Finally, we know that there is considerable overlap between CD and alcohol dependence, and that twin studies suggest shared genetic liability across these disorders, as described in the ‘Introduction’ section.12 The extensive comorbidity between the disorders, both in our sample (83% of the CD cases also met criteria for alcohol dependence) and in the general population, makes it unrealistic to completely tease apart genetic effects on the two disorders. However, we did want to determine whether our results for CD were driven largely by association with alcohol dependence. The correlation between the natural log transformations of CD symptom count and alcohol dependence symptom count was 0.45 (P < 0.001) in our sample. We reran the top SNPs from Table 2 including the log-transformed alcohol dependence symptom count as a covariate. Results are listed in Table 4. As expected, the P-values dropped in magnitude, though all were still significant.
