Authors' reply

We are confident that the relatively low response rates to lithium in our study relate to the narrow definition of lithium response, rather than to characteristics of the included patients. ¹ Thus, we intended to characterise patients who had an excellent response to lithium monotherapy; that is, patients who were ‘cured’ from further affective episodes following a start-up period of lithium as in a prior study. ² We used two robust clinical indicators to define excellent lithium response: (a) lithium prescribed in monotherapy; and (b) no need for psychiatric hospital admission. By doing this, we defined lithium response in a rather rigorous way, resulting in relatively low rates of response. We do not find that our definition of lithium response hampered the finding of the study that early treatment with lithium was associated with increased probability of excellent lithium response compared with delayed treatment, or hampered the generalisability of this finding. Although cycle acceleration occurs on average in bipolar disorder ^3,4 the results of our study may suggest that early treatment with lithium might prevent progression of bipolar disorder.