Dening (See C 28851 fo) thinks that the increase in the risk of crashes observed in our study may be due to the disease (bipolar disorder) itself and not lithium, hence confounding by indication. We acknowledged this possibility in our paper. Although our data did not allow identification of subjects with bipolar disorder, another approach to control for confounding by indication is to study a different drug for the same condition, which in this case was carbamazepine. Although an optimal comparative drug would have been valproic acid, the limited number of users of valproic acid in our study did not permit this comparison. The lack of increased risk associated with carbamazepine supports our conclusions of an increased risk of crashes with lithium use, especially given that carbamazepine, valproic acid, and lithium are all considered mainstay therapy for bipolar disorder in older adults. This was especially true in the early 1990s, the time span of our study, as newer pharmacological agents were not yet available. We can only presume that our paper passed statistical review because the reviewers recognised that confounding by indication can be addressed with a proper comparison drug with similar indication. Given the plausible biological mechanism for a potential association of a motor vehicle crash and lithium use (delayed reaction time) and the results of our study, we believe that elderly drivers taking lithium should be informed of this potential risk. Further studies of the effect of lithium and other psychotropic drugs on the risk of crashes would be valuable. (Author/publisher)
Abstract