Some caveats to this study are noteworthy. First, data on cannabis use disorders are not available in control subjects. Second, we elected to use data collected with the DIGS 4.0 as this provided the most comprehensive assessment and also excluded any heterogeneity attributable to assessment – thus, individuals assessed using prior versions of the DIGS were excluded. This did not influence the association between BD case status and cannabis use. Retaining only those subjects with a Bipolar I diagnosis also did not modify the association. Finally, the retrospective nature of the data precludes tests of causality.
