Citation

Moncrieff J, Ploderl M, Nabi Z, Stansfeld J, Wood L. Epidemiol. Psychiatr. Sci. 2022; 31: e88.

Copyright

(Copyright © 2022, Cambridge University Press)

DOI

10.1017/S2045796022000737

PMID

36510829

Abstract

Bschor et al.'s conclusion that the anti-suicidal effect of lithium is 'well-established' is not justified, and is challenged even by other proponents of lithium (Baldessarini and Tondo, Reference Baldessarini and Tondo2022). We conducted an updated meta-analysis of data on suicide from randomised trials of lithium because previous meta-analyses have excluded the majority of recent data due to using the Peto method, which cannot handle trials with zero events Therefore, these meta-analyses were based on small numbers of participants - 244 on lithium and 241 on placebo in the meta-analysis of placebo-controlled trials by Cipriani et al. (Cipriani et al., Reference Cipriani, Hawton, Stockton and Geddes2013), for example. In the case of an infrequent outcome like suicide, it is now recognised that including data from trials with zero events leads to more reliable estimates (Ren et al., Reference Ren, Lin, Lian, Zou and Chu2019). Our analyses were based on data from 1278 people allocated to lithium and 1300 on placebo.

Concluding that lithium has anti-suicidal effects involves relying on older and thus potentially unreliable studies or lower-level evidence from observational studies and, ignoring studies without events, potential biases in the literature that usually lead to overestimations of treatment effects and results for suicide attempts. But even with such an approach, uncertainty remains high due to the low overall number of suicides and wide confidence intervals...

Language: en