Fig. 1.

Diagrams of the primary analyses. These diagrams show the pathways being tested in the analysis only. There is evidence from the literature that the smoking and body mass index (BMI) relationship is bi-directional, however we have not shown this in the figure. Figure A shows the observational analysis conducted, which examined (1) The relationship between BMI at age 10.5 years and patterns of smoking between the ages of 13 and 18 years and (2) the relationship between body dissatisfaction (BD) at age 10.5 years and patterns of smoking between the ages of 13 and 18 years. Figure B demonstrates the Mendelian randomization analysis (3) which used a genetic risk score as a proxy for BMI (BMI-G) to examine the relationship with patterns of smoking between the ages of 13 and 18 years and with body dissatisfaction.