University of Southern California Dissertations and Theses

Examining exposure to extreme heat and air pollution and its effects on all-cause, cardiovascular, and respiratory mortality in California: effect modification by the social deprivation index

(USC Thesis Other)

Examining exposure to extreme heat and air pollution and its effects on all-cause, cardiovascular, and respiratory mortality in California: effect modification by the social deprivation index

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content

Examining Exposure to Extreme Heat and Air Pollution and its Effects on All-
Cause, Cardiovascular, and Respiratory Mortality in California: Effect
Modification by the Social Deprivation Index

by

Zainab Hasan

A Thesis Presented to the
FACULTY OF THE USC KECK SCHOOL OF MEDICINE
UNIVERSITY OF SOUTHERN CALIFORNIA
In Partial Fulfillment of the
Requirements for the Degree
MASTER OF SCIENCE
(APPLIED BIOSTATISTICS AND EPIDEMIOLOGY)

August 2022

Copyright 2022 Zainab Hasan
ii

Dedication
I would like to dedicate this study to my parents. To my Mamma and Abba, Samia Majid
Hasan and Syed Neyaz Hasan, for their endless sacrifice and resilience that provided me with the
opportunity to pursue higher education in a field I am passionate about. Thank you for teaching
me to be curious, and for providing me unwavering support to do so. To my Nanna and Nanu,
Niloofer Husain and Syed Husain Majid; to my Dadda and Dadajaan, Rafiqun Nisa (Sitara) and
Syed Mohammad Zaki Hasan. Thank you for teaching me the value of education and the
responsibility that comes with it – to leave the world a better place than I found it. Your presence
and memory will always be an inspiration. I am here because of the struggles, sacrifices, values,
hopes, and dreams of those who came before me. It will continue to guide me in my career and
in my life. I am forever grateful and indebted.
iii

Acknowledgements
Firstly, I would like to express great appreciation for my committee chair and
mentor Dr. Erika Garcia. For the past year and a half you have been an outstanding mentor,
taking an aspiring researcher through the world of environmental health and epidemiology
amidst a pandemic. Thank you for everything you have taught me. I would like to specially
acknowledge my committee members Dr. Rob McConnell and Dr. William Gauderman. Thank
you for your expertise and guidance throughout the development of my thesis. Dr. Mostafijur
Rahman – thank you for your guidance and resourcefulness in this research. I would like to
acknowledge the EH MATTERS program and its directors Dr. Jill Johnston and Edward Avol.
Thank you for an incredibly rewarding and educational internship experience that guided me to
this project. I would also like to acknowledge all of my advisors and professors in the program
that helped me along the way. A special shoutout to my amazing friends at USC and from
childhood who are like a second family and supported me generously throughout this process.
Finally, I would like to acknowledge my siblings Umar and Ameena for years of friendship,
loyalty, and laughs. I only hope to be as much of an inspiration to you as you are to me.

iv

Table of Contents

Dedication ....................................................................................................................................... ii
Acknowledgements ........................................................................................................................ iii
List of Tables ...................................................................................................................................v
List of Figures ..................................................................................................................................v
Abstract ........................................................................................................................................ viii
Chapter 1: Background ....................................................................................................................1
Chapter 2: Methods ..........................................................................................................................4
Exposure and Outcome Data 4
Social Deprivation Index 5
Statistical Analysis 7
Chapter 3: Results ..........................................................................................................................10
Descriptive Statistics 10
All-Cause Mortality 12
Cardiovascular Mortality 15
Respiratory Mortality 18
Ad Hoc Analysis 21
Chapter 4: Discussion ....................................................................................................................24
References ......................................................................................................................................28
Supplementary Material .................................................................................................................31

v

List of Tables
Table 1. Domain and Variable Description for the Social Deprivation Index (SDI)

Table 2. Table 2. N observations for each SDI category

Table 3. Table 3. Baseline Characteristics of the Study Population (N = 1,514,292)

List of Figures
Figure 1. Case-Crossover Design Visualization
Figure 2. Forest plot of odds ratios (ORs) including 95% confidence intervals (CIs) and p-
interaction values with binary categorization of SDI for the association between each exposure
(heat only, combined heat and PM2.5, and PM2.5 only) on all-cause mortality risk at the 90
th
, 95
th
,
and 97
th
percentiles with lags 0 and 1.
Figure 3. Forest plot of ORs including 95% CIs and p-interaction values with ternary
categorization of SDI for the association between each exposure (heat only, combined heat and
PM2.5, and PM2.5 only) on all-cause mortality risk at the 90
th
, 95
th
, and 97
th
percentiles with lags 0
and 1.
Figure 4. Forest plot of ORs including 95% CIs and p-interaction values with binary
categorization of SDI for the association between each exposure (heat only, combined heat and
PM2.5, and PM2.5 only) on cardiovascular mortality risk at the 90
th
, 95
th
, and 97
th
percentiles with
lags 0 and 1.
Figure 5. Forest plot of ORs including 95% CIs and p-interaction values with ternary
categorization of SDI for the association between each exposure (heat only, combined heat and
PM2.5, and PM2.5 only) on cardiovascular mortality risk at the 90
th
, 95
th
, and 97
th
percentiles with
lags 0 and 1.
vi

Figure 6. Forest plot of ORs including 95% CIs and p-interaction values with binary
categorization of SDI for the association between each exposure (heat only, combined heat and
PM2.5, and PM2.5 only) on respiratory mortality risk at the 90
th
, 95
th
, and 97
th
percentiles with
lags 0 and 1.
Figure 7. Forest plot of ORs including 95% CIs and p-interaction values with ternary
categorization of SDI for the association between each exposure (heat only, combined heat and
PM2.5, and PM2.5 only) on respiratory mortality risk at the 90
th
, 95
th
, and 97
th
percentiles with
lags 0 and 1.
Figure 8. Forest plot of ORs including 95% CIs and p-interaction values with binary
categorization of SDI for the association between each exposure (heat only, combined heat and
PM2.5, and PM2.5 only) on respiratory mortality risk under age 65 at the 90
th
, 95
th
, and 97
th

percentiles with lags 0 and 1.
Figure 9. Forest plot of ORs including 95% CIs and p-interaction values with ternary
categorization of SDI for the association between each exposure (heat only, combined heat and
PM2.5, and PM2.5 only) on respiratory mortality risk under age 65 at the 90
th
, 95
th
, and 97
th

percentiles with lags 0 and 1.
Supplementary Figure 1. Histogram of Age at Death for High SDI
Supplementary Figure 2. Histogram of Age at Death for Medium SDI
Supplementary Figure 3. Histogram of Age at Death for Low SDI
Supplementary Figure 4. Estimated excess risk (ER) of all-cause mortality (and 95% CI) on A)
co-occurrence of extreme maximum temperature and PM2.5 days compared to non-extreme days;
B) co-occurrence of extreme minimum temperature and PM2.5 days compared with non-extreme
days. Results are shown for same day exposure (lag-0) to 3 days prior exposure (lag-3).
vii

Supplementary Figure 5. Estimated excess risk (ER) of cardiovascular mortality (and 95% CI)
on A) co-occurrence of extreme maximum temperature and PM2.5 days compared to non-extreme
days; B) co-occurrence of extreme minimum temperature and PM2.5 days compared with non-
extreme days. Results are shown for same day exposure (lag-0) to 3 days prior exposure (lag-3).
Supplementary Figure 6. Estimated excess risk (ER) of respiratory mortality (and 95% CI) on
A) co-occurrence of extreme maximum temperature and PM2.5 days compared to non-extreme
days; B) co-occurrence of extreme minimum temperature and PM2.5 days compared with non-
extreme days. Results are shown for same day exposure (lag-0) to 3 days prior exposure (lag-3).
Supplementary Figures 7-9. Missing Data Flowcharts
viii

Abstract
Background: The climate crisis is increasing the frequency of extreme heat and air pollution
events. Previous research has found associations between heat and air pollution with increased
risk of mortality, especially for cardiovascular and respiratory mortality. Several factors
including access to healthcare, adaptation resources, and population mobility lead to increased
vulnerability for socioeconomically marginalized communities to climate change events
including extreme heat and air pollution. This study aimed to investigate whether increased
neighborhood-level social deprivation modifies the effect of co-exposure to extreme heat and
PM2.5 pollution on mortality risk.
Methods: A time-stratified case-crossover study design was used to assess the effect of daily
exposure to PM2.5 pollution and heat on all-cause, cardiovascular, and respiratory mortality in
California from 2014-2019 based on residential addresses. Air pollution data was assessed using
inverse distance-squared weighted observations from up to 4 nearby US EPA air monitoring
locations. Heat data was retrieved from a spatiotemporal model developed by the University of
Idaho. Extreme PM2.5 and heat on the date of death (lag 0) and the day before (lag 1) were used
as the exposures. Socioeconomic burden was evaluated using the census tract level Social
Deprivation Index (SDI) with binary and ternary categories. Conditional logistic regression was
used to analyze the associations between extreme heat and PM2.5 air pollution exposure and
mortality risk with effect modification by SDI.
Results: We found no consistent effect modification by SDI on the combined or separate effects
of extreme heat and PM2.5 pollution exposure and mortality with lag 0 and 1 days. Similar null
results were found for cardiovascular and respiratory mortality risk.
ix

Conclusion: We observed no pattern of effect modification by neighborhood-level social
deprivation on the effect of heat and air pollution exposure on mortality risk in California, a state
that experiences high levels of heat and air pollution especially with the ongoing climate crisis.
This is consistent with other studies using neighborhood level SES; further analyses need to be
conducted to explore this socioeconomic burden more.

1

Chapter 1: Background
Climate change is emerging as the world’s greatest public health challenge. There is
increasing evidence that the climate crisis will increase the intensity and frequency of heat waves
around the world (Tuholske et al. 2021). Many studies have established that exposure to high
temperatures is associated with increased mortality rates; more specifically, heat waves are
associated with increased mortality from cardiovascular and respiratory diseases (Zanobetti and
Schwartz 2008, Lay et al. 2021). Extreme heat reduces the body’s ability to thermoregulate and
produce sweat, leading to heat stroke. High temperatures may also strain the cardiovascular
system by increasing cardiac output and consequently increasing risk for heart failure and stroke
(Cheshire 2016). Continued exposure to extreme heat may damage vital organs such as the heart
and lungs, weakening the body’s adaptive responses to heat (Basu and Samet 2002).
Increased particulate matter (PM) air pollution is also a key factor in the global climate
crisis. Wildfires, especially in the state of California, are becoming more frequent and increasing
PM2.5 pollution in urban and rural settings. It is well-established that wildfire related PM2.5
pollution has significant effects on respiratory health (Yang et al. 2020), and emerging evidence
is showing that it has significant effects on cardiovascular health as well (Bevan et al. 2021).
PM2.5 pollution is shown to cause deterioration of lung function and exacerbation of respiratory
conditions such as asthma (Zheng et al. 2015); it is also demonstrating increased risk for cardiac
arrest (Jones et al. 2020).
While there are several studies discussing the effects of heat and the effects of air
pollution on mortality, there are very few studies that have looked at the co-exposure to extreme
heat and air pollution. There is some evidence that extreme heat and air pollution combined can
have larger effects compared to the exposures occurring separately (Ji et al. 2020). The
2

interactive effects of extreme heat and air pollution are a key question in evaluating the health
effects and mortality risk associated with the climate crisis. This thesis will evaluate the joint
exposure of extreme heat and PM2.5 pollution along with their separate exposures for the effect
on all-cause, cardiovascular, and respiratory mortality with effect modification by the Social
Deprivation Index – a score derived by the American Community Survey Data that combines
seven key socioeconomic variables by census tract.
There is a large body of evidence showing that socioeconomically deprived communities
have much higher exposure to particulate matter air pollution and heat. Exposure to extreme heat
and air pollution is exacerbated by the urban heat island effect and consequent lack of
greenspace, which disproportionately affects socioeconomically deprived urban populations
(Tuholske et al. 2021). Previous studies also note that mortality due to exposure from heat and
from air pollution disproportionately affects underserved populations due to limitations regarding
access to healthcare, resources to prepare for and adjust to the environment, and population
mobility (Li et al. 2019, Tibaukuu et al. 2018).
Furthermore, socioeconomic deprivation can contribute to poor baseline health (Wang
and Geng 2019), which makes underserved populations especially vulnerable to mortality risk
from heat and air pollution. Conditions that are more prevalent in communities with low
socioeconomic status (SES) due to historic marginalization include asthma, chronic obstructive
pulmonary disease (COPD), lung cancer, diabetes, and heart disease (Medina-Ramon et al. 2006,
Zhao et al. 2019). Living with these chronic health conditions can make one more susceptible to
health effects and mortality risk from climate change (Witt et al. 2015). For example, if a child
has asthma and they do not have access to healthcare or cannot afford medications or supportive
care, their asthma may go untreated and worsen as the child gets older. This can lead to reduced
3

lung function; exposure to heat and air pollution will exacerbate these problems and increase
mortality risk (Zheng et al. 2015). Socioeconomically deprived communities tend to live in food
deserts (Walker et al. 2010), which means they do not have as much access to healthy and fresh
foods compared to high SES communities. This can lead to conditions such as diabetes and heart
disease. Air pollution promotes cardiovascular disease risk factors and increases systemic
inflammation – it is estimated that half of the health burden of air pollution is through
cardiovascular disease (Tibaukuu et al. 2018). People living with diabetes and cardiovascular
disease can therefore be more sensitive to extreme air pollution and experience increased
symptoms. Underserved populations also have decreased protection from heat, with less access
to air conditioning and greenspace (Mears et al. 2020). This can lead to an increased risk of
mortality from heat-related illness (Tuholske et al. 2021).
The projection that heatwaves and extreme air pollution events will rapidly increase and
intensify (Rossati 2017) makes it all the more important to enact change that will prevent further
climate degradation. It is equally important to understand the health effects from these events to
determine proper prevention and adaptation methods that serve the unique needs of each
community. Understanding the effect that historic social and economic marginalization has on
the mortality risks from extreme heat and air pollution will provide needed evidence for policy
initiatives to support underserved communities and promote environmental health justice.

4

Chapter 2: Methods
Exposure and Outcome Data
All-cause, cardiovascular, and respiratory mortality data was obtained from the California
Department of Public Health’s Vital Statistics. All-cause mortality includes all observed deaths,
cardiovascular events include ICD-10 codes I00-I99, and respiratory events include ICD-10
codes J00-J99. During the study period, there were 1,514,292 all-cause, 492,513 cardiovascular,
and 139,116 respiratory deaths. Individual-level traits such as age, sex, race/ethnicity, and
education level were retrieved from the death certificate data.
Hourly and daily observations of air pollution data was retrieved from the US
Environmental Protection Agency’s Air Quality System. This system has over 150 stations
throughout California. Pollutant concentrations at each census block group were estimated by
calculating inverse distance-squared weighted observations from up to 4 nearby monitoring
locations, which were then scaled up to the census tract level using population-weighting
averaging. 24-hour average PM2.5 concentrations were used for the analysis; estimates for PM2.5
exposure was assigned for those of the same day (lag 0) and 1 day before the exposure (lag 1)
Heat and relative humidity data from a spatiotemporal model developed at the University
of Idaho was used for this analysis (Abatzoglou, 2013). Hourly temporal data using 12 km grids
and spatial data with 800 m resolution was combined to form 4 km grids that accounted for
weather factors such as wind, local topography, and downward shortwave radiation. Correlations
for validation of temperature was high, with median correlation of 0.94 (Abatzoglou, 2013). The
maximum temperature for each day was used for heat exposure and it was assigned for lag 0 and
lag 1 days before each date used in analysis. Decedents’ residential addresses are used for
temperature and air pollution exposure assessments.
5

We defined exposure to extreme heat and PM2.5 pollution using 90, 95, and 97
th

percentile thresholds. Extreme heat is defined by a day with daily temperature that is above the
specified percentile of daily temperature for 2014 – 2019 for that specific census tract. Extreme
PM2.5 pollution is defined by a day with daily mean PM2.5 above the specified percentile of daily
mean PM2.5 for 2014 – 2019 for all census tracts in California. Extreme and PM2.5 pollution will
be defined by a day with both temperature and mean PM2.5 pollution above the specified
percentile values from 2014 – 2019. We are interested in co-exposure to these extremes, so we
defined exposure as a day with extreme PM only, extreme heat only, both extreme PM and heat,
or neither extreme PM nor heat. Exposure was included in our models as a categorical variable
with these four levels, with the neither extreme group used as the referent category.
Social Deprivation Index
The Social Deprivation Index (SDI) was used as the variable to assess effect modification
by socioeconomic status of the effect of exposure to extreme heat and air pollution on all-cause,
cardiovascular, and respiratory mortality. SDI combines 7 indicators from the American
Community Survey: percent living in poverty, percent with less than 12 years of education,
percent single parent household, percent living in rented housing units, percent living in
overcrowded housing units, percent of households without a car, and percent non-employed
adults under 65 years of age. SDI scores range from 1 to 100 with a higher SDI score indicating a
more socially deprived the neighborhood. SDI scores are available at the census tract level and
they were linked to the study population using the decedents’ residential addresses. Table 1
describes the SDI variable in greater detail.
6

Table 1. Domain and Variable Description for the Social Deprivation Index (SDI)
Domain Variable
Income Percent population less than 100% FPL (population under 0.99 /total
population)
Education Percent population 25 years or more with less than 12 years of education
(population with less than high school diploma or 12 years of education/total
population)
Employment Percent Non-employed (not in labor force + unemployed) / (civilian + not in
the labor force) for the population 16-64 years
Housing Percent population living in renter-occupied housing units (Renter occupied
housing units/ (Owner-occupied housing units + Renter occupied housing
units))
Percent population living in crowded housing units (Tenure by Occupants Per
Room – a population with ≥ 1.01 occupants per room in Owner-occupied
housing units and Renter occupied housing units) / total population
Household
Characteristics
Percent single-parent households with dependents < 18 years (total single-
parent households (male and female) with dependents <18 years)/total
population)
Transportation Percent population with no car (population with no vehicle available/total
population)
Demographics
Percent high needs population – (population under 5 years of age + women
between the ages of 15-44 years + everyone 65 years and over)/total
population
* obtained from 2015-2019 American Community Survey 5-Year Summary File (Sequence
Number Table ID Lookup Table)

SDI was analyzed as a binary variable and as a ternary variable. These variables were
created by calculating quantile groups. For analysis with SDI as a binary variable, high SDI
includes scores above the 50
th
percentile among decedents with the outcome of interest (all-
cause, cardiovascular, or respiratory event mortality); low SDI includes scores at and below the
50
th
percentile. For analysis with SDI as a ternary variable, SDI was divided into tertiles; high
SDI includes scores above the 66.7
th
percentile of scores, medium SDI includes scores above the
33.3
rd
percentile and at and below the 66.7
th
percentile, and low SDI includes scores at and
below the 33.3
rd
percentile. Details of the range of SDI scores in each category along with the
7

number of observations for each category by mortality outcome are shown in Table 2. Missing
data flowcharts are detailed in the supplementary materials figures 7-9.
Table 2. N observations for each SDI category

All-Cause Mortality N = 1,514,292
Binary SDI
Low SDI [1,60] 760,424
High SDI (60, 100] 753,715
Ternary SDI
Low SDI [1,43] 505,777
Medium SDI (43,76] 509,799
High SDI (76, 100] 498,563
Cardiovascular Mortality N = 492,513
Binary SDI
Low SDI [1,60] 246,294
High SDI (60, 100 246,157
Ternary SDI
Low SDI [1,44] 163,227
Medium SDI (44,76] 166,756
High SDI (76, 100] 162,468
Respiratory Mortality N = 139,116
Binary SDI
Low SDI [1,61] 69,045
High SDI (61, 100] 70,061
Ternary SDI
Low SDI [1,44] 45,267
Medium SDI (44,76] 47,626
High SDI (76, 100] 46,213

Statistical Analysis
Descriptive statistics for exposure, outcome, and individual-level population data is
calculated as N counts and percent frequency for categorical variables and means with standard
deviation for continuous variables. This analysis used a time-stratified case-crossover design to
estimate all-cause, cardiovascular, and respiratory mortality risk on days with exposure to
extreme PM2.5 air pollution and/or extreme heat compared with non-extreme days in the state of
California from 2014-2019. A case-crossover design is a form of a case-control design, with
8

subjects being matched as their own controls. Control for individual-level, time-independent
factors, such as sex and race/ethnicity, is embedded within this design. In this study, a case is a
day where a person’s death has occurred, and a control is a day where that same person’s death
did not occur. For each case day (where death has occurred), there are about 4 control days on
the same day of the week within the same month. Then, these case days and controls days are
compared by level of exposure using conditional logistic regression on the outcomes of all-cause,
cardiovascular, and respiratory even mortality. Figure 1 is a visualization of this design, showing
an example of a case day and its respective control days.
Figure 1. Case-Crossover Design Visualization

This method also controls for time and season trends because it compares exposure levels
between the same days within each month of each year.
As mentioned previously, for the statistical analysis of this data PM2.5 and heat exposure
were evaluated as exposures above the 90
th
, 95
th
, and 97
th
percentile. These thresholds were
determined to assess the effect of extreme exposures on mortality, while maintaining enough
dates for power in the analysis. For this reason, we did not include the 99
th
percentile exposure
threshold. The exposure percentile variable had four categories: neither extreme heat nor extreme
PM2.5 pollution, extreme heat only, extreme PM2.5 pollution only, or extreme heat and extreme
PM2.5 pollution. Separate models were fitted for lag 0 exposures and lag 1 exposures. All models
were adjusted for relative humidity.
Effect modification by SDI was assessed by adding an interaction term with the indicator
variables for extreme exposure in the model. A main effect for SDI is not included in the model
9

because in a case-crossover design the cases serve as their own control, so there is no variability
for SDI of each decedent; therefore, the main effect of SDI on mortality risk cannot be estimated.
Below is the equation for the statistical model for this analysis:
𝐿𝐿𝐿𝐿 𝐿𝐿𝐿𝐿𝐿𝐿 (𝑌𝑌 𝑖𝑖 𝑖𝑖 ) = 𝛼𝛼 + 𝐵𝐵 1
𝑍𝑍 𝑝𝑝 𝑖𝑖𝑖𝑖
+ 𝐵𝐵 2
𝑍𝑍 ℎ
𝑖𝑖𝑖𝑖
+ 𝐵𝐵 3
𝑍𝑍 ℎ 𝑝𝑝 𝑖𝑖𝑖𝑖
+ 𝐵𝐵 4
𝑍𝑍 𝑝𝑝 𝑖𝑖𝑖𝑖
∗ 𝑀𝑀 𝑖𝑖 + 𝐵𝐵 5
𝑍𝑍 ℎ
𝑖𝑖𝑖𝑖
∗ 𝑀𝑀 𝑖𝑖 + 𝐵𝐵 6
𝑍𝑍 ℎ 𝑝𝑝 𝑖𝑖𝑖𝑖
∗ 𝑀𝑀 𝑖𝑖 + 𝑛𝑛𝑛𝑛 (𝑍𝑍 𝑟𝑟 ℎ
𝑖𝑖𝑖𝑖
)
Y = outcome
i = person
j = date
Zh = extreme heat only
Zhp = extreme heat and PM2.5 pollution
Zp = extreme PM2.5 only
Zrh = relative humidity - natural cubic spline with 3 degrees of freedom
Mi = SDI score effect modifier
Separate models were run for analysis of effect modification by SDI score with SDI as a binary
variable and then as a ternary variable.
All analyses were performed in R (version 4. 1.2) and significance was declared if the p-value for
a given outcome-exposure association was less than 0.05.

10

Chapter 3: Results
Descriptive Statistics
Descriptive statistics for the study population along with means and standard deviations
for exposures and SDI score are detailed in Table 3. Table 3 presents data on the entire study
population, consisting of 1,514,292 all-cause deaths, by levels of SDI score. About 33% of these
deaths are cardiovascular events (n = 492512) and about 9% of these deaths are respiratory
events (n = 139116). This ratio is relatively similar across all levels of SDI. The average age at
death for this population is 74 years old, with about 57% of ages at death being above 75 years
old. However, as the level of SDI score increases, the average age at death decreases with a
difference of 8 years between the low SDI and high SDI group. The average maximum daily
temperature increases slightly with increasing SDI score. Average daily PM2.5 concentration
exposure follows a similar pattern. Education level also decreases with increasing levels of SDI
score, with 32% less than high school education in the high SDI group and 10.2% less than high
school education in the low SDI group. The percentage of non-Hispanic whites decreases
significantly from the low SDI group to the high SDI group. Frequency of 90
th
, 95
th
, and 97
th

percentile exposures to neither heat nor PM2.5 air pollution decreases slightly with increasing
SDI levels. All three percentile exposures to heat only and to combined heat and PM2.5 air
pollution remain relatively constant across levels of SDI. However, 90
th
, 95
th
, and 97
th
percentile
exposure to PM2.5 air pollution increases with increasing level of SDI; frequencies increase by
about 4, 3, and 2% respectively from the low SDI group to the high SDI group.

11

Table 3. Baseline Characteristics of the Study Population (N = 1,514,292)

Variable - Mean (SD) All Low SDI
(n = 505777)
Medium SDI
(n = 509799)
High SDI
(n = 498563)
Age – years 74.5 (18.4) 78.0 (16.5) 74.9 (18.0) 70.4 (19.7)
Maximum Temperature - °C 23.8 (7.1) 23.5 (6.9) 23.6 (7.1) 24.4 (7.2)
PM2.5 Concentration - µg/m³ 10.3 (8.2) 9.7 (7.4) 10.2 (8.3) 11.1 (8.8)
Social Deprivation Index Score 57.7 (28.3) 24.0 (11.8) 60.4 (9.5) 89.1 (7.0)
Variable - No. (% Frequency) All Low SDI Medium SDI High SDI
Sex
Female 391 (64.7) 250716 (49.6) 249324 (48.9) 233309 (46.8)
Male 213 (35.3) 255056 (50.4) 260466 (51.1) 265237 (53.2)
Unknown 30 (0.0) 5 (0.0) 8 (0.0) 17 (0.0)
Age Group
0 - 25 33435 (2.2) 6962 (1.4) 10136 (2.0) 16334 (3.3)
26 - 50 104134 (6.9) 22897 (4.5) 33152 (6.5) 48069 (9.6)
51 - 75 514804 (34.0) 143166 (28.3) 171961 (33.7) 199635 (40.0)
> 75 861828 (56.9) 332727 (65.8) 294517 (57.8) 234492 (47.0)
Education Level
Less than high school 303230 (20.0) 51403 (10.2) 94068 (18.5) 157745 (31.6)
High school graduate 520270 (34.4) 162026 (32.0) 184028 (36.1) 174167 (34.9)
Some college 249177 (16.5) 91542 (18.1) 87914 (17.2) 69689 (14.0)
Associate Degree 87887 (5.8) 33406 ( 6.6) 31198 (6.1) 23274 (4.7)
Bachelor’s Degree 196181 (13.0) 95622 (18.9) 63979 (12.5) 36548 (7.3)
Graduate Degree 121087 (8.0) 65953 (13.0) 37431 (7.3) 17690 (3.5)
Unknown 36460 (2.4) 5825 (1.2) 11181 (2.2) 19450 (3.9)
Hispanic Ancestry
Cuban 3610 (0.2) 938 (0.2) 1164 (0.2) 1508 (0.3)
Mexican 200338 (13.2) 25909 (5.1) 60592 (11.9) 113828 (22.8)
Not Hispanic 1220200 (80.6) 461766 (91.3) 420077 (82.4) 338216 (67.8)
Other 82872 (5.5) 15541 (3.1) 25403 (5.0) 41925 (8.4)
Puerto Rican 4563 (0.3) 1136 (0.2) 1648 (0.3) 1779 (0.4)
Unknown 2709 (0.2) 487 (0.1) 915 (0.2) 1307 (0.3)
Race/Ethnicity
NH White 928651 (61.3) 388657 (76.8) 327051 (64.2) 212824 (42.7)
NH Black 114044 (7.5) 16967 (3.4) 30801 (6.0) 66266 (13.3)
American Indian/Alaska Native 8049 (0.5) 1516 (0.3) 3037 (0.6) 3495 (0.7)
Asian/Hawaiian/Pacific Islander 154882 (10.2) 50115 (9.9) 53910 (10.6) 50847 (10.2)
Hispanic 289246 (19.1) 42994 (8.5) 88058 (17.3) 158182 (31.7)
Multiracial/Other/ Unknown 19420 (1.3) 5528 (1.1) 6942 (1.4) 6949 (1.4)
90
th
Percentile Exposure
Neither 1229593 (81.2) 419917 (83.0) 416192 (81.6) 393360 (78.9)
Heat Only 133196 (8.8) 44826 (8.9) 44691 (8.8) 43671 (8.8)
Heat and PM2.5 11306 (0.7) 3288 (0.7) 3903 (0.8) 4114 (0.8)
PM2.5 140197 (9.3) 37746 (7.5) 45013 (8.8) 57418 (11.5)
95
h
Percentile Exposure
Neither 1366050 (90.2) 462294 (91.4) 461715 (90.6) 441901 (88.6)
Heat Only 69992 (4.6) 23570 (4.7) 23421 (4.6) 22996 (4.6)
Heat and PM2.5 2793 (0.2) 888 (0.2) 961 (0.2) 944 (0.2)
PM2.5 75457 (5.0) 19025 (3.8) 23702 (4.6) 32722 (6.6)
97
th
Percentile Exposure
12

Neither 1423102 (94.0) 479652 (94.8) 480425 (94.2) 462878 (92.8)
Heat Only 42654 (2.8) 14319 (2.8) 14217 (2.8) 14116 (2.8)
Heat and PM2.5 1365 (0.1) 475 (0.1) 472 (0.1) 418 (0.1)
PM2.5 47171 (3.1) 11331 (2.2) 14685 (2.9) 21151 (4.2)
Cardiovascular Cases 492512 (32.5) 163227 (32.3) 166756 (32.7) 162468 (32.6)
Respiratory Cases 139116 (9.2) 45267 (8.9) 47626 (9.3) 46213 (9.3)
*NH = non-Hispanic
**Low SDI = [1,43], Medium SDI = (43,76], High SDI = (76, 100]

All-Cause Mortality
At the 90
th
, 95
th
, and 97
th
percentile exposures, the association of exposure to extreme
heat only and exposure to combined extreme heat and PM2.5 pollution showed no consistent
trend of effect modification by SDI; the results are null with only spurious, marginally
statistically significant p-values for interaction between the exposures and SDI. The effect of 90
th

percentile exposure to extreme PM2.5 only with odds of mortality shows a faint trend of effect
modification by level of SDI. For binary categorization of SDI, odds of mortality risk with
exposure to extreme PM2.5 at the 90th percentile lag 0 days was elevated for decedents who were
within the high SDI group (OR = 1.019 [1.009, 1.028]) compared to the low SDI group (OR =
1.005 [0.995, 1.016]); however, this positive interaction between SDI and 90
th
percentile
exposure to extreme PM2.5 was only marginally statistically significant (p=0.070). Similar results
were observed for lag 1 day (p-interaction=0.073). Similar results were also observed for the
same analysis with ternary SDI for lag 1 day, with a marginally statistically significant p-value
for interaction between low SDI and high SDI groups (p-interaction=0.072). Interaction in the
medium SDI group for 90
th
percentile PM2.5 exposure lag 0 was also marginally statistically
significant (p-interaction=0.060). The same analyses at the 95
th
and 97
th
percentile exposures to
PM2.5 (lag 0 and 1 days) showed little evidence of effect modification by SDI group, with
spurious marginally statistically significant p-values for interaction and no consistent trend.
Results with binary SDI are shown in figure 2; results with ternary SDI are show in figure 3.
13

Figure 2. Forest plot of odds ratios (ORs) including 95% confidence intervals (CIs) and p-
interaction values with binary categorization of SDI for the association between each exposure
(heat only, combined heat and PM2.5, and PM2.5 only) on all-cause mortality risk at the 90
th
,
95
th
, and 97
th
percentiles with lags 0 and 1.

14

Figure 3. Forest plot of ORs including 95% CIs and p-interaction values with ternary
categorization of SDI for the association between each exposure (heat only, combined heat and
PM2.5, and PM2.5 only) on all-cause mortality risk at the 90
th
, 95
th
, and 97
th
percentiles with lags
0 and 1.

15

Cardiovascular Mortality
There was no evidence of effect modification by level of SDI score (binary and ternary)
on the effect of exposure to extreme heat and PM2.5 pollution, exposure to extreme heat only, or
exposure to extreme PM2.5 pollution only on cardiovascular mortality risk. This was the case for
all percentile level exposures. There was one statistically significant positive interaction between
the medium SDI level and 90
th
percentile exposure to combined extreme heat and PM2.5
pollution lag 1 day (OR:1.13[1.06, 1.20], p-interaction=0.028). However, this statistically
significant interaction was not part of a larger trend. Results are summarized in figure 4 for
binary SDI and figure 5 for ternary SDI.

16

Figure 4. Forest plot of ORs including 95% CIs and p-interaction values with binary
categorization of SDI for the association between each exposure (heat only, combined heat and
PM2.5, and PM2.5 only) on cardiovascular mortality risk at the 90
th
, 95
th
, and 97
th
percentiles
with lags 0 and 1.

17

Figure 5. Forest plot of ORs including 95% CIs and p-interaction values with ternary
categorization of SDI for the association between each exposure (heat only, combined heat and
PM2.5, and PM2.5 only) on cardiovascular mortality risk at the 90
th
, 95
th
, and 97
th
percentiles
with lags 0 and 1.

18

Respiratory Mortality
At the 90
th
, 95
th
, and 97
th
percentile level, exposure to combined heat and PM2.5 pollution
and exposure to only PM2.5 pollution did not show any statistically significant trends of effect
modification by SDI score for the effects on respiratory mortality. Exposure to extreme heat
showed a faint trend of negative interaction with SDI. The 97
th
percentile exposure to extreme
heat shows a statistically significant negative interaction between high SDI (binary) and the
extreme heat exposure with lag 1 day (p-interaction=0.014). The high SDI group had an odds
ratio below 1 (the null), indicating a protective effect, although this result was not statistically
significant (0.97 [0.91, 1.04]). The low SDI group had a statistically significant odds ratio of
1.10 [1.02, 1.18]. A similar result of negative interaction was observed with medium SDI and the
97
th
percentile exposure to extreme heat on lag 0 (p-interaction =0.006) and lag 1 (p-
interaction=0.014) days. At lag 0, the odds ratio for the medium SDI group was also below 1
(0.99 [0.92, 1.06]), but not statistically significant. The odds ratio for the low SDI group was
elevated above 1 and statistically significant (1.14 [1.06, 1.22]).
19

Figure 6. Forest plot of ORs including 95% CIs and p-interaction values with binary
categorization of SDI for the association between each exposure (heat only, combined heat and
PM2.5, and PM2.5 only) on respiratory mortality risk at the 90
th
, 95
th
, and 97
th
percentiles with
lags 0 and 1.

20

Figure 7. Forest plot of ORs including 95% CIs and p-interaction values with ternary
categorization of SDI for the association between each exposure (heat only, combined heat and
PM2.5, and PM2.5 only) on respiratory mortality risk at the 90
th
, 95
th
, and 97
th
percentiles with
lags 0 and 1.
21

Ad Hoc Analysis
The pattern of negative interaction with SDI for the exposure to extreme heat on odds of
respiratory mortality prompted further investigation, as it was in the opposite direction of the
original hypothesis. After looking into age at death by ternary levels of SDI, we found that the
distribution of age at death was quite different for each level of SDI (see figures 1-3 in the
supplementary materials); the mean age at death also decreases with increasing levels of SDI as
shown in Table 3. While individual age is controlled for in the case-crossover design, differences
in age distributions between the SDI groups may introduce potential confounding. To account for
this, we stratified the data by age and conducted a separate analysis. We separated the data into
decedents above and below 65; literature has shown that mortality from respiratory diseases is
significantly higher above age 65 (Osman et al. 2017). However, age stratification did not reveal
consistent trends with positive or negative interaction and also produced mostly null results,
except for one statistically significant negative interaction for exposure to extreme heat only at
the 97
th
percentile lag 1 (p-interaction=0.045). Results for this analysis are summarized in Figure
8 and 9.
22

Figure 8. Forest plot of ORs including 95% CIs and p-interaction values with binary
categorization of SDI for the association between each exposure (heat only, combined heat and
PM2.5, and PM2.5 only) on respiratory mortality risk under age 65 at the 90
th
, 95
th
, and 97
th

percentiles with lags 0 and 1.

23

Figure 9. Forest plot of ORs including 95% CIs and p-interaction values with ternary
categorization of SDI for the association between each exposure (heat only, combined heat and
PM2.5, and PM2.5 only) on respiratory mortality risk under age 65 at the 90
th
, 95
th
, and 97
th

percentiles with lags 0 and 1.

24

Chapter 4: Discussion
For the analysis of this large dataset consisting of all deaths in California from 2014-2019, a
time-stratified case-crossover study using conditional logistic regression for the effect of
exposure to extreme PM2.5 pollution and heat on mortality risk found no statistically significant
consistent pattern of effect modification by levels of SDI scores. Similar inconsistent and null
results were found for the effect of SDI score on odds of mortality from cardiovascular and
respiratory mortality. There were two faint signals, however, due to the inconsistency of the
results they did not indicate a larger trend of effect modification by SDI. One trend was
marginally statistically significant p-interaction values for effect modification by SDI on the
effect of exposure to PM2.5 pollution at the 90
th
percentile on all-cause mortality risk; effect
estimates steadily increased by increasing level of SDI. This may indicate a slight signal of a
higher all-cause mortality risk from extreme exposure to PM2.5 with a higher SDI; however, it
was not backed by statistical significance at the 𝛼𝛼 = 0.05 level and the trend was not reflected as
strongly in the 95
th
and 97
th
exposure levels, revealing lack of consistency.
A second trend in the data was a few statistically significant negative interactions between
exposure to extreme heat and SDI for respiratory mortality risk. We conducted an ad hoc
analysis with age stratification to attempt accounting for differences in age structures between
low and high SDI groups. The high SDI group had a lower mean age at death with more
concentration of younger decedents between the ages of 0 and 65 (Supplementary Figure 1). The
low SDI group had a higher mean age at death with the most left skewed distribution of all the
groups (Supplementary Figure 3). This is consistent with literature that suggests that life
expectancies tend to be lower in more socioeconomically marginalized neighborhoods (Xie et al.
2021). Due the inconsistent and only spurious statistically significant interactions with and
25

without age stratification, no conclusive conclusions can be drawn from the results for
respiratory mortality. Further investigation and more advanced statistical modeling techniques,
such as including more complex interaction terms with age, should be considered in future
studies to account for differences in age structures among groups of different socioeconomic
status.
While it is hard to compare results for studies of this nature due to differing methodologies
regarding air pollution and heat exposure assessment, we found that previous research shows
mixed results for effect modification by socioeconomic variables for the effect of environmental
exposures such as heat and air pollution on mortality. Basu and Ostro (2008) conducted a case
crossover study design for the effect of exposure to average daily temperature in 9 California
counties on all-cause and cardiovascular mortality. The study used education level as a proxy for
socioeconomic status and found no significant evidence of elevated mortality risks for groups
with lower education level. However, another case-crossover study by Xu et al. (2020) examined
differences in effects of extreme heat on mortality risk among cities in Brazil with differing
levels of socioeconomic indicators such as income and literacy. These results showed that cities
with generally lower socioeconomic status (i.e., lower literacy, income, etc.) had increased
mortality risk with exposure to extreme heat. An important systematic review investigated effect
modification by socioeconomic variables for the effect of exposure to air pollution on mortality
(Laurent et al. 2007). After reviewing 15 different studies with differing methodologies, they
found that studies that analyzed socioeconomic variables as group-level variables found mostly
null results. The results of these studies are consistent with our findings. Laurent et al. found that
some studies that analyzed socioeconomic variables on the individual-level saw some evidence
of effect modification.
26

Our study was unique in that it examined the role socioeconomic indicators had on the joint
effect of extreme heat and air pollution, which has not been done in previous studies. Strengths
of our study include that we were able to control for time-invariant individual-level factors such
as gender, race/ethnicity, and education level within the case-crossover design. We also worked
with a very large sample size consisting of 1.5 million subjects and over 6 million data points,
allowing for greater power than many previous studies. A limitation of the exposure metric could
be that when dividing the exposure into categories, there are smaller N observations for each
group. A continuous exposure could be an alternative method, however this was already
explored in a prior analysis (Rahman et al. 2022). Further limitations of our analysis include
inability to account for three important variables: air conditioning (AC) use, houseless decedents,
and population mobility. Access to data on AC use was beyond the scope of this project.
However, it is important to consider since it is related to socioeconomic status and can mitigate
health effects from extreme heat. Additionally, decedents were identified based on residential
addresses. Those who did not have housing were therefore not included in the study.
Houselessness and housing insecurity is very important when considering the role of
socioeconomic status on mortality risk from environmental exposures because a houseless
person may have increased exposure, particularly to extreme heat, and decreased access to health
resources (Shwarz 2022). Finally, this study used heat and PM2.5 exposure based on residential
address, thus we do not account for personal exposures subjects experience outside the house
such as during work or daily activities outside of the home.
Our study found effects in individual groups to be mostly significant, indicating that
exposure to extreme heat and extreme PM2.5 pollution, including our particular interest in the
joint exposure, is significantly associated with all-cause, cardiovascular, and respiratory
27

mortality. However, we did not find consistent evidence that this effect was different based on
neighborhood-level social deprivation. This indicates that the base models, with no interaction
with SDI, is a good representation of the associations between these exposures and mortality
risk. This is consistent with prior analysis; the results from the base model are summarized in
supplementary figures 4-6 (Rahman et al. 2022). It is important to note that although the overall
N is quite large for the exposure groups, when stratifying by SDI these groups become smaller
and limit power to detect effect modification.
We, along with other studies including Bevan et al. 2021 and Ueji et al. 2011, did find that
exposure to air pollution and heat was generally higher for groups with lower socioeconomic
status. In future studies, we want to further examine the role that socioeconomics plays on
exposure to climate change events and how exactly the exposures affect the health of different
populations. With such limited research on the joint effect of heat and air pollution, more
thorough analyses are needed on the role that socioeconomic variables, such as SDI, play on the
risk of mortality with joint exposure. This will improve adaptation and management of climate
change health effects, and it will provide evidence for policy initiatives to support
socioeconomically marginalized communities – especially with the rise of extreme
environmental events due to climate change.
28

References

“American Community Survey Summary File.” Census.gov, US Census Bureau, 7 Jan. 2022,
https://www.census.gov/programs-surveys/acs/data/summary-file.html.
Abatzoglou, J. T. (2013). Development of gridded surface meteorological data for ecological
applications and modelling. International Journal of Climatology, 33(1), 121-131.
doi:https://doi.org/10.1002/joc.3413.
Basu R, Ostro BD. A multicounty analysis identifying the populations vulnerable to mortality
associated with high ambient temperature in California. Am J Epidemiol. 2008 Sep
15;168(6):632-7. doi: 10.1093/aje/kwn170. Epub 2008 Jul 28.
Basu R, Samet JM. Relation between elevated ambient temperature and mortality: a review of
the epidemiologic evidence. Epidemiol Rev. 2002;24(2):190-202. doi:
10.1093/epirev/mxf007.
Bevan GH, Freedman DA, Lee EK, Rajagopalan S, Al-Kindi SG. Association between ambient
air pollution and county-level cardiovascular mortality in the United States by social
deprivation index. Am Heart J. 2021 May;235:125-131. doi: 10.1016/j.ahj.2021.02.005.
Epub 2021 Feb 13.
Cheshire WP Jr. Thermoregulatory disorders and illness related to heat and cold stress. Auton
Neurosci. 2016 Apr;196:91-104. doi: 10.1016/j.autneu.2016.01.001. Epub 2016 Jan 6.
Ji S, Zhou Q, Jiang Y, He C, Chen Y, Wu C, Liu B. The Interactive Effects between Particulate
Matter and Heat Waves on Circulatory Mortality in Fuzhou, China. Int J Environ Res
Public Health. 2020 Aug 18;17(16):5979. doi: 10.3390/ijerph17165979.
Jones CG, Rappold AG, Vargo J, Cascio WE, Kharrazi M, McNally B, Hoshiko S; with the
CARES Surveillance Group. Out-of-Hospital Cardiac Arrests and Wildfire-Related
Particulate Matter During 2015-2017 California Wildfires. J Am Heart Assoc. 2020 Apr
21;9(8):e014125. doi: 10.1161/JAHA.119.014125. Epub 2020 Apr 15.
Laurent, O., Bard, D., Filleul, L., & Segala, C. (2007). Effect of socioeconomic status on the
relationship between atmospheric pollution and mortality. J Epidemiol Community
Health, 61(8), 665-675. doi:10.1136/jech.2006.053611
Lay CR, Sarofim MC, Vodonos Zilberg A, Mills DM, Jones RW, Schwartz J, Kinney PL. City-
level vulnerability to temperature-related mortality in the USA and future projections: a
geographically clustered meta-regression. Lancet Planet Health. 2021 Jun;5(6):e338-
e346. doi: 10.1016/S2542-5196(21)00058-9. Epub 2021 May 20.
Li Y, Akkus C, Yu X, Joyner A, Kmet J, Sweat D, Jia C. Heatwave Events and Mortality
Outcomes in Memphis, Tennessee: Testing Effect Modification by Socioeconomic Status
and Urbanicity. Int J Environ Res Public Health. 2019 Nov 18;16(22):4568. doi:
10.3390/ijerph16224568.
Mears M, Brindley P, Jorgensen A, Maheswaran R. Population-level linkages between urban
greenspace and health inequality: The case for using multiple indicators of
neighbourhood greenspace. Health Place. 2020 Mar;62:102284. doi:
10.1016/j.healthplace.2020.102284. Epub 2020 Jan 13.
Medina-Ramón M, Zanobetti A, Cavanagh DP, Schwartz J. Extreme temperatures and mortality:
assessing effect modification by personal characteristics and specific cause of death in a
multi-city case-only analysis. Environ Health Perspect. 2006 Sep;114(9):1331-6. doi:
10.1289/ehp.9074.
29

Osman S, Ziegler C, Gibson R, Mahmood R, Moraros J. The Association between Risk Factors
and Chronic Obstructive Pulmonary Disease in Canada: A Cross-sectional Study Using
the 2014 Canadian Community Health Survey. Int J Prev Med. 2017 Oct 24;8:86. doi:
10.4103/ijpvm.IJPVM_330_17.
Rahman MM, McConnell R, Schlaerth H, Ko J, Silva S, Lurmann F, Palinkas L, Johnston J,
Hurlburt M, Yin H, Ban-Weiss G, Garcia E. The Effects of Co-Exposure to Extremes of
Heat and Particulate Air Pollution on Mortality in California: Implications for Climate
Change. Am J Respir Crit Care Med. (In Press)
Rossati A. Global Warming and Its Health Impact. Int J Occup Environ Med. 2017 Jan;8(1):7-
20. doi: 10.15171/ijoem.2017.963.
Schwarz L, Castillo EM, Chan TC, Brennan JJ, Sbiroli ES, Carrasco-Escobar G, Nguyen A,
Clemesha RES, Gershunov A, Benmarhnia T. Heat Waves and Emergency Department
Visits Among the Homeless, San Diego, 2012-2019. Am J Public Health. 2022
Jan;112(1):98-106. doi: 10.2105/AJPH.2021.306557.
Tibuakuu M, Michos ED, Navas-Acien A, Jones MR. Air Pollution and Cardiovascular Disease:
A Focus on Vulnerable Populations Worldwide. Curr Epidemiol Rep. 2018
Dec;5(4):370-378. doi: 10.1007/s40471-018-0166-8. Epub 2018 Sep 19.
Tuholske C, Caylor K, Funk C, Verdin A, Sweeney S, Grace K, Peterson P, Evans T. Global
urban population exposure to extreme heat. Proc Natl Acad Sci U S A. 2021 Oct
12;118(41):e2024792118. doi: 10.1073/pnas.2024792118.
Uejio CK, Wilhelmi OV, Golden JS, Mills DM, Gulino SP, Samenow JP. Intra-urban societal
vulnerability to extreme heat: the role of heat exposure and the built environment,
socioeconomics, and neighborhood stability. Health Place. 2011 Mar;17(2):498-507. doi:
10.1016/j.healthplace.2010.12.005. Epub 2010 Dec 13.
Walker RE, Keane CR, Burke JG. Disparities and access to healthy food in the United States: A
review of food deserts literature. Health Place. 2010 Sep;16(5):876-84. doi:
10.1016/j.healthplace.2010.04.013. Epub 2010 Apr 24.
Wang J, Geng L. Effects of Socioeconomic Status on Physical and Psychological Health:
Lifestyle as a Mediator. Int J Environ Res Public Health. 2019 Jan 20;16(2):281. doi:
10.3390/ijerph16020281.
Witt C, Schubert AJ, Jehn M, Holzgreve A, Liebers U, Endlicher W, Scherer D. The Effects of
Climate Change on Patients With Chronic Lung Disease. A Systematic Literature
Review. Dtsch Arztebl Int. 2015 Dec 21;112(51-52):878-83. doi:
10.3238/arztebl.2015.0878.
Xie Y, Bowe B, Yan Y, Cai M, Al-Aly Z. County-Level Contextual Characteristics and
Disparities in Life Expectancy. Mayo Clin Proc. 2021 Jan;96(1):92-104. doi:
10.1016/j.mayocp.2020.04.043.
Xu, R., Zhao, Q., Coelho, M. S. Z. S., Saldiva, P. H. N., Abramson, M. J., Li, S., & Guo, Y.
(2020). Socioeconomic level and associations between heat exposure and all-cause and
cause-specific hospitalization in 1,814 Brazilian cities: A nationwide case-crossover
study. PLOS Medicine, 17(10), e1003369. doi:10.1371/journal.pmed.1003369
Yang L, Li C, Tang X. The Impact of PM2.5 on the Host Defense of Respiratory System. Front
Cell Dev Biol. 2020 Mar 4;8:91. doi: 10.3389/fcell.2020.00091.
Zanobetti A, Schwartz J. Temperature and mortality in nine US cities. Epidemiology. 2008
Jul;19(4):563-70. doi: 10.1097/EDE.0b013e31816d652d.
30

Zhao Y, Huang Z, Wang S, Hu J, Xiao J, Li X, Liu T, Zeng W, Guo L, Du Q, Ma W. Morbidity
burden of respiratory diseases attributable to ambient temperature: a case study in a
subtropical city in China. Environ Health. 2019 Oct 24;18(1):89. doi: 10.1186/s12940-
019-0529-8.
Zheng XY, Ding H, Jiang LN, Chen SW, Zheng JP, Qiu M, Zhou YX, Chen Q, Guan WJ.
Association between Air Pollutants and Asthma Emergency Room Visits and Hospital
Admissions in Time Series Studies: A Systematic Review and Meta-Analysis. PLoS One.
2015 Sep 18;10(9):e0138146. doi: 10.1371/journal.pone.0138146.
31

Supplementary Material
Supplementary Figure 1.

Supplementary Figure 2.

32

Supplementary Figure 3.

33

Supplementary Figure 4. Estimated excess risk (ER) of all-cause mortality (and 95% CI) on A)
co-occurrence of extreme maximum temperature and PM2.5 days compared to non-extreme days;
B) co-occurrence of extreme minimum temperature and PM2.5 days compared with non-extreme
days. Results are shown for same day exposure (lag-0) to 3 days prior exposure (lag-3) (Rahman
et al. 2022).

34

Supplementary Figure 5. Estimated excess risk (ER) of cardiovascular mortality (and 95% CI) on
A) co-occurrence of extreme maximum temperature and PM2.5 days compared to non-extreme
days; B) co-occurrence of extreme minimum temperature and PM2.5 days compared with non-
extreme days. Results are shown for same day exposure (lag-0) to 3 days prior exposure (lag-3)
(Rahman et al. 2022).

35

Supplementary Figure 6. Estimated excess risk (ER) of respiratory mortality (and 95% CI) on A)
co-occurrence of extreme maximum temperature and PM2.5 days compared to non-extreme days;
B) co-occurrence of extreme minimum temperature and PM2.5 days compared with non-extreme
days. Results are shown for same day exposure (lag-0) to 3 days prior exposure (lag-3) (Rahman
et al. 2022).

36

*Mortality data: Death certificate data for all deaths occurring in California from 1/1/2014 to
12/31/2019 were obtained from the California Department of Public Health’s Vital Statistics.
During the study period, there were total 1,514,292 all-cause, 492,513 cardiovascular, and
139,116 respiratory deaths.

Supplementary Figure 7. All-Cause Mortality Missing Data Flowchart

Total case+control days
N = 6,660,568
Missing values for SDI
N = 672
Missing values for
lag PM days
N = 3,674
Final case+control days
N = 6,656,222
Final cases in analysis
N = 1,513,396
= 896 dropped cases
ALL CAUSE LAG
Total case+control days
N = 6,660,568
Missing values for SDI
N = 672
Final case+control days
N = 6,659,896
Final cases in analysis
N = 1,514,139
= 153 dropped cases
ALL CAUSE NO LAG
37

Supplementary Figure 8. Respiratory Mortality Missing Data Flowchart

Total case+control days
N = 611,350
Missing values for SDI
N = 42
Missing values for
lag PM days
N = 445
Final case+control days
N = 610,863
Final cases in analysis
N = 139,008
= 108 dropped cases
RESPIRATORY LAG
Total case+control days
N = 611,350
Missing values for SDI
N = 42
Final case+control days
N = 611,308
Final cases in analysis
N = 139,106
= 10 dropped cases
RESPIRATORY NO LAG
38

Supplementary Figure 9. Cardiovascular Mortality Missing Data Flowchart

Total case+control days
N = 2,166,512
Missing values for SDI
N = 280
Missing values for
lag PM days
N = 1,187
Final case+control days
N = 2,165,045
Final cases in analysis
N = 492,215
= 298 dropped cases
CARDIOVASCULAR LAG
Total case+control days
N = 2,166,512
Missing values for SDI
N = 280
Final case+control days
N = 2,166,232
Final cases in analysis
N = 492,451
= 62 dropped cases
CARDIOVASCULAR NO LAG

Abstract (if available)

Abstract Background: The climate crisis is increasing the frequency of extreme heat and air pollution events. Previous research has found associations between heat and air pollution with increased risk of mortality, especially for cardiovascular and respiratory mortality. Several factors including access to healthcare, adaptation resources, and population mobility lead to increased vulnerability for socioeconomically marginalized communities to climate change events including extreme heat and air pollution. This study aimed to investigate whether increased neighborhood-level social deprivation modifies the effect of co- exposure to extreme heat and PM2.5 pollution on mortality risk.
Methods: A time-stratified case-crossover study design was used to assess the effect of daily exposure to PM2.5 pollution and heat on all-cause, cardiovascular, and respiratory mortality in California from 2014-2019 based on residential addresses. Air pollution data was assessed using inverse distance-squared weighted observations from up to 4 nearby US EPA air monitoring locations. Heat data was retrieved from a spatiotemporal model developed by the University of Idaho. Extreme PM2.5 and heat on the date of death (lag 0) and the day before (lag 1) were used as the exposures. Socioeconomic burden was evaluated using the census tract level Social Deprivation Index (SDI) with binary and ternary categories. Conditional logistic regression was used to analyze the associations between extreme heat and PM2.5 air pollution exposure and mortality risk with effect modification by SDI.
Results: We found no consistent effect modification by SDI on the combined or separate effects of extreme heat and PM2.5 pollution exposure and mortality with lag 0 and 1 days. Similar null results were found for cardiovascular and respiratory mortality risk.
Conclusion: We observed no pattern of effect modification by neighborhood-level social deprivation on the effect of heat and air pollution exposure on mortality risk in California, a state that experiences high levels of heat and air pollution especially with the ongoing climate crisis. This is consistent with other studies using neighborhood level SES; further analyses need to be conducted to explore this socioeconomic burden more.

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

Assessment of the mortality burden associated with ambient air pollution in rural and urban areas of India

PDF

Disparities in exposure to traffic-related pollution sources by self-identified and ancestral Hispanic descent in participants of the USC Children’s Health Study

PDF

Ambient air pollution and lung function in children

PDF

Long term effects of air pollution on lung function in the European community respiratory health survey

PDF

Spatial analysis of PM₂.₅ air pollution in association with hospital admissions in California

PDF

Air pollution, smoking, and multigenerational DNA methylation Signatures: a study of two southern California cohorts

PDF

Air pollution and breast cancer survival in California teachers: using address histories and individual-level data

PDF

Personal exposure to particulate matter PM2.5 sources during pregnancy and birthweight

PDF

The effects of heat and air pollution on mental-health related mortality

PDF

Evaluation of new methods for estimating exposure to traffic-related pollution and early health effects for large population epidemiological studies

PDF

Early life air pollution exposure, gestational diabetes mellitus, and autism spectrum disorder

PDF

Prenatal environmental exposures and fetal growth in the MADRES cohort

PDF

Airway inflammation and respiratory health in the Southern California children's health study

PDF

Associations between ambient air pollution and hypertensive disorders of pregnancy

PDF

Effects of stress and the social environment on childhood asthma in the children' s health study

PDF

Assessing the impact of air pollution on adverse birth outcomes in a low resource setting

PDF

A pilot study of a global approach to assessing air pollution exposure in port communities: passive air monitoring of nitrogen dioxide concentrations

PDF

Prenatal air pollution exposure, newborn DNA methylation, and childhood respiratory health

PDF

Susceptibility to autism likelihood associated with prenatal ambient air pollution exposures: identifying vulnerable populations

PDF

Effectiveness of individual- and household-level protective actions in reducing symptoms associated with hydrogen sulfide chronic low-level exposure

Asset Metadata

Creator Hasan, Zainab Saira (author)

Core Title Examining exposure to extreme heat and air pollution and its effects on all-cause, cardiovascular, and respiratory mortality in California: effect modification by the social deprivation index

School Keck School of Medicine

Degree Master of Science

Degree Program Applied Biostatistics and Epidemiology

Degree Conferral Date 2022-08

Publication Date 07/23/2022

Defense Date 07/23/2022

Publisher University of Southern California (original), University of Southern California. Libraries (digital)

Tag Air pollution,climate change,effect modification,environmental health disparities,extreme heat,mortality,OAI-PMH Harvest,PM2.5

Format application/pdf (imt)

Language English

Contributor Electronically uploaded by the author (provenance)

Advisor Garcia, Erika (committee chair), Gauderman, William (committee member), McConnell, Rob (committee member)

Creator Email zainabshasan@gmail.com,zhasan@usc.edu

Permanent Link (DOI) https://doi.org/10.25549/usctheses-oUC111375210

Unique identifier UC111375210

Legacy Identifier etd-HasanZaina-10946

Document Type Thesis

Format application/pdf (imt)

Rights Hasan, Zainab Saira

Type texts

Source 20220728-usctheses-batch-962 (batch), University of Southern California (contributing entity), University of Southern California Dissertations and Theses (collection)

Access Conditions The author retains rights to his/her dissertation, thesis or other graduate work according to U.S. copyright law. Electronic access is being provided by the USC Libraries in agreement with the author, as the original true and official version of the work, but does not grant the reader permission to use the work if the desired use is covered by copyright. It is the author, as rights holder, who must provide use permission if such use is covered by copyright. The original signature page accompanying the original submission of the work to the USC Libraries is retained by the USC Libraries and a copy of it may be obtained by authorized requesters contacting the repository e-mail address given.

Repository Name University of Southern California Digital Library

Repository Location USC Digital Library, University of Southern California, University Park Campus MC 2810, 3434 South Grand Avenue, 2nd Floor, Los Angeles, California 90089-2810, USA

Repository Email cisadmin@lib.usc.edu