Delaney · Environmental epidemiology (Philadelphia, Pa.) 2026 · Retrospective cohort study · n=10,366,083

Long-term cumulative associations of annual air pollution exposure and hospitalization with Parkinson's Disease.

Cited 0 times in the scientific literature.

Level 3 - non-randomized controlled study

Retrospective longitudinal cohort study linking environmental exposures to Medicare claims data

PubMed 42583532 · doi:10.1097/EE9.0000000000000519 · record verified 2026-08-27

What was done

Researchers analyzed Medicare fee-for-service claims from 10,366,083 beneficiaries aged 65 and older across the contiguous United States from 2000 to 2016. The primary outcome was a beneficiary's first hospitalization claim with a diagnosis code for Parkinson's disease (PD). Ten-year exposure histories were modeled for fine particulate matter (PM2.5), nitrogen dioxide (NO2), and summer ozone (O3). The authors used discrete-time survival analysis with distributed lag models to evaluate non-linear, lagged associations between air pollution exposure and PD hospitalization.

What was found

Elevated PM2.5 and NO2 exposures occurring at least 4 years prior to hospitalization were significantly associated with higher odds of first PD hospitalization. Comparing 10 years of continuous exposure at the 90th percentile versus the 0.5th percentile: - PM2.5 (11.8 vs. 3.0 µg/m³): Odds Ratio = 1.634 (95% CI: 1.489, 1.792) - NO2 (31.7 vs. 3.7 ppb): Odds Ratio = 1.474 (95% CI: 1.379, 1.575) Evidence for an association between summer O3 exposure and PD hospitalization was more limited.

Why it matters

This study provides population-scale evidence that long-term, cumulative exposure to common traffic- and combustion-related air pollutants years before clinical presentation is linked to Parkinson's disease morbidity severe enough to require hospitalization.

Limits

The outcome definition relies exclusively on inpatient hospitalization claims, missing outpatient diagnoses, early-stage PD, and community-managed disease. Exposure was estimated from spatial models rather than individual-level monitoring, creating potential exposure misclassification. The analysis is subject to unmeasured residual confounding common in administrative Medicare datasets, such as lifestyle factors, occupation, and smoking.

Cited by