Prevalence of Depression Symptoms in US Adults Before and During the COVID-19 Pandemic

IMPORTANCE The coronavirus disease 2019 (COVID-19) pandemic and the policies to contain it have been a near ubiquitous exposure in the US with unknown effects on depression symptoms. OBJECTIVE To estimate the prevalence of and risk factors associated with depression symptoms among US adults during vs before the COVID-19 pandemic. DESIGN, SETTING, AND PARTICIPANTS This nationally representative survey study used 2 population-based surveys of US adults aged 18 or older. During COVID-19, estimates were derived fromtheCOVID-19andLifeStressorsImpactonMentalHealthandWell-beingstudy,conductedfrom March31,2020,toApril13,2020.BeforeCOVID-19estimateswerederivedfromtheNationalHealth and Nutrition Examination Survey, conducted from 2017 to 2018. Data were analyzed from April 15 to 20, 2020. EXPOSURES The COVID-19 pandemic and outcomes associated with the measures to mitigate it.


Introduction
Coronavirus disease 2019 (COVID-19) and the policies to contain it have been a near ubiquitous exposure for people in the US in 2020.As an event that can cause physical, emotional, and psychological harm, the COVID-19 pandemic can itself be considered a traumatic event. 1 In addition, the policies created to prevent its spread introduced new life stressors and disrupted daily living for most people in the US.As of April 13, 2020, 42 states were under stay-at-home advisories or shelterin-place policies, affecting at least 316 million people in the US, or approximately 96% of the population. 2 The unemployment rate was reaching record highs in the US, with more than 20 million people filing for unemployment between the start of COVID-19 and mid-April 2020.
Mental health is sensitive to traumatic events and their social and economic consequences.
Previous studies on disruptions to life owing to disasters, epidemics, or civil unrest suggest that exposure to large-scale traumatic events are associated with increased burden of mental illness in the populations affected. 3For example, after September 11, 2001, 9.6% of Manhattan residents reported symptoms consistent with depression and 7.5% reported symptoms consistent with posttraumatic stress disorder. 4Residents living closer to the World Trade Centers had higher prevalence of mental illness. 4,5Similarly, after natural disasters, populations affected by hurricanes report an increase in symptoms consistent with mental illness. 3Increases in mental illness have also been documented after other epidemics, such as the Ebola virus and SARS outbreak. 6,7In addition, social disruptions in day-to-day living after civil unrest, for example, have been found to be associated with mental illness.Data from Hong Kong show greater levels of depression, anxiety, and psychological distress during the 2019 Hong Kong civil protests. 8,9portantly, the mental health consequences of mass traumatic events are not evenly distributed across populations.Having lower income and less wealth are associated with greater burden of mental illness. 10[13][14] Early evidence from published studies suggests that COVID-19 is associated with mental illness. 15Among health care workers in China who were exposed to patients with COVID-19, 50.4% reported symptoms of depression. 15A study of medical students in China identified elevated prevalence of anxiety. 16Another study found that reduced sleep was associated with greater levels of anxiety and stress among health care workers in China. 17To date, most studies regarding mental health and COVID-19 have been conducted in Asia and have focused on specific subpopulations, such as college students 16 and medical workers. 15,17Published studies from the US on mental health have been in purposive samples.Most relevant to this study, a study by Nelson et al 18

Methods
This study was approved by the institutional review boards of NORC at the University of Chicago and Boston University.All AmeriSpeak participants provided written informed consent during the enrollment process to join the AmeriSpeak standing panel.All NHANES participants provided written informed consent first for the household interview and then for the health examination.This study followed the American Association for Public Opinion Research (AAPOR) reporting guideline.

Population Sample During the COVID-19 Pandemic
The primary sample for this study was a nationally representative group of US adults aged 18 years or older using the AmeriSpeak standing panel.The AmeriSpeak panel is a probability-based panel that is representative of the US population by design.Households are randomly selected with a known, nonzero probability from the NORC National Frame, covering approximately 97% of US households.
People excluded from the sampling frame were people with PO box-only addresses, some addresses not listed in the US Postal Service Delivery Sequence File, and some newly constructed homes.
Adults aged 18 years or older who could speak English and who had completed an AmeriSpeak survey in the past 6 months were eligible to take the survey.In total, 1470 participants completed the survey, representing a survey completion rate of 64.3% of sampled panelists.The survey was conducted mainly over the internet (1385 participants [94.2%]) with a small number conducted over the telephone (85 participants [5.8%]).Twenty-nine participants missing data for depression were excluded from the sample.The final COVID-19 study sample included 1441 participants.
The survey was distributed by NORC at the University of Chicago from March 31, 2020, through April 13, 2020, assessing COVID-19 exposure, stressors, and mental health using the COVID-19 and Life Stressors Impact on Mental Health and Well-being (CLIMB) study questionnaire.Participants were contacted via web survey and follow-up was conducted via telephone interview.Several demographic questions (eg, sex, age, self-reported race/ethnicity, educational status, and marital status) were previously assessed for all AmeriSpeak panel members.As members of the AmeriSpeak panel, participants are invited to participate in several surveys per month and were paid a cash equivalent of $3 for completing this survey.

Comparison Sample Before the COVID-19 Pandemic
The comparison sample for this study, measuring mental health before COVID-19, was the National

Key Definitions Depression Symptoms
Depression symptoms in both studies were assessed using the Patient Health Questionnaire-9, a clinically validated survey with a sensitivity of 88% and a specificity of 88% at a cutoff score of 10 or higher. 19Depression symptom categories were defined as none (score, 0-4), mild (score, 5-9), moderate (score, 10-14), moderately severe (score, 15-19), and severe (score, Ն20). 19Binary classification of depression symptoms was defined by a score of 10 or greater.

COVID-19 Stressors Score
We assessed 13 stressors based on prior studies conducted after traumatic events. 20,21Examples of COVID-19 stressors included losing a job, death of someone close to you owing to COVID-19, and having financial problems.We excluded stressors that were capturing constructs applicable only to specific groups and created a score ranging from 0 to 13 to measure cumulative exposure to COVID-19 stressors.We divided the scores into thirds to measure low, medium, and high exposure to COVID-19-induced stressors.Cutoffs for stressor score categories were low (score, 0-2), medium (score, 3-4), and high (score, 5-13).

Demographic Characteristics
Sex was defined as a binary variable for men or women.Age was defined as a categorical variable with 3 groups: 18 to 39 years, 40 to 59 years, or 60 years or older.Race/ethnicity was defined as a categorical variable across 5 mutually exclusive categories: non-Hispanic White, non-Hispanic Black, Hispanic, non-Hispanic Asian, and other race/ethnicity (including multiple races/ethnicities).
Education was defined as a categorical variable with 4 groups: less than high school graduate, high school graduate or general education diploma equivalent, some college, and college graduate or higher.Marital status was defined as a categorical variable with 4 groups: married; widowed, divorced, or separated; never married; and living with partner.Household income was defined as a categorical variable with 4 groups, divided approximately at the interquartile range: $0 to $19 999, $20 000 to $44 999, $45 000 to $74 999, and $75 000 or more.Household savings was defined as a binary variable with savings of at least $5000.Savings was defined as "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."Household size was defined as the number of people living in a home with categories from 1 to 7 or more to protect participant identity.

Statistical Analysis
First, we calculated the demographic characteristics of the NHANES and CLIMB samples.Mobile examination center survey weights were used for NHANES data, and probability survey weights were used for CLIMB data.Second, we conducted bivariable χ 2 analysis to assess the association between demographic characteristics and depression symptoms in CLIMB and NHANES samples.Third, we estimated the prevalence and 95% CIs of depression symptoms in the US across categories before and during COVID-19 using CLIMB and NHANES samples.Fourth, we calculated the difference and ratio between estimates of depression symptoms during and before COVID-19.Fifth, we estimated the distribution of depression symptom categories before and during COVID-19 in the US using CLIMB and NHANES samples.Sixth, we used multivariable logistic regression to estimate odds ratios (ORs) and 95% CIs for the association between COVID-19-induced life stressors and depression symptoms, controlling for demographic characteristics and resources; the model controlled for sex, age, race/ethnicity, household size, education, marital status, household income, household savings, and COVID-19 stressor score.We used complete case analysis for the multivariable analysis.Stata statistical software version 16.1 (StataCorp) was used for statistical analyses.P values were 2-sided, and statistical significance was set at P = .05.Data were analyzed from April 15 to 20, 2020.e Two-tailed χ 2 analysis conducted for significance testing.
prevalence was 1.5-fold higher for mild depression symptoms, 2.6-fold higher for moderate depression symptoms, 3.7-fold higher for moderately severe depression symptoms, and 7.5-fold higher for severe depression symptoms categories during COVID-19 compared with before COVID-19.
The Figure presents depression symptom scores from the Patient Health Questionnaire-9 grouped by category before and during COVID-19.There was a greater distribution of scores between 0 and 4 before COVID-19; the CLIMB sample had a greater distribution of depression symptoms than the NHANES sample for all scores greater than 5. Thus, we saw a rightward shift in symptom burden in the sample during COVID-19 compared with the sample before COVID-19.
Among the CLIMB sample, after controlling for sex, age, race/ethnicity, household size, education, marital status, household income, household savings, and COVID-19 stressor score, participants with lower social and economic resources and with higher COVID-19 stressor scores had higher odds of depression symptoms compared with participants with higher social and economic resources or lower COVID-19 stressor scores (

Discussion
This survey study found that prevalence of depression symptoms in the US increased more than knowledge, this is the first nationally representative study that assessed depression symptoms using the Patient Health Questionnaire-9 in US adults before and during the COVID-19-pandemic.We found a shift in depression symptoms, with fewer people with no symptoms and more people with more symptoms during COVID-19 than before COVID-19.We also found that lower income groups were at greater risk of depression symptoms than higher income groups, and that having less than $5000 in household savings was associated with 1.5-fold increased odds of depression symptoms, or 50% greater risk.Additionally, we found that people with exposure to more stressors had greater odds of depression symptoms.
Findings from a 2014 review 3 on trauma and mental health suggest that depression increases during and after traumatic events; our study adds to this literature.A 2020 study by Ni et al 8 analyzed depression symptoms before and after political unrest in Hong Kong using the same measure of depression symptoms we deployed in this study.They reported national depression symptoms prevalence before the unrest to be 6.5% (compared with 8.5% in our pre-COVID-19 US sample) and 11.2% in 2019 during unrest (compared with 27.8% in our during-COVID-19 sample).This suggests that the impact of COVID-19 on the US population may be substantially larger than that after other large-scale events.This may reflect the greater ubiquity of COVID-19 and its effects on the US population than prior recorded large-scale traumatic events.
analyzed concerns about COVID-19, symptoms, and responses to the pandemic across 9009 completed surveys distributed over social media.They found that 67.3% of participants were very or extremely concerned about COVID-19 and that 48.8% of participants were self-isolating most of the time to avoid COVID-19.To our knowledge, the mental health of the broader US population during COVID-19 has not been documented.Aiming to address this gap in understanding, we assessed the burden of depression symptoms in the US during COVID-19 using the same measures deployed in representative national surveys before COVID-19 began.We also aimed to understand the factors associated with depression symptoms during and before COVID-19.JAMA Network Open | Public Health Prevalence of Depression Symptoms in US Adults Before and During the COVID-19 Pandemic Health and Nutrition Examination Survey (NHANES), a nationally representative group of noninstitutionalized civilian US adults aged 18 years or older.The NHANES is an annual crosssectional survey conducted by the US government.Participants are selected through a 4-stage probability sampling design, selecting primary sampling units by the county-level, then by census block-level, and then by households in the 50 states and the District of Columbia.The AmeriSpeak panel's sampling frame included 97% of US households, in all 50 states and the District of Columbia, and used a 2-stage probability sample design, first at the county-level and then at the census blocklevel.Thus, in this way, the NHANES sample is a suitable comparison group for the nationally representative AmeriSpeak panel.Collection of NHANES is administered through household interviews and interviews in a mobile examination center, measuring physical and mental health with a Computer Assisted Personal Interview for the depression screener, similar to an online survey questionnaire.There were 9254 participants in the NHANES 2017 to 2018 cycle.The NHANES sample used in this study excluded 3398 individuals (36.7%) younger than 18 years and 791 individuals (8.5%) missing depression data.The final NHANES sample included 5065 participants.JAMA Network Open | Public Health Prevalence of Depression Symptoms in US Adults Before and During the COVID-19 Pandemic

Table 1 .
Prevalence of Depression Symptoms in US Adults Before and During the COVID-19 Pandemic Results A total of 1470 participants completed the CLIMB survey for a completion rate of 64.3%.Of these, 1441 participants were included in the final sample.A total of 619 participants (43.0%unweighted, 38.0% weighted) in the total sample were aged 18 to 39 years, 723 (50.2% unweighted, 48.1% weighted) were men, and 933 (64.7% unweighted, 62.9% weighted) were non-Hispanic White.The pre-COVID-19 NHANES sample included 5065 participants (1704 participants [37.8%] aged 18-39 years; 2588 [51.4%] women; 1790 [62.9%] non-Hispanic White).Table 1 presents demographic characteristics of the NHANES and CLIMB study participants, prevalence of depression symptoms for each sample weighted to the US population, and distribution of depression symptoms by Demographic Characteristics of Nationally Representative Samples of US Adults Before and After the COVID-19 Pandemic and Association With Depression Symptoms

Table 3 .
Odds of Depression Symptoms by Resources and Exposure to COVID-19 StressorsComplete case analysis used for multiple logistic regression resulting included 1386 participants for this model.Model controls for demographic characteristics (ie, sex, age, race/ethnicity, and household size).Depression symptoms defined as PHQ-9 score cutoff of 10 or greater.COVID-19 stressor score calculated from stressor summation ranging from 0-13; categories represent low (score, 0-2), medium (score, 3-4), and high (score, 5-13) exposure to stressors due to COVID-19.Data on household income were missing for 34; on household savings, 45 participants; and COVID-19 stressor score, 3 participants. a