We performed an additional post hoc analysis correcting for recruitment site, as the French sample had a slightly higher number of rapid-cycling subjects than the Norwegian sample, and the Norwegian sample had a slightly lower age at onset than the French sample. Reassuringly, after we corrected for recruitment site, sexual abuse was still significantly associated with rapid cycling (P = .006), and sexual abuse (P = .002) and emotional abuse (P = .002) were significantly associated with earlier age at onset (data not shown). Last, we performed a multivariate analysis controlling for alcohol and cannabis use, and our results remained the same.