Urban air pollution and emergency department visits related to central nervous system diseases.

Ambient air pollution has been associated with adverse neurological health outcomes. Ambient pollutants are thought to trigger oxidative stress and inflammation to which vulnerable populations, such as elderly may be particularly susceptible. Our study investigated the possible association between concentrations of ambient air pollutants and the number of emergency department (ED) visits for nervous system disorders among people residing in a large Canadian city. A time-stratified case-crossover study design combining data from the National Ambulatory Care Reporting System (NACRS) and the National Air Pollution Surveillance (NAPS) between 2004 and 2015 was used. Two air quality health indices were considered in additional to specific pollutants, including carbon monoxide (CO), nitrogen dioxide (NO2), sulfur dioxide (SO2), ozone (O3) and fine particulate matter (PM2.5). Weather condition data were included in the models. ED visits with a discharge diagnosis were identified using ICD-10 codes (G00-G99). The analysis was stratified by sex and age, also by seasons. The associations were investigated in arrays organized as 18 strata and 15 time lags (in days) for each pollutant. Overall, 140,511 ED visits were included for the analysis. Most ED visits were related to episodic and paroxysmal diagnoses (G40-G47, 64%), with a majority of visits for migraines (G43, 39%). Among females, an increase of 0.1ppm ambient CO was associated with an increased risk of paroxysmal diagnoses at day 1 (RR = 1.019 (95%CI 1.004-1.033)), day 6 (1.024 (1.010-1.039)) and day 7 (1.022 (1.007-1.036). PM2.5 and SO2, and air quality indices were similarly associated with ED visits for episodic and paroxysmal disorders in days 6 and 7. Findings highlight that ambient air pollution is associated with an increased number of ED visits for nervous system disorders, particularly visits for paroxysmal diagnoses.

1. Introduction

Air pollution is the fifth-leading cause of mortality, accounting for nearly 9% of associated mortalities worldwide [1]. The study of the detrimental health effects of ambient air pollution is complex, as the environment contains a mixture of various particles and gases, including volatile organic compounds, particulate matter (PM), nitrogen oxides (NOx), ground-level ozone (O3), sulfur dioxide (SO2), carbon monoxide (CO) and others. Moreover, pathogenesis is also dependent on weather conditions, including humidity, wind speed, atmospheric pressure, precipitation, temperature, and seasons. Fine PM with a particle aerodynamic diameter of 2.5μm (PM2.5) or smaller, NO2, and O3 are often studied because such pollutants pose a significant threat to human health [2]. Both NO2 and PM2.5 are emitted directly into the atmosphere from industrial sources (i.e., burning of fossil fuels, including for power generation), forest fires, and vehicle exhaust, including heavy duty and diesel engine exhaust, whereas ground-level O3 is formed mostly by photochemical reaction [3].

Exposure to air pollutants may result in more visits to doctors or emergency department (ED), hospital admissions, frequent use of prescription medications, loss of productivity, and general changes to personal quality of life. In the last decade it was discovered that exposure to some major air pollutants may lead to various neurological outcomes, especially in older adults, including dementia, Alzheimer’s disease (AD) and Parkinson’s disease (PD) [4-6], headaches and migraines [7-9], epilepsy and seizures [10, 11], overall cognitive decline [12], and depression [13]. For example, long-term exposure to PM2.5 for women over 70 years of age was associated with cognitive decline [14], as well as decreases in brain gray and white matter volume [15]. However, short-term exposure to ambient NO2, PM2.5, SO2, and CO was associated with some neurological outcomes, including but not limited to depression [13], migraines [7], and epilepsy [10, 11, 16]. Two Canadian studies conducted in the largest and most populous provinces, Ontario and Quebec, found that when older Canadians are long-term exposed to ambient PM2.5 and NO2, and other traffic-related air pollutants (i.e., ultrafine particles and black carbon), over time they may develop dementia [6, 17]. In a study conducted in five major Canadian cities, exposure to ambient SO2 and PM2.5 was associated with more ED visits for headaches and migraine attacks, especially among females [7]. In China, the frequency of outpatient visits for epilepsy was higher for men exposed to ambient SO2 and O3, whereas women were more sensitive to NO2 exposure [16]. However, in another Chinese study, acute exposure to ambient NO2, CO, and PM2.5 was associated with increased hospitalization for epilepsy [10]. A recent review of studies in Europe, Asia, and America highlighted the relationship between exposure to ambient air pollution and neurologic development among the young, as well as cognitive decline among the elderly [18].

Vulnerable populations, including elderly people, newborns and young children, pregnant women, as well as those with underlying health conditions, appear to be particularly sensitive to the deleterious health effects of ambient air pollution [14, 19–22].

Understanding the relationship between air pollutants and human health is essential to estimate economic impacts and the burden of health system resource utilization. To date, there is a paucity of information addressing the association of air pollution exposure with a wide range of disorders of the nervous system over the full life course. Previous studies have focused predominantly on specific vulnerable populations and health outcomes [19, 20, 23], or on particular pollutants [4, 14, 23–26]. The objective of this study was to comprehensively examine potential associations between concentrations of ambient air pollution and the number of ED visits for central and peripheral nervous system diseases in a large urban Canadian city.

2. Materials and methods

2.1 Studied population

Health data on the number of daily ED visits related to physician-diagnosed central and peripheral nervous system diseases were based on the International Classification of Diseases 10 Revision (ICD-10) codes G00-G99 (Chapter VI: “Diseases of the nervous system”) [27], and were obtained from the National Ambulatory Care Reporting System (NACRS) database [28] for the period of April 01, 2004 to December 31, 2015, inclusively (overall 4,292 days or 140 months). Briefly, the NACRS database is based on the Canadian Institute of Health Information (CIHI) reporting system in Canada, where the data are collected from hospitals and ambulatory care centres in the province of Ontario [29]. In 2016, the enumerated population of Toronto was 2,731,571 according to the Canadian Census Division 2016. The population density of this region is 4,334 people per square kilometer. The study population included all individuals from newborn to >60 years with a home addresses located in the area determined by the Census Division of Toronto. The primary outcome was the number of ED visits with a discharge diagnosis of all nervous system disorders (ICD-10 codes; G00-G99), and sub-analyzed according to diagnostic sub-categories, including “inflammatory diseases of the central nervous system” (G00-G09), “systemic atrophies primarily affecting the central nervous system” (G10-G14), “extrapyramidal and movement disorders” (G20-G26), “other degenerative diseases of the nervous system” (G30-G32), “demyelinating diseases of the central nervous system” (G35-G37), “episodic and paroxysmal disorders” (G40-G47), “nerve, nerve root and plexus disorders” (G50-G59), “polyneuropathies and other disorders of the peripheral nervous system” (G60-G64), “diseases of myoneural junction and muscle” (G70-G73), “cerebral palsy and other paralytic syndromes” (G80-G83), and “other disorders of the nervous system” (G90-G99).

2.2 Air pollutants and meteorological conditions data

Long-term air pollution data were obtained from the National Air Pollution Surveillance (NAPS) program, maintained by Environment and Climate Change Canada [30]. Hourly data from seven automated fixed-site monitoring stations with approximate maximum distance among them of 15 kilometers were averaged to estimate daily air pollution concentrations for the whole Census Division of Toronto. Five major ambient air pollutants were studied, which included CO, NO2, PM2.5, O3, and SO2. For each pollutant, daily measurements taken at hourly intervals were obtained. A 24-hour average was used for all air pollutants with the exception to O3, which was calculated as a daily maximum 8-hour average. Additionally, the Air Quality Health Index (AQHI) was calculated, based on the composite of three pollutants, NO2, PM2.5, and O3, measured as 24-hour averages in Toronto. The AQHI incorporates air pollutant concentrations and health risk estimations determined by mortality rates in large Canadian cities [31]. For the Canadian public, these values are rounded and shown as integer numbers on a scale (1–10 and 10+, and also expressed by colours from blue to brown) to demonstrate the risk related to ambient air quality, where lower-scale numbers represent lower health risks and higher-scale numbers represent higher health risks. The AQHI values are generated using the following formula: where the coefficients for air pollutants were estimated using mortality risks [31]. In addition, another value of the index, AQHI-x, was calculated using the same three air pollutants as AQHI, but O3 was on a maximum 8-hour average. This index has a stronger representation of O3 than in the AQHI calculations, because ground-level O3 tends to reach higher concentrations during daytime, especially on hot sunny days. The constructed indices were applied to investigate simultaneous effects of three air pollutants on human health.

To control for seasonal fluctuations in the concentrations of some air pollutants, data were analyzed by season, defined as cold (October to March) and warm (April to September) periods, classified according to the mean temperatures for each month recorded in Toronto [30]. To control for variations in weather [32, 33], daily relative humidity and ambient temperature data were collected from one weather monitoring station located at the Toronto International airport, details of which are described elsewhere [34], and the values were stratified as potential confounders.

2.3 Statistical analyses

A time-stratified case-crossover analytical design was used for all measured and unmeasured time-invariant factors and confounders, such as socioeconomic factors or comorbidities [35]. The analyzed data on associations between air pollutants concentrations and the number of ED visits for nervous system outcomes were organized as time-series values with a day as the time unit. These data consist of daily counts of ED visits, daily concentrations of air pollutants, and daily values of ambient temperature and relative humidity. A time-stratified approach was applied to cluster the data using a calendar hierarchical structure, where days were nested in day of weeks, then further nested in months and then in years [36]. Such clusters grouped as four or five days and were separated from each other. Controlling for time variables (i.e., any trends or fluctuations, etc.) was undertaken as described previously [37]. In the fitted statistical models, temperature and relative humidity were added in the form of natural splines of three degrees of freedom. The concentrations and their lagged values were assigned for the corresponding days. A two-tailed test at the 0.05 significance level was applied. The coefficients (slope, Beta) related to air pollutants and their standard errors (SEBeta) were estimated by the applied statistical models. Using these values (i.e., Beta, SEBeta), relative risks (RR) could be calculated. The statistical analyses were performed as conditional quasi-Poisson regression models [37, 38]. The numerical calculations were done employing R statistical software using the procedure for Generalized Non-linear Models (the package GNM) with the option “quasipoisson” [39]. The realized statistical models have the following form: “ModelFit = gnm(EDVisits~AirPollutant + ns(RelativeHumidity,3) + ns(Temperature,3)”. The options family = quasipoisson, eliminate = factor (Cluster) were included in the specification. The conditional Poisson model avoids estimating cluster parameters. It is conditional upon the total counts in each constructed cluster.

The conditional Poisson model is represented as a multinomial model and it is given by the formula where N is the number of events (counts) on the cluster and N., = ∑ N is the sum of events in each cluster. The parameters related to cluster are eliminated by conditioning on the sum of events on each cluster [37]. Here i is a day when ED visits occurs, β is a row vector of the coefficients, and T denotes transposition. A vector x contains variables: air pollutant concentrations and weather factors.

In total, 2,160 statistical models {15 (time lags expressed as days) x 18 (strata) x 8 (air pollutants and air quality health indices)} were applied. Strata covered patients’ demographic characteristics, including age group (0–10, 11–60 and >60 years old) and sex group (all, males and females), as well as seasonality expressed dichotomously. The numerical results from all models are listed in the Supplementary Materials and at the online location https://github.com/szyszkowiczm/NERVEToronto. This location also contains histograms (air pollutants, temperature, and relative humidity) and the map of Toronto.

2.4 Research ethics

All datasets are publicly available and de-identified, as such this study was deemed exempt from review by the Health Canada Research Ethics Board.

3. Results

Between April 1, 2004 and December 31, 2015, there were a total of 140,511 ED visits related to diseases of the nervous system (all G00-G99 codes) (Table 1). Overall, a majority of ED visits were by females (59.5%), particularly in the age ranges of 11–60 years old and >60 years. A total of 89,708 visits (64% of all ED visits) for the nervous system diseases were for episodic and paroxysmal disorders (G40-G47), among which nearly 40% were related to migraine attacks (G43) (Table 2).

Table 1

Descriptive statistics on the number of ED visits for diseases of the nervous system (ICD-10 codes: G00 –G99) collected in the city of Toronto between April 1, 2004 and December 31, 2015.

	ED visits for nervous system diseases	%
All individuals	140,511	100
Sex:
Male (M)	56,909	40.5
Female (F)	83,602	59.5
Ages (in years):
0–10	2,314 (M) and 1,978 (F)	53.9 (M) and 46.1 (F)
11–60	36,177 (M) and 57,456 (F)	38.6 (M) and 61.4 (F)
60+	18,418 (M) and 24,168 (F)	43.2 (M) and 56.8 (F)
Seasons (months)a:
Cold	67,631	48.1
Warm	72,880	51.9

a—cold season (October-March), warm season (April-September), F- female, M- male.

Table 2

Comparison of frequencies for each diagnostic sub-code (G00-G99).

Code	Meaning	ED visits (%)
G00-09	Inflammatory diseases of the central nervous system	1,913 (1.4)
G10-14	Systemic atrophies primarily affecting the central nervous system	472 (0.3)
G20-26	Extrapyramidal and movement disorders	4,075 (2.9)
G30-32	Other degenerative diseases of the nervous system	1,704 (1.2)
G35-37	Demyelinating diseases of the central nervous system	1,816 (1.3)
G40-47	Episodic and paroxysmal disorders:	89,708 (63.8):
	• Epilepsy/seizures (G40 and G41)	18,670 (20.8)
	• Migraine (G43)	34,864 (38.9)
	• Other headaches syndromes (G44)	8,159 (9.1)
	• Cerebral ischaemic attacks and other related syndromes (G45)	21,914 (24.4)
	• Sleep disorders (G47)	6,101 (6.8)
G50-59	Nerve, nerve root and plexus disorders	30,607 (21.8)
G60-64	Polyneuropathies and other disorders of the peripheral nervous system	3,436 (2.5)
G70-73	Diseases of myoneural junction and muscle	889 (0.6)
G80-83	Cerebral palsy and other paralytic syndromes	1,244 (0.9)
G90-99	Other disorders of the nervous system	4,650 (3.3)
	TOTAL	140,514 (100)

Table 3 summarizes the descriptive statistics on five major air pollutants, two air quality health indices and weather variables for the entire study period and by warm (April to September) and cold (October to March) season. During cold season, the mean ambient temperature was captured at 1.6°C (-22.2–23.5°C), while during warm season, the mean ambient temperature was 17.0°C (-4.2–31.2°C) (Table 2). The mean levels of some air pollutants, NO2, PM2.5, and SO2, were slightly higher in warm season than in cold season; however, none of the annual mean values for the studied air pollutants was above the respective Canadian Ambient Air Quality Standards (CAAQS) [40]. Descriptive statistics for all major pollutants and weather factors are also available in S2 Table in S1 File.

Table 3

Descriptive data on ambient air pollutants and meteorological conditions on a daily basis, collected in Toronto, Canada between April 1, 2004 and December 31, 2015.

Factors (units)	Seasonsa				All months		CAAQS d
	Cold		Warm
Air pollutants	Mean	Min/Maxb	Mean	Min/Maxb	Mean	IQRc
PM2.5 (μg/m3)	8.5	0.9/35.6	9.2	0.1/65.5	8.9	6.5	8.8
NO2 (ppb)	13.8	3.2/46.5	18.0	4.3/59.8	16.1	8.8	17.0
O3 (ppb)	23.9	1.7/56.6	23.2	2.4/62.2	23.5	12.8	N/A
O3H8 (ppb)	42.3	11.0/94.0	44.9	9.0/107.0	43.7	19.0	62.0 (8-hour)
SO2 (ppb)	0.9	-0.5/5.3	1.8	0/ 12.0	1.4	1.2	5.0
CO (ppm)	0.3	0/0.7	0.3	0/1.1	0.3	0.1	N/A
Air quality indices
AQHI	2.8	1.1/5.8	3.2	1.1/7.6	3.0	1.0	N/A
AQHI-x	4.0	1.6/8.0	4.7	1.7/10.3	4.4	1.5	N/A
Weather
Temperature (° C)	1.6	-22.2/23.5	17.0	-4.2/31.2	9.5	16.7	-
Relative Humidity (%)	72.7	31.7/98.8	68.8	35.4/96.7	70.7	14.3	-

Notes:

a- cold season (October to March) and warm season (April to September),

b- Min is minimum and Max is maximum,

c- IQR is difference between the third (75th percentile) and first (25th percentile) quartiles,

d- Canadian Ambient Air Quality Standards for annual values of PM2.5, NO2, and SO2 (CAAQS) (available on https://www.ccme.ca/en/air-quality-report#slide-7) are from 2020 for comparison purposes solely,

N/A-not applicable.

Nervous system diseases (G00-G99): When examining ED data for all G00-G99 codes, a total of 115 consistent positive associations were found, especially for exposure to ambient NO2 (24 positive associations) and ambient CO (25 positive associations) (S1 Fig in S1 File), with higher number of associations on lag days 0 and 7 with corresponding 27 associations and 33 associations, respectively (S2 Fig in S1 File). More ED visits were noted during the warm season (S3 Fig in S1 File).

Episodic and paroxysmal disorders (G40-G47): ED visits for episodic and paroxysmal disorders were more frequent at higher levels of CO and PM2.5, with 29 and 27 positive associations, respectively (Fig 1). The majority of cases (39%) were due to all classes of migraines (G43), which was associated with higher ambient CO and PM2.5 exposure, especially on lag days 6 (54 associations) and 7 (47 associations) (Fig 2) and during colder season for females especially (Fig 3). Ambient SO2 exposure (20 positive associations) and the AQHI (27 positive associations) were found to be related to ED visits for episodic and paroxysmal disorders (Fig 1). The RR and associated 95% confidence intervals (CI) were then calculated based on the interquartile range (IQR), which was 0.1 ppm for CO. For episodic and paroxysmal disorders, the RR and associated 95%CIs were determined for females exposed to CO on lag day 1 (1.019 (1.004–1.033), lag day 6 (1.024 (1.010–1.039)), and lag day 7 (1.022 (1.007–1.036) (Fig 4).

Fig 1

Total frequencies of all associations: 18 strata (rows), five air pollutants and two air quality health indices (columns) between ambient air pollutants levels and the number of ED visits for episodic and paroxysmal disorders (G40-G47) in Toronto, Canada from April 1, 2004 to December 31, 2015.

For visual representation: 0 (green) colour represents others than positive statistically significant associations.

Fig 2

Total frequencies for all associations: Five air pollutants + two air quality health indices (rows) and 15 time lags of 0–14 days (columns), between exposure to urban air pollutants levels and the number of ED visits for episodic and paroxysmal disorders (G40-G47) in Toronto, Canada between April 1, 2004 and December 31, 2015.

For visual representation: 0 (green) colour represents others than positive statistically significant associations.

Fig 3

Total frequencies of all associations for ambient air pollutants and the number of ED visits for related episodic and paroxysmal disorders (G40-G47).

Eighteen strata (classified by patients’ sex and age, as well as cold vs. warm seasons) examined and arranged in the rows, and 15 lags (expressed as days) are arranged in columns. For visual representation: 0 (green) colour represents other than positive statistically significant associations.

Fig 4

Relative risks (RR) and 95% confidence intervals (95%CI) for an increase in a one interquartile range (CO, IQR = 0.1 ppm).

ED visits diagnosed with the ICD-10 codes G40-G47 for the entire study period in Toronto between 2004 and 2015.

There was no apparent relationship between seasonality, however, older age contributed to the exposure-effect association, especially among women ≥60 years old exposed to CO on the concurrent day (0 lag day) (Fig 4).

4. Discussion

4.1 Summary of findings

The present study found a significant association of air pollutants and ED visits for nervous system diseases. In particular, episodic and paroxysmal disorders were most frequent, and visits for migraines accounted for nearly 40% of all paroxysmal disorders. The results revealed that females were more prone to visit the ED after being acutely (lags of 0–1 or 6–7 days) exposed to ambient CO, SO2, and PM2.5 pollutants. There was a consistent effect of ambient CO exposure on the number of ED visits for all nervous system diseases and episodic and paroxysmal disorders.

4.2 Potential mechanisms underlying the identified associations

Previous studies have shown a positive association between ambient air pollution and recurring headaches and migraine attacks [7–9, 32, 33, 41], especially when individuals are exposed to CO [42-44], PM2.5 [8, 41], and SO2 [7, 9]. Headache pathophysiology, including migraines and vascular headaches, may be due to nociceptive stimuli, which may trigger changes in the vasodilation of the cranial blood vessels and/or sensory nerve fibers [45, 46], and air pollution is known to cause changes to vascular and neural activity [47, 48]. Ambient air pollutants may be related to systemic inflammation, oxidative stress, apoptosis [49, 50] or brain oxygenation and metabolism [45].

Results suggest a difference in response to air pollutants between males and females, and among various age groups. There were more ED encounters for CNS disorders by females (between 11 and 60 years of age compared to males, overall (Table 1) and for episodic and paroxysmal disorders (Fig 3)). Observed differences may be attributable to biological, physiological, and behavioral differences among sexes. Population-based studies suggest that air pollution is associated with headaches and migraine attacks, especially in women [7–9, 41, 51].

Previous environmental health studies showed stronger effects from acute air pollutants exposure on human health. In a European study with 12 volunteers of mean age of 24 years, nine volunteers developed prolonged headaches following CO inhalation during the two study days [51]. Additionally, significant increases in heart rate and facial skin blood flow were observed among the exposed volunteers [51]. In a cross-sectional study with 4,073 patients (mean age: 40 years), increased incidence of headaches was detected after CO poisoning, with further symptoms of nausea, dizziness, and shortness of breath [52]. In an American study, higher hospitalization rates and mortality due to CO poisoning were found, especially among older individuals of 75 years of age or above [53]. However, symptoms from prolonged CO exposure normally occurs when levels exceed 70 ppm [43], which is very high compared to levels found in the current study (0.3 ppm).

Several studies have demonstrated a negative impact of air pollution exposure on development and progression of neurodegenerative disorders, including PD [54] and multiple sclerosis [55, 56], as well as potential for violent behavior resulting in 3.13% increase in risk for homicide and inflicted injury [57]. Such increase in ED visits was observed especially among males, children and Hispanics exposed to ambient CO during warm season in California [57]. Possible explanation can be because of combination effect of the air pollutant and heat stress [57]. Such results are in agreement with another Canadian study showing that even small increases in ambient CO levels may be related to an increases in substance abuse visits [58]. There might be a clear hormesis associated with CO exposure, as such higher concentrations and prolonged exposure may result in neuroinflammation and disruption to blood-brain barrier and transfer of oxygen to the brain [42, 45, 50], but very small quantities may be used for therapeutic benefit in some cases [59].

Exposure to other pollutants, such as PM2.5 and SO2 may also contribute to central nervous system disorders. Exposure to both pollutants may result in higher prevalence of multiple sclerosis [55, 56] and higher risk of progression of PD [54] and AD [4]. In a Canadian study, exposure to ambient SO2 and PM2.5 was associated with increased risks of ED visits for depression [60]. Short-term exposure to ambient SO2 resulted in increased 1.12 odds ratio of suicide risks, especially among males and well-educated individuals [61].

4.3 Strengths and limitations of the current study

The present study has several strengths. First, this study used ten years of longitudinal data on exposure and incorporated daily fluctuations in ambient air pollution concentrations into models estimating ED visits for nervous system diseases. To estimate the relative risks of a particular health outcome, the frequency of exposure to pollution prior to the health outcome is compared to control times (for example, a health outcome taken on a Monday is compared to those on other Mondays of the month). In addition, a case-crossover approach controls for all measured and unmeasured confounding factors, including socioeconomics and co-morbidities. The study considered the individual five major ambient air pollutants, as well as multi-pollutants expressed in AQHI (and AQHI-x) indices. The current study also considered both sexes across the lifespan (from newborn to ≥60 years of age) for patients attending the hospitals’ ED.

There are some limitations to the present study. This study was unable to control for individuals’ exposure duration, as there are disparities in individuals’ lifestyles and habits. The study was also unable to control for personal exposure, including indoor and in-vehicle exposure, as well as proximity to major and local roads [62], as the current study was solely dependent on air pollution concentrations captured by seven fixed-site automated monitoring stations. Secondly, it is possible that some admission data reported to the national NACRS database by facilities may be incomplete or inaccurate due to human error [29]. Additionally, previous patients’ health information was not available, and thus, the study assessed the daily changes in air pollution concentrations and the number of ED visits for nervous system diseases in the given study period. Lastly, the multiplicity of comparisons among all five studied air pollutants, time lags of 14 days and 18 strata may introduce a possibility of erroneous associations (type 1 error). However, the study found consistent positive association between exposure to ambient air pollutants and nervous system diseases, as well as for the episodic and paroxysmal disorders alone.

5. Summary and future perspectives

In the present study, consistent positive exposure-effect associations were found between acute (i.e., concurrent day or a weeklong) exposure to ambient CO, PM2.5 and SO2 concentrations and more frequent utilization of ED for CNS diseases (i.e., episodic and paroxysmal disorders), and perhaps, attributable to migraine attacks among patients residing in such a large city as Toronto. Although the biological plausibility of the observed association between ambient air pollution and episodic and paroxysmal disorders, including headaches and migraine attacks, is still remaining unclear. Additionally, further studies should focus on other geographical regions with similar environmental exposures to either confirm or refute the consistency of the findings presented here.

(DOCX)

Click here for additional data file.

30 Jul 2021

PONE-D-21-18477

Urban air pollution and emergency department visits related to central nervous system diseases

PLOS ONE

Dear Dr. Szyszkowicz,

Thank you for submitting your manuscript to PLOS ONE. After careful consideration, we feel that it has merit but does not fully meet PLOS ONE’s publication criteria as it currently stands. Therefore, we invite you to submit a revised version of the manuscript that addresses the points raised during the review process.

Please submit your revised manuscript by Sep 13 2021 11:59PM. If you will need more time than this to complete your revisions, please reply to this message or contact the journal office at plosone@plos.org. When you're ready to submit your revision, log on to https://www.editorialmanager.com/pone/ and select the 'Submissions Needing Revision' folder to locate your manuscript file.

Please include the following items when submitting your revised manuscript:

A rebuttal letter that responds to each point raised by the academic editor and reviewer(s). You should upload this letter as a separate file labeled 'Response to Reviewers'.

A marked-up copy of your manuscript that highlights changes made to the original version. You should upload this as a separate file labeled 'Revised Manuscript with Track Changes'.

An unmarked version of your revised paper without tracked changes. You should upload this as a separate file labeled 'Manuscript'.

If you would like to make changes to your financial disclosure, please include your updated statement in your cover letter. Guidelines for resubmitting your figure files are available below the reviewer comments at the end of this letter.

If applicable, we recommend that you deposit your laboratory protocols in protocols.io to enhance the reproducibility of your results. Protocols.io assigns your protocol its own identifier (DOI) so that it can be cited independently in the future. For instructions see:  http://journals.plos.org/plosone/s/submission-guidelines#loc-laboratory-protocols . Additionally, PLOS ONE offers an option for publishing peer-reviewed Lab Protocol articles, which describe protocols hosted on protocols.io. Read more information on sharing protocols at  https://plos.org/protocols?utm_medium=editorial-email&utm_source=authorletters&utm_campaign=protocols .

We look forward to receiving your revised manuscript.

Kind regards,

Xiaohui Xu, PhD

Academic Editor

PLOS ONE

Journal requirements:

When submitting your revision, we need you to address these additional requirements.

1. Please ensure that your manuscript meets PLOS ONE's style requirements, including those for file naming. The PLOS ONE style templates can be found at

https://journals.plos.org/plosone/s/file?id=wjVg/PLOSOne_formatting_sample_main_body.pdf and

https://journals.plos.org/plosone/s/file?id=ba62/PLOSOne_formatting_sample_title_authors_affiliations.pdf

[Note: HTML markup is below. Please do not edit.]

Reviewers' comments:

Reviewer's Responses to Questions

Comments to the Author

1. Is the manuscript technically sound, and do the data support the conclusions?

The manuscript must describe a technically sound piece of scientific research with data that supports the conclusions. Experiments must have been conducted rigorously, with appropriate controls, replication, and sample sizes. The conclusions must be drawn appropriately based on the data presented.

Reviewer #1: No

Reviewer #2: Partly

**********

2. Has the statistical analysis been performed appropriately and rigorously?

Reviewer #1: Yes

Reviewer #2: No

**********

3. Have the authors made all data underlying the findings in their manuscript fully available?

The PLOS Data policy requires authors to make all data underlying the findings described in their manuscript fully available without restriction, with rare exception (please refer to the Data Availability Statement in the manuscript PDF file). The data should be provided as part of the manuscript or its supporting information, or deposited to a public repository. For example, in addition to summary statistics, the data points behind means, medians and variance measures should be available. If there are restrictions on publicly sharing data—e.g. participant privacy or use of data from a third party—those must be specified.

Reviewer #1: No

Reviewer #2: No

**********

4. Is the manuscript presented in an intelligible fashion and written in standard English?

PLOS ONE does not copyedit accepted manuscripts, so the language in submitted articles must be clear, correct, and unambiguous. Any typographical or grammatical errors should be corrected at revision, so please note any specific errors here.

Reviewer #1: Yes

Reviewer #2: No

**********

5. Review Comments to the Author

Please use the space provided to explain your answers to the questions above. You may also include additional comments for the author, including concerns about dual publication, research ethics, or publication ethics. (Please upload your review as an attachment if it exceeds 20,000 characters)

Reviewer #1: Lukina et al. examined the associations between short-term air pollution and emergency department visits related to neurological disorders. The manuscript needs more work in the Introduction, Methods, and Results.

1. Introduction. As mentioned in the Introduction, several studies have examined the associations between air pollution and neurological outcomes. The Introduction would benefit from discussing the maturation of the current literature, the research gaps, and the need to conduct the present study.

2. Methods. Please list out all the CNS disorders included in this study and their corresponding ICD codes.

3. Statistical models. Instead of describing the slope and standard errors, it would be clear to write out the formula of the conditional quasi-Poisson regression. It would also be helpful to include the statistical testing and measures of effects.

4. Figure 1. Please write out the effect estimates in the main results. The figure is innovative yet confusing.

Reviewer #2: The authors conducted a case-crossover study in Canada on the associations between ED visits for CNS diseases and air pollution. Overall, the quality of this work needs to be improved, in both methodological and structural aspects. There are also weird grammars and expressions in text that need to be fixed by experts in this field. I listed my detailed comments as follows.

1. Line 44, the aim of the study should be presented in a complete sentence.

2. Line 56, result description should accompany values with their significance.

3. Line 79, I would rather say PM, NO2 and O3 acquired most study interest.

4. Line135, how was AQHI and AQHI-x calculated? To be exact.

5. Line 148, why and how to match cases with controls should be stated clearly.

6. The generalized non-linear models are rarely used for case crossover designs, suggest to try the common conditional logistic regression models.

7. How were metrological factors controlled in the models? The non-linear associations should be considered by natural splines or DLNM cross-basis patterns.

8. What are the missing rates for air pollutants on both case/control days?

9. Line 157-158, suggest to change this expression. Also, if hundreds of models are performed, multiple testing is needed for type-I error.

10. Line 194, it was also found? Here and other grammars/expressions should be checked by peer scientists in this field.

11. It is not appropriate and offers minimal knowledge by presenting the frequencies in tested associations for each exposure. Reshape this by forest plots or tables that reveals more information.

**********

6. PLOS authors have the option to publish the peer review history of their article (what does this mean?). If published, this will include your full peer review and any attached files.

If you choose “no”, your identity will remain anonymous but your review may still be made public.

Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy.

Reviewer #1: No

Reviewer #2: No

[NOTE: If reviewer comments were submitted as an attachment file, they will be attached to this email and accessible via the submission site. Please log into your account, locate the manuscript record, and check for the action link "View Attachments". If this link does not appear, there are no attachment files.]

While revising your submission, please upload your figure files to the Preflight Analysis and Conversion Engine (PACE) digital diagnostic tool, https://pacev2.apexcovantage.com/. PACE helps ensure that figures meet PLOS requirements. To use PACE, you must first register as a user. Registration is free. Then, login and navigate to the UPLOAD tab, where you will find detailed instructions on how to use the tool. If you encounter any issues or have any questions when using PACE, please email PLOS at figures@plos.org. Please note that Supporting Information files do not need this step.

9 Sep 2021

Reviewer's Responses to Questions (reviewer comments (in black text) and our responses (in blue text):

Comments to the Author

1. Is the manuscript technically sound, and do the data support the conclusions?

The manuscript must describe a technically sound piece of scientific research with data that supports the conclusions. Experiments must have been conducted rigorously, with appropriate controls, replication, and sample sizes. The conclusions must be drawn appropriately based on the data presented.

Reviewer #1: No

Reviewer #2: Partly

________________________________________

2. Has the statistical analysis been performed appropriately and rigorously?

Reviewer #1: Yes

Reviewer #2: No

________________________________________

3. Have the authors made all data underlying the findings in their manuscript fully available?

The PLOS Data policy requires authors to make all data underlying the findings described in their manuscript fully available without restriction, with rare exception (please refer to the Data Availability Statement in the manuscript PDF file). The data should be provided as part of the manuscript or its supporting information, or deposited to a public repository. For example, in addition to summary statistics, the data points behind means, medians and variance measures should be available. If there are restrictions on publicly sharing data—e.g. participant privacy or use of data from a third party—those must be specified.

Reviewer #1: No

Reviewer #2: No

________________________________________

4. Is the manuscript presented in an intelligible fashion and written in standard English?

PLOS ONE does not copyedit accepted manuscripts, so the language in submitted articles must be clear, correct, and unambiguous. Any typographical or grammatical errors should be corrected at revision, so please note any specific errors here.

Reviewer #1: Yes

Reviewer #2: No

________________________________________

5. Review Comments to the Author

Please use the space provided to explain your answers to the questions above. You may also include additional comments for the author, including concerns about dual publication, research ethics, or publication ethics. (Please upload your review as an attachment if it exceeds 20,000 characters)

Reviewer #1: Lukina et al. examined the associations between short-term air pollution and emergency department visits related to neurological disorders. The manuscript needs more work in the Introduction, Methods, and Results.

Thank you very much for your positive comments and constructive feedback.

1. Introduction. As mentioned in the Introduction, several studies have examined the associations between air pollution and neurological outcomes. The Introduction would benefit from discussing the maturation of the current literature, the research gaps, and the need to conduct the present study.

Response: As suggested, we have expanded our introduction section by discussing the current literature, the research gaps and adjusting our study objectives for better read (lines: 109-122-marked version).

2. Methods. Please list out all the CNS disorders included in this study and their corresponding ICD codes.

Response: We have included all 11 diagnostic sub-codes in the text (lines: 154-162-marked version).

3. Statistical models. Instead of describing the slope and standard errors, it would be clear to write out the formula of the conditional quasi-Poisson regression. It would also be helpful to include the statistical testing and measures of effects.

Response: As requested, we provided the model presentation (using R statistical software). It was shown (Ref: Armstrong et al. 2014) that such realization is equivalent to conditional logistic regression realized in the standard case-crossover method. We calculated relative risks with associated 95% confidence intervals for all models and the numerical results are presented at the location: https://github.com/szyszkowiczm/NERVEToronto. We did two analyses; for G00-G99 (all nervous system diseases) and G40-G47 separately, where G40-G47 (episodic and paroxysmal disorders) is 64% of all ED visits, but this group is more related to air pollution (publications on migraine, headache previously done by Szyszkowicz et al 2008, 2009a,b,c and many others). For this subgroup we see more (from 115 to 161) positive associations, especially consistent effect of ambient CO exposure for both G00-G99 and G40-G47.

Reference: Armstrong BG, Gasparrini A, Tobias A. Conditional Poisson models: a flexible alternative to conditional logistic case cross-over analysis. BMC Med. Res. Methodol. 2014; 14:122. doi: 10.1186/1471-2288-14-122.

4. Figure 1. Please write out the effect estimates in the main results. The figure is innovative yet confusing.

Response: As requested, we included three figures instead outlining by time lags, air pollutants/air quality health indexes, and strata classified by patients’ age and sex, and seasons (warm versus cold).

Reviewer #2: The authors conducted a case-crossover study in Canada on the associations between ED visits for CNS diseases and air pollution. Overall, the quality of this work needs to be improved, in both methodological and structural aspects. There are also weird grammars and expressions in text that need to be fixed by experts in this field. I listed my detailed comments as follows.

Thank you very much for your positive comments and constructive feedback. We have carefully gone throughout the manuscript and corrected / improved sentences (all changes tracked). We note that Reviewer #1 indicated that the writing is generally good.

1. Line 44, the aim of the study should be presented in a complete sentence.

Response: We reworded the objective of the study: “The present study investigated all possible associations between concentrations of air pollutants and the number of emergency department (ED) visits for nervous system disorders among people residing in Toronto, Canada.” (lines: 44-47-marked version).

2. Line 56, result description should accompany values with their significance.

Response: We expanded the results portion in abstract section: “For all ED visits (G00-G99) and for G40-G47, positive statistically significant associations were 115 and 161, respectively. For G40-G47, among females an increase in one interquartile range (IQR=0.1ppm) of ambient CO gives the following relative risks and 95%CIs [lag1: 1.019(1.004-1.033), lag6 (1.024(1.010-1.039)) and lag7 (1.022(1.007-1.036)]. Women older than 60 years of age were also affected by CO on lag0 (1.036(1.007-1.065)). Other pollutants, PM2.5 and SO2, and air indices were also associated with ED visits for episodic and paroxysmal disorders in lag6 and lag7.” (lines: 64-73-marked version).

3. Line 79, I would rather say PM, NO2 and O3 acquired most study interest.

Response: We reworded the sentence. However, we respectfully disagree with this reviewer on this point. In the statement, we refer to three major air pollutants that indeed have been studied quite a lot, because many epidemiological studies have consistently shown associated mortality and morbidity from exposure to major air pollutants. According to the World Health Organization (WHO), air pollution is the fifth-leading mortality health risks in the world and responsible to nearly 4.9 million premature deaths in 2017. Similar estimates were done by the Global Burden of Diseases (GBD), where both PM2.5 and O3 was responsible for 4.5 million premature deaths. We cannot eliminate the statement that major pollutants like PM2.5, O3 and NO2 pose a significant threat to human health. The reference already provided presents comprehensive estimate of all health outcomes related to such three air pollutants.

4. Line135, how was AQHI and AQHI-x calculated? To be exact.

Response: We included the formula: The AQHI values are generated using the following formula:

AQHI=1000/(10.4)×(e^(0.000537*O3) ┤+e^(0.000871*NO2)+e^(0.000487*PM2.5)-├ 3),

where the coefficients for air pollutants were estimated using mortality risks [25]. Lines: 183-185-marked version.

AQHIx is using the same three pollutants but ozone is used at an 8-hr average because it requires sunlight, which already mentioned in lines186-191-marked version.

5. Line 148, why and how to match cases with controls should be stated clearly.

Response: As requested, we expanded on such matter: “The time-stratified approach was applied to cluster the data using a calendar hierarchical structure, where days were nested in day of weeks, which further nested in months and then in years [28]. Such clusters grouped four or five days and were separated from each other. Controlling of time variables (i.e., any trends or fluctuations, etc.) was done according to the applied methodology as described previously [30].” (Lines: 206-211-marked version).

6. The generalized non-linear models are rarely used for case crossover designs, suggest to try the common conditional logistic regression models.

Response: We added the model description. We are using conditional Poisson regression. As Armstrong et al. 2014 presented, this approach is equivalent to the conditional logistic regression analysis.

Reference: Armstrong BG, Gasparrini A, Tobias A. Conditional Poisson models: a flexible alternative to conditional logistic case cross-over analysis. BMC Med. Res. Methodol. 2014; 14:122. doi: 10.1186/1471-2288-14-122.

7. How were metrological factors controlled in the models? The non-linear associations should be considered by natural splines or DLNM cross-basis patterns.

Response: We added the presentation of the used model (in R statistical software). Meteorological factors (ambient temperature and relative humidity) are presented as natural spline with three degrees of freedom. This study belongs to more general approach under assumption “Ambient air pollution may be damaging every organ and virtually every cell in the human body”. We still fully don’t know which health problems are related to ambient air pollution concentration levels. As the applied methodology allows relatively fast “scan” of the associations we mainly did this. Our findings are supported that for the subgroup (G40-G47-episodic and paroxysmal disorders) we observe more positive statistically significant associations (161 in total) versus more broad G00-G99-nervous system diseases, with 115 positive statistically significant associations. What was proposed (say distributed lag non-linear model or DLNM) can be sued to the detailed study, say migraine attacks.

8. What are the missing rates for air pollutants on both case/control days?

Response: We used the case-crossover (CC) design but didn’t realize it as case/control, rather daily count of events on the clusters. Such approach is (purely) equivalent to the CC method (described in details in Armstrong et al. 2014).

9. Line 157-158, suggest to change this expression. Also, if hundreds of models are performed, multiple testing is needed for type-I error.

Response: The numerical calculations were done in R statistical software using the package GNM with the option “quasipoisson”. The quasi-Poisson regression is realized to model an over dispersed count variable.

As in raised point 7, we realize “to scan” of the potential associations. The number of the correlations increase (from 115 to 161 positive statistically significant associations) if we used ED visits (G40-G47-episodic and paroxysmal disorders) with more health conditions already linked with ambient air pollution (assuming migraine attacks-39% as majority of cases). Also we applied the “quasi-Poisson” option which gives wider 95% confidence intervals. We listed all results, the proposed coloured heat map, which allows to identify the patterns of the associations. These associations can be studied more in details, say ambient air pollution concentration levels and number of ED visits for migraine attacks (from common (without aura) to classical (with aura) to unspecified migraines to complicated migraines/status migrainosus) among females.

In our approach, even the used data from the NACRS are daily counts and time-series type, we don’t have issues of the used degree of freedom for time and/or others often present in GAM methodology.

10. Line 194, it was also found? Here and other grammars/expressions should be checked by peer scientists in this field.

Response: We are not sure we understand the posed question. The manuscript was thoroughly checked by the authors, which are Research Scientist and Principal Investigator (Dr. Szyszkowicz), Study Coordinator (A. Lukina) and Medical Doctor (Dr. Burstein). Additionally, an independent colleague have read the paper and stated that “the paper is well written with no major revisions needed”. However, some smoothing with minor edits were suggested and the paper was corrected accordingly. We hope that this newest version will be satisfactory.

11. It is not appropriate and offers minimal knowledge by presenting the frequencies in tested associations for each exposure. Reshape this by forest plots or tables that reveals more information.

Response: We have changed the Fig 1 into three additional Figs 1, 2 and 3 that provides fully information for G40-G47. We also expanded our Supplementary Information section by adding three more Figs (S2, S3 and S4)-for G00-G99. We also expanded the results by adding RR and associated 95% CI (lines: 271-275-marked version). ________________________________________

6. PLOS authors have the option to publish the peer review history of their article (what does this mean?). If published, this will include your full peer review and any attached files.

If you choose “no”, your identity will remain anonymous but your review may still be made public.

Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy.

Reviewer #1: No

Reviewer #2: No

Submitted filename: PONE-D-21-18477-Rebuttal Letter-Response to Reviews.docx

Click here for additional data file.

31 Jan 2022

7 Mar 2022

Response to Reviews:

Journal Requirements:

Comment: Please review your reference list to ensure that it is complete and correct. If you have cited papers that have been retracted, please include the rationale for doing so in the manuscript text, or remove these references and replace them with relevant current references. Any changes to the reference list should be mentioned in the rebuttal letter that accompanies your revised manuscript. If you need to cite a retracted article, indicate the article’s retracted status in the References list and also include a citation and full reference for the retraction notice.

Response: We have made some significant changes to the Reference list mostly by moving some of the references that already existed in the previous version of the paper but for more consistent flow and also by adding a few additional references as requested by one of the reviewers. References that have been moved are: #17 (old #52): Chen H, Kwong JC, Copes R. et al. Exposure to ambient air pollution and the incidence of dementia: a population-based cohort study. Environ. Int. 2017; 108: 271-277, #18 (old #53): Clifford A, Lang L, Chen R. et al. Exposure to air pollution and cognitive functioning across the life course – a systematic literature review. Environ. Res. 2016; 147: 383-398, and # 34 (old #51): Vicedo-Cabrera AM, Sera F, Liu C. et al. Short term association between ozone and mortality: global two stage time series study in 406 locations in 20 countries. BMJ 2020, 368: m108. doi:10.1136/bmj.m108.), which made many references to shift in their consecutive numbers as well. New references that have been added are #54: Hu C, Fang Y, Li F, Dong B, Hua X, Jiang W, Zhang H, Lyu Y, Zhang X. Association between ambient air pollution and Parkinson’s disease: systematic review and meta-analysis. Environ. Res. 2019; 168: 448-459, #55: Lavery AM, Waubant E, Casper TC, Roalstad S, Candee M, Rose J, Belman A, Weinstock-Guttman B, Aaen G, Tillema J, Rodriguez M, Ness J, Harris Y, Graves J, Krupp L, Charvet L, Benson L, Gorman M, Moodley M, Rensel M, Goyal M, Mar S, Chitnis T, Schreiner T, Lotze T, Greenberg B, Kahn I, Rubin J, Waldman AT. Urban air quality and associations with pediatric multiple sclerosis. Ann. Clin. Transl. Neur. 2018; 5(10): 1146-1153, #56: Abbaszadeh S, Tabary M, Aryannejad A, Abolhasani R, Araghi F, Khaheshi I, Azimi A. Air pollution and multiple sclerosis: a comprehensive review. Neurol. Sci. 2021; 42: 4063-4072, #57: Thilakaratne RA, Malig BJ, Basu R. Examining the relationship between ambient carbon monoxide, nitrogen dioxide, and mental health-related emergency department visits in California, USA. Sci. Total Environ. 2020; 746: 140915, #58: Szyszkowicz M, Thompson EM, Colman I, Rowe BH. Ambient air pollution exposure and emergency department visits for substance abuse. PLoS ONE 2018; 13(6): e0199826, #59: Goebel U, Wollborn J. Carbon monoxide in intensive care medicine-time to start the therapeutic application. Intens. Care Med. Exp. 2020; 8:2, #60: Szyszkowicz M, Kousha T, Kingsbury M, Colman I. Air pollution and emergency department visits for depression: a multicity case-crossover study. Environ. Health Insights 2016; 10: 155-161, and #61: Lin G, Li L, Song Y, Zhou Y, Shen S, Ou C. The impact of ambient air pollution on suicide mortality: a case-crossover study in Guangzhou, China. Environ. Health 2016; 15:90.). To accompany all these new references, the Özkaynak H, Baxter LK, Burke J. Evaluation and application of alternative air pollution exposure metrics in air pollution epidemiology studies. J. Expo. Sci. Env. Epid. 2013; 23: 565 reference have been shifted to its last position as #62.

We also removed a few references, including: Kim CS, Hu SC. Regional deposition of inhaled particles in human lungs: comparison between men and women. Appl. Physiol. 1998; 84(6): 1834-1844. (#51) and Lipton RB, Mazer C, Newman LC, Solomon S. Sumatriptan relies migraine like headaches associated with carbon monoxide. Headache 1997; 37: 392-395. (#53)

Additional Editor Comments:

Nil

[Note: HTML markup is below. Please do not edit.]

Reviewers' comments:

Reviewer's Responses to Questions

Comments to the Author

1. If the authors have adequately addressed your comments raised in a previous round of review and you feel that this manuscript is now acceptable for publication, you may indicate that here to bypass the “Comments to the Author” section, enter your conflict of interest statement in the “Confidential to Editor” section, and submit your "Accept" recommendation.

Reviewer #1: All comments have been addressed

Reviewer #2: All comments have been addressed

Reviewer #3: (No Response)

Reviewer #4: All comments have been addressed

Reviewer #5: (No Response)

________________________________________

2. Is the manuscript technically sound, and do the data support the conclusions?

The manuscript must describe a technically sound piece of scientific research with data that supports the conclusions. Experiments must have been conducted rigorously, with appropriate controls, replication, and sample sizes. The conclusions must be drawn appropriately based on the data presented.

Reviewer #1: Yes

Reviewer #2: Yes

Reviewer #3: Yes

Reviewer #4: Yes

Reviewer #5: Partly

________________________________________

3. Has the statistical analysis been performed appropriately and rigorously?

Reviewer #1: Yes

Reviewer #2: Yes

Reviewer #3: No

Reviewer #4: Yes

Reviewer #5: No

________________________________________

4. Have the authors made all data underlying the findings in their manuscript fully available?

The PLOS Data policy requires authors to make all data underlying the findings described in their manuscript fully available without restriction, with rare exception (please refer to the Data Availability Statement in the manuscript PDF file). The data should be provided as part of the manuscript or its supporting information, or deposited to a public repository. For example, in addition to summary statistics, the data points behind means, medians and variance measures should be available. If there are restrictions on publicly sharing data—e.g. participant privacy or use of data from a third party—those must be specified.

Reviewer #1: No

Reviewer #2: Yes

Reviewer #3: No

Reviewer #4: Yes

Reviewer #5: No

________________________________________

5. Is the manuscript presented in an intelligible fashion and written in standard English?

PLOS ONE does not copyedit accepted manuscripts, so the language in submitted articles must be clear, correct, and unambiguous. Any typographical or grammatical errors should be corrected at revision, so please note any specific errors here.

Reviewer #1: Yes

Reviewer #2: Yes

Reviewer #3: Yes

Reviewer #4: Yes

Reviewer #5: Yes

________________________________________

6. Review Comments to the Author

Please use the space provided to explain your answers to the questions above. You may also include additional comments for the author, including concerns about dual publication, research ethics, or publication ethics. (Please upload your review as an attachment if it exceeds 20,000 characters)

Reviewer #1: (No Response)

Reviewer #2: The authors have responded to my previous comments, but I have to mention that this manuscript still can be polished by editing service.

Response: We have thoroughly gone throughout the manuscript and corrected/improved sentences (all changes tracked). We hope that this new version is finally to your satisfaction.

Reviewer #3: The authors made a sincere effort in addressing the questions raised in the previous round. I have the following additional comments/clarifications:

Comment 1. A conditional quasi-Poisson regression model was used for fitting the data. Was there any count of value = 0? In such a situation, zero-augmented models need to be sued.

Response: Thank you for this question. Count value = 0 does not create the problems in the used method. We have 1008 Levels: 2004:1:1, 2004:1:2, …, 2015:12:7 – where each “year:month:day of week” may have 4 or 5 days. It may happen that for a specific stratum we have all days with 0 counts. In the used approach, it is not the problem. This situation creates problem in the standard case-crossover method, where the Cox proportional-hazards model is applied (conditional logistic regression). Before applying such models, we need to remove each stratum with all zeros, but not in here realized method.

Comment 2. The authors state: "In total, 2,160 statistical models {15 (time lags expressed as days) x 18 (strata) x 8 (air pollutants and air quality health indices)} were applied". If so many models were run, how was the best-fitting model determined (via model comparison statistics, such as AIC/BIC, etc)?

Response: We organized our data as time-series daily counts (and daily environmental factors). The model was fitted to the triple combination (i.e., stratum, air pollutant (five individual pollutants and two air quality health indices), and time lag expressed as days); therefore, we have such 2,160 statistical combinations of the triples. We do not validate (test AIC/BIC) the constructed model. In the used approach we do not have “time” as it is used in the GAM (GLM), where we need to model time variable (and validate the fitted models). Here, the constructed clusters controls for “time” – we conditioned to the sum of counts for each cluster. We added a mathematical description (lines 214-221-marked version).

Reviewer #4: The authors have responded well to the statistical issues raised in the previous review. There is no further statistical concern about this revised manuscript.

Response: Thank you for reviewing our manuscript and providing constructive comments.

Reviewer #5: Comment 1. The literature on previous studies was not properly surveyed. A thorough review is recommended.

Response: We have expanded on the discussion section and added more information on health effects associated with CO exposure and other ambient air pollutants (fine particulate matter and sulfur dioxide), the literature that was included is recent starting from 2016 (Szyszkowicz M, Kousha T, Kingsbury M, Colman I. Air pollution and emergency department visits for depression: a multicity case-crossover study. Environ. Health Insights 2016; 10: 155-161 and Lin G, Li L, Song Y, Zhou Y, Shen S, Ou C. The impact of ambient air pollution on suicide mortality: a case-crossover study in Guangzhou, China. Environ. Health 2016; 15:90.) and ending with the most recent from 2020 and 2021 (Abbaszadeh S, Tabary M, Aryannejad A, Abolhasani R, Araghi F, Khaheshi I, Azimi A. Air pollution and multiple sclerosis: a comprehensive review. Neurol. Sci. 2021; 42: 4063-4072, Thilakaratne RA, Malig BJ, Basu R. Examining the relationship between ambient carbon monoxide, nitrogen dioxide, and mental health-related emergency department visits in California, USA. Sci. Total Environ. 2020; 746: 140915., and Goebel U, Wollborn J. Carbon monoxide in intensive care medicine-time to start the therapeutic application. Intens. Care Med. Exp. 2020; 8:2.) (lines: 333-350-marked version).

Comment 2. In line number 160, the authors used the formula for generating AQHI values involving an exponential function. Would the authors justify the basis for taking exponential function? Is it possible to generate for AQHI values by using a logarithmic function?

Response: Thank you for this question. The air quality health index (or simply AQHI) is calculated as specified in the cited publication (Stieb DM et al, A new multipollutant, no-threshold air quality health index based on short-term associations observed in daily time-series analyses. J. Air Waste Manag. Assoc. 2008; 58(3):435-450. doi:10.3155/1047-3289.58.3.435). Since 2008, the AQHI is a standard scale in Canada designed for general Canadian population to understand how air quality means to their health. It is scaled as 1-10 or 10+ and is hourly reported to Canadian population. It is a preventive tool – the higher the number, the greater the health risk associated with the air quality, meaning that persons are advised to limit their outdoor activities (i.e., 1-3 mean “low health risk”, 4-6 mean “moderate health risk”, 7-10 mean “high health risk”, and 10+ mean “very high health risk”). In summary – we have not changed the definition (exp(x) vs. log(x)). It is an interesting proposition to apply “log”. We cannot modify the representation – as it will no longer be AQHI, which is an approved/accepted scale in Canada.

Comment 3. The authors should clearly explain the Generalized Non-linear Model. Is it possible to replace the ‘Quasi Poisson’ distribution by any other discrete or continuous distribution?

Response: Thank you for this question. There are a few methods, which can be used – main categories of the methods are: time-series (TS) or case-crossover (CC) designs.

In TS, daily counts are analyzed, usually generalized linear model (GLM) is used and Quasi-Poisson is applied as a link function.

In the standard CC method, individual events and matched controls period are analyzed. The conditional logistic regression is usually applied.

In the present study, we analyze daily counts (as in TS), but we are using the time-stratified CC design to control for “time” variable. The applied method and comparison with other techniques (+ software) are presented in Armstrong et al. Conditional Poisson models: a flexible alternative to conditional logistic case cross-over analysis. BMC Med. Res. Methodol. 2014; 14:122. doi: 10.1186/1471-2288-14-122.

We added some mathematical descriptions. The methodology is well represented/described in the above-cited reference.

Comment 4. The results of the statistical analysis (Table 3, Table S1, and Table S2) were not represented diagrammatically. The authors did not clearly explain the descriptive statistics. What could be concluded about the data from mean, median, and quartile values.

Response: The descriptive statistics for air pollutants used in our study are described on lines 242-253-marked version), mean, maximum and minimum levels are presented for each air pollutant (and air health index) for two seasons (warm and cold) and year round. As for supplemental materials, the ED visits (as daily counts) and environmental characteristics data are mainly for later publications and for other centers to compare/reference– we characterized the environmental conditions in Toronto, Canada. In addition, on the e-location already provided in the manuscript (https://github.com/szyszkowiczm/NERVEToronto), which contains the histograms of all used air pollutants, weather factors and the map of Toronto (population density and monitoring stations).

The results are generated in 3D space with the coordinates (stratum, air pollutant, lag – 3 axes). The figures S1, S2, and S3 are the projections on 2D – show scores of the significant results.

Comment 5. The novelty of the proposed approach has not been highlighted. Moreover, the authors did not emphasize on the significance of the proposed approach over the approaches discussed in the previous studies.

Response: We emphasised the uniqueness of the present study in introduction (lines 119-129-marked version); however, we added a bit more details of our study compared to previous studies (lines 123-125-marked version).

Comment 6. It would be better if the authors represent total frequencies (as in Figures 1, 2, 3, S1, S2, and S3) by a frequency diagram.

Response: We tabulated all the results from 2,160 models (RR and 95%CIs), also we included histogram/frequencies of the environmental factors and the map of Toronto in e-location mentioned on 227-marked version. Figure S4 and S5 are maps to the numerical results (RR, 95%CIs) for G00-G99 “all nervous system diseases” and G40-G47 “episodic and paroxysmal disorders”, respectively.

________________________________________

7. PLOS authors have the option to publish the peer review history of their article (what does this mean?). If published, this will include your full peer review and any attached files.

If you choose “no”, your identity will remain anonymous but your review may still be made public.

Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy.

Reviewer #1: No

Reviewer #2: No

Reviewer #3: No

Reviewer #4: No

Reviewer #5: No

Submitted filename: PONE-D-21-18477R1-Rebuttal_Letter.docx

Click here for additional data file.

12 Jun 2022

Urban air pollution and emergency department visits related to central nervous system diseases.

PONE-D-21-18477R2

Dear Dr. Szyszkowicz,

We’re pleased to inform you that your manuscript has been judged scientifically suitable for publication and will be formally accepted for publication once it meets all outstanding technical requirements.

Within one week, you’ll receive an e-mail detailing the required amendments. When these have been addressed, you’ll receive a formal acceptance letter and your manuscript will be scheduled for publication.

An invoice for payment will follow shortly after the formal acceptance. To ensure an efficient process, please log into Editorial Manager at http://www.editorialmanager.com/pone/, click the 'Update My Information' link at the top of the page, and double check that your user information is up-to-date. If you have any billing related questions, please contact our Author Billing department directly at authorbilling@plos.org.

If your institution or institutions have a press office, please notify them about your upcoming paper to help maximize its impact. If they’ll be preparing press materials, please inform our press team as soon as possible -- no later than 48 hours after receiving the formal acceptance. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information, please contact onepress@plos.org.

Kind regards,

Julian Aherne

Academic Editor

PLOS ONE

Additional Editor Comments (optional):

All three reviewers' recommend 'accept' as all comments have been addressed. Well done. I look forward to seeing your published manuscript.

Reviewers' comments:

Reviewer's Responses to Questions

Comments to the Author

1. If the authors have adequately addressed your comments raised in a previous round of review and you feel that this manuscript is now acceptable for publication, you may indicate that here to bypass the “Comments to the Author” section, enter your conflict of interest statement in the “Confidential to Editor” section, and submit your "Accept" recommendation.

Reviewer #2: All comments have been addressed

Reviewer #3: All comments have been addressed

Reviewer #4: All comments have been addressed

**********

2. Is the manuscript technically sound, and do the data support the conclusions?

The manuscript must describe a technically sound piece of scientific research with data that supports the conclusions. Experiments must have been conducted rigorously, with appropriate controls, replication, and sample sizes. The conclusions must be drawn appropriately based on the data presented.

Reviewer #2: Yes

Reviewer #3: (No Response)

Reviewer #4: Yes

**********

3. Has the statistical analysis been performed appropriately and rigorously?

Reviewer #2: Yes

Reviewer #3: (No Response)

Reviewer #4: Yes

**********

4. Have the authors made all data underlying the findings in their manuscript fully available?

The PLOS Data policy requires authors to make all data underlying the findings described in their manuscript fully available without restriction, with rare exception (please refer to the Data Availability Statement in the manuscript PDF file). The data should be provided as part of the manuscript or its supporting information, or deposited to a public repository. For example, in addition to summary statistics, the data points behind means, medians and variance measures should be available. If there are restrictions on publicly sharing data—e.g. participant privacy or use of data from a third party—those must be specified.

Reviewer #2: No

Reviewer #3: (No Response)

Reviewer #4: Yes

**********

5. Is the manuscript presented in an intelligible fashion and written in standard English?

PLOS ONE does not copyedit accepted manuscripts, so the language in submitted articles must be clear, correct, and unambiguous. Any typographical or grammatical errors should be corrected at revision, so please note any specific errors here.

Reviewer #2: Yes

Reviewer #3: (No Response)

Reviewer #4: Yes

**********

6. Review Comments to the Author

Please use the space provided to explain your answers to the questions above. You may also include additional comments for the author, including concerns about dual publication, research ethics, or publication ethics. (Please upload your review as an attachment if it exceeds 20,000 characters)

Reviewer #2: The authros have responded well to all my previous comments in R1. While the new text is appropriate in content, there are multiple problems with English syntax. Copyediting is required.

Reviewer #3: (No Response)

Reviewer #4: The authors have responded well to the statistical issues raised in the previous review. There is no further statistical concern about this revised manuscript.

**********

7. PLOS authors have the option to publish the peer review history of their article (what does this mean?). If published, this will include your full peer review and any attached files.

If you choose “no”, your identity will remain anonymous but your review may still be made public.

Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy.

Reviewer #2: No

Reviewer #3: No

Reviewer #4: No

**********

17 Jun 2022

PONE-D-21-18477R2

Urban air pollution and emergency department visits related to central nervous system diseases.

Dear Dr. Szyszkowicz:

I'm pleased to inform you that your manuscript has been deemed suitable for publication in PLOS ONE. Congratulations! Your manuscript is now with our production department.

If your institution or institutions have a press office, please let them know about your upcoming paper now to help maximize its impact. If they'll be preparing press materials, please inform our press team within the next 48 hours. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information please contact onepress@plos.org.

If we can help with anything else, please email us at plosone@plos.org.

Thank you for submitting your work to PLOS ONE and supporting open access.

Kind regards,

PLOS ONE Editorial Office Staff

on behalf of

Dr. Julian Aherne

Academic Editor

PLOS ONE