Statistics from Altmetric.com
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.
Systematic review with meta-analysis.
The Cochrane Collaboration databases, reference lists, textbooks, relevant journals, and conference abstracts were searched. Trial authors were contacted, as were other experts and pharmaceutical companies.
Study selection and analysis:
Randomised controlled trials (RCTs) of ⩾3 months, comparing lithium with placebo for the maintenance of bipolar disorder were included. Trials where one group abruptly discontinued lithium and those with mixed groups of people with bipolar and unipolar disorder were excluded. Study quality was assessed to determine allocation concealment and blinding. STATA software was used to conduct meta-analyses, leading to pooled fixed effects and random effects relative risk (RR) ratios.
Risk of relapse (total, manic, depressive), total withdrawal rates, risk of specific adverse events.
Five RCTs met inclusion criteria (n = 770), with follow up periods of 11 months to four years. Total relapses: lithium prevented significantly more relapses compared with placebo (risk of relapse, lithium v placebo: 40% v 60%; RR 0.65, 95% CI 0.50 to 0.84). Manic episodes: lithium prevented significantly more manic episodes compared with placebo (risk of relapse, lithium v placebo: 14% v 24%; RR 0.62, 95% CI 0.40 to 0.95). Depressive relapses: there were no significant differences between groups (risk of relapse lithium v placebo: 25% v 30%; RR 0.85, 95% CI 0.66 to 1.11). Overall withdrawal from trials was lower with lithium (RR = 0.86, 95% CI 0.80 to 0.93). Adverse effects more common with lithium were somnolence, nausea, and diarrhoea.
Recent trial evidence shows that lithium reduces the rate of relapse for people with bipolar disorder. A meta-analysis of five trials showed a significant reduction in relapses in the lithium group overall, more pronounced for manic episodes than depressive episodes.
Although lithium has been in use for over 50 years, much of the RCT evidence is new, having been published from 2000 onward. In these trials, lithium has been used in a comparison arm when new drug agents are being studied for management of bipolar disorder. Bias in favour of lithium is therefore unlikely. In the original paper, there was an error in table 1: the previous lithium use in Kane et al’s 1982 study should have been listed as “unclear”. Also, in figure 3, the rate of depressive relapse with placebo in Bowden et al’s 2003 study should have been 30% (n = 21 of 70). The pooled depressive relapse rate with placebo was therefore 30% (n = 89 of 297) rather than 32%. The pooled fixed effect relative risk (reported on p 219) was 0.83 (95% CI 0.63 to 1.08) and the random effects relative risk was 0.85 (95% CI 0.66 to 1.11). These changes do not materially change the conclusions of the article.
For more than 50 years, lithium has been used for acute and long term treatment of mood disorders. Worldwide, many evidence-based treatment guidelines plus clinicians and researchers consider lithium to be the mood stabiliser of choice for maintenance treatment of bipolar1 and unipolar2 disorder. However, until recently, the number of large controlled studies with sound methodology has been limited. Lithium, being an inexpensive drug, has not been a focus of pharmaceutical companies for obvious reasons for many years. Due to new drug developments this situation has changed. As several recent large scale trials investigating the efficacy of newer compounds for bipolar disorder included lithium as a comparator drug, it seems more than timely to re-examine lithium’s efficacy in relapse prevention in bipolar disorder. The number of studies and subjects included in these trials is now sufficiently high to perform a meta-analysis.
The paper by Geddes et al is, to the best of my knowledge, the first published, methodologically rigorous meta-analysis of randomised, double blind, placebo controlled trials comparing lithium with placebo in the prevention of relapse in bipolar disorder. The overall results of this meta-analysis give firm evidence that lithium reduces the risk of affective relapse in bipolar disorder. Specifically, there is reasonable evidence that lithium prevents manic relapse, but less evidence that lithium prevents depressive relapses. These results will reassure clinicians that lithium still “works”, particularly for the prevention of mania and also for relapse prevention of depression, albeit to a lesser extent compared with mania. The latter result suggests that clinicians may want to consider alternative mood stabilisers3 if lithium fails to prevent depression in individual patients, or a combination of another mood stabiliser with lithium. Unfortunately, the evidence from controlled trials on the efficacy of such combination treatments is very limited to date, but the combination of two or even more mood stabilisers seems clinically reasonable and therefore worth recommending to effectively prevent depressive in bipolar disorder.
For correspondence: John Geddes, Department of Psychiatry, University of Oxford, Warneford Hospital, Oxford OX3 7JX, UK;
Sources of funding: one author was funded by the Oxfordshire Mental Healthcare Trust. The study was carried out under the auspices of the Depression, Anxiety and Neurosis Group of the Cochrane Collaboration.