Please use this link to access this publication.
Abstract
Background: Cannabis use is associated with outcomes like income, legal problems, and psychopathology. This finding rests largely on correlational research designs, which rely at best on statistical controls for confounding. Here, we control for unmeasured confounders using a longitudinal study of twins.
Method: In a sample of 4,078 American adult twins first assessed decades ago, we used cotwin control mixed effects models to evaluate the effect of lifetime average frequency of cannabis consumption measured on substance use, psychiatric, and psychosocial outcomes.
Results: On average, participants had a lifetime cannabis frequency of about one to two times per month, across adolescence and adulthood. As expected, in individual-level analyses, cannabis use was significantly associated with almost all outcomes in the expected directions. However, when comparing each twin to their cotwin, which inherently controls for shared genes and environments, we observed within-pair differences consistent with possible causality in three of the 22 assessed outcomes: cannabis use disorder symptoms (βW-Pooled = .15, SE = .02, p = 1.7 × 10−22), frequency of tobacco use (βW-Pooled = .06, SE = .01, p = 1.2 × 10−5), and illicit drug involvement (βW-Pooled = .06, SE = .02, p = 1.2 × 10−4). Covariate specification curve analyses indicated that within-pair effects on tobacco and illicit drug use, but not cannabis use disorder, attenuated substantially when covarying for lifetime alcohol and tobacco use.
Conclusions: The cotwin control results suggest that more frequent cannabis use causes small increases in cannabis use disorder symptoms, approximately 1.3 symptoms when going from a once-a-year use to daily use. For other outcomes, our results are more consistent with familial confounding, at least in this community population of twins.
