[nashke]

We can't show causation using this kind of methodology. Longitudinal association may suggest causation, but it does not imply it. In the discussion we offer some possible mechanisms for this association, not all of them causal in nature. We do know from previous studies that the endocannabinoid system is altered in suicidal individuals and that sex hormones both affect suicidality and have a reciprocal relationship with the endocannabinoid system. These suggest how and why intense cannabis use can affect suicidal behavior.

I've posted a link which should allow for full-text access for the next few weeks.