Migraine and Air Pollution A Systematic Review

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 17

Received: 9 March 2023 | Accepted: 4 August 2023

DOI: 10.1111/head.14632

REVIEW ARTICLE

Migraine and air pollution: A systematic review

​​
Andrea E. Portt MSc1 | Christa Orchard MPH1 | Hong Chen PhD1,2,3,4 |
Erjia Ge PhD1 | Christine Lay MD, MSc, FRCP, FAHS5,6 | Peter M. Smith PhD1,7

1
Dalla Lana School of Public Health,
University of Toronto, Toronto, Ontario, Abstract
Canada
Objective: To systematically synthesize evidence from a broad range of studies on the
2
Environmental Health Science and
Research Bureau, Health Canada, Ottawa,
association between air pollution and migraine.
Ontario, Canada Background: Air pollution is a ubiquitous exposure that may trigger migraine attacks.
3
Public Health Ontario, Environmental and There has been no systematic review of this possible association.
Occupational Health, Toronto, Ontario,
Canada Methods: We searched for empirical studies assessing outdoor air pollution and any
4
Populations & Public Health Research quantified migraine outcomes. We included short-­and long-­term studies with quanti-
Program, ICES, Toronto, Ontario, Canada
5
fied air pollution exposures. We excluded studies of indoor air pollution, perfume, or
Centre for Headache, Women's College
Hospital, Toronto, Ontario, Canada tobacco smoke. We assessed the risk of bias with the World Health Organization's
6
Temerty Faculty of Medicine, Neurology, bias assessment instrument for air quality guidelines.
University of Toronto, Toronto, Ontario,
Results: The final review included 12 studies with over 4,000,000 participants.
Canada
7
Institute for Work & Health, Toronto,
Designs included case-­crossover, case–­control, time series, and non-­randomized
Ontario, Canada pre–­post intervention. Outcomes included migraine-­related diagnoses, diary records,

Correspondence
medical visits, and prescriptions. Rather than pooling the wide variety of exposures
Andrea E. Portt, Dalla Lana School of and outcomes into a meta-­analysis, we tabulated the results.
Public Health, University of Toronto, 608-­
35 Charles Street West, Toronto, ON M4Y
Point estimates above 1.00 reflected associations of increased risk. In single-­pollutant
2R4, Canada. models, the percent of point estimates above 1.00 were carbon monoxide 5/5 (100%),
Email: andrea.portt@gmail.com
nitrogen dioxide 10/13 (78%), ozone 7/8 (88%), PM2.5 13/15 (87%), PM10 2/2 (100%),
Funding information black carbon 0/1 (0%), methane 4/6 (75%), sulfur dioxide 3/5 (60%), industrial toxic
University of Toronto Open Fellowship;
Ontario Graduate Scholarship; University
waste 1/1 (100%), and proximity to oil and gas wells 6/13 (46%). In two-­pollutant
of Toronto Data Science Institute Doctoral models, 16/17 (94%) of associations with nitrogen dioxide were above 1.00; however,
Fellowship
more than 75% of the confidence intervals included the null value.
Most studies had low to moderate risks of bias. Where differences were observed,
stronger quality articles generally reported weaker associations.
Conclusions: Balancing the generally strong methodologies with the small number of
studies, point estimates were mainly above 1.00 for associations of carbon monox-
ide, nitrogen dioxide, ozone, and particulate matter with migraine. These results were
most consistent for nitrogen dioxide.

Abbreviations: ED, emergency department; HR, hazard ratio; OR, odds ratio; PM10, particulate matter 10 microns or less in diameter; PM2.5, particulate matter 2.5 microns or less in
diameter; ppb, parts per billion; RR, risk ratio or relative risk; WHO, World Health Organization.

This is an open access article under the terms of the Creative Commons Attribution-NonCommercial License, which permits use, distribution and reproduction
in any medium, provided the original work is properly cited and is not used for commercial purposes.
© 2023 The Authors. Headache: The Journal of Head and Face Pain published by Wiley Periodicals LLC on behalf of American Headache Society.

Headache. 2023;63:1203–1219.  wileyonlinelibrary.com/journal/head | 1203


|

15264610, 2023, 9, Downloaded from https://headachejournal.onlinelibrary.wiley.com/doi/10.1111/head.14632 by Cochrane Hungary, Wiley Online Library on [31/12/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
1204 HEADA​CHE​

KEYWORDS
ambient air pollution, emergency department visits, environment, headache, headache diary,
migraine

I NTRO D U C TI O N M E TH O D S

Migraine is a brain disease with neurological sequelae that affects Types of studies
over one billion people and is the second leading cause of years lived
with disability worldwide.1 Environmental exposures might be a con- The protocol for this review was registered with the PROSPERO In-
tributing factor to migraine attacks among those who experience ternational prospective register of systematic reviews in November
the condition; therefore, identifying and avoiding triggers whenever 2021 (registration number CRD42021275394).17 As described in the
possible is an important prevention strategy for many people with protocol, we included peer-­reviewed studies that analyzed the as-
2
this condition. sociation between migraine and air pollution in humans. Recognizing
Air pollution is a ubiquitous exposure that impacts respiratory the dearth of information in this important area of headache medi-
and cardiovascular health, 3 and has been proposed as a poten- cine, we set out to summarize as much as possible of the available
tial trigger for migraine.4–­6 The potential biological mechanism data. As such, we accepted a wide range of populations, outdoor
through which air pollution may trigger migraine attacks is not air pollutants, comparators, migraine outcomes, and study designs.
yet fully understood. Most of the world's population is exposed We did not restrict based on short-­term or long-­term exposure and
to elevated ozone, nitrogen oxides, sulfur oxides, carbon mon- outcome measures. We limited exposures to quantitative (including
oxide, and particulate matter (PM), which are produced by traf- binary yes/no) measures of outdoor air pollution. Studies of indoor
7
fic, industrial processes, and burning fossil fuels. As we breathe, air pollution, tobacco smoke, and perfume were excluded. Studies
these pollutants may enter the bloodstream via the lungs, or that did not specify or quantify migraine were excluded, as were
move directly to the brain via the olfactory nerves. 8 After enter- studies that combined migraine with other headaches. Studies were
ing the body, reactive pollutant chemicals provoke inflammatory published in English up to February 2023. We excluded conference
5,9
processes that may lead to migraine attacks. In support of this abstracts or presentations, dissertations, case studies, editorials,
biological mechanism, an animal study found that rats exposed to and reviews.
increasing levels of nitrogen dioxide within realistic air pollution
ranges displayed increasing migraine-­like behavior with higher
levels of nitrogen dioxide. 5 Understanding the relationship be- Search and study selection
tween air pollution and migraine is important because if certain
levels of air pollution are associated with increased frequency of We developed our search strategy, including key terms and data-
migraine attacks, it may provide opportunities for interventions bases, in consultation with a librarian. We searched EMBASE and the
at an individual or community level to reduce migraine burden. Allied and Complementary Medicine Database (AMED) via Ovid, the
While respiratory risks are likely highest for children and peo- Cochrane Library, the Latin American and Caribbean Health Sciences
ple with preexisting conditions, migraine peaks during the third Literature (LILACS) database, and the Cumulative Index to Nursing
to fourth decades and often affects otherwise healthy individ- and Allied Health Literature (CINAHL). Our key words included “mi-
10,11
uals ; therefore, people with migraine may represent a previ- graine” as well as “air pollution” and/or “ambient” and/or “ambient air.”
ously unconsidered group who benefit from improved air quality. Two reviewers independently screened all titles and abstracts.
Given that individuals have limited control over the air they Where there was disagreement, the article was included in the next
breathe day-­to-­d ay,12 policy changes are the most effective way screening step.
to reduce air pollution exposure.
This review builds upon limited existing work summarizing asso-
ciations between air pollution and migraine and/or headache.13–­15 Data extraction
Specifically, we believe this review represents the first systematic
search for studies of air pollution and migraine, assessment of risk As well as pollution exposure and effect estimate data for single-­and
of bias, and synthesis of the available data. Reflecting the studies to two-­pollutant models, we also extracted data on covariates. Several
date, our review includes short-­ and long-­term studies and a wide studies examined effect modification by temperature, sex, age, and
variety of air pollution exposures and migraine outcomes. More time lag between exposure and outcome; therefore we extracted
broadly, this work builds on the existing evidence of air pollution as a these data as well. Data were extracted into an Excel document by
hazard not only for individuals with preexisting respiratory or heart either C.O. or A.P., and verified by A.P.
conditions,16 but also for those susceptible to neurologic conditions Effect estimates included odds ratios (ORs), hazard ratios (HRs),
including migraine. relative risks/risk ratios (RRs), percent changes, and percent relative
|

15264610, 2023, 9, Downloaded from https://headachejournal.onlinelibrary.wiley.com/doi/10.1111/head.14632 by Cochrane Hungary, Wiley Online Library on [31/12/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
H​E AD​ACH​E 1205

risks. Percent relative risks were converted to relative risks for ease Data collection templates, extracted data, and analytic code are
of comparison. It is commonly accepted that ORs approximate RRs available upon request.
when the incidence is less than 5%.18,19 Given the rare frequency
of migraine outcomes, we assumed that odds ratios approximated
risk ratios, and tabulated them side-­by-­side18 For simplicity, we also R E S U LT S
included HRs and percentages in the tables.
Where clarification was required or quantitative data were miss- Description of studies
ing, we emailed the corresponding authors. 20–­22
Our initial searches identified 124 records. After removing dupli-
cates, 118 titles and abstracts remained. We identified six more
Assessment of risk of bias records from searching references within initial search results and
later searches. We piloted our screening process in an open process
We piloted an assessment tool provided by the World Health Organ- on the first 14 publications.
ization (WHO), the Risk of bias assessment instrument for systematic We retrieved 24 full texts for further assessment. Twelve articles
reviews informing the WHO global air quality guidelines, 23 with three were excluded due to language, 29 study type (case study,30 edito-
selected studies to identify and resolve challenges. This instru- rial, or review31–­33), exposure or outcome not being quantified,34,35
ment includes assessment of bias in confounding, selection, expo- headache and migraine being pooled,36 or due to the data being used
sure, outcome, missing data, and selective reporting. For short-­term in a larger, included study37–­40 (see Figure 1).
studies, critical confounders are temperatures, seasonality, day of Table 1 summarizes the characteristics of the 12 remaining stud-
the week, and long-­term trends. Other potential confounders in the ies. Reporting was generally incomplete for cohort size, patient, and
WHO instrument are holidays and influenza epidemics (not used, migraine attack or event numbers. In terms of reported numbers
see modifications below). For long-­term studies, critical confound- alone, the total cohort size was 4,084,937, with 112,642 patients
ers are age, sex, socioeconomic status, and body mass index. Other with migraine, and 84,919 migraine events. Note that the Chen and
potential confounders are year of enrolment, ethnicity, diet, physical Chiu papers used the same cohort and migraine sample for analysis
activity, marital status, and smoking. 23 of different pollutants, and therefore participants and outcomes for
We applied the WHO instrument with the following modifications: this pair of papers are only counted once in Table 1.
Populations included public or private insurance program pa-
1. Influenza epidemics are seasonal and are associated with respi- tients, patients with migraine recruited from outpatient clinics
ratory24 and cardiovascular25 outcomes and mortality26 and are in-­person or by mailed questionnaire, and in one case, residents
therefore included in “other potential confounders” in the WHO of a single small town. Whereas one study specifically focused on
instrument for air quality guidelines. Given the lack of known patients with episodic migraine, 20 characteristics of the sample
or hypothesized relationship between migraine and influenza, patients and their migraine conditions were otherwise largely unre-
we did not include influenza in “other potential confounders” ported (Table 1). Presumably, migraine attacks recorded in diaries at
in our assessment of risk of confounding bias. home are often less severe than those recorded during emergency
2. Given the established impact of weather on air pollution, 27 and department (ED) visits.
the hypothesized association of migraine with humidity and baro- Table 2 summarizes exposures, comparators, outcomes, and
metric pressure, 28 these two variables were considered “other study designs. Exposures were quantified measures of black carbon,
potential confounders.” carbon monoxide, methane, nitrogen dioxide, ozone, sulfur dioxide,
residential proximity to oil and gas wells, PM 10 microns or less in
For each included study, two reviewers independently assessed diameter (PM10), and PM 2.5 microns or less in diameter (PM2.5). One
the risk of bias. Consensus was achieved through discussion. study applied closure of an industrial waste landfill as a binary proxy
for exposure to volatile organic compounds.
Outcomes were quantified migraine events: ED visits, hospital-
Data synthesis izations, headache diaries, reported migraine symptoms, migraine
diagnoses, migraine-­specific drug prescriptions, and estimated mi-
We set out to summarize the modest amount of available informa- graine severity.
tion on air pollution and migraine outcomes, with the understanding Study designs were mainly case-­crossover, but also included
that many of the studies would not be directly comparable. Indeed, case–­control, time series, repeated measures analysis, and non-­
given the wide variety of exposures and exposure ranges, outcomes, randomized pre–­post intervention (see Table 2).
confounding variables, effect measures, and study designs, the re- Reporting of the shape of the pollution–­outcome relationship
sulting data could not reasonably be pooled for meta-­analysis; there- was incomplete across studies. Studies that did make such a report
fore, we tabulated the results and summarized them narratively. We employed either linear relationships or non-­linear relationships with
also summarized effect modification by lag, temperature, and sex. splines.
|

15264610, 2023, 9, Downloaded from https://headachejournal.onlinelibrary.wiley.com/doi/10.1111/head.14632 by Cochrane Hungary, Wiley Online Library on [31/12/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
1206 HEADA​CHE​

F I G U R E 1 PRISMA flow diagram of the systematic search for migraine and air pollution and/or ambient and/or ambient air, leading to 12
full-­text studies for review. Modified from Page et al.57 [Color figure can be viewed at wileyonlinelibrary.com]

Transient health outcomes are sometimes reported after a the studies met the criteria for moderate risk of confounding bias
delay (lag) following environmental exposures.41 Among short-­ simply due to lack of control for holidays. Given that holidays are
term studies, lags varied by length and method of calculation. Lag relatively rare, this would not be expected to strongly influence
lengths spanned from same day to 7 days. The lagged exposures results. One short-­term study met the criteria for high risk of
were either calculated as simple average pollution over the given confounding bias due to missing the critical confounder tempera-
days or as distributed lags, which parameterize the time-­response ture,44 while a long-­term study from 1994 did not report on the
41
relationship. critical confounder body mass index.45 Finally, one study scored
Timeframes for long-­term studies ranged from 1 to 5 years (see low risk of bias by including all of the confounders listed as critical
Table 2). or potential.46
Given the minimal variation in levels of risk of bias, our ability to
assess effect estimates according to risk of bias was limited. Arti-
Risk of bias cles with low risk of bias tended to report weak positive associations
between pollutant exposures and migraine outcomes. Variation in
Figure 2 illustrates the proportion of the 12 included studies point estimates was greater in studies with moderate or high risks of
within each risk of bias category. The studies were largely well bias (see Tables S1–­S6).
designed with minor methodological challenges resulting in mod-
erate risk categorization. Risks of bias by study are summarized in
Table 3. Estimated effects of pollutants
In most categories of risk of bias (selection, exposure, outcome,
missing data, selective reporting) these articles fell within low or Short-­term single pollutant studies
moderate risk, with minor transgressions leading to moderate risk
ratings. For example, one study had a moderate risk of selection bias Short-­term studies tested the hypothesis that exposure to air pollu-
due to low response rates in surveys,42 whereas others had a moder- tion may increase the likelihood of migraine onset or severity within
ate risk of bias from missing data due to lack of reporting of missing the next 0–­7 days, as measured by clinic visits, hospitalizations, ED
data for exposures 43 or outcomes. 21 visits, or migraine diaries. Exposures included black carbon, carbon
In contrast to the other bias categories, the risk of confound- monoxide, ozone, PM10 and PM2.5, nitrogen dioxide, ozone, and sul-
ing bias was generally moderate to high. For example, eight of fur dioxide.
H​E AD​ACH​E

TA B L E 1 Characteristics of included studies including population and sample sizes, where reported.

Sample size

Size of Number of patients with Number of


Author Jurisdiction Year Population cohort migraine migraine events

Chen Taipei, Taiwan 2015 Random sample of National Health Insurance (NHI) program 1,000,000 NR 13,676
patients
Chiu Taipei, Taiwan 2015 Random sample of National Health Insurance (NHI) program 1,000,000 NR 13,676
patients
Dales Santiago, Chile 2009 Patients hospitalized for migraine as reported by the Instituto 2,699,000 NR NR
Nacional de Estadisticas
Elser Northern California, USA 2021 Adults with migraine covered by a single insurance company 360,139 89,575 NR
Elser California, USA 2023 Adult members of a commercial and Medicare claims database 9898 NR 5534
Koo Ansan-­city, South Korea 2010 Patients with migraine recruited from an outpatient clinic 245 33 NR
Lee Seoul, South Korea 2018 Patients who visited the ED for migraine as reported by the NR 18,921 NR
National Emergency Department Information System
Li Boston, USA 2019 Adults with episodic migraine recruited from hospitals and clinics NR 98 870
Mukamal Boston, USA 2009 Patients seen in a single ED 7054 2250 NR
Szyszkowicz Multi-­city Canada 2009 Patients who visited the ED for migraine NR NR 64,839
Tustin Pennsylvania, USA 2017 Adult clinic patients who replied to a mailed, self-­administered 7785 1765 NR
questionnaire
Zmirou Montchanin, France 1994 Residents of the town Montchanin 816 NR NR
a
Total 4,084,937 112,642 84,919

Abbreviations: ED, emergency department; NR, not reported.


a
Note that Chen 2015 and Chiu 2015 studied the same cohort and sample of migraine events.
| 1207

15264610, 2023, 9, Downloaded from https://headachejournal.onlinelibrary.wiley.com/doi/10.1111/head.14632 by Cochrane Hungary, Wiley Online Library on [31/12/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
TA B L E 2 Summary of study methods including designs, lags, exposures, comparators, and outcomes.
| 1208

Pollution-­
migraine
Author Year Study design Time frame Models Lag type relationship Exposure Comparator Outcome

Chen 2015 Case-­crossover Daily Logistic 3-­day average Linear PM2.5 Odds of clinic visits for migraine Clinic visits
regression on days with 75th percentile
pollution compared to 25th
percentile

Chiu 2015 Case-­crossover Daily Logistic 24-­h average Linear Carbon monoxide, Odds of clinic visits for migraine Clinic visits
regression nitrogen dioxide, on days with 75th percentile
ozone, PM10, sulfur pollution compared to 25th
dioxide percentile

Dales 2009 Time-­series Daily Mixed effects 0–­5 days Linear Carbon monoxide, Risk of hospitalization for migraine Hospitalizations
with undistributed & nitrogen dioxide, on days with 75th percentile
Poisson distributed ozone, PM2.5, PM10, pollution compared to 25th
distribution sulfur dioxide percentile

Elser 2021 Case–­control 5 years Generalized NR/NA Non-­linear: Methane, nitrogen Odds of migraine diagnosis at higher Migraine diagnosis
linear mixed Splines dioxide, oil and gas levels of residential pollutant
wells, PM2.5 exposure compared to lower

Case-­crossover 5 years Negative NR/NA Non-­linear: Methane, nitrogen Risk of migraine outcomes at higher Migraine probability,
binomial Splines dioxide, oil and gas levels of residential pollutant triptan
mixed wells, PM2.5 exposure compared to lower prescriptions,
neurology visits,
urgent care visits,
ED visits

Elser 2023 Case-­crossover Daily Logistic 7-­day average NR/NA PM2.5, overall and from Odds of ED visits for migraine on ED visits
regression wildfire days with 10 μg/m3 higher PM2.5
in the previous week compared
to days with lower PM2.5

Koo 2010 Case-­crossover Daily Logistic 0–­4 day distributed NR/NA Carbon monoxide, Hazard of self-­reported migraine Migraine diaries
regression nitrogen dioxide, in diaries on days with one
ozone, PM10, sulfur unit (ppb or μg/m3) increase in
dioxide pollution compared to days with
one unit lower pollution

Lee 2018 Case-­crossover Daily Logistic Single-­day lag and Linear Carbon monoxide, Odds of ED visits for migraine ED visits
regression moving average nitrogen dioxide, on days with 75th percentile
lag structures up ozone, PM2.5, PM10, pollution compared to 25th
to 7 days prior sulfur dioxide percentile

Li 2019 Repeated measures Daily Fixed effects 1–­3 day averages NR/NA Carbon monoxide, Odds of self-­reported migraine Migraine diaries
analysis logistic nitrogen dioxide, in diary on days with 75th
regression ozone, PM2.5, sulfur percentile pollution compared to
dioxide 25th percentile

Mukamal 2009 Case-­crossover Daily Logistic 0–­2 day distributed NR/NA Black carbon, nitrogen Odds of ED visits for migraine ED visits
regression dioxide, PM2.5, sulfur on days with 75th percentile
dioxide pollution compared to 25th
percentile
HEADA​CHE​

15264610, 2023, 9, Downloaded from https://headachejournal.onlinelibrary.wiley.com/doi/10.1111/head.14632 by Cochrane Hungary, Wiley Online Library on [31/12/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
|

15264610, 2023, 9, Downloaded from https://headachejournal.onlinelibrary.wiley.com/doi/10.1111/head.14632 by Cochrane Hungary, Wiley Online Library on [31/12/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
H​E AD​ACH​E 1209

Our main analysis focused on results with pooled sexes and sea-

Population-­level change

Abbreviations: ED, emergency department; NR/NA, not reported or not applicable; PM2.5, particulate matter 2.5 microns or less in diameter; PM10, particulate matter 10 microns or less in diameter; ppb,
headaches in the

in migraine drug
son or temperatures. We summarized these results in Table 4. Point

past 12 months
Yes/No: Migraine

prescriptions
estimates for the likelihood of migraine events and relatively higher
carbon monoxide, nitrogen dioxide, or ozone ranged from 0.97 to
Outcome

ED visits
1.17.20–22,46–48 The portion of estimates above the null value was 6/6
(100%) for carbon monoxide, 10/13 (77%) for nitrogen dioxide, and 7/8
(88%) for ozone; however, most confidence intervals included 1.00.

in migraine drug prescriptions in


exposure to gas wells compared
the mean pollutant level change

the affected town compared to


Informal comparison of the change
Point estimates for the association between migraine and rel-
Odds of self-­reported migraine by
with 75 percentile residential
amount compared to on days
days with higher pollution by

to 25th percentile exposure


Risk of ED visits for migraine on

atively higher PM10 or PM2.5 ranged from 0.97 to 3.09. 20,22,44,46–49

nation-­wide in France
For PM10 and PM2.5, 3/3 (100%) and 10/11 (91%) of estimates, re-
with lower pollution

spectively, were above the null. Seven of the confidence intervals


excluded the null value.
Comparator

Three of the included studies reported on the association be-


tween sulfur dioxide and migraine outcomes. 20,22,46 The point es-
timates were distributed around the null value, with several falling
below 1.00 (see Table 4).
Mukamal et al.47 estimated the OR for ED visits when black car-
Industrial waste landfill
Ozone, PM2.5, PM10,

bon was relatively higher by 0.51 μg/m3 to be 1.00 (0.93, 1.07) (see
sulfur dioxide

Table 4).
Exposure

Gas wells

Long-­term single pollutant studies

Three studies tested the hypothesis that people living in areas


relationship
Pollution-­
migraine

with higher long-­term exposure to air pollution may have higher


NR/NA
Linear

Linear

numbers of migraine outcomes than those living in areas with


lower pollution. Exposures were measures of methane gas, prox-
imity to oil and gas wells, and industrial landfill closures. Outcomes
included migraine diagnoses or self-­reports, migraine probability,
migraine drug prescriptions, neurology visits, urgent care visits,
Lag type

0–­2 day

NR/NA

NR/NA

and ED visits.
Oil and gas wells release various air pollutants including methane,
nitrogen oxides, PM, petroleum hydrocarbons, sulfur oxides, and vol-
linear mixed

Multiple linear
regression

regression

atile organic compounds.42,48,50 Two studies looked at the association


Generalized

models

between oil and/or gas wells and migraine. Elser et al. found point
Logistic
Models

estimates at or below the null value for the association between 4/5
(80%) of migraine outcomes and a continuous measure of proximity
to oil and gas wells in California.48 In contrast, they reported point
after landfill
before and
12 months

estimates above 1.00 for neurology visits, urgent care visits, and ED
Time frame

closure
18 months
Previous

visits, and migraine probability using a binary measure of wells within


Daily

10 km of residence. Tustin et al. studied survey reports of migraine in


association with unconventional oil and gas well development includ-
ing fracking in Pennsylvania. Oil and gas well exposure was an aggre-
Non-­r andomized,

intervention
Case-­crossover

gate estimate including well activity dates, volume of production, and


pre–­post
Study design

Time series

proximity to participants’ residences. Compared to participants in the


lowest quartile of exposure, those in the second, third, and highest
TA B L E 2 (Continued)

quartiles had 1.14 (0.74, 1.75), 0.89 (0.58, 1.36), and 1.43 (0.94, 2.18)
odd ratios, respectively, for reporting migraine attacks42; therefore,
2009

2017

1994
Year

based on these studies, estimates of potential associations between


parts per billion.

oil and gas extraction and migraine outcomes are mixed (see Table 4).
Szyszkowicz

Point estimates by Elser et al. of the association between long-­


Author

Zmirou
Tustin

term exposure to 10,000 kg/h of methane and 6 migraine outcomes


(diagnoses, triptan prescriptions, neurology visits, urgent care visits,
|

15264610, 2023, 9, Downloaded from https://headachejournal.onlinelibrary.wiley.com/doi/10.1111/head.14632 by Cochrane Hungary, Wiley Online Library on [31/12/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
1210 HEADA​CHE​

F I G U R E 2 Results of the risk of bias assessments. Two reviewers applied the Risk of bias assessment instrument for systematic reviews
informing WHO global air quality guidelines23 to all 12 included studies. For selective reporting, missing data, outcome measurement, exposure
assessment, and selection bias, the majority of studies met the criteria for low risk of bias. In contrast, 9 of 12 studies had moderate risk of
confounding bias and two met the criteria for high risk. [Color figure can be viewed at wileyonlinelibrary.com]

ED visits, and migraine probability) were evenly distributed around Six studies reported on potential effect modification by season/
the null value 1.00 (see Table 4). temperature with both sexes combined. The point estimates do not
Zmirou et al. described changes in prescribing of migraine drugs appear to change according to temperature, with the possible ex-
by unit per resident following closure of an industrial toxic waste ception of a stronger relationship with ozone during cold weather
landfill in Montchanin, France, in 1988. The landfill had received ad- (Table S9).
hesives, varnishes, and paint waste from 1979 to 1988. Over 300 There were limited data to extract on potential effect modifica-
volatile organic compounds were identified at the site including tion by age. One study reported generally higher point estimates and
benzene, chloroform, methylene chloride (still found in conven- tighter confidence intervals in the <40 and 40–­6 4 age groups com-
tional paint strippers in some jurisdictions), trichloroethylene, chlo- pared to the 65+ age group for carbon monoxide, nitrogen dioxide,
45
robenzene, 1,1,1-­trichloroethane, and 1,1-­dichloroethane. In the ozone, PM2.5, PM10, and sulfur dioxide.46 Another study reported
18 months following the closure, the number of units of migraine a stronger association with sulfur dioxide in warm weather for fe-
drugs prescribed per resident decreased by 17.9% in Montchanin males, and also reported that the association with nitrogen dioxide
compared to a 1.01% decrease nationwide (see Table 4). for males in warm weather was strongest below age 20 and above
age 40.49
We also assessed whether specific migraine outcomes were
Effect modification in single-­pollutant models more strongly associated with air pollution exposures. We focused
on nitrogen dioxide and PM2.5 as these pollutants were tested in re-
Potential effect modifications by lag time, sex, season or tempera- lation to many outcomes. We did not observe any clear pattern (see
ture, and age were examined in single-­pollutant models. Table S10).
Ten studies reported estimates for specific lags or long-­term
associations between single pollutant exposures and migraine out-
comes. Most point estimates for all time frames were above the null Two-­pollutant studies
value of 1.00. The confidence intervals from the estimates with a
48-­h lag were more likely than other timeframes to exclude the null. Four studies controlled for co-­pollutants while assessing the asso-
This suggests that if there is an association between air pollution and ciation between air pollutants and clinic visits, ED visits, or hospitali-
migraine outcomes, it may be strongest approximately 2 days follow- zations for migraine. 22,43,46,51 Overall results suggest that estimates
ing exposure (see Table S7). Where distributed and non-­distributed for the association between migraine outcomes and carbon monox-
lags were reported, results were similar regardless of lag distribution ide, ozone, and sulfur dioxide may be confounded by co-­pollution
type.45 with nitrogen dioxide and PM. Estimates for PM may be confounded
Dales et al. reported on effect modification by sex in all seasons. by co-­pollution with carbon monoxide, sulfur dioxide, and nitrogen
The results showed no evidence of effect modification by sex (see dioxide. In contrast, estimates for nitrogen dioxide remained stable
Table S8). in two-­pollutant models (see Table 5).
|

15264610, 2023, 9, Downloaded from https://headachejournal.onlinelibrary.wiley.com/doi/10.1111/head.14632 by Cochrane Hungary, Wiley Online Library on [31/12/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
H​E AD​ACH​E 1211

DISCUSSION

Reporting

Moderate
Moderate

Moderate
The objective of this study was to summarize the association be-

Low
Low

Low
Low
Low

Low
Low
Low
Low
tween air pollution exposures and migraine in humans. The over-
all quality of the included studies was high. Nonetheless, the small
Missing Data

number of studies and the wide variety of pollution exposures and


Moderate

Moderate

Moderate
migraine outcomes make it difficult to draw conclusions. To sum-
marize the studies we reviewed, point estimates for single-­pollutant
Low

Low
Low

Low

Low

Low
Low
Low
Low
models between carbon monoxide, nitrogen dioxide, ozone, PM2.5,
and PM10 and migraine outcomes were mainly above the null value.
Moderate

Moderate
Outcome

In contrast to several sub-­studies37–40,49 of a larger study49 included


in our review, estimates for the effect of exposure to sulfur diox-
Low
Low
Low

Low
Low
Low
Low
Low
Low
Low
ide does not suggest increased risk of migraine outcomes. 20,22,43,46
This disagreement may be due to different co-­pollutants or ranges
Moderate

Moderate

Moderate
Moderate
Exposure

of sulfur dioxide. Individual studies found point estimates around


Other sources of bias

the null value for an association between black carbon,47 methane,48


Low
Low
Low

Low

Low
Low
Low
Low

and migraine. Also based on a single study, a population exposed


to volatile industrial waste had a relatively higher level of migraine
drug prescription.45 Results are mixed regarding proximity to oil and
Moderate

Moderate
Moderate
Selection

gas wells.42,48 A lack of available data also limits conclusions about


Low
Low
Low
Low

Low
Low
Low

Low
Low

effect modification by sex, weather, and age. Results from four stud-
ies of two-­pollutant models suggest that positive point estimates of
physical activity, marital status

association with migraine outcomes is most consistent for exposure


Year of enrolment, ethnicity, diet,
Year of enrolment, diet, physical
Missing potential confounders

to nitrogen dioxide.43,46
TA B L E 3 Summary of the studies' risks of bias, including missing critical and other potential confounders.

Holidays, barometric pressure


Holidays, barometric pressure

Holidays, barometric pressure

We chose not to meta-­analyze our results due to lack of com-


parability in comparison ranges, outcomes, and adjustment for
Diet, physical activity
activity, smoking

confounders within individual pollutants. These differences make


it difficult to pool the results and to find a single effect estimate.
Disagreements in results between studies may stem from these dif-
Holidays
Holidays
Holidays

Holidays
Holidays

ferences, or variation in pollutant sources and related unmeasured


None

co-­pollutants. Some studies unsurprisingly found limited evidence


of associations given extremely narrow exposure ranges, for exam-
ple 1 part per billion (ppb) of ozone21 or 0.6 ppb of sulfur dioxide20
(see Table 4). We recommend that future research include prelim-
inary assessments of the population's range of pollution exposure
Body mass index
Missing critical

and limit analyses to areas where the range of exposure is likely to


Temperature
confounders

be clinically relevant. Results from this review suggest that nitrogen


dioxide should be included as a potential pollutant that triggers mi-
None
None
None
None

None
None
None
None
None
None

graine attacks.
With regards to exposure-­response time frames, short-­term
Risk of confounding

studies should include 48-­h lags, as this time frame appears to be


Confounding bias

the most significant for short-­term migraine outcomes. Although


averages and distributed lags yielded similar results in these stud-
Moderate
Moderate
Moderate
Moderate

Moderate

Moderate
Moderate
Moderate
Moderate

ies, we endorse the use of distributed lags. This method parame-


High

High
Low
bias

terizes the lag-­o utcome relationship based on the data, and can
be less biased when the lag between exposure and outcome is
unknown.41
2009

2009
2009
2023
2015
2015

2021

2018
2010

2017
2019

1994
Year

Care should be taken to include the critical and potential con-


founders listed in the WHO's bias assessment instrument, 23 as well
as barometric pressure changes. Finally, we agree with previous
Szyszkowicz
Mukamal

recommendations that analysis of ED visits should be restricted to


Zmirou
Author

Tustin
Dales
Chen

migraine as the main reason for the visit, excluding secondary head-
Elser

Elser
Chiu

Koo
Lee
Li

ache disorders.52
TA B L E 4 Summary of estimates of the association between air pollution and migraine outcomes in single-­pollutant models.
| 1212

Effect
Pollutant Author Year Type Range Units Outcome Lag Estimate (95% CI) Measure

Black carbon Mukamal 2009 Interquartile range 0.51 μg/m3 ED visits 24 h 1.00 (0.93, 1.07) OR
Carbon monoxide Lee 2018 Interquartile range 2.88 ppm ED visits 24 h 1.03 (1.01, 1.05) OR
Li 2019 Interquartile range 0.15 ppm Migraine diaries 24 h 1.05 (0.94, 1.16) OR
Li 2019 Interquartile range 0.15 ppm Migraine diaries 48 h 1.03 (0.90, 1.19) OR
Dales 2009 Interquartile range 1.15 ppm Hospitalizations 48 h 1.11 (1.06, 1.17) RR
Dales 2009 Interquartile range 1.15 ppm Hospitalizations 48 h: distributed lag 1.14 (1.04, 1.24) RR
Li 2019 Interquartile range 0.15 ppm Migraine diaries 72 h 1.02 (0.85, 1.21) OR
Gas wells Tustin 2017 Quartiles 3rd quartile Weighted distance to Self-­reported Long-­term 0.89 (0.58, 1.36) OR
wells migraine
Tustin 2017 Quartiles 2nd quartile Weighted distance to Self-­reported Long-­term 1.14 (0.74, 1.75) OR
wells migraine
Tustin 2017 Quartiles 4th quartile Weighted distance to Self-­reported Long-­term 1.43 (0.94, 2.18) OR
wells migraine
Oil and gas wells—­binary Elser 2027 Set range 1000-­unit Weighted sum of wells Triptan Long-­term 0.99 (0.94, 1.04) RR
increase within 10 km prescriptions
Elser 2031 Set range 1000-­unit Weighted sum of wells Migraine Long-­term 1.02 (0.97, 1.07) RR
increase within 10 km probability
Elser 2028 Set range 1000-­unit Weighted sum of wells Neurology visits Long-­term 1.09 (1.03, 1.16) RR
increase within 10 km
Elser 2029 Set range 1000-­unit Weighted sum of wells ED visits Long-­term 1.11 (1.00, 1.24) RR
increase within 10 km
Elser 2030 Set range 1000-­unit Weighted sum of wells Urgent care visits Long-­term 1.43 (1.21, 1.70) RR
increase within 10 km
Oil and gas wells—­continuous Elser 2026 Set range 1000-­unit Weighted sum of wells Migraine Long-­term 0.99 (0.99, 1.00) RR
increase within 10 km probability
Elser 2024 Set range 1000-­unit Weighted sum of wells ED visits Long-­term 0.99 (0.98, 1.01) RR
increase within 10 km
Elser 2025 Set range 1000-­unit Weighted sum of wells Urgent care visits Long-­term 0.99 (0.98, 1.01) RR
increase within 10 km
Elser 2023 Set range 1000-­unit Weighted sum of wells Neurology visits Long-­term 1.00 (1.00, 1.00) RR
increase within 10 km
Elser 2022 Set range 1000-­unit Weighted sum of wells Triptan Long-­term 1.00 (1.00, 1.01) RR
increase within 10 km prescriptions
Industrial waste landfill closure Zmirou 1995 Before/After Not applicable Not applicable Migraine drug Long-­term −0.010 (NA, NA) % change
Nationwide, France prescriptions
Zmirou 1994 Before/After in Not applicable Not applicable Migraine drug Long-­term −0.179 (NA, NA) % change
Montchanin, France prescriptions
HEADA​CHE​

15264610, 2023, 9, Downloaded from https://headachejournal.onlinelibrary.wiley.com/doi/10.1111/head.14632 by Cochrane Hungary, Wiley Online Library on [31/12/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
TA B L E 4 (Continued)

Effect
Pollutant Author Year Type Range Units Outcome Lag Estimate (95% CI) Measure
H​E AD​ACH​E

Methane Elser 2021 Set range 100,000 kg/h ED visits Long-­term 0.88 (0.63, 1.21) RR
Elser 2021 Set range 100,000 kg/h Neurology visits Long-­term 0.95 (0.85, 1.05) RR
Elser 2021 Set range 100,000 kg/h Migraine Long-­term 1.01 (0.94, 1.09) RR
probability
Elser 2021 Set range 100,000 kg/h Triptan Long-­term 1.03 (0.95, 1.12) RR
prescriptions
Elser 2021 No 100,000 kg/h Migraine diagnosis Long-­term 1.04 (1.00, 1.08) OR
Elser 2021 Set range 100,000 kg/h Urgent care visits Long-­term 1.12 (0.92, 1.36) RR
Nitrogen dioxide Mukamal 2009 Interquartile range 8.1 ppm ED visits 24 h 0.97 (0.89, 1.06) OR
Lee 2018 Interquartile range 17.58 ppb ED visits 24 h 1.05 (1.02, 1.09) OR
Li 2019 Interquartile range 11.7 ppb Migraine diaries 24 h 1.05 (0.93, 1.19) OR
Mukamal 2010 Interquartile range 8.1 ppm ED visits 48 h 1.07 (1.00, 1.14) OR
Li 2019 Interquartile range 11.7 ppb Migraine diaries 48 h 1.09 (0.93, 1.28) OR
Dales 2009 Interquartile range 15.43 ppb Hospitalizations 48 h 1.11 (1.06, 1.17) RR
Dales 2009 Interquartile range 15.43 ppb Hospitalizations 48 h: distributed lag 1.19 (1.06, 1.31) RR
Elser 2021 Set range 5.0 ppb Neurology visits Long-­term 0.99 (0.94, 1.05) RR
Elser 2021 Set range 5.0 ppb Migraine Long-­term 1.00 (0.96, 1.05) RR
probability
Elser 2021 Set range 5.0 ppb Triptan Long-­term 1.01 (0.98, 1.06) RR
prescriptions
Elser 2021 Set range 5.0 ppb Migraine diagnosis Long-­term 1.02 (1.00, 1.05) OR
Elser 2021 Set range 5.0 ppb ED visits Long-­term 1.16 (1.05, 1.29) RR
Elser 2021 Set range 5.0 ppb Urgent care visits Long-­term 1.22 (1.02, 1.46) RR
Ozone Li 2019 Interquartile range 14.0 ppb Migraine diaries 24 h 1.17 (1.00, 1.36) OR
Koo 2010 Set range 1.0 ppb Migraine diaries 48 h 1.01 (0.97, 1.05) HR
Li 2019 Interquartile range 14.0 ppb Migraine diaries 48 h 1.15 (0.95, 1.39) OR
Dales 2009 Interquartile range 69.51 ppb Hospitalizations 48 h 1.17 (1.08, 1.26) RR
Koo 2010 Set range 1.0 ppb Migraine diaries 72 h 1.00 (0.97, 1.04) HR
Lee 2018 Interquartile range 21.9 ppb ED visits 72 h 1.03 (1.00, 1.07) OR
Li 2019 Interquartile range 14.0 ppb Migraine diaries 72 h 1.17 (0.93, 1.47) OR
Koo 2010 Set range 1.0 ppb Migraine diaries 96 h 1.01 (0.98, 1.05) HR
3
PM10 Lee 2018 Interquartile range 29.23 μg/m ED visits 144 h 1.03 (1.01, 1.06) OR
Dales 2009 Interquartile range 37.79 μg/m3 Hospitalizations 48 h 1.10 (1.04, 1.15) RR
|

3
Dales 2010 Interquartile range 37.79 μg/m Hospitalizations 48 h: distributed lag 1.11 (1.04, 1.18) RR
1213

(Continues)

15264610, 2023, 9, Downloaded from https://headachejournal.onlinelibrary.wiley.com/doi/10.1111/head.14632 by Cochrane Hungary, Wiley Online Library on [31/12/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
TA B L E 4 (Continued)
| 1214

Effect
Pollutant Author Year Type Range Units Outcome Lag Estimate (95% CI) Measure

PM2.5 Szyszkowicz 2009 Mean pollutant level 8.3 μg/m3 ED visits 24 h 1.02 (1.00, 1.03) RR
change
Mukamal 2009 Interquartile range 6.7 μg/m3 ED visits 24 h 1.02 (0.96, 1.09) OR
Li 2019 Interquartile range 3.6 μg/m3 Migraine diaries 24 h 1.03 (0.93, 1.15) OR
Li 2019 Interquartile range 3.6 μg/m3 Migraine diaries 48 h 1.02 (0.90, 1.17) OR
3
Lee 2018 Interquartile range 15.22 μg/m ED visits 48 h 1.03 (1.01, 1.05) OR
Dales 2009 Interquartile range 21.51 μg/m3 Hospitalizations 48 h 1.11 (1.00, 1.19) RR
3
Dales 2009 Interquartile range 21.51 μg/m Hospitalizations 48 h: distributed lag 1.14 (0.98, 1.30) RR
Szyszkowicz 2009 Mean pollutant level 8.3 μg/m3 ED visits 72 h 1.02 (1.00, 1.03) RR
change
Li 2019 Interquartile range 3.6 μg/m3 Migraine diaries 72 h 1.05 (0.90, 1.23) OR
Elser 2023 Set range 10.0 μg/m3 ED visits 168 h (7-­day) 0.97 (0.86, 1.10) OR
average
Elser 2021 Set range 5.0 μg/m3 Migraine diagnosis Long-­term 1.00 (0.97, 1.10) OR
3
Elser 2021 Set range 5.0 μg/m Triptan Long-­term 1.03 (0.97, 1.10) RR
prescriptions
Elser 2021 Set range 5.0 μg/m3 Migraine Long-­term 1.14 (1.07, 1.22) RR
probability
Elser 2021 Set range 5.0 μg/m3 Neurology visits Long-­term 1.18 (1.09, 1.29) RR
3
Elser 2021 Set range 5.0 μg/m Urgent care visits Long-­term 3.09 (2.28, 4.2) RR
PM2.5—­Wildfire Elser 2023 Set range 10.0 μg/m3 ED visits 168 h (7-­day) 1.07 (0.83, 1.37) OR
average
Sulfur dioxide Li 2019 Interquartile range 0.6 ppb Migraine diaries 24 h 1.01 (0.93, 1.09) OR
Lee 2018 Interquartile range 2.37 ppb ED visits 24 h 1.02 (0.99, 1.05) OR
Dales 2009 Interquartile range 6.2 ppb Hospitalizations 48 h 1.10 (1.04, 1.17) RR
Dales 2009 Interquartile range 6.2 ppb Hospitalizations 48 h: distributed lag 1.13 (1.05, 1.22) RR

Abbreviations: CI, confidence interval; ED, emergency department; NA, not applicable; OR, odds ratio; ppb, parts per billion; ppm, parts per million; RR, risk ratio or relative risk.
HEADA​CHE​

15264610, 2023, 9, Downloaded from https://headachejournal.onlinelibrary.wiley.com/doi/10.1111/head.14632 by Cochrane Hungary, Wiley Online Library on [31/12/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
|

15264610, 2023, 9, Downloaded from https://headachejournal.onlinelibrary.wiley.com/doi/10.1111/head.14632 by Cochrane Hungary, Wiley Online Library on [31/12/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
H​E AD​ACH​E 1215

TA B L E 5 Summary of estimates of the association between air pollution and migraine outcomes in two-­pollutant models.

Pollutant Co-­pollutant Author Year Season Interquartile range Units Estimate (95% CI) Effect measure

Carbon None Chiu 2015 Cold 0.29 ppm 1.02 (0.98, 1.07) OR
monoxide Lee 2018 All 2.88 ppm 1.03 (1.01, 1.05) OR
Chiu 2015 Warm 0.29 ppm 1.20 (1.14, 1.26) OR
Nitrogen dioxide Chiu 2015 Cold 0.29 ppm 0.76 (0.69, 0.84) OR
Lee 2018 All 2.88 ppm 1.02 (0.98, 1.05) OR
Chiu 2015 Warm 0.29 ppm 1.18 (1.06, 1.30) OR
Ozone Lee 2018 All 2.88 ppm 1.03 (1.01, 1.05) OR
Chiu 2015 Cold 0.29 ppm 1.15 (1.09, 1.21) OR
Chiu 2015 Warm 0.29 ppm 1.19 (1.14, 1.25) OR
PM10 Chiu 2015 Cold 0.29 ppm 0.99 (0.93, 1.04) OR
Lee 2018 All 2.88 ppm 1.02 (0.99, 1.05) OR
Chiu 2015 Warm 0.29 ppm 1.19 (1.12, 1.27) OR
PM2.5 Lee 2018 All 2.88 ppm 1.01 (0.98, 1.04) OR
Sulfur dioxide Lee 2018 All 2.88 ppm 1.04 (1.01, 1.07) OR
Chiu 2015 Cold 0.29 ppm 1.09 (1.03, 1.16) OR
Chiu 2015 Warm 0.29 ppm 1.22 (1.15, 1.28) OR
Nitrogen dioxide None Lee 2018 All 17.58 ppb 1.05 (1.02, 1.09) OR
Chiu 2015 Cold 8.49 ppb 1.1 (1.05, 1.15) OR
Chiu 2015 Warm 8.49 ppb 1.16 (1.11, 1.21) OR
Carbon monoxide Chiu 2015 Warm 8.49 ppb 1.02 (0.94, 1.11) OR
Lee 2018 All 17.58 ppb 1.06 (1.02, 1.11) OR
Chiu 2015 Cold 8.49 ppb 1.40 (1.27, 1.55) OR
Ozone Chiu 2015 Cold 8.49 ppb 0.98 (0.93, 1.03) OR
Lee 2018 All 17.58 ppb 1.05 (1.02, 1.08) OR
Chiu 2015 Warm 8.49 ppb 1.15 (0.99, 1.09) OR
PM10 Lee 2018 All 17.58 ppb 1.05 (1.01, 1.08) OR
Chiu 2015 Cold 8.49 ppb 1.09 (1.03, 1.14) OR
Chiu 2015 Warm 8.49 ppb 1.15 (1.08, 1.21) OR
PM2.5 Lee 2018 All 17.58 ppb 1.04 (1.00, 1.08) OR
Sulfur dioxide Lee 2018 All 17.58 ppb 1.06 (1.03, 1.1) OR
Chiu 2015 Cold 8.49 ppb 1.20 (1.13, 1.27) OR
Chiu 2015 Warm 8.49 ppb 1.20 (1.14, 1.27) OR
Ozone None Lee 2018 All 21.9 ppb 1.03 (1.00, 1.07) OR
Chiu 2015 Warm 12.5 ppb 1.07 (1.02, 1.12) OR
Chiu 2015 Cold 12.5 ppb 1.23 (1.16, 1.31) OR
Carbon monoxide Lee 2018 All 21.9 ppb 1.03 (1.00, 1.07) OR
Chiu 2015 Warm 12.5 ppb 1.05 (1.00, 1.10) OR
Chiu 2015 Cold 12.5 ppb 1.36 (1.26, 1.47) OR
Nitrogen dioxide Chiu 2015 Warm 12.5 ppb 1.02 (0.97, 1.08) OR
Lee 2018 All 21.9 ppb 1.03 (1.00, 1.06) OR
Chiu 2015 Cold 12.5 ppb 1.43 (1.33, 1.54) OR
PM10 Chiu 2015 Warm 12.5 ppb 1.03 (0.97, 1.08) OR
Lee 2018 All 21.9 ppb 1.03 (1.00, 1.06) OR
Chiu 2015 Cold 12.5 ppb 1.24 (1.16, 1.32) OR
PM2.5 Lee 2018 All 21.9 ppb 1.03 (0.99, 1.06) OR
Sulfur dioxide Lee 2018 All 21.9 ppb 1.03 (1.00, 1.07) OR
Chiu 2015 Warm 12.5 ppb 1.06 (1.01, 1.11) OR
Chiu 2015 Cold 12.5 ppb 1.22 (1.14, 1.31) OR

(Continues)
|

15264610, 2023, 9, Downloaded from https://headachejournal.onlinelibrary.wiley.com/doi/10.1111/head.14632 by Cochrane Hungary, Wiley Online Library on [31/12/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
1216 HEADA​CHE​

TA B L E 5 (Continued)

Pollutant Co-­pollutant Author Year Season Interquartile range Units Estimate (95% CI) Effect measure
3
PM10 None Lee 2018 All 29.23 μg/m 1.03 (1.01, 1.06) OR
Chiu 2015 Cold 26.81 μg/m3 1.05 (1.01, 1.10) OR
Chiu 2015 Warm 26.81 μg/m3 1.11 (1.06, 1.17) OR
Carbon monoxide Chiu 2015 Warm 26.81 μg/m3 1.01 (0.95, 1.07) OR
3
Lee 2018 All 29.23 μg/m 1.03 (1.00, 1.05) OR
Chiu 2015 Cold 26.81 μg/m3 1.06 (1.01, 1.11) OR
Nitrogen dioxide Chiu 2015 Warm 26.81 μg/m3 1.02 (0.95, 1.08) OR
Lee 2018 All 29.23 μg/m3 1.02 (1.00, 1.05) OR
Chiu 2015 Cold 26.81 μg/m3 1.03 (0.98, 1.07) OR
Ozone Lee 2018 All 29.23 μg/m3 1.03 (1.01, 1.06) OR
Chiu 2015 Cold 26.81 μg/m3 1.06 (1.02, 1.10) OR
3
Chiu 2015 Warm 26.81 μg/m 1.10 (1.04, 1.16) OR
Sulfur dioxide Lee 2018 All 29.23 μg/m3 1.03 (1.01, 1.06) OR
Chiu 2015 Warm 26.81 μg/m3 1.11 (1.05, 1.18) OR
Chiu 2015 Cold 26.81 μg/m3 1.13 (1.07, 1.18) OR
PM2.5 None Chen 2015 Cold 17.48 μg/m3 1.02 (0.98, 1.07) OR
Lee 2018 All 15.22 μg/m3 1.03 (1.01, 1.05) OR
3
Chen 2015 Warm 17.48 μg/m 1.13 (1.08, 1.19) OR
Carbon monoxide Chen 2015 Cold 17.48 μg/m3 1.02 (0.96, 1.08) OR
Chen 2015 Warm 17.48 μg/m3 1.03 (0.97, 1.09) OR
Lee 2018 All 15.22 μg/m3 1.04 (1.01, 1.07) OR
Nitrogen dioxide Chen 2015 Cold 17.48 μg/m3 0.97 (0.92, 1.02) OR
Lee 2018 All 15.22 μg/m3 1.02 (1.00, 1.05) OR
3
Chen 2015 Warm 17.48 μg/m 1.03 (0.97, 1.10) OR
Ozone Chen 2015 Cold 17.48 μg/m3 1.03 (0.98, 1.08) OR
Lee 2018 All 15.22 μg/m3 1.03 (1.01, 1.05) OR
Chen 2015 Warm 17.48 μg/m3 1.13 (1.07, 1.20) OR
Sulfur dioxide Lee 2018 All 15.22 μg/m3 1.04 (1.02, 1.07) OR
Chen 2015 Cold 17.48 μg/m3 1.13 (1.06, 1.20) OR
Chen 2015 Warm 17.48 μg/m3 1.15 (1.09, 1.22) OR
Sulfur dioxide None Chiu 2015 Cold 2.14 ppb 0.94 (0.89, 0.99) OR
Lee 2018 All 2.37 ppb 1.02 (0.99, 1.05) OR
Chiu 2015 Warm 2.14 ppb 1.05 (1.00, 1.10) OR
Carbon monoxide Chiu 2015 Cold 2.14 ppb 0.88 (0.82, 0.94) OR
Chiu 2015 Warm 2.14 ppb 0.97 (0.92, 1.02) OR
Lee 2018 All 2.37 ppb 1.01 (0.97, 1.04) OR
Nitrogen dioxide Chiu 2015 Cold 2.14 ppb 0.84 (0.79, 0.89) OR
Chiu 2015 Warm 2.14 ppb 0.93 (0.88, 0.99) OR
Lee 2018 All 2.37 ppb 1.00 (0.97, 1.03) OR
Ozone Chiu 2015 Cold 2.14 ppb 0.98 (0.93, 1.03) OR
Lee 2018 All 2.37 ppb 1.02 (0.99, 1.05) OR
Chiu 2015 Warm 2.14 ppb 1.04 (0.99, 1.09) OR
PM10 Chiu 2015 Cold 2.14 ppb 0.86 (0.8, 0.92) OR
Chiu 2015 Warm 2.14 ppb 1.00 (0.95, 1.06) OR
Lee 2018 All 2.37 ppb 1.00 (0.97, 1.04) OR
PM2.5 Lee 2018 All 2.37 ppb 0.99 (0.96, 1.03) OR

Abbreviations: OR, odds ratio; PM2.5, particulate matter 2.5 microns or less in diameter; PM10, particulate matter 10 microns or less in diameter; ppb,
parts per billion; ppm, parts per million.

We hope that laboratory studies with animals or humans will possibly via inflammatory mediators.4,5 Most studies reported re-
further improve our understanding of the biological mechanisms un- sults using interquartile ranges for individual pollutant exposures.
derlying possible triggers including air pollution to migraine onset, Reporting relationships between exposures and outcomes using
|

15264610, 2023, 9, Downloaded from https://headachejournal.onlinelibrary.wiley.com/doi/10.1111/head.14632 by Cochrane Hungary, Wiley Online Library on [31/12/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
H​E AD​ACH​E 1217

specific intervals would have increased comparability across studies; of data: Christa Orchard, Andrea E. Portt, Peter M. Smith. Drafting
however, research in other outcomes suggests that since the inter- of the manuscript: Andrea E. Portt. Revising it for intellectual con-
quartile range represents deviation from local norms, it may be more tent: Hong Chen, Erjia Ge, Christine Lay, Christa Orchard, Andrea
physiologically relevant to triggering adverse reactions.53 E. Portt, Peter M. Smith. Final approval of the completed manuscript:
Another area that remains unclear is the effect of co-­pollution. Hong Chen, Erjia Ge, Christine Lay, Christa Orchard, Andrea E. Portt,
Future multi-­pollutant modeling would increase clarity about the in- Peter M. Smith.
dividual and additive impacts of the complex chemical mixtures in
the air we breathe. Measurement of co-­pollutants beyond those re- AC K N OW L E D G M E N T S
ported here is often challenging; however, interesting work is being We are grateful to Ms. Maggie Tiong, Manager of Library Services
carried out to estimate the reactive oxygen potential of air pollu- at the Institute for Work & Health, for her support with the search
tion,54 which could be applicable to migraine.4,6,9 strategy.
Two recent publications have briefly reviewed associations be-
tween air pollution and headache and/or migraine.13,55 The overall F U N D I N G I N FO R M AT I O N
conclusions of these papers are in agreement with our results. A Andrea E. Portt is financially supported by the University of Toronto
scoping review that included a summary of air pollution and head- Open Fellowship. Christa Orchard is financially supported by the
ache55 reported point estimates above the null value for associations University of Toronto Open Fellowship, the University of Toronto
between migraine outcomes and carbon monoxide, nitrogen diox- Data Science Institute Doctoral Fellowship, and the Ontario Gradu-
ide, ozone, PM2.5, and PM10. A recent narrative review of air pollu- ate Scholarship.
tion and headache disorders56 cited 10 studies in total in relation to
headache. Two reported on headache as opposed to or combined C O N F L I C T O F I N T E R E S T S TAT E M E N T
with migraine, and two were excluded from our review as sub-­ Dr. Christine Lay participated in advisory boards for Lundbeck,
37,38 49
analyses of a larger study. They also included seven of the 12 TEVA, Miravo, and AbbVie and has received research grants from
articles included in our review. Lundbeck, Teva, Miravo, Lilly, and AbbVie. Ms. Andrea E. Portt, Ms.
As with any review, there are potential biases in our process. The Christa Orchard, Dr. Hong Chen, Dr. Erjia Ge, and Dr. Peter M. Smith
WHO risk of bias instrument is not designed for migraine outcomes23; have no conflicts of interest to declare.
our modifications may not completely adjust for differences between
the included migraine outcomes and established air pollution-­related P R OTO C O L R EG I S T R AT I O N
outcomes such as cardiovascular morbidity or all-­cause mortality. The protocol for this review was registered with the PROSPERO In-
We were unable to properly screen an article published in Arabic, ternational prospective register of systematic reviews in November
we were not able to contact some corresponding authors, and some 2021. Systematic review registration number: CRD42021275394.
requested supplementary data were no longer available.21,22
ORCID
Andrea E. Portt https://orcid.org/0000-0002-3393-1448
CO N C LU S I O N Christa Orchard https://orcid.org/0000-0002-2868-5748
Hong Chen https://orcid.org/0000-0003-0353-3622
Overall, our systematic review provides initial evidence that in addi- Erjia Ge https://orcid.org/0000-0002-3622-3287
tion to well-­established cardiovascular and respiratory risks,16 air pol- Christine Lay https://orcid.org/0000-0002-6281-4857
lution may trigger neurologic effects in the form of migraine attacks. If Peter M. Smith https://orcid.org/0000-0001-8286-4563
air pollution does trigger migraine attacks, both policy and individual
factors would impact migraine frequency. Policy changes are the most REFERENCES
effective way to reduce air pollution exposure. Despite downward 1. James SL, Abate D, Abate KH, Abay S, Abbafati C, Abbasi N. The
trends, population-­weighted exposure to traffic-­related nitrogen diox- global, regional, and national incidence, prevalence, and years
lived with disability for 354 diseases and injuries for 195 countries
ide remains high in Europe and North America.3 On an individual level,
and territories, 1990 to 2017: a systematic analysis for the Global
the take-­home message is that if these results are confirmed in future Burden of Disease Study 2017. Lancet. 2018;392:1789-­1858.
studies, clinicians and patients may begin to consider air pollution ex- 2. World Health Organization, ed. Atlas of Headache Disorders and
posure as a potential migraine trigger. Patients may take steps to (1) Resources in the World. World Health Organization; 2011.
3. World Health Organization. WHO Air Quality Guidelines. Particulate
reduce exposure, for example by choosing active transport over driv-
Matter (PM2.5 and PM10), Ozone, Nitrogen Dioxide and Sulfur Dioxide.
ing when possible and monitoring air quality reports12; or (2) mitigate World Health Organization; 2021.
air pollution impacts through acute or preventive treatment. 4. Borkum JM. The migraine attack as a homeostatic, neuroprotective
response to brain oxidative stress: preliminary evidence for a the-
ory. Headache. 2018;58:118-­135.
AU T H O R C O N T R I B U T I O N S
5. Ye S, Li S, Ma Y, et al. Ambient NO2 exposure induces migraine in
Study concept and design: Andrea E. Portt, Peter M. Smith. Acquisition rats: evidence, mechanisms and interventions. Sci Total Environ.
of data: Christa Orchard, Andrea E. Portt. Analysis and interpretation 2022;844:157102.
|

15264610, 2023, 9, Downloaded from https://headachejournal.onlinelibrary.wiley.com/doi/10.1111/head.14632 by Cochrane Hungary, Wiley Online Library on [31/12/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
1218 HEADA​CHE​

6. Molot J, Sears M, Marshall LM, Bray RI. Neurological susceptibil- a Public Health Perspective. 1st ed. American Public Health
ity to environmental exposures: pathophysiological mechanisms in Association Press; 2013:6-­8.
neurodegeneration and multiple chemical sensitivity. Rev Environ 28. Hoffmann J, Schirra T, Lo H, Neeb L, Reuter U, Martus P. The in-
Health. 2022;37:509-­530. fluence of weather on migraine—­are migraine attacks predictable?
7. Phalen RF, Phalen RN. Important properties of air pollutants. In: Ann Clin Transl Neurol. 2015;2:22-­28.
Gartside M, Falivene C, eds. Introduction to Air Pollution Science: 29. Mehrifar Y, Pirami H, Farhang DS. The relationship between ex-
A Public Health Perspective. 1st ed. American Public Health posure to manganese in welding fumes and incidence of migraine
Association Press; 2013:41-­78. headache symptoms. Tehran Univ Med J. 2018;76:135-­141.
8. Phalen RF, Phalen RN. Human exposures to air pollutants. In: 30. Chalela JA. New onset migraine associated with a civilian burn pit.
Gartside M, Falivene C, eds. Introduction to Air Pollution Science: Mil Med. 2017;182:e1812-­e1813.
a Public Health Perspective. 1st ed. American Public Health 31. Kessler R. Dryer vents: an overlooked source of pollution? Environ
Association Press; 2013:187-­188. Health Perspect. 2011;119:A474-­A475.
9. Borkum JM. Migraine triggers and oxidative stress: a narrative re- 32. Dandona L, Dandona R, Kumar GA, et al. Nations within a na-
view and synthesis. Headache. 2016;56:12-­35. tion: variations in epidemiological transition across the states of
10. Ramage-­Morin PL, Gilmour H. Prevalence of migraine in the India, 1990–­2016 in the Global Burden of Disease Study. Lancet.
Canadian household population. Health Rep. 2014;25:10-­16. 2017;390:2437-­2460.
11. Linde M, Dahlöf C. Attitudes and burden of disease among self-­ 33. Bridges B. Fragrances and allergic reactions. J Am Board Fam Pract.
considered migraineurs—­a nation-­wide population-­based survey in 2001;14:400-­4 01.
Sweden. Cephalalgia. 2004;24:455-­465. 34. Chakravarty A, Mukherjee A, Roy D. Trigger factors in childhood
12. Carlsten C, Salvi S, Wong GWK, Chung KF. Personal strategies to migraine: a clinic-­based study from eastern India. J Headache Pain.
minimise effects of air pollution on respiratory health: advice for 2009;10:375-­380.
providers, patients and the public. Eur Respir J. 2020;55:1902056. 35. Lukina AO, Burstein B, Szyszkowicz M. Urban air pollution and
13. Garg D, Mehndiratta MM, Wasay M, Aggarwal V. Air pollution and emergency department visits related to central nervous system
headache disorders. Ann Indian Acad Neurol. 2022;25:S35-­S 40. diseases. PloS One. 2022;17:e0270459.
14. Loder EW, Cardona L, Rizzoli P. Outdoor air quality and headache: 36. Zock JP, Verheij R, Helbich M, et al. The impact of social capital,
a systematic review. Headache. 2010;50:S1-­S64. land use, air pollution and noise on individual morbidity in Dutch
15. Portt A. Scoping review of migraine and air pollution studies. neighbourhoods. Environ Int. 2018;121:453-­460.
Proceedings of the 61st Annual Scientific Meeting American 37. Szyszkowicz M, Rowe BH, Kaplan GG. Ambient sulphur dioxide ex-
Headache Society, Philadelphia 2019. posure and emergency department visits for migraine in Vancouver,
16. World Health Organization. Occupational and Environmental Canada. Int J Occup Med Environ Health. 2009;22:7-­12.
Health Team. WHO Air Quality Guidelines for Particulate Matter, 38. Szyszkowicz M, Stieb DM, Rowe BH. Air pollution and daily ED vis-
Ozone, Nitrogen Dioxide and Sulfur Dioxide: Global Update 2005: its for migraine and headache in Edmonton, Canada. Am J Emerg
Summary of Risk Assessment. Report No. WHO/SDE/PHE/ Med. 2009;27:391-­396.
OEH/06.02. World Health Organization; 2006. Accessed 39. Szyszkowicz M, Colman I, Wickham M, Rowe BH. Ambient sulphur
November 15, 2022. https://apps.who.int/iris/handl​e/10665/​ dioxide and emergency department visits for migraine and depres-
69477 sion. Int Neuropsychiatr Dis J. 2013;1:77-­88.
17. PROSPERO: International Prospective Register of Systematic Reviews. 40. Szyszkowicz M, Porada E. Ambient sulphur dioxide and female ED
University of York; 2021. https://www.crd.york.ac.uk/prosp​ero/ visits for migraine. ISRN Neurol. 2012;2012:279051.
18. Zhang Y, Yu KF. What's the relative risk? A method of correct- 41. Gasparrini A. Modelling lagged associations in environmental time
ing the odds ratio in cohort studies of common outcomes. JAMA. series data: a simulation study. Epidemiology. 2016;27:835-­8 42.
1998;280:1690-­1691. 42. Tustin AW, Hirsch AG, Rasmussen SG, Casey JA, Bandeen-­Roche K,
19. Dohoo I, Martin W, Stryhm H. In: McPike M, ed. Methods in Schwartz BS. Associations between unconventional natural gas de-
Epidemiologic Research. VER Inc; 2012. velopment and nasal and sinus, migraine headache, and fatigue symp-
20. Li W, Bertisch SM, Mostofsky E, Buettner C, Mittleman MA. toms in Pennsylvania. Environ Health Perspect. 2017;125:189-­197.
Weather, ambient air pollution, and risk of migraine headache onset 43. Chiu HF, Yang CY. Air pollution and daily clinic visits for migraine in a sub-
among patients with migraine. Environ Int. 2019;132:105100. tropical city: Taipei, Taiwan. J Toxicol Environ Health A. 2015;78:549-­558.
21. Koo YS, Kwon DY, Yang KS, Park MH. The effect of weather 44. Elser H, Rowland ST, Marek MS, et al. Wildfire smoke exposure and
and air pollution on the prevalence of headaches. Neurol Asia. emergency department visits for headache: a case-­crossover anal-
2010;15:245-­251. ysis in California, 2006–­2020. Headache. 2023;63:94-­103.
22. Dales RE, Cakmak S, Vidal CB. Air pollution and hospitalization for 45. Zmirou D, Deloraine A, Saviuc P, Tillier C, Boucharlat A, Maury
headache in Chile. Am J Epidemiol. 2009;170:1057-­1066. N. Short-­term health effects of an industrial toxic waste landfill: a
23. WHO Global Air Quality Guidelines Working Group on Risk of retrospective follow-­up study in Montchanin, France. Arch Environ
Bias Assessment. Risk of Bias Assessment Instrument for Systematic Health. 1994;49:228-­238.
Reviews Informing WHO Global Air Quality Guidelines. World Health 46. Lee H, Myung W, Cheong HK, et al. Ambient air pollution expo-
Organisation; 2020. sure and risk of migraine: synergistic effect with high temperature.
24. Chow EJ, Rolfes MA, O'Halloran A, et al. Respiratory and nonrespi- Environ Int. 2018;121:383-­391.
ratory diagnoses associated with influenza in hospitalized adults. 47. Mukamal KJ, Wellenius GA, Suh HH, Mittleman MA. Weather and air
JAMA Netw Open. 2020;3:e201323. pollution as triggers of severe headaches. Neurology. 2009;72:922-­
25. Kwong JC, Schwartz KL, Campitelli MA, et al. Acute myocardial 927. Erratum appears in Neurology. 2009;73(17):1428.
infarction after laboratory-­confirmed influenza infection. N Engl J 48. Elser H, Morello-­Frosch R, Jacobson A, et al. Air pollution, meth-
Med. 2018;378:345-­353. ane super-­emitters, and oil and gas wells in northern California: the
26. Behrouzi B, Araujo Campoverde MV, Liang K, et al. Influenza vacci- relationship with migraine headache prevalence and exacerbation.
nation to reduce cardiovascular morbidity and mortality in patients Environ Health. 2021;20:45.
with COVID-­19. J Am Coll Cardiol. 2020;76:1777-­1794. 49. Szyszkowicz M, Kaplan G, Grafstein E, Rowe B. Emergency de-
27. Phalen RF, Phalen RN. Introduction to air pollution science. In: partment visits for migraine and headache: a multi-­city study. Int J
Gartside M, Falivene C, eds. Introduction to Air Pollution Science: Occup Med Environ Health. 2009;22:235-­242.
|

15264610, 2023, 9, Downloaded from https://headachejournal.onlinelibrary.wiley.com/doi/10.1111/head.14632 by Cochrane Hungary, Wiley Online Library on [31/12/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
H​E AD​ACH​E 1219

50. Adgate JL, Goldstein BD, McKenzie LM. Potential public health haz- 56. Garg AX, Hackam D, Tonelli M. Systematic review and meta-­
ards, exposures and health effects from unconventional natural gas analysis: when one study is just not enough. Clin J Am Soc Nephrol.
development. Environ Sci Technol. 2014;48:8307-­8320. 2008;3:253-­260.
51. Chen CC, Tsai SS, Yang CY. Association between fine partic- 57. Page MJ, McKenzie JE, Bossuyt PM, et al. The PRISMA 2020 state-
ulate air pollution and daily clinic visits for migraine in a sub- ment: an updated guideline for reporting systematic reviews. BMJ.
tropical city: Taipei, Taiwan. Int J Environ Res Public Health. 2021;372:n71.
2015;12:4697-­4708.
52. Riddle EJ. Effect of wildfire smoke on primary headache disorders
remains unclear. Headache. 2023;63:4-­5. S U P P O R T I N G I N FO R M AT I O N
53. Guo Y, Gasparrini A, Armstrong B, et al. Global variation in the ef- Additional supporting information can be found online in the
fects of ambient temperature on mortality: a systematic evaluation.
Supporting Information section at the end of this article.
Epidemiology. 2014;25:781-­789.
54. Weichenthal S, Shekarrizfard M, Kulka R, et al. Spatial variations
in the estimated production of reactive oxygen species in the ep-
ithelial lung lining fluid by iron and copper in fine particulate air
pollution. Environ Epidemiol. 2018;2:e020. How to cite this article: Portt AE, Orchard C, Chen H, Ge E,
55. Louis S, Carlson AK, Suresh A, et al. Impacts of climate change Lay C, Smith PM. Migraine and air pollution: A systematic
and air pollution on neurologic health, disease, and practice: a
review. Headache. 2023;63:1203-1219. doi:10.1111/
scoping review. Neurology. 2022;100:474-­483. doi:10.1212/
WNL.0000000000201630 head.14632

You might also like