A systematic review searched to December 2024 identified 23 primary studies of lithium exposure and dementia, 11 with extractable estimates. Four cohort studies pooled to a hazard ratio of 0.633 (95% CI 0.517 to 0.775), while three randomised trials of cognitive outcomes in 229 participants gave a standardised mean difference of -0.410 (95% CI -0.810 to -0.020). Two ecological studies gave an incidence rate ratio of 0.830, heterogeneity was substantial, and the authors state that causality is not established.
Why it is interesting: The pooled observational hazard ratio sits beside randomised evidence limited to three trials of cognitive outcomes.