Evidence for lithium as a maintenance treatment for bipolar disorder type II remains limited since most treatment-prevention studies focus on bipolar disorder type I or do not distinguish between types of bipolar disorder.

To compare the impact of lithium discontinuation on hospital utilisation in patients with bipolar disorder type I or schizoaffective disorder and patients with bipolar disorder type II or other bipolar disorder.

Mirror-image study, examining hospital utilisation within two years before and after lithium discontinuation as part of LiSIE, a retrospective cohort study into effects and side-effects of lithium for the maintenance treatment of bipolar disorder as compared to other mood stabilisers.

For the whole sample, the number of admissions increased from 86 to 185 admissions after lithium discontinuation, with the mean number of admissions/patient/review period doubling from 0.44 to 0.95 (p < 0.001). The number of bed days increased from 2218 to 4240, with the mean number of bed days/patient/review period doubling from 11 to 22 (p = 0.025). This increase in admissions and bed days was exclusively attributable to patients with bipolar disorder type I or schizoaffective disorder.

Our findings suggest that due to a higher relapse risk in patients with bipolar disorder type I or schizoaffective disorder there is a need to apply a higher threshold for discontinuing lithium than for patients with bipolar disorder type II or other bipolar disorder.

Michael Ott has been a scientific advisory board member of Astra Zeneca Sweden, Ursula Werneke has received funding for educational activities on behalf of Norrbotten Region (Masterclass Psychiatry Programme 2014–2018 and EAPM 2016, Luleå, Sweden): Astra

Currently, the evidence for lithium as a maintenance treatment for bipolar disorder type II (BD-II) remains limited. Guidelines commonly extrapolate recommendations for BD-II from available evidence for bipolar disorder type I (BD-I). Comparing the impact of lithium discontinuation is one way of assessing effectiveness in both groups.

To compare the impact of lithium discontinuation on hospital admissions and self-harm in patients with BD-I or schizoaffective disorder (SZD) and patients with BD-II or other bipolar disorder.

Mirror-image study, examining hospital admissions within 2 years before and after lithium discontinuation in both patient groups. This study was part of a retrospective cohort study (LiSIE) into effects and side-effects of lithium for maintenance treatment of bipolar disorder as compared with other mood stabilisers.

For the whole sample, the mean number of admissions/patient/review period doubled from 0.44 to 0.95 (P<0.001) after lithium discontinuation. The mean number of bed days/patient/review period doubled from 11 to 22 (P = 0.025). This increase in admissions and bed days was exclusively attributable to patients with BD-I/SZD. Not having consulted with a doctor prior to lithium discontinuation or no treatment with an alternative mood stabiliser at the time of lithium discontinuation led to more admissions.

The higher relapse risk in patients with BD-I/SZD suggests a higher threshold for discontinuing lithium than for patients with BD-II/other bipolar disorder. In patients with BD-II/other bipolar disorder, however, judged on the impact of discontinuation alone, lithium did not appear to prevent more severe depressive episodes requiring hospital admission.