Parkinson’s disease (PD) is the second most common neurodegenerative disorder after Alzheimer’s disease.1 The etiology of PD is still unknown, and it seems the result of the interplay of genetic and environmental factors.2 Age is the most important risk factor, and given the increasingly aged population, PD incidence and prevalence will have major public health implications.
Parkinson’s disease tend to be more frequent in men than in women (with a male to female ratio in the range from 1.3 to 2) suggesting both the presence of an hormonal component and lifestyle risk factors that have a different distribution in men and in women.3 Consumption of dairy products, exposure to some pesticides, and a history of melanoma are well-documented risk factors.3 Prior head injuries have been associated to PD in several studies, but the possibility of reverse causation is still debated. Other factors have been reported but are considered not well established such as adipose distribution, β-blocker use, well-water consumption, use of postmenopausal estrogens, infections, and early life factors.3 There are also factors associated with a decreased risk of PD like physical activity, uric acid, smoking,4,5 caffeine intake,6 moderate alcohol consumption, or the use of anti-inflammatory drugs.2 Some studies have reported an inverse association with calcium channel blockers, statins, flavonoids, and healthy dietary patterns.3,7,8 Few studies have analyzed the association between PD and long-term exposure to air pollution, and they had conflicting results.9–18 Only three studies investigated the association between PD and ozone exposure.14,16,19
It is recognized that older brains are more vulnerable to proinflammatory stimuli, and one of the possible mechanisms of air pollution’s adverse effects on brain health is oxidative stress.20 A 2012 workshop on the topic recommended further research on the effects of air pollution on the central nervous system and indicated the need to evaluate whether the effects occur after cardiovascular damage or whether they are independent.21
The aim of the present study is to evaluate the association of air pollution exposure to several air pollutants, that is, particulate matter (with a diameter of <10 μm: PM10, <2.5 μm: PM2.5, and 2.5–10 μm: PM2.5–10), PM2.5 absorbance as proxy of elemental carbon, nitrogen oxides (NOx), nitrogen dioxide (NO2) and summer ozone (O3), and the incidence of Parkinson’s disease. We conducted the investigation using the established dataset of the Rome Longitudinal Study22,23 and the exposure assessment used within the European study of cohorts for air pollution effects (ESCAPE) project.24,25
The data are from the Rome Longitudinal Study, the 2001 census cohort of Rome ascertained from the Municipal Register.22,23 It includes all residents on the census reference day (21 October 2001) who were not living in institutions (prisons, hospitals, or nursing homes). Data were available on gender, age, birthplace, marital status, education, and occupation. Through a record linkage with Municipal Register data, using a personal identifier under strict control to protect individual privacy, data were available on residential history from 1996. In addition, the date in which the subject started to live at the baseline address was available. Vital status was available as of 31 December 2013.
We followed the subjects of the cohort for the specific period from 1 January 2008 to 31 December 2013 because the database of pharmaceutical prescriptions (see below) were available only since 2006. We selected subjects free from Parkinson’s disease (see below), who are 50+ years of age at inclusion or who turned 50 during the follow-up period (2008–2013). We followed the subjects until 31 December 2013, or 100 years of age, death, migration from Rome, or occurrence of Parkinson’s disease, whichever came first.
Sources of health data
We used data from the Regional Health Information System (HIS) which covers both public and private healthcare providers. The HIS includes the Hospital Discharge Registry (HDR), the Exempt from Copays Registry (ECR), and the Regional Drug Registry (RDR) database. The HDR collects discharge data from all hospitals with up to six diagnoses and procedural codes International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) for each record. The ECR includes data on all residents who qualified for free healthcare services for specific conditions, for example, disability, chronic diseases including Parkinson’s, low income, or old age. The RDR database includes all individual records for each drug prescription dispensed by general practitioners, by hospitals at discharge, and by specialist outpatient clinics. Drugs are identified through the national drug register code, which refers to the International Anatomical Therapeutic Chemical Classification System. Every prescription reports individual patient information and drug-dispensing dates. Each patient had an identification code and was linked to the Rome Longitudinal Study.
Case definition of Parkinson’s disease
To identify people affected by PD, we used the HDR, the ECR, and the RDR. From the HDR, we selected all subjects who were hospitalized from 2001 to 2013 with a primary or a secondary diagnosis of PD (ICD-9-CM: 332.0) (note: we excluded a diagnosis of Parkinsonism—ICD-9-CM: 332.1—in any diagnostic position). From the ECR, we selected all subjects registered with the specific PD code (038.332) from 2005 to 2013. Finally, from the RDR, we selected all patients with at least two anti-Parkinson pharmacy claims in 1 year: levodopa or dopamine agonists (pramipexole, ropinirole, and pergolide) or Catechol-O-methyltransferase inhibitors (entacapone) (see eTable 1; http://links.lww.com/EE/A15 for ATC codes); the RDR was available from 2006 to 2013.
In order to exclude prevalent cases from the analysis of incidence, we excluded subjects identified as PD cases before 2008 and identified new PD cases as those at the first selection in the Regional Information System (HDR, ECR, RDR, whichever came first) starting the 1 January 2008 (to 2013).
We used the Land Use Regression (LUR) models developed for the city of Rome within the ESCAPE project to assess exposure to particulate matter (<10 μm: PM10, <2.5 μm: PM2.5, and 2.5–10 μm: PM2.5–10), PM2.5 absorbance, and nitrogen oxides (NOx and NO2).24,25 The models and the measurement campaigns are described in detail elsewhere.24–27 Briefly, during 2010, nitrogen oxides were measured in 40 sites, whereas particulate matter and PM2.5 absorbance were measured in 20 sites. The sites were chosen to represent the spatial distribution of residential addresses, following a strict protocol.26,27 Nitrogen oxides were measured using passive samplers, whereas particulate matter was measured using Harvard impactors. PM2.5 absorbance, as proxy elemental carbon,28 was obtained as a result of the transformation of reflectance measurements on PM2.5 filters, according to the International Standardization Organization.27 Measurements were taken in three 2-week periods and were averaged using a continuous background monitoring site that operated the entire year to estimate an annual average measure of pollutants’ concentration at each site.24,25 Several traffic and land use variables were used to predict pollutants’ concentrations in the LUR models using strict protocols. The models’ R2 ranged from 71% (PM2.5) to 86% (NO2) and a cross-validated R2 from 59% (PM10) to 79% (PM2.5 absorbance). The equations from the LURs were applied to the coordinates of all the residential addresses of the cohort members on 1 January 2008.
We used a chemical dispersion model to estimate daily (8 hours) summer (May to September) ozone exposure. The Flexible Air quality Regional Model, a three-dimensional Eulerian model of the transport and multiphase chemistry of pollutants in the atmosphere, was applied to the city of Rome with a grid resolution of 1 × 1 km, based on the year 2005 measurements (see eFigure 1; http://links.lww.com/EE/A15).29 We estimated exposure to each pollutant at subjects’ address on 1 January 2008.
We calculated crude and standardized incidence rates of PD per 100,000 person-years. We used as standard population the 2008 Italian population (www.demo.istat.it) by sex and age group (50–64; 65–74; 75–84; 85+).
To estimate the association between air pollution exposure and the incidence of Parkinson’s disease, we used a Cox proportional hazards regression model, with age as the time scale. We tested the proportional hazards assumption using the Schoenfeld test, and we stratified the baseline hazard function by sex, the only variable that did not satisfy the proportionality assumption.
We considered several variables as potential confounders or effect modifiers: sex, age, level of education (primary school, junior high school, high school, and university), marital status (married, single, separated/divorced, and widowed), place of birth (Rome or other places), and a small-area (census block, average 500 inhabitants) index of socioeconomic position (SEP).30 The small-area index of SEP is a composite indicator based on 2001 census data. Briefly, it was obtained performing a factor analysis using variables that represent various dimensions of deprivation (education, occupation, family composition, housing tenure, and migration). The indicator categorized the census blocks of Rome, according to the quintiles of its distribution, into five levels (very high, high, medium, low, and very low).30
We estimated hazard ratios (HRs) of the association between each pollutant (PM10, PM2.5–10, PM2.5, PM2.5 absorbance, NO2, NOx, summer O3) and incidence of Parkinson’s disease. We calculated HRs adjusted for all the variables indicated above. We considered air pollutants as continuous variables. To be able to compare the results with previous studies based on ESCAPE project LUR models, we used different fixed increases to evaluate the association with the outcome: 10 μg/m3 for exposure to PM10, NO2, and Summer O3, 5 μg/m3 for PM2.5 and PM2.5–10, 20-μg/m3 for NOx, and an increase of 1 unit (10−5/m) for PM2.5 absorbance. In addition, to study the shape of the relationship between air pollution exposure and PD incidence, we used splines with 2 knots.31 To investigate a possible nonlinear association, we performed a likelihood ratio test between the model with and without spline.
To explore the possible effect modification by sex and age, we performed the likelihood ratio test to compare the goodness of fit of the models with or without an interaction term. In addition, we have considered the possibility to use previous cardiovascular disease (CVD) hospital admissions as a possible effect modifier to evaluate whether the effect of air pollution on PD was lower or higher in patients with preexisting CVD; this approach was followed by Chen and colleagues.15 However, adjusting for CVD, or stratifying for it, could introduce a collider bias.32,33 This is due to the fact that exposure to air pollution and smoking are both risk factors for cardiovascular diseases34,35 and smoking can also be inversely associated with PD.36,37 Therefore, adjusting for cardiovascular diseases (the collider) can induce a false, strengthened, or reversed association between air pollution and PD.38 Thus, we performed an additional analysis changing the case definition to include subjects who had previous acute cardiovascular events, that is, subjects who had at least one hospitalization for acute myocardial infarction (ICD-9-CM: 410 or 412 in any diagnostic position) or stroke (ICD9-CM: 431, 433.x1, 434, 436, 438 in any diagnostic position) before PD (“PD with CVD hospitalization”). As an alternative case definition, we considered also cases of PD that had at least three cardiovascular drug prescriptions (Antihypertensive, diuretics, β-blocking agents, calcium channel blockers, and agents acting on the renin–angiotensin system) in the year before identification of PD (“PD with CVD drug prescriptions”).
To investigate the role of different data sources in our results, we performed a sensitivity analysis limiting cases from the HDR only, from the ECR only, and then from the RDR database only, respectively.
The choice of studying the population from 1 January 2008 was driven by the availability of drug prescriptions from 2006. This implicated the selection, as the study population, of subjects included in the Rome Longitudinal Study who were alive and still residents in Rome as of 1 January 2008. This selection could have introduced a selection bias. To investigate this possibility, we used an inverse probability weighting approach.39,40 We used a logistic regression model to predict the probability of being alive and resident in Rome on the 1 January 2008 considering the following variables in the model: sex, age, age squared, marital status, educational level, and birthplace. We calculated the weights as the inverse of the probability of being included. We stabilized the weights and we used them in a sensitivity analysis to take into account the different characteristics of the selected population compared with the subjects included in the original Rome Longitudinal Study.
To account for the possible bias due to deaths that occurred before PD onset, we performed a competing risks analysis using the Fine and Grey method, with death as competing event.41
In addition, we restricted our population to subjects who are 50+ years of age at inclusion and did not include subjects who turned 50 during the follow-up.
To investigate a possible exposure misclassification bias, due to a long latency period for PD onset, we performed a subgroup analysis on long-term residents, that is, subjects who were living at the baseline address since 1986, and a sensitivity analysis using the exposure assessment on 2001 addresses.
Only study participants with complete exposure and confounder information in the main model were included in all analyses. All analyses were conducted using STATA13 (StataCorp. 2013; Stata Statistical Software: Release 13; StataCorp LP, College Station, TX).
After excluding prevalent Parkinson’s cases (N = 9,460, 1%), subjects with missing exposures due to lack of geocoding of some addresses (N = 20,170, 2%), and those with information missing on confounders (12 subjects), we selected 1,008,253 subjects.
Table 1 shows the characteristics of the study population. It was composed by 443,956 males (44%) and 564,297 females (56%). The mean age of the study population at the beginning of the follow-up was 63 years (SD = 12). During the follow-up, we identified 13,104 new cases of Parkinson’s disease, 6,222 males and 6,882 females. The majority of the cases (87.5%) were selected from the drug prescriptions data, 10.3% of incident cases were selected from the ECR, and 2.3% of the incident cases were from the Hospital Discharge Registry. The standardized incidence rate (by sex and age) was 298 per 100,000 person-years; it was higher in males than in females, and did not vary across socioeconomic groups of the population.
Table 2 shows the average annual exposures of the study population, with mean, standard deviation, and interquartile range. The correlations among the annual exposures were relatively high between particulate matter and nitrogen oxides (ranging from 0.52 [NOx−PM2.5 absorbance] to 0.70 [PM2.5–10−NO2]), while they were low between O3 and other pollutants (ranging from −0.13 [NOx−O3] to −0.02 [PM2.5−O3]) (see eTable 2; http://links.lww.com/EE/A15).
Table 3 shows the association (HRs) between residential exposure to air pollutants and incidence of Parkinson’s disease. There was no evidence of association between exposure to PM10 and incidence of PD. There was some evidence of a negative association between PD incidence and both PM2.5 and PM2.5–10. There was a negative association between traffic-related air pollutants (PM2.5 absorbance, NO2, and NOx) and the incidence of PD. We found an HR = 0.94 (95% confidence interval [CI] = 0.91, 0.98) for each 10−5/m increase in PM2.5 absorbance, HR = 0.97 (95% CI = 0.96, 0.99) for each 10 μg/m3 increase in NO2, and HR = 0.97 (95% CI = 0.96, 0.98) for each 20 μg/m3 increase in NOx. On the contrary, we found a positive association between exposure to ozone and PD incidence with an HR = 1.02 (95% CI = 1.00, 1.05) for each 10 μg/m3 increase in O3 at residence. The association with ozone remained when we considered confounding by NO2 in the same model (HR = 1.02; 95% CI = 0.99, 1.04, for 10 μg/m3 increase in O3), although with larger confidence intervals.
eFigure 3; http://links.lww.com/EE/A15 reports the spline curves for those pollutants for which the likelihood ratio test indicated a departure from linearity (O3, NO2, PM2.5 absorbance).
Table 4 shows the results of the effect modification analyses by sex and age. We found a lower risk of PD for increasing exposure to PM (PM10, PM2.5–10, PM2.5) in women and not in men. There was no evidence of effect modification by age, nor by educational level and small-area socioeconomic position (not shown).
We found similar results in the additional analyses we made. When we considered 1,513 PD incident cases with cardiovascular conditions (PD with CVD) as the outcome, the association with air pollution exposure was in the same direction of those reported in the main results (see eTable 3; http://links.lww.com/EE/A15). The results were similar when we analyzed subjects who had at least three prescriptions for drugs for the cardiovascular system during the year before identification of PD (see eTable 4; http://links.lww.com/EE/A15). Similarly, when we selected new cases only from the Hospital Discharge Registry (3,872 new cases during the follow-up), or when we selected new cases using the Regional Drug Registry (12,706 incident cases), or using the Exempt from Copays Registry (1,513 incident cases), we found results comparable to the main analysis (see eTable 5; http://links.lww.com/EE/A15). Moreover, we found similar results when we added in the model an inverse probability weighting to consider possible differences from the source population (see eTable 6; http://links.lww.com/EE/A15). The same we found when we considered possible competing risks (see eTable 7; http://links.lww.com/EE/A15) and subpopulation with only subject 50+ years of age at inclusion (see eTable 8; http://links.lww.com/EE/A15). Similarly, the additional analysis performed on long-term residents (subjects who were living in the baseline address since 1986) showed the same findings (see eTable 9; http://links.lww.com/EE/A15), as those performed using the exposure models at 2001 address (see eTable 10; http://links.lww.com/EE/A15).
Although we found a suggestion of a positive association between residential exposure to ozone and incidence of PD, the most striking results of our study were the negative association between long-term exposure to NO2, NOx, soot (absorbance of PM2.5), and incidence of Parkinson’s disease in the population of Rome. The results were robust after several sensitivity analyses, and the findings suggest that previous cardiovascular diseases do not play a role in the association between air pollution and PD.
The effects of ambient air pollution on mortality and morbidity for cardiovascular diseases, lung cancer, and coronary events are well known.23,34,42–47 Conversely, only a few studies (see Table 5 for a review) have investigated the possible association between air pollution exposure and Parkinson’s disease, making compelling the rationale for the present study.9–19
We found a 2% higher risk of Parkinson’s disease (95% CI = 1.00, 1.05) for each 10 μg/m3 increase in ozone exposure. Currently, only three studies have investigated a possible role of ozone exposure. The first study, conducted in North Carolina and in Iowa, suggested a possible association between ozone exposure and PD,19 although the authors found an association in North Carolina but not in Iowa. The other two studies, both conducted in Taiwan, showed no association between ozone exposure and PD.14,16 Animal studies found that ozone exposure is associated with cumulative damage to the brains of rodents, and ozone inhalation seemed to produce impaired nigral cell morphology and loss of dopamine neurons in rats.48–50 In humans, ozone activates proinflammatory genes,51 and both acute and long-term impacts on mortality have been reported.52,53 A study in Mexico City found a relationship between exposure to concentrations of air pollutants and neuroinflammation and innate immune responses in crucial brain target anatomical areas in children and young adults, which could have a causative role in Parkinson’s disease.54 One study in Taipei showed that two major traffic-related pollutants impacting these effects were sulfate and ozone.55
In contrast with our results of a negative association between exposure to particulate matter and PD in women, other studies did not find evidence of an association between PM10 exposure and PD. Palacios and colleagues11,18 followed a prospective cohort of 121,700 female nurses 30–55 years of age, and they found no evidence that exposure to air pollution (PM10 and PM2.5) is a risk factor for PD and limited evidence for the association between exposure to metals and risk of the disease. Also Liu and colleagues,10 in their case–control study in the United States, did not observe a statistically significant association between exposure to particulate matter and Parkinson’s disease. On the contrary, Kioumourtzoglou and colleagues,9 in a cohort of subjects living in 50 cities in the northeastern United States, found strong evidence of an association between PM2.5 exposure and first hospitalization for Parkinson’s diseases, HR = 1.08 (95% CI = 1.04, 1.12) per 1 μg/m3 increase in annual PM2.5 concentrations, using city-wide annual concentrations. Chen and colleagues,15 in a cohort of subject in Canada, found an HR = 1.03 (95% CI = 1.02, 1.05) per 3.4 μg/m3 increase in annual PM2.5 concentrations. In a nested case–control study in Taiwan, Chen and colleagues16 found an odds ratio (OR) = 1.01 (95% CI = 1.00, 1.02) per 1 μg/m3 increase in PM10 concentrations. Lee and colleagues,14 in a case–control study in Taiwan, showed a protective association OR = 0.93 (95% CI = 0.90, 0.96) between PM10 exposure and incidence of Parkinson’s diseases. When, in the sensitivity analysis, we considered only first hospitalizations, we could not find any evidence of an association with PM2.5 (HR = 0.93; 95% CI = 0.85, 1.01). It is worth noting that besides the overall meta-analytic result, among the 50 US cities they considered, the association was negative in three cities, and not statistically significant in 29 locations.9 While we did not find any association between particulate matter exposure and PD in men, the recent study on male health professionals in the United States17 found a negative association with hazard ratios below one (HR = 0.98; 95% CI = 0.97, 0.99; for 10 μg/m3 increase in PM10). Because the mechanisms through which air pollution affects health (inflammation, oxidative stress, etc.) are similar in men and women,56 we do not have an explanation for the different results we obtained for particulate matter in the two genders.
In contrast to previous case–control studies,10,12,13 we found a statistically significant negative association between exposure to NO2, NOx, and incidence of Parkinson’s disease, not explained by the known inverse relationship between ozone and nitrogen oxides. We estimated a 3% decrease in the risk of incidence of Parkinson’s (95% CI = 0.96, 0.99) for each 10 μg/m3 increase in NO2 and a 3% decrease (95% CI = 0.96, 0.99) for each 20 μg/m3 increase in NOx. On the contrary, the studies carried out in Canada,15 in Taiwan,16 and in the United States10 did not find any association with traffic exposure and NO2, and a case–control study in Denmark showed a 9% higher risk (95% CI = 3%, 16%) per interquartile range increase in NO2 exposure.12 As regards NOx exposure, the Taiwan study did not show any association,16 while in the Denmark study a positive association was observed (OR = 1.06; 95% CI = 1.02, 1.11, per interquartile range increase).12
Generally, epidemiological studies on the effects of air pollution find associations with health outcomes for which smoking is an important risk factor (cardiovascular disease, in particular ischemic heart disease, lung cancer, and respiratory disease). In Parkinson’s disease, however, smoking has been reported as a protective factor. In a systematic review and meta-analysis of PD and cigarette smoking carried out by Breckenridge and colleagues,36 a protective role of smoking was found in the onset on PD, with an overall relative risk (RR) = 0.41 (95% CI = 0.34, 0.48). Two studies confirmed the protective effect of both active and passive smoking,37,57 while one study found a protective effect of active smoking, but did not find evidence of an association for passive smoking.58 The mechanism of the protective effect of smoking is attributed to the role of nicotine receptors; several studies showed that nicotine reduced the effect of toxic insults and protected against nigrostriatal damage in Parkinsonian monkey, rat, and mouse models.59–61 However, given the only modest amount of urinary cotinine levels in passive smokers, there might be other factors, in addition to nicotine, that influence the protective effect of smoke on PD incidence. However, it has been hypothesized that smoking cessation could be part of the preclinical PD, with a decreased responsiveness to nicotine in this phase of the disease.3
This is the first study that found an inverse association between exposure to nitrogen oxides and the incidence of Parkinson’s disease, and similar results were reported by only one study for particulate matter; thus, further research is needed. This work has several strengths. We used a very large cohort to explore the effects of air pollution on the population. We had several individual characteristics, taken from the 2001 census, and we estimated different pollutants (PM10, PM2.5–10, PM2.5, PM2.5 absorbance, NO2, NOx, summer O3) at the residential addresses.
This study also has several limitations. We used a cohort based on administrative data, and information on important risk factors for Parkinson’s disease such as diet and past occupational exposures was not available. We adjusted for two measures of socioeconomic status: the individual level of education, and the census-block socioeconomic position, which are strongly related to diet and occupation; however, residual confounding cannot be ruled out.
For 6,984 people in this cohort, information about smoking habits was available from an ancillary dataset.23 We found no statistically significant association between smoking and several pollutants, that is, PM2.5, PM2.5 absorbance, NO2, NOx, summer O3. We found slightly less risk of being a smoker with increasing level of PM10 and PM2.5–10, but this marginal association is unlikely to affect our results (see eTable 11; http://links.lww.com/EE/A15). We excluded from the study subjects with missing exposure assessment and we considered losses to follow-up those who migrate from Rome in the study period. Since the characteristics of these subjects were similar to those included in the study, it is unlikely we have introduced a bias. We can also exclude a possible attrition bias.32 In fact, we did not find differences in main characteristics (sex, socioeconomic status, level of education, and exposure to air pollution) between the population at the inclusion of this study (1 January 2008) and the 2001 census cohort of Rome (Rome Longitudinal Study), where the study population derived from. Similar results were found when we added in the model an inverse probability weighting to consider these possible differences.
A possible bias is due to the probable underestimation of identification of Parkinson’s disease cases because we used administrative data instead of clinical information. Although health administrative databases in the Lazio region are complete and of good quality,62–64 and although we used a combination of multiple sources of data, only cases diagnosed and recorded in administrative databases could be included in our study. Using administrative datasets, we identified the date of pathology onset as the date recorded in the dataset. In PD it is difficult to establish an exact date of onset, and likely we overestimated the number of person-years free from disease. Although we cannot exclude these possible misclassification biases, we believe that these biases would have been nondifferential between the group of high exposed and the group of low exposed to air pollution.
An important limitation is the exposure assessment. The prodromal phase of PD can last over 20 years, and we used land use regression models based on air pollution measurement campaigns implemented in 1 year (2010), and a dispersion model for ozone based on 2005 data. However, for nitrogen dioxide, we demonstrated that the spatial distribution of concentrations did not change in Rome over a period of more than 10 years,65 we can assume that the same phenomenon occurred in the past. For exposure to summer ozone, given the stability of built environment in Rome and its local meteorology, we believe that the spatial distribution of summer ozone is nearly constant over time. To investigate the possible exposure misclassification we analyzed the subgroup of people who did not change their residence address since 1986, obtaining similar findings.
Overall, we found some evidence that long-term exposure to ozone was positively associated with incidence of PD in the population of Rome, while long-term exposure to traffic-related pollutants (PM absorbance and nitrogen oxides) was negatively associated. We were not able to find a plausible explanation of the negative associations based on the data we have and the issue is to be explored in more detail. Because there are few studies on the association between ozone and PD, and only three cohort studies on the topic, further research should explore the association between air pollution and Parkinson’s disease in different settings.
Conflicts of interest statement
The authors declare that they have no conflicts of interest with regard to the content of this report.
The authors thank Margaret Becker for her help in editing the manuscript
Data and code availability: The Rome Longitudinal Study dataset, due to national privacy policies, cannot be shared. The scripts written to analyze the data would not themselves be informative without these data and therefore also not supplied.
1. De Lau LML, Breteler MMB. Epidemiology of Parkinson’s disease
2. Kalia LV, Lang AE. Parkinson’s disease
3. Ascherio A, Schwarzschild MA. The epidemiology of Parkinson’s disease
: risk factors and prevention.Lancet Neurol201615121257–1272
4. Ritz B, Ascherio A, Checkoway H, et al. Pooled analysis of tobacco use and risk of Parkinson disease.Arch Neurol2007647990–997
5. Sugita M, Izuno T, Tatemichi M, Otahara Y. Meta-analysis for epidemiologic studies on the relationship between smoking and Parkinson’s disease
.J Epidemiol200111287–94Available at: http://www.ncbi.nlm.nih.gov/pubmed/11388498
. Accessed 13 May 2016
6. Costa J, Lunet N, Santos C, Santos J, Vaz-Carneiro A. Caffeine exposure and the risk of Parkinson’s disease
: a systematic review and meta-analysis of observational studies.J Alzheimers Dis201020(1 supplS221–S238
7. Allen MT, Levy LS. Parkinson’s disease
and pesticide exposure--a new assessment.Crit Rev Toxicol2013436515–534
8. Pezzoli G, Cereda E. Exposure to pesticides or solvents and risk of Parkinson disease.Neurology201380222035–2041
9. Kioumourtzoglou M-A, Schwartz JD, Weisskopf MG, et al. Long-term PM2.5 Exposure and Neurological Hospital Admissions in the Northeastern United States.Environ Health Perspect2016124123–29
10. Liu R, Young MT, Chen J-C, Kaufman JD, Chen H. Ambient air pollution
exposures and risk of Parkinson disease.Environ Health Perspect2016124111759–1765
11. Palacios N, Fitzgerald KC, Hart JE, et al. Particulate matter and risk of Parkinson disease in a large prospective study of women.Environ Health20141380
12. Ritz B, Lee P-C, Hansen J, et al. Traffic-related air pollution
and Parkinson’s Disease
in Denmark: a case-control study.Environ Health Perspect20161243351–356
13. Finkelstein MM, Jerrett M. A study of the relationships between Parkinson’s disease
and markers of traffic-derived and environmental manganese air pollution
in two Canadian cities.Environ Res20071043420–432
14. Lee P-C, Liu L-L, Sun Y, et al. Traffic-related air pollution
increased the risk of Parkinson’s disease
in Taiwan: a nationwide study.Environ Int20169675–81
15. Chen H, Rey J, Kwong C, et al. Living near major roads and the incidence of dementia, Parkinson’s disease
, and multiple sclerosis: a population-based cohort study
16. Chen C-Y, Hung H-J, Chang K-H, et al. Mortimer K. Long-term exposure
to air pollution
and the incidence of Parkinson’s disease
: a nested case-control study.PLoS One2017128e0182834
17. Palacios N, Fitzgerald KC, Hart JE, et al. Air pollution
and risk of Parkinson’s disease
in a large prospective study of men.Environ Health201788180
18. Palacios N, Fitzgerald K, Roberts AL, et al. A prospective analysis of airborne metal exposures and risk of Parkinson disease in the nurses’ health study cohort.Environ Health Perspect20141229933–938
19. 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.Occup Environ Med201557537–54
20. Sparkman NL, Johnson RW. Neuroinflammation associated with aging sensitizes the brain to the effects of infection or stress.Neuroimmunomodulation2008154–6323
21. Block ML, Elder A, Auten RL, et al. The outdoor air pollution
and brain health workshop.Neurotoxicology2012335972–984
22. Cesaroni G, Badaloni C, Romano V, Donato E, Perucci CA, Forastiere F. Socioeconomic position and health status of people who live near busy roads: the Rome Longitudinal Study (RoLS).Environ Health2010941
23. Cesaroni G, Badaloni C, Gariazzo C, et al. Long-term exposure
to urban air pollution
and mortality in a cohort of more than a million adults in Rome.Environ Health Perspect20131213324–331
24. Beelen R, Hoek G, Vienneau D, et al. Development of NO2 and NOx land use regression models for estimating air pollution
exposure in 36 study areas in Europe – The ESCAPE project.Atmos Environ20147210–23
25. Eeftens M, Beelen R, De Hoogh K, et al. Development of land use regression models for PM2.5, PM 2.5 absorbance, PM10 and PMcoarse in 20 European study areas; Results of the ESCAPE project.Environ Sci Technol2012462011195–11205
26. Cyrys J, Eeftens M, Heinrich J, et al. Variation of NO2 and NOx concentrations between and within 36 European study areas: results from the ESCAPE study.Atmos Environ201262374–390
27. Eeftens M, Tsai M-Y, Ampe C, et al. Spatial variation of PM2.5, PM10, PM2.5 absorbance and PMcoarse concentrations between and within 20 European study areas and the relationship with NO2 – results of the ESCAPE project.Atmos Environ201262303–317
28. Cyrys J, Heinrich J, Hoek G, et al. Comparison between different traffic-related particle indicators: elemental carbon (EC), PM2.5mass, and absorbance.J Expo Anal Environ Epidemiol2003132134–143
29. Gariazzo C, Silibello C, Finardi S, et al. A gas/aerosol air pollutants study over the urban area of Rome using a comprehensive chemical transport model.Atmos Environ200741347286–7303
30. Cesaroni G, Agabiti N, Rosati R, Forastiere F, Perucci CA. An index of socioeconomic position based on 2001 Census, Rome.Epidemiol Prev2006306352–357
31. Eisen EA, Agalliu I, Thurston SW, Coull BA, Checkoway H. Smoothing in occupational cohort studies: an illustration based on penalised splines.Occup Environ Med20046110854–860
32. Weuve J. Accounting for bias due to selective attrition: the example of smoking and cognitive decline.Epidemiology2012231119–128
33. Day FR, Loh P-R, Scott RA, Ong KK, Perry JRB. A robust example of collider bias in a genetic association study.Am J Hum Genet2016982392–393
34. Gan WQ, Koehoorn M, Davies HW, Demers PA, Tamburic L, Brauer M. Long-term exposure
to traffic-related air pollution
and the risk of coronary heart disease hospitalization and mortality.Environ Health Perspect20111194501–507
36. Breckenridge CB, Berry C, Chang ET, Sielken RL, Mandel JS. Association between Parkinson’s disease
and cigarette smoking, rural living, well-water consumption, farming and pesticide use: systematic review and meta-analysis.PLoS One20161141–42
37. Searles Nielsen S, Gallagher LG, Lundin JI, et al. Environmental tobacco smoke and Parkinson’s disease
38. Sperrin M, Candlish J, Badrick E, Renehan A, Buchan I. Collider bias is only a partial explanation for the obesity paradox.Epidemiology2016274525–530
41. Fine JP, Gray RJ. A proportional hazards model for the subdistribution of a competing risk.J Am Stat Assoc199994446496–509
42. Beelen R, Hoek G, van den Brandt PA, et al. Long-term effects of traffic-related air pollution
on mortality in a Dutch cohort (NLCS-AIR study).Environ Health Perspect20081162196–202
43. Dockery DW, Stone PH. Cardiovascular risks from fine particulate air pollution
.N Engl J Med20073565511–513
44. Lepeule J, Laden F, Dockery D, Schwartz J. Chronic exposure to fine particles and mortality: an extended follow-up of the Harvard Six Cities study from 1974 to 2009.Environ Health Perspect20121207965–970
45. Lipsett MJ, Ostro BD, Reynolds P, et al. Long-term exposure
to air pollution
and cardiorespiratory disease in the California teachers study cohort.Am J Respir Crit Care Med20111847828–835
47. Pope CA, Burnett RT, Thurston GD, et al. Cardiovascular mortality and long-term exposure
to particulate air pollution
: epidemiological evidence of general pathophysiological pathways of disease.Circulation2004109171–77
48. Angoa-Pérez M, Jiang H, Rodríguez AI, Lemini C, Levine RA, Rivas-Arancibia S. Estrogen counteracts ozone
-induced oxidative stress and nigral neuronal death.Neuroreport2006176629–633Available at: http://www.ncbi.nlm.nih.gov/pubmed/16603924
. Accessed 29 July 2016
49. Guevara-Guzmán R, Arriaga V, Kendrick K, et al. Estradiol prevents ozone
-induced increases in brain lipid peroxidation and impaired social recognition memory in female rats.Neuroscience2009159940–950
50. Pereyra-Muñoz N, Rugerio-Vargas C, Angoa-Pérez M, Borgonio-Pérez G, Rivas-Arancibia S. Oxidative damage in substantia nigra and striatum of rats chronically exposed to ozone
.J Chem Neuroanat200631114–123
51. Mirowsky J, Dailey L, Devlin R. Differential expression of pro-inflammatory and oxidative stress mediators induced by nitrogen dioxide and ozone
in primary human bronchial epithelial cells.Inhal Toxicol2016288374–382
52. Jerrett M, Richard BT, Pope AC III, et al. Long-term ozone
exposure and mortality.N Engl J Med2009360111085–1095
53. Atkinson RW, Butland BK, Dimitroulopoulou C, et al. Long-term exposure
to ambient ozone
and mortality: a quantitative systematic review and meta-analysis of evidence from cohort studies.BMJ Open20166e009493
54. Calderón-Garcidueñas L, Solt AC, Henríquez-Roldán 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 youn.Toxicol Pathol2008362289–310
55. Chuang KJ, Chan CC, Su TC, Lee CT, Tang CS. The effect of urban air pollution
on inflammation, oxidative stress, coagulation, and autonomic dysfunction in young adults.Am J Respir Crit Care Med20071764370–376
56. Clougherty JE. A growing role for gender analysis in air pollution
epidemiology.Environ Health Perspect20101182167–176
57. Mellick GD, Gartner CE, Silburn PA, Battistutta D. Passive smoking and Parkinson disease.Neurology2006671179–180
58. Tanaka K, Miyake Y, Fukushima W, et al. Active and passive smoking and risk of Parkinson’s disease
.Acta Neurol Scand20101226377–382
59. Quik M, Bordia T, Zhang D, Perez XA. Nicotine and nicotinic receptor drugs: potential for Parkinson’s disease
and drug-induced movement disorders.Int Rev Neurobiol2015124247–271
60. Stuckenholz V, Bacher M, Balzer-Geldsetzer M, et al. The α7 nAChR agonist PNU-282987 reduces inflammation and MPTP-induced nigral dopaminergic cell loss in mice.J Parkinsons Dis201332161–172
61. Suzuki S, Kawamata J, Matsushita T, et al. 3-[(2,4-dimethoxy)benzylidene]-anabaseine dihydrochloride protects against 6-hydroxydopamine-induced parkinsonian neurodegeneration through α7 nicotinic acetylcholine receptor stimulation in rats.J Neurosci Res2013913462–471
62. Cascini S, Agabiti N, Incalzi RA, et al. Pneumonia burden in elderly patients: a classification algorithm using administrative data.BMC Infect Dis201313559
63. Di Domenicantonio R, Cappai G, Arc M, et al. Occurrence of inflammatory bowel disease in central Italy: a study based on health information systems.Dig Liver Dis2014469777–782
64. Pinnarelli L, Nuti S, Sorge C, et al. What drives hospital performance? The impact of comparative outcome evaluation of patients admitted for hip fracture in two Italian regions.BMJ Qual Saf201221September 2011127–134
65. Cesaroni G, Porta D, Badaloni C, et al. Nitrogen dioxide levels estimated from land use regression models several years apart and association with mortality in a large cohort study