Fine Particulate Matter and Parkinson Disease Risk Among Medicare Beneficiaries
This article has been corrected.
VIEW CORRECTIONAbstract
Background and Objectives
Numerous studies suggest that environmental exposures play a critical role in Parkinson disease (PD) pathogenesis, and large, population-based studies have the potential to advance substantially the identification of novel PD risk factors. We sought to study the nationwide geographic relationship between PD and air pollution, specifically PM2.5 (particulate matter with a diameter <2.5 micrometers), using population-based US Medicare data.
Methods
We conducted a population-based geographic study of Medicare beneficiaries aged 66–90 years geocoded to US counties and zip+4. We used integrated nested Laplace approximation to create age, sex, race, smoking, and health care utilization–adjusted relative risk (RR) at the county level for geographic analyses with PM2.5 as the primary exposure of interest. We also performed an individual-level analysis using logistic regression with cases and controls with zip+4 centroid PM2.5. We adjusted a priori for the same covariates and verified no confounding by indicators of socioeconomic status or neurologist density.
Results
Among 21,639,190 Medicare beneficiaries, 89,390 had incident PD in 2009. There was a nationwide association between average annual PM2.5 and PD risk whereby the RR of PD was 56% (95% CI 47%–66%) greater for those exposed to the median level of PM2.5 compared with those with the lowest level of PM2.5. This association was linear up to 13 μg/m3 corresponding to a 4.2% (95% CI 3.7%–4.8%) greater risk of PD for each additional μg/m3 of PM2.5 (ptrend < 0.0001). We identified a region with high PD risk in the Mississippi-Ohio River Valley, where the risk of PD was 19% greater compared with the rest of the nation. The strongest association between PM2.5 and PD was found in a region with low PD risk in the Rocky Mountains. PM2.5 was also associated with PD in the Mississippi-Ohio River Valley where the association was relatively weaker, due to a possible ceiling effect at average annual PM2.5 levels of ∼13 μg/m3.
Discussion
State-of-the-art geographic analytic techniques revealed an association between PM2.5 and PD that varied in strength by region. A deeper investigation into the specific subfractions of PM2.5 may provide additional insight into regional variability in the PM2.5-PD association.
Introduction
Several studies have linked air pollution in the form of aerosolized particulate matter to various adverse health outcomes. Recent investigations identified associations between fine particulate matter, that is, particulate matter with diameter ≤2.5 micrometers (PM2.5), and neurologic disease, including dementia1 and stroke.2 The ultrafine particles (≤0.1 micrometers) in PM2.5 cross the blood-brain barrier in humans.3 In addition, some subcomponents of PM2.5 are more neurotoxic than others. In particular, PM2.5 can contain heavy metals, including arsenic and manganese, which have been implicated in the neuropathogenesis of basal ganglia degeneration.4,5 Despite this, epidemiologic investigations of PM2.5 and Parkinson disease (PD) have yielded mixed results,6-20 with marked discrepancies in the magnitude, shape, and even direction of PM2.5-PD associations. One possible contributor is the use of different PD-related outcomes and widely varying definitions of “incidence.” In addition, the range of PM2.5 levels across studies varies widely, and differences in the size and content of particulate matter in different regions might influence PM2.5-PD associations. Accordingly, several prior studies found PM2.5-PD associations to differ by geographic area10-12 or urban/rural land use.16,18 Several powerful spatial analytic methods offer to advance our understanding of the role of PM2.5 in PD by enabling high-resolution, population-based investigations of the geographic distribution of PD, to characterize patterns of incidence and their relation to PM2.5 exposure.
Similar to studies of PM2.5 and PD, existing research into the national distribution of PD primarily consists of studies of mortality and prevalence.21-23 One study found a North-South gradient in PD mortality and prevalence, which likely reflects the general burden of disease but does not capture the geographic patterns of incidence necessary for understanding the role of environmental risk factors. In addition, most nationwide studies of PD rely on state-level data.22,24 To date, only one nationwide, county-level study of incident PD has been conducted in the United States.21 This study found nonrandom clustering of PD in the Midwest and East South Central United States, which likely reflects, in part, the effect of environmental exposures on PD risk. Deeper exploration of these PD clusters, and the potential role of PM2.5 in contributing to those clusters, requires further investigation with advanced geographic methods and geostatistical approaches not yet applied to neurodegenerative disease. We conducted a US population-based geographic study of PD risk to examine spatial patterns of newly diagnosed PD and relationships with PM2.5, using a multimethod approach that included spatial analytic and statistical methods. We hypothesized that we would identify spatial clustering of PD and observe a positive association between PM2.5 and PD, which would vary by region.
Methods
This study was approved by the Human Research Protection Office at Washington University School of Medicine in St. Louis and by the Centers for Medicare & Medicaid Services. Participant consent in this records-based study was not required.
Study Population and Case Ascertainment
Our eligibility criteria were designed to ensure a population-based sample with complete data: age-eligible for Medicare ≥2 years before diagnosis/selection (66 years and older), no Part C (Medicare Advantage/health maintenance organization) coverage, 90 years and younger, and US residence, all in 2009, without additional (e.g., medical) inclusion/exclusion criteria.25 Incident PD cases included all study-eligible beneficiaries with at least one International Classification of Diseases, Ninth Revision, diagnosis code of 332 or 332.0 in 2009, but no prior year. This case definition, similar to a few prior studies of PM2.5 and PD,17,19,20 which maximizes sensitivity without materially affecting specificity,26 aims to ensure representativeness of cases, including with regard to exposures. Beneficiaries with a diagnosis of atypical parkinsonism (333.0) or Lewy body dementia (331.82) were excluded if diagnosed in the year of PD diagnosis or earlier (4.6% potential cases).25 We geocoded beneficiaries to their residential zip+4 after applying a two-year lag in their residential history, which we obtained from the 2007 Medicare beneficiary annual summary file. For maps and geographic analyses, we linked beneficiaries to county of residence to avoid zero inflation.
PM2.5 Exposure Estimation
Our exposure of a priori interest was average annual PM2.5 from 1998 to 2000,27 a period largely before PD onset. We used this period because it is likely etiologically relevant and 90% of beneficiaries maintained the same county code (84% had the same 5-digit zip code) for all years available (2004–2009). PM2.5 data,27 which were available in 1-kilometer grids, were based on several predictors, including satellite, meteorologic, land use, and elevation data with varying resolutions.28 This PM2.5 model achieved a cross-validated R2 of 0.89. All 3 prior PD studies that used this or a similar PM2.5 model7,16,17 observed significant associations between PM2.5 and PD, consistent with an acceptable degree of exposure measurement error in the context of air pollution health effects research, where associations are generally difficult to detect. We then used geographic information systems to estimate average annual PM2.5 at 2 different geographic levels: (1) county for mapping and spatial analyses and (2) zip+4 (centroid) for individual-level regression analysis.
Assessment of Covariates
For all essential covariate data, we started with data at the individual (beneficiary) level and only collapsed to the county level for mapping and spatial analyses. Covariates included beneficiary demographic information (age, sex, and race) and measures of health care utilization in the year before diagnosis/reference from the beneficiary annual summary file. We defined health care utilization as the number of physician visits (carrier) and outpatient visits either for the individual or for county-level analyses per county for all Medicare beneficiaries in our study. We obtained the county-level current prevalence of smoking cigarettes29 for county-level/geographic analyses and estimated the probability of ever smoking at the individual level for individual-level analyses. We developed this smoking variable for use in geographic studies through a multivariable linear regression model with our validated claims-based probability of smoking30 as the outcome. This gold-standard variable replicates the well-established relationship between smoking and PD.25 Predictors of smoking probability in the new model were county-level prevalence of current smoking29 and individual-level data from the beneficiary annual summary file. Similar to the original smoking model based on detailed claims, these individual-level predictors included sex, race, birth cohort, and selected medical conditions—here chronic obstructive pulmonary disease, lung cancer, stroke, acute myocardial infarction, other ischemic cardiovascular disease, stroke/transient ischemic attack, chronic kidney disease, osteoporosis, and depression. This smoking variable along with use of care, age, sex, and race represented our core set of covariates. As additional covariates for sensitivity analyses, we obtained the following ecologic data: census tract airborne trichloroethylene,31 county-level area deprivation,32 census block group median household income,33 neurologists per county per 100,000 Medicare beneficiaries,21 and county-level agricultural pesticide use.34 In addition, from the beneficiary annual summary file, we obtained individual-level data on acute myocardial infarction, ischemic heart disease, stroke/transient ischemic attack, congestive heart failure, chronic obstructive pulmonary disease, and diabetes. These 6 conditions might be caused by PM2.5 exposure and subsequently lead to care that facilitates diagnosis of PD, that is, act as nuisance mediators.
Estimation of PD Relative Risk for Counties
We used integrated nested Laplace approximation for Bayesian inference in R-Project (R-INLA)35 using R version 4.1.2 to estimate county-PD relative risks (RR) that account for known demographic risk factors36-40 and spatial dependency. Specifically, these R-INLA–derived RRs were based on Gaussian distribution using indirect age-sex-race standardized incidence ratios, health care utilization, and smoking as input. We calculated the standardized incidence ratio by dividing the number of observed cases by expected counts and then multiplying by 100, based on 4 age (65–69, 70–74, 75–79, and 80+ years), sex, and 5 race (White, Black, Asian, Hispanic, and other) strata. To address spatial autocorrelation, we integrated spatial dependency into the R-INLA model using conditional autoregressive distribution to smooth risks according to the standardized incidence ratios of neighboring counties. The R-INLA–derived PD RR for each county was available for all county-level analyses.
Assessment of Spatial Clustering of PD
To formally test for spatial clustering and identify high and low PD risk counties, we used univariate local indicators of spatial association (LISAs) to map PD hot and cold spots.41 Hot-spot counties have above-average PD risk (RR) and share boundaries with counties that all have above-average PD risk. Cold-spot counties have below-average PD risk and share boundaries with counties that have below-average PD risk. LISAs identify where high and low-risk counties form contiguous clusters rather than only focusing on counties with significant RRs. LISAs also provide Global Moran's I value to describe the nationwide presence or absence of clustering.
Examination of the PM2.5-PD Association
We used 3 approaches to examine the relationship between PM2.5 and PD—a regression model for assessing the nationwide relationship and 2 geographic approaches for assessing regional relationships. Specifically, for our nationwide assessment, we performed traditional multivariable regression at the individual level using Stata/MP version 17. For regional assessment, to explore whether the PM.2.5-PD association differed by region, we used bivariate LISAs implemented in GeoDa version 1.2041 and geographically weighted regression (GWR) implemented in ArcGIS. Bivariate LISAs were used to assess and visualize local spatial correlation between PD and PM2.5 while GWR was used to determine and quantify the direction and strength of local associations.
Logistic Regression Analysis
We performed logistic regression with PD as the outcome and zip+4 PM2.5 as the independent variable, adjusted for the same a priori covariates as above but assessed at the individual level. To initially examine the association while allowing for nonlinear associations, we modeled PM2.5 as deciles, with the lowest decile of PM2.5 as the reference group. Based on these results, we then sought to develop a more parsimonious model using linear splines. We selected the final model using the Akaike information criterion while also examining the sensitivity of results to knot number and placement. To assess whether restriction to Medicare-aged beneficiaries could limit generalizability, we tested whether the PM2.5-PD association differed by age. To assess whether associations might be due to occupational rather than environmental exposures, we tested whether the association was stronger in men than women. We tested for interaction on the multiplicative scale while including main-effects terms in the model. As a sensitivity analysis, we included Lewy body dementia cases. Because PD is relatively rare, the odds ratio provides an accurate estimate of the RR.
Spatial Correlation
We used bivariate LISAs to overlay PM2.5 and PD hot and cold spots and assess local spatial correlation.41 A bivariate LISA map is the convergence of 2 univariate LISA maps into a single map (i.e., a map of PD hot and cold spots + a map of PM2.5 hot and cold spots). A bivariate LISA map delineates 4 cluster categories (high-high, low-low, low-high, and high-low). The high-high category represents counties where PD risk and PM2.5 levels are both high relative to their means, and the low-low category represents counties where PD risk and PM2.5 levels are both low; these categories are consistent with a positive correlation between PM2.5 and PD. The 2 discordant (low-high and high-low) categories are inconsistent with a positive correlation.
Spatial Regression
We used GWR to determine the direction and strength of the local associations between PM2.5 and PD (ArcGIS). In GWR, local regression is performed for each county using surrounding counties. We used an adaptive spatial bandwidth to define the latter. GWR computed a regression coefficient for each county using county-PD RR as the dependent variable and PM2.5 as the independent variable. We retained RR as a continuous outcome and modeled PM2.5 in deciles in a single continuous variable. Thus, the GWR beta coefficients represent the absolute difference in the PD RR when going from any decile of PM2.5 to the decile below. We then computed 95% confidence intervals (CIs) and mapped the GWR coefficients for counties where the CI excluded zero (significant at a two-sided alpha = 0.05). This allowed us to include both the positive and negative coefficients in a map, as well as to produce a map focused on regions where GWR coefficients were positive, providing a more conservative reflection of where the association between PM2.5 and PD was positive. In addition, we used the Monte Carlo test of spatial variability to assess objectively whether the relationship between PM2.5 and PD varied across the nation.
Data Availability
The Centers for Medicare & Medicaid Services does not permit data sharing under the data use agreement.
Results
After excluding prevalent PD, 21,639,190 US Medicare beneficiaries from Medicare research files42 met initial study eligibility criteria,43 including 89,790 PD incident cases and 21,549,400 non-cases. Of these, we had 65,180 cases (73%) and 15,561,435 non-cases (72%) with high-resolution residence information available (zip+4) 2 years before PD diagnosis or the control reference date. We were restricted to those with zip+4 information so that our ecologic exposure estimates would better represent individual-level exposures. We observed no demographic differences between the full data set43 vs this sample (Table 1), overall or by case status. In both samples, 89% of cases and 86% of controls were non-Hispanic White and 6% of cases and 8% of controls were Black. Sex and age distribution for cases and controls were essentially identical between the full and present sample as well.
| PD cases (N = 65,180) | Controls (N = 15,561,435) | |
|---|---|---|
| Female, %b | 51% | 58% |
| Race/ethnicity, % | ||
| White | 89% | 86% |
| Black | 6% | 8% |
| Hispanic | 2% | 2% |
| Asian | 2% | 2% |
| Native American | 0.1% | 0.2% |
| Pacific Islander/other/unknown | 1% | 2% |
| Age, mean (standard deviation) | 79 (6.1) | 76 (6.2) |
Abbreviations: PD = Parkinson disease; PM2.5 = particulate matter <2.5 micrometers in diameter.
a
All Medicare beneficiaries with incident PD or without PD in 2009 who met basic criteria; methods and sample are described in Silver et al. 2020.43 Excludes 19,144 cases and 4,693,525 controls without geocodable zip+4 data for residence 2 years before PD diagnosis or the control reference date and 3,140 cases and 740,179 controls without PM2.5 data at the geocoded location in the United States.
b
Percentage excludes 47 controls with unknown sex.
We found a positive association between PM2.5 and PD where the RR for PD was 11%–28% higher in each of the 9 upper deciles of PM2.5 relative to the lowest decile (Table 2). The association was strictly linear in the lower 5 deciles up to at least 13 μg/m3 and then began to weaken as differences in PM2.5 levels became more similar across deciles. Nonetheless, risk continued to increase generally, with exception of the highest 2 deciles, resulting in a significant trend overall (ptrend < 0.0001) (Table 2). When fitting a model with 2 linear splines, the model with the lowest (best) Akaike information criterion (814702.9) incorporated a knot at 13 μg/m3 with 4.2% greater risk of PD for each additional μg/m3 of PM2.5 up to this level and then a weak, nonsignificant increase with increasing PM2.5 levels thereafter resulting in a plateau (Table 2, overall PM2.5-PD association p < 0.0001). The spline model was sensitive to knot placement, with 2 significant positive splines observed when the knot was at 12 μg/m3 (RR = 1.048, 95% CI 1.041–1.055 per μg/m3 of PM2.5 up to this level and 1.007, 95% CI 1.003–1.012 per μg/m3 of PM2.5 thereafter, Akaike information criterion 814710.6), for example, but equally good yet contrasting models with a knot at either 14 or 15 μg/m3. A simple linear model yielded a higher (poorer) Akaike information criterion (814770.9), and a likelihood ratio test confirmed a poorer fit relative to our best spline model (p < 0.0001). Based on the best spline model, the PD RR was 1.56 (95% CI 1.47–1.66) when comparing beneficiaries exposed to 12.93 μg/m3 of PM2.5 (median PM2.5) with beneficiaries exposed to 2.16 μg/m3 of PM2.5 (minimum PM2.5). With adjustment for 6 medical conditions associated with PM2.5 that might facilitate PD diagnosis, the RR was 1.43 (95% CI 1.35–1.51). Adjustment for socioeconomic variables, access to neurologists, and pesticide use did not affect the RR for PM2.5 and PD (eTable 1, links.lww.com/WNL/D163). Airborne trichloroethylene also did not confound the PM2.5-PD association (Table 2). In addition, the results were unchanged when Lewy body dementia was included in the case definition. We found no evidence that the PM2.5-PD association differed by age (pinteraction = 0.77). The pattern of association was the same in men and women, but with slightly stronger associations in women for both splines (pinteraction = 0.001). The counties with average annual PM2.5 levels that fell within Spline 1 were largely in the Western half of the United States (Figure 1). However, the results were fairly similar in urban areas, suburban areas, small towns, and rural areas, with significant RRs ranging from 1.040 to 1.047 per μg/m3 of PM2.5 up to 13 μg/m3 and then relatively flat thereafter for each of land use type (likelihood ratio pinteraction = 0.69).
| PM2.5 Decile | Zip+4 level PM2.5, μg/m3, range | RR (95% CI)a | Also adjusted for airborne trichloroethylene RR (95% CI)a,b |
|---|---|---|---|
| D1 | 2.16 to 9.47 | 1.00 (Reference) | 1.00 (Reference) |
| D2 | 9.50 to 10.87 | 1.11 (1.07–1.15)c | 1.11 (1.07–1.15)c |
| D3 | 10.90 to 11.90 | 1.16 (1.12–1.20)c | 1.16 (1.12–1.20)c |
| D4 | 11.93 to 12.90 | 1.22 (1.17–1.28)c | 1.21 (1.17–1.26)c |
| D5 | 12.93 to 13.70 | 1.24 (1.19–1.28)c | 1.23 (1.19–1.28)c |
| D6 | 13.73 to 14.27 | 1.23 (1.19–1.28)c | 1.23 (1.18–1.27)c |
| D7 | 14.30 to 14.73 | 1.26 (1.22–1.31)c | 1.26 (1.21–1.30)c |
| D8 | 14.77 to 15.33 | 1.28 (1.24–1.33)c | 1.28 (1.24–1.33)c |
| D9 | 15.37 to 16.20 | 1.26 (1.21–1.30)c | 1.26 (1.21–1.30)c |
| D10 | 16.23 to 23.30 | 1.24 (1.19–1.28)c | 1.23 (1.19–1.28)c |
| ptrend < 0.0001 | ptrend < 0.0001 |
| PM2.5 Spline | RR (95% CI)a per μg/m3 PM2.5 | RR (95% CI)a per μg/m3 PM2.5 | |
|---|---|---|---|
| Spline 1 | 2.16 to 13.00 | 1.042 (1.037–1.048)c | 1.042 (1.037–1.048)c |
| Spline 2 | 13.03 to 23.30 | 1.002 (0.997–1.008) | 1.001 (0.996–1.007) |
| p < 0.0001d | p < 0.0001d | ||
| Median vs minimum PM2.5 | 12.93 vs 2.16 | 1.56 (1.47–1.66)c | 1.56 (1.47–1.66)c |
Abbreviations: μg/m3 = =micrograms per cubic meter; CI = confidence interval; PD = Parkinson disease; PM2.5 = particulate matter <2.5 micrometers in diameter; RR = relative risk.
a
RRs estimated from odds ratios and associated 95% CIs from logistic regression with PD as the outcome and zip+4 level PM2.5 as the independent variable, adjusted for individual-level variables (age, sex, race, ever/never smoking, and use of medical care), based on 65,180 cases and 15,561,435 controls with geocodable zip+4 data for residence 2 years before PD diagnosis or the control reference date and with PM2.5 data at the geocoded location in the United States.
b
Cases and controls were assigned the trichloroethylene value of the census tract that corresponded to their zip+4 center.
c
p < 0.001.
d
Overall p-value for the PM2.5-PD association, assessed using a likelihood ratio test comparing models with and without both splines.
The relationship between individual-level (zip+4) PM2.5 exposure and risk of PD among US Medicare beneficiaries in 2009 was well-described by a logistic regression model with 2 linear splines with a knot at 13 μg/m3 PM2.5. The adjusted relative risk of PD was 1.56 (95% CI 1.47–1.66) when comparing beneficiaries exposed to 12.93 μg/m3 of PM2.5 (the median of PM2.5) with beneficiaries exposed to 2.16 μg/m3 of PM2.5 (the lowest level of PM2.5 and the reference group for the individual-level line in the plot). The adjusted relative risk of PD was 1.60 (95% CI 1.51–1.70) when comparing beneficiaries exposed to the highest PM2.5 with this same reference group. A locally weighted scatterplot smoothing curve between county-level mean PM2.5 for the smoothed, standardized county-PD relative risk (with all other counties in the contiguous United States as the reference group). Each dot represents a county.
In univariate LISA (hot and cold spot) analysis, we found moderate clustering of similar values for PD risk across the contiguous United States (Global Moran's I value of 0.500, p < 0.05). The associated LISA map revealed an S-shaped pattern of high PD risk in the Mississippi-Ohio River Valley and a PD cold spot in a large portion of the Western part of the nation (Figure 2). PD risk was 19% greater in the Mississippi-Ohio River Valley hot spot compared with the rest of the nation. Bivariate LISAs revealed local spatial correlation characterized by high-high clusters (counties with above average PD risk and above average PM2.5) and low-low clusters (counties with below average PD risk and below average PM2.5) in the above respective areas (Figure 3). Several other smaller high-high or low-low areas and only a small number of discordant areas were observed. The Monte Carlo test of spatial variability further revealed spatial variation in the local parameter estimates for PM2.5 as a PD risk factor (p = 0.0001). GWR identified specific clusters of counties where the positive association between PM2.5 and PD risk was the strongest (Figure 4). The strongest positive coefficients formed a cluster of 51 counties within our PD cold spot in the mountainous regions of Colorado and Wyoming, where the PM2.5-PD RR increased, in absolute terms, by approximately 0.15–0.16 per decile of PM2.5 exposure. Although much weaker, we also identified a positive association between PD and PM2.5 in our PD hot spot: For 118 counties North of the Mississippi-Ohio river confluence (in Indiana, Ohio, Illinois, and Missouri), the RR of PD increased by approximately 0.03–0.04 with each additional decile of PM2.5 exposure. Four regions of the country demonstrated null or inverse associations between PM2.5 and PD. These included the North Dakota-Minnesota border; parts of the Mid-Atlantic; South Atlantic; and a region that spans part of Washington State, Idaho, and Montana (eFigure 1, links.lww.com/WNL/D162).
Areas of high (hot spots) and low (cold spots) risk of PD among Medicare beneficiaries in the contiguous United States in 2009. Hot and cold spots were identified using univariate LISA analyses and county-PD relative risks that account for age, sex, race, smoking, health care utilization, and spatial dependency. LISA = local indicator of spatial association; PD = Parkinson disease.
Cluster map showing how PD hot and cold spots among US Medicare beneficiaries in 2009 overlay with PM2.5 hot and cold spots in 1998–2000. Clusters were identified with bivariate LISA using (1) relative risks that account for age, sex, race, smoking, health care utilization, and spatial dependency and (2) average annual PM2.5. The high-high category represents counties where risk of PD and PM2.5 exposure are both high relative to means. The low-high and high-low clustering categories describe spatial outliers where a low-high county is one with below-average PD risk and above-average PM2.5. The low-low category represents counties where PD risk and exposure are both less relative to their means. LISA = local indicator of spatial association; PD = Parkinson disease; PM2.5 = particulate matter <2.5 micrometers in diameter.
GWR coefficient map showing location and strength of significant positive coefficients for the association between county-level average annual PM2.5 and Parkinson disease relative risk in the United States that account for age, sex, race, smoking, health care utilization, and spatial dependency. Each coefficient represents the absolute difference in the relative risk with each additional decile of PM2.5 exposure. The strongest associations appear in Colorado and Wyoming, which have relatively low risk of Parkinson disease and relatively low PM2.5 levels compared with the rest of the nation. This pattern of coefficients is an artifact of a possible ceiling effect in the Mississippi-Ohio River Valley, which has above-average Parkinson disease risk and some of the highest PM2.5 levels in the nation. These patterns of coefficients are consistent with the observed positive association between PM2.5 and PD nationwide overall that is stronger at lower PM2.5 levels than higher PM2.5 levels. GWR = geographically weighted regression; PM2.5 = particulate matter <2.5 micrometers in diameter.
Discussion
In this multimethod study investigating the nationwide patterns of PD risk and its relation to PM2.5, we found an S-shaped pattern of high PD risk (hot spot) in the Mississippi-Ohio River Valley and low PD risk (cold spot) in the Western United States. This geographic pattern aligns broadly with our prior study of incident PD21 while providing a more refined pattern. Our bivariate LISA correlation map for PD and PM2.5 closely overlapped with our PD risk map, in that PD hot spots generally aligned with PM2.5 hot spots and PD cold spots generally aligned with PM2.5 cold spots. Although bivariate LISA maps are exploratory and cannot alone confirm relationships at the nation level, we confirmed a PM2.5-PD association nationally. In our regression analysis using individual-level data, the association between PD and PM2.5 was linear up to at least 13 μg/m3 PM2.5. At the highest levels of PM2.5, the relationship between PM2.5 and PD appeared to plateau, but the overall association remained positive. Our GWR results also suggested a possible ceiling effect as the association weakened in the Mississippi-Ohio River Valley where some of the highest levels of PM2.5 in the nation overlay regions with some of the highest PD risk in the nation. Although the reason for the plateau is unclear, several studies report effect estimates consistent with a similar plateau.12-14,16 It is more important that the robust dose-response association at the lower levels of PM2.5 potentially has substantial public health relevance. Recently, the US Environmental Protection Agency proposed to revise the primary (health-based) annual PM2.5 standard level from 12 μg/m3 to 9–10 μg/m3 because of growing evidence of health effects at levels lower than the previous regulatory standard.44 Our study provides important additional evidence supporting this proposal. Previous studies,8,16,20 but not all,15 demonstrated clear linear associations at this lower range of PM2.5 levels while the observed range of PM2.5 levels or analytic approaches in all or most of the remainder could have obscured these associations. Therefore, our study provides insights that strengthen the interpretation of the broader literature as generally consistent with PM2.5 as an exposure that increases risk of PD.
Key strengths of our study are that we used large, population-based data and were restricted to incident disease, aspects particularly important for geographic studies designed to inform disease etiology. Many prior studies used other PD-related outcomes that might be affected by PD progression or survival, not just risk. We chose to maximize sensitivity of identification of incident PD to ensure that our work was particularly resilient to selection bias.26 Selection bias might occur in other studies with more restrictive definitions of incident PD if, for example, PD cases from higher vs lower PM2.5 areas are less able to access neurologist care or survive long enough to begin anti-parkinsonian medications, that is, be ascertained as a case. In addition, our results are less vulnerable to biases from exposure measurement error because our zip+4 PM2.5 data serves as a stronger proxy for individual-level exposure compared with environmental studies that use broader units of geography-based exposure assignment. Furthermore, our study leveraged new and innovative geographic information system methods. Geographic approaches offer insight beyond multilevel (cluster) regression modeling by enabling investigators to adjust for confounders within their maps and smooth extreme values by taking into account neighboring values, thus refining local patterns of disease. Another important strength is that, within the context of our population-based sample, we also aimed to be as inclusive as possible with not only case ascertainment but also study eligibility, applying only scientifically essential criteria. This allowed us to include minority, poor, and rural populations—and the PD cases that arise within these populations—that are often missed in epidemiologic studies of PD.45
We found a clear linear dose-response relation between PM2.5 and PD up to at least 13 μg/m3, and others have also observed a linear association across a similar range of PM2.5 levels as in our study.8,16,20 Two of these studies found a similar magnitude of effect, estimating ∼50% greater risk of PD when comparing ∼13 μg/m3 of PM2.5 with the lowest levels of exposure of ∼2 μg/m3.8,16 Within this approximate PM2.5 range, the third, smaller study with local air monitoring data reported a linear association of ∼50% greater magnitude.20 If the PM2.5-PD association is causal, future studies investigating the association between PD and PM2.5 subfractions might demonstrate an even greater effect size and might provide additional insights to PD etiology. A multicountry study in Europe confirms the importance of considering the subfractions of PM2.5.8
Accordingly, one potential explanation for the heterogeneity in PM2.5-PD associations across the many studies conducted to date6-20 is that PM2.5 is not a homogenous exposure, but rather a mixture of chemical components that varies by source. PM2.5 can be produced from a variety of sources, including industrial emissions,8 motor vehicle traffic,19 and farming practices.10 PM2.5 may, therefore, have varied effects on the development of neurodegenerative disease depending on the chemical composition of PM2.5 in different regions. Of particular relevance, studies have identified heavy metals within PM2.5 in the Ohio River Valley, including manganese and zinc linked to iron and steel manufacturing.46,47 In addition to the type of, and proximity to, emission sources, the physical environment may also influence levels of relevant PM2.5 components and, hence, health effects. For example, the S-shaped pattern of high risk in the Mississippi-Ohio River Valley follows the low-elevation valleys along the Appalachian Mountains, suggesting a potential role of regional wind patterns and topography in the relationship between geographic exposures such as PM2.5 and PD. Our methods allowed for the detection of differential effects across the subregions of the United States. In most regions, we detected a strong positive relationship between PM2.5 and PD, whereas in a few regions, there was no detectable effect or there was evidence of inverse associations. These results might inform efforts to identify natural and built environment conditions that could provide some protections against air pollution, including specific climates and urban layouts. However, we cannot rule out the possibility that regional differences could be partly due to gene-environment interaction48 and/or interactions with various nongenetic factors including diet.49 The differential effects across regions emphasize the value of taking a geographic approach in large studies. Thus, we encourage the application of this approach in other parts of the world, including densely populated regions in South Asia where PM2.5 levels can reach ∼60 μg/m3 on average50 (compared with 14 μg/m3 for the contiguous United States at the time of our study).
There are several limitations of this work. Our geographic analyses relied on aggregate data subject to the ecological fallacy. Nevertheless, regression based on individual-level outcome data corroborated our county-level geographic findings. At the same time, we acknowledge that the separate issue of exposure measurement error (likely nondifferential) remains to some extent. Any PM2.5 exposure model is imperfect, even the one we used, which achieved a cross-validated R2 of 0.89.28 In addition, we linked the objective PM2.5 estimates from this model using zip+4 rather than exact addresses. Another limitation of Medicare data is the restriction on coverage to patients 65 years and older, although the PM2.5-PD association did not differ by age. We excluded patients with Medicare Advantage plans from analyses because these plans do not report claims to Medicare, but during the study years, they only represented a quarter of beneficiaries,25 so we anticipate that our results remain generalizable. We also excluded a small percentage of potential cases whose PD incident status was uncertain because of the diagnosis of Lewy body dementia and/or atypical parkinsonism. However, the results of our sensitivity analysis found no difference in the results when Lewy body dementia was included in our case definition. Our results also assume that beneficiaries were nonmobile during the 10 years before diagnosis; we acknowledge that our methods were unable to capture early-life exposures that might be relevant. That said, the overwhelming majority of beneficiaries in our study were nonmobile for at least the 5 years of data which were available to us. Owing to the long PD prodromal period,25 we applied as much exposure lagging as possible and used PM2.5 estimates from up to 10 years before diagnosis. The true period of relevant exposure may be longer, but we were unable to extend our exposure period or explore time windows of exposure given that earlier years of residence data were not available. We also acknowledge that the association between PM2.5 and PD could result, at least in part, from a correlated, environmental exposure, but we ruled out confounding by both pesticides and trichloroethylene. In addition, while we found no evidence that the PM2.5-PD association was stronger in men than women, we cannot rule out confounding by occupational exposures. Finally, we also note that the validity of our data on PD diagnosis requires individuals to obtain medical care and relies on competent diagnoses and data entry by health care providers and staff. A prior study suggested that the case identification method we used in our study had 82.7% sensitivity and 99.7% specificity among primary care patients,26 and through our design and analysis choices, we sought to minimize the potential for selection bias and confounding by use of medical care. Despite limitations, our study advances current knowledge of PM2.5 in relation to PD and demonstrates the utility of using a multimethod geographic approach for investigating environmental risk factors.
Using state-of-the-art geographic analytic techniques, we identified strong regional associations between PM2.5 and PD in the United States. A deeper investigation into the subfractions of PM2.5 in those regions may provide insight into PD risk factors.
Glossary
- μg/m3
- micrograms per cubic meter
- CI
- confidence interval
- GWR
- geographically weighted regression
- LISA
- local indicator of spatial association
- PD
- Parkinson disease
- PM2.5
- particulate matter <2.5 micrometers in diameter
- R-INLA
- Laplace approximation method for Bayesian inference in R-project
- RR
- relative risk
- U.S.
- United States
Appendix Authors
| Name | Location | Contribution |
|---|---|---|
| Brittany Krzyzanowski, PhD | Barrow Neurological Institute, Phoenix, AZ | Drafting/revision of the manuscript for content, including medical writing for content; major role in the acquisition of data; study concept or design; analysis or interpretation of data |
| Susan Searles Nielsen, PhD | Washington University in St. Louis, MO | Drafting/revision of the manuscript for content, including medical writing for content; major role in the acquisition of data |
| Jay R. Turner, DSc | Washington University in St. Louis, MO | Drafting/revision of the manuscript for content, including medical writing for content |
| Brad A. Racette, MD | Barrow Neurological Institute, Phoenix, AZ | Drafting/revision of the manuscript for content, including medical writing for content; major role in the acquisition of data |
Footnotes
Editorial, page 921
CME Course: NPub.org/cmelist
Supplementary Material
References
1.
Shaffer RM, Blanco MN, Li G, et al. Fine particulate matter and dementia incidence in the adult changes in thought study. Environ Health Perspect. 2021;129(8):87001.
2.
Yuan S, Wang J, Jiang Q, et al. Long-term exposure to PM(2.5) and stroke: a systematic review and meta-analysis of cohort studies. Environ Res. 2019;177:108587.
3.
Calderon-Garciduenas L, Solt AC, Henriquez-Roldan C, et al. Long-term air pollution exposure is associated with neuroinflammation, an altered innate immune response, disruption of the blood-brain barrier, ultrafine particulate deposition, and accumulation of amyloid beta-42 and alpha-synuclein in children and young adults. Toxicol Pathol. 2008;36(2):289-310.
4.
Lee CP, Zhu CH, Su CC. Increased prevalence of Parkinson's disease in soils with high arsenic levels. Parkinsonism Relat Disord. 2021;88:19-23.
5.
Willis AW, Evanoff BA, Lian M, et al. Metal emissions and urban incident Parkinson disease: a community health study of medicare beneficiaries by using geographic information systems. Am J Epidemiol. 2010;172(12):1357-1363.
6.
Cerza F, Renzi M, Agabiti N, et al. Residential exposure to air pollution and incidence of Parkinson's disease in a large metropolitan cohort. Environ Epidemiol. 2018;2(3):e023.
7.
Chen H, Kwong JC, Copes R, et al. Living near major roads and the incidence of dementia, Parkinson's disease, and multiple sclerosis: a population-based cohort study. Lancet. 2017;389(10070):718-726.
8.
Cole-Hunter T, Zhang J, So R, et al. Long-term air pollution exposure and Parkinson's disease mortality in a large pooled European cohort: an ELAPSE study. Environ Int. 2023;171:107667.
9.
Jo S, Kim YJ, Park KW, et al. Association of NO2 and other air pollution exposures with the risk of Parkinson disease. JAMA Neurol. 2021;78(7):800-808.
10.
Kirrane EF, Bowman C, Davis JA, et al. Associations of ozone and PM2.5 concentrations with Parkinson's disease among participants in the Agricultural Health Study. J Occup Environ Med. 2015;57(5):509-517.
11.
Kioumourtzoglou MA, Schwartz JD, Weisskopf MG, et al. Long-term PM2.5 exposure and neurological hospital admissions in the Northeastern United States. Environ Health Perspect. 2016;124(1):23-29.
12.
Liu R, Young MT, Chen JC, Kaufman JD, Chen H. Ambient air pollution exposures and risk of Parkinson disease. Environ Health Perspect. 2016;124(11):1759-1765.
13.
Palacios N, Fitzgerald KC, Hart JE, et al. Particulate matter and risk of Parkinson disease in a large prospective study of women. Environ Health. 2014;13:80.
14.
Palacios N, Fitzgerald KC, Hart JE, et al. Air pollution and risk of Parkinson's disease in a large prospective study of men. Environ Health Perspect. 2017;125(8):087011.
15.
Salimi F, Hanigan I, Jalaludin B, et al. Associations between long-term exposure to ambient air pollution and Parkinson's disease prevalence: a cross-sectional study. Neurochem Int. 2020;133:104615.
16.
Shi L, Wu X, Danesh Yazdi M, et al. Long-term effects of PM(2.5) on neurological disorders in the American Medicare population: a longitudinal cohort study. Lancet Planet Health. 2020;4(12):e557-e565.
17.
Shin S, Burnett RT, Kwong JC, et al. Effects of ambient air pollution on incident Parkinson's disease in Ontario, 2001 to 2013: a population-based cohort study. Int J Epidemiol. 2018;47(6):2038-2048.
18.
Toro R, Downward GS, van der Mark M, et al. Parkinson's disease and long-term exposure to outdoor air pollution: a matched case-control study in the Netherlands. Environ Int. 2019;129:28-34.
19.
Yu Z, Wei F, Zhang X, et al. Air pollution, surrounding green, road proximity and Parkinson's disease: a prospective cohort study. Environ Res. 2021;197:111170.
20.
Yuchi W, Sbihi H, Davies H, Tamburic L, Brauer M. Road proximity, air pollution, noise, green space and neurologic disease incidence: a population-based cohort study. Environ Health. 2020;19(1):8.
21.
Wright Willis A, Evanoff BA, Lian M, Criswell SR, Racette BA. Geographic and ethnic variation in Parkinson disease: a population-based study of US Medicare beneficiaries. Neuroepidemiology. 2010;34(3):143-151.
22.
Lux WE, Kurtzke JF. Is Parkinson's disease acquired? Evidence from a geographic comparison with multiple sclerosis. Neurology. 1987;37(3):467-471.
23.
Kurtzke JF, Goldberg ID. Parkinsonism death rates by race, sex, and geography. Neurology. 1988;38(10):1558-1561.
24.
Lanska DJ. The geographic distribution of Parkinson's disease mortality in the United States. J Neurol Sci. 1997;150(1):63-70.
25.
Searles Nielsen S, Warden MN, Camacho-Soto A, Willis AW, Wright BA, Racette BA. A predictive model to identify Parkinson disease from administrative claims data. Neurology. 2017;89(14):1448-1456.
26.
Butt DA, Tu K, Young J, et al. A validation study of administrative data algorithms to identify patients with Parkinsonism with prevalence and incidence trends. Neuroepidemiology. 2014;43(1):28-37.
27.
Di Q, Wei Y, Shtein A, et al. Daily and Annual PM2.5 Concentrations for the Contiguous United States, 1-km Grids, v1 (2000-2016); 2021. Palisades, New York: NASA Socioeconomic Data and Applications Center (SEDAC). Accessed October 10, 2023. https://sedac.ciesin.columbia.edu/data/set/aqdh-pm2-5-concentrations-contiguous-us-1-km-2000-2016
28.
Di Q, Amini H, Shi L, et al. An ensemble-based model of PM(2.5) concentration across the contiguous United States with high spatiotemporal resolution. Environ Int. 2019;130:104909.
29.
Dwyer-Lindgren L, Mokdad AH, Srebotnjak T, Flaxman AD, Hansen GM, Murray CJ. Cigarette smoking prevalence in US counties: 1996-2012. Popul Health Metr. 2014;12(1):5.
30.
Faust I, Warden M, Camacho-Soto A, Racette BA, Searles Nielsen S. A predictive algorithm to identify ever smoking in medical claims-based epidemiologic studies. Ann Epidemiol. 2023;85:59-67.e6.
31.
U.S. Environmental Protection Agency. National Air Toxics Assessment (NATA): 2005 Census Tract Modeled Ambient Concentrations, Exposures, and Risks; 2005. Accessed March 4, 2023 epa.gov/national-air-toxics-assessment/2005-nata-assessment-results.
32.
Kind AJH, Buckingham WR. Making neighborhood-disadvantage metrics accessible - the neighborhood Atlas. N Engl J Med. 2018;378(26):2456-2458.
33.
U.S. Census Bureau. 2009-2011 American Community Survey 3-year Public Use Microdata Samples [SAS Data file]; 2010. Accessed October 20, 2022. factfinder.census.gov/faces/nav/jsf/pages/searchresults.xhtml?refresh=t
34.
U.S. Geological Survey (USGS). Estimated Annual Agricultural Pesticide Use: County-Level Data. National Water-Quality Assessment (NAWQA) Project; Pesticide National Synthesis Project. U.S. Department of the Interior; 2018. Accessed October 18, 2019. water.usgs.gov/nawqa/pnsp/usage/maps/county-level/.
35.
Rue H, Martino S, Chopin N. Approximate Bayesian inference for latent Gaussian models by using integrated nested Laplace approximations. J R Stat Soc Ser B: Stat Methodol. 2009;71(2):319-392.
36.
Reeve A, Simcox E, Turnbull D. Ageing and Parkinson's disease: why is advancing age the biggest risk factor? Ageing Res Rev. 2014;14(100):19-30.
37.
Gillies GE, Pienaar IS, Vohra S, Qamhawi Z. Sex differences in Parkinson's disease. Front Neuroendocrinol. 2014;35(3):370-384.
38.
Branson CO, Ferree A, Hohler AD, Saint-Hilaire M. Racial disparities in Parkinson disease: a systematic review of the literature. Adv Parkinson's Dis. 2016;05(4):87-96.
39.
Nefzger MD, Quadfasel FA, Karl VC. A retrospective study of smoking in Parkinson's disease. Am J Epidemiol. 1968;88(2):149-158.
40.
Gross A, Racette BA, Camacho-Soto A, Dube U, Searles Nielsen S. Use of medical care biases associations between Parkinson disease and other medical conditions. Neurology. 2018;90(24):e2155–e2165.
41.
Anselin L, Syabri I, Kho Y. GeoDa: an introduction to spatial data analysis. Geogr Anal. 2006;38(1):5-22.
42.
Centers for Medicare and Medicaid Services (CMS). Research Identifiable Files (RIFs). Centers for Medicare and Medicaid Services (CMS); 2009.
43.
Silver MR, Racette BA, Dube U, Faust IM, Searles Nielsen S. Well water and Parkinson's disease in medicare beneficiaries: a Nationwide Case-Control Study. J Parkinsons Dis. 2020;10(2):693-705.
44.
U.S. Environmental Protection Agency. National Ambient Air Quality Standards (NAAQS) for PM: Rule Summary; 2023. Accessed February 16, 2023 epa.gov/pm-pollution/national-ambient-air-quality-standards-naaqs-pm#rule-summary.
45.
Willis AW, Schootman M, Evanoff BA, Perlmutter JS, Racette BA. Neurologist care in Parkinson disease: a utilization, outcomes, and survival study. Neurology. 2011;77(9):851-857.
46.
Vedantham R, Landis MS, Olson D, Pancras JP. Source identification of PM2.5 in Steubenville, Ohio using a hybrid method for highly time-resolved data. Environ Sci Technol. 2014;48(3):1718-1726.
47.
Kim M, Deshpande SR, Crist KC. Source apportionment of fine particulate matter (PM2.5) at a rural Ohio River Valley site. Atmos Environ. 2007;41(39):9231-9243.
48.
Lee PC, Raaschou-Nielsen O, Lill CM, et al. Gene-environment interactions linking air pollution and inflammation in Parkinson's disease. Environ Res. 2016;151:713-720.
49.
Chen C, Hayden KM, Kaufman JD, et al. Adherence to a MIND-like dietary pattern, long-term exposure to fine particulate matter air pollution, and MRI-based measures of brain volume: the Women's Health Initiative Memory Study-MRI. Environ Health Perspect. 2021;129(12):127008.
50.
van Donkelaar A, Martin RV, Brauer M, et al. Global estimates of fine particulate matter using a combined geophysical-statistical method with information from satellites, models, and monitors. Environ Sci Technol. 2016;50(7):3762-3772.
Information & Authors
Information
Published In
Copyright
Copyright © 2023 The Author(s). Published by Wolters Kluwer Health, Inc. on behalf of the American Academy of Neurology. This is an open access article distributed under the terms of the Creative Commons Attribution License 4.0 (CC BY), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Publication History
Received: July 6, 2023
Accepted: August 3, 2023
Published online: November 21, 2023
Published in issue: November 21, 2023
Disclosure
S. Searles Nielsen receives research support from the following government and nongovernmental organizations: US National Institutes of Health (NIH)—National Institute of Environmental Health Sciences (NIEHS) (R01ES029524, R01ES026891, R01ES026891-S1, R01ES025991, R01ES025991-S1, K01ES028295), the US Department of Defense (PD190057), and Cure Alzheimer's Fund, Michael J. Fox Foundation (MJFF) (000939) (020718). J.R. Turner receives research support from the Michael J. Fox Foundation (MJFF) (000939). B.A. Racette receives research support from the following government and non-governmental organizations: Michael J Fox Foundation (MJFF) (000939) (020718), National Institute of Environmental Health Sciences (NIEHS) (R01ES025991, R01ES025991-02S1, R01ES030937-S1,R01ES029524), National Institute of Occupational Safety and Health (NIOSH) (R01OH011661), Cure Alzheimer's Fund, Department of Defense (PD190057), Hope Center for Neurologic Disorders (Washington University). B.A. Racette has received honoraria (personal compensation) for service on the National Advisory Environmental Health Sciences Council for NIEHS. Go to Neurology.org/N for full disclosures.
Study Funding
This study was supported by the Michael J. Fox Foundation for Parkinson's Research (MJFF000939), the US Department of Defense (PD190057), and the National Institutes of Health National Institute of Environmental Health Sciences (K01ES028295).
Authors
Metrics & Citations
Metrics
Citation information is sourced from Crossref Cited-by service.
Citations
Download Citations
If you have the appropriate software installed, you can download article citation data to the citation manager of your choice. Select your manager software from the list below and click Download.
Cited By
- Sleep and Arousal Hubs and Ferromagnetic Ultrafine Particulate Matter and Nanoparticle Motion Under Electromagnetic Fields: Neurodegeneration, Sleep Disorders, Orexinergic Neurons, and Air Pollution in Young Urbanites, Toxics, 13, 4, (284), (2025).https://doi.org/10.3390/toxics13040284
- Cropland associated with risk of Parkinson's disease in the northern Great Plains, Parkinsonism & Related Disorders, 132, (107288), (2025).https://doi.org/10.1016/j.parkreldis.2025.107288
- Parkinson's Disease Detection Using Deep Learning, 2024 First International Conference on Data, Computation and Communication (ICDCC), (186-191), (2024).https://doi.org/10.1109/ICDCC62744.2024.10961363
- Environmental Risk Factors for Parkinson's Disease: A Critical Review and Policy Implications, Movement Disorders, 40, 2, (204-221), (2024).https://doi.org/10.1002/mds.30067
- The effect of air pollution on hospitalizations with Parkinson’s disease among medicare beneficiaries nationwide, npj Parkinson's Disease, 10, 1, (2024).https://doi.org/10.1038/s41531-024-00815-x
- Particulate Matter-Induced Emerging Health Effects Associated with Oxidative Stress and Inflammation, Antioxidants, 13, 10, (1256), (2024).https://doi.org/10.3390/antiox13101256
- Reader Response: Fine Particulate Matter and Parkinson Disease Risk Among Medicare Beneficiaries, Neurology, 103, 7, (2024)./doi/10.1212/WNL.0000000000209911
- Author Response: Fine Particulate Matter and Parkinson Disease Risk Among Medicare Beneficiaries, Neurology, 103, 7, (2024)./doi/10.1212/WNL.0000000000209412
- Reader Response: Fine Particulate Matter and Parkinson Disease Risk Among Medicare Beneficiaries, Neurology, 103, 7, (2024)./doi/10.1212/WNL.0000000000209374
- Author Response: Fine Particulate Matter and Parkinson Disease Risk Among Medicare Beneficiaries, Neurology, 103, 7, (2024)./doi/10.1212/WNL.0000000000209334
- See more
Loading...
View Options
View options
PDF and All Supplements
Download PDF and Supplementary MaterialShort Form
Short FormFull Text
View Full TextLogin options
Check if you have access through your login credentials or your institution to get full access on this article.
Personal login Institutional LoginPurchase Options
The neurology.org payment platform is currently offline. Our technical team is working as quickly as possible to restore service.
If you need immediate support or to place an order, please call or email customer service:
- 1-800-638-3030 for U.S. customers - 8:30 - 7 pm ET (M-F)
- 1-301-223-2300 for customers outside the U.S. - 8:30 - 7 pm ET (M-F)
- [email protected]
We appreciate your patience during this time and apologize for any inconvenience.
We appreciate the interest of You et al. in our work1 and agree that historical and modern institutions shape communities in ways that warrant further exploration. In our study, we adjusted for socioeconomic status in the form of median household income, area deprivation, and access to health care. Nevertheless, we agree that incorporating a broader set of social determinants of health variables into environmental studies is necessary to better understand the impact of environmental pollutants on the most vulnerable populations.
Reference
Author disclosures are available upon request ([email protected]).
We read with great interest the recent article by Krzyzanowski et al., “Fine Particulate Matter and Parkinson Disease Risk Among Medicare Beneficiaries,” in which they highlight a key association between environmental exposures and neurologic health outcomes.1 Using population-based US Medicare data, they identified a positive, region-specific association between exposure to particulate matter <2.5 micrometers (PM2.5) and risk of Parkinson disease (PD).1 These findings join a growing body of evidence establishing a link between environmental exposures and PD risk.2
Modern society is such that our zip code may hold implications for health and life expectancy. As we contend with the longitudinal impact of environmental pollutants, we must remain mindful of the historical and modern institutions that have shaped the communities within the geographic regions where such pollutants are the most prevalent. Socioeconomic status, race, and ethnicity are some factors associated with exposure to specific environmental toxins.3-5 While identifying “hotspots” of pollution is critical to understanding environmental determinants of brain health, particularly PD, we must also address the unique needs of these vulnerable communities. Only by bridging knowledge between environmental exposures and socioeconomic inequities can we begin to take steps toward remediation through public health strategies.
References
Author disclosures are available upon request ([email protected]).
We appreciate Dr. Brenner's interest in our recent article “Fine Particulate Matter and Parkinson Disease Risk Among Medicare Beneficiaries.”1 We agree that nanoplastic contamination is an intriguing potential environmental neurotoxicant that warrants further exploration with regard to PD pathogenesis. Of note, Costa-Gómez et al. speculate that the source of polystyrene was “local agricultural practices,”2 and some of the PD hotspot regions align with areas with high agricultural activity. Ultimately, studies in human populations will be critical to demonstrate a causal association between PD and nanoplastics.
References
Author disclosures are available upon request ([email protected]).
The study by Krzyzanowski et al. regarding Parkinson disease (PD) risk associated with fine particulate matter (PM 2.5) was interesting with reference to PD pathogenesis.1 Increasing nanoplastics contamination in the environment, including in the PM2.5 (alveolar fraction), specifically polystyrene, has been found to induce alpha-synuclein (AS) fibrillization, while in mice, nanoplastics induced alpha-synuclein inclusions in dopaminergic neurons of the substantia nigra.2 Other types of plastic waste such as polycarbonates, polyethylenes, or polypropylene could also interact with AS, but further studies would be necessary to decide.2 Quantification of the concentration of the polystyrene (PS) in the PM2.5 in a medium-sized coastal city in Spain was 1.8 ng/m-3.3 It was thought the main source of the PS was "local agricultural practices."3 The PD hot spot, 118 counties North of the Mississippi-Ohio river confluence identified in the study,1 also appears to be a region of extensive agriculture, so there could be some relationship to agricultural practices producing microplastic contamination in the PM2.5. Microplastics are widespread in the environment and may be implicated in development of neurodegenerative diseases such as Parkinson disease.