When air pollution correlates with Alzheimer's disease, the natural assumption is that the link runs through cardiovascular damage. Particulate matter harms blood vessels; damaged blood vessels harm the brain. The causal story is reassuringly indirect: fix the heart problem and the brain problem follows.
A study of 27.8 million Medicare beneficiaries tested this assumption by measuring how much of the PM2.5–Alzheimer's association passes through three obvious mediators: hypertension, stroke, and depression (Loughborough University–led team, PLOS Medicine 2026). Hypertension mediated 1.6%. Stroke mediated 4.2%. Depression mediated 2.1%. Together, the conditions that should carry the signal carried almost none of it. The association — HR 1.085 per 3.8 µg/m³ increase in five-year average exposure — remained essentially intact after accounting for every plausible intermediary.
What remains is the direct neurotoxic pathway: ultrafine particles crossing the blood-brain barrier or entering through the olfactory nerve, accumulating in brain tissue, driving neuroinflammation without first damaging the cardiovascular system. The particles don't need a middleman. They are small enough to arrive on their own.
The general principle: when a causal link between A and C survives controlling for every known B, the comfortable assumption that A→B→C was never the main story. The intermediaries were bystanders — real phenomena, genuinely correlated, but not carrying the weight the narrative assigned them. The directness of a causal path is not a function of how many steps seem plausible between endpoints. It is an empirical question, and the answer can be that most of the plausible steps are scenery.