A rapidly evolving evidence base suggests that exposure to outdoor air pollution is a risk factor for the onset of dementia, with an upturn in publications since 2022. We sought to synthesise and critically assess this evidence base accounting for the latest studies.

In this systematic review and meta-analysis, we searched MEDLINE, Embase, Cochrane Library, CINAHL, Global Health, PsycINFO, Scopus, and Web of Science Core Collection from database inception up to Oct 23, 2023, for primary observational studies of adults (aged ≥18 years) that provided a quantitative analysis of the association between long-term (≥1 year) exposure to outdoor air pollutants and a subsequent physician diagnosis of dementia. When three or more independent studies reported an exposure–outcome pair, effect estimates of the association were extracted and harmonised to a prespecified exposure increment, and included in inverse-variance weighted random-effects meta-analyses. Between-study inconsistency was assessed using the I² statistic and the Cochran Q test. Study-level risk of bias and confidence in the overall body of evidence were assessed with the Office of Health Assessment and Translation tool, and publication bias was examined. The protocol for this review was registered with PROSPERO, CRD42023414413.

The search generated 15 619 records, of which 51 studies met the inclusion criteria for data extraction. After excluding studies due to population overlap and missing continuous effect estimates, 32 studies reported on exposure–outcome pairs that met the threshold of three or more studies, and were included in meta-analyses of adjusted effect estimates for incident dementia and/or in subgroup analyses of dementia subtypes. In meta-analyses of incident dementia, we identified a dementia diagnosis to be significantly associated with long-term exposure to PM_2·5 (21 studies, n=24 030 527, pooled adjusted hazard ratio (HR) per 5 μg/m³ increase in exposure, 1·08 [95% CI 1·02–1·14]; I²=95%), nitrogen dioxide (16 studies, n=17 228 429, pooled adjusted HR per 10 μg/m³ increase, 1·03 [1·01–1·05]; I²=84%), and black carbon/PM_{2·5 absorbance} (six studies, n=19 421 865, pooled adjusted HR per 1 μg/m³ increase, 1·13 [1·01–1·27]; I²=97%). We found no significant association for exposure to nitrogen oxides (five studies, n=241 409, pooled adjusted HR per 10 μg/m³ increase, 1·05 [0·97–1·13]; I²=44%), PM₁₀ (four studies, n=246 440, pooled adjusted HR per 15 μg/m³ increase, 1·52 [0·80–2·87]; I²=82%), or annual ozone (four studies, n=419 972, pooled adjusted HR per 45 μg/m³ increase, 0·82 [0·35–1·92]; I²=69%), with moderate to considerable heterogeneity between studies in these pooled analyses. Of the 32 studies overall, three (9%) had a probably high risk of bias in one of seven domains; all other studies had ratings of probably to definitely low risk of bias. The overall certainty of evidence of studies in the systematic review was moderate.

This analysis adds to the body of evidence that outdoor air pollutants are risk factors for dementia, indicating that reduced exposure to pollution could reduce dementia rates and stricter air quality standards would likely provide substantial health, social, and economic benefits.