The American Journal of Geriatric Psychiatry · Jun 27, 2026

Bayesian Reanalysis of a Pilot Randomized Clinical Trial of Low-Dose Lithium in Mild Cognitive Impairment: A Probabilistic Re-Expression of the LATTICE Trial

By da Silva AMP, Haddad-Santos D, de Siqueira Lima DV et al.

OBJECTIVE: The LATTICE pilot randomized clinical trial found no statistically significant effect of low-dose lithium on 6 coprimary outcomes in mild cognitive impairment (MCI). We aimed to re-express these results in Bayesian terms and estimate the probability of benefit, negligible difference, or harm across cognitive, neuroimaging, biomarker, and safety outcomes. METHODS: We conducted a Bayesian reanalysis of published aggregate data from LATTICE, a single-site, randomized, double-blind, placebo-controlled 2-year trial of low-dose lithium carbonate versus placebo. The trial randomized 83 older adults with MCI, of whom 80 initiated treatment. Outcomes included verbal and visuospatial delayed recall, Preclinical Alzheimer Cognitive Composite, hippocampal and cortical volume measures, brain-derived neurotrophic factor, and serious adverse events. Posterior means, 95% credible intervals (CrI), and posterior probabilities were estimated using weakly informative priors. RESULTS: Lithium was associated with a posterior mean standardized effect of 0.38 SD for CVLT-II delayed recall (95% CrI, -0.03 to 0.80), corresponding to a 96.5% probability of benefit and an 80.6% probability of exceeding 0.20 SD. The raw between-group difference was 1.47 points (95% CrI, 0.18-2.75). Bayesian re-expression of the intention-to-treat model showed an annualized CVLT-II treatment-by-time effect of 0.69 points per year (95% CrI, 0.02-1.35). BVMT-R remained near null. Hippocampal outcomes were directionally favorable but imprecise. Other cognitive, biomarker, imaging, and safety outcomes showed no robust signal. CONCLUSIONS: This Bayesian reanalysis supports an inconclusive interpretation of LATTICE, while identifying a probabilistic signal for verbal delayed recall-not seen in the other outcomes-that warrants testing in adequately powered trials.

doi.org/10.1016/j.jagp.2026.06.014PubMed 42521537Interventions

AgingResearch.News