Testing the association between tobacco and cannabis use and cognitive functioning: findings from an observational and Mendelian randomization study

Background Although studies have examined the association between tobacco and cannabis use in adolescence with subsequent cognitive functioning, study designs are usually not able to distinguish correlation from causation. Methods Separate patterns of tobacco and cannabis use were derived using longitudinal latent class analysis based on measures assessed on five occasions from age 13–18 in a large UK population cohort (Avon Longitudinal Study of Parents and Children). Cognitive functioning measures comprised of working memory, response inhibition, and emotion recognition assessed at 24 years of age. Mendelian randomization was used to examine the possible causal relationship between smoking initiation, lifetime cannabis use and cognitive functioning. Results We found evidence of a relationship between tobacco and cannabis use and diminished cognitive functioning for each of the outcomes in the observational analyses. There was evidence to suggest that late-onset regular tobacco smokers (b=-0.29, 95 %CI=-0.45 to -0.13), early-onset regular tobacco smokers (b=-0.45, 95 %CI=-0.84 to -0.05), and early-onset regular cannabis users (b=-0.62, 95 %CI=-0.93 to -0.31) showed poorer working memory. Early-onset regular tobacco smokers (b = 0.18, 95 %CI = 0.07 to 0.28), and early-onset regular cannabis users (b = 0.30, 95 %CI = 0.08 to 0.52) displayed poorer ability to inhibit responses. Late-onset regular (b=-0.02, 95 %CI=-0.03 to - 0.00), and early-onset regular tobacco smokers (b=-0.04, 95 %CI=-0.08 to -0.01) showed poorer ability to recognise emotions. Mendelian randomization analyses were imprecise and did not provide additional support for the observational results. Conclusion There was some evidence to suggest that adolescent tobacco and cannabis use were associated with deficits in working memory, response inhibition and emotion recognition. Better powered genetic studies are required to determine whether these associations are causal