaviors, and internalizing symptoms in a birth cohort of over 7,000 Finnish children followed over 16 years from the prenatal period to adolescence (about 80% retention). We conducted secondary analyses of data from the 1986 Northern Finland Birth Cohort Study. CR indicators (e.g., teen pregnancy, parent unemployment, paternal substance misuse) were measured via questionnaires administered to mothers during pregnancy; dichotomous indicators were summed into a CR count index. Adolescent outcomes were measured via the Youth Self Report and standard substance use (alcohol, cigarettes, and marijuana/other illicit drugs) items. Primary analyses were conducted using multivariate structural equation modeling in Mplus 7.11 in two steps. First, the CR index, gender, and a CR*gender product term were examined as predictors of the latent outcome variables. Model fit was acceptable (e.g., RMSEA=.045; CFI=.953). None of the gender interaction effects were significant. Thus, a second model was estimated in which the gender product term was replaced by a CR*CR product term. Model fit was acceptable (e.g., RMSEA=.043; CFI=.958). There was a significant (p<.05) interaction effect on internalizing symptoms (b=-.04, se=.02, β=-.10), indicating a leveling off of prediction as the number of risks increased. Moreover, CR was a significant (p<.05) positive predictor of both substance use (b=.62,  se=.61; β=.30) and externalizing problems (b=.22, se=.06; β=.16). Results support the CR hypothesis in the Finnish context and extend prior research by showing that CR effects may be non-linear for internalizing symptoms. This suggests that the burden of multiple risks may have a ceiling effect for certain outcomes. Findings have implications for preventive interventions targeted at youth in greatest need to bolster protection in the face of risk.