Among 721 adults in the Democratic Republic of the Congo (489 with dementia, 232 cognitively unimpaired), poverty carried the strongest association with dementia (adjusted OR 2.34), followed by low resilience (1.84) and depression (1.53). Each standard-deviation increase in education cut odds by roughly half (aOR 0.51). Adding 16 social, behavioral and environmental factors raised Nagelkerke pseudo-R² from 0.106 (age and sex alone) to 0.368. Alcohol and PM2.5 showed no significant link. An apparent protective effect of obesity (aOR 0.31) rests on data that were 79.9% missing.

The education and depression findings echo the Lancet Commission's modifiable-risk framework, built largely on high-income cohorts, so this adds rare Central African data where dementia research is scarce. But the design limits what readers can take away. The cohort was non-representative and case-heavy, roughly two-thirds with dementia, so odds ratios do not translate into population risk. Cross-sectional timing leaves reverse causation wide open: dementia can cause depression, job loss and poverty, and low education may partly reflect cognitive reserve or measurement bias in testing. The obesity result is almost certainly an artifact of missingness or weight loss in dementia, and mean imputation adds further uncertainty. The authors themselves caution that the improved model fit does not estimate preventable burden. This is a preprint that has not been peer reviewed, so figures may shift. Assessment: confirmatory and hypothesis-generating, useful for context but not evidence that fixing poverty or depression prevents dementia.