We congratulate Duffy et al on their paper. ^{Reference Duffy, Horrocks, Doucette, Keown-Stoneman, McCloskey and Grof1} We have long argued that bipolar disorder is often underdiagnosed by community mental health teams, and that the reason for this is often failure to assess the longitudinal trajectory of patients with recurrent depression. ^{Reference Rogers and Agius2,Reference Rogers, Agius and Zaman3} We have attempted to remedy this by developing 29 questions to be used in the history-taking of all patients with depression and recurrent depression to demonstrate the developmental trajectory of the illness. ^{Reference Agius and Murphy4} These questions are presently being field tested in Bedford, UK, and at the University of Perugia, Italy. We have also demonstrated that when the systematic assessment of the trajectory of bipolar disorder is carried out in a community mental health team, the number of patients with bipolar disorder among the patients assessed by the team increases, but there remain a number of patients who do have unipolar depression; ^{Reference Bongards, Zaman and Agius5} in other words, the assessment of the trajectory of patients with mood disorder enables the discrimination between bipolar and unipolar depression.

We would comment that Duffy et al raise an important point in suggesting that a history of use of lithium by relatives of the patients changes the trajectory of bipolar disorder; however, in our experience it is very difficult to collect this information from patients, who often do not know details of their relatives’ illnesses. Furthermore, Duffy et al are right in proposing that it is possible to suggest a staging of bipolar disorder similar to McGorry’s staging of schizophrenia, but the schizophrenia staging is underpinned by Pantelis’ neuroimaging of the different stages of schizophrenia. To propose a staging model of bipolar disorder, we require similar neuroimaging results describing the differences between the individual stages.