Ultrafine particles and late-life cognitive function: Influence of stationary mobile monitoring design on health inferences.

Growing evidence links ultrafine particles (UFP) to neurotoxicity, but human studies remain limited. Various mobile monitoring approaches have been used to develop air pollution exposure models. However, whether design choices impact epidemiology, including for UFP and cognitive function, remains unclear. We evaluated the adjusted association between 5-year average UFP number concentration (PNC) and late-life cognitive function (Cognitive Abilities Screening Instrument - Item Response Theory [CASI-IRT]) in the Adult Changes in Thought cohort (N = 5283) by leveraging an extensive roadside mobile monitoring campaign specifically designed for epidemiology. To assess the impact of reduced monitoring approaches on this association, we repeatedly subsampled UFP measures from the campaign, developed exposure models, and evaluated the degree to which associations were impacted. In the primary analysis, each 1900 pt/cm increment in PNC was associated with an adjusted mean baseline CASI-IRT score that was 0.002 (95 % CI: -0.016, 0.020) higher, which was not statistically significant. Point estimates were consistent across sampling designs with fewer visits per site (≤12), fewer seasons (1-3), and unbalanced visit frequency across sites. Sampling designs restricted to rush hours were more similar (median point estimate 0.002, IQR of point estimates: 0.000, 0.003) than business hour designs (0.006, IQR: 0.005, 0.007), but the opposite was true when temporal adjustments were applied (rush: -0.003, IQR: -0.005, -0.001; business: 0.002, IQR: 0.001, 0.004). We observed similar results in sensitivity and secondary analyses. We did not find evidence of an association between UFP and cognitive function in fully adjusted models. Monitoring design had minimal impact on the inferential results in this setting, which may have been caused by the lack of association. Secondary analyses in a reduced model that is potentially confounded suggest that monitoring design might have a greater impact in other datasets. Further research is needed, particularly in contexts with robust statistically significant health associations.