Available via license: CC BY 4.0
Content may be subject to copyright.
Ettman et al., Sci. Adv. 8, eabm9737 (2022) 2 March 2022
SCIENCE ADVANCES | RESEARCH ARTICLE
1 of 9
CORONAVIRUS
Assets, stressors, and symptoms of persistent
depression over the first year of the
COVID-19 pandemic
Catherine K. Ettman1,2*, Gregory H. Cohen1, Salma M. Abdalla1, Ludovic Trinquart1,3,4,
Brian C. Castrucci5, Rachel H. Bork5, Melissa A. Clark2, Ira B. Wilson2,
Patrick M. Vivier2,6, Sandro Galea1
The coronavirus disease 2019 (COVID-19) pandemic has been accompanied by an increase in depression in U.S.
adults. Previous literature suggests that having assets may protect against depression. Using a nationally repre-
sentative longitudinal panel survey of U.S. adults studied in March and April 2020 and in March and April 2021, we
found that (i) 20.3% of U.S. adults reported symptoms of persistent depression in Spring 2020 and Spring 2021,
(ii) having more assets was associated with lower symptoms of persistent depression, with financial assets—
household income and savings—most strongly associated, and (iii) while having assets appeared to protect
persons—in particular those without stressors—from symptoms of persistent depression over the COVID-19 pan-
demic, having assets did not appear to reduce the effects of job loss, financial difficulties, or relationship stress on
symptoms of persistent depression. Efforts to reduce population depression should consider the role played by
assets in shaping risk of symptoms of persistent depression.
INTRODUCTION
The first year of the coronavirus disease 2019 (COVID-19) pan-
demic presented unprecedented challenges for population mental
health. The threat and fear of a new devastating infectious disease
(1), millions of deaths globally, and unprecedented reductions in
social interactions were each stressors that could be expected to in-
fluence mental health. In addition, the efforts to mitigate the pandem-
ic were accompanied by an economic downturn that contributed to
poor mental health (2–5). An increase in poor mental health at the
start of the COVID-19 pandemic relative to before it has been doc-
umented across multiple studies. Daly etal. (6) reported an increase
in symptoms of depression from 8.7% in 2017 to 2018 to 14.4% in
April 2020 using the Patient Health Questionnaire-2 (PHQ-2) screener
for depressive symptoms. Czeisler etal. (7) documented a popula-
tion prevalence of 24.3% for symptoms of depressive disorder in
June 2020 also using the PHQ-2. Using the nine-question PHQ-9,
Ettman etal. (8) reported an increase in elevated symptoms of prob-
able depression from 8.5% in 2017 to 2018 to 27.8% in March and
April 2020, suggesting a potential threefold increase in symptoms
of probable depression at the start of the COVID-19 pandemic.
McGinty etal. (9) measured symptoms of serious psychological dis-
tress using the Kessler 6 Psychological Distress Scale and documented
in an increase from 3.9% in 2018 to 13.6% in April 2020, suggesting
a 3.5-fold increase in symptoms of psychological distress. While the
increase in depression at the onset of the pandemic may not have
been unexpected given what we knew about the risks for depression
before the pandemic, depression remained high through the end of
2020 as the COVID-19 pandemic continued (10). Vahratian etal. (10)
reported a continued increase in symptoms of depressive disorder
from 24.5% in August 2020 to 30.2% in December 2020. Ettman etal.
(11) reported that 32.8% of U.S. adults reported elevated symptoms
of probable depression in March and April 2021.
The continued high prevalence of depression is unusual. In the
aftermath of other mass traumatic events, population mental health
improved in the months that followed the large-scale trauma. For
example, after an initial increase, population depression decreased
substantially in the first six months after Hurricane Ike (12), the 1999
Mexico floods (13), and the September 11th attacks (13,14). The
chronic and continued exposure to the COVID-19 pandemic through-
out 2020 may have resulted in the observed persistence of high levels
of depression over time at the population level during the pandemic.
Comparison with population mental health following the last pan-
demic of similar scale, namely, the 1918 Flu Pandemic, is challenging,
given advances in the field on mental health screening instruments
and classification of conditions. Even so, publication from the time
suggested that while around one-fourth of cases at a Boston-based
hospital showed depression at any time, it was not chronic or per-
sistent when present (15). Persistent depression among individuals
(that is, unrelenting depression, or depression expressed by the same
person across multiple times) is particularly concerning given its po-
tential for ongoing health and economic consequences in popula-
tions (16). It was estimated that depression cost the United States
over $210 billion in 2010 including absenteeism (missed work), pre-
senteeism (underproduction while at work), and costs for treatment,
among others (17). The economic toll of depression could be con-
siderably larger with an increased prevalence of the population re-
porting symptoms of depression.
The role of assets may be essential to understanding the per-
sistently high burden of depression in the population during the
COVID-19 pandemic. Assets can protect against poor mental health,
as noted before the COVID-19 pandemic (18–20). In particular, fi-
nancial assets, physical assets, and social assets may all protect against
depression (19,21). For example, having family savings was associated
1Boston University School of Public Health, Boston, MA, USA. 2Brown University
School of Public Health, Providence, RI, USA. 3Institute for Clinical Research and
Health Policy Studies, Tufts Medical Center, Boston, Massachusetts, USA. 4Tufts
Clinical and Translational Science Institute, Tufts University, Boston, Massachu-
setts, USA. 5de Beaumont Foundation, Bethesda, MD, USA. 6Hassenfeld Child
Health Innovation Institute, Providence, RI, USA.
*Corresponding author. Email: cettman@bu.edu
Copyright © 2022
The Authors, some
rights reserved;
exclusive licensee
American Association
for the Advancement
of Science. No claim to
original U.S. Government
Works. Distributed
under a Creative
Commons Attribution
License 4.0 (CC BY).
Ettman et al., Sci. Adv. 8, eabm9737 (2022) 2 March 2022
SCIENCE ADVANCES | RESEARCH ARTICLE
2 of 9
with 150% greater odds of symptoms of depression relative to not
having family savings in 2015 to 2016 (18). Beyond just having family
savings, owning a home was also associated with lower odds of symp-
toms of depression: Homeowners without savings had 2.15 times
the odds of symptoms of depression relative to homeowners with
$5000in family savings, and home renters without family savings
had 3.65 times the odds of symptoms of depression relative to home
renters with home savings (20). The association of assets with prob-
able depression is so strong that it may, in fact, account for much of
the difference in population-level depression between racial-ethnic
groups (19). Having more assets was associated with a lower preva-
lence of probable depression at the start of COVID-19, including in
the face of stressors (22). It is possible that having access to assets may
also protect persons against persistent or chronic depression over
time. We do not know the prevalence of symptoms of persistent de-
pression, nor do we know the factors associated with greater risk for
symptoms of persistent depression following exposure to stressors
after the presence of the COVID-19 pandemic for 1 year.
Therefore, in this work, we aimed to understand the following:
(i) the population prevalence of symptoms of persistent depression
at two time points, 1 year apart, during the COVID-19 pandemic;
(ii) the relative influence of particular types of assets on symptoms
of persistent depression over that time; and (iii) whether having
assets reduced the effect of stressors on symptoms of persistent de-
pression over the course of the COVID-19 pandemic. This paper
addresses gaps in the literature by using a nationally representative,
longitudinal panel study to measure symptoms of persistent depres-
sion 1 year into the COVID-19 pandemic, measuring symptoms of
persistent depression in March and April 2020 and in March and
April 2021in U.S. adults. We use detailed assets and stressor expo-
sures measured at the start of the COVID-19 pandemic to predict
symptoms of persistent depression 1 year later.
RESULTS
Table1 shows the prevalence of symptoms of persistent depression
by gender, age, and race/ethnicity. Twenty percent of U.S. adults re-
ported symptoms of persistent depression, reporting elevated symp-
toms of probable depression in both March and April 2020 and March
and April 2021. Among women, 24.8% reported symptoms of per-
sistent depression and among men, 15.4% reported symptoms of
persistent depression (P<0.01). Persons ages 18 to 39 years reported
the highest prevalence of symptoms of persistent depression (26.8%)
relative to persons ages 40 to 59 years (20.7%) and persons ages
60 years and older (10.2%) (P<0.01). We found no evidence of
differences in the prevalence of symptoms of persistent depression
across race/ethnicity.
Figure1 shows a visual representation of the prevalence of symp-
toms of persistent depression by three types of assets, which are de-
scribed in greater detail in Materials and Methods: financial assets,
physical assets, and social assets. Financial assets include household
income and household savings; physical assets include homeowner-
ship; and social assets include educational attainment and marital
status, as published previously (19). The graph shows that as each
asset type increased, the prevalence of persistent depression decreased.
While persons with more social assets reported lower prevalence of
Table 1. Symptoms of persistent depression in March and April 2020 (T1) and March and April 2021 (T2) by gender, age, and race/ethnicity. Note: T1
demographic characteristics reported. Other race includes multiple races and non-Hispanic Asian race. Column percentages provided for total; row percentages
provided for persistent depression. Symptoms of persistent depression defined as presence of PHQ-9 score of 10 or greater at T1 and T2. n unweighted,
% weighted using T2 survey weights. P value reflects the two-sided 2 test between persistent depression and all other categories (people with no
depression, depression only at T1, or depression only at T2). P values <0.05 suggest significance in differences between persistent depression and all other
categories by demographic characteristics.
Total Symptoms of persistent
depression All other categories
n%n%n%P value
Total 1139 208 20.3 931 79.7
Gender 0.008
Female 563 51.8 130 24.8 433 75.2
Male 576 48.2 78 15.4 498 84.6
Age <0.001
18–39 years 458 40.3 102 26.8 356 73.2
40–59 years 380 32.0 76 20.7 304 79.3
≥60 years 301 27.7 30 10.2 271 89.8
Race/ethnicity 0.602
Black, non-
Hispanic 95 11.9 13 16.2 82 83.8
Hispanic, any race
or races 186 16.4 39 23.9 147 76.1
White, non-
Hispanic 773 63.1 139 20.7 634 79.3
Other race,
non-Hispanic 85 8.6 17 15.8 68 84.2
Ettman et al., Sci. Adv. 8, eabm9737 (2022) 2 March 2022
SCIENCE ADVANCES | RESEARCH ARTICLE
3 of 9
depression than persons with fewer social assets, the spread between
the groups was greatest between high- and low-asset holders for fi-
nancial assets and physical assets. Persistent depression was highest
among persons with low household income, with 40.9% of persons
with $0 to $19,999 and 9.5% of persons with $75,000 or more in an-
nual household income reporting persistent depression (P<0.01).
Thirty-one percent of persons with less than $5000in household
savings relative to 13.2% of persons with $5000 or more in household
savings reported persistent depression (P<0.01). Twenty-five percent
of home renters and 16.3% of homeowners reported persistent de-
pression (P<0.01). More than 25% percent of persons without a high
school degree reported persistent depression, while 11.9% of persons
with a college degree or higher reported persistent depression. Persons
who were not married reported a higher prevalence of persistent de-
pression than persons who were married; 28% of persons who had
never married and 28% of those who were living with a partner versus
13.2% of persons who were married reported persistent depression.
Table2 shows the odds of symptoms of persistent depression by
financial assets (income and savings), physical assets (homeowner-
ship), and social assets (education and marital status). Persons with
a household income of $0 to $19,999 relative to $75,000 or more
had 6.8 times the odds of symptoms of persistent depression, ad-
justing for demographic characteristics (model 2). Persons with less
than relative to more than $5000in household savings had 2.7 times
the odds of symptoms of persistent depression when adjusting for
demographics (model 3). Persons with a high school degree or grad-
uate equivalency degree/general educational diploma (GED) had
2.9 times the odds of symptoms of persistent depression as persons
with a college degree or more (model 5). Persons who were never
married had 2.1 times the odds and persons who were widowed,
divorced, or separated had 2.0 times the odds of symptoms of per-
sistent depression as persons who were married, when controlling
for demographics (model 6). Model 7 shows the adjusted odds of
symptoms of persistent depression, adjusting for all assets (which
are correlated with each other; see table S1) and demographic char-
acteristics. When adjusting for financial, physical, and social assets
at the same time, having a household income of $0 to $19,999 rela-
tive to $75,000 or more was associated with 3.5 times the odds of
symptoms of persistent depression and having household savings of
less than $5000 was associated with 1.7 times the odds of symptoms
of persistent depression.
Figure2 shows the predicted probability of symptoms of per-
sistent depression adjusted for demographic characteristics and for
the interaction of job loss, financial difficulties, and relationship prob-
lems with each asset type. Assets did not appear to reduce the effect
of job loss, financial difficulties, or relationship problems on symp-
toms of persistent depression. Assets were, however, associated with
lower symptoms of persistent depression among persons who did not
report stressors. In particular, persons who had more than $5000in
savings and did not report job loss had substantially lower symptoms
of persistent depression than persons who had less than $5000in
savings and did not report job loss; similarly, persons with $5000 or
more in savings and did not report relationship problems had sub-
stantially lower symptoms of persistent depression than their coun-
terparts with less than $5000in savings.
Table S1 shows the correlation of assets with each another, and
table S2 shows the prevalence of symptoms of persistent depression
by financial assets, physical assets, and social assets. Table S3 shows
0.0%
5.0%
10.0%
15.0%
20.0%
25.0%
30.0%
35.0%
40.0%
45.0%
$0
-
$19,999
$20,000
-
$44,999
$45,000
-
$74,000
≥$75,000
$0
-
$4999
≥$5000
Occupied without payment
Rented for cash
Homeowner
Less than high school graduat
e
High school grad or GED
Some college
College grad or more
Living with partner
Never married
Widowed, divorced, separated
Married
Income
Savings
Homeownership
Education
Marital status
Financial assets Physical assets Social assets
Prevalence of persistent depression symptoms
Fig. 1. Prevalence of symptoms of persistent depression in March and April 2021 (T2) by financial assets, physical assets, and social assets in March and April 2020
(T1). Note: T1 assets reported. Symptoms of persistent depression defined as presence of PHQ-9 score of 10 or greater at T1 and T2. GED, graduate equivalency degree/
general educational diploma. Percentages weighted using T2 survey weights.
Ettman et al., Sci. Adv. 8, eabm9737 (2022) 2 March 2022
SCIENCE ADVANCES | RESEARCH ARTICLE
4 of 9
the adjusted predicted probabilities and 95% confidence intervals
(CIs) for all interaction pairs. Table S4 shows the results of likeli-
hood ratio tests for the interaction between each stressor (job loss,
financial problems, and relationship problems) and each asset type.
At the 0.05 level of significance, there was evidence of interaction
between savings and job loss and between savings and relationship
problems on symptoms of persistent depression. Figure S1 shows the
unadjusted, unweighted prevalence of symptoms of persistent de-
pression across each stressor and asset type combination.
DISCUSSION
In a nationally representative longitudinal panel study of U.S. adults
1 year into the COVID-19 pandemic, we describe three main findings:
(i) 20.3% of surveyed U.S. adults reported symptoms of depression
in both March and April 2020 and March and April 2021; (ii) finan-
cial assets, physical assets, and social assets were each associated with
a lower likelihood of symptoms of persistent depression 1 year into
the COVID-19 pandemic, with the strongest associations among fi-
nancial assets; and (iii) persons with fewer assets and more stressors
in March and April 2020 were more likely to report symptoms of per-
sistent depression 1 year later, in March and April 2021, controlling
for gender, race/ethnicity, age, and household size. Having assets was
particularly important for reducing symptoms of persistent depres-
sion 1 year into the COVID-19 pandemic in the absence of stressors.
People who experienced stressors had greater symptoms of persistent
depression than persons who did not. Persons with lowest risk of
symptoms of persistent depression in Spring 2021 were those with
high assets in Spring 2020 and no exposure to job loss, financial dif-
ficulties, or relationship problems.
Table 2. Odds ratios of symptoms of persistent depression in March and April 2021 (T2) by assets in March and April 2020 (T1). Note: Odds radio (OR),
adjusted odds ratios (aOR), and 95% confidence interval (CI) presented. Model 1: unadjusted. Models 2 to 6: adjusted for household income, household savings,
homeownership, education, or marital status, respectively, and gender, age, race/ethnicity, and household size. Model 7: multivariable model adjusted for
gender, age, race/ethnicity, household size, and all assets (household income, household savings, homeownership, education, and marital status). Symptoms of
persistent depression defined as presence of the PHQ-9 score of 10 or greater at T1 and T2. Data weighted using T2 survey weights. Ref, reference.
Model 1 Model 2 Model 3 Model 4 Model 5 Model 6 Model 7
OR (95% CI) aOR (95% CI) aOR (95% CI) aOR (95% CI) aOR (95% CI) aOR (95% CI) aOR (95% CI)
Household income
$0–$19,999 6.6 (3.5–12.2) 6.8 (3.7–12.6) – – – – 3.5 (1.6–7.6)
$20,000–$44,999 2.6 (1.5–4.6) 2.7 (1.5–4.9) – – – – 1.7 (0.9–3.3)
$45,000–$74,999 1.8 (1.0–3.4) 1.8 (1.0–3.2) – – – – 1.3 (0.7–2.4)
≥$75,000 Ref Ref – – – –
Household savings
$0–$4999 3.0 (1.9–4.7) – 2.7 (1.7–4.2) – – – 1.7 (1.0–2.8)
≥$5000 – Ref – – –
Homeownership
Occupied without
payment 3.7 (1.1–2.6) – – 3.1 (1.3–7.5) – – 0.9 (0.5–1.5)
Rented for cash 1.7 (1.4–9.9) – – 1.4 (0.9–2.3) – – 1.6 (0.6–4.2)
Homeowner – – Ref – –
Education
Less than high
school
graduate
2.5 (1.0–6.6) – – – 2.6 (1.0–6.6) – 1.1 (0.4–2.8)
High school
graduate or
GED
2.7 (1.6–4.5) – – – 2.9 (1.7–5.1) – 1.6 (0.8–3.0)
Some college 2.2 (1.4–3.4) – – – 2.2 (1.4–3.5) – 1.6 (0.9–2.6)
College graduate or more – – – Ref –
Marital status
Living with
partner 2.7 (1.3–5.5) – – – – 2.0 (1.0–4.0) 1.4 (0.7–2.9)
Never married 2.5 (1.5–4.2) – – – – 2.1 (1.3–3.5) 1.5 (0.8–2.7)
Widowed,
divorced,
separated
2.0 (1.2–3.2) – – – – 2.0 (1.2–3.5) 1.5 (0.8–2.7)
Married – – – –
Ettman et al., Sci. Adv. 8, eabm9737 (2022) 2 March 2022
SCIENCE ADVANCES | RESEARCH ARTICLE
5 of 9
These findings are consistent with some but not all other studies
that have reported a consistently high prevalence of symptoms of de-
pression during the COVID-19 pandemic in population-level longi-
tudinal cohorts. Most of the studies published to date on longitudinal
cohorts have been conducted outside of the United States, in Australia,
Austria, and the United Kingdom, among others. Czeisler etal. (23)
found an unchanged prevalence of probable depression in the Australian
population in April 2020 and September 2020. However, Australia
Fig. 2. Predicted probability of symptoms of persistent depression in March and April 2021 (T2) by stressors and assets in March and April 2020 (T1). Note: T1
stressors and assets reported. Symptoms of persistent depression defined as presence of the PHQ-9 score of 10 or greater at T1 and T2. Models adjusted for gender, age,
race/ethnicity, and household size. Unweighted.
Ettman et al., Sci. Adv. 8, eabm9737 (2022) 2 March 2022
SCIENCE ADVANCES | RESEARCH ARTICLE
6 of 9
maintained its lockdown during the two survey periods, unlike the
United States, whose COVID-19 restriction policies were largely being
lifted around the timing of the COVID-19 and Life stressors Impact
on Mental Health and Well-being (CLIMB) Time 2 (T2) survey. In
Austria, which did lift its lockdown policies before the CLIMB T2
collection, Pieh etal. (24) found no significant change in probable de-
pression between April 2020 and September 2020; they found that
18.3 and 19.7% of the sample reported PHQ-9 scores of 10 or greater
at T1 and T2, respectively. Several studies conducted on longitudi-
nal cohorts in the United Kingdom reported that depressive symp-
toms improved after the initial start of the COVID-19 pandemic.
Fancourt etal. (25) and Pierce etal. (26) reported that depressive
symptoms improved from the start of the COVID-19 pandemic and
August 2020 and October 2020, respectively. However, these studies
were conducted during summer months, which may have been con-
founded by seasonal effects resulting in improved affect. In addition,
a requirement that participants provide at least three repeated mea-
sures (i.e., participants had to respond to at least three of the weekly
surveys between 23 March 2020 and 9 August 2020 to be included
in the sample) in the study by Fancourt etal. may be susceptible to
survivorship bias (27), which could lead to an underreporting of
adverse mental health symptoms at the population level. Given that
the CLIMB survey had a response rate of 81.1% at T2, our data may
be less susceptible to survivorship bias.
Within the United States, there are few studies that have longitu-
dinally followed populations during the COVID-19 pandemic. While
several studies compared adult depressive symptoms relative to the
start of the COVID-19 pandemic (6,28,29), none to our knowledge
has reported on persistent depression (presence of repeatedly report-
ing probable depression in the same persons) reported out as far as
April 2021. Of the longitudinal studies conducted at the start of the
COVID-19 pandemic, findings suggested no change or a slight de-
cline in depressive symptoms during the first half of 2020 (30–33). For
example, Shuster etal. (32) found that anxiety and depressive symp-
toms declined after the initial weeks of COVID-19, measured be-
tween April 2020 and June 2020. Their study differs from ours in that
the population was not representative of U.S. adults, captured a
10-week span (relative to our 12-month comparison), and may have
seen a decline in depressive symptoms due to loss to follow-up of
persons with depression and seasonal effects, with affect improv-
ing during summer months. Our findings were consistent with theirs
in that they reported that female gender, younger age, and lower
household income were associated with increased depression across
time. They also found that worsening economic situation due to
COVID-19 was associated with increased depression over time (32).
The most recent longitudinal study of which we are aware measured
depressive symptoms over the previous 7 days using the PHQ-2 with
the last reporting period being from 20 January to 1 February 2021
(10). While the authors used an abbreviated form of the PHQ-9 and
did not capture detailed asset or stressor information, they reported
an increase in the percentage of adults with recent symptoms of
an anxiety or depressive disorder (10), with 30.2% of U.S. adults
reporting symptoms of a depressive disorder as of January and
February 2021 (relative to our finding that 20.3% of U.S. adults re-
ported symptoms of depressive disorder at both March and April 2020
and March and April 2021). According to the Centers for Disease
Control National Center for Health Statistics Household Pulse
Survey, which used the shorter-form PHQ-2, 24.7% of U.S. adults
reported symptoms of depressive disorder between 17 March and
29 March 2021, which closely aligned with our survey collection at
T2 (34).
Our findings were consistent with other studies that have ad-
dressed the stressors of job loss, financial strain, and relationship
conflict during COVID-19. In a cross-sectional study conducted in
April 2020, McDowell etal. (35) reported an increase in symptoms of
depression among persons who reported job loss. Hertz-Palmor etal.
(36) assessed relations between financial strain and depressive symp-
toms in March and April 2020 and 1 month later in samples in the
United States and Israel. They found that income loss and financial
strain were associated with exacerbated depressive symptoms in their
1-month follow-up sample. Lee etal. (37) studied relationship con-
flict during COVID-19 from March and April 2020 and found that
in the weeks studied, relationship conflict increased. Although their
findings did not show a significant association between relationship
conflict and depressive symptoms, this may have been due to limita-
tions in sample size (N=291) (37). Nonetheless, these studies show
early evidence that exposure to stressors during the COVID-19 pan-
demic was associated with depressive symptoms.
Our findings that having more assets was associated with lower
depression are novel in the context of COVID-19, even if consistent
with studies conducted after other traumatic and stressful events. For
example, Gallo etal. (38) found that persistent depression lowered
over time following involuntary job loss but remained highest among
low-wealth persons. Tracy etal. (39) found that low–socioeconomic
status persons were more likely to report depressive symptoms fol-
lowing Hurricane Ike. Thus, although the conditions of COVID-19
were unique, these findings provide support for the notion that eco-
nomic conditions can buffer the effects of stressors on depression.
The COVID-19 pandemic in particular was unique in its wide-
ranging scope, its ongoing nature, and the economic inequities that
it produced (40). As a result, persistent depression may be higher
1 year into the COVID-19 pandemic than that documented after other
large-scale events.
This study has three main limitations. First, the PHQ-9, which
was used to assess symptoms of probable depression at each time
point, is a depression screener, which does not replace the gold stan-
dard of clinical diagnosis. However, given a sensitivity and specificity
of 88% relative to clinical diagnosis, the PHQ-9 is the best available
measure that allows large-scale assessments to provide estimates of
symptoms of depression consistent with probable depression at the
national level across time (41). Using the PHQ-9 allowed us to esti-
mate the burden of depression at the population level at the begin-
ning of the pandemic, setting a baseline for symptoms of persistent
depression, which was defined as presence of probable depression at
T1 and T2. The study was not designed to measure continuous prob-
able depression throughout the 12-month follow-up period but rather
to measure presence of probable depression at both T1 and T2. Sec-
ond, similar to all longitudinal studies following the same persons
over time, we experienced some loss to follow-up at T2. However,
with 81.1% of respondents replying at T2, we had a relatively high
response rate, particularly given potential for survey response fatigue
during the COVID-19 pandemic. It is possible that nonresponders at
T2 had a higher prevalence of depression at T2 than responders (27),
suggesting that the documented symptoms of persistent depression
presented may represent an underestimate of the prevalence of prob-
able depression at T2 and therefore symptoms of persistent depres-
sion. Survey weights accounting for nonresponse at T2 were used,
allaying these concerns. Third, our sample size may have limited our
Ettman et al., Sci. Adv. 8, eabm9737 (2022) 2 March 2022
SCIENCE ADVANCES | RESEARCH ARTICLE
7 of 9
ability to detect significant associations in the interactive effects
of assets in protecting against persistent depression across stressor
groups. However, that we were able to detect significant associa-
tions in the interactions of savings with relationship problems and
job loss speaks to the magnitude of these stressors on poor men-
tal health.
We found a high prevalence of persistent depression across a na-
tionally representative group of U.S. adults measured longitudinally
after 1 year of follow-up during the COVID-19 pandemic through
April 2021. Our results highlight the importance of assets as a po-
tential protective mechanism against ongoing probable depression.
Exposure to stressors and having fewer assets were both associated
with greater persistent depression 12 months into the COVID-19 pan-
demic. While assets did not appear to reduce the effect of stressors
on persistent depression, having more assets and not experiencing
stressors was associated with significantly less persistent depression
than having fewer assets. These findings highlight the deleterious
effect of stressors on mental health and the potential protective ef-
fect of assets against the COVID-19 pandemic in the absence of re-
ported job loss, financial problems, and relationship stressors.
Given the high prevalence of depression in U.S. adults 1 year into
the COVID-19 pandemic, with one in five surveyed U.S. adults screen-
ing positive for symptoms of probable depression both in March
and April 2020 and in March and April 2021, finding multiple ways
to address and mitigate the burden of poor mental health will be
critical. These findings suggest that interventions to shore up eco-
nomic contexts that people live in, in particular bolstering financial
assets and reducing stressors, may serve to lessen persistent depres-
sion over time. Efforts to improve the economic status of low-asset
populations may lead to improved mental health.
MATERIALS AND METHODS
Experimental design
This study used a nationally representative sample of U.S. residents
ages 18 years and older followed longitudinally over 1 year of the
COVID-19 pandemic. Participants were surveyed in March and
April 2020 (T1) and in March and April 2021 (T2) as part of the CLIMB
study (11). Participants were drawn from the AmeriSpeak panel, which
is a nationally representative standing panel whose sampling frame
covered 97% of U.S. households and used a two-stage probability-
based sample design to recruit members. The AmeriSpeak panel has
a household response rate of 34.2% (42). Participants provided con-
sent at induction into the AmeriSpeak panel and at the beginning
of each CLIMB survey. Participants were contacted over email and
over the telephone if they did not respond to email outreach. The
CLIMB survey completion rate for participants at T1 was 64.3% (8),
and among those participants, the response rate at T2 was 81.1% (11).
Details on sampling, demographics, and characteristics can be found
in other published work (8,11,22,43). The final analytic sample in
this paper included 1139 participants who responded to all depres-
sion questions at T1 and T2. Forty-four persons were removed from
the analysis because they did not respond to all depression questions
at T1 and/or T2. The participants who were not included in the anal-
ysis because of missing depression values did not differ significantly
across gender, race/ethnicity, marital status, age, education, or house-
hold income status from participants who were included in the anal-
ysis. Survey weights accounted for nonresponse at T1 and at T2 and
aligned the sample with the U.S. adult population according to
the 2010 U.S. Census (1) using the following variables: age, gender,
Census Division, race/ethnicity, education, housing type, and house-
hold phone status. The institutional review boards at NORC at the
University of Chicago and Boston University approved this study.
Key covariates
Symptoms of persistent depression
Participants completed the PHQ-9, a validated screening tool for
depression at both T1 and T2. The PHQ-9 is a nine-item screening
tool measuring probable depression based on the DSM-IV: Diag-
nostic and Statistical Manual of Mental Disorders, fourth edition.
Participants responded to nine questions about their affect over the
last 2 weeks; responses were tallied for corresponding scores ranging
from 0 to 27. Using a PHQ-9 score of 10 or greater has a sensitivity
of 88% and a specificity of 88% when tested against clinical diag-
nosis of depression (41). Having symptoms of persistent depression
was defined as the presence of a PHQ-9 score of 10 or greater at both
T1 and T2. We compared persons with persistent depression to all
others in the sample, given the increased burden of poor mental
health for persons reporting probable depression chronically across
multiple time points (16). Because the PHQ-9 is a screening tool, it
cannot replace official diagnosis of depression by a clinician (44).
Throughout this paper, symptoms of persistent depression refer to
the presence of elevated symptoms of probable depression in both
March and April 2020 and March and April 2021 as indicated by a
PHQ-9 score of 10 or greater at both times. This definition differs
from persistent depressive disorder, reflecting the DSM-5 concepts
of dysthymia and chronic major depression, which would be diag-
nosed by a clinician (45). Here, symptoms of persistent depression
reflect the presence of elevated symptoms of probable depression at
two points one year apart during the COVID-19 pandemic.
Assets
To consider the role different types of assets, assets were grouped into
three categories: financial assets, physical assets, and social assets,
as previously published (19). Financial assets included household
income and household savings. Household income was defined as
a categorical variable: $0 to $19,999, $20,000 to $44,999, $45,000 to
$74,999, and≥$75,000 (11), with categories divided roughly at the
interquartile range. To determine household savings, participants
were asked to list total money in all types of accounts, including “cash,
savings, or checking accounts, stocks, bonds, mutual funds, retirement
funds (such as pensions, IRAs, 401Ks, etc.), and certificates of deposit”
as consistent with national surveys (18). A binary variable was then
created: $0 to $4999 or $5000 or more, as used previously (8,11,22).
Physical assets referred to homeownership, which was defined as a
categorical variable: homeowner, rented for cash, and occupied with-
out payment of cash rent (11). Social assets included educational
attainment and marital status. Educational attainment was defined
as a categorical variable: less than high school graduate, high
school graduate or GED, some college, including vocational/tech
school, and college graduate or more (11). Marital status was
defined as a categorical variable: married; widowed, divorced, or
separated; never married; and living with partner (11). Assets at T1
were reported.
Demographic characteristics
Key demographic characteristics include gender (female/male), age
category (18 to 39 years, 40 to 59 years, and 60 years or older), race/
ethnicity (Hispanic, non-Hispanic black, non-Hispanic white, and
Ettman et al., Sci. Adv. 8, eabm9737 (2022) 2 March 2022
SCIENCE ADVANCES | RESEARCH ARTICLE
8 of 9
other race, including non-Hispanic Asian and multiple races), and
household size (continuous variable, capped at 7).
Stressors
To examine the role of assets in protecting against persistent de-
pression, including in the face of stressors, we selected three stressors
as examples of COVID-19–induced stressors experienced during
the pandemic. We also aimed to understand whether assets modified
the relation between stressors and persistent depression. We used
three stressors that were highly associated with probable depression
and, therefore, were candidates for potential effect measure modi-
fication. Job loss, financial difficulties, and relationship problems
were each defined as a binary variable in response to the following
question: “Have any of the following affected your life as a result of
the coronavirus or COVID-19 outbreak?” Responses included the
following: “losing a job,” “having financial problems,” and “family
or relationship problems (for example, with your spouse or kids).”
Stressors at T1 were reported.
Statistical analysis
First, we calculated the prevalence of symptoms of persistent de-
pression by gender, age, and race/ethnicity. Prevalence measures of
symptoms of persistent depression were weighted unless otherwise
noted. We used complex probability weights to account for nonre-
sponse at T1 and T2 and to align with the U.S. adult population; the
survey weights allowed for estimates to represent the U.S. national
adult population (11). We conducted two-tailed 2 analyses to measure
the difference in distribution of symptoms of persistent depression
across groups. Significance was set at P<0.05. Demographic and
asset variables at T1 were used to predict symptoms of persistent
depression across time. Second, we calculated the weighted preva-
lence and odds along with their 95% CIs of symptoms of persistent
depression across financial assets, physical assets, and social assets.
Model 1 reported the unadjusted odds of symptoms of persistent de-
pression by each asset; thus, model 1 showed the bivariable relation
between symptoms of persistent depression and each asset, unad-
justed for any other variable. Models 2 through 6 reported the odds
of symptoms of persistent depression by each asset type, adjusting
for demographic characteristics of age, gender, race/ethnicity, and
household size (to account for sharing of assets within a household).
Thus, models 2 to 6 show the adjusted odds of symptoms of per-
sistent depression adjusting for age, gender, race/ethnicity, and house-
hold size, along with household income (model 2), household savings
(model 3), homeownership (model 4), education (model 5), and
marital status (model 6), respectively. Model 7 adjusted for all assets
and demographic characteristics together. Third, we assessed the
effect modification of assets on the relation between stressors and
symptoms of persistent depression and tested for interaction. To do
this, we estimated the predicted probability of symptoms of per-
sistent depression by the stressors of job loss, financial difficulties,
and relationship problems across each asset type, adjusting for de-
mographic characteristics and for the interaction of each asset with
each stressor. We used the margins command in STATA to calculate
the predicted probabilities of the interaction term combinations.
To estimate the predicted probabilities of symptoms of persistent
depression on interaction pairs, we used unweighted regres-
sion models in our margin commands, as consistent with recom-
mendations in the literature (46, 47) and reported unweighted
probabilities across interaction pairs for relevant comparison.
We graphed the predicted probabilities in a figure with 15 panels
representing each of the five asset and three stressor category
combinations.
A correlation table across different asset types is listed in table S1,
and a table with symptoms of persistent depression by asset types
and corresponding P value is listed in table S2. A table with predicted
probabilities and corresponding 95% CIs is listed in table 3. We con-
ducted maximum likelihood ratio tests to measure the difference
between each interaction term model and the relevant nested model
with only main terms; we determined significance of the interaction
terms using a P value cutoff of 0.05. Last, we calculated the prevalence
of symptoms of persistent depression by interaction group catego-
ries to understand the direction of the associations. Maximum like-
lihood ratio test results and unadjusted prevalence of symptoms of
persistent depression by asset and stressor groups are listed in table
S4 and fig. S1, respectively. Use of survey weights is described in
respective note of each figure and table. Analyses were conducted in
STATA 16.1.
SUPPLEMENTARY MATERIALS
Supplementary material for this article is available at https://science.org/doi/10.1126/
sciadv.abm9737
REFERENCES AND NOTES
1. A. Arora, A. K. Jha, P. Alat, S. S. Das, Understanding coronaphobia. Asian J. Psychiatr. 54,
102384 (2020).
2. D. Witteveen, E. Velthorst, Economic hardship and mental health complaints during
COVID-19. Proc. Natl. Acad. Sci. U.S.A. 117, 27277–27284 (2020).
3. E. L. Groshen, COVID-19’s impact on the U.S. labor market as of September 2020. Bus.
Econ. 55, 213–228 (2020).
4. M. H. Brenner, D. Bhugra, Acceleration of anxiety, depression, and suicide: Secondary
effects of economic disruption related to COVID-19. Front. Psychol. 11, 592467 (2020).
5. B. L. Perry, B. Aronson, B. A. Pescosolido, Pandemic precarity: COVID-19 is exposing
and exacerbating inequalities in the American heartland. Proc. Natl. Acad. Sci. U.S.A. 118,
e2020685118 (2021).
6. M. Daly, A. R. Sutin, E. Robinson, Depression reported by US adults in 2017-2018
and March and April 2020. J. Affect. Disord. 278, 131–135 (2021).
7. M. É. Czeisler, R. I. Lane, E. Petrosky, J. F. Wiley, A. Christensen, R. Njai, M. D. Weaver,
R. Robbins, E. R. Facer-Childs, L. K. Barger, C. A. Czeisler, M. E. Howard,
S. M. W. Rajaratnam, Mental health, substance use, and suicidal ideation during
the COVID-19 Pandemic — United States, June 24–30, 2020. MMWR Morb. Mortal. Wkly
Rep. 69, 1049–1057 (2020).
8. C. K. Ettman, S. M. Abdalla, G. H. Cohen, L. Sampson, P. M. Vivier, S. Galea, Prevalence
of depression symptoms in US adults before and during the COVID-19 pandemic.
JAMA Netw. Open 3, –e2019686 (2020).
9. E. E. McGinty, R. Presskreischer, H. Han, C. L. Barry, Psychological distress and loneliness
reported by US Adults in 2018 and April 2020. JAMA 324, 93–94 (2020).
10. A. Vahratian, S. J. Blumberg, E. P. Terlizzi, J. S. Schiller, Symptoms of anxiety or depressive
disorder and use of mental health care among adults during the COVID-19 pandemic —
United States, August 2020–February 2021. MMWR Morb. Mortal. Wkly Rep. 70, 490–494
(2021).
11. C. K. Ettman, G. H. Cohen, S. M. Abdalla, L. Sampson, L. Trinquart, B. C. Castrucci,
R. H. Bork, M. A. Clark, I. Wilson, P. M. Vivier, S. Galea, Persistent depressive symptoms
during COVID-19: A national, population-representative, longitudinal study of U.S. adults.
Lancet Reg. Health Am. 5, 100091 (2021).
12. O. Gruebner, S. R. Lowe, M. Tracy, S. Joshi, M. Cerdá, F. H. Norris, S. V. Subramanian,
S. Galea, Mapping concentrations of posttraumatic stress and depression trajectories
following Hurricane Ike. Sci. Rep. 6, 32242 (2016).
13. F. H. Norris, M. Tracy, S. Galea, Looking for resilience: Understanding the longitudinal
trajectories of responses to stress. Soc. Sci. Med. 68, 2190–2198 (2009).
14. R. C. Silver, E. A. Holman, D. N. McIntosh, M. Poulin, V. Gil-Rivas, Nationwide longitudinal
study of psychological responses to September 11. JAMA 288, 1235–1244 (2002).
15. K. A. Menninger, Psychoses associated with influenza: I. General data: Statistical analysis.
JAMA 72, 235–241 (1919).
16. R. C. Kessler, The costs of depression. Psychiatr. Clin. N. Am. 35, 1–14 (2012).
17. P. E. Greenberg, A.-A. Fournier, T. Sisitsky, C. T. Pike, R. C. Kessler, The economic burden
of adults with major depressive disorder in the United States (2005 and 2010). J. Clin.
Psychiatry 76, 155–162 (2015).
Ettman et al., Sci. Adv. 8, eabm9737 (2022) 2 March 2022
SCIENCE ADVANCES | RESEARCH ARTICLE
9 of 9
18. C. K. Ettman, G. H. Cohen, S. Galea, Is wealth associated with depressive symptoms
in the United States? Ann. Epidemiol. 43, 25–31.e1 (2020).
19. C. K. Ettman, G. H. Cohen, S. M. Abdalla, S. Galea, Do assets explain the relation between
race/ethnicity and probable depression in U.S. adults? PLOS ONE 15, e0239618 (2020).
20. C. K. Ettman, G. H. Cohen, P. M. Vivier, S. Galea, Savings, home ownership, and depression
in low-income US adults. Soc. Psychiatry Psychiatr. Epidemiol. 56, 1211–1219 (2020).
21. S. Diwan, S. S. Jonnalagadda, S. Balaswamy, Resources predicting positive and negative
affect during the experience of stress: A study of older Asian Indian immigrants
in the United States. Gerontologist 44, 605–614 (2004).
22. C. K. Ettman, S. M. Abdalla, G. H. Cohen, L. Sampson, P. M. Vivier, S. Galea, Low assets
and financial stressors associated with higher depression during COVID-19 in a nationally
representative sample of US adults. J. Epidemiol. Community Health 75, 501–508 (2021).
23. M. É. Czeisler, J. F. Wiley, E. R. Facer-Childs, R. Robbins, M. D. Weaver, L. K. Barger,
C. A. Czeisler, M. E. Howard, S. M. W. Rajaratnam, Mental health, substance use,
and suicidal ideation during a prolonged COVID-19-related lockdown in a region
with low SARS-CoV-2 prevalence. J. Psychiatr. Res. 140, 533–544 (2021).
24. C. Pieh, S. Budimir, E. Humer, T. Probst, Comparing mental health during the COVID-19
lockdown and 6 months after the lockdown in Austria: A longitudinal study. Front.
Psychol. 12, 625973 (2021).
25. D. Fancourt, A. Steptoe, F. Bu, Trajectories of anxiety and depressive symptoms during
enforced isolation due to COVID-19 in England: A longitudinal observational study.
Lancet Psychiatry 8, 141–149 (2021).
26. M. Pierce, S. McManus, H. Hope, M. Hotopf, T. Ford, S. L. Hatch, A. John, E. Kontopantelis,
R. T. Webb, S. Wessely, K. M. Abel, Mental health responses to the COVID-19 pandemic:
A latent class trajectory analysis using longitudinal UK data. Lancet Psychiatry 8, 610–619
(2021).
27. M. É. Czeisler, J. F. Wiley, C. A. Czeisler, S. M. W. Rajaratnam, M. E. Howard, Uncovering
survivorship bias in longitudinal mental health surveys during the COVID-19 pandemic.
Epidemiol. Psychiatr. Sci. 30, e45 (2021).
28. J. Breslau, M. L. Finucane, A. R. Locker, M. D. Baird, E. A. Roth, R. L. Collins, A longitudinal
study of psychological distress in the United States before and during the COVID-19
pandemic. Prev. Med. 143, 106362 (2021).
29. J. M. Twenge, T. E. Joiner, U.S. Census Bureau-assessed prevalence of anxiety and
depressive symptoms in 2019 and during the 2020 COVID-19 pandemic. Depress. Anxiety
37, 954–956 (2020).
30. E. Robinson, M. Daly, Explaining the rise and fall of psychological distress during
the COVID-19 crisis in the United States: Longitudinal evidence from the Understanding
America Study. Br. J. Health Psychol. 26, 570–587 (2021).
31. M. Daly, E. Robinson, Psychological distress and adaptation to the COVID-19 crisis
in the United States. J. Psychiatr. Res. 136, 603–609 (2021).
32. A. Shuster, M. O’Brien, Y. Luo, L. A. Berner, O. Perl, M. Heflin, K. Kulkarni, D. Chung, S. Na,
V. G. Fiore, X. Gu, Emotional adaptation during a crisis: Decline in anxiety and depression
after the initial weeks of COVID-19 in the United States. Transl. Psychiatry 11, 435
(2021).
33. E. E. McGinty, R. Presskreischer, K. E. Anderson, H. Han, C. L. Barry, Psychological distress
and COVID-19–related stressors reported in a longitudinal cohort of US adults in April
and July 2020. JAMA 324, 2555–2557 (2020).
34. CDC National Center for Health Statistics, Mental Health - Household Pulse Survey - COVID-19
(2021); www.cdc.gov/nchs/covid19/pulse/mental-health.htm.
35. C. P. McDowell, M. P. Herring, J. Lansing, C. S. Brower, J. D. Meyer, Associations between
employment changes and mental health: US data from during the COVID-19 pandemic.
Front. Psychol. 12, 631510 (2021).
36. N. Hertz-Palmor, T. M. Moore, D. Gothelf, G. E. DiDomenico, I. Dekel, D. M. Greenberg,
L. A. Brown, N. Matalon, E. Visoki, L. K. White, M. M. Himes, M. Schwartz-Lifshitz, R. Gross,
R. C. Gur, R. E. Gur, I. M. Pessach, R. Barzilay, Association among income loss, financial
strain and depressive symptoms during COVID-19: Evidence from two longitudinal
studies. J. Affect. Disord. 291, 1–8 (2021).
37. S. J. Lee, K. P. Ward, C. M. Rodriguez, Longitudinal analysis of short-term changes
in relationship conflict during COVID-19: A risk and resilience perspective. J. Interpers.
Violence , 8862605211006359 (2021).
38. W. T. Gallo, E. H. Bradley, J. A. Dubin, R. N. Jones, T. A. Falba, H.-M. Teng, S. V. Kasl, The
persistence of depressive symptoms in older workers who experience involuntary job
loss: Results from the health and retirement survey. J. Gerontol. B Psychol. Sci. Soc. Sci. 61,
S221–S228 (2006).
39. M. Tracy, F. H. Norris, S. Galea, Differences in the determinants of posttraumatic stress
disorder and depression after a mass traumatic event. Depress. Anxiety 28, 666–675
(2011).
40. H. Long, A. van Dam, A. Fowers, L. Shapiro, The covid-19 recession is the most unequal in
modern U.S. history. Washington Post; www.washingtonpost.com/graphics/2020/
business/coronavirus-recession-equality/.
41. K. Kroenke, R. L. Spitzer, J. B. W. Williams, The PHQ-9: Validity of a brief depression severity
measure. J. Gen. Intern. Med. 16, 606–613 (2001).
42. J. M. Dennis, “Technical overview of the AmeriSpeak panel: NORC’s probability-based
household panel” (NORC at the University of Chicago, 2021); https://amerispeak.norc.
org/Documents/Research/AmeriSpeak%20Technical%20Overview%202019%2002%2018.pd f.
43. S. M. Abdalla, C. K. Ettman, G. H. Cohen, S. Galea, Mental health consequences
of COVID-19: A nationally representative cross-sectional study of pandemic-related
stressors and anxiety disorders in the USA. BMJ Open 11, e044125 (2021).
44. W. Reynolds, The PHQ-9 works well as a screening but not diagnostic instrument
for depressive disorder. Evid. Based Ment. Health 13, 96–96 (2010).
45. R. K. Patel, G. M. Rose, Persistent Depressive Disorder (StatPearls Publishing, 2021);
www.ncbi.nlm.nih.gov/books/NBK541052/.
46. C. Winship, L. Radbill, Sampling weights and regression analysis. Sociol. Methods Res. 23,
230–257 (1994).
47. L. Avery, N. Rotondi, C. McKnight, M. Firestone, J. Smylie, M. Rotondi, Unweighted
regression models perform better than weighted regression techniques for respondent-
driven sampling data: Results from a simulation study. BMC Med. Res. Methodol. 19, 202
(2019).
Acknowledgments: We thank L. Sullivan for consulting on the methods used in this study.
Funding: Funding for T1 of the CLIMB survey came from the Rockefeller Foundation-Boston
University 3-D Commission grant number 2019 HTH 024 (S.G.). Funding for T2 of the CLIMB
survey came from the de Beaumont Foundation grant AGMT DTD 11/16/2020 (S.G.). Research
reported in this publication was supported by the National Institute on Minority Health and
Health Disparities of the National Institutes of Health under award number F31MD017133
(C.K.E.). The content is solely the responsibility of the authors and does not necessarily
represent the official views of the National Institutes of Health. Author contributions:
Conceptualization: C.K.E., P.M.V., and S.G. Data curation: S.G. and C.K.E. Formal analysis: C.K.E.
and L.T. Funding acquisition: C.K.E., S.G., B.C.C., R.H.B., and S.M.A. Methodology: C.K.E., P.M.V.,
S.G., L.T., I.W., and M.A.C. Supervision: S.G. and P.M.V. Writing–original draft: C.K.E. Writing–
review and editing: C.K.E., G.H.C., S.M.A., L.T., B.C.C., R.H.B., M.A.C., I.W., P.M.V., and S.G.
Competing interests: The authors declare that they have no competing interests. Data and
materials availability: All data needed to evaluate the conclusions in the paper are
present in the paper and/or the Supplementary Materials.
Submitted 31 October 2021
Accepted 6 January 2022
Published 2 March 2022
10.1126/sciadv.abm9737