Search

The high association of substance use and mental health has been extensively researched, however there remains conflicting evidence in the temporal direction of this relationship. This thesis aims to investigate this association using a range of different methods to examine the direction of association between substance use and mental health problems in adolescence, and whether these are likely to be causal. I also examined the possible role of social cognition in this relationship due to its common associations with both substance use and mental health problems. First, systematic review is used to identify any patterns in the current mental health and substance use literature. Here, I find the evidence is largely mixed and there is a general lack of bidirectional studies and null results reported. Secondly, I conducted a series of longitudinal studies in the Avon Longitudinal Study of Parents and Children (ALSPAC) comparing trends of substance use, mental health, and social cognition in both temporal directions. My results suggested two possible pathways (a) substance use impairs social cognitive performance via poor mental health or (b) substance use independently impairs both social cognitive performance and mental health. Finally, to investigate the causality of these associations I conducted a Mendelian randomisation analyses in the most robust observational results (tobacco initiation, externalising behaviour, and social communication). Here, I found some evidence of an association that genetic risk of tobacco initiation is causally associated with externalising disorders, but no evidence of a causal association of genetic risk of tobacco initiation with social cognition. The evidence here suggests some evidence of a causal association of tobacco initiation with externalising behavior. However, the observed associations of tobacco on social cognition may be due to environmental or confounding factors. This thesis further highlights the importance of using range of difference methodological and statistical techniques each with differing underlying assumptions when investigating causal inferences.